What Is Osteoporosis?

Why Remodeling?

It appears that this continuous remodeling is characteristic of the big, densely "Haversian" bones of long-lived humans, horses, and elephants and ornithischian, saurischian, and certain therapsid dinosaurs (Bakker, 1986; Enlow and Brown, 1957, 1958; Rensberger and Watanabe, 2000), but not the tiny non-"Haversian" bones of short-lived mice and rats. The reason for this is simple. Big bones have more flaws than small bones and will therefore have shorter fatigue lives than smaller bones equally loaded/unit volume (R.B. Martin, 2003; D. Taylor et al., 1999). R.B. Martin (2003) has given a dramatic example of this size effect. He has estimated that the volumes of the femurs of a shrew, a human, and an elephant are in the relative proportions of 1:900:30,000. If these bones couldn't remodel, the fatigue life of the elephant's femur would be extremely short compared to the shrew's femur when repeatedly subjected to the same physiological strain. In the same paper, R.B. Martin gave another example of the relation between the probabilities of failure of equally loaded, equidiameter iron wires 1, 2, and 3 units in length. The failure probabilities were .63, .87, and .95, respectively. Therefore, the progressively larger and longer bones of the vertebrates that evolved in the Dinosauria and later in the Mammalia were intrinsically more prone to continuous fatigue damage. On the other hand, despite increasingly heavy musculature, the skeletons of these larger animals were selected to be as gracile as possible to give their owners the maximum mobility for their size and thus the maximum possible survivability (R.B. Martin, 2003). But there was an obvious and dangerous downside to this failure to increase bone size in step with muscle mass—increasing bone microdamage during physiological activity. The solution to this problem was the increasing use of the remodeling mechanism during normal activity. Thus, they had to routinely mobilize BMUs to repair the microcracks continuously appearing at high strain sites in the big bones caused by their owners' heel, hoof, and paw strikes (Burr, 1993; R. B. Martin, 2000a, 2000b,2003; Tami et al., 2002; Verborgt et al., 2000; Whitfield et al., 2002b). Actually, the continual pulling on the ribs by muscles just during breathing causes them to have more microcracks, and thus the greatest remodeling rate and production of Haversian canals, of all the cortical bone in the body (Frost, 1960)! This of course would explain Enlow and Brown's comment (Enlow and Brown (1958)) that "…if a form does possess Haversian tissue in any part of its skeleton, the rib is, with very few exceptions, Haversian". However, the BMU remodeling mechanism was probably available for exploitation, at least in the small mammalian ancestors, because rat bones do resort to Haversian remodeling when needed. If a rat bone such as the ulna, which has no osteons and normally no microdamage or remodeling, is cyclically super-fatigue-loaded enough to produce linear microcracks and more diffuse patches of damage, typical intracortical remodeling starts in a few days to remove the linear microcracks (Bentolila et al., 1998).

But as with all physiologically good things, we shall soon see that remodeling can be dangerous when its level rises above a certain point, as does in postmenopausal women because, for example, the remodeling-driven loading of cortical bone with new BMU-produced Haversian canals increases bone porosity and hence fragility (R.B. Martin et al., 1998).

The Road into the Boney Interior, Sloshing Fluids, Flow-metering Cilia, Tugged Integrins, Osteointernet Chattering

Figure 2

The quiet before the storm. The big bones of large (more...)

Figure 2

.

The quiet before the storm. The big bones of large animals such as humans and elephants, unlike the tiny bones of mice and rats which keep growing throughout their short lives, are prone to microcracking at muscle-loading sites and routinely use a mechanism to repair the cracks as soon as possible to prevent the microcracks accumulating up to macrocracks and mechanical failure. (Rats can switch on the same device if the damage is extensive enough.) In a normal intact human bone, there is a vast cellular signaling syncytium, the osteo-internet, which consists of osteocytes in their lacunocanalicular network that opens out onto the surface of a trabecula or the wall of a Haversian or Volkmann canal, where it contacts retired osteoblasts known as bone-lining cells—a kind of cellular wallpaper. When the network is intact the blood flowing through the canal's blood vessels or sinusoids near the trabeculae has osteoclast precursors sailing peacefully by among the various leucocytes and carries nutrients and oxygen for the network's osteocytes. The nutrients and dissolved oxygen cross over the lining and are sucked through the tiny openings and travel along the "external pathway" of the canaliculi (the canalicular pores) (E-N (nutrient) left side line) when the network expands and wastes are pushed along the "external pathway" and expelled through the ports and into the blood (E-WD (waste disposal) right side line) when the network is squeezed by the alternating expnasion and compression of the bone in an activity cycle. The canaliculi are busy places—it has been said that canaliculi can be cleared as many as 2100 times in 24 hours (reviewed by Knothe Tate, 2000). Along with the traffic moving along the external pathway (E-N lines), the osteocyte pushes components along its processes and through gap junctions with other osteocytes and lining cells with waves of contraction of actin microfilaments (nanomuscles) (cell-cell connector lines). Aside from this very busy supply and waste disposal traffic, the lining cells are tranquillized by a stream of signalers being sent along the internal and external pathways from the normally strained osteocytes.

fig. 2

The osteocytes are the bones' professional strain monitors that are programmed to respond when the strain is above or below a particular bone's expected range (Burger, 2001; Klein-Nulent et al., 2003). Osteocytes are the chosen few—the 4% of the osteoblast teams that made the bone but did not have to self destruct after a cortical tunnel or trabecular trench was filled and the big, no longer needed, osteoblast crew had to be laid off. In other words they are retired osteoblasts that were saved from apoptotic self destruction by being locked up in the new bone. But they had to undergo a career change and shrink down from plump osteoblasts into lacuna-locked, spider-shaped chemical and strain sensors as they put away their bone-making tools.

fig. 2

fig. 2

fig. 2

The osteocytes' processes tend to be directed perpendicularly to the bone surface, i.e., the vascular surface, which means that their prime communication is with lining cells, and moreover some processes can extend beyond the bone surface into the adjacent blood vessels and in the case of adjacent marrow to send information about loading directly to nests of early osteoblast and osteoclast progenitors (Kamioka et al., 2001; Klein-Nulent et al., 2003; Palumbo et al., 1990). In other words, they are part of a giant cellular syncytium gated by large gap junctional pores. However, molecules that are too big to pass through the gap junctions can diffuse along or, in an actively bending bone, be swept along in the sloshing extracellular fluid, via the external pathway at a rate depending on canalicular diameter and pushing and sucking intensities (Knothe-Tate, 2001; Tami et al., 2002; L. Wang et al., 2000). However, the efficiency of the nutrient and factor delivery and waste removal falls off with the distance of the osteocytes from the osteonal or marrow canalicular ports, and these distant cells may therefore be more likely to be more vulnerable to stress and less responsive to any injected or swallowed factor (L. Wang et al., 2000). However, it must be noted that osteocytes do try to compensate for this long distance (or "remoteness") factor by having more processes and canaliculi than those nearest the Haversian canal, much as trees increase their root systems to get enough water and nutrients in dry regions (Knothe-Tate, 2001).

Let's now follow molecules, such as one of the PTHs, that will figure so prominently further on from syringe in skin to the nutrient artery of a femur. Their itinerary has been supplied byKnothe-Tate (2003). The molecules in the marrow blood vessels first see the cells lining the trabeculae, where some will stay and stimulate while others pass through into the trabeculae. The rest are carried by the capillaries into the Haversian channels of the cortical bone. The PTHs are pushed through the vascular walls under a pulsing head of pressure as cyclic bone loading periodically squeezes the bone marrow and its blood vessels. The peptides pass through the layer of bone-lining cells that serve as endosteal and osteonal wall "tiles", regulators of the movements of Ca²⁺ and K⁺ into and out of the canalicular fluid, and reversibly retired osteoblasts that can be a rapidly mobilizable "National Guard" to provide osteoblasts for crack-repairing BMUs. The fluid carrying the PTHs then passes through the 500-600-nm canalicular ports. The interlamellar matrix is also a molecular sieve, a slow pathway, through which the 3-10 kDa PTH fragments (but not molecules larger than 40 to 70 kDa) can diffuse. The fluid then flows along the space between the osteocyte surface and the canaliculus wall. The space is not empty—it contains collagens and proteoglycans that impede the PTH progression. However, in the case of the PTHs, they can help their progression along the canaliculus by increasing the porosity of the collagen-proteoglycans by stimulating osteocytes to line the lacuocanalicular walls with hyaluronan, a macroporous (big-pored) molecule (Midura et al., 2003). The osteocye also pushes the fluid along its canaliculi with its pulsing microvilli (Knothe-Tate, 2003). When the fluid and its peptide passengers near the lacuna containing the osteocyte's cell body with its sensory cilium that we will talk about later, they hit a kind of gate or valve formed by the cell processes which favors flow out of, rather than into, the lacuna (Knothe-Tate, 2003).

While the cells and their fluid-filled lacunocanalicular network make up only about 1% of the bone volume, the total surface area of these pipelines is immense! In an adult male skeleton, it is 1200 m², while the surface areas of the Haversian and Volkmann's canals and trabeculae are only 3.2 and 9m², respectively (Johnson, 1966; R.B. Martin et al., 1999). Thus, bone is really a stiff, dense, fluid-filled, calciferous sponge. And the rate of flow of fluids and their cargoes into, through, and out of the lacunocanalicular sponge is determined by how often and how tightly the sponge is squeezed by breathing in the case of the ribs and by walking, running, and other activities (Burger et al., 2003). This immense osteocyte internet is the "cell phone" with which they tell each another and lining cells about strains and microcracks and mobilize the appropriate responses.

fig. 2

Figure 3

The storm hits! A microcrack slices through the dense (more...)

Figure 3

.

The storm hits! A microcrack slices through the dense tangle of canalicular communication lines through which the osteocytes received nutrients and oxygen, got rid of their wastes, and sent a stream of strain-generated signals to the retired osteoblasts, the B-LCs, lining the osteonal (or trabecular) surface. The cutting off of communications causes the osteocytes (OCs) to begin running out of fuel, asphyxiating and drowning in their own wastes. The stressed osteocytes trigger the self-destructive apoptosis mechanism and in the process send distress and dying signals and cellular scents that cause the blood vessel lining cells to velcroize and grab appropriately addressed osteoclast precursors from the circulation. These grabbed cells then invade the damaged patch and fuse with each other to form the first wave of osteoclast diggers of a new BMU. One of the things that attracts these and subsequent osteoclast precursors and vulturelike mature osteoclasts is the "scent" of the phosphatidyl serine-decorated membranes of the dying osteocytes' dendrites poking out of the canalicular pores in the osteonal or trabecular wall.

Figure 4

Because of ruptured supply lines, the starving and (more...)

Figure 4

.

Because of ruptured supply lines, the starving and suffocating osteocytes bordering a microcrack start the self-destructive apoptotic mechanism, which is driven in part by the increasingly expressed Bax protein. However, to stop the death zone from spreading, a living anti-apoptosis firewall is set up when signals from the dying osteocytes induce outlying osteocytes to turn on their anti-apoptosis Bcl-2 and anti-Bax genes.

fig. 3

fig. 4

The normal patterns of squeezing and stretching of the bones of an active skeleton, particularly in the hip and distal leg bones by heel striking and in the ribs by breathing, pushes and sucks the fluid in the osteocytes' lacunas and canaliculi back and forth and into and out of the canaliculi's pores in the osteonal walls to produce pulsing shear forces on the cells and their processes (Cowin, 1999; Burr et al., 2002; Ehrlich and Lanyon, 2002; Gooch and Tennant, 1997; Klein-Nulent et al., 2003; Kufahl and Saha, 1990; Knothe-Tate, 2001,2003; Knothe-Tate et al., 2004; L. Wang et al., 1999,2000; You et al., 2001). According to Burger et al. (2003), Knothe-Tate (2003), Knothe-Tate et al. (1998, 2004), Piekarski and Munro (1977), Smit et al. (2002) and L. Wang et al. (2000) strain pulses in long bones, such as tibia and femur from heel strikes during the walking cycle or in the ribs caused by muscle pulling during breathing cycles, cause fluid to flow through the 3-D lacunar-canalicular network to the bone surface and back again—it sloshes back and forth. Static strain is a one-shot suck or push with a change in position without the subsequent pumping action needed to drive osteocyte signaling—the fluid in the lacunas and canaliculi must slosh back and forth within a certain frequency range to keep the osteocytes informed of the level of bone function and signaling appropriately to its fellows in the osteointernet (Burr et al., 2002; Ehrlich and Lanyon, 2002; Gooch and Tennant, 1997). As I said above, cyclic pumping is what keeps food and oxygen coming to the osteocytes and their wastes being carried away. Thus, stopping the pumping strangles the osteocytes. This is why Lanyon and Rubin (1984) found that cyclically loading turkey wing ulna induced new bone formation while stopping the pumping by static loading caused holes to appear in the cortex.

But there is an interesting aspect of strain cycles that affects osteogenic signaling in long bones. Srinivasan and Gross (2000), using the mid-diaphysis of the avian wing ulna, reported that flow rates are maximum during the first load cycle in which fluid is squeezed out of the osteocyte lacunae into the canaliculi. When the load is lifted, the fluid flows back into the lacunae. But if the next loading happens before the lacunae are refilled, there is less fluid to be squeezed into the canaliculi. If this continues, fluid flows and with them osteocyte signaling drop off to produce a steady state equilibrium. This means that if fluid pulsing and signaling are to be maintained at an optimal level, there must be rest periods between the loadings to allow the lacunae to be fully refilled. Srinivasan et al. (2002, 2003) confirmed this by measuring the effects of cyclic loading with and without rest periods on bone formation in avian ulna and the tibias of young, adult, and aged mice. In all cases, inserting rest periods between load cycles enhanced the osteogenicity of low-level loading protocols. For example, 5 consecutive days of 100 low-level loading did not affect the tibial periosteal bone formation rate, but putting 10 seconds of rest between every 10 cycles increased the signaling so that the periosteal bone formation rate increased 8-fold (Sirinvasan et al., 2002).

Figure 5

Most cells, osteoblasts, and osteocytes included (more...)

Figure 5

.

Most cells, osteoblasts, and osteocytes included (Ilaria del Pra, Ubaldo Armato and your author, as yet unpublished observations) appear to have a strain-sensing device known as the primary or (solitary) cilium that extends into the extracellular fluid and matrix from the mother-daughter pair of centrioles in a cell's microtubule-organizing centrosome lying beside the nucleus. A striking example of one of these cilia is illustrated here. This one, which belongs to a Ptk kangaroo rat's kidney epithelial cell, looks a lot like a car aerial and is at least the cell's fluid flowmeter, but it might also be equipped with receptors to detect various hormones and other things flowing by in the fluid. The bending back and forth of such a cilium by sloshing extracellular fluid, for example, perhaps in an osteocyte's lacuna (Whitfield, 2003a), sends a stream of signals into and through gap junctionally connected cells like flashes of light through a fiberoptic network (Nauli et al., 2003; Praetorius and Spring, 2001, 2003; see also (fig. 6). The frequency and intensity of the signal pulses depend on the frequency and extent of bending of the cilium. This photograph was generously sent to me by Dr. Sam Bowser of the Wadsworth Center, Albany, NY.

Figure 5

Figure 6

How the bending of a femoral (for example) osteocyte's (more...)

Figure 6

.

How the bending of a femoral (for example) osteocyte's primary cilium by fluid sloshing back and forth in the cell's lacuna, when the bone's owner is walking, running, or breathing can send signals streaming through the osteointernet. The bending of the cilium prys open Ca2+ channels in the cilium's membrane through which surges a bolus of Ca²⁺. This Ca²⁺ bolus "lights up" the cell. It triggers a shower of events such as the stimulation of nuclear CaMKIV (Ca²⁺•calmodulin-dependent protein kinase IV), a surge of cyclic AMP-dependent protein kinase catalytic subunits into the nucleus which stimulates various gene activities, the stimulation of NO (nitric oxide) by Ca²⁺•CaM-responsive eNO-synthase, and the release of glutamate to stimulate cells such as preosteoblasts and osteoblasts waiting at the canalicular ports in the osteonal or trabecular wall that have glutamate receptors. The Ca²⁺-triggered signal is sent through gap junctions to adjacent cells by the IP3 produced by the chopping up of the membrane phospholipid PIP₂ by PLC activated by the ciliary stretching. The Ca²⁺ bolus cannot pass into other cells because Ca²⁺ closes gap junctions. However, the IP₃ takes over and moves through the junctions and indirectly transmits the Ca²⁺ signal from the bending cilium by easily passing through the gap junctions and triggering the release of Ca²⁺ from the neighbors'calcium-storage vesicles. The reader must not forget that the signaling from the cilium on the cell body in the lacuna collaborate with signals from the tugging of the integrins anchoring the cell's processes to the walls of the canaliculi.

fig. 6

fig. 6

fig. 6

fig. 6

fig. 6

Disrupting the kidney cell's cilium eliminates its ability to sense changes in extracellular fluid flow rate (Praetorius and Spring, 2003a). The horrendous consequence of this "deafening" in both humans and mice—is polycystic kidney disease (PKD) caused by disrupting the kidney tubule cells' gene for PC-1 or PC-2 (Calvert, 2002; Pazour and Witman, 2003; Yoder et al., 2002). If I am right and the osteocyte's primary cilium is used to measure bone strain, and if the osteoblasts (remember they too have them) use their cilia for detecting and initiating the reponse to various things such as growth factors and hormones in their extracellular fluid, PKD should be accompanied by severe skeletal deformation. And this is precisely what happens when the mouse's PC-1-encoding Pkd1 gene is disabled (Boulter et al., 2001; W. Lu et al., 2001)!

Signals from the osteocytes' primary cilia would be accompanied especially at higher strains by the tugging on the signal-generating integrins and associated cytoskeleton that attach the cell and its processes to the osteopontin lining the walls of the lacunae and canaliculi like waves cause a ship to pull on its mooring lines (matricrine signaling). But in this case, the cells' mooring lines are linked to sophisticated signaling instruments (Denhardt et al., 2002; Miranti and Brugge, 2002). This tugging causes the clustering of integrins with signalers and periodic collection of the actin cytoskeleton into thick, oriented stress fibers with a consequent activation of signaling enzymes on actin-linked cell membrane rafts and mechanosensitive Ca²⁺ channels (Burr et al., 2002; N.X. Chan et al., 2000; Duncan and Misler, 1989). The upshots of this normal pumping and shearing are Ca²⁺ ions surging through mechanosensitive channels and spiking signalers such as NO from the Ca²⁺-stimulated NO-synthase, prostacyclin (PGI₂) and prostaglandin-E₂ (PGE₂) produced specifically by the mechosensitive COX-2 (cyclooxygenase-2), and, as we will soon see, glutamate (the well-known excitatory neurotransmitter until recently believed to be used only by central neurons) and serotonin, that passes from osteocyte to osteocyte through gap junctions an/or bounces from receptor to receptor to keep the bone lining cells from unnecessarily ordering vascular endothelial cells to start the repair process by grabbing passing osteoclast precursors (A.D. Bakker et al., 2003; Bliziotes et al., 2001; N.X. Chan et al., 2000; Chenu, 2002; Duncan and Misler, 1989; Edlich et al., 2001; Mason, 2004; Ehrlich and Lanyon, 2002;Kufahl and Saha, 1990; Manolagas, 2000; R.B. Martin, 2000; Noble, 2000; Noble and Reeve, 2000; Noda et al., 1996; Nomura and Takano-Yamamoto, 2000; Parfitt, 1998; Pavalko et al., 1998, 2003; Ryder and Duncan, 2001; Schaffler, 2000; Skerry, 1999; Skerry and Genever, 2001; A.F. Taylor et al., 2002; Westbroek et al., 2001).

The signals from bending cilia, stretched cell membranes, and tugged cadherins and integrins turn on osteogenic genes such as the gene for the strongly osteogenic PTHrP, which will be a very important part of things to come further on in our story (X. Chen, C. Macica et al., 2003). How? The stimulation of the PTHrP gene is mediated by the opening of stretch-activated TREK-family K⁺ channels (X. Chen, C. Macica et al., 2003). The ciliary bending and membrane deformations also open the cilium's PC-2 channel and another mechanosensitive ion channel in the cell membrane known as ENaC (SA-CAT-stretch-activated cation channel) through which Na⁺ surges into the cell to depolarize the membrane and thus open L-type VSCCs (voltage-sensitive Ca²⁺ channels) (Kizer et al., 1997; Pavalko et al., 2003). The result of this ionic bombardment is a cascade of Ca²⁺-driven and other events that create complexes called mechanosomes that convert the mechanical deformation into biochemical responses and gene expressions and the generation of osteogenic factors such as PTHrP (Pavalko et al., 2003).

The mechanism goes something like this. The nonreceptor tyrosine kinase FAK and cSrc in the tugged integrin-associated focal adhesion plaques are activated and phosphorylate the Crk and p130^Cas in Crk•p130^Cas complexes in the plaques. The phospho-Tyr-p130^Cas attaches to Nmp4/CIZ and kicks Crk out of the complex to form one of the "mechanosomes"— phospho-Tyr-p130^Cas•Nmp4/CIZ (Kirsch et al., 2000; Nakamoto et al., 2002; Pavalko et al., 2003; Yamada and Even-Ram, 2002). The activated FAK kinase triggers the Ras → Raf → MEK 1/2 → ERK 1/2 cascade, the last of which phosphorylates and thus enhances the activity of the Cbfa1/Runx-2 transcription factor that drives the expression of key osteoblast genes (Franceschi and Xiao, 2003). Meanwhile, the tugged cadherins in a lining cell's adherens attachment junctions with its neighbors also release β-catenins, which associate with LEF-1 protein to form a β-catenin•LEF1 mechanosome.

The basic "take-home" message from this complicated cluster of cell surface deformation-triggered events is really quite simple to remember. Fluid tugging enhances the activity of the master osteoblast transcription factor Cbfa1/Runx-2 and generates at least two mechanosomes that are fired into the nucleus. There they bind to target sites on the nuclear matrix and then together with Cbfa1/Runx-2 grab, bend and twist the promoters of their bone-making target genes into shapes that enable them to be switched on. One of these mechanosensitive genes codes for PTHrP which, as we shall see further on, is a towering figure in osteoblast differentiation (X. Chen, C. Macica et al., 2003; Daifotis et al., 1992; Pirola et al., 1994; Steers et al., 1998). The steady strain-driven maintenance of a supply of PTHrP by osteocytes could, for example, facilitate crack repair by promoting the flow of osteoprogenitors into the mature BMU osteoblast pool.

To make the strain-response story more complicated, stretching does something rather astonishing—it can activate MLO-Y4 osteocytic cells' nongenomic (membrane-based) estrogen receptors, the signals from which send ERK2 into their nuclei and oppose the initiation of apoptosis by etoposide (Aguirre et al., 2003). In other words, strain can activate estrogen receptors without estrogen (i.e., ligand-naive, receptors)!

In general terms, it seems as if moderate normal body movements keep signals flowing steadily through the extensive osteocyte network (which because of gap junctions is really a giant gated syncytium) from bending cilia. But more strenuous body movements add signals from cadherin and integrin tugging.

As we shall see in the next section, in addition to reshaping gene promoters the intercellular chattering caused by mechanical strain looks a lot like neural network crosstalking. Maybe osteocytes could be called "honorary neurons". An activated neuron sends a wave of Na⁺ influx-induced membrane depolarization along its axon, which when it hits the axon terminus, opens Ca²⁺ channels. This triggers the release into the synaptic cleft of packets of a neurotransmitter such as L-glutamate, which activates receptors on a post-synaptic neuron. Amazingly, the same Na⁺/Ca²⁺ shifts caused by the punching and pulling of an osteocyte and the tweaking of its cilium by its owner's heel strikes may also cause the cell to release glutamate (Mason, 2004; A.F. Taylor, 2002)!

fig. 3

Calling Diggers with the Neuron-Like Signaling Machinery of Osteocytes, Osteoblasts, and Osteoclasts

"It is clear that neurotransmitters have profound effects on bone, influencing the differentiation, proliferation, activity, and apoptosis of osteoblasts and osteoclasts."—Spencer et al. (2004).

You may remember that I pointed out at the beginning of this chapter that bone cells unexpectedly share with neurons such things as memory and the ablility to exchange information with each other through extensive networks. Well, this sharing looks as if it runs much deeper than expected. It seems that the major neurotransmitters, such as glutamate and serotonin, may be major osteoctye→osteoblast "osteotransmitters" as well as nerve→lining cell "neuro-osteotransmitters" in the osteonal canals that help initiate the response to loading and microcracking (Bliziotes et al., 2001; Chenu, 2002; Hinoi et al., 2004; Skerry and Genever, 2001; Serre et al., 1999; Skerry and Taylor, 2001; Spencer et al., 2004; Westbroek et al., 2001). This raises the possibility of there being osteosynapses that are similar to neuronal synapses and T-lymphocye-dendritic (Langerhans) cell synapses (Dustin and Colman, 2002). But so far no such synapses have yet been seen, although osteocyte processes have been found terminating "suggestively" on osteoblasts during bone formation (Menton et al., 1984).

fig. 6

These receptors and the synapse-like connections they rather loudly suggest are not mere osteoblast or preosteoblast ornaments. Inhibiting the AMPA/kainate and NMDA receptor signaling reduces the production and differentiation of osteoblasts and causes precursor cells to become adipocytes rather than osteoblasts (Dobson and Skerry, 2000; Skerry and Genever, 2001; Taylor et al., 2000). All of this evokes the image of a network of nerves, precursor cells, osteoblasts, osteocytes, and maybe lining cells talking to each other in fluent glutamate!

O.K., but what has all of this to do with "CALLING THE DIGGERS", the title of this section? While a strain-induced glutamate surge stimulates osteoblast generation, it also stimulates osteoclast generation via the baby osteoclasts' wet nurses—immature bone marrow osteoblastic cells (Atkins et al., 2003; Kitazawa et al., 1999; G.P. Thomas et al., 2001). It stimulates the immature, osteoblastic marrow stromal cells to put RANKL on their surfaces while mature, matrix-making osteoblasts cannot make RANKL apparently because its gene's promoter is locked shut by having its CpG sequences methylated at some point during maturation (Atkins et al., 2003; Corral et al., 1998; Kitazawa et al., 1999; G.P. Thomas et al., 2001). RANKL is the ligand for the osteoclast precursors' RANK (Receptor Activator of N F-κB transactivator) receptors the signals from which stimulate osteoclast differentiation (A. Taylor et al., 2000). And the osteoclasts' glutamate-activated NMDA receptor/channels also stimulate differentiation via the NF-κB transactivator exactly as do the signals from RANKL-activated RANK receptors, (Black et al., 2002, 2003; Chenu, 2002; Espinosa et al., 1999; Hinoi et al., 2004; Itzstein et al., 2000; Laketic-Ljubojevic et al., 1999; Mentaverri et al., 2003; Merle et al., 2003; Patton et al., 1998; Peet et al., 1999; Skerry and Genever, 2001). Osteoclasts seem to need the Ca²⁺ flowing through these activated NMDA receptor/channels in order to mature and later to assemble the actin ring needed to seal off the excavation site (Itzstein et al., 2000; Mason et al., 1997). And the osteoclasts, short-lived cells that they are, also need the Ca²⁺ surges to stimulate their nNOS (neural nitric oxide synthase) to make the NO they need to hold off self-destructive apoptosis (Mentaverri et al., 2003). Indeed, blocking the NMDA receptors with MK801 or DEP (±-1-[1,2-dipenylethyl]piperidine) cause the cellular Ca²⁺ content to drop and the cells to self-destruct (Mentaverri et al., 2003).

What Summons the Fillers?

As the the vulturing "truffle hound" osteoclasts are tunneling through the cortical bone looking for dying osteocytes and in the process digging out the microcrack and a large patch of bone around it by spraying the protein-shredding cathepsin K and mineral-dissolving HCl onto the bone, they release a pack of factors (e.g., bone morphogenic protein (BMP)-2, fibroblast growth factor (FGF)-2, insulin-like growth factors (IGFs)-I and -II, IGF-binding protein (IGFBP)-5, TGF-βs) that were deposited in the matrix 2 to 5 years earlier by the osteoblasts of an ancient BMU. It is widely assumed in the Bone Community that these liberated factors, particularly the FGF-2, and the IGFs, form an immediately accessible store of osteoblast generators that can be drawn on by a subsequent BMU to locally increase osteoblast generation by stimulating osteoprogenitor proliferation to start new bone growth for repairing a microcrack. The TGF-βs would assist by stimulating osteoblastic stromal cells via Cbfa1/Runx-2 and TGF receptor-activated Smad proteins to reduce osteoclast generation by stimulating OPG expression and reciprocally reducing RANKL expression (Takai et al., 1998; Thirunavukkarnasu et al., 2000, 2001; Troen, 2003). However, they may merely be litter from a past repair job—the remnants of the autocrine/paracrine factors that the past osteoblasts were using to stimulate themselves—that may not be able to withstand the osteoclasts' cathepsin K and acid spray.

But the osteoclasts leave a trail of chemokine droppings that attract osteoblastic cells. This cytokine is Mim-1 (myb-induced myeloid protein 1) (Falany et al., 2001; Ponomareva et al., 2002; Troen, 2003). This peptide lures the preosteoblasts to the excavation site and at the same time stimulates the ability of the Cbfa1/Runx-2 to get at, and stimulate the expression of, key osteoblastic genes such as osteocalcin and then drive the mineralization of the new matrix (Ponomareva et al., 2002). Mim-1 can facilitate Cbfa1/Runx-2's access to the osteoblast-specific genes' promoters ("switch boxes") because it is a chromatin-decondensing, gene-accessing histone acetyltransferase.

Serotonin may also be put into the matrix by osteoblasts and subsequently released by the diggers from a future BMU. Osteoblasts and periosteal fibroblastic osteoblast precursors express serotonin receptors and have the machinery to release and reaccumulate it (Bliziotes et al., 2001; Westbroek et al., 2001). And Chopra and Anastassiades (1992) have reported that one of the matrix components—bone sialoprotein (BSP)—binds serotonin, receptors for which might be on the osteoblast primary cilium as they are in neurons in certain parts of the brain (Brailov et al., 2000; Hamon et al., 1999). Although what serotonin does in bone is unknown, it may be another of the several factors that stimulate the proliferation of serotonin-receptor-bearing osteoprogenitor cells which in turn generate crack-filling osteoblasts (Westbroek et al., 2001).

A lot of Ca²⁺ is also pumped out of the hole when the osteoclasts dissolve the bone mineral (hydroxyapatite) by directing a stream of H⁺ into the hole with their massively expressed vacuolar H⁺-ATPase and Cl^- counterions though Cl^- channels in the ruffled border (Blair, 1998)—as I said a couple of paragraphs ago, they actually spray the bone mineral with hydrochloric acid (HCl)! They ingeniously get the H⁺ of the HCl from water by first using their carbonic anhydrase to make H⁺ (HCO₃)^- (carbonic acid) by combining CO₂ with water, then they use their membrane (HCO₃)^-/Cl^- exchanger to pump out the (HCO₃)^- through their upper surfaces away from the hole and bring in Cl^- in exchange for the (HCO₃)^- and finally pump the H⁺ out with the electrogenic ATPase, which to maintain electroneutrality pulls Cl^- through its channels into the hole and onto the hydroxy apatite. This plume of Ca²⁺ is a very important two-way switch—it is an "off" switch for the osteoclasts and an "on" switch for osteoprogenitors coming behind them.

When the Ca²⁺ concentration in a tunnel or trench gets high enough, it activates Ca²⁺ sensors on the osteoclast's apical H⁺- and protease-pumping, finger-like ruffles sticking into the hole (Blair et al., 2002; Kameda et al., 1998; T. Yamaguchi, 2003; Zaidi et al., 1999). The resulting Ca²⁺ surges and signals cause the osteoclast to pull its acid-dripping fingers out of the hole, pull up the actin sealing ring it had put around the hole, and glide over to another part of the patch and start digging again. But if it cannot find another place to attach itself by its integrins, in time to generate the matricrine signals neeed to hold off apoptosis, it will self-destruct—a drastic response to unemployment and homelessness known as anoikis or homelessness-induced apoptosis (Frisch and Screaton, 2001; Lorget et al., 2000; Sakai et al., 2000; Stupack and Cheresh, 2002).

Figure 7

The signals from the CaRs (Ca²⁺-sensing receptors) (more...)

Figure 7

.

The signals from the CaRs (Ca²⁺-sensing receptors) sticking out of the cell membrane triggers the expression of BMPs (bone morphogenic proteins) which drive the development of osteoblasts by switching on the gene for Cbfa1/Runx-2, which in turn configures the promoters of various osteoblast-specific genes for activation by specific transcription factors. However, Cbfa1/Runx-2 also stimulates BMP expression, thereby setting up a self-sustaining maturation cycle.

fig. 7

fig. 7

As the osteoclast-attracting osteocyte killing stops and the load-driven sloshing resumes, the Ca²⁺ sensor-driven stimulation of ERK (extracellular signal regulated kinases) 1/2 in the cells stimulates the expression of eNOS and the production of NO (J. Rubin et al., 2003), the gaseous signaler which diffuses out of the cells to prevent local osteocytes from killing themselves and becoming osteoclast-attracting corpses. This NO also stimulates the enzyme, guanylyl cyclase, in nearby vascular cells by attaching to the enzyme's heme group (Lincoln et al., 1997). The cyclic GMP produced by the cyclase relaxes arterioles, the overall effect of which is to increase the delivery of blood-borne nutrients through the advancing blood capillaries to the filler crews (Lincoln et al., 1997). The NO-stimulated cyclic GMP in the osteoblastic cells also stimulates the cells' PKG (protein kinase G) which in turn may, like the Ca²⁺ sensor signals, also activate the ERKs 1/2, which in turn activate the cFos/Jun (AP-1) transcription factor complex and the various matrix-related genes it switches on (Pfeilschrifter et al., 2001; Zaragoza et al., 2002). And this in turn also shuts down the expression of RANKL, which turns off osteoclast generation (Rubin et al., 2003).

Where Do the Fillers Come From and Where Do They Go?

This is all very interesting, but where do the osteoblasts come from to make a new Haversian canal surrounded by concentric layers of bone? They come from a group of constantly proliferating, spindle-shaped mesenchymal cells that follow the osteocyte cadaver-hunting osteoclasts along the walls of the cutting cone, driven by the various mitogenic and chemotactic factors, mixed with the large amounts of Ca²⁺ and phosphate released from the bone by the osteoclasts to make a tasty osteogenic "soup" (Bianco et al., 1993; Doherty and Canfield, 1999; R.B. Martin et al., 1998; Parfitt, 2000a, 2000b, 2001; Schaffler, 2000). The band of proliferating mesenchymal cells advancing along the tunnel wall throw off preosteoblasts that dismantle their mitogenic machinery and become osteoblasts which start depositing layers or lamellae of bone around the tunnel walls. Unlike their progenitors, the new osteoblasts stay in place to lay down concentric 15-μm-thick strips of bone around the tunnel wall, while the proliferating progenitors move on behind the osteoclasts, throwing off new bands of osteoblasts as they go. The result of this is the layering of lamellae around the new blood vessels and nerves.

The question remains as to whether one generation or band of osteoblasts fills the tunnel (except of course the vascular "right-of way"), taking rest breaks between each pair of lamellae or whether several generations of osteoblasts are needed do the job, which would mean that osteoblasts are still being generated by progenitors on the new surfaces. Probably several osteoblast generations are needed for the refill job (Martin et al., 1998). As the refilling progresses, the rate of formation drops for at least two reasons. First, as the lamellan build-up approaches the central blood vessel, the shrinking space somehow interferes with osteoblast differentiation, and second, the progressive drop in stresses with the increasing lamellar build-up results in a fading of the strain signals needed for osteoblast differentiation that were at their peak in the microcrack and at the head of the cutting cone (R.B. Martin et al., 1998).

At a repair site on the trabecular or endosteal surface, the osteoblasts come from the preosteoblasts from the adjacent marrow and the lining cells that make up the canopy of the repair "blister" mentioned above (i.e., the BRC canopy) (Hauge et al., 2001).

If any of the BMP-2 liberated by the osteoclasts from the matrix has escaped being shredded by the osteoclasts' proteases, it would kick osteoprogenitors along the osteoblast differentiation pathway (Chen et al., 1998). They would then be stimulated to start working by the signals from other receptors which include the CaRs in the caveolar pouches in their cell membranes (Brown and MacLeod, 2001; Kifor et al., 1998) and the type 1 PTH/PTHrP (PTHR1) receptors (Aubin, 1998, 2000, 2001; Aubin and Triffitt, 2002; McCauley et al., 1996), the expression of which shoots up about 10-fold during the transition from mature osteoprogenitor to preosteoblast.

Another factor that drives the osteoprogenitor proliferation and maturation comes from the cells of the advancing blood vessels. That factor is endothelin-1 which binds to, and activates, the osteoprogenitors' endothelin-A receptors (Mohammad et al., 2003; von Schroeder et al., 2003).

Signaling by glutamate from the dying or hypoxic osteocytes or from invading or adjacent nerves is also needed. However, the osteoblasts also release glutamate using exactly the same vesicular release machinery as neurons (Bhangu et al., 2001). As we learned above, the osteoblasts are also equipped with neuron-like NMDA-type glutamate receptor/channels (Gu et al., 2002; Hinoi et al., 2003; Skerry and Taylor, 2001; Spencer et al., 2004), AMPA/kainate ionotropic receptor channels as well as the G-protein-coupled metabotropic mGluR1b receptors with their 7 transmembrane α-helices, which belong to the same GPCR (G-protein-coupled receptor) family as the Ca²⁺ receptors (Gu and Publicover, 2000; Hinoi et al., 2001; Skerry and Taylor, 2001; Spencer et al., 2004). Moreover, they have Na⁺-dependent GLAST (EAA1) transporters to suck the glutamate back into the cell, just as they do in neurons to keep the external glutamate low, which keeps the signaling noise down and prevents dangerously prolonged receptor activation by released glutamate (Chenu, 2002; Mason et al., 1997). The activated mGluRs, like activated Ca²⁺ sensors and the related PTHR1 receptors, trigger a burst of PLC (phospholipase C) activity, which triggers a prompt release of Ca²⁺from internal stores followed by the flow of Ca²⁺ through opened membrane channels and a surge of PKCs (protein kinase Cs) activity (Gu and Publicover, 2000).

The importance of activated NMDA receptor/channels at a critical point in the osteoblast maturation program for driving post-proliferative osteoblast differentiation by stimulating the expression and possibly the transport of the master osteoblast differentiation driver Cbfa1/Runx-2 to its target genes has been demonstrated by Hinoi et al. (2003) using freshly isolated rat calvarial osteoblastic cells. Thus, NMDA receptor/channel inhibitors interfere with the binding of the master osteoblast differentiation driver Cbfa1/Runx-2 to its DNA targets and thus the expression of key osteoblast genes such as alkaline phosphatase and osteocalcin and ultimately the arrival of the cell at the stage where it accumulates Ca²⁺ for matrix mineralization (Hinoi et al., 2003). The NMDA receptor/channels on the calvarial osteoprogenitors in Hinoi et al.'s experiments (A.F. Taylor, 2002) must have been continuously stimulated by the 500-μM glutamate in the α-MEM culture medium, with no relief being possible from the GLAST transporters, as would be the case in a bone. (This incidentally suggests the extremely important possibility that glutamate has been a covert signaler for osteoblasts in standard culture media.) In a bone, the osteoblastic cells' NMDA receptor/channels would only be stimulated by glutamate boluses from glutamatergic nerves responding to strain or to glutamate released by damaged and dying osteocytes in microcracks.

Despite this, the importance of glutamate as an osteogenic osteotransmitter is still uncertain, because although Dobson and Skerry (2000) and Taylor et al. (2000) have found that inhibiting the glutamate receptors reduce bone formation in vivo and in vitro, Gray et al. (2001) have reported that high doses of NMDA receptor/channel inhibitors did not affect bone formation by cultured rat osteoblasts and bone formation in GLAST transporter knock-out mice was normal. But Skerry et al. (2001) have listed the serious shortcomings of Gray et al.'s GLAST knockout mice which I will not repeat here. Osteoblasts also have AMPA/kainate receptors, which also activate glutamate receptor/channels (Skerry and Genever, 2001), and thus might take over from blocked NMDA receptor/channels, and we do not know whether GLAST activity and glutamate reuptake are as critical for bone making as they are for neurotransmission. One of the most serious errors in Gray et al.'s experiments is their use of competitive NMDA receptor/channel antagonists. These antagonists simply could not compete with the high levels of glutamate in a culture medium that are needed to support the growth of osteoblastic cells (A.F. Taylor, 2002).

The upshot of all of this is that osteoblast differentiation and osteocyte functions, such as sensing and memorizing strain pulse frequency, involves glutamate, and to do their job these cells use exactly the same signaling machinery and transmitters as central neurons (Skerry and Taylor, 2001; Spencer et al., 2004; Turner et al., 2002). In other words, it looks as if we have been looking at "osteoneurons"—what a fantastic change in our view of these cells and the smart bones they make!!

These unexpected neural similarities are not restricted to glutamate receptors and transporters! Active osteoblasts also make BDNF (brain-derived neurotrophic factor) and the TrkB receptors needed to bind it (Yamashiro et al., 2001). Of course, you might say that this BDNF is simply used to stimulate the innervation of new bone. However, the osteoblasts' TrkB receptors mean that they too use their BDNF to stimulate themselves and their neighbors for some aspect of bone making.

Before the osteoblasts arrive on the scene to start filling in the trench or tunnel, the lining cells must sweep up the litter left by the untidy osteoclasts on the resorption cavity floor (Everts et al., 2002). There are collagen "bristles" sticking out of the cavity floor. The lining cells then move onto the surface to give it a good shave and make a smooth surface upon which they slap a layer of osteopontin to glue the new collagen that will be made by the incoming osteoblasts (Everts et al., 2002; McKee and Nanci, 1996). As we shall see below, the residual phosphate from the dissolved bone mineral in the resorption cavity could be the stimulator of this first burst of osteopontin expression (Beck, 2003; Beck and Knecht, 2003; Beck et al., 2000, 2003). These lining cell janitors are in fact reversibly retired osteoblasts who when stimulated by agents such as PTHs can revert to full osteoblasthood to give a first wave of osteoblasts to "kindle" bone building.

fig. 4

While there were as few as 10 osteoclasts doing the digging, there may be hundreds of osteoblasts working in a tunnel or trench (R.B. Martin et al., 1998). Despite their numbers, the osteoblasts take about 4-8 times longer to fill the cortical tunnels and trabecular "blister"-covered trenches than the osteoclasts took to dig them. Therefore, at any moment in the microfracture sites, a mature bone has holes that are being dug or have been recently dug by the microcrack-repairing BMUs—these holes together make up the remodeling space. As the osteoblasts are slowly filling these holes with new factor-loaded matrix (‘osteoid’), some of them will be trapped in it like insects in amber, as it is gradually mineralized with apatite-like Ca-phosphate. When the large amount of Ca²⁺ in the excavation site has been used by the osteoblasts to mineralize the new matrix and the filling has been finished at the endocortical and trabecular worksites, the blister canopy collapses, and the old lining cells, together with new ones from a lucky few osteoblast survivors, connect to the entrapped osteocytes and cover the repaired patch (Hauge et al., 2001). Walled up in their matrix-lined cells, they reduce their PTHR1 receptor density (Aubin, 1998, 2000, 2001; Aubin and Triffitt, 2002) and become osteocytes that connect to the osteointernet and start sending messages from their bending cilia and tugged and twisted integrins to the overlying lining cells and from them to adjacent marrow stromal cells and Haversian blood vessels to again summon BMUs to repair damage when needed or mobilize Ca²⁺ if needed to restore the circulating level (Marotti, 1996; Martin, 2000a, 2000b, 2002; Hauge et al., 2001; Skerry, 1999; Whitfield et al., 1998b; Yellowley et al., 2000).

When the new patch is finally in place 3-9 months later and the osteocytes start sending the signals needed to prevent lining cells from unnecessarily summoning osteoclast precursors from the blood vessels running through the cortical Haversian canals or from forming a BMU-activating blister on endosteal or trabecular surfaces, the members of the last osteoblast crew are now out of work—they have become redundant! Those that have failed to find a free space on which to settle as lining cells or osteocytes to keep their surface-sensing integrins emitting their survival-promoting signals trigger apoptosis and self-destruct like the unemployed osteoclasts before them (Frisch and Screaton, 2001; Stupack and Cheresh, 2002; Whitfield et al., 1998b, 2000a). However, the doomed, self-destructing osteoblasts make one final contribution to bone formation, specifically mineralization, by dumping alkaline phosphatase into the new matrix in vesicles released from their blebbing surfaces (Farley and Stilt-Coffing, 2001). Some of this alkaline phosphatase will get into the blood to tell the outside World of bone being made and osteoblasts dying.

Most of the self-destructing (apoptosing) cells in bone are located in microcrack sites being repaired by BMUs. The large amount of inorganic phosphate (P_i), released along with Ca²⁺ from the apatite crystals dissolved by the osteoclasts (Gupta et al., 1996), may increase the chance of osteoblasts triggering apoptosis, unless they are attached and protected by survival factors such as Bcl-2 (Allen et al., 2002; Meleti et al., 2000). The reason for this is likely to be the Na⁺ gradient-powered pumping of P_i into the osteoblasts by their Pit 1 and 2 type III Na⁺/Pi transporters, particularly the Pit 1 transporter that they selectively upregulate during maturation for matrix mineralization (Adams et al., 2001; Meleti et al., 2000; Nielsen et al., 2001; Takeda et al., 1999). When the phosphate reaches a critical level, it would be carried into the mitochondria by the P_i^-/OH^- exchanger and trigger the assembly of the huge mitochondrial permeability transition pore and through this the leakage of cytochrome c, AIF (apoptosis-inducing factor), procaspase-9, and Diablo into the cytoplasm from the intermembrane space. The appropriately named and deadly Diablo neutralizes a group of caspase inhibitors, and cytochrome c forms a complex with APAF-1, which causes caspase-9 to auto-activate and trigger a lethal cascade of so-called "executioner" caspase protein shredders (Crompton, 2000; Finkel, 2001). (This ability of P_i to greatly promote apoptogenesis was first noticed by me nearly 40 years ago, when I was searching for a way to enhance the radiation-induced apoptosis of rat thymic lymphocytes and human peripheral blood lymphocytes for an ultra-sensitive radiation bio-dosimeter (e.g., Whitfield et al., 1967). Meleti et al. (2000) have shown that inhibiting the transporter with low concentrations of phosphonoformic acid (PFA) prevents P_i from killing primary human osteoblasts, and Mansfield et al. (2001) have shown that PFA also prevents P_i from triggering apoptosis in chondrocytes. It follows from this that inhibiting the Na⁺/P_i transporter might prolong the lifetime of osteoblasts by shielding them from the high P_i in their workplace. However, as we shall see later on, this could be dangerously counterproductive, because the transporter is also a key player in osteoid mineralization.

Figure 8

The relation between proliferative activity and PTHR1 (more...)

Figure 8

.

The relation between proliferative activity and PTHR1 and PTHrP expressions at the various stages of osteoblastic differentiation, using the information provided by Aubin (2000, 2001) and Aubin and Triffitt (2002) as the foundation of the scheme. The appearance and disappearance of the various other things as the differentiation progresses are discussed in the text.

fig. 8

Finally, osteoblasts are recruited in load-bearing mature human bones only for remodeling BMUs to repair microcracks. But the osteoblast recruitment and osteoclast activity are not coupled in growing bone. Clusters of osteoblasts operate independently from osteoclasts in the growing, so-called modeling, bones of rat pups and human children (Frost, 1997; R.B. Martin et al., 1998; Selye, 1932). And as we shall see below, PTHs can stimulate a massive layering of osteoblasts on trabeculae and bone formation without a prior activation of osteoclasts. Indeed, growing mutant mice, which cannot generate functional osteoclasts, become osteopetrotic due to unopposed bone building by osteoblasts; and mutant mice that cannot generate functional osteoblasts become the opposite, osteoporotic, due to the unopposed osteoclast activity (Karsenty, 1999; Whitfield et al., 1998b).

Dangerous Osteoblastic Transvestites

At this point, we should take a quick look at osteoblasts which don't live in bone, but instead appear in and ossify stressed and injured arteries and aortic valves and thus cause heart attacks and death of atherosclerotics and persons with end-stage kidney disease as well as cause the vascular impairment and consequent amputation of diabetics' limbs (reviewed by T.M. Doherty et al., 2004; Whitfield, 2005). They come from a subpopulation of redifferentiation-competent vascular smooth muscle cells known as "calcifying vascular smooth muscle cells" or CVCs (T.M. Doherty et al., 2004; Whitfield, 2005). They lay down sheets of bone and deposit bone nodules in the walls of affected vessels—they literally make stiff armored blood vessels (T.M. Doherty et al., 2004; Whitfield, 2005).

Factors from LDL particle-stimulated plaque/atheroma endothelial cells and macrophage invaders diapedesing through the excited endothelium into the vascular intima, high concentrations of circulating inorganic phosphate, and the intranuclear action of the cells' hyperexpressed PTHrP stimulate the proliferation and migration of vascular smooth muscle cells from the vascular media into the intima and the redifferentiation of the CVCs among the migrating cells into functioning osteoblasts (T.M. Doherty et al., 2004; Jono et al., 2000; Steitz et al., 2001; Whitfield, 2005). Thus, the CVCs lose their specific smooth muscle lineage markers and α-actin and then go about making bone just like normal osteoblasts (T.M. Doherty et al., 2004; Stein et al., 1996). They express the master osteoblast transcription factor Cbfa1/Runx-2, make type I collagen, alkaline phosphatase, and then ostocalcin and osteopontin when mineralizing the matured matrix (Jono et al., 2000; Steitz et al., 2001). A very strong support for what follows further on in our discussion of PTH's cyclic AMP (cyclic adenosine-3',5'-monophosphate)-triggered bone-building action is the ability of a short exposure to cyclic AMP to cause CVCs to stop proliferating and assume a classical cuboidal osteoblast shape and functions (Tintutt et al., 1998).

Leptin, Fat, Brains, and Bones

Figure 10

The overall restraining of bone growth from the brain (more...)

Figure 10

.

The overall restraining of bone growth from the brain by leptin and neuropeptide Y (NPY) via local, hypothalamic nuclei-dependent β-adrenergic agents and the direct positive, brain-independent control of bone growth by leptin and its receptors. Leptin from fat cells, osteoblasts, and even brain cells, or leptin that is injected intracerebrovascularly, stimulates receptor-bearing neurons of the VMH, ventromedial hypothalamic nuclei. These neurons project to the preganglionic sympathetic neurons of the lateral cell columns of the spinal cord, which in turn project to the local β-adrenergic nerves in the bones which fire their β-adrenergic transmitters at osteoblasts which have β2 AR, β2-adrenergic receptors on their surfaces. Details of these two control mechanisms may be found in the text. The broken lines mean inhibitory, and the solid lines mean stimulation. LepRb, the large completely functional leptin receptor; Osteoprogens, osteoprogenitors; Y2R, neuropeptide Y (NPY) receptor-2.

fig. 10

What Does Leptin Do to Bone?

It isn't the arcuate nucleus through which leptin operates on bone from the brain. Karsenty's group have recently found the boney consequences of leptin stimulating the neurons of the ventral medial hypothalamic nuclei (VMHN), which projects to the sympathetic nervous system (S. Takeda et al., 2003). The hypothalamus is really the head nucleus of the autonomic nervous system, and leptin keeps the β-adrenergic nervous system functioning by stimulating VMHN neurons, which project to the preganglionic sympathetic neurons of the lateral cell columns of the spinal cord. And directly tweaking the VMHN neurons by squirting leptin into the cerebral ventricles of a mouse affects the beast's bones. But I am getting ahead of myself.

Of course bones are affected by leptin's stimulation of GnRH (gonadotropin-releasing hormone) secretion and the estrogen-dependence of the Lep gene (originally called the Ob gene for the obese mouse in which it was discovered) gene expression (Ahima and Flier, 2000; Brann et al., 1999; Chu et al., 1999; J. Friedman, 2000; Himms-Hagen, 1999). Therefore, when either the Lep (Ob) gene or the LepRb receptor (Sweeny, 2002) is disabled, the animal will have reduced sympathetic activity and thus be hypotensive (Mark et al., 1999). And it will also make too much cortisol and too little estrogen and therefore should be osteopenic. This should be very bad for bones!

But instead of being severely osteopenic as expected, Ob(Lep)^-/- obese mice that cannot make leptin and Db^-/- mice with disabled LepR receptors have a high vertebral bone mass (HBM) (Ducy et al., 2000a, 2000b; Karsenty, 2000b; S. Takeda et al., 2003). Since the lack of estrogen increases osteoclast production, they do have more osteoclasts than normal, lean animals but only a normal number of hyperactive osteoblasts that override the diggers by making twice as much bone matrix as the osteoblasts in lean mice (Ducy et al., 2000a, b; Karsenty, 2000b).

This supernormal bone production, despite the soaring osteoclast population, is due to a hypothalamic response to the lack of leptin production throughout the body in Ob(Lep)^-/- obese mice or the incompetence of leptin receptors in Db^-/- obese mice, and it can be stopped in Ob(Lep)^-/- mice (which have functional leptin receptors) by injecting leptin into the animals' cerebral ventricles (Ducy et al., 2000a, 2000b; Karsenty, 2000b). Moreover, bone loss can be caused by intracerebroventricular leptin injection into normally boned, lean mice—leptin seems to cause the release from the brain of an osteoblast suppressor. Perhaps the osteoblast hyperactivity is due to the dramatic surge of NPY gene expression in the leptin-lacking Ob(Lep)^-/- mouse's hypothalamus (Wilding et al., 1993). But no—cerebrally injected NPY does the same thing as leptin—it causes bone loss (Ducy et al., 2000; Karsenty, 2000)!

Thus, it seems that in mice, a set-point level of leptin might stimulate the neurons in some hypothalamic nucleus to produce a hypothalamic osteoblast inhibitory factor, a HOBIF, that somehow holds osteoblast activity at some optimal level (Ducy et al., 2000a,b; Karsenty, 2000b; Whitfield, 2002b; Whitfield et al., 2002) (fig. 10). It appeared at one point that the nNOS (neural nitric oxide synthase) might be part of this brain-based leptin signaling mechanism that dampens osteoblast activity, because knocking out the nNOS gene causes large increases in bone mass (van't Hof et al., 2002b). But this is not the HOBIF. At any rate, without signals from leptin-activated receptors on hypothalamic neurons, there is no osteoblast restraining HOBIF production, and bone growth climbs along with the food consumption and body weight (fig. 10).

But is all of this just much ado about nothing?! The bone growth in a fat mouse could simply be driven by the escalating loading by the massive increase in body weight and the consequent operation of the body's "mechanostat" to make more bone to reduce the strain as we have learned in the last chapters. But the bone mass starts rising before the weight rises. Furthermore, it also rises in A-ZIP/F-1 mice that do not have the leptin-making, white-fat adipocytes, without which the animals cannot gain weight (Ducy et al., 2000a; Karsenty, 2000b). But readers should beware of this model—all of the other sources of leptin must still be functioning in the A-ZIP/F-1 mice, and as we shall see further on, this bone growth may not be due to a lack of leptin and HOBIF.

The Karsenty group (S. Takeda et al., 2002,2003) now knows what HOBIF is. First, it is not a circulating agent, because when the circulations of two high-bone-mass Db^-/- mice are connected (parabiotic pairing) and one receives intracerebral ventricle leptin, only the injected mouse loses bone—the factor does not spread via the circulation (S. Takeda et al., 2002,2003). Second, selectively destroying a normal mouse's ventromedial hypothalamic nuclei (VMH) with gold-thioglucose, but not destroying the arcuate nuclei with monosodium glutamate, drives the bone mass up to the high level in Db^-/- mice and prevents the bone mass from dropping after intracerebroventricular leptin injection (S. Takeda et al., 2002,2003). Third, knocking out dopamine β-hydroxylase, and with it the ability of β-adrenergic nerves in the bones to make epinephrine and norepinephrine, does not increase body weight, but it does increase bone mass and blunts the bone-reducing response to intracerebroventricular leptin (S. Takeda et al., 2002,2003). Therefore, HOBIF is a noncirculating, i.e., within-bone, product of dopamine β-hydroxylase in the bones' nerves.

In mice, and probably humans, leptin operates on bone independently from the appetite and body weight controlling hypothalamic neurons of the arcuate nuclei (which, as noted above, can be destroyed without affecting bone mass) by stimulating the sympathetic nervous system via the LepRb receptor-bearing cells of the ventromedial hypothalamic nucleus (Funahashi et al., 2003; S. Takeda et al., 2002,2003), which increase catecholamine secretion by peripheral nerves (S. Takeda et al., 2002,2003). It appears that it is the VMHN-dependent, nonmyelinated sympathetic nerves in the bone that terminate near the trabeculae and follow the osteoblasts filling osteonal tunnels and locally bombard these osteoblasts with β-adrenergic agonists. And osteoblasts are certainly well-armed with β1- and/or β2-adrenoreceptors (Kellenberger et al., 1998; Majeska et al., 1992; Moore et al., 1993; Spencer et al., 2004; Takeuchi et al., 2000; Togari et al., 1997). For example, in my experience the β-adrenergic agonist isoproterenol very strongly stimulates adenylyl cyclase activity in ROS 17/2 rat osteoblasts. Activation of β-adrenoreceptors increases the expression of the pro-osteoclastogenic RANKL in MC3T3-E1 preosteoblasts and mouse bone marrow cells, and chemical sympathectomy inhibits preosteoclast differentiation and impairs bone resorption in adult rats (Cherruau et al., 1999; Spencer et al., 2004; Takeuchi et al., 2000). Also, we know that β-adrenergic stimulators, such as clenbuterol and salbutamol, decrease bone mass in leptin-deficient mice and female Wistar rats by reducing bone cell proliferation; that a β-blocker such as propranolol increases bone mass in intact and ovariectomized wild-type mice; and that taking β-adrenergic blockers reduces bone fracturing in women (Bonnet et al., 2003; Pasco et al., 2004; S. Takeda et al., 2002,2003).

To summarize—injecting leptin into the brain of a mouse stimulates the sympathetic nervous sytem and the adrenergic nerves in the bones via the lepRb-receptor-bearing cells of the ventromedial hypothalamic nuclei, and this raises the level of β-adrenergic transmitters in the bone that slow the proliferation of osteoprogenitor cells (certainly not the terminally post-proliferative osteoblasts as claimed by S. Takeda et al. (2002)) (fig. 10). But there is something inconsistent with their earlier paper. According to Ducy et al. (2000a), the high bone mass in the Ob(Lep)^-/- is not due to there being more osteoblasts making bone as claimed by S. Takeda et al. but to the same number of osteoblasts making more bone.

On the basis of leptin's indirect, hypothalamically mediated, osteoblast-suppressing action, Karsenty (2000b) has called it the osteoblast-suppressing equivalent of the osteoclast-suppressing OPG. But it isn't (Gordeladze et al., 2001; Reseland and Gordeladze, 2002; Gordeladze et al.,2002a, 2002b; Whitfield, 2001, 2002)!! While the primary adult mouse osteoblasts in the experiments of Ducy et al. (2000a) apparently did not express the fully signaling Lep receptor—LepRb (Funahashi et al., 2003; Sweeny, 2002; Zabeau et al., 2003)—and thus could not respond directly to leptin, others have found that mouse bone cells can directly and positively respond to the cytokine both in culture and, most importantly, in the mouse. MC3T3-E1 mouse preosteoblasts and MCC-5 mouse chondrocytes both have LepRb receptors, make leptin, and dump it out into the culture medium (Kume et al., 2002). Liu et al. (1997) have reported that noncerebrally injected leptin stimulates endocortical bone formation in obese mice. Steppan et al. (1998, 2000) have also reported the results of experiments showing that intraperitoneally injected, instead of intracerebrally injected, leptin directly and potently stimulates bone growth in Ob(Lep)^-/- mice. And could a direct action of the leptin be responsible for the high bone mass in A-ZIP/F-1 mice, which do not have white fat cells, but must have all of the other leptin makers?

As could have been expected from these other reports of leptin's bone-stimulating action in mice, it is now becoming clear that the Karsenty story is misleading because of an inadequate examination of their Ob(Lep)^-/- mice and a failure to pay attention to the earlier reports of Steppan et al. (1998, 2000). But Hamrick et al. (2004) have now properly looked at Ob(Lep)^-/- mice. It turns out that Steppan et al. (1998, 2000) were right for the appendicular skeleton! Leptin lack does indeed shorten femurs and reduce BMD, bone mineral content, cortical thickness, trabecular thickness, and trabecular volume in the femur. The expression of TGF-β1, an osteoprogenitor stimulator and strong anti-adipogenesis agent, is also reduced, which is consequently associated with a massive increase in femoral adipocytes and a drop in the osteocyte population density. However, the Karsenty group focused on an increased lumbar vertebral bone mass in their Ob(Lep)^-/- mice. And according to Hamrick et al. (2004), they were also right! Hamrick et al. found increased bone mineral content, BMD, and trabecular bone volume in the lumbar vertebrae of their Ob(Lep)^-/- mice. The Ob(Lep)^-/- lumber vertebrae were also longer and broader than in normal lean mice. This suggested to Hamrick et al. that the chondrocytes of the vertebral growth plates respond differently from the femoral growth plate—maybe they are more sensitive to, or more controlled by, leptin/VMNH-dependent β-adrenergic activity. The loss of long bone (appendicular skeleton) mass is likely due in part to severely reduced loading because of a significant drop in muscle (e.g., quadriceps) mass as well as the basic inactivity of such obese beasts. As discussed in detail in Chapter 1, this would stress osteocytes and ultimately trigger their self-destruction and osteoclast scavenging of the cadavers by restricting the transcanalicular provision of nutrients and clearance of waste. And of course, another contributor would be the lack of leptin's direct anabolic action discovered in mice by Liu et al. (1997) and Steppan et al. (1998, 2000). Therefore, all that remains of the Karsenty mouse story is the ability of an intracerebroventricular shot of leptin to stimulate the VMHN-neurons and through them the preganglionic sympathetic neurons of the lateral cell columns of the spinal cord and the bones to which they project.

Cornish et al. (2001) have reported that leptin directly stimulates fetal rat osteoblasts as strongly as IGF-I and increases bone strength in mature male mice by 20% or more. And Picheret et al. (2003) have shown that obese female and male Zuker fa/fa rats with their disabled LepR receptors that cannot respond to the large amounts of leptin pouring out of their fat cells have significantly lower femoral bone density and osteoblast activity (as indicated by a significantly lower blood osteocalcin concentration) than lean control rats. This is supported by Tamasi et al. (2003) who have reported that in these rats tibial trabecular bone volume, trabecular number, and trabecular thickness are also lower than in normal Zuker rats, although the formation rates in the normal and Zuker fa/fa rats were not different. Moreover, Maor et al. (2002) have reported that leptin stimulates endochondral bone formation in cultured murine mandibular condyle and humeral growth plate by causing the cells to express and secrete IGF-I which, as we shall see further on, is also stimulated by PTHs.

Leptin is also one of the drivers of endochondral bone formation in the mouse. AndDumond et al.(2003) have found that leptin stimulates rat chondrocyes to grow and make cartilage by making IGF-I and TGF-β1. Large amounts of the cytokine are made specifically by hypertrophic chondrocytes (see Chapter 5v to find out what these cells do) near invading capillaries and osteoblasts in the primary spongiosa extending from the growth plate of long bones (Kerr, 1999). It seems that the chondrocytes use leptin to get the blood vessel cells moving and proliferating and making matrix metalloproteinases-2 and —9 which they use as drills for tunneling through the collagenous matrix to deliver blood-borne osteoprogenitor cells and other essentials to the construction sites (Kume et al., 2002).

Freshly isolated human osteoblasts (e.g., from trabecular bone in the head of the femur or the iliac crest), normal rat osteoblasts, and ROS 17/2.8 rat osteoblastic osteosarcoma (bone cancer) cells express the all-important big, signaling LepRb receptors (Bassilana et al., 2000; Enjuanes et al., 2002; Evans et al., 2001; Gordeladze et al., 2001; Y.-J. Lee et al., 2002; Reseland and Gordeladze, 2002a, 2002b; Reseland, et al., 2002; Steppan et al., 1999, 2000). Signals from these receptors obviously must stimulate bone growth because Thomas et al. (2001) have reported that intraperitoneally infusing human leptin prevents disuse ("pseudo-microgravity")-induced bone loss in tail-suspended rats. And Burguerera et al. (2001) and Gordeladze et al. (2002a, 2002b) have shown that leptin promotes rat bone growth by suppressing the osteoclast-promoting RANKL expression and stimulating osteoclast-suppressing OPG expression in osteoblastic stromal cells (fig. 10). Iwaniec et al. (1998) have reported that low concentrations (1-100 ng/ml of medium) of leptin increase bone nodule formation in human bone cultures. Leptin also makes human marrow stromal cells differentiate into osteoblasts instead of adipocytes (Thomas et al., 1999), and it stimulates the proliferation of cultured (i.e., probably incompletely matured simply because normal mature osteoblasts cannot proliferate) human osteoblastic cells and causes human marrow stromal osteoprogenitor cells to express alkaline phosphatase, collagen I, osteocalcin and to mineralize matrix (Gordeladze et al., 2002a, 2002b; Evans et al., 2001; Thomas et al., 1999). Leptin also enhances the expression of the genes for the major osteoblast differentiation driver, the master Cbfa1/Runx-2 gene transactivator, as well as alkaline phosphatase and osteocalcin, and it protects the human osteoblasts from self destructing by suppressing the expression of the proapoptotic Bax protein and stimulating the expression of anti-apoptotic Bcl-2 protein (Gordeladze et al., 2002a, 2002b). Leptin also facilitates the transition of cultured osteoblasts into preosteocytes (Gordeladze et al., 2002). Leptin's ability to facilitate the osteoblast→osteocyte transition and stimulate matrix mineralization by human osteoprogenitors fits in nicely with the fact that that the LepRb receptor-bearing human osteoblasts start making and secreting autocrine (self-stimulating)/paracrine (neighbor-stimulatimg) leptin when they are either in the mineralization or early osteocyte stage (Reseland et al., 2001) (fig. 10).

Most important for our story is the very exciting possibility that signals from type 1 PTH/PTHrP receptors (the so-called PTHR1s we will soon be getting to know rather intimately) stimulate the expression and secretion of leptin which, from the accumulating evidence, could at least partly mediate PTH's stimulation of bone growth (fig. 10). This has been suggested by Torday et al.'s (2002) report that PTHrP stimulates the leptin gene, leptin production, and leptin secretion in lung lipofibroblasts and by Gordeladze et al.'s (2002a) report that the effects of leptin on human osteoblasts mimic those of hPTH-(1-84).

The amazing story emerging from all of this goes something like this. Leptin is expressed first in mesenchymal stem cells to steer them onto the osteoblastic pathway, but this stops in proliferating osteoprogenitors. Then it starts again, after proliferation stops, and PTHrP and PTHR1 receptors appear and osteoblasts have matured, in order to help drive matrix deposition and mineralization. The leptin from the osteoblasts stimulates more osteoprogenitors to advance along the walls of the osteoclast tunnels and trenches and stimulates the activity and increase the longevity of their osteoblast friends and neighbors.

All of this means that human osteoblasts, like the cells of blood vessels, muscles, and many other organs throughout the body, are controlled, in the particular case of the osteoblasts negatively controlled, by the body-wide leptin/hypothalamic VMH-promoted β-adrenergic nerve activity. But they are also directly and positively controlled by leptin itself from the white fat adipocytes, local yellow marrow adipocytes, and most importantly from the late-stage osteoblasts themselves (Laharrague et al., 1998; Reseland et al., 2001, 2002; S. Takeda et al., 2003; Whitfield, 2002b) (fig. 10). Therefore, the question arises whether there is an adrenergic inhibition of bone formation or a strong direct stimulation by leptin of osteoblast activity in people with common (as opposed to the much rarer genetic (e.g., Ob(Lep)^-/-) obesity who make a lot of the peptide that can't slam the brakes on eating but can still stimulate VMH-driven β-adrenergic activity and cause obesity-dependent hypertension (Rahmouni and Haynes, 2002). The reader should stay tuned in order not to miss the exciting things coming in the Leptin World, especially if an osteogenic leptin analog can be made that cannot pass the blood-brain barrier and trigger the negative VMH-driven β-adrenergic action.

But how does all of this interesting stuff fit into a postmenopausal story? The postmenopausal estrogen drop might turn down both of the leptin mechanisms (Brann et al., 1999; Chu et al., 1999), which would cause a rise in weight and help promote the osteoclast population explosion caused by the lack of estrogen. Thus, for example, in the rat model ovariectomy silences Ob(Lep) gene expression, and leptin consequently vanishes from their circulation (Chu et al., 1999). And the animals get hungry and heavier, while their trabecular bone is being chewed up by osteoclasts and their marrow is getting fattier (Whitfield et al., 1995). The leptin drop and the resulting surge of NPY expression in the hypothalamus cause the hunger and weight gain. At first sight, this seems to conflict with the amount of circulating leptin expression being positively correlated to the body fat load—leptin should rise. But Ob(Lep) gene expression is estrogen-dependent, and giving estrogen to the OVXed rats prevents the leptin drop, overeating, and weight gain (Chu et al., 1999). Therefore, without enough estrogen to maintain the Ob(Lep) gene expression, leptin virtually vanishes from the circulation by 7 weeks after OVX, but the increasing accumulation of white fat in its swelling adipocytes eventually overrides the estrogen lack, leptin production rebounds, and its level soars above the sham OVX level between 9 and 13 weeks after the operation (Chu et al., 1999). Therefore, the bones of an overeating OVXed rat do not at first have to face the debilitating double whammy of leptin-induced, VMH-driven β-adrenergic-suppressed osteoblast activity as well as an osteoclast feeding frenzy, but the direct stimulation of osteoblasts by the eventual leptin surge should slow or stop the bone loss. Alternatively, any positive action of the leptin surge might be counterbalanced by an adrenergic surge. It follows from this that it would be important to find out whether a leptin loss might also affect bone loss in postmenopausal women and whether leptin injection would stimulate bone growth in these women as it does in rats. Indeeed, Ke et al. (2001, 2002) have anticipated leptin's admission into the exclusive Bone Growers Club by patenting it and its analogs for treating osteoporosis and all other conditions causing bone loss and for accelerating fracture healing.

But the hypothalamis has a lot of jobs to do, and there seems to be another player in the brain-and-bones saga—NPY's Y2 receptor. Why do we think so? Well, if the Y2 receptor gene is selectively knocked out in mouse hypothalamic neurons, osteoblasts make more matrix just as they do in Ob(Lep)^-/- mice (Baldock et al., 2002; Herzog, 2002)—signals from the Y2 receptors in hypothalamic neurons somehow lead downstream to the reduction of osteoblast activity. And of course, this why Ducy et al. (2000a) found that directly injecting NPY into the cerebral ventricles caused bones to lose mass in their Ob(Lep)^-/- mice, just as did leptin injection. Thus, when the leptin level is "normal", NPY secretion and Y2 receptor signaling are minimal or suppressed. But any possible upsurge of osteoblast activity caused by shutting down NPY-Y2 receptor activity is prevented by leptin-driven VMH-mediated β-adrenergic secretion. However, the direct osteogenic action of leptin would still be working in the bones.

But things are complicated by the fact that NPY also directly targets osteoblasts. Indeed, osteoblasts have receptors for the NPY released into the bone by the same sympathetic nerves that are driven by leptin via the VMHN neurons (Bjurholm, 1991; Ekblad et al., 1984; Togari et al., 1997). But we don't yet know how NPY receptors signals directly affect osteoblast activity. All we know at the moment is that NPY inhibits the abilities of noradrenaline and PTH to stimulate adenylyl cyclase in osteoblastic cells (Bjurholm, 1991). However, it is clear that there is a complex control system operating at several levels running from the brain and spinal cord to the bones and osteoblast surfaces to control bone formation.

But before moving on to other things, it must be clearly understood that as matters stand in late 2004, leptin is a direct bone anabolic agent which may indirectly, but only under severely nonphysiological conditions, limit bone growth by increasing local bone β-adrenergic activity via the VMHN neurons. Thus, it would seem to be unreasonable to try to treat osteoporotic patients without cardiovascular disorders with potent β-blockers as suggested by S. Takeda et al. (2003). But Pasco et al. (2004) have presented data indicating that women with cardiovascular disease who have been taking such drugs do have higher bone mineral densities and fewer fractures.

How Does Estrogen Loss Cause an Osteoclast Population Explosion?

The ability of osteocytes to know when the strain exceeds a site-appropriate level and to react to this by signaling through the osteointernet the need to make the various factors such as PGE₂, PGI₂, IGFs, and TGF-βs as well as our new friend leptin to mobilze osteoblasts to make more bone to bring the strain to a subthreshold level depends on estrogen (Ehrlich and Lanyon, 2002). But as the estrogen level starts falling, estrogen receptor-α (ERα) signals start fading, as must leptin on the approach to menopause, the osteocytes become less responsive to strain, i.e., they become less able to feel the sloshing of the canalicular fluid. This results in fewer calls for resources to maintian bone and increasing uncompensated strain from muscle pulling and limb impacts with a preferential loss of minimally strained bone. But along with the osteocytes becoming "strain-deaf", the estrogen drop also causes an upsurge of osteoclast generation.

However, the vigilant reader will challenge this blanket assertion that the signals from estrogen receptors fade as the estrogen level drops because strain by itself can activate estrogen receptors without estrogen (Aguirre et al., 2003). Thus, I must now qualify this by saying simply that estrogen seems to be replacable in some cells such as osteocytes by strain.

Osteoclast progenitors respond to estrogen through their conventional "genomic" estrogen receptors, which work on the chromosomes in the nucleus to activate target genes with so-called EREs (estrogen response elements) in their "signal boxes" as well as through what appear to be "nongenomic (genotropic)" receptors that are translated into protein from the same primary transcript as the genomic receptors and fire signals into the cell from little caves (caveolae) or pouches in the plasma membrane (Chambliss et al., 2000; Coleman and Smith, 2001; Falkenstein et al., 2000; Fionelli et al., 1996; Kelly and Levin, 2001; Kim et al., 1999;Krishnan et al., 2001; Levin, 2000; Manolagas et al., 2002; Morley et al., 1992; Razandi et al., 1999, 2002; Segars and Driggers, 2002; Toran-Allerand, 2000; C.S. Watson, 2003).

Figure 11

How estrogen and things made by T-lymphocytes control (more...)

Figure 11

.

How estrogen and things made by T-lymphocytes control osteoclast generation and bone turnover. The (-) symbols indicate that estrogen inhibits expression of the indicated factor, such as MCSF (macrophage colony-stimulating factor), and that lowering the estrogen level releases the expression and production of the factor. It is also shown that T-cells make RANKL and TNFα that stimulate osteoclast generation and FasL (Fas ligand) that causes ostoblastic cells to self destruct by switching on the apoptotic mechanism.

Figure 12

Summary of some of the many consequences of an estrogen (more...)

Figure 12

.

Summary of some of the many consequences of an estrogen crash on bone.

fig. 11

fig. 12

fig. 12

fig. 11

fig. 12

fig. 11

fig. 12

fig. 11

fig. 11

fig. 12

fig. 11

The ability of osteoblastic precursor cells to make RANKL falls as the rankl gene's promoter-switch box is locked shut by methylation of its CpG domains while the anti-osteoclast opg gene stays active as the progeny differentiate into mature bone-making osteoblasts—a wise stratagem to protect the new bone they are making (Atkins et al., 2003; Gori et al., 2000; Kitazawa et al., 1999; G.P. Thomas et al., 2001). To understand the control of osteoclast differentiation, we must know that the expressions of the genes for RANKL and OPG in the individual cultured cells are logically linked by a reciprocal control mechanism, which will be discussed in Chapter 5. Suffice it to know here that when one gene's expression is turned up, the other's expression is turned down (e.g., Ma et al., 2001; Yamagishi et al., 2001).

fig. 11

fig. 11

The TGF-β stores in the bone matrix are also depleted (reviewed by Jilka, 1998; Whitfield et al., 1998b). Therefore, there is even less of it that can evade the proteases and reduce osteoclast generation and kill mature osteoclasts when they dig it out of the matrix (Jilka, 1998). And there is also less of it to attract immature osteoblasts into the excavation sites, although TGF-β receptor signals may be less important for this than the migration-stimulating signals from the osteoblastic cells' Ca²⁺-sensing receptors stimulated by the very large amount of Ca²⁺ in the excavation sites (Brown and MacLeod, 2001; T. Yamaguchi et al., 1998).

fig. 12

Since estrogen also lengthens the working lives of osteoblasts and the microcrack-detecting osteocytes by suppressing their apoptotic self destruct mechanism, the estrogen loss unleashes this mechanism and reduces the number of osteocytes (Manolagas, 2000; Manolagas et al., 1999; Tomkinson et al., 1998). This loss of osteocytes would reduce the inhibitory signaling through the osteocyte-lining cell network, which would be equivalent to severing the signaling network by microcracks (R.B. Martin, 2000, 2002). This pseudo-microcrack signaling would increase BMU activation and therefore bone turnover and the remodeling space—functionally it would be equivalent to a widespread microcracking, like the shattering of an automobile windshield.

Perhaps counterintuitively, such BMU activation should, and in my experience does, increase the number of osteoblasts. However, the hole-filling efforts of these osteoblasts can't keep up with the hole-digging zeal of the large osteoclast crews!

Under these low-estrogen conditions, osteoblast working life and activities, such as laying down bone matrix and secreting various factors, may also be reduced, and the BMU replacement deficit thus increased, by the loss of signals from the nongenomic estrogen receptors such as those found on the surfaces of cultured female rat osteoblasts (Falkenstein et al., 2000; Lieberherr et al., 1993; Manolagas, 1999). These surface receptors appear either to be the products of the gene for the nucleus-seeking, gene-activating ERα genomic receptor that have been modified to stay at the cell surface and fire Ca²⁺ and cyclic AMP signals into the cell when activated by 17β-estradiol (Chambliss et al., 2000; Falkenstein et al., 2000; Fiorelli et al., 1998; Kelly and Levin, 2001; Lieberherr et al., 1993; Razandi et al., 1999; C.S. Watson, 2003; Whitfield et al., 1998b). Or they might be the entirely unrelated ER-X receptors (Toran-Allerand, 2000) or the "γ-adrenergic" receptors that are activated by dopamine, epinephrine, and norepinephrine as well as 17β-estradiol and various xenoestrogens (Benten et al., 2001; Nadal et al., 1998, 2000).

fig. 12

fig. 9

How to Stop (or at Least Slow) Menopausal Bone Loss

fig. 9

Obviously, bone loss can be stopped by giving the post-menopausal woman estrogens or one of the partially estrogen-mimicking SERMs (selective estrogen receptors modulators) such as raloxifene (Eli Lilly's Evista™) (Roe et al., 2000). But this does not cause a sustained increase in bone formation—it works by reducing remodeling. Giving one of these, raloxifene, to OVXed rats does not affect the expression of bone formation-related genes such as alkaline phosphatase, biglycan, collagen I, IGF- II, collagen-binding protein, and osteocalcin in their femurs, but it actually reduces the expression of decorin and IGF-I (Onyia et al., 2002). On the other hand, it reduces osteoclast generation by reducing the expression of CSF-1 (colony-stimulating factor-1).

Bone loss can also be reduced or even stopped with one of the many bisphosphonates such as alendronate (Merck & Co.'s very popular Fosamax™) (Fleisch, 1997, 1998), or either oral or nasally sprayed calcitonin peptide that shuts down osteoclast activity by activating receptors on the cells of the osteoclast lineage (Roodman, 1996; Suda et al., 1999).

The bisphosphonates are Ca-apatite-loving, bone-binding analogs of pyrophosphate (P-O-P) in which the oxygen has been replaced by a carbon atom, the so-called geminal carbon, which makes them nonmetabolizable. They preferentially attach to osteoclast excavation sites and attack the working mature osteoclast crews (Rogers, 2003). Then, upon the arrival of osteoblasts where the bone mineral is exposed, they are buried in the new bone. There they lie in wait, unchanged for months to years, until the poisonous bone patch is once again cracked and marked for removal by osteoclasts. The osteoclasts inadvertently release the deadly bisphosphonates and suicidally gobble them up along with other debris they want to expel from the hole by carrying them from their apical acid-protease-dripping fingers poking into the holes to their upper basal surfaces by a process called "transcytosis". (The terms apical (upper end) and basal (bottom end) may be a bit confusing here because osteoclasts stick their apical fingers into the bone and their basal tails in the "air".) Because the bisphosphonates lock onto the exposed mineral surfaces of osteoclast-excavating microcrack repair sites, and since trabecular bone reputedly has the highest turnover rate (though see Parfitt (2002)), the bisphosphonates are best able to prevent resorption and remodeling and increase the strength of trabeculae-rich bones such as vertebrae and femoral trochanter, which have the largest numbers of attractive remodeling excavation sites at any given time (Rodan and Reszka, 2002).

Figure 13

The ways statins and N (nitrogen)-bisphosphonates (more...)

Figure 13

.

The ways statins and N (nitrogen)-bisphosphonates (N-BPs) disable and kill osteoclasts. See fig. 14 and the text for further details of how N-BPs and other bisphosphonates kill osteoclasts.

fig. 13

fig. 13

Let's summarize what we have learned so far. A mature osteoclast, the specific target of an N-BP, does its job with its cytoskeletal machinery. It lays down an actin ring—a bit like a micro-octopus's suction cup. Inside the ring it sets up its ruffled border loaded with molecular hoses for spraying the HCl and matrix-chewing enzymes for digging the hole. These cytoskeletal creations are driven and shaped by the various isoprenylated small GTP-ases. When a N-BP sits down on a bone patch and is swallowed (by fluid-phase endocytosis) by the osteoclast, the whole cytoskeletal machinery grinds to a stop as the cell runs out of its supply of isoprenylated GTPases. So the first effect of a N-BP is to seek out and disable mature osteoclasts wherever they are digging. But something even more sinister follows the shutdown.

Figure 14

The mechanisms by which N-bisphosphonates drive osteoclasts (more...)

Figure 14

.

The mechanisms by which N-bisphosphonates drive osteoclasts to self destruction by causing them to switch on their apoptosis mechanism.

fig. 14

fig. 14

But alendronate (Fosamax™) unfortunately does not leave osteoblasts alone, at least in rats. Far from it—the orally delivered bisphosphonate actually turns down the expression of genes for bone formation in the proximal femoral metaphysis of ovariectomized rats (Onyia et al., 2002)! The inhibited genes include the ones for alkaline phosphatase, biglycan, collagen I, decorin, IGFs-I and II, collagen-binding protein, and osteocalcin. This strong suppression of osteoblast genes explains the strong suppression of bone formation activity by alendronate in rats (Onyia et al., 2002). And this happens not only in rats. Black et al. (2003), Finkelstein et al. (2002), and Neer et al. (2002) have found that alendronate also inhibits alkaline phosphatase and prevents hPTH-(1-34)(Forteo™) and hPTH-(1-84) from stimulating it and bone growth in hip and vertebrae in osteoporotic men and postmenopausal women.

But as I said at the start of this section, alendronate does appear to do something more positive to osteoblasts and osteocytes. It can protect osteoblasts and osteocytes from apoptotic self destruction. The cytoplasms of these cells are normally directly interconnected in a functional syncytium by gap junctions which are pipes made of end-to-end joined halves ("half-pipes") or connexons, each of which is a bundle of six, 4-pass (tetraspanning) transmembrane proteins, such as Cx (connexin) 43, through which passes a 1.5-μm channel through which ions (except Ca²⁺, which causes the channel to slam shut) and a wide variety of small molecules such as cyclic nucleotides and ATP can pass (Goodenough and Paul, 2003). But these cells also have uncoupled, or extra-junctional (EJ), "half- pipe" connexons that independently stick out of the cell and serve as nonspecific, large channels which slam shut when the external Ca²⁺ concentration rises above a critical level(Goodenough and Paul, 2003). It seems that alendronate operates by binding to and opening these "1/2 pipes", then passing through them into the cell and activating the cSrc protein-tyrosine kinase, which triggers an apoptosis-blocking MAP/ERK kinase cascade that closes the pipes by phosphorylating their Cx43 proteins (Goodenough and Paul, 2003).

Because these antiresorptives (or antiremodelers) suppress osteoclasts and may also reduce osteoblast action, they only cause, at the most, a limited amount of growth in mature human bones as the osteoblasts continue filling in the microcrack repair holes (the remodeling space), which account for 10% or more of the current total bone volume at any given moment (Dempster, 1995, 1997; Fleisch, 1997, 1998). But this is the limit—the gains in bone mineral density (BMD) in lumbar spine caused by optimal doses of alendronate, etidonrate, HRT, raloxifene, and residronate range only from 2.6 to 6.8% and from 1.0 to 4.8% in the femoral neck by 3 years (Pearson and Miller, 2002).

Because they stop osteoclast development and therefore micocrack-triggered remodeling, bisphosphates such as alendronate at least temporarily reduce fracture rates by prolonging the so-called secondary mineralization (i.e., by hardening the remaining bones and increasing the bone mineralization density level and therefore strength), but they do so without affecting trabecular thickness or number, i.e., without increasing the amount of bone tissue (Boivin and Meunier, 2002; Boivin, et al., 2000; T.J. Martin, 2004; Pearson and Miller, 2002). They may also improve hole-filling by making osteoblasts live and work longer (Allen et al., 2002; Manolagas, 2000) by preventing them from self-destructing by apoptosis by sucking up the large amounts of phosphate in the osteoblast excavations with their P_i transporters (Mansfield et al., 2001). Nevertheless, they can't raise bone mass above the normal level in rats or reduce the fracture risk in the mature human skeleton by more than 50% of the baseline risk (Rosen and Bilezekian, 2000; Seeman, 2001). However, a wrong impression of anabolic ability can be obtained by extrapolating the responses to antiresorptives of the continuously growing bones of mice and rats to the nongrowing mature remodeling human bones, where, because of osteoblast generation's tight coupling to osteoclast activity, the reduction of osteoclast generation ultimately reduces osteoblast generation. Thus, rodents with their always growing, low-microcracking bones treated with antiresorptives are like mutant mice, which cannot produce functional osteoclasts and become osteopetrotic as their contiuously and independently generated osteoblasts continue their unopposed bone making (Karsenty, 1999).

There is a potentially very serious problem with bisphosphonate therapy. Microcracking (i.e., microcrack density) increases exponentilally with age in bones such as the femur, probably because declining bone mass increases strain on the remaining bone and thus cracking (R.B. Martin et al., 1998; Schaffler et al., 1995). Therefore, murdering mature osteoclasts and preventing their replacement might be counterproductive because it will impede repair of the rising microcracks, which could with time lead to a resurgence of fracturing, the seriousness of which would be mitigated of course by the weakening muscle-imposed strain on the aging body by its shrinking muscles (Mashiba et al., 2001; R.B. Martin et al., 1998).

Anabolics—The Answers to the Osteoporotics' Prayers

fig. 9

Figure 15

A striking demonstration of the potent ability of (more...)

Figure 15

.

A striking demonstration of the potent ability of hPTH-(1-31)NH₂ (Ostabolin™ ) to strongly stimulate trabecular growth in a rat femur. A) Trabecular bone from the femur of an untreated OVXed rat. Typical trabeculae are seen in the ample normal marrow space. B) The greatly thickened trabeculae and almost totally bone-filled marrow space by the end of 6 weeks of single daily subcutaneous injections of 0.8 nmole of hPTH-(1-31) that were started 2 weeks after ovariectomy. These are scanning electron micrographs of demineralized bone obtained with a Philips SEM-505. The bar in A represents 100 μm.

fig. 15

Another property expected of an anbolic agent is a much greater stimulation of trabecular and endosteal growth than cortical growth because growth can only be started on bone surfaces, and the total trabecular surface in an adult human male is 9 m², while the total surface area of the Haversian and Volkmann's canals inside the cortex is only about 3.2 m² (Johnson, 1966; R.B. Martin et al., 1999). However, as I pointed out in Chapter 2i the greatest total surface area, 1200 m², is in the 3-D osteocyte lacunar-canalicular network, but it is obviously inaccessible—it is stuffed with cell bodies and processes (Johnson, 1966; R.B. Martin et al., 1999).

As we shall now see, the PTHs—the big native parathyroid hormone and certain of its small fragments and at least one of the fragments' lactamized (cyclic) analogs—are the first and currently the foremost bone anabolics (Whitfield et al., 1998b, 2002a, 2002b, 2002c).

The Ancient Origins of PTH and Its Receptor

The receptor for PTH and the PTH-like N-terminal 13 amino acids of PTHrP (PTH-related protein), which we will soon be meeting, belong to an ancient family that likely appeared hundreds of millions of years ago in the first animals in the Precambrian seas. The descendents of the ancient Mother Receptor are the modern receptors for calcitonin, corticotrophin-releasing hormone, gastric-inhibitory protein, glucagons, glucagons-like peptide-1, growth hormone-releasing hormone, an insect (the tobacco hornworm Manduca sexta) diuretic hormone, insect (Drosophila) calcitonin-like hormone and vasopressin-like hormone, pituitary adenylyl cyclase-activating protein, secretin, vasoactive intestinal peptide (Brody and Cravchik, 2000; Whitfield et al., 1998b; Wong, 2003). This family is an example of Nature using a basic design to do many different jobs. Thus, although these receptors have different amino acid sequences, they have similar higher order structures. Indeed, they are so alike that changing only one amino acid enables the secretin receptor to be activated by PTH and switching the equivalent amino acid in the PTH/PTHrP receptor to the secretin receptor's amino acid enables the receptor to be activated by secretin!

Hints of the antiquity of the PTH family and their receptors are the ability of mammalian PTH to increase the Ca²⁺ concentration in crustacean hemolymph by mobilizing the ion from the exoskeleton just as they do from our endoskeletons; the presence of a molecule in the neuroendocrine cells of the pond snail that cross-reacts with anti-mammalian PTH antibodies; and the ability of PTH to stimulate Ca²⁺ influx into snail neurons (Whitfield et al., 1998b). Is it possible that the Family was emerging in trilobites and Anomalocaris? Certainly our Cambrian ancestors, the notochord-bearing Myllokunmingia fengjiaoa and Pikaia (from Chengjian, China (Xian-Guang et al., 2004) and Alberta, Canada (Long, 1995)), respectively, must have been able to swim 500,000 years ago in the warm Ca²⁺-rich (ca. 10 mM) Panthalassic and Iapetus oceans, because their Precambrian ancestor(s) had found a way to avoid lethal hypercalcemia from Ca²⁺ flowing in through their gills, guts, and other portals. The solution was proto-calcitonin, which was made and secreted by the first ultimobranchial bodies that later became the thyroid parafollicular cells of the terrestrial animals and the installation of calcitonin—specific offspring of the ancient mother receptor on the surfaces of gill and proto-kidney cells (Whitfield et al., 1998b). A later invention, specifically by ray-fin fishes and our lobe-fin fish ancestors, was stanniocalcin which was designed to prevent Ca²⁺ from getting through the gills'chloride cells (Whitfield et al., 1998b).

But things changed radically during the Devonian period when the rapidly evolving lobe fins, sporting their new many-toed limbs instead of fins, left the ocean and invaded weedy coastal lagoons and river deltas (Zimmer, 1998). Eventually, they invaded fresh water and crawled out onto dry land. Then the challenge switched from preventing hypercalcemia to preventing hypocalcemia because it is harder to get this essential, yet dangerous, ion if mismanaged, in fresh water and on dry land (Whitfield and Chakravarthry, 2001). They had already prepared themselves for invading dry land and losing the protection of "anti-gravity" underwater buoyancy by arming themselves with dramatically strengthened endoskeletal ribs and vertebrae made of true "mesodermal cellular calcified bone" (Smith and Hall, 1990). These pre-adaptations enabled them to breathe without being crushed by their own weight like modern beached whales as well as to serve as large Ca²⁺ storehouses (bones have 99% of the body's calcium) from which the ion could be withdrawn when needed. But they also had to redesign their kidneys so that they could dump water without losing Ca. Around this time, calcitonin was fired from its job as the prime controller of Ca²⁺ levels. Its place was taken by PTH.

The two great engines of evolution have been gene duplication and gene fusion, the first of which gives the opportunity for safely getting novel mutant genes (i.e., new "ideas") while keeping at least one copy of the original gene information and letting the duplicates mutate freely and safely. The second of which produces novel proteins with new functions. Fusions would put a string of once individually controlled genes or gene duplicates under the command of one promoter as in the case of PTHrP. But where did PTH come from? To try to answer this, we must travel back maybe 540 million years to Cambrian Pikaia and Myllokunmingia.

The first PTH was probably what we now know as the N-terminal part of PTH and PTHrP. The gene for this peptide duplicated and perhaps fused with others to make the string of different factors—"Ur" (German for original)-PTHrP—a modern version of which, like the mammalian PTHrP, is produced by brain cells and various other organs, but, unlike the current mammalian PTHrP which doesn't circulate except in cancer patients, is a true hormone that normally circulates in the blood of fish from lampreys and sharks to sea bream (Sparus aurata) and Puffer fish (Fugu rubripes) (Danks et al., 1998; Ingleton, 2002; Ingleton et al., 2002; Trivett et al., 2002). The ancestral hormone was probably designed first to drive the development of many different tissues of the early beasts in the Precambrian seas, just as its descendent does in modern beasts. It was meant to drive Ca²⁺-dependent epithelial differentiation in various organs, later the formation by neural crest cells of tough mineralized epidermal structures such as the ancient "jawless fishes" boney head shields, dermal denticles (skin teeth), and their then-new mesenchymal cartilagenous skeletons. It also used its N-terminal region in kidneys, gills, brains, and nerves to internal control Ca²⁺ and phosphate levels in the very high, indeed the dangerously high, Ca²⁺ (e.g., 10 mM) seawater.

"Ur-PTHrP" might have been a much slimmer molecule than today's mammalian molecule (Ingleton, 2002). Indeed judging from the modern fish PTHrPs, it was rather like the mammalian PTH. It had the same N terminus, nuclear transport, and RNA-binding domains as the "modern" mammalian PTHrP (Ingleton, 2002). But it did not have the mammalian "osteostatin" domain with the 107-111 (TRSAW) sequence. On the other hand, the mammalian molecule does not have the modern fish and frog PTHrPs' functionally mysterious 38-54 domains (Ingleton, 2002). The ancient PTHrP's gene was located beside the thymopoietin gene, and the mammals have kept their PTHrP gene there during the subsequent hundreds of millions of years of evolution. This enduring linkage seems to be due to thymus epitlelial cells and parathyroid gland cells being derived from the same group of pharyngeal endodermal cells in the embryonic third branchial pouch (Gordon et al., 2001; Günther et al., 2000; Su et al., 2001).

Then something very big happened in a certain line of fish—something needed to avoid a debilitating, if not lethal, Ca²⁺ shortage which would kill anything armed only to avoid hypercalcemia that tried to leave the Ca²⁺-rich ocean for fresh water or dry land. If this had not happened, we would not be here. There was another duplication of the N-terminal part of the Ur-PTHrP gene, which was to give rise to PTH. Then there appeared another gene, gcm2 (one of the two known homologs of the Drosophila's "glial cells missing", gcm, gene (J. Kim et al., 1998)), the product of which, when the gene was switched on, staked out a part of the thymus gland's primordium for cells that made the new PTH. This gcm2-expressing part of the combined thymus-parathyroid structure was the forerunner of the separate, encapsulated parathyroid gland of terrestrial vertebrates (Gordon et al., 2001; Su et al., 2001). Echoes of the parathyroid glands originating from an ancient thymic primordium are the fact that although knocking out gcm2 in mice eliminates the parathyroid glands, the thymic epithelial cells can take over and make PTH and, of course, the fact that the mammalian PTHrP gene is still located beside the thymopoietin gene (Günther et al., 2000).

With time, the emerging pharyngeal PTH became the circulating Ca²⁺ and phosphate managing hormone. Big PTHrP then lost the job of being the circulating Ca²⁺ controller, but it kept its many other jobs of being a multipurpose, widely expressed and acting autocrine, paracrine, "intracrine", and nuclear gene-activaing polyprotein—a noncirculating string of cytokines instead of a single circulating hormone—that mediates various mesenchymally driven epithelial differentiation programs, endochondral bone formation, smooth muscle relaxation, placental Ca²⁺ transport, and neural Ca²⁺ channel repression (Chatterjee et al., 2002; Macica and Broadus, 2003; Nissenson, 2000; Re, 2002a, 2002b; Whitfield and Chakravarthy, 2001). In fact, the various add-ons (gene fusions and/or mutant duplicates of the original gene) to the big new multiple cytokine PTHrP molecule, marked out by 8 endoproteolytic consensus sites, could be cut out of the cytokine chain by processing protease scissors and then used separately to activate different receptors and stimulate the functions of the cells expressing them. On the other hand, the smaller, new PTH took over as the circulating hormonal controller of the body's all-important Ca²⁺ and phosphate levels via the bone reserves, kidney tubules, and gut epithelium. But it could also functionally overlap with big PTHrP's ancient N-terminal domain because it could still target and affect cells with the PTHR1 receptor which had been used by the Ur-PTHrP. Thus, when the first Devonian vertebrates invaded dry land 350,000,000 years ago, they already had parathyroid glands, the cells of which have Ca²⁺ sensors/receptors housed in the flask-shaped caveolae of their surface membranes with which they monitor the blood Ca²⁺ concentration (Kifor et al., 1998, 2002, 2003). If the signals from these sensors should start fading, the cells try to restore the volume of sensor signaling by releasing enough PTH to raise the circulating Ca²⁺ concentration by increasing the outflow of phosphate, stemming the outflow of Ca²⁺ through the kidney, increasing Ca²⁺ uptake through the gut, and commanding the members of the osteocyte internet and bone lining cells to pull Ca²⁺ out of the bone stores if the other two fail to do the job (Kifor et al., 2002, 2003).

PTHs Induced Bone Growth and Fracture Mending in Animals

Seventy-five years ago, the Gospel carved in stone or perhaps the Gospel-carved-in-Bone was that PTH's job in bone was to extract Ca²⁺ and dump it into the blood when needed. But Bauer et al. (1929) got a big surprise when they treated female rats with the only available PTH (a bovine parathyroid gland extract from Eli Lilly) to assess the role of trabecular bone as a readily resorbable, accessible Ca²⁺ reserve, but instead they saw—bone growth! But they didn't know what to make of it. There was an excellent reason for their reluctance to believe that PTH could be osteogenic. Even the intermittent injections that we now know strongly stimulate bone growth still stimulate osteoclasts with their acid hoses and cathepsin K shovels, but their destructive digging can be overridden by the bone-building arm, provided the injections are sufficiently far apart. However, if the PTH is injected at close intervals or continuously infused into mice or rats, the response is net bone resorption instead of net growth (Schneider et al., 2003; Tam et al., 1982).

With all due respect to Bauer et al., the bone-growing PTHs story was really kick-started by a Parisian group who boldly went where no one had gone before—they suggested that a young boy's ultimately lethal bone overgrowth (osteopetrosis) was due to the PTH secreted by his parathyroid tumor (Péhu et al., 1931)! Hans Selye was fascinated and fathered the field of PTHs as anabolics when he announced a year later (Selye, 1932) that daily injections of the Eli Lilly extract into rat pups caused a massive pile up of bone-making (osteogenic) osteoblasts on the surfaces of the trabeculae in cancellous bone, just as had happened in the unfortunate French boy. The Selye experiment gave birth to a paradox—PTH was supposed to be a potent bone destroyer, not a potent bone builder. As Riordan (1987) says on page 31 of his book "The Hunting of the Quark"—"A paradox is a sure sign that nature is trying to tell us something completely new and different about herself".

At this point, we should remember from Chapter 2i that when something like a PTH is injected, it first enters the bone via the marrow blood vessels, which means that the cells in the cortical shell will see it only after cells in the trabeculae and endosteum have had the first grab at it. Therefore, a recurring theme will be the PTHs' apparent "trabeculophilia"—its ability to stimulate trabecular bone growth more than cortical bone growth.

An incredible thing about the dramatic accumulation of osteoblasts induced by the parathyroid extract in Selye's rat pups was that it happened directly without being primed by osteoclasts as happens in human BMUs—there was a direct stimulation of osteoblast precursors in the little beasts' rapidly growing bones which we now know would have been loaded with PTH receptor-loaded, bone-building osteoblasts! But who would have believed it? In those days, people had to use a bovine parathyroid gland extract. The extract undoubtedly contained intact bovine PTH-(1-84) and some of its various proteolytic fragments, but it almost certainly contained other things (Bringhurst, 2002). And it took almost 40 years for the bone people to get around to proving that it was indeed the PTHs that did the job (Kalu et al., 1970; Walker, 1971)! Since then, the basic features of PTH's osteogenic actions described by Selye have been confirmed by the results of scores of experiments on cynomolgus monkeys, dogs, ferrets, rabbits, and sheep, but most often on OVXed rats, using a synthetic piece of PTH, h(human)PTH-(1-34), that has been wrongly believed for decades to be fully bioactive (comprehensively reviewed by Dempster et al. (1993) and Whifield et al. (1998a, 2002a, 2002b, 2002c)).

Actually, hPTH-(1-34) is not fully bioactive—the rest of the native hormone is not just functionless filler as has been generally believed! It now appears that following the gene duplication that put PTH and PTHrP on their separate evolutionary pathways, PTH acquired an additional target and functions. In the mid-1980s, T.M. Murray and his group in Toronto, Canada published almost completely ignored (by me included) evidence for there being another receptor for a region of the native hormone's C-tail (McKee et al., 1985; Rao et al., 1985). Then, in 1998, Erdmann et al. (1998) reported that the big native PTH's residues 73-76 form the core of a region that activates a receptor, the signals from which cause Ca²⁺ to surge into growth plate chondrocytes. Divieti et al. (2001, 2002a, 2002b) have also found more evidence of this receptor, which they have named the CPTHR (C-terminal PTH receptor), that sees C-terminal PTH fragments. The binding affinity is maximum (20-30 nM) for hPTH-(24-84), but drops 25-fold when residues 25-27 are missing, i.e., the affinity for hPTH-(28-84) is 500-700 nM (Bringhurst, 2002). This receptor is apparently maximally expressed by the spidery osteocytes, i.e., the locked-in-bone cells with extended processes that make mRNAs for gap junctions' connexin 43 and the mineralization-modulating osteocalcin but not the osteoblast-specific alkaline phosphatase or Cbfa1/Runx-2 (Bringhurst, 2002; Divieti et al., 2001). And the CPTHR signals promote the formation of gap junctions, which are such an important part of the intercellular connections in the osteocyte-osteointernet signaling syncytium that registers strain and microfracture (Burger, 2001; D'Ippolito et al., 2002). If the big PTH reaches the osteocytes in their lacunocanalicular hideaways, the signals from this receptor can cause these osteocytes to turn on their self-destruct apoptogenic mechanism (Bringhurst, 2002; Divieti et al., 2001). But this CPTHR-driven osteocyte killing is counterbalanced by the reduction of osteoclast generation by the activation of the CPTHRs on osteoclast progenitors (Divieti et al., 2002a, 2002b). However, the differentiating osteoclasts can be protected from lethal CPTHR signaling by having their CPTHR expression suppressed when the differentiating osteoclasts bind to immature osteoblastic stromal cells' PTH-stimulated RANKLs (Divieti et al., 2002b). Thus, the intact PTH-(1-84), unlike hPTH-(1-34), has many targets—preosteoblasts, osteoblasts, osteocytes, and of course, the many other cells in the body that have the N-terminal PTH-binding PTHR1 receptors as well as osteocytes and osteoclast precursors that have CPTHRs.

The intact PTH that has not clamped onto its target cells is promptly broken down into C-terminal but not N-terminal fragments in the liver by Kupffer cells—the half-life of circulating exogenous or endogenous intact hPTH-(1-84) is just 2 minutes (Bringhurst, 2002; Bringhurst et al., 1988; Kronenberg et al., 2001). Indeed, N-terminal fragments seem to be selectively destroyed (Bringhurst, 2002; Bringhurst et al., 1988; Kronenberg et al., 2001). One of the large circulating C-terminal fragments is hPTH-(7-84) (Kronenberg et al., 2001), which can promote osteocyte apoptosis and impair osteoclastogenesis (Bringhurst, 2002; Divieti et al., 2002). However, as we shall see, a bolus of normally rare N-terminal PTH fragments by subcutaneously injecting the hPTH-(1-34) fragment can obviously override any destructive, apoptogenic action of CPTHR signaling from the circulating large C-terminal breakdown products of the big native PTH and strongly stimulate bone formation.

These observations suggest that the selective destruction of N-terminal fragments (e.g., amino acids 1-34) of the native PTH-(1-84) could be meant to prevent circulating PTH from interfering with differentiation mechanisms driven in the cells of various tissues, including bones, by signals from the cells' PTHR1 receptors activated by the hPTH-(1-34)-like N-terminal domain of locally produced and secreted PTHrP.

Table 1

The anabolic/osteogenic PTHs

Table 1

The anabolic/osteogenic PTHs

rhPTH-(1-84) (Preos™)
hPTH-(1-38)
hPTH-(1-36)
rhPTH-(1-34) ( Forteo™)
rhPTH-(1-31)NH2
hPTH-(1-31)NH2 (Ostabolin™)
[Leu27]cyclo(Glu22-Lys26)hPTH-(1-31)NH2 (Ostabolin-C™)
hPTH-(1-30)NH2
[Leu27]cyclo(Glu22-Lys26)hPTH-(1-28)NH2 (mini-C)

Starting daily subcutaneous injections of 30 or 75 μg of hPTH-(1-34)/kg of body weight into a young female rat when she is only 5-8 weeks old and continuing for the rest of her life will produce big, grossly abnormal bones (Sato et al., 2002). The PTH injections will keep stimulating bone growth throughout the rat's life. The injections will push the growth of bones such as the femur far above normal values—indeed, so far that the marrow spaces will be nearly filled with bone, the difference between cortical and trabecular bone will disappear in a kind of fusion, and the vertebrae and femora will be greatly stiffened. Moreover, although PTH normally stimulates only the growth of lining cell-covered, marrow-bathed endosteal and trabecular bone (Andreassen and Oxlund, 2000), the lifelong massive stimulation will eventually include periosteal apposition. But, as we shall see further on, this relentless, lifelong battering by daily PTH injections into Fisher rats and the excessive amounts of PTH constantly being made in humans with primary hyperparathyroidism can have a deadly consequence—osteostarcomas.

According to the results of experiments on rats, humans, and cultured human and rat bone cells, the PTHs owe their ability to stimulate bone growth by dramatically stimulating the proliferation of marrow osteoprogenitor cells, followed by the layering of osteoblasts on quiescent bone surfaces without prior signaling from osteoclast activity exactly as Selye reported 72 long years ago (Hodsman and Steer, 1993; Kostenuik et al., 1999). Indeed a contribution of osteoprogenitor proliferation to this layering is indicated by PTH-induced stimulation of DNA synthesis in the the tibias of female rats, a response that is nearly doubled by mechanically loading the tibias (Barry et al., 2002). But there is also a direct and reversible stimulation of the ‘reversion’ of retired osteoblasts—the quiescent lining cells covering the osteonal walls, the endosteum and trabeculae—to active osteoblasts, an increase in osteoblast activity, and an increase in the osteoblast lifespan by preventing apoptotic self-destruction (Dobnig and Turner, 1995; Hock, 1999; Manolagas, 2000; Jilka et al., 1999; Leaffer et al., 1995).

The PTHs are most potently anabolic for the bones of the fast-growing fetal and newborn animals (Walker, 1971) simply because the marrow is still red and loaded with stem cells and osteopogenitor cells, and the growing bones are covered with gangs of mature osteoblasts bristling with PTHR1 receptors (Fermore and Skerry, 1995). Because the growing bones are in this super-responsive state, PTHs can so greatly stimulate trabecular bone growth that it fills the marrow spaces and strangles blood formation in these young animals (and, as we shall soon see, in at least one historically-very-significant young French boy) (Walker, 1971).

Of course, this is why Selye (1932) found that the PTH in the Lilly bovine parathyroid extract so greatly stimulated osteoblast accumulation and bone growth in his rat pups without first stimulating osteoclast activity. As fatty yellow marrow replaces red marrow (Bianco and Riminucci, 1998) and bone growth slows with age, so does the responsiveness of bone to PTH in rats (Walker, 1971). In big animals such as humans with microcracking-prone bones at their loading sites (R.B. Martin, 2002), the availability of PTHR1 receptor-bearing preosteoblasts and mature osteoblasts for an immediate or first response to PTH injection in the mature skeleton is a function of the number of BMUs (Manolagas, 2000), which is in turn linked to osteoclast activation.

Figure 16

A possible model of the PTHR1 receptor. Cohen and (more...)

Figure 16

.

A possible model of the PTHR1 receptor. Cohen and Stewart (2003, p.165) have described receptors such as PTHR1 as proteins laced through the cell membrane and that "pushing a receptor button on the outside rings a chemical bell inside". PTHR1 is a barrel with 7 α-helical staves laced through the cell membrane with a N-terminal "nose" and stave-connecting loops sticking out of the cell, as well as connecting alternating loops hanging down into the cell's cytoplasm from the bottoms of the staves. And then there is the receptor's "magic tail" (Bockaert et al., 2003) hanging out of stave 7 that attaches to a NHERF-1 and/or NHERF-2 scaffold lying beside the receptor and bearing a cluster or network of interacting signal-transmitting enzymes and factors. Depending on what's in them, these clusters can increase the variety of signals that the receptor sends into the cell when the PTH pushes the receptor's button—its N-nose—which causes the receptor to ring the chemical bell by forming its intracellular loops into a pocket for grabbing G-proteins hanging from the cell membrane by their βγ subunits and slapping its tail on the NHERF scaffold if there is one. As indicated in the drawing, the signaling starts when the PTH attaches its C-tail (C) to the receptor's N-"nose" (and perhaps to some of the barrel loops sticking out of the membrane) and then sticks its own nose (N) onto the Leu²³²-Lys²⁴⁰ patch on barrel stave 2, which pushes against extracellular loop 2, linking staves 4 and 5, and causes the staves to tilt into the active configuration as described in the legend of fig. 17. As said in the text, the signal mix emitted by the activated PTHR1 depends on whether the PTH has a normal or shortened N-"nose" and C-tail and on what scaffolds the cell has on its membrane within reach of the receptor tail as well as what things the cell has loaded onto the scaffolds. As discussed in the text, this receptor may also dimerize with another PTHR1 or even heterodimerizes with another kind of receptor to send a different mixture of signals. Abbreviations: AC, adenylyl cyclase; βγ subunits of Gi G-protein; DAG, diacylglycerol; ERK1/ERK2, extracellular signal-regulated protein kinases also known as MAP kinases; Gi, the AC-inhibiting subunit of Giβγ G-protein heterotrimer; Gs, the AC-stimulating subunit of Gsβγ G-protein heterotrimer; LPA, lysophosphatidic acid; IP₃, inositol-1,4,5-trisphosphate; LPAR, lysophosphatidic acid receptor; NaPi-CoTransp., Na⁺-phosphate cotransporter; PAPhHydl, phosphatidic acid phosphyhydrolase; PKCs, various PKC isoforms; PLA₂, phospholipase A₂; PLC-β1, phospholipase-Cβ1; PLD, phospholipase D; PtdCh, phosphatidyl choline; Rho-PK, rho (a G-protein)-controlled protein kinase.

fig. 1

fig. 16

fig. 1

Daily injections of hPTH-(1-34) also stimulate the layering of new bone, primarily on the inner cortical tunnels and trabecular trenches, and thus raise the bone mass and bone strength in ovariectomized nonhuman primates, such as Macaca fascicularis cynomolgous monkeys, above the normal level in sham-operated monkeys (Turner et al., 2001; Sato et al., 2000). But this PTH does something in OVXed monkeys that it can't do in rats—it increases the number of trabeculae (Jerome et al., 2001). But it does not do this by making new trabeculae from scratch. Instead, it does this by increasing the thickness of existing trabeculae, which are then sliced in two by osteoclastic ‘tunneling’ when they reach a certain size (Jerome et al., 1994, 2001). Trabecular tunneling is thus a primate's ingenious way of restoring the original trabecular thickness and number without making new trabeculae from scratch.

Without estrogens, the new bone in a PTH-treated OVXed rat should be, and apparently is in some cases, as much a victim of overdigging BMUs as was the old bone, when the PTH injections are stopped (reviewed by Whitfield et al., 1998b, 2000c, 2000d). But according to the results of most of the relevant experiments on rats, the new "PTH bone", be it endocortical or trabecular, can be effectively protected by an osteoclast suppressor such as a bisphosphonate, calcitonin, or an estrogen (Samnegard et al., 2001; reviewed by Whitfield et al., 1998b, 2000c, 2000d). However, the new bone's need for protection would depend on its rate of turnover (i.e., the size and activity of the osteoclast population). Thus, for example, Mosekilde et al. (1997) found that the new "PTH bone" in the sluggishly turning over bones of aged osteopenic OVXed rats did not need protection by the bisphosphonate, risedronate. And the positive effect of a 12-month treatment with PTH-(1-34) on the cancellous bone of femoral neck, iliac crest, and vertebrae of OVXed cynomolgus monkeys was maintained for 3-6 months after the injections were stopped (Jerome et al., 2001). But new ‘PTH bone’ can face another threat. If it should have grown beyond mechanical needs, it will be under-strained and consequently contain a growing number of dying osteoclasts and then be destroyed by a flock of vulturing osteoclasts (Dempster et al., 1997).

Cotreatment with an antiresorptive to block the resorption component theoretically should enhance the effectiveness of the PTH. But the maximum increase in trabecular and cortical bone mass in rats treated with an optimal dose of hPTH-(1-34) or hPTH-(1-38) alone is the same as in rats cotreated with PTH and a bisphosphonate, calcitonin, or an estrogen (Hodsman et al., 1999a, 1999b). However, cotreatment with the SERM raloxifiene reduces the dose of the PTH fragment needed to get this maximum effect. But Boyce et al. (1996) found that cotreating beagles with hPTH-(1-34) and a bisphosphonate, risedronate, was osteogenically better than treating with hPTH-(1-34) alone. On the other hand, Delmas et al. (1995) found that a bisphosphonate, tiludronate, eliminated the otherwise strong bone-building action of hPTH-(1-34) in old sheep with their slowly remodeling, hence sparsely BMUed, bones. Evidently, the only source of PTH receptor-bearing preosteoblasts and mature osteoblasts in these old beasts were BMUs. So getting rid of BMUs with the bisphosphonate eliminated the waves of osteoclasts needed to kindle an osteogenic response. In other words, there was no prior prompting from osteoclasts, few osteoblasts, hence few PTH targets to start bone growth.

OPG (osteoprotegerin), the physiological inhibitor of osteoclast generation, should potently enhance PTH osteogenicity by selectively stopping the PTH from stimulating osteoclast generation without affecting the stimulation of osteoblast accumulation and activity (Capparelli et al., 2003; Hofbauer et al., 2000; Kostenuik and Shaloub, 2001; Kostenuik et al., 2000a, 2000b). Nor would it have any of the side effects of the other less selective artificial antiresorptives. This expectation has been confirmed by Kostenuik and his colleagues (Kostenuik and Shaloub, 2001; Kostenuik et al., 2001). They OVXed 3-month-old rats and began treating them with PTH and/or OPG 15 months later. OVX significantly reduced the bone mineral density (BMD) in the distal femurs and lumbar vertebrae. Subcutaneously injecting hPTH-(1-34) (0.08 mg/kg) three times a week for 5-1/2 months into the aging rats increased the osteoclast-gnawed surface by 50%, but it also significantly increased the BMD in the distal femurs and lumbar vertebrae. Subcutaneously injecting recombinant OPG (10 mg/kg of body weight) stopped OVX from increasing osteoclast activity. The femoral and vertebral mineral density then rose as the unaffected osteoblasts continued making bone. But the OPG was less effective than the PTH injections in increasing the BMD. Injecting the recombinant OPG along with the PTH prevented the PTH from increasing osteoclast activity without affecting its osteogenic activity and consequently caused a greater increase in BMD than did the injections of either peptide alone. These observations again underscore Selye's original observations that PTH does not need a prior activation of osteoclasts to stimulate bone growth (i.e., it stimulated bone growth despite an essentially total osteoclast shut down) and that OPG strongly enhances a PTH's bone-building action. PTH only needs enough receptor-bearing preosteoblasts and osteoblasts to start the process.

It is important for managing postmenopausal bone loss to know that OPG can prolongedly inhibit bone resorption in postmenopausal women. One subcutaneous injection of an OPG dose, such as 3.0 mg/kg of body weight, reduced the urinary levels of collagen (i.e., bone) breakdown products such as N-terminal telopeptides by as much as 80% by the 4^th day after injection, and the levels were still down by as much as 14% as long as 6 weeks after the injection (Bekker et al., 2001). This prolonged suppression of osteoclast activity reduced the blood Ca²⁺ concentration. This drop turned off the parathyroid glands' Ca²⁺ sensors, which caused a prolonged elevation of the circulating PTH concentration (from a mean normal 40 pg/ml to about 80 pg/ml at 5 days and still about 55 pg/ml by 30 days) to try to bring the blood Ca²⁺ concentration back to the normal level. Of course, the hormone couldn't do this by stimulating the osteoclasts to chew up more bone, so it had to rely on reducing phosphate uptake by proximal kidney tubules and increasing the Ca²⁺ uptake by distal kidney tubules, which resulted in an 80% drop in the urinary Ca²⁺ concentration by 5 days, a 50% drop by 20 days, and a lingering 10% drop by 30 days.

As expected, the osteogenic PTHs accelerate fracture mending in both normal and osteoporotic bones. Indeed, a fracture enhances the effectiveness of PTH by triggering a shower of BMUs with their PTH receptor-bearing, hence PTH-stimulable, osteoblasts (R.B. Martin et al., 1998). For example, daily subcutaneous injections of 200 μg of hPTH-(1-34)/kg of body weight increased both the ultimate load that could be tolerated before breaking and the callus volume of rat tibial fractures by as much as 75% and 95%, respectively, over the control values by 20 days after fracturing and by 175% and 72%, respectively, over the controls during the next 20 days (Andreassen et al., 1999). A similar acceleration of fracture mending by Ostabolin-C™ ([Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂) has been reported by Andreassen et al. (2004). Kim and Jahng (1999) have also reported that a daily injection of recombinant hPTH-(1-84) for 30 days significantly accelerated the mending of surgically produced bilateral tibial shaft fractures in OVXed rats. A. Nakajima et al. (2002) have found that daily injections of 10 μg of hPTH-(1-34)/kg of body weight into male Sprague-Dawley rats with unilateral femoral fractures stimulated the probably IGF-I-mediated proliferation of subperiosteal osteoprogenitor cells in the callus, as indicated by the fraction of cells expressing PCNA (proliferating cell nuclear antigen), the DNA polymerase-δ processivity clamp (Whitfield and Chakravarthy, 2001)). By 28 days, the PTH injections had increased BMC (bone mineral content) 61%, BMD 46%, and load to failure 32% more than in control vehicle (the solution in which the PTH was dissolved)-treated rats. By 42 days, these values had risen, respectively, 119%, 74%, and 55% above those in the control untreated rats. Finally, Alkhiary et al. (2003) have reported a similar enhancenment of femoral fracture healing in male Sprague-Dawley rats.

In a different and very elegant way to accelerate fracture mending, Bonadio and colleagues implanted a degradable GAM (gene activated collagen matrix) sponge loaded with a plasmid containing DNA encoding hPTH-(1-34) into surgical breaks in rat femurs as well as beagle femurs and tibias that were so wide that they normally would have healed only very slowly or not at all (Bonadio, 2000; Bonadio et al., 1999; Fang et al., 1996; Goldstein and Bonadio, 1998). But the fibroblasts in the fractures' granulation tissue picked up the plasmid DNA and briefly became mini-bioreactors, making and secreting hPTH-(1-34) that dramatically stimulated fracture healing. Implanting a second GAM with DNA coding for BMP-4 (bone morphogenic protein-4), along with the PTH-GAM, accelerated mending even more.

PTHs can also dramatically accelerate bone formation in and around an implant—an ability of immense orthopedic importance. Indeed, screwing an implant into a bone, like a fracture, should make more PTHR1 receptor-bearing PTH targets and therefore increase the primary response to a PTH in humans by driving squadrons of osteoblast-spawning BMUs out in all directions from the screwholes (R.B. Martin et al., 1998). Skripitz et al. (2000a, 2000b) showed the implant-fixing power by inserting perforated hollow titanium chambers into the cortices of the proximal tibias of male Sprague-Dawley rats and followed the penetration of endosteal cells into the chambers and the growth of bone in the chambers. In untreated rats by 6 weeks, the bone in the chamber had been hollowed out by osteoclasts to form a marrow space with few trabeculae, but in rats which had been injected daily and subcutaneously with hPTH-(1-34), the chamber was filled with thick trabecular struts and plates. Then Skriptz and Aspenberg (2001a, 2001b) showed that subcutaneously injecting hPTH-(1-34) only on Mondays, Wednesdays, and Fridays stimulated the fixation of stainless steel screw implants to rat tibias. In just 2 weeks, the PTH injections so strongly stimulated the growth of dense fibrous bone-anchoring tissue around the screws that it took twice as much force to pull them out of the tibias than to pull the implants out of the control rats' tibias. Shirota et al. (2003) found that hPTH-(1-34) reversed the thinning of bone around titanium screw implants on ovariectomized rats. All of these may herald a new era for people needing new joints because a long lifetime of an implant needs, at the outset, a lot of new bone to build up around and cling to the implant surface without spaces that could fill up with wear debris (Davies, 2000,). This could herald a new era for people needing new joints. More recently, M. Allen et al. (2004) have reported that [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin-C™) is significantly better than hPTH-(1-34) (Forteo™) at stimulating the growth of anchoring bone around a methacrylate pin in rat tibias.

So far, we have been considering the effects of injected PTHs. What would happen if we blinded the parathyroid cells to circulating Ca²⁺ by disabling their CaRs? The cells would interpret the CaR silence as being due to a large drop in the blood Ca²⁺ concentration that urgently warrants PTH release. NPS 2143 is a CaR silencer (a "calcilytic"), which when given orally in a dose of 100 μmol/kg causes the PTH concentration in the blood of OVXed Sprague-Dawley rats to shoot up from 25 pg/ml to about 110 pg/ml in 30 minutes and then stay there for at least 4 hours (Gowan et al., 2000; Nemeth and Fox, 2003). This prolonged, indeed continuous, release of endogenous PTH dramatically increases both bone turnover and bone formation in the proximal tibia, but without a net bone loss or gain or without significantly affecting the OVX-induced trabecular crash. However, bone mineral density as well as trabecular thickness and area, but of course not trabecular number, can be increased by giving 17β-estradiol along with the calcilytic to suppress osteoclasts and resorption. But the NPS 2143/estrogen combination is still nowhere nearly as osteogenic as the subcutaneous injections of 10 nmoles/kg of hPTH-(1-34)NH₂ or [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin C™) into OVXed rats (Whitfield et al., 1998b). However, maybe the osteogenicity of the calcilytics can be improved by shortening their circulating half-life and therefore the duration of the PTH surge, which would shift the balance from resorption to bone-building. Remember the Principal Golden Rule for PTHs—they are most effective when given in well-separated brief boluses.

While it is clear that exogenous PTHs have strongly stimulated bone growth in every one of many animal models, the question remains as to whether normal circulating PTH affects bone growth and maintenance. According to Chow et al. (1998), the answer is a resounding—Yes! First, they found that removing the thyroid-parathyroid complexes from young female Wistar rats dramatically reduced bone growth in the caudal vertebrae. And they also showed that circulating PTH is required for the mechanical responsiveness. Mechanical stimulation of the caudal vertebrae in normal rats induced osteogenesis. But the vertebrae in the thyro-parathyroidectomized rats could not respond to the mechanical stimulation.

PTHs Induced Bone Growth in Humans

It was Bauer et al.'s 1929 passing mention of the Lilly bovine parathyroid extract's unexpected stimulation of bone formation in rats, but much more the case of a 8-year-old osteopetrotic boy who was killed by the aplastic anemia caused by wildly growing bone clogging his marrow spaces and stifling his blood cell production (Péhu et al., 1931), that prompted Selye's dramatic demonstration of the awesome osteoblast-generating, osteogeneic action of the Eli Lilly bovine parathyroid extract in rat pups that was to be repeated hundreds of times in the next 72 years (Selye, 1932). The thing that riveted Selye's attention to the French paper was the authors' iconoclastic suggestion that the cause of the unfortunate boy's lethal bone growth was the abnormal amount of PTH pouring out of the boy's parathyroid adenoma—a PTH-producing tumor. But as Enlow and Brown warned 46 years ago: "A great many experimental studies, conclusions, and generalizations concerning bone tissue are based on observations of laboratory animals which do not possess typical human-like bone tissue" (Enlow and Brown, 1958). However, with the advantage of 2005 hindsight, PTH's osteogenicity was already staring the Bone Community in its collective face—postmenopausal women with mild (and mild is the key word) primary hyperparathyroidism do not lose trabecular bone mass. Indeed, these women have more trabecular bone packets with greater wall width, higher bone apposition rates, and active formation periods than their healthy postmenopausal peers (Dempster et al., 1999).

Table 2

Clinical studies and trials with PTH

Table 2

Clinical studies and trials with PTH

Author	PTH (dose)	Duration	Co-Therapy	Primary Endpoint
Reeve (1976)	500 IU hPTH-(1-34)	6 months	none	histology
Reeve (1976)	500-2000 IU hPTH-(1-34)	1-6 months	none	calcium balance
Reeve (1980)	500 IU hPTH-(1-34)	6-24 months	none	histology
Hesp (1981)	500 IU hPTH-(1-34)	12 months	none	BMD
Reeve (1981)	500 IU hPTH-(1-34)	6 months	none	calcium balance
Slovik (1981)	450-750 IU hPTH-(1-34)	1 month	none	calcium balance
Slovik (1986)	400-500 IU hPTH-(1-34)	12 months	calcitriol	BMD
Reeve (1987)	1000-1500 IU hPTH-91-34)	12 months	calcitriol	BMD
Neer (1987)	400-500 IU hPTH-(1-34)	18-24 months	calcitriol	BMD
Hesch (1989)	750 IU hPTH-(1-38)	14 months	calcitonin	BMD
Reeve (1990)	500 IU hPTH-(1-34)	12 months	estrogen	BMD
Hodsman (1990)	400 IU hPTH-(1-38)	6 months	calcitonin	biochemistry
Reeve (1991)	500 IU hPTH-(1-34)	12 months	estrogen	calcium balance
Hodsman (1991)	400 IU hPTH-(1-38)	6 months	calcitonin	BMD
Neer (1991)	400-500 IU hPTH-(1-34)	12-24 months	calcitriol	BMD
Bradbeer (1992)	450-750 IU hPTH-(1-34)	6-12 months	estrogen	histology
Hodsman (1993)	800 IU hPTH-(1-34)	3 months	calcitonin	markers
Reeve (1993)	500 IU hPTH-(1-34)	12 months	estrogen	BMD
Finklestein (1994)	40 mg hPTH-(1-34)	6 months	naferelin	BMD
Lindsay (1995)	400 IU hPTH-(1-34)	36 months	estrogen	BMD
Sone (1995)	20 IU hPTH-(1-34)	6 months	none	BMD
Hodsman (1997)	800 IU hPTH-(1-34)	24 months	calcitonin	BMD
Lindsay (1997)	25 mg hPTH-(1-34)	36 months	estrogen	BMD
Lindsay (1998)	50-100 mg rhPTH-(1-84)	12 months	none	BMD
Fujita (1999)	50-200 IU hPTH-(1-34)	12 months	none	BMD
Cann (1999)	400 IU hPTH-(1-34)	24 months	none	BMC
Roe (1999)	400 IU hPTH-(1-34)	24 months	estrogen,calcitriol	BMD
Rittmaster	(2000) rhPTH-(1-84)	12 months	none	BMD
Cosman (2001)	400IU hPTH-(1-34)	36 months	estrogen	BMD
Neer (2001)	20 or 40 mg hPTH-(1-34)	19 months	none	decreased fractures
Bilezikian (2001)	400IU hPTH-(1-34)	18 months	calcitriol	BMD, markers
Dempster (2001)	400IU hPTH-(1-34)	18 months (men) 36 months (women)	cortical	increased cortical width and trabecular connectivity
Kurland (2000)	400 IU hPTH-(1-34)	18 months (men)	none	lumbar bone mass and femoral BMD
Orwoll (2003)	20, 40 μg hPTH-(1-34)Forteo™	11 months (men)	none	lumber spine, proximal femur BMD
Hodsman (2003)	100 μg hPTH-(1-84)	12 months(women)

To start the story of PTHs and humans, let's look at the rise and fall of the hPTH-(1-34) fragment in the circulation after it is injected subcutaneously. For example, Lindsay et al. (1993) have reported that injecting 25 μg of this peptide once a day into women with osteoporosis (20 μg is the recommened daily dose today for treating osteoporosis) caused a peak of circulating fragment at 30 minutes later at a level which averaged 10-times the basal level followed by an exponential drop in the level with a mean half-time of 75 minutes. Of course, this brief surge of the fragment caused the level of circulating endogenous, native hPTH-(1-84) to drop immediately by 35% as the parathyroid gland cells responded to the sudden loading of the blood with normally very rare N-terminal fragments. All of the many studies, starting in 1976, using this fragment have shown that such a daily bolus of a PTH can potently stimulate bone formation in humans as it does in dogs, mice, monkeys, and rats.

But let's go back and start at the beginning. In 1976, Reeve et al. (1976a, 1976b) reported that single daily injections of 100 μg/kg of hPTH-(1-34) stimulated iliac trabecular bone growth, but they did not affect femoral BMD. In a subsequent uncontrolled, multicenter study, they injected 16 osteoporotic women and 5 osteoporotic men once each day with 50 to 100 μg/kg of hPTH-(1-34) for 6 months to 2 years (Reeve et al., 1980). The treatment produced a significant amount of new, normally mineralized iliac lamellar trabecular bone.

However, before continuing, I must also note Heaney's warning that BMD (areal density) is the "misbegotten surrogate of bone mass" (Heaney, 2003). Actually as discussed above, the level of remodeling activity is a better indicator of bone fragility than bone mass or BMD.

In Lindsay et al.'s 3-year study, single daily subcutaneous injections of 25 μg of hPTH-(1-34) increased the vertebral BMD by 12.8% in 27 postmenopausal osteoporotic women who also received estrogen, which by itself did not affect the BMD in the 25 women of the control group (Lindsay et al., 1997). As expected, the vertebral BMD rise was accompanied by a drop in vertebral fracturing, which was indicated by the PTH-treated women not losing as much radiographically measured vertebral height as the untreated women. Just as in the animal models, the increases were less in bones with smaller first-to-get-PTH cancellous (trabecular) compartments, such as hip (4.4%) and forearm (1.0%). The whole body BMD increased by 8%.

In a follow-up to this study, Cosman et al. (2001a) reported that the BMD did not change during the 3-year treatment and for 1 year afterwards in the control group receiving HRT (hormone (estrogen) replacement therapy). In the PTH-HRT group, the biochemical markers of both bone formation (bone-specific alkaline phosphatase) and resorption (urinary crosslinked collagen N-telopeptides)—peaked at 6 months and returned to baseline levels by 30 months. During this period, the high cancellous spinal bone mass increased 13.4%, the lower cancellous hip bone mass increased 4.4%, and total body bone mass increased 3.7%. The bone masses did not change significantly during the first year after stopping the PTH injections but continuing HRT. The PTH-HRT treatment dramatically and significantly reduced the number of "incident" (in other words "spontaneous" or muscle-pull or bending-crush) vertebral fractures during the follow-up year. The frequency of incident fractures in the PTH-HRT group was reduced to as little as 0% or 25% of the frequency in the HRT-only group, depending on the choice of height reduction (their measure of vertebral crushing) ‘cut-off point’.

A larger increase in vertebral BMD was seen by Hesch et al. (1989) in a smaller group of 13 osteoporotic women. After 14 months of single daily subcutaneous injections of 54 μg (about 700-750 Units of activity) of hPTH-(1-38), the mean vertebral BMD had increased by 20%. This large response might have been due to the post-PTH prevention of a resumption of resorption by the nasally sprayed anti-osteoclast calcitonin that was also given to the women.

In a randomized, controlled 2-year trial with 30 osteoporotic women receiving 6 treatment cycles, each one of which consisted of 1 month (28 days) of single daily subcutaneous injections of 800 U (or about 60 μg)/day of hPTH-(1-34) followed by a 3-month rest period (to restore full responsiveness to the PTH for the next round of injections), there was a 10.1% increase in the BMD of lumbar vertebrae, a nonsignificant 2.4% increase in the femoral neck BMD, and a 80% drop in the vertebral fracture incidence (Hodsman et al., 1997). However, in this study treating the osteoporotic women with the antiresorptive calcitonin during the rest periods did not enhance PTH's osteogenicity.

Rittmaster et al. (2000) have reported the results of a study in which 75 postmenopausal women were treated for 1 year with daily injections of 50, 70, or 100 μg of recombinant hPTH-(1-84) and then with a daily dose of 10 mg of alendronate for the next year. This treatment stopped cortical bone loss and increased the spinal bone mass by 14% as compared to 6.9 to 9.2% increase in patients without subsequent alendronate treatment.

In a much larger study, Fujita et al. (1999) treated 220 osteoporotic patients with weekly subcutaneous doses of 50 U (3.7 μg; 0.8 nmole), 100 U (7.4 μg; 1.6 nmole), or 200 U (14.8 μg; 3.2 nmole) of hPTH-(1-34) for 48 weeks. The highest dose caused a significant 8.1% increase in the lumbar vertebral BMD, without affecting cortical thickness or metacarpal BMD. The increase in vertebral BMD was accompanied by a 30-40% reduction in "backache".

In another study (Roe et al., 1999), osteoporotic women injected themselves with a placebo or 400 U (about 30 μg) of hPTH-(1-34) subcutaneously once each day for 2 years and also took an oral estrogen (0.625 mg Premarin/day) with or without methoxyprogesterone. They all received 800 U of vitamin D and 1500 mg of calcium/day. By the end of 2 years, the overall (cortical plus trabecular) lumbar vertebral BMD in the PTH/estrogen-treated group was a dramatic 28.3% higher than in the placebo/estrogen-treated control group. The femoral neck BMD in the PTH/estrogen-treated group was 10.8% higher than the femoral neck BMD in the placebo/estrogen control group. The density of the trabecular (cancellous) compartments of the L1 and L2 vertebrae in the PTH/estrogen-treated women (as measured selectively by QCT) had increased even more dramatically—it had increased by 74% during the 2 years while the mean density in the vertebrae of the placebo/estrogen-treated controls had dropped 2.1%.

At the end of 1998, Lindsay et al. (1998) summarized the results of a one-year, Phase II, multicenter (18 centers in Canada and the USA), placebo-controlled, double-blind study of the effects of recombinant rhPTH-(1-84) (Preos™; NPS Pharmaceuticals Inc.'s ALX-1-11 from the now-defunct Allelix Biopharmaceuticals) on BMD and serum indicators of bone resorption and formation in 217 postmenopausal osteoporotic women between 50 and 75 (average, 64.5) years of age. The results of the prior Phase I study had indicated that one subcutaneous injection of 0.02 to 5.0 μg of ALX-1-11/kg into healthy postmenopausal women did not produce a frank hypercalcemia, even at the highest dose (though you will learn further on that this is not the case), which indicated that the hormone was safe and well-tolerated in this dose range (Schweitert et al., 1997). The failure to cause hypercalcemia could be due to the killing of osteoclasts by the stimulation of their CPTHR receptors for a C-terminal domain of the large native PTH (Divieti et al., 2002). The Phase II patients were given single daily injections of 50, 75, or 100 μg of hormone (≈ 0.8 to 1.6 μg/kg). The largest dose increased the vertebral BMD by 6.9%, while the densities of the vertebrae in the placebo-treated women did not change. However, the BMD dropped in the arms and legs by 0.3% (75 μg) and 0.9% (100 μg), respectively. The serum levels of both formation indicators (bone-specific alkaline phosphatase, osteocalcin) and resorption indicators (deoxypyridinolines, N-terminal cross-linked collagen peptides) increased 100 to 200%. There were no serious side effects, nor did the patients make antibodies to the recombinant human hormone. The authors believed that the drops in arm and leg BMDs were only transient, and that PTH appeared to be a promising novel treatment for osteoporosis.

Transient though they might have been, the drops in the BMDs of the arms and legs of Lindsay et al.'s patients are reminiscent of the results of an earlier study by Neer et al. (1991) that nearly kicked the PTHs off the list of credible therapeutics for osteoporosis. As expected, giving hPTH-(1-34) and calcitriol (1α,25(OH)₂vitamin D₃) to 15 osteoporotic women for 1 to 2 years increased the lumbar vertebral BMD by a whopping 32%! But at the same time there was a net 4% drop in the radial BMD, which was what set off the alarm bells. The great fear was that although PTH can reduce the unpleasant vertebral crushing, it might increase the far more crippling hip fracturing. To explain this unsettling result, they suggested that PTH somehow steals cortical bone to make trabecular bone. Fortunately for the PTHs, this ‘Cortical Steal’ hypothesis as it was called has mercifully died, although there is a reason why there is often a transient drop in cortical BMD. So the common, though not universal (e.g., Lindsay et al.'s (1998) study) experience has been virtually the same as the experience with nonhuman bones—PTHs strongly stimulate trabecular bone growth but either do not affect or less strongly stimulate cortical bone growth. To understand this, remember that the trabecular cells get the first grab at an injected PTH coming into the bone in the nutrient blood vessels.

Neer et al. (2001) and Zanchetta et al. (2003) have reported the effects on bone geometry and fracturing of daily subcutaneous injections of a placebo, or 20 or 40 μg of Eli Lilly's recominant hPTH-(1-34) (Forteo™) into 1637 post-menopausal women who had already suffered vertebral fractures. These numbers were from this venerable, 33-year-old peptide's phase III clinical trial. The relative risk of vertebral fracturing was reduced to 0.35 (95% CI, 0.22-0.55) in women receiving daily injections of 20 μg of the peptide and to 0.31 (95% CI, 0.19-0.50) in women receiving daily injections of 40 μg of the peptide. The relative risk of nonvertebral fracturing was 0.47 in women receiving 20 μg of the PTH (95% CI, 0.25-0.88) and 0.46 in women receiving 40 μg (95% CI, 0.25-0.86). Both doses of the PTH increased the BMD of hip and spine by 2 to 4%, but again there was the usual drop in the BMD of the radial shaft (i.e., cortical bone) during the first year. However, according to Zanchetta et al. (2003), using pQCT instead of DEXA, hPTH-(1-34) actually significantly increased the BMD and area of the distal radius. Thus, hPTH-(1-34), Lilly's Forteo™, improves bone geometry as indicated by increased axial and polar cross-sectional moments of inertia, decreases the risk of wrist, vertebral and nonvertebral fracturing, increases femoral and vertebral BMDs, and is well tolerated, although there was some (29%) hypercalcemia with 40 μg, which, because it causes such unpleasant things as muscle weakness, fatigue, nausea, vomiting, and kidney stones, unfortunately limits the dose that can be used. Indeed, according to Lane et al. (2003), a daily injection of 40 μg of hPTH-(1-34) caused the circulating level of the osteoclastogenic soluble RANKL to peak broadly around 250% of the baseline value between the 3^rd and 9^th month and the osteoclastogenic IL-6 to peak at about 150% of the baseline value at the 3^rd month. Obviously, the new potently anabolic PTHs such as oral Ostabolin™ or Ostabolin-C™ (Jolette et al., 2003) that apparently do not cause hypercalcemia, even at very high doses, will have a considerable clinical advantage over Forteo™ when they complete their clinical trials.

The answer to the crucial question of what the PTH treatment does to cortical bone seems to have been provided by Cann et al. (1999). They used 3D-QCT to separately track the hPTH-(1-34) (Forteo™)-induced changes in the density and mass of the cortical and trabecular (cancellous) parts of the proximal femurs of the osteoporotic women in Roe et al.'s (1999) study. By the end of the second year, the PTH injections had increased the spinal BMD by a very substantial 74% and the BMD of the trabecular compartments of the proximal femur from 10.5% in the trochanter to 12.4% in the whole hip. On the other hand, the Density Devil reappeared here again—the cortical BMD dropped by less than 3.5% during the first year. But, as in Lindsay et al.'s (1998) study, the drop then stopped. So were Neer et al. (1991) right? Yes, but PTH treatment is certainly not hazardous for femoral necks. Despite the small drop in cortical bone density, the use of 3D-QCT enabled Cann et al. to see that the cortical bone mass in the proximal femurs had actually risen 10.9% in the trochanter and by an awesome 20.7% in the femoral neck by the end of the second year. These large increases in cortical mass took place on the endosteum (just as Andreassen and Oxlund (2000) found in the femurs of hPTH-(1-34)-treated OVXed rats) because the endosteum, unlike the periosteum, is both bathed in marrow and covered with PTH receptor-bearing lining cells, which like the equally marrow-bathed trabecular lining cells, can throw PTH-induced growth factors directly at osteoblastic progenitors in the adjacent marrow stroma (Kostenuik et al., 1999; Pun et al., 2001). Thus, as expected in a mature skeleton, PTH rapidly stimulated osteoclast generation by RANKL-producing immature PTHR1 (the common PTH receptor)-poor osteoblastic stromal cells, which would reduce the cortical density by increasing the porosity (i.e., the number of osteoclast-dug tunnels). But this was far more than balanced by a delayed, large production of endocortical bone by increased numbers of osteoblasts resulting from the PTH's three osteoblast-generating actions that will be unveiled in the next chapter.

The same changes have been seen by Burr et al. (2001) in our OVXed relatives, cynomolgus monkeys (Macaca fascicularis), that received one daily subcutaneous injection of 1 or 5 μg of hPTH-(1-34)/kg for 12 or 18 months. The injections increased the cortical porosity, especially in the inner third of the mid-diaphysis (the middle part (see Netter, 1997) for beautiful drawings) of the left humerus, where it was increased 5 to 16 times above the porosity in the sham-operated monkeys. (Ostabolin-C™ causes far less porosity in cynomolgous monkeys (Paul Morley, personal communication).) Surely this must have reduced the bone's mechanical strength. But it did not! The strength actually increased, because although the osteoclasts were digging more holes in the endocortical zone, the PTH-stimulated osteoblasts were busily increasing the mineralizing surface 11-fold more than in the controls and making bone 4 times faster than their peers in control monkeys.

Sato et al. (2004) have also seen the same thing in 9-year-old, OVXed cynolmolgous monkeys that had been subcutaneously injected daily for 18 months with 1 or 5 mg of hPTH-(1-34) (Forteo™). The PTH increased osteon activation, osteonal bone formation, and cortical width and area, but it also increased bone turnover and cortical porosity. However, again the increased bone growth more then compensated for the increased porosity and the bone strength (as indicated by the ultimate load to failure) ranged above the weakened OXV value and the normal sham value.

A similar conclusion was reached by Hirano et al. (2000) from a study of the effects of an osteogenic dose of hPTH-(1-34) on cortical bone porosity and bending rigidity of rabbit tibias. They found that here too the PTH greatly increased the porosity, mostly on the endocortical surface, but this was more than offset by the formation of new endocortical bone as well as periosteal bone, which increased the tibial bending rigidity. The same coincidence of bone loss and large bone building has been seen by Whitfield et al. (1998b) in OVXed rats where hPTH-(1-34) injections did not stop the estrogen-deprived, stimulated osteoclasts from destroying the femoral central metaphyseal trabeculae but they increased thickness of the surviving, laterally sited trabeculae 1.65 times.

At this point, the reader might ask whether preventing the PTH-induced surge of osteoclasts boring holes in the cortex would increase the potency of the anabolic response to a PTH. And the answer is almost certainly yes. "Yes" if one uses the potent, long-acting, and safe anti-osteoclastogenic osteoprotegerin (Kostenuik et al., 2001; Smith et al., 2003) instead of the bisphosphonate alendronate (Fosamax™), which does kill osteoclasts but unfortunately as we shall see below also interferes with the expression of osteoblast-specific genes (Onyia et al., 2002).

Accumulating mature osteoblasts should release an indicator of their presence, such as osteocalcin into the blood. Indeed, this is what happens in PTH-treated osteoporotic humans. The hPTH-(1-34) injections in the study of Lindsay et al. (1997) caused the serum osteocalcin content to rise for the first 6 months, after which it slowly subsided. Cosman et al. (1998, 2000) found that the PTH injections in these experiments caused the serum osteocalcin and pro-collagen I-C-terminal propeptide (released during collagen fiber processing and assembly) to rise during the first 4 to 6 weeks, which indicated a rapid build-up of osteoblasts. There should also have been evidence of PTH-stimulated osteoclasts. But the osteoclast response, as indicated by the release of cross-linked N-telopeptides from osteoclast-cleaved collagen, was not detectable until 6 months and then dropped to baseline values by 24 months.

Reeve et al. (2001) have reported the effects of treating severely osteoporotic women with hPTH-(1-34) or hPTH-(1-38). They found that the PTHs improved the "body calcium balance", "impressively" increased the spinal BMD, and stimulated smaller increases in proximal femur and radius. They concluded "hPTH or comparable PTH receptor activators remain the most promising anabolic treatment for osteoporosis currently under clinical evaluation …".

In mid-2003, Cosman et al. reported the preliminary results of a study on osteoporotic patients that could revolutionize the PTH treatment protocol. Normally, an osteoporotic patient self-injects 20 μg of hPTH-(1-34) (Forteo™) once daily for no more than 2 years. However,Cosman et al. (2003) have tested in osteoporotic women the cyclic protocol first tried by Hodsman et al. (1997). The first reason for cyclic treatment is based on the fact that daily injection of hPTH(1-34) (Forteo™) initially stimulates bone formation more than resorption, but resorption eventually catches up by 6 months. (This might be due to target cells such as the anti-osteoclast, OPG-producing, bone-building osteoblasts and potential osteoblast lining cells having enough receptors to make them more immediately responsive than the pro-osteoclast, RANKL-producing progenitor cells (Atkins et al., 2003)). So if there is a sufficiently long rest period between injections, ostoclast production and resorption would not have a chance to catch up. The second reason is that if cyclic treatment works at least as well as continuing daily injections, it could be only half as expensive/year for the patient. The preliminasry results, like those of Hodsman et al. (1997), are encouraging. When hPTH-(1-34) (Forteo™) was "on" for 3 months, the serum markers of bone formation rose, dropped when it was "off " for the next 3 months, but then rose robustly when turned on again during the next 3 months. By the end of 9 months (one cycle), there was the same approximately 4% increase in the patients' lumbar vertebral BMD that had been stimulated by either the costly, continuing-daily treatment or the cheaper, cyclic treatment.

Glucocorticoid therapy, like aging, wandering around in space, or lying immobilized in bed, can also cause osteoporosis. It does this by inhibiting osteoblastogenesis (apparently by suppressing Cbfa1/Runx-2 and Cbfa1/Runx-2-stimulated TGF-β type I receptor expression (Chang et al., 1998; Ducy, 2000)), promoting osteoblast and osteocyte apoptosis, and inhibiting intestinal Ca²⁺ uptake (Lane et al., 2000; Weinstein et al., 1998). Injecting a PTH can overcome the glucorticoid action and stimulate bone growth. This is shown in a recent experiment involving 51 women (mean age of 63 years) receiving glucocorticoids and estrogens (Lane et al., 2000). Twenty-eight injected themselves with 40 μg (400 U) of hPTH-(1-34) (Forteo™) once each day for 12 months and continued taking estrogen, while 23 just continued taking estrogen. Here again as expected, PTH preferentially stimulated growth of cancellous (trabeculae-rich) bone. By 12 months, the PTH injections had increased the trabecular mass (as measured by QCT) mainly in cancellous lumbar vertebrae by about 35%, which continued rising to 45% during the next 12 months as the PTH-enhanced remodeling space (remember that PTH also increases osteoclast generation and thus excavations) continued to be filled up after the injections were stopped. By 24 months, the overall BMD (as measured by dual energy X-radiation absorptiometry instead of QCT), which is only a function of the extent of mineralization and the amount of remodeling space in the increasingly massive vertebral bone, had increased by only 12.6%. The hip mass rose by 4.7% over the 24-month period. However, the treatment did not affect the cortical bone mass in the forearm.

Rehman et al. (2003) have also found that treating women with glucocorticoid-induced osteoporosis for 1 year with hPTH-(1-34) (Forteo™) plus hormone-replacement therapy (HRT) increased the cross-sectional areas of their L1 and L2 vertebrae by 4.8% over the areas in patients given HRT only. hPTH-(1-34) (Forteo™) also increased the compressive strength of the vertebrae by 200% over the baseline levels, while HRT alone had no effect.

PTH can also treat idiopathic (a fancy word for "cause is unknown") osteoporosis in middle-aged men (Kurland et al., 2000). Unlike postmenopausal osteoporosis in women, which is associated with high bone turnover due to osteoclastic overactivity, male idiopathic osteoporosis is associated with low bone turnover and thus is not as amenable to the drugs designed to suppress osteoporotic postmenopausal women's demonic bone diggers. Kurland et al. (2000) reported the results of an experiment on 23 men (50 ± 1.9 years of age) with idiopathic osteoporosis. They gave 10 of the men a daily subcutaneous injection of 400 IU (about 30 μg of hPTH-(1-34)(Forteo™) for 18 months. While the lumbar spine bone mass did not change in the 13 placebo-treated control patients, the PTH increased it increased by a significant 13.5%.

A very important important question at this point is—how does the human skeletal response to Lilly's hPTH-(1-34) (Forteo™) compare with the response to the Merck-Frosst's bisphosphonate alendronate (Fosamax™)? Body et al. (2002) treated 73 postmenopausal osteoporotic women with 10 mg of oral alendronate plus subcutaneously self-injected placebo (instead of PTH) and oral calcium (1000 mg) and vitamin D (400-1200 IU) each day for 14 months. Another 73 postmenopausal osteoporotic women injected themselves subcutaneously once daily with 40 μg hPTH-(1-34) (Forteo™), took an oral placebo (instead of alendronate), and the same calcium and vitamin D supplements for the same period of time. The greater effectiveness of PTH appeared as early as 1 month, and by 14 months alendronate had increased the BMD of the lumbar spine by only 5.6%, while hPTH-(1-34) had increased it by 12.2% (p < 0.001). The PTH also increased the femoral neck BMD by about 6%, while alendronate increased it by only about 2%. And by 14 weeks, PTH had increased total hip BMD by about 4.5%, while alendronate had increased it by about 2.5%. And the nonvertebral (ankle, foot, radius, ribs, toe) fracture incidence was 4.1% in the PTH-treated group compared with 13% in the alendronate-treated group. Interestingly, the PTH promptly started to increase bone alkaline phosphatase, an indicator of bone growth, by about 50 to 100%, while alendronate equally promptly caused the alkaline phosphatase level to drop by 50%. This would be expected because of alendronate's ability to inhibit the genes for bone formation such as alkaline phosphatase, collagen I, and IGFs-I and II in the proximal femoral metaphyses of OVXed rats (Onyia et al., 2002). Of course, the Density-Loss Devil appeared again—there was the usual small drop (about 4%) in the distal radial BMD in the PTH-treated group, which was expected from other studies in humans, monkeys, and rabbits, and was probably transient and not reflective of the actual increase in bone strength (Burr et al., 2001; Cann et al., 1999; Hirano et al., 2000; Neer et al., 1991, 2001). Clearly, hPTH-(1-34) (Forteo™) is much better than alendronate at increasing BMD at various skeletal sites and decreasing nonvertebral fractures in humans (D.M. Black et al., 2003; Finkelstein et al., 2003).

Alendronate cotreatment actually blunts the anabolic actions on hips and vertebrae of daily injections of 40 μg of hPTH-(1-34) (Forteo™) into osteoporotic men (Finkelstein et al., 2003) and 100 μg of hPTH-(1-84) into osteoporotic women (D.M. Black et al., 2003). The reason for this has been found in rat femurs by Onyia et al. (2002)—alendronate prevents hPTH-(1-34) (Forteo™) from stimulating key bone-growth-driving genes.

Finally, an equally important question is—Does prior exposure to bisphosphonates or SERMs compromise the anabolic action of a subsequent treatment with hPTH-(1-34) (Forteo™)? This is very important because most if not all osteoporotic patients coming to the clinician's office for PTH treatment will have been receiving one of these antiresorptives. Ma et al. (2003) have reported that prior treatment of OVXed Sprague-Dawley rats for 10 months with alendronate (Fosamax™), 17α-ethinyl estradiol, or raloxifene before switching over to hPTH-(1-34) (Forteo™) did not affect the responsiveness of the rats' skeletons to the PTH. But this result would be expected because rats retain a substantial pool of PTH-responsive osteoblastic cells for their continuously growing bones. However, this might not be the case with old humans in whom an antiresorptive-induced lack of BMUs and microcrack repair would blunt the response to a PTH, as happened in Delmas et al.'s (1995) antiresorptive-treated old sheep. And this appears to be the case for the potent alendronate (Fosamax™) in humans.

Ettinger et al. (2004) measured the effects of prior treatment with alendronate (Fosamax™) or raloxifene on the anabolic action of daily injections of hPTH-(1-34) (Forteo™) in postmenopausal, 60-87-year-old women for 18 months after stopping the bisphosphonate treatment. Briefly, raloxifene pretreatment did not affect the ability of the PTH peptide to increase the BMD in hip and spine, but the BMDs in the alendronate-pretreated women increased later and peaked at lower levels than in the raloxifene-pretreated women. The authors suggested that the much longer-lasting suppression of remodeling by alendronate pretreatment than by raloxifene pretreatment would more severly reduce the pool of PTHR1-bearing target osteoblasts for stimulation by the PTH injections in the alendronate-pretreated women as probably happened in Delmas et al.'s antiresorptive-treated old sheep.

The Starter Gun: The PTHR1 Receptor and the Signals It Fires

"…the receptor protein sticks its molecular equivalent of head and arms out to catch passing…molecules and keeps its foot on the molecular gears inside the cell."—L.H. Caporale (2002; p. 139).

"…it is getting more and more complex and we are frequently overloaded with information. In addition, we have moved from absolute fidelity (one ligand, one receptor, one G protein, one effector, etc.) to almost complete promiscuity (multiple ligands, receptors, G proteins, effectors, and all interacting among them)."—J. Vazquez-Prado et al. (2003; p. 556).

To tackle the formidable job of understanding how the PTHs stimulate bone growth in humans, rodents, and other animals, we must know where and how things start. What signals do they send into their target cells via the PTHR1 receptors to trigger osteogenesis? However, before going on I must warn the reader that most people call the receptor PTH1R. Since there is no type 1 PTH (i.e., PTH1) but there is a type 1 PTH receptor, I use what surely must be the more accurate name for the receptor—PTHR1 (i.e., PTH's receptor No.1).

But aside from the semantics, how do I know that the PTHR1 receptor is the molecular gun which the PTHs use to fire their osteogenic bullets into the bone cells? To show this, Calvi et al. (2001) and Schipani et al. (2002) coupled a mutant human PTHR1 gene (HKrk-H223R) encoding a permanently switched-on receptor to the mouse a(I) collagen gene promoter in order to restrict the gene's expression to the osteoblasts in a "transgenic" mouse, otherwise there would have been widespread chaos in the many other PTHR1-expressing tissues. The receptor encoded by this gene was "frozen" in an active configuration—it didn't need a PTH to activate it. The switched-on PTHR1, like normal PTH-activated PTHR1s, increased endosteal bone formation, caused an accumulation of inter-trabecular preosteoblastic marrow stromal cells and a build-up of hyperactive trabecular osteoblasts in both proximal tibia and cranial bones, and dramatically increased trabecular bone volume. As must happen with PTH infusion or, as in this case, nonstop PTHR1 signaling, there was also a large increase in cortical and trabecular osteoclasts, a lack of periosteal growth, increased cortical porosity, and a resulting drop in cortical bone mass. However, there was no bone loss, and the trabecular volume was even higher in doubly-mutant mice making collagenase-3-resistant a(I) collagen along with the switched-on PTHR1 (Schipani et al., 2002). In these resistant beasts, the diggers couldn't reduce the bone build up by chopping up the collagen matrix.

Figure 17

What the PTH-PTHR1 complex might look like from (more...)

Figure 17

.

What the PTH-PTHR1 complex might look like from above after the attachment of the PTH to the receptor. The PTH are the two cylindrical α-helices connected by thick black lines. In this cartoon, the PTH has just landed on the receptor's N-nose and shoves its nose against extracellular loop 2 linking staves 4 and 5, which strains the S-S bond linking barrel staves (i.e., α-helices) 3 and 4. This causes barrel stave 3 to tip and trigger a rearrangement of the staves that opens the way for a nearby, or in fact associated, GDP•Gs-protein to be pulled into a pocket formed by the intracellular stave-linking loops, where it is activated by having the GDP removed and a GTP put in its place (Karnik et al., 2003). The GTP•Gs subunit then activates the nearby adenylyl cyclase hanging from the cell membrane (Cooper, 2003). But this will depend on whether the receptor tail is attached to a NHERF scaffold lying to one side and what other signalers are on the scaffold as in fig. 16.

fig. 16

fig. 17

fig. 17

fig. 16

fig. 17

fig. 17

fig. 16

fig. 16

The PTHR1 signaling must not be allowed to continue beyond a certain time if it is to be optimally effective. As we shall see further on, one way of doing this is to simply drag the PTHR1•PTH complex into the cell on a β-arrestin platform, pull the PTH off and destroy it, and put the cleaned up receptor back to work on the surface. However, the job of being the cell's general GPCR receptor receptor off-switch has been given to RGS-2, a small member of the B/R4 clan of the 26-member family of RGS (regulators of G-protein signaling) proteins (Hollinger and Hepler, 2002; Petsko and Ringe, 2004). PTHs rapidly stimulate RGS-2's gene via adenylyl cyclase/cyclic AMP (but we don't know what happens when PLC but not adenylyl cyclase is stimulated), and the resulting RGS-2 protein shuts the signaling off by stimulating the conversion of active GTP•Gαs and GTP•Gαq to inactive GDP•Gαs and GDP•Gaq as well as by directly inhibitng AC (Hollinger and Hepler, 2002; Miles et al., 2000; Thirunavukkarnasu et al., 2002; Tsingotjidou et al., 2002). However, RGS-2, like RGS-4, another small member of the B/R4 clan, might have a more positive role, such as triggering Ca²⁺ oscillations (Hollinger and Hepler, 2002). But it appears that RGS-2 might either require help from β-arrestin to shut the receptor signaling off or it is not really needed for this, because knocking out the β-arrestin-2 gene in mouse osteoblasts prolongs PTH-triggered cyclic AMP signaling, which increases cortical (but not trabecular) bone growth (Pierroz et al., 2003).

So far, the effort to find out how a PTH such as hPTH-(1-34) causes the PTHR1 barrel loops and staves to tumble into an active G-protein-grabbing state has focused only on the stimulation of AC as the activation indicator. Peptides such as hPTH-(3-34), hPTH-(7-34), hPTH-(13-34), hPTH-(28-32), hPTH-(28-42), or hPTHrP-(5-36), which have been N-terminally truncated (i.e., have had their normal N-terminal noses cut off ) are very wrongly labeled antagonists only because they cannot activate AC. But they can activate PTHR1 receptors and stimulate PLC-β1 and PKCs (osteoblasts variously express PKCs-α -β, -δ, -ε, -η, -θ, -ζ, -ι/λ!) as strongly as hPTH-(1-34) in some cells, such as freshly isolated human foreskin fibroblasts, OK opossum kidney cells, ROS 17/2 cells, UMR-106 cells, and S49 murine T-lymphoma cells (Azarani et al., 1995a, 1995b; Chakravarthy et al., 1990;Jouishomme et al., 1992; Whitfield et al., 1998b, 2001). Because of this persistent mistake, there has been a large gap in our understanding of PTHR1 signaling. But this gap is now being narrowed by the discovery in the last couple of years of PTH's "magic tail" to useBockaert et al.'s (2003) happy term.

The PKCs activation story has now become even more complicated. It appears that hPTH-(1-34) has two separable ways to stimulate PKCs (Yang et al., 2004). When wt9 murine osteoblasts with 800,000 to 1,000,000 PTHR1 receptors/cell are exposed to hPTH-(1-34), they activated AC and phosphorylated and activated their PKC-d. [G1, R19]hPTH-(1-28), which lacks both the PLC-activating and the 29-34 PLC-independent PKC-activating regions (Takasu et al., 1999; Whitfield, Isaacs et al., 2001), did not persuade the wt9 cells to activate PLC or phoshorylate PKC-δ (Yang et al., 2004). On the other hand, the AC-activating [G1, R19]hPTH-(1-34), which still lacks the PLC-activation region, caused the cells to activate PKC-d because it has the 29-34 PLC-independent PKC-activating region that was discovered in my laboratory by Jouishomme et al. (1992)).

fig. 17

fig. 16

fig. 16

Because of its PDZ patch, PTHR1's tail, like a dog's tail, is a sophisticated signaling device. The signaling by the PTHR1 receptor, like several other GPCR receptors, such as the β₂-AR, is in some cells mediated by a large particle called a "transducosome" consisting of a core scaffold and several attached and interacting signal components (Kreienkamp, 2002). These transducosomes can be awfully complex—the 5-HT_2c receptor's tail, for example, can associate with as many as 15 different proteins (Becamel et al., 2002)! And PTHR1 plugs into a transducosome with its tail. One of the cores of PTHR1's transducosomes is NHERF-2 , a kind of scaffold or circuit board attached to the cell membrane, loaded with a set of interconnected signaling enzymes, and is tethered to underlying cytoskeletal actin cables via an "ERM" (ezrin, radixin, moesin) bridge (Harris and Lim, 2001; Kreienkamp, 2002; Mahon and Segre, 2002; Ponting et al., 1997; Voltz et al., 2001). Thus, if the cell has NHERF-2s , each of its PTHR1 receptors can be part of a signal-transducing supercluster on the NHERF-2 circuit board perhaps along with another GPCR receptor as well as several membrane-associated G-proteins and other components such as PLC-β1.

The NHERF-2 scaffold in cells such as PS120 fibroblasts grabs PLC-β1 by the tail with its PDZ-I domain and puts the phospholipase within easy reach of the active PTHR1 tail, and subunits from the associated membrane-anchored Gi/o or Gq/11 heterotrimers (Mahon et al., 2002; Suh et al., 2001). It appears that when the target cell has NHERF-2 (and it is very important to know that not all cells do!) and PTH-activated PTHR1 attaches to the scaffold with its cytoplasmic C-tail's PDZ-II-binding patch, AC stimulation is suppressed, and PLC-β1 stimulation is greatly amplified. Why?

The reactions triggered by the activated PTHR1 are taking place in a dense cluster of enzymes, G proteins, and receptors on the underside of the cell membrane. In PS120 cells, the mechanism seems to be due to the NHERF-2-connected PTHR1 receptor generating active GTP•Gi/oa subunits that find, bind, and inhibit adenylyl cyclase and the NHERF-2 complex abetting this cyclase suppression by hindering the coupling of the AC—activating Gs in the pocket formed by the intracellular loops to the adenylyl cyclase. Meanwhile, the βγ subunits released from the Gi/o or Gq/11 complexes stimulate the PLC-β1 sticking to the NHERF-2 circuit board's PDZ-I domain (Mahon et al., 2002).

This means that cells with lots of NHERF-2s will respond to a PTH very differently from those with no NHERF-2. Thus, ROS 17/2.8 rat osteosarcoma cells without NHERF-2 respond to PTH-(1-34) with as much as a 30-fold burst of adenylyl cyclase activity but no PLC-β1 activity, but if NHERF-2 be put into these cells, the PTH can stimulate PLC and PKC (Mahon et al., 2002). By contrast hPTH-(1-34) can stimulate PLC activity 3-4 fold and adenylyl cyclase by only 1.2-fold or less in ECV304 cells, which have lots of NHERF-2 (and NHERF-1) (Mahon et al., 2002). And then there are the lymphocytes with lots of NHERF-2s that respond to hPTH-(1-34) with a burst of PLC-β1 and membrane-associated PKC activity but with no increase in AC activity (Mahon et al., 2002; Whitfield et al., 1998b, 1999a). Of course, this could explain why PTH cannot directly stimulate adenylyl cyclase in lymphocytes (Whitfield et al., 1971, 1999a).

This could also explain why N-terminally truncated peptides such as hPTH-(3-34), hPTH-(13-34), hPTH-(28-34) can significantly stimulate PKC in OK opossum kidney cells and ROS 17/2 rat osteosarcoma cells (the parents of the NHERF-2-less ROS 17/2.8 cells), but not in HKRK-B7 cells which have only a low level of NHERF-2 (Azarani et al., 1995a, 1995b; LiChong et al., 1998; Whitfield et al., 2001; M.J. Mahon, personal communication to F.R. Bringhurst).

The take-home message from all of this is that the PTHR1 is basically an adenylyl cyclase activator, a cyclic AMP pulse generator that must be plugged into a NHERF-2 circuit board if it is to activate PLC-β1 and its downstream effectors such as PKCs. Put in another way, PTHR1s by themselves are adenylyl cyclase stimulators, but they are PLC/PKC stimulators only when plugged into a NHERF-2 cluster. Moreover, there is also a revolutionary postscript to this take-home message—N-terminally truncated PTHs can still stimulate PLC/PKC, though not adenylyl cyclase, if the PTHR1 is connected to a NHERF-2 transducosome, but they will not signal at all in cells without a NHERF transducosome.

fig. 16

A model for this action might be what happens when β₂-AR receptor sticks to NHERF's PDZ-I domain (Ahn et al., 1999). The activated β₂-AR receptor in this cluster causes the dimerization (pairing) and transactivation of the EGF receptor. The β₂-AR receptor also activates the c-Src tyrosine protein kinase, which causes the β2-AR receptor to cluster with the EGF receptor pair. This huge β₂-AR •[EGFR]₂ cluster is then put by a β-arrestin phosphoprotein into a tear drop-shaped clathrin-coated pit, the neck of which is grabbed and pinched off by a dynamin "pinchase" and the pinched-off "tear drop" vesicle is delivered into the cytoplasm (Smythies, 2002). But to get to the point–the active c-Src and the still-functioning pair of EGF receptors trigger the ERK1/2 cascade and its many effects (Maudsley et al., 2000; Friedman et al., 2002). So if PTHR1 attached to NHERF-2 can do the same things, this would be how hPTH-(1-34) could stimulate ER1/2 phosphorylation/activation.

The attachment of the PTHR1 to the NHERF-2 scaffold in some cells means that there might be yet another signal—an internal pH spike—when PTHR1 is activated as happens when the β2-AR is activated. NHERF-2 inhibits the Na⁺/H⁺ exchanger when the exchanger is sitting on NHERF's PDZ1 domain. But the activated β₂-AR knocks the exchanger off NHERF's PDZ-I (Barber and Ganz, 1992; Barber et al., 1992; Bockaert et al., 2002). The displaced and rejuvenated exchanger then causes a pH spike in the cell by pumping protons (H⁺s) out of the cell. So PTH might also stimulate the Na⁺/H⁺ exchanger, because when the activated PTHR1 receptor's tail attaches to PDZ-II , though this is not the PDZ-I to which β₂-AR attaches, it might also displace Na⁺/H⁺ exchanger from its inhibitiory attachment to NHERF (Voltz et al., 2001). Indeed, PTH fragments do trigger a PKA-mediated extracellular acidification response (ECAR; a squirt of protons out of the cell) in neonatal mouse calvaria and ROS 17/2 rat cells, but an apparently PKC-mediated ECAR in human SaOS-2 cells (Belinsky and Tashjian, 2000; Belinsky et al., 1999). But we don't know whether these different cells have different amounts of NHERFs. However, this is probably a "false trail" because proton-squirting by human SaOS-2 osteosarcoma cells cannot be the work of the Na⁺/H⁺ exchanger because it is not affected by the exchanger's specific inhibitors (Barrett et al., 1997).

fig. 16

fig. 16

fig. 16

fig. 16

fig. 16

But there could be even more! Maybe, just maybe, PTHR1 receptor homodimerizes (twins with another PTHR1) or far more important heterodimerizes (couples with different kinds of receptors). It has long been an article of faith that GPCR receptors like PTHR1 don't do either of these. They were believed to be solitary signalers that would never pair with each other or with any others, as do receptors such as the EGF receptor and its several kin. But there is now a growing number of examples of GPCR homodimers and heterodimers that profoundly affect signaling. For example, there is the adenosine A1 receptor—dopamine 1 receptor heterodimer, the dopamine D2R receptor—somatostatin 5 receptor heterodimer, and the opioid ? receptor —opioid δ receptor heterodimer (Filizola et al., 2002; Jordan et al., 2000, 2001; Marshall, 2001; S.P. Lee et al., 2000; Rocheville et al., 2000a, 2000b; Vazquez-Prado et al., 2003). And of course we must not forget the ability of β₂-adrenergic GPCR receptors to activate the entirely different EGF tyrosine kinase receptors by physically joining them in a heteromultimeric transducosome that produces active EGF dimers (Maudsley et al., 2000)! If PTHR1 should have the same unsuspected gregarious tendencies as other GPCRs, its pairings with other family members, along with clusters of various signalers on NHERFs, could produce dazzling displays of multi-track signaling fireworks, depending on the target cell's make up that could be ignited by PTH and, here is the important thing—the ligand of the other member of the couple.

In summary, what PTHR1, like other GPCR receptors, surrounded by a crowd of hangers-on and their widely ramifying actions, is telling us through the fog of ignorance is actually very clear—there is no such thing as standard PTH receptor signals! The kinds of membrane scaffolds and the signal transmitters and trannsducers on them determine what signals a PTHR1 (or, for that matter, any other receptor) gives its cell. But what determines the kind of G-protein the activated receptor couples with? There are several ways that a GPCR receptor like PTHR1 might couple to more than one kind of G-protein and activate different signal mechanisms. There is the sequential model in which the activated receptor initiates a signal cascade that in the process switches back and modifies the receptor to activate a different G-protein and start a second cascade (Schoneberg et al., 2002). As an example of this, phosphorylation of a β-adrenergic receptor by PKA can shift its coupling from Gs to Gi (Daaka et al., 1997; Zamah et al., 2002). Likewise, the activated prostacyclin receptor couples first to Gs, which through its Gsa subunit stimulates adenylyl cyclase, cyclic AMP production, and the cyclic AMP activates PKA, which then turns around and phosphorylates the receptor on its Ser³⁵⁷, which enables the receptor to couple to Gi and Gq and trigger other signal cascades (Lawler et al., 2001). Then there is the parallel model in which the receptor is in equilibrium with two possible activation states (Schoneberg et al., 2002). Thus, in one state PTHR1 without access to a NHERF-2 "chip" would couple to and activate Gs and the Gs would activate adenylyl cyclase, while activated PTHR1 attached to a NHERF-2 "chip" would activate GTP•Gi/o or GTP•Gq complexes, which would prevent Gs from activating adenylyl cyclase and at the same time stimulate PLC-β1 on the NHERF's PDZ-I domain with the βγ subunits libertated from the GTP•Gi/o or GTP•Gq/11 complexes. Perhaps PTHR1 behaves like the β₂-adrenergic receptor, which when activated unfurls a motif on its C-tail with which it binds to the appropriate PDZ domain on NHERF-2's tail and thus displaces and releases the Na⁺/H⁺ exchanger from its inhibitory linkage to NHERF (Schoneberg et al., 2002; Voltz et al., 2001). The relative responses of adenylyl cyclase and PLC-β1 to a PTH would depend on how many receptors the cell has and how many of these can attach to NHERF-2.

In other words, the adenylyl cyclase/PLC-β1/??? contents of the primary signal from PTHR1s in a cell and whether bone growth will be started by this signal will depend on the cell's PTHR1:NHERF-2 ratio.

fig. 16

fig. 17

But wait! As usual there is a problem. According to Errazahi et al. (2003), keratinocytes from newborn rats seem to make functional PTHR1s. However, there is something peculiar about them because they do not respond to bPTH-(1-34) as do other rat cells such as ROS 17/2 cells. Unfortunately, these authors also don't say whether PTHrP also stimulated adenylyl cyclase in their keratinocytes. Thus, rat keratinocytes may have a modified PTHR1 which just cannot activate adenylyl cyclase or perhaps is prevented from doing so by associating with NHERF-2 (if keratinocytes have NHERF-2). So stay tuned to the so far slowly unfolding saga of keratinocytes and their adenylyl cyclase-shunning PTH/PTHrP receptor.

Of course, we must not ignore other members of the PTH receptor clan—yes, indeed, there are others! There is PTH2, which, unlike PTHRK, is a potent adenylyl cyclase stimulator designed to be activated by the PTH-related, 39-residue tuberoinfundibular peptide, TIP39, and is expressed most abundantly in the nervous system and to a lesser extent in lung, the pancreatic islets' somatostatin-producing D-cells and placenta (Usdin et al., 2002). However, the human PTHR2, but not the rat PTHR2, can also be strongly activated by PTH but not by PTHrP, while TIP39 binds to, but doesn't activate, PTHR1 (Usdin et al., 2002). Then there is the zebrafish's zPTHR3, which shares 67% of its amino acid sequence with hPTHR1 but has a 50-fold greater affinity for hPTHrP than hPTH (Rubin and Jüppner, 1999). Like zPTHR1, zPTHR3 can activate adenylyl cyclase, but unlike zPTHR1, it cannot stimulate PLC (i.e., it cannot stimulate IP3 accumulation) (Rubin and Jüppner, 1999). Then there is a mysterious receptor in the brain (specifically the supraoptic nucleus) that is activated by PTHrP-(1-34) but not by hPTH-(1-34) or hPTHrP-(7-34) (Yamamoto et al., 1998). The signals from this receptor stimulate the expression and secretion of arginine vasopressin by the supraoptic nucleus cells (Yamamoto et al., 1998).

Now enough is enough! Let's see what all of this receptor lore means to bone making.

What Signals Switch on the Bone-Making Machine?

fig. 16

fig. 8

If it is cyclic AMP, bursts of cyclic AMP-stimulated PKA activity and the direct targets of cyclic AMP-activated GEFs (guanine nucleotide exchange factors) such as Epacs 1 and 2 (de Rooij et al., 2000) that trigger bone growth, and if a PTH fragment could be made that needs only to activate adenylyl cyclase to stimulate bone growth, it could have potentially fewer side effects resulting from PKCs' activity in the many different cells expressing PTHR1 receptors. Such a fragment could be made if the multi-signaling hPTH-(1-34) could be cut into smaller, still partly functional, pieces each of which stimulates either one or the other signal mechanism. And this we seemed to have been done in the early '90s when PTHR1 signaling seemed to be so direct and simple without the complications of NHERFs and receptor heterodimers.

The potently osteogenic (in OVXed rats and cynomolgous monkeys) Ostabolin™ (hPTH-(1-31)NH₂) and Ostabolin C™ ([Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂) are as effective adenylyl cyclase stimulators as hPTH-(1-34)NH₂ , but unlike hPTH-(1-34)NH₂, they do not stimulate PKC activity in cultured nontransformed murine MC3T3-E1 calvarial (skull bones) preosteoblasts, rat ROS 17/2, osteoblast-like osteosarcoma cells, UMR 106-01 rat osteoblast-like osteosarcoma cells, as well as primary rat osteoblasts (Barbier et al., 1997; Jouishomme et al., 1992, 1994; Ryder and Duncan, 2001; Singh et al., 1999; Swarthout et al., 2000; Whitfield et al., 1996b, 1999a). Cultured human fetal osteoblasts (hFOB cells) appear to respond to hPTH-(1-31)NH₂ in the same way. In these human osteoblasts, hPTH-(1-31)NH₂ does not stimulate the PKC-dependent expression of TGF-β1 (and presumably not PKC activity) in these human fetal osteoblasts, although it stimulates the cyclic AMP-dependent expression of TGF-β2 as strongly as hPTH-(1-34) (Wu and Kumar, 2000).

As might be expected from what the PTHR1 receptor has been trying to tell us, whether such C-terminally truncated fragments can also stimulate PKC activity depends on which cells' membranes the PTHR1 receptors find themselves. For example, hPTH-(1-31)NH₂ can increase membrane-associated PKC activity in pig kidney cells engineered to put a huge number (i.e., about 950,000) of human PTHR1 receptors on their surfaces and in normal human newborn foreskin fibroblasts, which have only a very few (in fact not detectable by competitive binding assay) human PTHR1 receptors (Whitfield et al., 2001). The failure of hPTH-(1-31)NH₂ (Ostabolin™) and [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin-C™) to increase membrane-associated PKC activity in ROS 17/2 osteoblast-like cells and fetal human osteoblasts means that while PTHR1 receptors on cultured osteoblasts can't activate PLC or PLD, they can activate one or both of these enzymes in densely human-receptored pig kidney cells and very sparsely receptored cultured human foreskin fibroblasts that are differently equipped with the necessary messengers, kinase isoforms, and NHERF scaffolds.

At this point, the reader may be muttering enough's enough with this horrendous signaling complexity. But there's even more! Even the once sort-of-simple and all-important adenylyl cyclase-cyclic AMP mechanism has become formidably complicated (Chin et al., 2002). There is a very large number of possible cell type-specific combinations of different types of PKA R (regulatory) and C (catalytic) subunits in or on the cell membrane or in or around the nucleus. There are cyclic AMP-binding G-protein-activating guanine nucleotide exchange-stimulating factors (GEFs) factors. There are ion channels and transcription factors, such as the very important NF-κB that directly bind and respond to cyclic AMP independently from PKA activity. And R subunits can function independently from the catalytic subunits. For example, the cyclic AMP•RIIβ complex can bind to the CREs (cyclic AMP-responsive elements) in the promoters (the transcription switch boxes) of some genes. Therefore, the cell's response to the cyclic AMP pulses triggered by activated PTHR1 receptors depends on the whether the cells have NHERF scaffolds, what kinds of cyclic AMP-binding targets are available, and what PKA R and catalytic subunit isoforms the cell has and where they are in the cell.

GPCRs such as PTHR1 do not directly activate receptor tyrosine kinase receptors (RTKs). But surprisingly they can indirectly transactivate RTK receptors such as the EGF receptor (Cole, 1999; Daub et al., 1997; Iwamoto and Mekada, 2000; Prenzel et al., 1999)! At first, it seemed that in these cases the EGF receptor was not activated by its EGF ligand, but it now appears that the activated GPCR stimulates a transmembrane metalloproteinase that cuts soluble sHB-EGF (a heparin-binding protein with EGF-like motifs) from membrane-anchored pro-HB-EGF (precursor of heparin-binding EGF) (Iwamoto and Mekada, 2000; Prenzel et al., 1999). The liberated sHB-EGF then activates EGF receptors on its own cell or adjacent cells. When the PTH/PTHR1-"transactivated" EGF receptors phosphorylate themselves, they become sticky flypaper-like or Velcro-like scaffolds for collecting and clustering the components of the Ras/Raf-1-mechanism that triggers the multipurpose MAP kinase cascade (Cole, 1999; Daub et al., 1997; Pearson et al., 2001). However, we don't know what the Ras/Raf/MAP kinase cascades triggered by PTHR1-transactivated RTKs, or whatever other receptor with which PTHR1 might have paired with, might contribute to osteogenesis. Nevertheless, the important message is that the signals triggered by a PTH could come from both PTHR1 receptors and any transactivatable RTKs the cell might be expressing.

Figure 18

The travels of a PTHR1 (shaded arrows) in a target (more...)

Figure 18

.

The travels of a PTHR1 (shaded arrows) in a target cell such as an osteoblast. (1) The receptor is activated by either endocrine, injected, or swallowed pill-borne PTH from the blood or by autocrine/paracrine PTHrP from the cell or its neighbors. The receptor fires signals into the cell, and the signals feed back to shut down the receptor. (2) PTHR1 and the attached PTH are attached to β-arrestin-2 scaffold (solid rectangle) containing a cluster of signalers (circles) and carried into the cell in a clathrin-coated endocytic vesicle where the PTH is destroyed and the receptor is reactivated. The reactivated PTHR1 is returned to the surface ready to go again and the empty β-arrestin-2 scaffold (A) is left behind in the cytoplasm. (3) The receptor might enter the nucleus because it has a NTS (nuclear transport sequence) and activate genes. (4) One of the signalers that had been activated when the β-arrestin-2 scaffold had grabbed PTHR1 is a JNK (c-Jun kinase—one of the circles brought in on the arrestin rectangle) which gets into the nucleus and stimulates gene transcription. (5) The PTHR1 gene transcript is translated into PTHR1 in the cytoplasm by the ribosomal translating machinery. The newborn PTHR1 may directly get into the nucleus instead of being plugged into the membrane. (6,7) Genes are stimulated by PTHR1 or JNK.

fig. 18

fig. 18

fig. 18

Figure 19

The dense tangle of transmitters, transcription factors, (more...)

Figure 19

.

The dense tangle of transmitters, transcription factors, transcripts and translations of the information encoded by the scores of genes activated by signals from PTH-activated PTHR1 receptors. The stream of signals from the receptors includes a burst of cyclic AMP-dependent protein kinase (PKA) which is needed to start the bone-building machinery. But the signal mix and the cascades it triggers depend on to what other receptors the PTHR1s are linked and to what signalers-loaded scaffolds, such as NHERFs -1 and -2, the PTHR1s are attached. The upshot of this is a seemingly ever-growing intricate web of responses hidden among which are the actual bone-building drivers. The challenge is to separate the driving "wheat" from irrelevant "chaff ".

fig. 19

fig. 18

In the late 1980s and early 1990s, we set out to find whether the osteogenic signal from PTHR1 receptors required the stimulation of adenylyl cyclase, membrane-associated PKCs, or both to start making bone. So we needed fragments that stimulate membrane-associated PKC but not adenylyl cyclase. Unlike the C-terminally cropped fragments, no N-terminally truncated PTH fragment can stimulate AC. However, as we learned above, the ability of a N-terminally truncated PTH to stimulate PKC activity also depends on the kind of cell expressing the PTHR1 receptors. For example, b(bovine)PTH-(3-34)NH₂, hPTH-(3-34)NH₂, hPTH-(13-34)OH, hPTH-(28-48)OH, hPTH-(8-84)OH, and even the tiny hPTH-(28-34)OH, hPTH-(29-32)OH can increase membrane-asociated PKC activity in cultured ROS17/2 osteoblastic cells, UMR-106 rat osteoblastic cells, human fetal osteoblasts, and human dermal fibroblasts, and hPTH-(8-84) can strongly stimulate membrane-associated PKC activity in spleen cells in rats, but they are at best only very weak stimulators of membrane-associated PKC activity in the pig cells with nearly a million human PTHR1 receptors (Fujimori et al., 1992; Jouishomme et al., 1992, 1994; LiChong et al., 1998; Neugebauer et al., 1995; Swarthout et al., 2001; Whitfield et al., 2001; Wu and Kumar, 2000). This of course could be due to the artificially engineered pig cells having only very little NHERF-2, which is required for PTHR1 to activate PLC and thus stimulate PKC activity (M. Mahon, personal communication to F.R. Bringhurst).

hPTH-(1-31)NH₂ (Ostabolin™), its lactam derivative, [Leu²⁷]cyclo[Glu²²-Lys²⁶]hPTH-(1-31)NH₂ (Ostabolin-C™ ), as well as [Leu²⁷]cyclo[Glu²²-Lys²⁶]hPTH-(1-28)NH₂, which stimulate AC but not membrane-associated PKC activity in cultured rat and human fetal osteoblasts, strongly stimulate bone growth in OVXed rats (Rixon et al., 1994; Whitfield et al., 1996b, 1997b, 1997c, 1997d, 1998b, 1999c, 2000a, 2000b, 2000c). On the other hand, the N-terminally disabled 1-desamino-hPTH-(1-34) and the N-terminally truncated hPTH-(28-48) and hPTH-(8-84) that strongly stimulate membrane-associated PKC activity, but not adenylyl cyclase, in cultured osteoblasts and in rats do not stimulate bone growth in the OVXed rats (Rixon et al., 1994; Strein, 1994). Therefore, as things stand at the moment, activating adenylyl cyclase and the cascades it triggers seems to be all a PTH needs to do to start, but maybe not subsequently drive, bone formation (at least in rats). This is supported by the ability of selective cyclic AMP-elevating, cyclic nucleotide phosphodiesterase inhibitors (Pentoxyfylline and Rolipram) to stimulate bone growth in normal male mice (Kinoshita et al., 2000). But (that caveating word again!) Locklin et al. (2003) have found that hPTH-(3-34), which cannot stimulate adenylyl cyclase or bone formation, nevertheless can stimulate the expression of that master conductor of osteoblastic genesæ bfa1/Runx-2æ in cultured C57BL/6 mouse marrow cells as effectively as the cyclase-stimulating, osteogenic hPTH-(1-34) and hPTH-(1-31).

However, AC stimulation might not be enough to start osteogenesis in ovariectomized rats and mice. MPTH-(1-21) (i.e., [Ala,^3,12Nle,⁸Gln,¹⁰ Har,¹¹Trp,¹⁴Arg,¹⁹Tyr²¹]rPTH-(1-21)) can bind to PTHR1 and strongly stimulate adenylyl cyclase, but it cannot stimulate bone formation in OVXed mice and rats and was unexpectedly much less potent in affecting alkaline phosphatase activity in human HOB-03-C5 and Saos bone cells (Murrills et al., 2002). Thus, for some reason the cyclic AMP signals from PTHR1 activated by this short, AC-stimulating PTH can't start osteogenesis. But wait! Did the fragment reach the bones in an active state?Rixon et al. (1994) had to answer the same question when they found that hPTH-(3-34) and hPTH-(13-34) did not stimulate bone formation in OVXed rats. Was this because they couldn't stimulate bone formation or they couldn't reach the bone in an active state? In this case, both turned out to be the case. The inability of Rixon et al.'s PTHs to reach their bone targets in the rat was shown by their abilities to strongly stimulate membrane-associated PKC activity, as did hPTH-(1-34) when added to spleen lymphocytes suspensions just a few minutes after they were taken from the rat, but they did not increase membrane-associated PKC activity in spleen lymphocytes still in the rat as did hPTH-(1-34). So one could ask Murrills and colleagues whether their little peptide actually reached bone cells?

The failure of small (less than 34 amino acids) N-terminally truncated PTH fragments to stimulate PKC activity in spleen cells in the living, breathing rat while strongly stimulating PKC activity in freshly isolated spleen cells suggests the existence of something that sees and trashes such small fragments but ignores a larger N-terminally truncated fragment such as rhPTH-(8-84) or a 34-amino acid fragment such as 1-desamino-hPTH-(1-34), with its quasi-normal N-terminus. This nemesis of small "nose-less" fragments might be the 230-kDa protein in rat and human serum that Kukreja et al. (1994) found to inactivate bPTH-(7-34)NH₂ and hPTHrP-(7-34)NH₂, but not hPTHrP-(1-34). As we learned above, the unselected rhPTH-(8-84) would be able to stimulate osteocytes having both the CPTHR (Bringhurst, 2002; Divieti et al., 2001, 2002) and in my experience (Rixon et al., 1994) the PLC-β1/PKC arm of the PTHR1 response. However, one must be careful of this because Bringhurst (2002) has said that hPTH-(7-84) does not bind to PTHR1 receptors. If this were true, it would mean that spleen lymphocytes responded by either the CPTHR or another receptor, as suggested by Whitfield et al. (1999a). Remember (if by now you need reminding) that nothing is simple with PTHs and their receptors.

fig. 1

fig. 1

As we shall see further on, this difference between the effects of hPTH-(1-34) and the two C-terminally truncated fragments in humans and rodents just might be due to lower doses of hPTH-(1-31)NH² (Ostabolin™) and Ostabolin-C™ not having hPTH-(1-34)'s ability to stimulate osteoblast progenitors to express the osteoclast differentiation stimulator RANKL and reduce the expression of the osteoclast-inhibiting OPG (Hofbauer et al., 2000; Suda et al., 1999). Indeed maybe Ostabolin™ and Ostabolin C™ increase the OPG/RANKL expression ratio to a level that actually inhibits OVX-induced osteoclast activation, as suggested by the findings of Jolette et al. (2003) with monkeys.

Before going any further, I must repeat that we do not have the full measure of the signals that the PTHs send into their target bone-making cells in the various provinces of the Real Bone World. For example, what NHERFs do the cells in the various stages of osteoblastic development in a PTH-treated bone have, and what are plugged into their NHERFs that can be tweaked by activated PTHR1s? In other words what do the PTHR1-bearing cells have that can be targeted by the primary, osteogenesis-kick-starting cyclic AMP signal?

PTHs Speak to Bones with Forked Tongues—One Prong Says Make It!

Generalities

Obviously, we must find out how the adenylyl cyclase-stimulating (i.e., the N-terminally intact) PTHs and the signals from their PTHR1 receptor stimulate bone growth in order to make even better anabolic drugs for the future. But this is not easy. The first problem is trying to identify the key players in the horde of genes and other things that can respond directly to PTHR1's promiscuous or "shotgun" signaling. Indeed, the PTHR1 signals with their many target genes, transcription factors, enzymes, enzyme regulators, cytoskeletal components and motors set off a barrage of bone-relevant as well as bone-irrelevant reactions. Within this tangle of events are the few that actually lead to bone formation. The challenge is to find the bone-building needles in this tangled haystack of responses (fig. 19).

As a tiny tantalizing taste of things to come, Qiu et al. (2003), starting by characterizing the promoters of 8 "training" genes known to be regulated by PTHs, have identified 31 transcription factors and 80 genes that could be affected directly by PTHs. And Onyia et al. (2001a) have found that 158 genes associated with bone growth are stimulated in rat metaphyseal bone by PTH. Added to these are those in the differently equipped and scaffolded members of the osteoblastic lineage in different parts of the bones that belong to the vast signaling "syncytium" stretching from the cilium-waving strain-sensing osteocytes locked in their lacunae, to their dendrites snaking through canaliculi, to the retired osteoblasts lining the bone surfaces, and from there to the osteoblastic and osteoclastic progenitors and precursors in the bone marrow (Marotti, 1996). We see osteoblastic cells signaling and, amazingly, remembering strain frequencies using the same machinery as neurons in the central nervous system (Skerry and Taylor, 2001; Spencer et al., 2004; Turner et al., 2002). But we are being swamped by an avalanche of reports of relevant, but undoubtedly many more irrelevant, gene transactivators, hundreds of gene expressions, dozens of enzymes and mechanisms in osteoblastic cells not working in their native bones but that have been ripped from their native signaling "syncytium" and planted on plastic or in cells of established or semi-established lines in which cell cycle controls, PTHrP and PTHR1 expressions, and the maturation program have been more or less disconnected. Indeed, such dysregulated cells are like crazy computers simultaneously running many normally mutually exclusive programs. Therefore, what follows can only be the vague outline of the mechanism(s) by which the PTHs cause bones to grow as seen from inside the swirling stream of events.

Figure 20

The osteoblastic maturation story and how a subcutaneously (more...)

Figure 20

.

The osteoblastic maturation story and how a subcutaneously injected PTH such as hPTH-(1-34) (Forteo™) or one of the Ostabolin™ family peptides such as hPTH-(1-31)NH₂ (Ostabolin™) swallowed in a pill might affect it. The blue (darker) cells are the proliferatively competent mesenchymal stem cell and osteoprogenitors, and the red (slightly lighter) cells are the proliferatively shutdown cells in the more advanced stages of osteoblastic differentiation. The white ovals on the cells are PTHR1 receptors, and their number indicates the relative densities which peak as the cells enter the preosteoblast and osteoblast stages and then fall if the cells are lucky enough to be chosen to be strain-sensing osteocytes or lining cells. Abbreviations: AR, amphiregulin; FGF-2, fiboblast growth factor-2, one of the missionary growth factors; IGF-I, insulin-like growth factor-I, another missionary growth factor; ImmPro, immature osteoprogenitor; MatPro, mature osteoprogenitor; PreOst, preosteoblast; PTHrPs, cytokines cut out of the PTHrP chain such as PTHrP-(107-111), PTHrP-(107-139), and PTHrP-(8-11) that is virtually identical to endothelin 1's residues 6-9 and thus would trigger osteogenic signals from endothelin A receptors; MSC, mesenchymal stem cell; TGF-βs, tumor growth factor-βs, still more missionary growth factors. The green lines are the effects of the PTH. But we must not forget that PTHrP is produced by the differentiating red cells and stimulates their and their neighbors' PTHR1 receptors and is thus one of the programmed drivers of osteoblastic maturation. A color version of this figure is available online at http://www.Eurekah.com.

When an adenylyl cyclase-stimulating PTH is injected subcutaneously, or, as may soon be happening, swallowed in a pill (fig. 20), it first meets and is grabbed by trabecular bone cells in the bone marrow and only later and probably more or less depleted does it reach cells in the Haversian canals and the lacunocanalicular network (see Chapter 2i; Knothe-Tate, 2001). The target cells in the various regions will have their PTHR1s in different superclusters, and some cells will have unconventional PTH/PTHrP receptors. But how many target cells with PTHR1 receptors are available to kindle an osteogenic response? This depends on the age of the rat or human. In a growing rat or human, there are packs of osteoblasts and osteoclasts working independently to build and shape the growing bones. But when growth stops in humans (and other big-boned animals), there are few or no independently working osteoblasts. The only osteoblasts will be in BMUs. So there won't be many PTHR1-bearing first responders in older humans. But if, for example, a long bone is broken and an implant is screwed into it, BMUs will be activated and fan out from the screwhole in all directions along the diaphysis on their several-millimeter journeys and affect the whole length of the bone for months afterwards (R.B. Martin et al., 1998). In other words, these BMUs will generate PTH targets—PTHR1-bearing osteoblasts—and thus enhance the responsivenss to PTH.

The potently osteogenic hPTH-(1-34) (Forteo™), but possibly not so much other equally potent bone-building PTHs such as hPTH-(1-31)NH₂ (Ostabolin™) and [Leu²⁷]cyclo[Glu²²-Lys²⁶]hPTH-(1-31)NH₂ (Ostabolin-C™), increases osteoclast generation like an estrogen crash and hence the number of holes being dug in the bone at any given moment (i.e., the remodeling space). It stimulates (probably via cyclic AMP) the proliferation of the pluripotent, spleen colony-forming, CFU-S, stem cells and other members of the rat hematopoietic cell populations, which need the microenvironment provided by docking on immature osteoblastic stromal cells (Byron, 1977; Gallien-Lartigue and Carrez, 1974; Perris et al., 1971). And Calvi et al. (2003) have much more recently confirmed that either hPTH-(1-34) or a permanently switched-on mutant PTHR1 receptor, which does not need PTH to activate, it, can enlarge a mouse's hemopoietic stem pool. It seems that PTHR1 receptor signals cause the osteoblastic bone-lining cells to make large amounts of surface Jagged 1, which in turn binds to Notch receptors on adjacent hemopoietic stem cells attached by N-cadherin to the osteoblastic lining cells (Calvi et al., 2004; Ohishi et al., 2003; Schweisguth, 2004; Zhu and Emerson, 2004). The resulting Notch signals in the stem cells increase the pool of hemopoietic stem cells. When cells from the enlarged stem cell pool are mobilized and leave their osteoblastic niche, they can generate increased numbers of "rookie"osteoclast recruits, which carry on the stem cell tradition of making PTHR1 mRNA and receptors, but hold these in their cytoplasm and around their nuclei and maybe even in their nuclei without putting them out onto their surfaces as they mature (Faucheux et al., 2002; Kanatami et al., 1998; Langub et al., 2001; Watson et al., 2002). The PTH-induced burst of cyclic AMP formation also stimulates osteoclast formation by causing some immature marrow osteoblastic stromal cells with PTHR1 receptors to express and deploy on their surfaces the osteoclastogenesis-stimulating RANKL (Atkins et al., 2003; Fu et al., 2001; Galvin et al., 2001; Kanazawa et al., 2000; K. Lee et al., 1994; Stilgren et al., 2001). The PTH also induces cultured osteoblasts and possibly some immature osteoblastic stromal cells to make autocrine (autostimulating)/paracrine (neighbor-stimulating) IL-6 which causes the immature osteoblastic stromal cells to make RANKL (Greenfield et al., 1999; Tamura et al., 1993) (fig. 20).

The Target Cells

The basic requirement for a PTH to stimulate bone growth is that there must be enough PTHR1-bearing cells with which to start building up a crew of active PTHR1 receptor-bearing, osteoblasts. The response to a PTH will initially depend on the size of this receptored team. As stated above, in the bones of human children or rat pups, the mature osteoblast teams will be large and located in osteoclast-free, bone-modeling clusters, but in an adult human, it should be much smaller because most of the cells will belong to microcrack-filling BMUs. But there are also PTH-responsive, BMU-independent retired osteoblasts in mature bones which have fewer but still effective PTHR1 receptors (fig. 20). These are the lining cells, covering the walls of the osteons' central canals as well as the endosteal and trabecular surfaces. The PTH will increase the number of the mature bone-making osteoblasts by increasing their generation, their lifespan, or persuading the retired lining cells to come out of retirement (fig. 20). A mature osteoblast is at the end of a long line of cells stretching back to an ALP (alkaline phosphatase)-expressing "grandmother" cell in the bone marrow stroma or the periosteum and ultimately a mesenchymally-derived "great-great-great-grandmother" stem cell, which had unlimited self-renewing potential and the options to found lines of adipocytes, chondrocytes, fibroblasts, or myoblasts (Aubin, 2000, 2001; Aubin and Triffitt, 2002). As is the case for all stem cells (Whitfield and Chakravarthy, 2001), these mesenchymally-derived stem cells normally proliferate only when needed to maintain a pool of extensively proliferating osteoprogenitors (i.e, transit amplifying cells) for growing bone in children and and removing muscle-inflicted microcracks in adults (fig. 20).

But let's look more closely at the PTH targets in the cortical bone of a mature skeleton. As we learned in the first chapter, a pack of continuously proliferating spindle-shaped immature osteoprogenitor cells (i.e., juvenile osteoblasts) capable of expressing RANKL chases the BMU osteoclasts as the osteoclasts are pushing the PTHR1-bearing lining cells aside and tunnelling through the exposed bone, and getting rid of, a patch of microcracked bone. The osteoprogenitor cells put a layer of mature osteoprogenitors around the tunnel wall. These mature osteoprogenitors switch off and dismantle their cell-cycle machinery and become PTHR1-bearing preosteoblasts that mature into "smart" receptor-bearing osteoblasts with primary cilia to measure, and respond to fluctuations in, the components and movements of the extracellular fluid (Tonna and Lampen, 1972) and start laying down the first of 7 or 8 pairs of lamellae around the tunnel wall to fill it in but they save space for the blood vessel's channel which makes a new osteonal central canal. These nearly mature and mature osteoblasts plus the surrounding lining cells are the targets of injected PTH or pill-delivered PTH diffusing from the blood vessel in the center of the tunnel (fig. 20). Presumably, the several growth factors from the PTH-stimulated osteoblasts will "contaminate or dope" the new bone, diffuse down through the stack of cells and stimulate the proliferation of the receptorless outer immature osteoprogenitors, and get into the blood to attract more progenitors to the construction site. PTH-induced factors diffusing out of the stimulated lining cells into the more extensive bone marrow and marrow sinusoids on the endosteal and trabecular surfaces should have greater overall impact because the products of the PTH-stimulated, PTHR1-expressing osteoblastic cells and bone lining cells on endosteal and trabecular surfaces will immediately reach and stimulate immature osteoprogenitor cells in the adjacent marrow which have not yet started deploying PTHR1 receptors. The fact that endosteal surfaces but not periosteal surfaces have lining cells with PTHR1 receptors (though fewer than when they were maximally receptored osteoblasts) explains why PTH stimulates endosteal but not periosteal bone growth as does growth hormone (Andreassen and Oxlund, 2000).

Autocrine PTHrP and Emerging PTHRs1—Proliferation Suppressors and Drivers of Normal Osteoblast Differentiation

And now let's look at the drivers of osteoblast differentiation. A menagerie of factors including hedgehogs (Indian and/or Sonic), Bapx 1, Msx-2 and signals from FGFR1 receptors promote a BMP-2 → BMP1B•II heterodimeric(2-component)receptor → phospho-Smad-1/5 second messenger-mediated surge of the type 2 Cbfa1/Runx-2 expression to close off the adipocyte option and start a clone of extensively proliferating "transit amplifying cells" with limited self-renewing abilitiy—the immature osteoprogenitors (Aubin, 2000, 2001; Aubin and Triffitt, 2002; Banerjee et al., 2001; Bidwell et al., 2001; Chen et al., 1998; Chung et al., 2001; Ducy, 2000; Ju et al., 2000; Karsenty, 2000; Lian and Stein, 2003; Spinella-Jaegle et al., 2001) (fig. 8 and fig. 20). It is essential to know that for the BMPR-triggered surge of phosphoSmads to switch target genes on or off, they must interact with Cbfa1/Runx-2 on the genes' promoters (Canalis et al., 2003). A premature stimulation of the normally cycle-stopping mature osteoblast genes by the type 2 Cbfa1/Runx-2 is prevented and the progenitors kept proliferating by the "homeobox" protein, Msx2, which is expressed during the skeletal growth phase and binds to and shuts down type 2 Cbfa1/Runx-2 (Dodig et al., 1999; Lian et al., 1998; Ryoo et al., 1997; Shirakabe et al., 2001). Indeed, MC3T3 preosteoblasts throw their Cbfa1/Runx-2 into the proteasome garburettor barrel before they start replicating DNA and forced hyperexpression of the Cbfa1/Runx-2 protein reduces proliferation (Galindo et al., 2003). The level of Msx-2 might be determined by the expression of antisense Msx1-AS which downregulates the constitutively expressed Msx1 and maybe also Msx2 (Blin-Wakkach et al., 2001). Overexpression of Msx-2 would keep the osteoprogenitors proliferating to maintain an optimally sized precursor pool and prevent type 2 Cbfa1/Runx-2 from driving their maturation into post-proliferative preosteobolasts and osteoblasts whereas disabling the gene would reduce osteoprogenitor proliferation and cause premature maturation (Dodig et al., 1999). But at a critical point the signals from collagen-stimulated integrins and other factors cause the appearance of another "homeobox protein", Dlx5, which downregulates Msx1 and pulls Msx2 off type 2 Cbfa1/Runx-2 which can then stimulate several key genes with the type 2 Cbfa1/Runx-2-binding, osteoblast-specifc element 2 (OSE2; the AACCACA DNA nucleotide sequence) in their promoter-signal boxes that include the osteoblastic-specific zinc finger protein transcription factor Osterix without which there can be no osteoblast maturation or bone formation (Blin-Wakkach et al., 2001; Ducy and Karsenty, 1995; Katagiri and Takahashi, 2002; Marijanovic et al., 2002; Nakashima et al., 2002; Zhang et al., 2002a). But Dlx5 also inhibits the expression of the osteocalcin gene by binding to a single "homeobox"-binding site in the gene's promoter-switch box (Ryoo et al., 1997). Why this seeming anti-differentiation action? As in all differentiation programs, timing of appearance is all-important. So this is one of two strategies (we will meet the other one soon) to prevent type 2 Cbfa1/Runx-2 from prematurely stimulating the appearance of osteocalcin, which is expressed later on in the ostoblast maturation process to switch off and prevent excessive mineralzation (Ducy et al., 1996).

At this point, we must not forget the surprising likelihood of signals from the osteoblastic cells' neuron-like NMDA glutamate receptors being involved in driving the pivotal expression of Cbfa1/Runx-2 (Hinoi et al., 2003). This possibly hugely important role of glutamate receptors has been hidden by the large amounts of glutamate that saturate the glutamate receptors of osteoblastic cells in the standard cell culture media (Hinoi et al., 2003, 2004). If we can somehow peel away the other functions of this multifunctional amino acid, we can expect to see another part of the machinery that starts and drives osteoblast maturation.

One of the type 2 Cbfa1/Runx-2's major targets is the PTHR1 receptor gene, specifically its P3 promoter (it has three such switch boxes) to which the protein binds as effectively as it does to the osteocalcin promoter-switch box (Karpieren et al., 2000). The now mature osteoprogenitors start expressing PTHrP and putting PTHR1 receptors out on their surfaces (Aubin, 2000, 2001; Aubin and Triffitt, 2002) (fig. 8 and fig. 20). The appearance of PTHrP and PTHR1 receptors coincides with the secretion of autocrine/paracrine PTHrP, the PTH-(1-34)-like N-terminus ("nose") of which both activates and stimulates the further expression of PTHR1 receptors. The signals from the PTHR1 receptors coincide with the shut down of the proliferative machinery, and start the procession of mature osteoblast-specific gene expressions (Aubin, 1998, 2000, 2001; Aubin and Triffitt, 2002 (fig. 8 and fig. 20).

To become a mature osteoblast, a progenitor may use its PTHrP and PTHR1s to make the promoter-switch boxes of key osteoblast-specific genes accessible to transcription factors by bending the DNA to bring originally separated regions together to make functional switch boxes (Alvarez et al., 1998). This is done with "architectural" transcription factors, such as NmP4, which we met in Chapter 2i. They can bind to AT-rich regions of the minor groove of DNA and twist and/or bend the DNA to make functional gene promoters (switch boxes), for example by binding to an AT-rich region (between nucleotides 1971-1574 and -1555) which is upstream from the coding region of the collagen-I gene (Alvarez et al., 1998; Feister et al., 2000). Indeed PTH, and therefore autocrine PTHrP, causes NmP4 to get into the nucleus where it bends and twists the collagen-I gene's promoter-switch box into a functional switch for the transcribing machinery (Alvarez et al., 1998).

Since signals from activated PTHR1 receptors cause substantial cytoskeletal shifts as well as potentiate cells' responses to strain, they may use the same c-Src-triggered, Cas-Crk-mediated, nucleus-seeking mechanosomes that we talked about in Chapter 2i that are assembled at the surface of strained osteocytes and then go to the nucleus to stimulate osteoblast-specific genes (Carvalho et al., 1994; Eagan et al., 1991; Goligorsky et al., 1986; Krempien et al., 1978; Lomri and Marie, 1990; Pavalko et al., 2003). In other words, PTHR1 signaling driven by autocrine PTHrP in preosteoblasts mimics strain deformation by sending architectural transcription factors such as NmP4 into the nucleus to bend and twist the promoters of osteoblast-specific genes into functional configurations that can receive and be activated by classical transcription factors.

A remarkable example of the ability of the PTH-triggered "shotgun" signals (fig. 19) to switch on some of the same genes they trigger in osteoblasts even in completely inappropriate cells has been reported by Hefferan et al. (2002). They found that blood leukocytes from patients given 400 IU (activity units) of hPTH-(1-34)/day were expressing typical osteoblast genes such as those for alkaline phosphatase, BMPs-2,-4,-6, collagen I, and IGF-I. This is a striking example of the leukocyte genome being turned on by PTH exactly like the peptide turns on the osteoblast's genome, but, of course, without being able to establish a functioning, osteoblast proteome in such an alien cell body. It also confirms my Group's now ancient, very pioneering, but equally very premature demonstrations of the ability of PTH to control and stimulate hemopoiesis and thymic lymphocyte generation (e.g., Perris et al., 1967, 1971;Whitfield et al., 1973, 1998b).

Zuscik et al. (2002) have shown that the expression of PTHrP's gene is negatively affected by the nuclear Ca²⁺ content. So the expression of the gene starting in preosteoblasts requires that the nuclear Ca²⁺ level be low enough. Unfortunately, we don't know how the nuclear Ca²⁺ level changes during the the progression from immature osteoprogenitor to osteoblast. But maybe when something such as APRO (TIS21), whom I will introduce further on, finally stops a mature osteoprogenitor proliferating, the Ca²⁺ transients which drive transit through the key stages of the cell cycle would stop, the nuclear Ca²⁺ level would drop, and the expression of the PTHrP gene would be unleashed. But it also seems that the secretion of the emerging whole PTHrP molecule and its cleaved components would be driven by the very high Ca²⁺ concentration (e.g., 40 mM!) that hits the preosteoblasts and osteoblasts when they start working in the osteoclast tunnels and trenches. This Ca²⁺ would activate the osteoblastic cells' Ca²⁺-sensing receptors, which have their C-tails tied along with MAPK to filamin-A in signaling clusters in caveolar "message centers" (Awata et al., 2001; Hjälm et al., 2001). The signals from these clusters stimulate the PKCs/MEK1,2/p38MAP kinases-mediated synthesis of PTHrP (MacLeod et al., 2003; Sanders et al., 2001; Tfelt-Hansen et al., 2002). Besides driving osteoblast differentiation, signals from PTHR1 receptors activated by the secreted PTHrP would protect the cells from being killed by the large amount of phosphate that accompanies the Ca²⁺ (Adams et al., 2001; Allen et al., 2002; Meleti et al., 2000).

In Chapter 2i, we learned that strain turns on the PTHrP gene, possibly by opening TREK stretch-sensitive K⁺ channels in osteoblastic cells (X. Chen et al., 2003). Therefore, the osteogenic PTHrP (Whitfield et al., 1997d) is very likely a principal mediator of loading-induced bone growth and maintenance. In other words, straining osteoblastic cells causes them to make and secrete PTHrP, much like pulling a skunk's tail triggers a spray of pungent perfume.

Indeed, the autocrine stimulation of preosteoblasts by their own PTHR1 signaling and secreted PTHrP is certainly a critical factor driving a progenitor along the road to functional osteoblasthood. Knocking out the PTHrP gene in mice impairs osteoblast development and bone formation (Miao et al., 2002). Schiller et al. (1999) have given an example of the coincidence of the emergence of PTHR1 receptors with proliferative shutdown, which was followed about 5 hours later by the initiation of bone nodule formation in growing cultures of MC3T3-E1 murine calvarial (skull bone) preosteoblasts. The cyclic AMP signals from the PTHR1 receptors activated by the emerging autocrine/paracrine PTHrP shut down the cell cycle machinery and further increase the expression of PTHR1 (Lu et al., 2002). They also push Cbfa1/Runx-2 into the cell nucleus (Fujita et al., 1999, 2001a) and there into the arms of pRb, in this case a cotransactivator. After being pushed into the nucleus by the PTHR1 signals, Cbfa1/Runx 2 doesn't just wander aimlessly around looking for pRb and client genes. It has a nuclear matrix-targeting sequence (NMTS) in its C-tail (amino acids 397-434) that cause it to seek out specific nuclear matrix sites, where it can form transcription complexes with suitably addressed osteoblast genes (Zaidi et al., 2001).The Cbfa1/Runx-2•pRb complex then promotes the effectiveness of the antiproliferative, cycle-blocking p21^CIP and p27^{KIP 1} proteins and stimulates the parade of gene expressions which ends with a mature fully mineralized matrix, loaded with various growth factors from the earlier stages (Lian et al., 1998; Stein et al., 1996; Thomas et al., 2001). The absence of the functional proliferation-restraining hypophosphorylated pRb that would otherwise work with Cbfa1/Runx-2 cotransactivator to open the mature post-proliferative osteoblast gene file explains the 500-times higher incidence of osteosarcoma associated with the disabled pRb in retinoblastoma patients (Gurney et al., 1995; Thomas et al., 2001).

Figure 21

The pivotal moment in the initiation of osteoblast (more...)

Figure 21

.

The pivotal moment in the initiation of osteoblast differentiation. The key event in this scheme is the stimulation of the cyclic AMP/PKA system by signals from PTHR1 receptors activated by autocrine (autostimulating)/paracrine (neighbor-stimulating) factors. The cyclic AMP system stops proliferation by stimulating APRO1, which in turn prevents the activation of the major cell cycle protein kinases, cyclin D1•PK and cyclin E•PK. The cylic AMP signaling also turns on Cbfa1/Runx-2 expression, which in turn sets up a self-reinforcing BMP2↔Cbfa1/Runx-2 cycle and stimulates the expression of the osteoblast-specific gene for osterix, which is absolutely essential for osteoblast differentiation.

This pivotal PTH/PTHrP- and Cbfa1/Runx-2-driven step in the progression of an osteoprogenitor cell such as MC3T3-E1 toward bone-making osteoblast includes a brief expression of the APRO1 (Antiproliferation1 or TIS21(mouse)/BTG2 (human) or PC3(rat)) gene (Matsuda et al., 2001; Raouf and Seth, 2002; Tirone, 2001) (fig. 8,fig. 20, and fig. 21), the product of which blocks the G₁ build-up to chromosome replication (Matsuda et al., 2001; Guardavaccaro et al., 2000; Tirone, 2001). The highly labile APRO1 (TIS21) protein, together with the Cbfa1/Runx-2•pRb complex's enhancement of the effectiveness of the p21^CIP and p27^Kip cell cycle blockers, prevents the transcription of the cyclin D1 gene and also directly interacts with, and inhibits, the CDK4•cyclin D1 protein kinase (Guardavaccaro et al., 2000; Matsuda et al., 2001; Tirone, 2001). Thus, there is no CDK4•cyclin D1 to start the G₁ build up to chromosome replication by hyperphosphorylating and breaking the pRb lock on the cell cycle machinery. Unphosphorylated, pRB locks the cycle machinery by shutting down the activator of cell cycle genes—the E2F protein—and joins with Cbfa1/Runx-2 to drive post-proliferative osteoblast gene expressions (Harbour and Dean, 2000; Whitfield and Chakravarthy, 2001). If there is no APRO1, pRb can be hyperphosphorylated by CDK4•cyclin D1, which makes it let go of E2F, which then collaborates with CDK4•cyclin D1's successor, the CDK2•cyclin E protein kinase, to turn on the genes for key chromosome-replicating proteins (Harbour and Dean, 2000; Matsuda et al., 2001; Tirone, 2001; Whitfield and Chakravarthy, 2001). However, APRO1 is doubly effective—it can also reduce the expression of cyclin E and consequently the CDK2•cyclin E protein kinase (Tirone, 2001).

This cycle-stopping expression of APRO1 (TIS21) is by no means unique to preosteoblasts. Still in the context of bone, it is also involved in osteoclast differentiation, where it is induced by RANKL-induced RANK receptor signaling to stop osteoclast precursors proliferating as a prelude to terminal maturation (S.W. Lee et al., 2002). But it also seems to be part of a widespread mechanism for shifting a proliferating precursor cell's progeny into a post-proliferative stage of development. Thus, for example, a mouse or rat neuroectodermal cell in the ventricular zone of the neural tube that has reached the point of generating a post-proliferative neuron switches on its gene for the APRO1 (TIS21) protein (Iacopetti et al., 1994, 1999; Tirone, 2001). One of its daughters will shut her (TIS21) gene off, but will still have some of mother's APRO1 (TIS21) protein and keep it during migration and the initial stage of neuronal differentiation. The appearance of APRO1 (TIS21), the resulting lack of CDK4•cyclin D1 kinase and hence the failure of cycle-suppressing pRb's hyperphosphorylation and inactivation, marks a neuron's birthday which over the subsequent decades the cell would celebrate as its "APRO1 (TIS21) day" (Iacopetti et al., 1994; Tirone, 2001). Knocking out the cycle-locking pRB gene prevents neuronal differentiation and keeps the permanently, cycle-unlocked neuroblasts proliferating, which ultimately results in a massive apoptosis of accumulating neuroblasts (Guardavaccaro et al., 2000; Lee et al., 1994). APRO1(TIS21) also associates with mCaf1, a component of the cell's mRNA deadenylase (Rouault et al., 1998; Tucker et al., 2001). This could mean that when the brief reign of APRO1(TIS21) stops the cell cycling, it ensures that the cycle block will continue by driving the deadenylation and breakdown of the cycle-related mRNAs, which clears the way for reloading the cell with mRNA messages from differentiation-related genes. By the time the APRO1 (TIS21) protein has gone, the cell, be it a neuron or osteoblast, is committed to becoming an irreversibly differentiated neuron or osteoblast.

One might ask whether preosteoblasts use the recently discovered gene-censoring siRNA (small interfering RNA) mechanism to sustain the shut down of the cycle engine. Such siRNAs could selectively cause the cleavage, or at least block the translation, of the mRNA transcripts of key cell cycle genes (Lau and Bartel, 2003).

APRO1(TIS21) is only one of a widely used set of genes the products of which, the maturation or "M" team, turn off the cell proliferation-drivers and irreversibly lock MC3T3-E1 cells into terminal maturation (fig. 20). The quiescence-induction-associated Q (quiescin) 6 sulfhydryl oxidase protein also peaks and drops along with APRO1 (TIS21) (Raouf and Seth, 2002). Indeed, Q6 expression is suppressed in tumor cells (Coppock et al., 1993, 2000). Another member of the M team is HAX-1, which is part of the EDC (epithelial differentiation gene) complex (e.g., involucrin, loricrin, S100s, trichohyalin) on human chromosome 1q21-22 (Marenholz et al., 2001; Raouf and Seth, 2002; Whitfield and Chakravarthy, 2001) (fig. 20). The Ca²⁺-triggered expression of this EDC gene complex is also at the heart of the terminal, apopotosis-like differentiation (i.e., diffpoptosis) of skin keratinocytes into keratin-loaded cadavers (Whitfield and Chakrvarthy, 2001).

In summary, the initiation of normal osteoblastic maturation might go as follows. The signals from collagen-stimulated integrins trigger the expression of PTHR1 receptor and PTHrP to stimulate the emerging PTHR1 receptors to give cyclic AMP surges for triggering the brief appearance of the very labile APRO1 (Fletcher et al., 1991; Tirone, 2001), which inhibits the expression of both cyclins D1 and E and the activity of any residual CDK4/6•cyclin D1 and CDK2•cyclin E kinases (fig. 21). And they drive Cbfa1/Runx-2 into the nucleus (fig. 21). The Ca²⁺ surge also caused by PTHR1 signaling may also stimulate the expression of HPX-1, which is another part of the differentiation machinery. The lack of CDK4/6•cyclin D1 prevents the phosphorylation of pRB, which, because of this won't let go of the replication genes' transactivator, E2F, but it can grab Cbfa1/Runx-2 by its C-tail to form a complex that promotes the expression of Cbfa1/Runx-2's target osteoblast-specific genes. Meanwhile, the briefly appearing APRO1 also de-adenylates and thus eliminates the mRNAs for the various parts of the cell cycle machinery.

The take-home message from all of this is—when APRO1 vanishes, the cell's key replication-driver genes are silent with their promoter-switch boxes buried in chromatin condensed by having CpG dinucleotide islands methylated and/or their histones deacetylated, but the cell emerges with a reconfigured nucleus with bone-making genes, such as osteocalcin, with their promoter-switch boxes now accessible for switching by having their chromatin decondensed by demethylation of CpG islands and their histones acetylated (Vilagra et al., 2002).

But (that ominous word again!) Fujita et al. (2002) have claimed that PTHR1 signaling does stimulate osteoblastic cell proliferation because they found that hPTH-(1-34) strongly stimulated cyclic AMP accumulation but only slightly stimulated (about 50%) proliferation of ATDC5, MC3T3-E1, and B-Raf-transformed MG63 bone cells. The proliferogenic mechanism appeared to have to be triggered by the cyclic AMP binding directly to and activating Epac, a cyclic AMP-GEF (Guanosine nucleotide Exchange Factor) (de Rooij et al., 2000; Stork and Schmitt, 2002; Zwartkruis and Bos, 1999). Cyclic AMP•Epac stimulates the exchange of GTP for GDP on GDP•Rap1, a Ras-like GTPase (Zwartkruis and Bos, 1999). GTP•Rap1 stimulates B-Raf protein kinase that sets off the mitogenic MEK1/2•ERK1/2 cascade (Fujita et al., 2002; Pearson et al., 2001; Stork and Schmitt, 2002; Zwartkruis and Bos, 1999). And Tawfeek (2002) has since shown that a PTH such as [Gly¹,Arg ¹⁹]hPTH-(1-28) which only stimulates adenylyl cyclase (Yang et al., 2004) does indeed stimulate the MEK•ERK cascade, which we would expect from a generation of cyclic AMP•Epac rather than a burst of PKA activity, which would have instead inhibited the MEK•ERK cascade. On the other hand, Mahon and Segre (2002) have found that the PTHR1 receptor's C-tail connection to NHERF-2 is needed for hPTH-(1-34) to stimulate the MEK mechanism in PS120 fibroblasts by a pathway not involving cyclic AMP (i.e., it is not stimulated by the AC-stimulator forskolin) or PKC. Significant though this variously triggered proliferogenic MEK-ERK cascade seems to be for PTH's actions in vitro, it is probably irrelevant to how PTH, PTHrP, and PTHR1 signalling actually stimulate normal osteoblast accumulation in a bone. Neoplastic or semi-neoplastic cells of various established bone cell lines, such as those used by Fujita et al. (2002), have their normally reciprocally controlled, mutually exclusive proliferogenic and maturation mechanisms only more or less disconnected. It is far more likely that the complex interactions of the cyclic AMP from activated PTHR1 receptors collaborate with APRO1 to permanently shut down the cell cycle machinery in normal bone cells.

PTHR1 Signals Turn on the Genes for Making Osteoblasts

Back in Section iiib, we learned that signals from PTHR1 can, like strain, load the nucleus with architectural transcription factors such as Nmp4 that bind target sites on chromosomes to produce genes with functionally reconfigured promoter-switch boxes accessible to being switched on by appropriate classical transcription factors. The accessibility of genes to these transcription factors driven into the nucleus by short bursts of PTHR1 signaling is also likely to need the brief (6-24 h) bursts of the enzyme UBP41 that could snip the ubiquitin off the nucleosomes' histone H2B, which has until now enhanced gene-silencing chromatin coiling by promoting the methylation of certain lysines equivalent to the lysine 79 in the "non-tail" core of the yeast nucleosomes' H3 histones (Briggs et al., 2002; Latchman, 2004; Miles et al., 2002; Osley, 2004; Sun and Allis, 2002). (There is also evidence for deubiquitination of lysine 4 in the H3 tail being involved in turning off certain genes (Osley, 2004)). As a result, the chromatin-uncoiling by the PTHR1-stimuated UBP41 exposes to transcription factors the promoters of the genes for factors such as FGF-2, FGF receptors and IGF-I, and further stimulate the expression of the whole PTHrP chain and one or more of its chopped-out cytokine components (Amizuka et al., 1996; Goltzman and White, 2000; Kartsogiannis et al., 1997; Walsh et al., 1997; Zhang et al., 1995) (fig. 20), and TGF-βs (TGF-β1 and β2 if the PTH is hPTH-(1-34) that stimulates adenylyl cyclase and membrane-associated PKC activity or only TGF-β2 if the PTH is hPTH-(1-31)NH₂ that seems only to stimulate adenylyl cyclase in human osteoblasts (Wu and Kumar, 2000)). I call these PTHR1-induced factors "missionaries" because they carry the PTH/PTHrP message, to multiply and get ready for bone making, to progenitors that are too young to have PTHR1 receptors. Some of these factors will be locked up in the new matrix and will not be released until the osteoclasts of a future microcrack-repairing BMU dig them out while digging a new tunnel or trench.

The production of mRNA for PTHrP-(1-139), -(1-141), and -(1-173) by alternate splicing (i.e., cutting and pasting) of the PTHrP gene transcripts in primary human osteoblasts increases a dramatic 38-fold by 45 to 90 minutes after hPTH-(1-34) is added to the culture medium (Walsh et al., 1997). While we know almost nothing about the contributions of PTHrP and its autocrine/paracrine components to mature osteoblast activity and the osteogenic response, there are reasons to believe that they are important. PTHrP's PTH-like 1-34 N-terminal region would bind to and activate the PTHR1 receptors, and through them adenylyl cyclase, as efficiently as a PTH (see Goltzman and White (2000), Lam et al. (1999), and Whitfield et al. (1998a) for reviews) (fig. 18). Indeed, because circulating PTH fragments with N-termini are normally rare, autocrine PTHrP must be the osteoblasts' main AC activator. But unlike PTH, the whole PTHrP molecule or a PTHrP fragment with the 85-107 NTS (nuclear-targeting sequence (Steggerda and Paschal, 2002)) that is either made in the cell or rides into the cell from the surface on its PTHR1 receptor can function as an ‘intrakine’ or "intracrine" factor that using its NLS and NES (nuclear export sequence) signal sequences is shuttled back and forth between the cytoplasm and nucleus (Aarts et al., 1999, 2001; Goltzman and White, 2000; Jans et al., 2003; Lam et al., 1999, 2000, 2002; Re, 2002a, 2002b; Watson et al., 2000).

But let's take a closer look at this ancient string of cytokines. When its mRNA is translated into a protein with a so-called secretory signal sequence, it is directed to the Golgi apparatus, where it is put into the secretory machinery. But the mRNA can also be translated by the ribosomes from an alternative translation-start site into a protein without a secretory signal sequence (Jans et al., 2003). This protein is not fed into the secretory machinery. Instead, it binds by its NLS sequence to the nuclear cargo carrier importin-β1 to form a complex that links to a microtubular trackway, where it is pulled by a dynein motor to the mouth of a pore in the nuclear envelope and handed over to the pore channel transport machinery. The complex is then ratcheted along the pore channel by series of dockings to so-called nucleoporins ("nups") lining the pore channel and then dumped into the nucleus, where it is separted from the importin carrier when the importin binds avidly to Ran•GTP and homes especially onto the nucleolus, where it can directly affect nuclear functions such as the ribosome-making machinery (Fried and Kutay, 2003; Jans et al., 2003; Lam et al., 2002).

It seems likely that this nuclear intrakine/intracrine stimulation would happen only in the mature, postmitotic preosteoblast or osteoblast because the cyclin-dependent protein kinases such as CDK4•cyclin D1 and CDK2•cyclin E that drive the cell cycle would disable the NTS by phosphorylating Thr⁸⁵ (Goltzman and White, 2000; Lam et al., 1999, 2000). Once in the nucleus and nucleolus, the peptide targets ribosome production and does its ancient job of coordinating the protein-synthesizing machinery with the signals from its activated surface receptors (Amling et al., 1997; Lam et al., 2000; Re, 2002a, 2002b). But it also seems likely that the nuclear PTHrP stimulates certain genes, among them the genes for the anti-apoptosis Bcl-2 and Bcl-XL proteins (Tovar-Sepulveda et al., 2002), that would prolong the cell's matrix-making activity by preventing it from killing itself by apoptosis (Antonsson and Martinou, 2000; Fadeel et al., 1999) (fig. 20). But in murine MC3T3-E1 osteoblasts, the PTHR1 receptor itself is also loaded into the nucleus during the late G₁ phase of the cell cycle (when the cyclin-dependent protein kinases are active and would stop NTS-bearing PTHrP from entering the nucleus) and then stays there throughout the rest of the cycle (Watson et al., 2000) (fig. 18). Therefore, the PTHR1 receptor either alone or with a N-terminal fragment lacking the Thr⁸⁵-containing region seems to be able to get into the preosteoblast's nucleus to carry out some job related to DNA replication, and in this it resembles the transcriptional activity of the EGF receptor (Lin et al., 2001; Ni et al., 2001).

PTHrP is not a hormone (i.e., it does not circulate throughout noncancer-bearing mammalian bodies as it does in fish, for example) or a single cytokine. It is really a string of different cytokines with 8 endoproteolytic consensus sites marking their borders where prohormone thiol protease and prohormone convertases can cut them out to operate as individual cytokines, each with its own receptor and specific mission (Deftos et al., 2001; Hook et al., 2001). One of these cytokines is the phospholipase-C-activating PTHrP-(67-86), the role of which in bone is unknown (Orloff et al., 1995). But another of these is hPTHrP-(107-139) ("osteostatin"), with its highly conserved 107-111 (i.e., Thr(T)¹⁰⁷-Arg(R)¹⁰⁸-Ser(S)¹⁰⁹-Ala(A)¹¹⁰-Trp(W)¹¹¹) region, that by itself can stimulate the proliferation of cultured fetal rat osteoblasts (Cornish et al., 1999) and possibly bone formation in 7- to 8-month-old OVXed rats (Roufflet et al., 1994). Moreover, concentrations of PTHrP-(107-139) and PTHrP-(107-111) as low as 10^-15 and 10^-13 M inhibit osteoclast generation and activity in vivo and in vitro (hence the name "osteostatin") (Fenton et al., 1991; Cornish et al., 1997; Zheng et al., 1994). And PTHrP-(107-139) is angiogenic—it could collaborate with the angiogenic action of leptin pouring out of hypertrophic chondrocytes (Kume et al., 2002) by stimulating human osteoblasts to make the angiogenic VEGFs, which would provide the blood vessels needed to feed the new bone being made in a trabecular/endocortical blister or cortical tunnel (Esbrit et al., 2000). And it might be the PTHrP-induced modulator of vascular invasion into cartilage that does not work through the PTHR1 receptor (Lanske et al., 1999). These fragments operate by stimulating PKCs on membrane scaffolds, not by activating PTHR1 but by activating an apparently very high-affinity "TRSAW" (Thr(T)¹⁰⁷-Arg(R)¹⁰⁸-Ser(S)¹⁰⁹-ALA(A)¹¹⁰-Trp(W)¹¹¹) receptor (Cuthbertson et al., 1999; Moonga and Dempster, 1995; Whitfield et al., 1996a, 1998a).

But there is something else lurking in the PTHrP string of factors which when it is cut out can drive bone formation. PTHrP's 8-11 sequence (LHDK) is virtually identical to endothelin-1's residues 6-9 (LMDK), which means that PTHrP-(8-11) can stimulate the endothelin A receptor (Mohammad et al., 2003). And endothelin-1 concentrations as low as 10^-10 do indeed stimulate both proliferation and differentiation of the endothelin-A receptor-expressing osteoblastic cells in rat calvarial cell cultures (von Schroeder et al., 2003). Moreover, stimulating the endothelin A receptor on mouse calvarial cells with either endothelin-1 or PTHrP-(8-11) strongly stimulates bone formation (Mohammad et al., 2003). Since endothelin-1 is made by vascular endothelial cells, this means, for example, that the blood vessels trailing behind the osteoclasts tunneling through bone can use it to stimulate osteoblasts to mature and start plastering new bone on the tunnel wall.

Figure 22

The points in the osteoblast maturation program where (more...)

Figure 22

.

The points in the osteoblast maturation program where PTHs and maybe statins control the progression to mature bone-making osteoblast. Abbreviations: ALP, bone-specific alkaline phosphatase; BSP, bone sialoprotein; Ca²⁺•PO₄, apatite-like calcium phosphate bone mineral; ECM, extracellular matrix; FGF-4, fibroblast growth factor-4; Ihh, Indian hedgehog; OC, osteocalcin; OP, osteopontin; OSX, osterix; Shh, sonic hedgehog.

You may remember that stromal cells are put on the osteoblast pathway when sonic hedgehog (Shh) stimulates them via FGF-4 to express BMP (bone morphogenic protein)-2 that stimulates them to express type 2 Cbfa1/Runx-2 (Ducy, 2000; Karsenty, 2000a; Katagiri and Takahashi, 2002; Komori, 2000) (fig. 22). As we shall see below, Cbfa1/Runx-2 is also involved in the driving of chondrocyte maturation and run-down of the prehypertrophic and hypertrophic chondrocytes cells to mineralization and coordinating this with the induction of osteoblasts in perichondrium/periosteum to make bone on calcified cartilage scaffolds (de Crombrugge et al., 2001). Cbfa1/Runx-2 interacts with the R-Smads specifically from activated BMPR-1B receptors to switch on the Osterix gene, which then steers the cell irrevocably onto the osteoblast pathway (Katagiri and Takahashi, 2002; Nakashima et al., 2002; Yagi et al., 2003). Another hedgehog—Indian hedgehog (Ihh)—produced by prehypertrophic and hypertrophic chondrocytes together with BMP-2, BMP-6, and Cbfa1/Runx-2 stimulates differentiation of a distinct population of perichondrial cells into osteoblasts in the collar of the developing bone (Chung et al., 2001; Karsenty, 2001). The osteoblast-determining potency of the Cbfa1/Runx-2 gene product is illustrated by the fact that forcing nonosteoblastic, primary fibroblasts to make it causes them to switch on their osteoblastic genes (Ducy, 2000; Lian and Stein, 2003). And just overexpressing the Cbfa1/Runx-2 gene in an adenovirus construct is enough to stimulate bone formation in a neonatal rat metatarsal organ culture (Krishnan et al., 2003).

Figure 23

The control of matrix mineralization. Things begin (more...)

Figure 23

.

The control of matrix mineralization. Things begin with the expression of Cbfa1/Runx-2 and its shipment into the nucleus, where it stimulates the expression of ALP (bone-specifc alkaline phosphatase) and Pit1 phosphate transporter. ALP is plugged into the cell membrane with its catalytic subunit (its businiess end) sticking out of the cell. ALP then clips phosphate off external compounds, and Pit1 carries it into the cell. But the cell won't immediately start mineralizing the matrix because the nuclear Cbfa1/Runx-2 has stimulated the production of osteocalcin (OC) that inhibits mineralization. However, the accumulation of phosphate causes Cbfa1/Runx-2 to be kicked out of nucleus, which stops OC and removes the block on mineralization. As phosphate is locked up in mineral, the cell phosphate content drops, Cbfa1/Runx-2 returns to the nucleus where it restores OC expression, stops mineralization. Now the cell has done its job and is ready to retire and become an osteocyte or bone lining cell or commit apopotic self destruction. Phosphate accumulation and exposure to the very high phosphate concentrations in osteoclast excavations can be deadly. But one of the many things PTHR1 signals can do is block phosphate-induced apoptosis.

As expected from this, one of the osteogenically determining things an injected adenylyl cyclase-stimulating osteogenic PTH does is push progenitor cells along the osteogenic path by stimulating them to express more PTHR1 receptors (Lu et al., 2001; von Stechow et al., 2003), and via bursts of PTHR1-driven adenylyl cyclase activity turn up the volume of Cbfa1/Runx-2 gene expression and the loading of the cells' nuclei with the Cbfa1/Runx-2 transactivator (Fujita et al., 2001a, 2001b; Krishnan et al., 2003; Moore et al., 2000; Selvamurugan et al., 2000) (fig. 22 and fig. 23). Cbfa1/Runx-2 has a NMTS (nuclear matrix-targeting sequence) domain in its C-tail (Zaidi et al., 2001) and a DNA-binding ‘Runt’ domain (Ducy, 2000; Ito, 1999). With its tail stuck to the matrix, Cbfa1/Runx-2 grabs the so-called OCE2 upstream sites of still silent osteoblast-specific genes such as the genes for ALP (alkaline phosphatase), BSP (bone sialoprotein), osteocalcin, osteopontin, OPG, proα1(I) and proa2(I) procollagen, PTHR1, and type I TGF-β receptor (Franceschi, 1999; Karperien et al., 2000; Kern et al., 2001; Komori, 2000) (fig. 22). Thus, Cbfa1/Runx-2 is one of the DNA-benders and twisters that pins these key genes to sites on the nuclear matrix containing the RNA-polymerase II-gene-transcribing machine (Lian and Stein, 2003; Turner, 2001; Vilagra et al., 2002). These now primed genes with their configured promoters exposed by being pinned down by Cbfa1/Runx-2 stay quiet until pRb and other classical stage-specific transcription factors appear to form transcription-activating multicomponent enhanceosomes that can switch on the transcription printing press (Turner, 2001). And the expression of another gene, the gene encoding the zinc-finger-containing transactivator OSX (osterix), which, as you will remember, is a "must" for the subsequent maturation of the cell into an osteoblast (Katagiri and Takahashi, 2002; Nakashima et al., 2002; Yagi et al., 2003; Zhang et al., 2002a) (fig. 22).

The "mature" or late osteoprogenitors emerging from the inter-trabecular pool of osteoprogenitors and just starting their spurt of PTHR1 receptor expression (one of the global markers of osteoblasts in all parts of the skeleton) and PTHR1-activating PTHrP (fig. 8) can go either directly or from the red marrow's sinusoidal vasculature into trenches dug into trabecular surfaces and the nearby cortical endosteum (Bianco and Riminucci, 1998). Presumably much less bone will be made at fatty marrow sites, as happens with FGF-2 treatment (Pun et al., 2001). However, they can reach cortical bone only from the marrow blood vessels after the cells in the trabeculae have had the first grab at them, which is one of the reasons why the adenylyl cyclase-stimulating PTHs are best at stimulating the growth of trabeculae-rich bones such as the vertebrae. Another reason for the PTHs' "trabeculophilia" besides the peptide first reaching the trabeculae after its injection or ingestion is that, except for their PTHR1 receptors, trabecular osteoblasts differ from cortical osteoblasts in their repertoire of expressed genes and their transcripts (i.e., their "transcriptomes") (Aubin, 2000, 2001; Aubin and Triffitt, 2002; Candeliere et al., 2001).

When the osteoprogenitor cells have arrived at the swept-up tunnel wall or trench and laid enough collagen on it, the signals from autocrine/paracrine PTHrP-activated PTHR1 receptors, the expressions of which have been stimulated by collagen-bound/activated integrins, have led the cells' "M teams" to shut down the cycle-driving machinery and enabled Cbfa1/Runx-2 to up-regulate the differentiation-related genes (fig. 22). But although they may look alike and make bone, the resulting post-mitotic mature osteoblasts do not have the same gene expression profile. The cells' gene expression profiles are specifically tailored to meet the different needs for working in different regions of the skeleton and even in different parts of a particular bone. Thus, for example, the relative expressions of bone sialoprotein, osteopontin, osteocalcin, and PTHrP at the mRNA and protein levels depend on which part of a developing bone (e.g., endocranial or ectocranial surface of mouse calvaria) the cell is located as well as who its immediate neighbors are and consequently the microenvironment in which it finds itself. In another example, osteoblasts in rat calvaria express glutamate NMDA receptor/channels with subunits NR2A, NR2B and NR2D, but not NR2C, while femoral osteoblasts express NR2C (Itzstein et al., 2001). But all osteoblasts, no matter where they are or who their neighbors are, express the genes for the PTHR1 receptor, type I collagen, and alkaline phosphatase (Candeliere et al., 2001). The products of these three genes are absolutely needed—they are the sine qua nons for becoming a mature, post-proliferative, matrix-making, and mineralizing osteoblast.

The PTHR1-expressing preosteoblasts and mature osteoblasts in modeling clusters making new bone and filling excavated microcracks in mature skeletons are the first or prime targets of injected PTH (K. Lee et al., 1994) (fig. 8). But it must be remembered that the signals from the PTHR1 receptors cannot stimulate the proliferation of maturing preosteoblasts and mature osteoblasts that have been proliferatively disabled by their PTH/PTHrP-induced M-teams' APRO1(TIS21)s. By the time they have got rid of their cell cycle machinery and become working osteoblasts, they have greatly (10-fold) increased the expression of PTHR1 receptors, which probably enables the autocrine/paracrine N-terminal PTHrPs that they have made and secreted to give themselves and their neighbors the cyclic AMP signals needed to drive their further maturation and bone-building (Aubin, 1998, 2000, 2001; Aubin and Triffitt, 2002).

Each daily PTH-triggered cyclic AMP transient releases active PKA catalytic subunits from inactive R•C holoenzymes (i.e., regulatory subunit–catalytic subunit) that surge from the cytoplasm into the target cell's nucleus through the nuclear pore complexes and turn on a set of early response genes with a CRE (cyclic AMP-responsive element) in their promoters and CREB•CBP/p300 complexes (cyclic AMP responsive element-binding protein•CREB-binding protein complexes) on the CREs (Ptashne and Gann, 2002; Quinn, 2002). However, after the nuclear catalytic subunit surge peaks around 15-30 minutes in cells such as MC3T3-E1 mouse preosteoblasts and ROS 17/2.8 osteoblastic cells, the catalytic subunits are no longer personae gratae. They are grabbed by the protein kinase A inhibitor PKIγ, which, like the various PKI isoforms, has a NES (nuclear export signal) domain, and are dumped out of the nucleus (X. Chen et al., 2002, 2003; Wen et al., 1995a, 1995b; Wiley et al., 1999). This, of course, stops the immediate-early gene expressions that the catalytic subunits were stimulating (X. Chen et al., 2003). The evicted catalytic subunits reassemble inactive R•C holoenzymes in the cytoplasm. Cyclic AMP transients may also feed back to stop the PTHR1 receptor attached to β-arrestin-2 from activating adenylyl cyclase by rapidly and briefly stimulating the expression of the gene for RGS-2, a GTPase activator, which converts receptor-generated adenylyl cyclase-activating GTP•Gsα to inactive GDP•Gsα (Pierroz et al., 2003; Thirunavukkarnasu et al., 2002).

One of the earliest of the many hundreds of genes directly and indirectly booted by the cyclic AMP signals from PTHR1s are genes of the Fos family (c-fos, fosB, fra1, fra2) the products of which can associate with the products of members of the Jun family (c-jun, junB, junD) of genes to form AP-1 (activator protein-1) transcription factor complexes that target an array of genes (Demiralp et al., 2001; K. Lee et al., 1994; McCauley et al., 1997, 2001; Onyia et al., 2001a; Tyson et al., 1999). The turning on of at least one of these genes, c-fos, is essential for PTHs to stimulate bone growth (at least in mice). Thus, knocking out the c-fos gene in mice also knocks out the ability of hPTH-(1-34) to stimulate femoral bone growth (Demiralp et al., 2001).

The switching on of the c-fos gene has been used to mark the sequence of responses of the various bone cells in 4-week-old rats to subcutaneously injecte full-length hPTH-(1-84) (K. Lee et al., 1994). The c-fos gene expression in trabecular osteoblasts, and the few spindle-shaped stromal cells that made PTHR1 receptors, peaked between 15 and 30 minutes and stopped by 60 minutes. Most stromal cells did not make PTHR1 receptors, nor could they, and the osteoclasts express their c-fos genes until 60 minutes by which time the osteoblasts and the few PTHR1-expressing stromal cells had stopped. This delayed c-fos switch-on in the receptorless stromal cells is best explained by missionaries sent to the PTHR1-lacking cells by PTHR1-loaded osteoblasts. Only after the PTHR1-receptorless stromal cells had stopped expressing c-fos at 2 hours were the osteoclasts expressing the gene. The very late burst of c-fos expression in osteoclasts is likely due to RANKL flowing from the immature osteoblstic stromal cells which, unlike the densely PTHR1-receptored mature osteoblasts with their shut-down rankl genes, are the only ones that can make this osteoclastogenic factor (Atkins et al., 2003; Corral et al., 1998).

These brief PTHR1-induced pulses of cyclic AMP and both cytoplasmic and nuclear PKA catalytic subunits and the c-Fos•Jun transcription factor complexes they generate only start the bone-making machine. Daily PTH-induced cyclic AMP spikes can't directly make immature marrow osteoprogenitors or preosteoblasts and mature osteoblasts proliferate (Kostenuik et al., 1999; K. Lee et al., 1994; Y. Wang et al. 2003; reviewed by Whitfield et al., 1998b). Indeed, a striking example of this has been given by Y. Wang et al. (2003), who has reported that PTH stimulated the progression of preosteoblasts to bone nodule-making mature osteoblasts in primary calvarial osteoblast cultures without stimulating proliferation. However, PTH can stimulate the PTHR1-expressing osteoblasts that have been proliferatively switched off by the "M-team" (fig. 20) to make factors that indirectly stimulate the proliferation of the immature osteoprogenitors, which do not yet have PTHR1 receptors.

Speading the PTH Message—The Missionary Factors That Suppress Cell Suicide and Stimulate Growth to Swell the Osteoblast Workforce

It is the products of the cascades of events triggered by the primary drivers such as amphiregulin (AR), FGF-2, IGF-I growth factors, and TGF-βs that grab the wheel and steer the machine in the receptor owners as well as start the generation of osteoblasts by stimulating the receptorless osteoprogenitors (Nishida et al., 1994; Whitfield et al., 1998b) (fig. 20).

Figure 24

Two of the three ways by which a PTH stimulates bone (more...)

Figure 24

.

Two of the three ways by which a PTH stimulates bone formation. (1) It stimulates the expression and release of missionary growth factors such as amphiregulin (AR) and short FGF-2, which can stimulate osteoprogenitors (Osteoprogen) that have not yet begun to express PTHR1s. (2) It extends the working lives of osteoblasts (OBs) by releasing the anti-apopotosis Bcl-2 protein from the stranglehold of the pro-apopotsis BAD protein, trapping the phosphorylated BAD protein in the cytosol bound harmlessly to 14-3-3 protein, and stimulating the expression of IGF-I which stimulates the expression of the anti-apopotosis Bcl-2 and Bcl-X_L proteins. The third way is to stimulate bone lining cells to come out of retirement and become working osteoblasts once again.

But the kick-starting PKA pulses also lengthen the working lifespans of mature osteoblasts by preventing them from committing scheduled apoptotic suicide (fig. 20). This increases the amount of bone they can make. Thus, as Gubrij et al. (2003) have shown, when osteoprogenitors can be made to constitutively overexpress an antiapoptotic protein such as Bcl-2 more of them survive to make more PTH-responsive, super-productive osteoblasts which make thicker trabeculae. The PTH/PTHR1-triggered PKA pulses' anti-apoptotic action results in part from phosphorylation and inactivation of the cells' apoptosis-initiating BAD protein, which makes BAD release the anti-apoptosis Bcl-X_L from its repressive grip and the phosphorylated BAD is trapped in the cytosol bound to 14-3-3 protein and thus unable to cause cytochrome c to leak from the mitochondria and trigger apoptosis-executing caspase proteases such as caspase-3 (J.M. Adams, 2003; Bellido et al., 2001; Chao and Korsmeyer, 1998; Stanislaus et al., 2000; Yin et al., 2003). Also they can be protected from suicidal thoughts and responses, maybe by the stimulation of sphingosine kinase that produces the anti-apoptosis sphingosine phosphate, but most definitely by the stimulation of the expression of the very important IGF-I, a stimulator of the expression of the apoptosis-preventing Bcl-2 and Bcl-X_L family proteins (Calvi et al., 2001; Hill et al., 1997; Johanson and Rosen, 1997; Locklin et al., 2003; Machwate et al., 1998; Pfeilschrifter et al. 1995; Pugazhenthi et al., 1999; Tovar-Sepulveda et al., 2002; Tumber et al. 2000; Virdee et al., 2000; Watson et al., 1995, 1999) (fig. 24).

The Bcl-2 gene is also turned on directly by the PTH-triggered cyclic AMP pulses. And its Bcl-2 protein product seems to be the sine qua non for PTH's anti-apopotsis action. Thus, preventing the translation of Bcl-2 gene transcripts with Bcl-2-specific, transcript-censoring siRNA (see the popular review of siRNAs' actions by Lau and Bartel, 2003) eliminates PTH's ability to prevent dexamethasone, etoposide, or the deprivation of substrate adhesion from triggering self destruction by apoptosis in OB-6 osteoblastic cells (Ali et al., 2003). Incidentally, this anti-apoptotic action of PTH plus its ability to stimulate the proliferation of hemopoietic CFU-S stem cells, which are normally clustered close to the PTHR1-bearing endocortical lining cells and responsive to factors such as Jagged1 produced by PTH-stimulated osteoblastic cells (Calvi et al., 2004; Cui et al., 1996; Gallien-Lartigue and Carrez, 1974; Lord, 1990; Perris et al., 1971), are probably why a subcutaneous injection of 50 to 200 USP units of Eli Lilly parathyroid extract (the same extract used by Selye in the 1930s) given 5 minutes to 1 hour after whole-body irradiation with 8.0 Gys of 2000 kvp x-rays increased the 30-day survival of 1296, 300-g male hooded rats from 33% to 73% (p<<<0.001) (Rixon and Whitfield, 1961).

Figure 25

How a PTH could directly stimulate the expression (more...)

Figure 25

.

How a PTH could directly stimulate the expression of the apoptosis-preventing Bcl-2 protein. The cyclic AMP surge from the PTH-activated PTHR1 causes PKA's (cyclic AMP-dependent protein kinase) catalytic subunits (C) to separate from the enzyme's regulatory (R) subunits and travel into the nucleus. The cyclic AMP signal also causes Cbfa1/Runx-2 to move into the nucleus. The C subunits phosphorylate the CREB (cyclic AMP response element-binding) protein on Bcl-2 gene's promoter and the several Cbfa1/Runx-2 proteins bound to their specifc sites on the gene's promoter. The result of these nuclear migrations and phosphorylations is the activation of the bcl-2 gene's promoter, Bcl-2 expression, and the suppression of apoptotic self-destruction.

How does PTH stimulate the Bcl-2 gene? The gene has one cyclic AMP-targeted CREB and several Cbfa1/Runx-2 sites in its promoter, and because of these, this anti-self-destruct gene is stimulated directly by the cyclic AMP boluses from PTHR1 (Plotkin et al., 2002). Thus, a cyclic AMP spike would phosphorylate and activate CREB and drive Cbfa1/Runx-2 into the nucleus and onto the Bcl-2 gene's promoter along, with the activated phospho-CREB (Fujita et al., 2001a). The PKA catalytic subunits that also surge into the nucleus would phosphorylate, and activate and thus increase the affinity of Cbfa1/Runx-2 for its target sites on the Bcl-2 promoter (Franceschi and Xiao, 2003; Selvamurugan et al., 2000) (fig. 24 and fig. 25). The PTH-induced, apoptosis-supressing phosphorylations of BAD by PKA and the stimulation of cyclic AMP/CREB/Cbfa1/Runx-2-dependent Bcl-2 expression are transient, i.e., self-limiting (Bellido et al., 2001, 2002). One reason for this transience is the silencing and expulsion of CREB-activating PKA catalytic subunits from the nucleus by PKI, which we met three paragraphs ago. Another reason is that the PTHR1 signaling also causes Cbfa1/Runx-2 to be dumped into the proteasome shredder (Bellido et al., 2002). Thus, only well-separated short bursts of PTHR1 can maintain the anti-apoptotic action. It follows from this that turning off the shredder with anti-protease, such as lactacystin, or overexpressing and swamping the system with Cbfa1/Runx-2 should, and in fact does, prolong PTH's anti-apoptotic action (Bellido et al., 2002).

By now the reader may be getting used to de-generalizing generalizations. In the case of PTH's anti-apoptosis action, two exceptions have been reported. The first exception has been found in murine C3H10T1/2 and MC3T3-E1 cells. As expected, hPTH-(1-34) triggered cyclic AMP-protected preconfluent cells from dexamethasone-induced apoptosis, but, and this is a very big but, the cyclic AMP surge triggered apoptosis in postconfluent C3H10T1/2 and MC3T3-E1 cells (H.-L. Chen et al., 2002). The second exception was found in the distal femurs of intact young Fisher 344 and Sprague-Dawley rats. Single daily injections of hPTH-(1-34) (Forteo™) into male rats that stimulate bone growth also stimulate a wave of osteoblast apoptosis in the proliferating zone just below the growth plate and osteocyte apoptosis in the osteocyte death zone that starts promptly during the first day, peaks around days 3-7, and then gradually drops off by 28 days (Stanislaus et al., 2000). There was no detectable increase in Bcl-2 expression. However, there does seem to be an anti-apoptosis effect in other regions of the femur, as indicated by a large drop in the activities of the apoptosis-driving caspases 2, 3, and 7 in the whole metaphysis. In fact, these drops are so large that they can obscure any more limited regional increase in the caspase activities. In other words, PTH stimulates apoptosis in certain zones of the rat femur, but it inhibits it elsewhere in the bone.

The IGF-I secreted by the PTH-stimulated, longer-lived, working osteoblasts also increases the osteoblast population by serving as a missionary that delivers PTH's call to proliferate to nearby proliferatively turned-on juvenile osteoprogenitors on the walls of new osteonal tunnels under construction or to osteoprogenitors in the bone marrow or trabeculat trenches, none of which have PTHR1 receptors or are not ready yet to make their own IGF-I (Calvi et al., 2001; Kostenuik et al., 1999). The importance of IGF-I for PTH's anabolic action is dramatically indicated by the fact that "knocking out" the IGF-I gene also knocks out the ability of hPTH-(1-34) to stimulate femoral bone growth in mice as well as the fact that PTH doesn't stimulate the proliferation of osteoblasts isolated from these IGF-I-less mice, unless the cells are given exogenous IGF-I (Miyakoshi et al., 2001).

But neither PTH, nor the IGF-I it induces, can stimulate osteoprogenitor proliferation or the growth of unloaded bones, such as hindlimb tibias, (Kostenuik et al., 1999), because the lack of loading disables the responsiveness of the osteoprogenitor cells to IGF-I (Bikle et al., 1994; Kostenuik et al., 1999)! In fact, if this missionary IGF-I is to carry the PTH message to proliferate and migrate to osteoprogenitor cells through the cells' IGF-1 receptors, the cells' aVβ3 integrin signaling tethers must be being tugged if the IGF-I receptors are to be activatable (Sakata et al., 2004). It seems that the activated integrins drive the collection of separately signaling-incompetent IGF-I receptors into signaling-competent clusters at the cell surface (Miyamoto et al., 1996; Sakata et al., 2004; Zheng and Clemmons, 1998). In other words—no loading, no integrin signaling, and so no signaling-competent IGF-1 receptors! As if this is not enough, stopping the waving of the cells' solitary cilial flowmeters and silencing the squeals from tugged integrins reduce the accessibility and responsiveness of osteoblast-specific genes to Cbfa-1/Runx2 and hinder classical transcription factors from stimulating MAP kinases and pumping architectural transcription factors like NMP4 into the nucleus to reconfigure and prime these key genes for responsiveness (Franceschi and Xiao, 2003; Latchman, 2004; Parvalko et al., 2003; Whitfield, 2003a). This reduced the ability of IGF-I to stimulate the Ras-ERK1/2-Akt mechanism and with it the expression of the genes for alkaline phosphatase, a-I collagen and osteocalcin.

This need for loading and the gene accessibility and responsiveness it creates is also seen on the cellular level. In MC3T3-E1 mouse calvarial preosteoblasts, simulated low or microgravity reduces the ability of hPTH-(1-34) to stimulate the expression of the c-Fos and c-Jun transcription factors (Ontiveros et al., 2002). Unloading also stresses and thus increases the incidence of apopototic suicide in the osteoprogenitor cells from unloaded bone (Sakata et al., 2002). But osteoprogenitors are not the only cells affected. As we learned in Chapter 2, the stress of canalicular fluid stasis drives osteocytes to commit apoptotic suicide and release lysophosphatidylcholine, which attracts carrion-hungry osteoclasts to their corpses. This means that there is an overwhelming surge of osteoclastic vultures that eliminate PTH/IGF-I targets and override any possible PTH action.

However, once again we must heed the terrible "BUT" word. Sakai et al. (1999) have found that unloaded tibias in the paralyzed, neurectomized hindlimbs of ddY male mice can still respond to hPTH-(1-34) as effectively as the tibias in the sham-neurectomized contralateral limbs!! Here, of course we are dealing with a particular strain of male mice, instead of Kostenuik et al.'s female rats.

As expected, A. Nakajima et al. (2002) have found that the cells in the PTH-stimulated calluses of rat femoral fractures express IGF-I, with the peak expression occurring between 4 and 7 days. But they have concluded that IGF-I was not responsible for the stimulation of osteoprogenitor proliferation because the surge of DNA-replicating cells expressing PCNA, the DNA polymerase-δ's processive replication factor (Whitfield and Chakravarthy, 2001), did not coincide with the surge of IGF-I expression. Unfortunately, they did not look at the expression of FGF-2, which, as we are about to see, could well have been the PTH-induced missionary mediator of osteoprogenitor proliferation.

IGF-I may also mediate the seemingly strong stimulation of DNA replication in the trabecular cells and chondrocytes of the rapidly developing bones of neonatal (2-day-old) mice injected with hPTH-(28-48) (Rihani-Bisharat et al., 1998). This PTH fragment binds to PTHR1, but as stated in sections i and ii of this chapter, it does not stimulate adenylyl cyclase or bone growth in OVXed rats, but it does stimulate PLC and PKCs (Jouishomme et al., 1992; Strein, 1994). And it is these signals that trigger IGF-I expression in the receptor-bearing, still proliferatively competent, late progenitors and mature, nonproliferating preosteoblasts and osteoblasts.

But cyclic AMP-stimulated IGF-I might not be as important for stimulating bone growth in humans as it seems to be in rodents (Johanson and Rosen, 1997). Human osteoblasts make much more IGF-II than IGF-I (Mohan and Baylink, 1999). Moreover, PTHs do not stimulate human osteoblasts to make IGF-II. Indeed, cyclic AMP actually seems to inhibit IGF-II expression. But the important thing is that the PTHs and cyclic AMP do trigger some production of IGF-I in these human cells (Mohan and Baylink, 1999). It would be interesting to know whether the cells have IGF receptors on their primary cilia to detect IGFs in their surroundings.

IGF-I collaborates with the TGF-βs that are also stimulated by the PTHs (Wu and Kumar, 2000) to strongly stimulate the procollagen α1[I] gene (Vermes et al. 2001). Therefore, these autocrine (self-stimulating)/paracrine (neighbor-stimulating) factors together vigorously drive osteoblasts' matrix production.

Figure 26

FGF-2 is one of the major missionary growth factors (more...)

Figure 26

.

FGF-2 is one of the major missionary growth factors sent out from a PTH-stimulated osteoblast. This composite cartoon shows how PTH stimulates the expression of this factor in osteoblasts with PTHR1 receptors and how it stimulates immature osteoblasts without PTHR1s but with FGF receptors. The cyclic AMP/PKA signal from PTHR1 stimulates the expression of short and/or long forms of FGF-2. The long form is translated, but the protein stays in the cell, and because it has a NLS (nuclear localization sequence), it moves into the nucleus where it stays and stimulates target genes. The short form is released from the cell and travels to cells with FGF receptors. There it activates the receptor that sends signals such as a burst of MAP protein kinase activity. But the short FGF-2 can also get into the target cell either by itself or with its receptor and use its NLS to get itself and its receptor into the nucleus to stimulate genes such as those for Cbfa1/Runx-2, collagen I, IFG-I, osteocalcin , PTHR1, ribosomes, and for itself (black circle) and others (white circle).

The cyclic AMP surges also cause osteoblasts to make two very important things besides IGF-I. One of these is the long (i.e., extended), 24-kDa, NLS (nuclear localization sequence)-bearing, isoform of FGF-2 (also known as bFGF (Okada-Ban et al., 2000; Ornitz and Itoh, 2001)) (Hurley et al., 1999; Liang et al., 1999; Montero et al., 2000; Power et al., 2002; Zhang et al., 2002b) (fig. 26). But this long FGF-2 is no traveling missionary. It is an intracrine factor that stays in the cell, where it operates mainly in the nucleus of rapidly proliferating cells, such as immature osteoprogenitor cells (Olsnes et al., 2003; Re, 2002a, 2002b) (fig. 26).

How does the long FGF-2 get its NLS? The FGF-2 mRNA transcript has three CUG start codons upstream from the "canonical" or conventional AUG start codon (Vagner et al., 1995). Usually, a ribosome binds to the transcript's 5' cap and "sniffs" along the transcript until it finds an AUG start codon and then settles down to translate the message. But upstream from the first start codon is a so-called IRES (internal ribosome entry sequence) with which the transcript plugs directly into the ribosome, which forthwith starts translating the message into the long FGF-2 without looking further on for a AUG start codon. The important point here is that between the transcript's three start sites and the downstream conventional AUG, there is a NLS before the secretion signal sequence (Jans et al., 2003; Stachowski et al., 2003). So the importins-α and -β override the secretion signal sequence, grab the long, FGF-2 by its NLS, and drag it into the nucleus. There it directly or indirectly stimulates the expression of a set of genes which includes the genes for Cbfa1/Runx-2, collagen-I, IGF-I, osteocalcin, and PTHR1, as well as the ribosome genes and its own fgf-2 gene, products of which can also inhibit apoptosis (Bouche et al., 1987; Debiais et al., 2002; Olsnes et al., 2003; Re, 2002a, 2002b) (fig. 26).

The other FGF-2 is the short (18-kDA) FGF-2, which stays mainly in the producer cell's cytoplasm and, like PTHrP, is exported as an intracrine/autocrine/paracrine factor that can activate both its own cell's and neighboring cells' receptors (Olsnes et al., 2003). The short FGF-2 is translated from the transcript's conventional downstream AUG start codon and so does not have the upstream NLS and therefore cannot be kidnapped by the importins and dragged into the nucleus. But it can still get into the nucleus because it is small enough to simply diffuse through the nuclear pores (Stachowski et al., 2003). This is the missionary FGF-2, which when secreted can activate FGF receptors on its producer cell as well as on immature (i.e., PTHR1-lacking) osteoblastic progenitors, or it can ride into the cell and into the nucleus on its receptor (fig. 26). The key point here is that the signals from the activated FGF protein kinase receptors drive the differentiation of osteoprogenitors, not necessarily by increasing the expression and production of Cbfa1/Runx-2, but by stimulating the phosphorylation and therefore the activity of existing Cbfa1/Runx-2 by stimulating ERK 1/2 protein kinases (Franceschi and Xiao, 2003).

The likelihood of FGF-2 being one of the prime mediators, if not the prime mediator, of PTH's bone-forming action is indicated by hPTH-(1-34)'s failure to stimulate bone formation in mice which have had their fgf-2 gene knocked out (Hurley et al., 2002). FGF-2 also stimulates cell migration, which means that it can lure osteoprogenitors to clusters of active FGF-2-producing osteoblasts in osteonal tunnels for example. It should be also be noted that while PTH works largely in OVXed rats by increasing trabecular thickness, injecting FGF-2 by itelf into such animals can also increase trabecular number and connectivity, probably by affecting a broader group of osteoblastic cells (Lane et al., 2003).

Therefore, PTH-induced surges of FGF-2 and IGF-I drive osteoblast accumulation and bone formation in part by stimulating the proliferation of juvenile osteoprogenitors and their ultimate differentiation into PTHR1-receptor-bearing osteoblasts with extended working lives and all thoughts of committing apoptotic suicide suppressed (fig. 20). As would be expected if FGF-2 mediates PTH action, pretreating OVXed rats for 2 weeks with daily intravenous boluses (250 μg/kg) of FGF-2 before 8-weeks of daily subcutaneous injections (80 μg/kg ) of hPTH-(1-34) and (of course a further round of FGF-2 boluses, this time from the PTH-treated cells) increases trabecular thickness and connectivity more than PTH injections alone (Iwaniec et al., 2002). This is probably due to the FGF-2 pretreatment increasing the pool of PTH-responsive osteoblastic cells.

PTHR1-induced bursts of PKA activity in UMR-106-01 rat osteosarcoma cells also rapidly (e.g., within 1 hour) and massively increase (e.g., by 12 times) the expression of the gene encoding amphiregulin (AR), a member of the EGF family, without affecting the other family members (EGF itself, betacellulin, epiregulin, HB-EGF, TGF-α) of the EGF and ErbB2 receptor-activating family (Qin et al., 2003) (fig. 21). Clearly such a potent proliferogen could collaborate with the other missionaries, such as short FGF-2, to stimulate osteoprogenitor proliferation (fig. 25).

Obviously, bone formation would also be enhanced by blocking osteoclast generation. As we have seen, mature osteoblasts have their RANKL genes shut off and instead make the anti-osteoclastogenic osteoprotegerin. But these amazing cells, when stimulated by PTH, give osteoclastogenesis a double whammy by also making the anti-osteoclast IL-18 , the gene for which is stimulated by cyclic AMP/protein kinase A signals from the activated PTHR1 receptors (Raggatt et al., 2003).

But RANKL has unexpectedly, indeed astonishingly, joined the ever-growing mob of factors (fig. 19) clamoring for precedence in the PTHs' osteogenetic action. It has been suggested that RANKL, or more precisely, oligomerized (aggregated, clustered) RANKL mediates PTH's cyclic AMP-driven anabolic action on primary mouse osteoblastic cells!! Why? Well, a cluster (oligomer) of RANKLs fused to the enzyme GST (glutathione-S-transferase) as a platform stimulates bone formation when injected into mice (Faccio et al., 2003). This cluster of RANKLs bound to RANK receptors stays on the surface of murine calvarial osteoblasts much longer than single molecules of RANKL bound to RANK receptors (Faccio et al., 2003)!!! The oligomerized RANKL stays there long enough to drive RANK receptor signaling that is sufficiently prolonged to stimulate collagen production via the ERK1/2 mechanism, which according to Franceschi and Xiao (2003) would among other things, phosphorylate and activate Cbfa1/Runx2. The mediation of PTH's anabolic action in mice by oligomeric RANKL•RANK complexes is also strongly indicated by the ability of osteoprotegerin to prevent PTH action (Faccio et al., 2003). But the question is what cells are stimulated by a PTH to make RANKL oligomers? Since mature osteoblasts' RANKL genes are locked shut by having CpG islands in their promoter-switch boxes methylated, the RANKL-making targets of PTH would have to be immature osteoblastic cells. However, both mouse osteoblasts and human osteoblasts (Reinholz et al., 2002) appear to have the RANK receptors needed to respond to RANKL, wherever it comes from.

There may be another unexpected source of missionary growth factors with which PTH boluses could drive osteoblast accumulation—developing osteoclasts! Kubota et al. (2002) have found that RANKL-stimulated developing osteoclasts to make PDGF (platelet-derived growth factor)-BB, which prevents MC3T3-E1 semi-transformed preosteoblasts from maturing by keeping them in the cycling mode. Therefore, the PDGF-BB from osteoclasts that have been induced to develop by RANKL from PTH-stimulated immature osteoblastic stromal cells could stimulate the proliferation of nearby osteoprogenitors. But this interaction would not last very long because the stimulated osteoprogenitors would increasingly make and release OPG and turn off their RANKL genes and RANKL production, which would combine to reduce osteoclast development and with it PDGF-BB production (Atkins et al., 2003; Kubota et al., 2002).

As if all of this is not enough, the reader may remember that the signals from PTH- or PTHrP-activated PTHR1 receptors probably stimulate the expression of leptin, which we met in Chapter 3i (Torday et al., 2002) (fig. 10). This autocrine and paracrine cytokine does a lot of things to promote bone formation (fig. 10). It opposes osteoclast generation by stimulating osteoblasts to make the anti-osteoclastogenic OPG and reduce the pro-osteoclastogenic RANKL production by immature osteoblastic cells, and it dissuades osteoprogenitors from choosing the adipocyte option, stimulate osteoblast maturation, stimulate the proliferation and migration of vascular endothelial cells to the bone construction site, and finally promote matrix mineralization.

The striking PTH-induced layering of rat bones with mature osteoblasts may not be due only to the stimulation of preosteoblast proliferation and maturation into apoptosis-resistant osteoblasts. In rats, it seems to be at least partly, if not wholly, due to the reversible, reversion of the thin, flat, post-mitotic, osteogenically inactive bone-lining cells to plump, bone-making, but of course still proliferatively disabled, osteoblasts (Dobnig and Turner, 1995; Leaffer et al., 1995). In other words, PTH summons lining cells from retirement to retool themselves for reentering the bone-making workforce. This transformation is likely due to a direct stimulation by PTH because, although the lining cells reduce PTHR1 receptor expression and other traces of their past osteoblast lives, they still express some receptors (at least in humans) (Langub et al., 2001). Presumably, the reverting cells would raise their PTHR1 receptor display to the level in mature osteoblasts and once again start making missionary growth factors such as FGF-2, IGF-I, and TGF-β(s) and releasing them into the adjacent bone marrow to expand the pool of osteoprogenitor cells. When the PTH injections stop, the ex-lining cells relax, stop making bone, and return to being flat lining cells doing their retirement job in the osteocyte-lining cell signaling network (Dobnig and Turner, 1995; Leaffer et al., 1995). The resulting drop in IGF-I production and secretion by the reverting lining cells and aging osteoblasts would restore their vulnerability to apoptosis.

But while the flat, resting lining cells are not "tooled up" to make the huge amounts of bone matrix they made when they were osteoblasts, they can still make some. They sweep up the collagenous litter left by the BMU osteoclasts on the floor of the excavated microcrack repair site (i.e., the blister) (Everts et al., 2002). While doing this, they must be exposed to a lot of inorganic phosphate as well as Ca²⁺. When enough of this phosphate is brought into the cells by their transporters, it directly activates the phosphate-responsive promoter-switch box of the osteopontin gene (Beck, 2003; Beck and Knecht, 2003; Beck et al., 2000, 2003), the product of which the cells use to mix up a layer of cement to glue the large amount of matrix made by the incoming mature osteoblasts to the old bone (Denhardt and Noda, 1998; Denhardt et al., 2001; McKee and Nanci, 1996). The janitorial lining cells also deposit a thin priming layer of collagen on top of the osteopontin cement to prepare for the arrival of the mature osteoblasts.

Matrix Mineralization

Needless to say, inorganic phosphate and the mature osteoblast's Na⁺ gradient-driven, type III Pit1 phosphate transporter that ships it into the cell are key players in bone formation (Caverzasio and Bonjour, 1996; Nielsen et al., 2001; Selz et al., 1989; Takeda et al., 1999). Moreover, we are now learning that inorganic phosphate is a major signaler, an activator of certain gene promoters (Beck, 2003; Beck and Knecht, 2003; Beck et al., 2003). The PTHrP-triggered signals from the emerging PTHR1 receptors on the young matrix-making cells drive Cbfa1/Runx-2 into the nucleus to stimulate the ALP (alkaline phosphatase) gene, the product of which is plugged into the cell membrane with its catalytic subunit (i.e., its "business end") sticking out into the extracellular space (Beck and Knecht, 2003; Beck, 2003; Beck et al., 1998, 2000, 2003). But the cell won't immediately start mineralizing matrix because the Cbfa1/Runx-2 pushed into the nucleus by PTHR1 signaling (Fujita et al., 2001a, 2001b) stimulates the expression of the mineralization-blocking osteocalcin (Ducy et al., 1996) (fig. 23). The ALP then snips phosphate out of external substrates such as the β-glycerophosphate that is added to culture media for bone nodule formation and in the Real World out of matrix components. The rising phosphate switches on the gene for the Pit 1 transporter (Beck et al., 2003) to carry phosphate into the cell. The incoming phosphate stimulates the expression of a group of genes whose products drive the transition from matrix-making to mineralization (Beck and Knecht, 2003; Beck et al., 2003). First, the incoming phosphate downregulates the expression of the major matrix components, collagen-1α1 and collagen-1a2, because the emphasis must now shift to mineralization—enough collagen has been made (Beck and Knecht, 2003; Beck et al., 2003). The incoming phosphate also stimulates the gene for Nrf2, the transcription factor that homes on the antioxidant response element (ARE) and the battery of genes that have it in their promoter-switch boxes (Nguyen et al., 2003). Presumably the Nrf2/ARE-driven genes and their products will help the cell survive the formidable challenges of a lot of phosphate and a blebbing membrane (Beck et al., 2003). Of course, the phosphate also stimulates the expression of calcyclin and annexin V which, as we shall see in the next paragraph, are needed for the mineral formation (Beck et al., 2003). And the phosphate also stimulates the gene for cyclin D1, which has some role in differentiating osteoblasts besides its more familiar role of triggering the build-up to DNA replication in proliferatively competent cells (which the mature osteoblast most certainly is not). However, when the Pit-1-pumped phosphate reaches a critical level it stimulates the nuclear export machinery, which empties the nucleus of its Cbfa1/Runx-2 (Fujita et al., 2001a, 2001b) (fig. 23). This dumping of Runx2/Cbfa1 out of the nucleus shuts down osteocalcin expression (Fujita et al., 2001b), which lifts the block to mineralization (Ducy et al., 1996; Fujita et al., 2001a) (fig. 23). (Of course, the block will have to be reimposed later on to avoid runaway mineralization.) Again one must remember that timing is everything.

To mineralize the freshly made matrix, the osteoblast loads Pit transporters, alkaline phosphatase, and annexin V Ca²⁺ channels into prevesicle patches of their cell membrane that will be budded off to make vesicles that are detached calcium-apatite-making engines. To start out, the large amount of CaR-activating Ca²⁺ released by the osteoclasts is sucked up, and the osteoblasts' CaRs are silenced when the cells combine the Ca²⁺ with phosphate to mineralize the osteoid matrix they have just finished making. The Cbfa1/Runx-2-stimulated alkaline phosphatase and Pit1 transporter (the gene for which has been stimulated by incoming phosphate) are now plugged into the prevesicle patches (Beck, 2003). But it seems that the transporter's operation must be protected by the cell's PHEX endopeptidase, which would destroy any transport-inhibiting FGF-23 that the cell might be making or be exposed to (Bowe et al., 2001; Shih et al., 2002). The new osteoid is mineralized when alkaline phosphatase-generated, Pit1 transporter-delivered phosphate and Ca²⁺ flowing through the channels formed in the vesicle walls by the AnxV, the gene for which has also been stimulated by incoming phosphate, combines to form needles of impure hydroxyapatite (Ca₁₀[PO₄]₆[OH]₂) in the matrix-bound vesicles (Bandorowicz-Pikula et al., 2001; Beck, 2003; Caverzasio and Bonjour, 1996; Hoshi and Ozawa, 2000; Selz et al., 1989). These vesicles armed with alkaline phosphatase, Pit1 transpoprters, and annexin V channels in their walls are now budded from the surfaces of the osteoblasts, and when they have loaded themselves to bursting with calcium apatite-like needles, they spill the needles out onto the osteoid (Caverzasio and Bonjour, 1996; Nielsen et al., 2001). Since PTH-induced cyclic AMP transients stimulate type III transporters, well-separated boluses of AC-stimulating PTHs promote mineralization by selectively increasing the V_max (the transporter's top carrying speed) of the transporter, which persists even after the vesicles bud from the osteoblast's surface (Caverzasio and Bonjour, 1996).

At this point, we might ask what happens to the PTHR1 receptors when the mineralizing osteoblast's surface is budding off vesicles. Are they budded off along with the vescles? Certainly it appears that the post-mineralization osteocytes have downregulated their PTHR1 receptor expression.

If an osteoblast is lucky enough to survive surging Ca²⁺ and phosphate and a roiling membrane to become an osteocyte, the incoming phosphate will have triggered a second burst of osteopontin gene expression and osteopontin synthesis to glue the cell and its processes to the walls of its cubicle and its dendritic processes to the walls of their canaliculi (Beck and Knecht, 2003; Beck et al., 1998, 2000, 2003; Denhardt and Noda, 1998; Denhardt et al., 2001; McKee and Nanci, 1996). The emerging osteopontin also prevents the incoming phosphate from inducing apoptosis (Koyama et al., 2003).

The incoming phosphate stimulates osteopontin expression by somehow causing the biphasic activation (by phosphorylation) specifically of ERK 1/2 (Beck and Knecht, 2003). In other words, there is an early surge of phosphorylated ERK 1/2 between 15 and 45 minutes after adding 10 mM phosphate to the MC3T3-E1 cells' culture medium and a second surge beginning at 8 hours and building up at least until 32 hours. Other members of the MAP kinase family such as p38 are not involved, but a separate stimulation of PKCs activity also seems to be needed.

A Very Quick Overview

Now let's take a short time out and briefly summarize how PTH stimulates bone growth in an adult bone such as a femur or tibia (fig. 20). At the time of the first PTH injection, there are osteoblasts in microcrack-repairing BMUs, busily laying layers of bone around the walls of recently dug tunnels in cortical bone and filling trabecular trenches. But then there is the large population of lining cells—retired osteoblasts—still with some PTHR1 receptors, covering the walls of the cortical canals, the endosteum, and trabeculae. The PTH diffusing from the blood vessels in the canals and the marrow sinusoids will hit all of these cells and cause them to come out of retirement and retool themselves with more PTHR1 receptors and the various other devices for making bone. And they will be kept from slipping back into retirement by periodic kicks from the daily PTH boluses. The reverted lining cells on the endosteal and trabecular surfaces are strategically placed for stimulating various progenitors—they are bathed in bone marrow, which means that the growth factors that the daily PTH boluses cause them to make and secrete can directly reach and stimulate the proliferation of immature osteoprogenitors. Of course, the PTH's antiapoptotic action could delay the inevitable mass (95%) self-destruction of the BMU osteoblast crews, but this would not apply to the stimulated osteoblast/lining cells, which would just stay where they are on the bone surface and slip back into the role of being lining cells connected to osteocytes and monitoring the continuous chattering of osteocytes for signs of fracture (fig. 2 and fig. 3).

PTHs Speak to Bones with Forked Tongues—The Other Prong Says Remove It!

While the osteogenic action of single daily injections of an adenylyl cyclase-stimulating PTH predominates over their osteoclast-stimulating, bone-demolishing action, the balance shifts to net resorption if the PTH is continuously infused for longer than 1 hour or injected too frequently (Dobnig and Turner, 1997). But it appears that although intermittent injections of hPTH-(1-34) have been found to at least transiently increase cortical porosity in humans, monkeys, and rabbits, it has been found in mice and rats that anabolic intermittent injections of medium and high doses of the peptide actually decrease osteoclast surface in OVX rats (Lane et al., 1996) and suppress the final steps of osteoclast differentiaition in male ddY mice (Sakai et al., 1999).

Why and How Does Prolonged PTH Infusion Destroy Bone?

Figure 27

The various things that happen to stimulate osteoclast (more...)

Figure 27

.

The various things that happen to stimulate osteoclast activity and bone resorption rather than bone growth when a PTH is continuously infused instead of being injected once a day or swallowed once a day in a pill.

First, Watson et al. (1999) found part of the answer in rats. As expected, intermittent injections of the full-size hPTH-(1-84) caused a massive accumulation of IGF-I-making osteoblasts (Watson et al., 1995, 1999). Continuously infusing the PTH still caused osteoblasts to accumulate, but now things were different. The cells made less autocrine/paracrine IGF-I and start making IGF-binding proteins, IGFBP-3, IGFBP-4, and IGFBP-5 (fig. 27). The IGF-I-trapping IGFBPs, particularly IGFBP-4 (Mohan et al., 1999), would have prevented the smaller amounts of IGF-I from stimulating bone formation. Then, Locklin et al. (2003) reported that intermittent exposure of cultured primary mouse marrow cells to hPTH-(1-34) caused an IGF-I-dependent stimulation of osteoblast differentiation markers while continuous exposure could not stimulate these markers and along with this reduced IGF-I mRNA production. Added to the IGF-I blockade is the switching on of the accumulating osteoblasts' matrix-destroying collagenase-3 by overproduction of FGF-2 (Hurley et al., 1995, 1999) (fig. 27). PTH itself would also have stimulated collagenase-3 expression by stimulating Cbfa1/Runx-2 expression (Locklin et al., 2003; Selvamurugan et al., 2000), but as noted in the next paragraph, this might not happen with excessive exposure to PTH. This combination of IGF-I blockage and a surging, matrix-destroying collagenase should be enough to tip the balance over to bone demolition.

Intermittent exposure of mouse marrow to adenylyl cyclase-stimulating PTHs stimulates Cbfa1/Runx-2 that drives osteoblast differentiation and bone building (Locklin et al., 2003). Single daily boluses of adenylyl cyclase-stimulating PTHs actually trigger the loading of nuclei with Cbfa1/Runx-2 and the expression of about 158 genes in rat femoral metaphyseal bone cells (Fujita et al., 2001a, 2001b; Moore et al., 2000; Onyia et al., 2001). Among these genes is the one encoding a still functionally mysterious, 145-amino acid cytoplasmic protein PAIGB—in the tibial cells of Sprague-Dawley rats (Robinson et al., 2003).

These daily boluses also promptly trigger only brief (6-24 h) bursts of ubiquitin-specific protease UBP41, expression of which could save key gene transactivators by removing any attached ubiquitin chains that would mark them for the proteasome shredder (Chung and Baek, 1999; Huang et al., 1995; Miles et al., 2002). However, as I said in section iiid, it seems likely that these brief bursts of UBP41 activity could promote the reconfiguration of chromatin that makes genes more accessible to appropriate transcription factors (Latchman, 2004). Ubiquitination of histone H2B facilitates the methylation on lysine residues 4 and 79 of histone H3, which promotes a more tightly packed chromatin structure and gene silencing (Latchman, 2004). Therefore, removing these ubiquitins with PTH-triggered, cyclic AMP-driven bursts of UBP41 could result in the demethylation of H3 histones and the unlocking of dozens of genes, such as the osteogenically key genes that are responsive to the Cbfa1/Runx2 that is, at the same time, being loaded into the nucleus by the PTH boluses.

But continuously infusing these PTHs into rats selectively switches on a different and very much larger, set of 759 genes in the metaphyseal cells, which, for example, no longer includes the gene for the PAIGB protein, but instead includes genes for bone matrix-shredding proteases (Onyia et al., 2001a, 2001b; Robinson et al., 2003). Continuous PTH exposure also locks on UBP41 expression (Miles et al., 2002; Murray et al., 1998), which, by increasing the free ubiquitin pool along with a cyclic AMP-induced stimulation of the proteasomal endopeptidases, would shift the cell's proteome (its protein complement) from one for bone making to one for bone destruction. This prolonged activity of UBP41 and increased proteasomal enzymes drives a PKA-stimulated ubiquitination and destruction of Cbfa1/Runx-2 with a consequent shutdown of Cbfa1/Runx-2-dependent osteoblast-specific genes such as the genes for Bcl-2 and OPG (Bellido et al., 2002; Tintut et al., 1999) (fig. 25). The upshot of these changing gene expressions during continuous PTH infusion are drops in the expression of the antiosteoclastogenesis OPG and osteoblasts' bone-making genes, such as those for BSP, collagen I, and osteocalcin, as well as a reciprocal increase in RANKL and the products of other genes that drive osteoclast differentiation and activation and bone resorption (Ma et al., 2001).

How could continuous PTH infusion turn on or off so many more genes than single daily boluses of the peptide? Evidently, continuous infusion activates additional transcription factors. According to Chauvin et al. (2001), spacing the boluses 24 hours apart would enable an osteoblast's PTHR1 receptors to fully recycle once and enable the cell to present a fully receptored surface to the next PTH bolus. Each bolus would cause a brief burst of cyclic AMP/PKA signaling from the full complement of PTHR1s on the various target cells. But a continuously infused PTH might cause a continuous, β-arrestin-2-mediated endocytic cycling of PTHR1s with reduced steady-state level of receptors and very low-level adenylyl cyclase/cyclic AMP signaling (Locklin et al., 2003) but with a steady, β-arrestin-2-mediated pumping of MKK4-activated JNK kinase (and maybe PTHR1 receptor itself ) into the nucleus (fig. 18). Such a sustained injection of a transcription factor(s) activator such as JNK into the nucleus could massively stimulate the expression of a much larger set of genes and their resorption-driving products.

The most important arbiter of whether a PTH builds or destroys bone is the RANKL/OPG expression ratio it establishes. Halladay et al. (2002) have used UMR106 rat osteoblast-like cells transfected with an opg gene promoter construct pOPG5.9βgal (-5917 to +19) to see the effects of short and long exposures to hPTH-(1-38) and a PTH of the Ostabolin™ Family, hPTH-(1-31) (Ostabolin™), on the activity of the opg gene's promoter as indicated by the extent of stimulation of a β-galactosidase "reporter gene" attached to the promoter construct. Exposing the UMR106 cells to the PTHs or to cyclic AMP-elevators such as the potent adenylyl cyclase-stimulating forskolin or to a cyclic AMP turnover inhibitor, the cyclic nucleotide phosphodiesterase inhibiting, IBMX, or to cyclic AMP itself for 8 hours nearly doubled the opg promoter-switch box activity, with the hPTH-(1-31) being slightly more effective than hPTH-(1-38). However, exposure to either PTH or one of the cyclic AMP elevators for 24 or 48 hours turned the promoter off. (Unfortunately, they did not look at the effects of the peptides at shorter intervals, such as the physiologically relevant 1 or 2 hours.) On the other hand, a PKC stimulator such as TPA (12-O-tetradecanoyl phorbol-13 acetate) (at 10^-8 or 10^-7 M) reduced the opg gene promoter activity in cells exposed to the drug for 8 hours but strongly stimulated the opg promoter in cells exposed for 24 or 48 hours. Thus, the stimulatory effect of a short exposure to an adenylyl cyclase-/PKC-stimulating PTH (remember that this of course depends on the cell's NHERF-2 content) on opg gene activity would be reduced by PKC stimulation, while the inhibitiory effect of a prolonged exposure would be blunted by PKC stimulation. This intervention of PKC into the process could at least partly explain why intermittent injections of members of the Ostabolin™ family (hPTH-(1-31NH2 (Ostabolin™) and [Leu²⁷]Glu²²-Lys²⁶hPTH-(1-31)NH₂)hPTH-(1-31) (Ostabolin-C™)), which tend not to stimulate PKC, are poor stimulators of osteoclast generation and do not cause hypercalcemia in various animal models and humans, even at very high doses (Fraher et al., 1999; Jolette et al., 2003; Mohan et al., 2000; Whitfield et al., 1998c).

Added to direct opg gene suppression by prolonged cyclic AMP pulsing is the ability of a prolonged PTH exposure to prevent osteoblastic cells from making OPG by stimulating the proteasome that destroys Cbfa1/Runx-2 which normally keeps the opg gene working by binding to the opg promoter's 12 so-called OSE2 sites (Bellido et al., 2002; Fu et al., 2001; Galvin et al., 2001; Hofbauer et al., 1999, 2000; Kanazawa et al., 2000; Miles et al., 2002; Murray et al., 1998; Onyia et al., 2000; Stilgren et al., 2001; Thirunavukkarnasu et al., 2000; Tintut et al., 1999).

But what about the other player, the pro-osteoclast RANKL? Within an hour of a single injection of the adenylyl cyclase/PKC-stimulating hPTH-(1-38) into rats, RANKL mRNA production surges and OPG mRNA reciprocally crashes in the distal femoral metaphysis and diaphysis, but OPG expression soon rebounds as its gene's promoter-switch box is turned on and rankl gene expression is reciprocally turned off by 3 hours, which means that the RANKL/OPG ratio must be low by this time (Ma et al., 2001; Onyia et al., 2000). Such intermittent OPG peaks could be why daily injections of hPTH-(1-34) (Forteo™) or [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin C™) decrease osteoclast activity or kill osteoclasts in OVX rats and cynomolgus monkeys (Jolette et al., 2003; Lane et al., 1996).

According to the observations of Locklin et al. (2003) on mouse marrow cells, if the PTH treatment is continuous, rankl gene expression should rise while the opg gene's promoter and thus its expression should be inhibited in the rat. Indeed, as expected, continuous infusion of hPTH-(1-38) into rats causes a sustained rise in rankl expression, a sustained drop in OPG expression, and increases in serum Ca²⁺ concentration and osteoclast generation (Ma et al., 2001). Locklin et al. (2003) have also found that although a continuous exposure to hPTH-(1-31) (Ostabolin™) is a 2 to 3 times more effective stimulator of rankl gene expression than continuous exposure to hPTH-(1-34) in their mouse marrow cultures, the two peptides only equally increased osteoclast generation. This indicates a large difference between the actions of factors controlling message processing. And it could mean that hPTH-(1-31) signaling lacks something needed to get the additional number of rankl gene transcripts translated into active RANKL. But as we will see further on, we are being confused by data from "pure" and multicomponent cultures and the whole animal.

Figure 28

The reciprocal control of osteoprotegerin (OPG) and (more...)

Figure 28

.

The reciprocal control of osteoprotegerin (OPG) and RANKL expression by a cyclic AMP/PKA signal from PTH-activated PTHR1 receptor. This mechanism would operate in a single cell, but the striking inability of PTHs belonging to the Ostabolin™ family to stimulate osteoclast activity and bone resorption is not due to this kind of mechanism but to the abilities of juvenile osteoblastic cells to make RANKL but not OPG and the abilities of mature osteoblasts to make OPG but not RANKL as discussed in the text.

At this point, the reader might ask why the expressions of RANKL and OPG are reciprocal in the same cell? The answer might lie in the phosphorylation of the CREB protein's Ser¹³³ by PKA activated by the cyclic AMP signals coming from PTH-activated PTHR1 (Fu et al., 2002). The PKA-phosphorylated/activated CREB plus a CBP/p300 (Ptashne and Gann, 2002; Quinn, 2002) on the rankl gene promoter's CRE turns on the rankl gene (fig. 28). At the same time, CREB•CBP/p300 on the c-fos gene's promoter stimulates this gene's expression (Tyson et al., 1999). The c-Fos product turns off the opg gene via the AP-1 complexes (Latchman, 2004) it forms with c-Jun on the opg gene's promoter (Fu et al., 2002) (fig.28). (Interestingly, the hPTH-(1-34)-induced increase in AP-1 activity and suppression of opg gene expression in osteoblasts could be enhanced by signals from serotonin receptors activated by serotonin coming from strain-stimulated osteocytes (Bliziotes et al., 2001). This mechanism explains the prompt stimulation of rankl and suppression of opg by the brief bolus of hPTH-(1-38) in the rats of Ma et al. (2001). But why did opg expression surge upwards and rankl crash during the next hour or so? Obviously the inhibitory AP-1 complexes on the opg promoter must rapidly disappear leaving a still activated opg promoter. According to this sort of scenario, a continuing supply of AP1 and suppression of opg, activation of rankl and the resulting increased osteoclast generation would require a sustained barrage of PTH boluses. But there is a problem here. I suspect that this reciprocity mechanism at the single-cell level may actually be irrelevant in the whole bone or primary mixed bone cell cultures, because immature osteoblastic cells express their rankl gene, but then turn the gene off and switch the opg gene on when they become functioning osteoblasts (Atkins et al., 2003; Corral et al., 1998).

But in mice the responses of OPG and RANKL expressions to PTH injections appear to depend on whether the animal is OVXed or not (Lu et al., 2001). They have found that OVX alone caused the level of RANK expression in the tibias of C57BL/36 mice to double by 4 weeks and then drop back to only one-third of the starting level during the next 7 weeks and the OPG expression dropped to only 20% of the starting level by 11 weeks. Injecting hPTH-(1-34) into intact mice once daily for 5 days a week caused RANKL expression to increase 4 times and PTHR1 expression to triple. But, OVX inverted the responses of RANKL and OPG to PTH. The same PTH injections into OVXed mice reduced RANKL mRNA production to 37% of the starting level and caused a 2- to 3-fold increase in the level of OPG mRNA and PTHR1 expression, probably because the PTH-stimulated the estrogen-lacking RANKL-competent immature cells to mature into RANKL-incompetent, OPG-making osteoblasts without being replaced by new RANKL-making immature cells. The estrogen lack could indeed have caused a shortage of immature osteoblastic stromal cells by diverting the flow of bi-potential precursor cells from the osteoblastic to adipocyte pathway (Okazaki et al., 2002). This suggests that PTHs might be less able to stimulate bone growth in ovary-intact than in OVXed rats. And this is indeed the case. The abilities of hPTH-(1-34) and the two members of the ostabolin family of PTHs to stimulate femoral trabecular growth in ovary-intact rats are only about half of their abilites to stimulate trabecular growth in OVXed rats (unpublished observations in my laboratory). It is obviously not a little important to know whether this inversion can be repeated and whether it happens in postmenopausal women.

Yet another factor will come into play when the PTH boluses are too closely spaced. Adenylyl cyclase-stimulating PTHs increase the top speed (V_max) of the osteoblasts' Pit1 phosphate transporter (Caverzasio and Bonjour, 1996; Nielsen et al., 2001; Selz et al., 1989). This transporter normally provides the inflow of phosphate needed to drive the expression of osteopontin and the accumulation of mineral in the matrix vesicles budded off the osteoblasts (Beck et al., 2000; Caverzasio and Bonjour, 1996; McKee and Nanci, 1996; Selz et al., 1989). Well-spaced cyclic AMP/PKA pulses would promote osteoid mineralization by causing phosphate to surge into the matrix vesicles. However, continuous infusion or more closely spaced boluses of a PTH might boost the flow of phosphate into the cell to an apoptosis-triggering level (Allen et al., 2002; Mansfield et al., 2001; Meleti et al., 2000).

The ability of PTHs to stimulate resorption has bedeviled the clinical application of these peptides since the 1930s. So the dream of all "PTH buffs" has been to make a PTH that could stimulate bone growth, but not bone resorption. Higher doses of such a PTH could be safely used to maximally accelerate fracture healing and get a bigger anabolic bang for the osteoporotics' buck. It appears that two such PTHs have at last been made. These PTHs belong to the Ostabolin™ family—hPTH-(1-31)NH₂ (Ostabolin™) and [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin-C™ i.e., cyclized Ostabolin™). The potently anabolic Ostabolin™ has been shown to be both orally administrable and a very poor stimulator of resorption in humans and mice (Fraher et al., 1999; Mehta et al., 2000; Mohan et al., 2000). And Jolette et al. (2003) have shown that the potent stimulation of bone formation in cynomologous monkeys, even by daily injections of such a large dose of Ostabolin-C™ as 80 μg/kg of body weight for 7 days, decreased the indices of bone resorption!

Moreover, Paul Morley (Zelos Terapeutics) has informed me that Ostabolin-C™ does not cause the burst of cortical porosity in cynomologus monkeys that is caused by hPTH-(1-34) (Forteo™) (Burr et al., 2001). Intermittent injections of Ostabolin-C™ may simply be better at suppressing osteoclast differentiation than Forteo™ (Lane et al., 1996; Sakai et al., 1999). Indeed during the first 8 weeks after OVX, the metaphyseal trabecular number in distal femurs of Sprague-Dawley rats fell 52% from the normal (sham-operated) 4.6 ± 0.1/mm to 2.2 ± 0.1(p<0.01). When hPTH-(1-34) (Forteo™) was injected once daily into the OVXed rats starting 2 weeks after the operation and ending 6 weeks later, the mean number of metaphyseal trabeculae had dropped from the normal (sham-operated) value of 4.6 ± 0.1 to 2.0 ± 0.1/mm (p<0.01), which was not significantly different from the 2.2 ± 0.1 in the untreated OVXed rats. However, when Ostabolin-C™ was injected instead of hPTH-(1-34) (Forteo™) the trabecular number dropped by only 28% to 3.3 ± 0.1/mm, which was 1.7-fold higher than the 2.0 ± 0.1/mm (p<0.01)in the hPTH-(1-34) (Forteo™)-treated rats. If the PTH injections were delayed for 9 weeks after OVX when the maximum osteoclast activity had passed, there was no difference between the trabecular numbers in the animals treated daily with either peptide between 9 and 15 weeks. In both kinds of experiment, the two peptides equally and greatly increased the mean trabecular thickness to a value about 1.7 times higher than in the femurs of the sham-operated animal. Thus, it seems from these numbers that Ostabolin-C™ and hPTH-(1-34) (Forteo™) are equally osteogenic, but Ostabolin-C™ is the better osteoclast suppressor or killer, at least in rats and monkeys.

Why would these Ostabolin™-family peptides be such poor stimulators of bone resorption? On the basis of the available information from cultured cells, it simply makes no sense! To try to answer this question, we must look more closely at what we have learned in the past few paragraphs. RANKL expression is unquestionably stimulated by cyclic AMP/PKA. And all of the osteogenic peptides, including the Ostabolin™ family peptides, activate PTHR1, which stimulates adenylyl cyclase which stimulates RANKL. So why would Ostabolin™ family members be poorer osteoclast stimulators than hPTH-(1-34) (Forteo™)? Why would the potent adenylyl cyclase-stimulator and potently anabolic [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin-C™) actually reduce osteoclast activity in OVXed rats as wells as monkeys?

The reason may stem from two things. First, the density of PTHR1 receptors on an osteoblastic cell depends on its "maturity" (Aubin, 1998, 2000, 2001; Aubin and Triffitt, 2002) (fig. 20). Second, immature progenitors and mature osteoblasts differently express RANKL and OPG (Atkins et al., 2003). RANKL is made by immature marrow osteoblastic cells with relatively few emerging PTHR1 receptors, while the mature osteoblasts with maximum PTHR1 receptor densities do not make RANKL but can make OPG (fig. 8 and fig. 20). As I said back in Chapter 2ii, this maturity-dependent expression of RANKL and OPG is a smart move on Nature's part because the two populations live in different places with different neighbors and jobs to do in the crowded spaces of an osteonal tunnel or a trabecular trench. This means that a PTH will stimulate RANKL expression in immature osteoblasts and OPG rather than RANKL in mature osteoblasts, and the relative levels of expression will depend on the population's maturity/receptor density profile. Thus, PTH-stimulated immature, RANKL-competent osteoblastic cells following osteoclasts in a new osteonal tunnel or in a trabecular trench would help the osteoclasts, but PTH-stimulated, mature, OPG-expressing, working osteoblasts with their RANKL genes shut down (Kitazawa et al., 1999) would stop osteoclasts with OPG to protect the bone they are making.

Now we come to the point. The Ostabolin™ family peptides have lower affinities for PTHR1 than does hPTH-(1-34) (Forteo™), with Ostabolin™ having the lowest value (Whitfield et al., 2000c). This means that during their brief appearance in a single injected bolus, the Ostabolin™ family peptides would be less able than hPTH-(1-34) (Forteo™) to grab onto enough of the fewer PTHR1s to stimulate the immature RANKL makers. But they would have enough time and not be significantly less effective than hPTH-(1-34) (Forteo™) in grabbing enough PTHR1s to stimulate the OPG-making preosteoblasts and mature osteoblasts with many more PTHR1 receptors on their surfaces. However, the advantage should vanish when the peptides are given much more time to grab PTHR1s during continuous infusion. This should be the case for Ostabolin-C™, which has a higher affinity for PTHR1 than Ostabolin™ but still a lower affinity for PTHR1 than hPTH-(1-34)NH₂ (Whitfield et al., 2000c). Indeed, that is exactly what happens. When Ostabolin™ is continuously infused (at a dose of 4 nmoles/kg per day) into a Sprague-Dawley rat, the blood Ca²⁺ concentration rises at about the same rate as in rats receiving hPTH-(1-34)NH² but then stops rising during the next 5 days while the blood Ca²⁺ concentration continues rising in the animals receiving hPTH-(1-34)NH². However, in striking contrast to its total inability to raise the blood Ca²⁺ level when injected intermittently in monkeys (Jolette et al., 2003), continuous Ostabolin-C™ infusion causes the blood Ca²⁺ concentration to rise as steadily and at least as high as, or even higher than, in the rats receiving hPTH-(1-34)NH₂.

A differential response of RANKL-expressing immature and OPG-expressing mature osteoblastic cells could also account for the strange observation of Fujita et al. (1999) that a once-a week injection of 60 μg of hPTH-(1-34) (Forteo)™ significantly increased spinal BMD but at the same time significantly reduced the level of collagen breakdown bits in osteoporotic patients. In this case, a weekly brief bolus of PTH would target most strongly the mature, receptor-rich, OPG-making osteosteoblasts, but it would be too ephemermal to significantly stimulate the immature RANKL-makers.

How an Odd Couple—PTHrP and an Indian Hedgehog—Build Bones

An odd couple, PTHrP and a hedgehog, actually an Indian hedgehog, controls the growth of a long bone such as the femur. Indian hedgehog is one of a family of several proteins that are called hedgehogs, because disabling the single hedgehog gene in the fruit fly Drosophila causes the fly's larvae to have a pattern of cuticular denticles that make it look like a prickly hedgehog (Bier, 2000).

The femur begins as a bone-shaped template, a condensation of mesenchymal cells expressing the master cartilage-inducing transcription factor known as Sox 9 (de Crombrugge et al., 2001). Sox 9 probably stimulates the expression of the genes for the cell surface components needed for chondrocytic condensation (de Crombrugge et al., 2001). Then Sox 9 causes the cells to express L-Sox 5 and Sox 6 that are "architectural" transcription proteins that reconfigure nuclear chromatin structures to enable Sox 9 and other classical transcription factors to turn on chondrocye-specific genes (de Crombrugge et al., 2001). The condensed mesenchymal cells differentiate into chondrocytes that lay out a cartilage model of the future bone and perichondrial cells that surround and enclose the emerging model. The cells at the center of the model stop proliferating, stop expressing Sox 9, and start maturing into terminal hypertrophic (swollen) cells, which calcify the matrix to enable it to be resorbed by osteoclasts and replaced by bone. While this is going on in the center, the model's perichondrium is invaded by capillaries and changes from a perichondrium into a periosteum. In other words, the perichondrial layer spawns osteoblasts that deposit bone on the inner surface of the periosteum to make a hollow cylinder—the bone collar. Meanwhile, the hypertrophic cells' matrix calcification reduces their already low oxygen and nutrient supplies, which causes them to trigger the suicidal apoptotic mechanism. The calcified matrix is removed by "chondroclasts", leaving only trabeculae-like calcified cores or spicules in a primary marrow space. Vascular and perivascular cells from the periosteum then invade this marrow space. The blood vessels produce the marrow sinusoids, and the perivascular invaders generate the bone marrow stromal cells, osteoprogenitors, and the first osteoblasts, which layer bone onto the calcified cartilage cores to make the first trabeculae. The intramembranous (i.e., directly from mesenchymal cells rather than from mesenchymal cells via cartilage) ossification by mesenchymal cells to make the bone collar and the endochondral (i.e., indirect via cartilage) bone formation in the center of the cartilage model together make up the so-called primary ossification center (Kerr, 1999).

As the bone lengthens, the cartilage layers attached to the inner surfaces of the bony epiphyses at both ends of the growing bone are reduced to thin growth plates (also called physes). Now the bone has proximal (nearer the body) and distal (farther from the body) growth plates, which contain chondrocyte stem cells and are the engines that drive the lengthening of the bone. The round, moderately sized chondrocytes at the top of a growth plate nearest the epiphysis, the reserve or resting zone, are the stem or chondroprogenitor cells (R.B. Martin et al., 1998). They are fed with nutrients diffusing down through the cartilage matrix from the blood vessels in the overlying epiphyseal bone. Thus, the core of the growth plate and the columns of cells extending down from it are hypoxic, and to do their jobs they have to turn to HIF-1α and HIF-1β (which we met in Chapter 1) and the large number of oxygen-regulated genes they control to stimulate the genes for the the glucose transporter 1 and the enzymes of the anerobic glycolysis machinery to replace the mitochondrial oxygen-driven ATP production (Schipani et al., 2001; Wenger, 2002). They must stop trashing HIF-1α. When the oxygen level in the core drops below a critical level, the cells' oxygen-driven protein hydroxylases stop working, which means that prolines in their HIF-1αs' oxygen-dependent destruction domains are no longer hydroxylated, and the peptides are no longer polyubiquitinated and dumped into the cells' protein-shredding proteasome carburetor barrels (Marz, 2004; Schipani et al., 2001; Wenger, 2002). The accumulating HIF-1α moves into the nucleus together with HIF-1β and stimulates the transcription of the genes for the components of the glycolytic machinery.

When a stem cell gets the signal to start its proliferative engine and divide, one of its two daughter cells stays home to keep up the self-renewing stem cell tradition, while the other daughter sheds it and leaves home to join the parallel columns or stacks of flat or disk-like "transit amplifying cells", cells in the proliferative zone. Eventually the great granddaughters of the original stem cell daughter reach the end of the proliferative column and plunge into the deadly ionic turbulence of the hypertophic zone (R.B. Martin et al., 1998), where they switch off and dismantle their proliferative engines and start their apoptosis-terminated hypertrophic maturation program in which they increase their volume (i.e., hypertrophy) 5-to 10-fold while increasing their expression of alkaline phosphatase, annexin V, BMP-6, PTHR1 receptors, type X collagen and, at this point, the very important Indian hedgehog (e.g., Zuscik et al., 2002). The purpose of all of this activity is to lay down cartilagenous matrix primed for calcification, which is necessary for osteoclasts to attach to the surface, remove it, and thus enable its ultimate relacement by bone.

Finally, the cells start depositing Ca²⁺-hydroxyapatite crystals on the matrix between the cell columns. The calcification of the cartilage drives the hypertrophic cells to apoptotic self destruction by impeding the diffusion of food supplies and disposal of wastes (there is no lacunocanalicular network in cartilage to efficiently distribute nutrients and dispose of waste via the marrow vasculature as in cortical bone). The dying (apoptosing) cells release proteases, which shred the cartilaginous matrix in the distal segments of the cell columns but leave the calcified matrix between the columns intact. The resulting proximal calcified "stalactites" and distal calcified "stalagmites" are targeted by osteoclasts and osteoblasts and become the foundations of the primary proximal and distal metaphyseal trabeculae. The invasion of these osteoclasts and osteoblasts is enabled by the proliferation of blood vessels driven by leptin-induced proteases and the HIF-1α-induced expression of the chondrocytes' vascular endothelial growth factor (VEGF) genes (Wenger, 2002).

Figure 29

The roles of PTHrP and its friend, Ihh (Indian hedgehog), (more...)

Figure 29

.

The roles of PTHrP and its friend, Ihh (Indian hedgehog), to steer the voyage of chondroblasts from the the stem cell niche to hypertrophic self destruction and cartilage calcification. HT Genes refer to the set of genes that are turned on in the swelling, hypertrophic cell.

fig. 29

fig. 29

fig. 7

But remember what Zuscik et al. (2002) said about Ca²⁺ and PTHRP gene expression. The intracellular Ca²⁺ concentration must be low enough to enable PTHrP expresson. This Ca²⁺ sensitivity resides in the -1498 to the -863 region of the PTHrP gene's promoter (Zuscik et al., 2002). This part of the PTHrP gene's promoter has consensus sites for eleven transactivators (Zuscik et al., 2002). One of these is YY1, a possible silencer of the Ca²⁺-driven binding of which to the PTHrP gene promoter interferes with a positive factor (Roy et al., 1996; Zuscik et al., 2002). Alternatively, one of these eleven may be a transcription promoter, the binding of which to its target site on the PTHrP gene promoter is reduced by nuclear Ca²⁺ just as nuclear Ca²⁺ reduces the binding of the transcription enhancer, CBF/NF-Y, to its target site on the grp78/BiP gene (Roy et al., 1996). In either case, the Ca²⁺ concentration in the peri-articular region must be kept low enough, or the cells' Ca²⁺-sequestering pumps must work hard enough, to keep the nuclear Ca²⁺ level low enough to allow PTHrP to be made in response to Ihh stimulation. The PTHrP from the peri-articular cells then percolates along the stacks of proliferating cells, which deploy increasing numbers of PTHR1 receptors on their surfaces as they get near the deadly, seething, hypertrophic zone (Amizuka et al., 1994; Karaplis, 2001; Lee et al., 1995; Weisser et al., 2002).

fig. 29

fig. 29

fig. 29

The younger proliferating cells push the older chondrocytes into an increasing external Ca²⁺ concentration and a strongly increasing, maturation-driving inorganic phosphate concentration on the approaches to the hypertrophic calcification zone (Wuthier, 1993).

fig. 29

As we learned above when the inorganic phosphate reaches a critical level in cells such as osteoblasts, it turns on the apoptotic mechanism (Meleti et al., 2000), unless stopped by PTH (Allen et al., 2002). The shutdown of PTHrP expression and PTHR1 signaling would thus enable the inorganic phosphate flowing into the hypertrophying cells through the type III Na-phosphate cotransporters to do its ultimately deadly job in collaboration with Ca²⁺ (Magne et al., 2003). The phosphate stimulates key genes such as the type II collagen gene and together with Ca²⁺ drops the anti-apoptosis Bcl-2 level and increases the pro-apoptosis Bax level (Coe et al., 1992; Magne et al., 2003).

The now gasping, hypoxic, hungry HT cells in the increasingly impermeable calcifying matrix, no longer protected by PTHR1 signals and with no longer adequate HIF-1α-driven glycolytic machinery and consequently dwindling glycogen reserves, rapidly emptying ATP fuel tanks, and spluttering ion pumps, are now plunging toward their self-inflicted apoptotic doom in the hypoxia of the lower hypertrophic zone as their internal phosphate level rises menacingly, and they fill their endoplasmic reticular storage vesicles with in-rushing Ca²⁺ and, in a last futile attempt to avoid a Ca²⁺ catastrophe, dump it into their leaking mitochondria (Brighton et al., 1976, 1978; Iannotti et al., 1994; Magne et al., 2003; Mansfield et al., 1999; Rajpurohit et al., 1999; L.N.Y Wu et al., 1995). As the hypertrophs start apoptosing, their membranes sprout blebs with annexin V plug-ins. These blebs are released as 100-nm vesicles, loaded with alkaline phosphatase, Ca²⁺-binding proteins, collagens II and X and proteases, with their membranes studded with channels formed by annexin V through which Ca²⁺ flows and type III Na-phosphate cotransporters through which phosphate flows to form hydroxyapatite crystals on the inner surface of the vesicle membranes. When the vesicles are loaded, their membranes are ruptured by the proteases and the hydoxyapatite needles are dumped into the extracellular fluid, where they trigger a self-sustaining mineral nucleation on the matrix using the ambient normal levels of Ca²⁺ and phosphate (H.C. Anderson, 1995; Bandorowicz-Pikula et al., 2001; Boskey, 1992; Hoshi and Ozawa, 2000; Kirsch et al., 1997, 2000; Plate et al., 1996). The cartilage calcifiers also now express collagen I, which bridges the transition from a mineralized type II/IX/XI collagen cartilage matrix to a type I collagen-rich bone matrix made by the progeny of the Ihh-stimulated perichondrial/periosteal cells (Kirsch et al., 1997).

As these hypoxic chondrocytes have been pushed downward, something else may happen that contributes to their injury and death. Their HIF-1αs have been busily stimulating the production of VEGF, which has been stimulating the extension of blood vessels into the bone to deliver osteoclast and osteoblast progenitors to, respectively, clear away the cartilage and replace it with bone. With these vessels comes oxygen, which instead of being a welcome relief for the hypoxic chondrocytes, delivers the coup de grace because by causing a wave of acetyl CoA from the accumulated lactate to surge into the reawakened mitochondria. This causes the mitochondria to spray deadly ROS (reactive oxygen species) into the cytoplasm, because the cytochrome oxidases are overwhelmed and cannot put all of the flood of electrons safely into water.

Interestingly, elasmobranchs (sharks and rays) appear to use part of the same PTHrP-dependent mechanism to make their cartilagenous skeleton (Trivett et al., 2002). But the final stages are missing. There is no cartilage calcification, destruction, and replacement by bone. Trivett et al. (2002) have suggested that ancestral elasmobranchs actually had the full endochondral bone-making process but lost it along the way. Indeed, there is still some calcification of these beasts' outer notochordal sheath.

Why don't cortical osteocytes strangle themselves like chondrocytes when buried in calcified matrix? The answer is really quite simple—that elegant network of canals, the lacunocanalicular osteointernet. The buried osteocytes are continuously supplied with food and oxygen from the blood flowing through the vessels of the cortical osteonal canals and their waste is carried out and dumped into the blood through their lacunocanalicular network that is connected to blood vessels and nerves, while chondrocytes are solitary, anti-social individuals who pay the price of their splendid isolation by having limited access to marrow blood vessels, which forces them to depend entirely on diffusion through the cartilage matrix to get the most of their supplies (Archer and Francis-West, 2003). While diffusion from the bone vascular "railheads" at the growth plate is at first quite adequate in the proliferative zone, because the matrix is largely water and the cells can operate under the hypoxia prevailing in the core regions conditions by upregulating their HIF-1α (R.B. Martin et al., 1998; Schipani et al., 2001), it becomes lethally inadequate when the matrix is mineralized in the hyertrophic zone.

Disabling the genes for Ihh, PTHrP, or PTHR1 releases the brakes on the proliferative "off-switch", and the cells prematurely rip open the "HT" envelope, commit apoptosis, and the bone under construction is abnormally short, and its owner is a dwarf (Amizuka et al., 1994; Karaplis, 2001; Karaplis et al., 1994; Lanske et al., 1996; St-Jacques et al., 1999). However, if the Ihh gene is still working, accelerating maturation, and therefore the production of Ihh-making prehypertrophic cells by disabling the PTHrP or PTHR1 genes, increases Ihh production by the prehypertrophic cells. The increasing Ihh tries to make up for the chondrocyte shortage caused by the premature maturation by increasing the flow of peri-articular cells into the dividing columns (T. Kobayashi et al., 2002). On the other hand, uncontrolled, persistent PTHR1 signaling delays the proliferative switch-off and maturation which would tend to reduce the appearance of Ihh-making prehypertrophic cells and thus slow up the flow of cells into the dividing columns (Schipani et al., 1995, 1997; Weir et al., 1996). Excessive Ihh production also inhibits the initiation of maturation by swiftly turning up the peri-articular cells' PTHrP production, which in turn would tend to feed back and reduce the appearance of Ihh-making prehypertrophic cells (T. Kobayashi et al., 2002; Minina et al., 2001; Vortkamp et al., 1996). Finally, Ihh-stimulated chondrocyte proliferation together with persistently firing, maturation-blocking PTHR1 receptors could be a deadly combination—it could cause tumors. And indeed it does! Hopyan et al. (2002) have found an abnormally signaling PTHR1 receptor in the cells of benign cartilage tumors from a human suffering from endochondromatosis (i.e., Ollier and Maffucci diseases).

Before leaving the cartilage and its chondrocytes, one might ask whether the chondrocytes in a permanent cartilage tissue such as articular (joint) cartilage, as opposed to the ephemeral cartilage in a growing femur, are equipped like osteocytes to measure and appropriately respond to strain. Articular chondrocytes, unlike osteocytes, do not form intercommunicating networks with direct access to blood vessels. They rely on diffusion of nutrients through the matrix. Therefore, they must function with very low oxygen levels. This means that they must use anerobic glycolysis instead of aerobic respiration to make their ATP fuel and thus do not have the large mitochondrial complements of osteoblasts and osteoclasts (Archer and Francis-West, 2003). Nevertheless they obviously do their job. To do this, they express the stress-reducing, function-maintaining, hypoxia-inducible factor compnents (HIF-1α and HIF-1β) and use the HIF-1α•HIF-1β pair and the AP-1 transcription factor complex to drive various gene expressions such as the genes for TGF-βs and for the components of the glycolytic mechanism (Archer and Francis-West, 2003; Bilton and Booker, 2003; L. E. Huang and Bunn, 2003; Lando et al., 2003). Moreover each articular chondrocyte has a single cilium which is pulled back and forth by attached collagen fibers and very likely sends Ca²⁺ signals to tell the cell to respond appropriately, when the cartilage is bent and stretched by joint bending and twisting (Archer and Francis-West, 2003; Gillespie, 1999; C. Jensen et al., 2004; Meier-Vismara et al., 1979; Poole et al., 1997, 2001; Whitfield, 2003a; Wilsman, 1978; Wilsman and Fletcher, 1978). But there is a huge difference between an articular chondrocyte and an osteocyte. The signal from a collagen-fiber-pulled cilium of an anti-social chondrocyte would not spread to its neighbors as would a signal triggered by the bending cilium flowmeter of an extensively networking osteocyte (Whitfield, 2003a).

Hopefully, the reader would like to take a quick glance at another of the many examples of PTHrP controlling tissue development—in this case, the skin epidermis. (We must not forget that some of our ancient predecessors had boney armorplates in their skin, although these were the products of neural crest epithelial cells (Smith and Hall, 1990).) Here, too, we see the same pattern as in growing bone—stem cells giving rise to rapidly proliferating transit amplifying progeny which stop proliferating, mature, and die in a curious hybrid process we have called diffpoptosis (differentiation and apoptosis) (Whitfield and Chakravarthy, 2001). The cycling transit amplifying basal keratinocytes start making and deploying PTH/PTHrP receptors that respond differently from the PTHR1 receptors of chondrocyte and osteoblast—normal keratinocytes may (Errazahi et al., 2003) or may not (Orloff et al., 1995; Whitfield and Chakravarthy, 2001) express their gene for the conventional PTHR1. At any rate, the signals from this receptor, whatever it may be, do not include adenylyl cyclase activation, although the cells can respond with a large burst of adenylyl cyclase activity to the β-adrenergic receptor-activating isoproterenol or to PTH-(1-34) when they have been artificially engineered to express PTHR1 (Orloff et al., 1995; Whitfield et al., 1992). However, the keratinocytes do not make the PTHrP to stimulate these unconventional PTH receptors until they lift off the basal lamina; otherwise, the PTHrP would interfere with their proliferation. Indeed, Menon et al. (1991) have shown that the proliferating basal cells are the only keratinocytes with substantial amounts of internal Ca²⁺, which according to the findings of Zuscik et al. (2002) should prevent PTHrP expression. These cells would be generating nuclear Ca²⁺ transients to start replicating their chromosomes and then to enter and complete mitosis (Whitfield and Chakravarthy, 2001). Then when the cells lift off the basal lamina, they stop proliferating, nuclear Ca²⁺ drops, and PTHrP is expressed (Juhlin et al., 1992; Menton et al., 1991; Whitfield et al., 2001). The amount of PTHrP coming down from the suprabasal cells tells the basal cells how many suprabasal cells are there and accordingly controls the rate of proliferation—a lot of cells in the stack means a lot of PTHrP coming down and a reduced production of basal cells, while a depleted cell stack means less PTHrP coming down and increased basal cell proliferation (Whitfield, 2004). However, the PTHrP expression in the lifted cells does not last. The external Ca²⁺ concentration triples or more when the cells rise through the granular layer and into the lower corneum and, as happens at the threshold of the hypertophic cartilage zone, this is enough to shut off PTHrP expression and trigger diffpoptosis, which converts granular keratinocytes into cornified corpses (Whitfield, 2004; Whitfield and Chakravarthy, 2001).

Before going any further, we must quickly look at what causes some basal stem cell daughters to lose their "stemcellness" and start differentiating. When these daughters reach a certain point on the border of the stem cell niche, they stop expressing Jagged or Delta molecules but express a receptor for these molecules known as Notch. This means that the Jagged or Delta on their stem cell sisters will bind to Notch, the signals from which causes the borderland daughters to become transit amplifying cells each of which produces a limited number of terminally differentiating offspring (Whitfield, 2004).

If the PTHrP from the suprabasal cells should also reach the dermal fibroblasts, it would activate their normal adenylyl cyclase-coupled PTHR1 receptors. This could cause the fibroblasts to make and maybe release, among other things, Jagged-1as happens in bone marrow (Calvi et al., 2004). This Jagged-1 (the skin's most expressed Notch ligand) would stimulate the Notch receptors on the overlying keratinocytes and drive their differentiation (Lowell et al., 2000; Nickoloff et al., 2002; Nicolas et al., 2003; Okuyama et al., 2004; Rangarajan et al., 2001; Thélu et al., 2002). This could be another way the amount of suprabasal PTHrP could control keratinocyte production. It would also mean that applying hPTH-(1-34) to the skin could reduce keratinocyte proliferation and drive differentiation (Whitfield, 2004).

It follows that a failure of keratinocytes to express PTHrP would cause a massively dysregulated cell production like that in psoriasis (Juhlin et al., 1992). The beefy red psoriatic patch is produced by the hyperproliferation of keratinocytes, which causes a pile up of cornified squames with an impaired ability to slough off (desquamate). It appears that the reason for the psoriatic hyperproliferation is underexpression of the Jagged-Notch-Fringe signaling system, which is upregulated in the basal and differentiating layers of the normal epidermis and maintains the normal relation between proliferation and differentiation (Thélu et al., 2002). The hyperproliferation of the basal keratinocytes and the suppression of the Delta-Notch control mechanism is somehow triggered locally by bacterially or otherwise activated Langerhans dendritic cells and driven for example by IL-8 from neutrophils streaming out of local blood vessels (Nickoloff et al., 2000). The keratinocytes in turn secrete autocrine (autostimulating)/paracrine (neighbor-stimulating) NGF that prevents them from self destructing while stimulating the T cells, which have been directed to the battle zone by the activated Langerhans cells coming to the lymph nodes from the skin, and recruiting other factor-spewing inflammatory cells (Lebwohl, 2004; Nickoloff et al., 2000). And the keratinocytes also emit NO that causes the local dermal blood vessels to dilate, which gives the patch its swollen redness and enhances the movement of activated Langerhans cells to the nodes, T-cells, and neutrophils to and from the battle zone (Nickoloff et al., 2000). This torrent of cytokines and other factors results in the basal layer having a proliferation-favoring, lower than normal, level of extracellular Ca²⁺ and the suprabasal cells having an abnormally high extracellular Ca²⁺ level, which could be why suprabasal psoriatic keratinocytes do not make PTHrP, the proliferative terminator (Holick et al., 1994; Menon and Elias, 1991; Zuscik et al., 2002).

Is this lack of PTHrP the reason for the excessive proliferation? Indeed, it seems to be! Topically applying an adenylyl cyclase-stimulating PTH such as hPTH-(1-34) or one of the newer smaller Ostabolin™ Family PTHs, but not hPTH-(7-34) which cannot stimulate adenylyl cyclase, stops the wild cell production and eliminates psoriatic plaques (Holick et al., 2003). How? Keratinocytes don't have a PTH/PTHrP receptor that activates adenylyl cyclase! The answer may lie with the dermal fibroblasts which do have the adenylyl cyclase-coupled PTHR1 and the ability of the closely related PTHR1-bearing, bone marrow stromal cells to make large amounts of Jagged-1 when treated with hPTH-(1-34) (Calvi et al., 2004; Whitfield, 2004). Such jagged-1 could restore order to the disturbed keratinocytes by reviving their Notch signaling-driven differentiaion mechanism (Lowell et al., 2000; Nickoloff et al., 2002; Nicolas et al., 2003; Okuyama et al., 2004; Rangarajan et al., 2001; Thélu et al., 2002; Whitfield, 2004).

PTHs and Rheumatoid Arthritis

In rheumatoid arthritis, a systemic autoimmune disease, immune-response cells such as macrophages, polymorphonuclear leukocytes, and T-lymphocytes invade the joints and start pumping out cytokines that produce an inflamed synovium, the membrane of the sac that encloses the fluid that lubricates the joints. In mice and rats, this cytokine storm can be started by injecting type II collagen intradermally, along with Freund's adjuvant, to induce an autoimmune reaction to the animal's own cartilage (Mori et al., 2002). Cytokines such as IL-1 and interferon-γ from this autoimmune response cause the formation of clusters of tall endothelial cells in the postcapillary venules of the synovial membrane that serve as the ports of entry of lymphocytes that form perivascular lymphoid aggregates or cuffs around the blood vessels like those of tonsil-like lymphoid tissues (Iguchi and Ziff, 1986; Ziff, 1993). These lymphocytes plus macrophages and plasma cells swarming through the layer of flat endothelial cells of the post-capillary venules release a swarm of cytokines which includes the mighty TNF-α, that pushes the synovial fibroblasts to proliferate and make things such as PTHrP and RANKL, which stimulate osteoclast generation and with it bone destruction (Funk, 2001; Funk et al., 2002; Goldring, 2002; Goldring and Gravallese, 2000, 2002; Gravallese and Goldring, 2000; Iguchi and Ziff, 1986; Mori et al., 2002). The surface of the underlying bone is primed for an osteoclast onslaught by proteases from both the invading leukocytes and the proliferating synovial fibroblasts. It becomes the site for the generation of osteoclasts from invading precursors that is driven initially by RANKL from the invading T-lymphocytes but later mostly from the cytokine-crazed, proliferating synovial fibroblasts (Goldring, 2002; Goldring and Gravallese, 2000, 2002; Gravallese and Goldring, 2000; Mori et al., 2002).

The synovium now relentlessly thickens into a predatory, giant ameba-like, bone-eating blanket or pannus (Latin for cover), a tumor-like blanket of inflammatory factor-secreting granulation tissue made up of polymorphonuclear leucocytes and lymphocyte invaders and the proliferating fibroblasts which vigorously make PTHrP that, like TNF-α, stimulates RANKL expression by the underlying osteoblastic cells (Funk et al., 2002; Goldring, 2002; Goldring and Gravallese, 2000, 2002; Gravallese and Goldring, 2000; Mori et al., 2002).

It must clearly be understood that inflammation and its associated cytokines do not by themselves cause the osteoclast attack and bone loss—something special is needed. This special something is the RANKL from T-cells and synovial fibroblasts as well as bone marrow stromal cells and osteoblasts (Bolon et al., 2002a, 2002b). Therefore, mice without functional RANKL genes can still develop severely inflammed joints in an autoimmune response response, but without RANKL no osteoclasts arrive on the scene to dig holes in the bones (Goldring, 2002; Goldring and Gravallese, 2000). However, the IL-1 cytokine from the immunocytes does stimulate osteoclast generation by its sharing with RANKL the ability to stimulate NF-κB in osteoclast precursors (Xing et al., 2003).

As the pannus grows, it, like a tumor, must make angiogenic factors such as VEGF (vascular endothelial growth factor) to make new blood vessels to get its oxygen, nutrients and along with these more blood-borne cytokine-producing lymphocytes, macrophages, neutrophils, and osteoclast precursors to help drive it forward (Funk et al., 2002). The endothelial cells of the pannus's blood vessels make PTHrP, but, unlike bone vascular endothelial cells, they don't have the PTHR1 receptors to respond to it, although they can ship some of it into their own nuclei instead of outside the cell to stimulate genes the products of which can prevent them from committing apoptotic suicide (Funk et al., 2002; Lam et al., 2002). The PTHrP from the endothelial cells hits the underlying vascular smooth mucle cells, which do have PTHR1 receptors (Funk et al., 2002). The cyclic AMP signals from the PTHR1 receptors cause the smooth muscle cells to relax and the blood vessels to dilate and thus carry more blood into the advancing pannus.

The steady pounding from the rising tide of RANKL-stimulating PTHrP from the pannus cells crawling over the joint causes the excessive osteoclast generation and digging to spread from the superficial subchondral bone to endosteal bone, which can eventually spread systemically to osteoporosis (Fu et al., 2002; Funk, 2001; Funk et al., 2002).

Obviously, the bone erosion by osteoclasts in a rheumatoid arthritic joint can be stopped, for example, with the very long-acting OPG, the RANKL decoy ligand, that prevents mature osteoclasts from setting up their sealing ring and resorbing bone by directly binding to a 140-kDa protein on their surfaces and takes RANKL out of action, which prevents osteoclast generation by depriving osteoclast precursors of signals from their RANK receptors (Bolon et al., 2002a, 2002b; Goldring, 2002; Goldring and Gravallese, 2002; Hakeda et al., 1998). Alternatively, osteoclasts can be disabled and killed by one of the bisphosphonate gang or calcitonin. It seems likely that a short treatment with single daily shots of one of the osteogenic PTHs, particularly hPTH-(1-31)NH₂ (Ostabolin™) or [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin-C™), that are very poor stimulators of osteoclasts (Fraher et al., 1999; Jolette et al., 2003; Mohan et al., 2000; Whitfield, 2003b) should be able to do something that OPG or the antiresorptives cannot do—regrow the bone in the osteoclast-gnawed joints and as well as restore bone lost throughout the skeleton because of the spread of osteoclastogenic factors from the damaged joints, the lack of activity due to severe joint pain, and the use of osteoporosis-inducing glucocorticoids (Lane et al., 2000; Whitfield, 2003b). Indeed, the very promising first indications of this have been provided by Fukata et al.'s (2004) report that intermittent hPTH-(1-34) injections (3 times per week for 4 or 6 weeks) can prevent or stop the drop in cortical and trabecular BMD, increase trabecular thickness, and increase the trabecular mineral apposition rate, i.e., increase trabecular bone formation, in the tibias of female Sprague-Dawley rats with collagen-induced arthritis. But the PTH treatment did not affect the arthritis itself.

PTHs and Cancer—A Concern with All Growth Stimulators

Figure 30

How a lifelong barrage of PTH boluses can induce osteosarcoma. (more...)

Figure 30

.

How a lifelong barrage of PTH boluses can induce osteosarcoma. Basically, the signals from PTHR1 receptors and the release of the several missionary growth factors listed in the panel (note that PTHrP Cfs refers to cleavage products of the PTHrP chain of factors) from preosteoblasts and mature osteoblasts drive the proliferation of immature osteoprogenitors. Over time, mutations happen in the continuously proliferating cells and accumulate (as symbolized by the progression of nuclear circles) up to the point where cells appear that continuously make their own autocrine growth factors, which lock them into the cycling mode (cycle lock ) and prevent them from shutting down their cell cycle machinery and mature. At some point in this carcinogenic promotion, they also lose their ability to check for, and eliminate, mutations or self destruct if a mutation can't be eliminated. This loss causes the mutation load in previously arrested and discarded, but now surviving, cells to freely generate metastatic clones and cancer. Abbreviations: AGF, autocrine (autostimulating) growth factors; OB, mature osteoblasts; Op, juvenile osteoprogenitor; PreOb, preosteoblast.

fig. 30

fig. 30

This appears to be what happens in a rat subjected to a life-long treament with a large dose of rhPTH-(1-34) (Forteo™). At the July 27, 2001 meeting of the FDA Endocrinologic and Metabolic Drugs Advisory Committee, it was revealed that a lifelong treatment with Lilly's rhPTH-(1-34) (Forteo™) is a potent osteosarcoma inducer in both female and male Fisher rats (Center for Drug Evaluation and Research, 2001; Vahle et al., 2002). For example, subcutaneously injecting 75 μg of Forteo™/kg of body weight each day into 60 rats starting when they were 6 to 8 weeks of age and continuing for 2 years increased the incidence of osteosarcoma from the 0.2 to 0.4% in the untreated control animals to 40% in the treated females and 60% in the treated males. At 13 months, tumors began appearing at several sites throughout the skeleton and then spread to various soft tissues in about one-third of the animals.

At first glance, this seems to mean that PTH treatment is downright dangerous. But it really is not! This kind of demonstration is ludicrously unrealistic. The daily dose of the fragment given to the rats (75 μg/0.4-0.6 kg of body weight) was about 400 times the 20 μg/40-60 kg of body weight recommended for treating osteoporosis (Center for Drug Evaluation and Research, 2001; Neer et al., 2001; Vahle et al., 2002). Although the extensively proliferating rat and human osteoprogenitors obey the same rules, they are probably more genomically unstable in an inbred rat than in a "wild-type" human, as indicated by the very high basal osteosarcoma incidence of 2000-4000 x 10^-6 in the Fisher rats compared to the 1000-fold lower (3-5 x 10^-6) incidence in humans (Gurney et al., 1995). Moreover, the short-lived, small-boned rats keep growing throughout their lives (though at a decelerating rate and with a decreasing osteogenic responsiveness to PTH (Walker, 1971)) and therefore maintain a substantial pool of immature osteoprogenitors for continuous bone making. But because the long-lived, large-boned humans close their bones' growth plates and stop growing, the size of their osteoprogenitor pools are limited to providing osteoblasts for the microcrack-repairing BMUs. Indeed, the osteosarcoma incidence in humans peaks in growing children around 13-15 years of age and then virtually vanishes after growth stops (Gurney et al., 1995), while a rat, especially a male rat, stays equivalent to a human child throughout its life. Neither the humans nor 80 cynomolgus monkeys treated in the trials with 8 times the human dose of Forteo™ for 18 months developed osteosarcoma (Center for Drug Evaluation and Research, 2001). Of course, if osteosarcomas are as rare in monkeys as in humans, the Lilly Forteo™ team would have needed a huge number of monkeys to detect an increased osteosarcoma incidence.

Of course, the fallacy of extrapolating the effects of a lifelong treatment with PTH in rats to humans is that a large dose of the peptide would never be given to a human for 60-90 years—only 20 μg will be given daily for no longer than 2 years! And in a talk (unfortunately unaccompanied by a hard copy record) given at the 24^th meeting of the American Society for Bone and Mineral Research in San Antonio Texas, J. Vahle (Eli Lilly and Company) reported that when the rats were given 30 μg of hPTH-(1-34) for 2 years, 27.5% of them developed osteosarcomas, but when they were given the same dose for only 6 months between the ages of 2 and 8 months or between 6 and 12 months, the osteosarcoma incidence was only 3%, which was virtually the same as in untreated rats. In other words, a treatment with hPTH-(1-34) that was short enough not to trigger osteosarcoma was far more than enough to strongly stimulate bone growth. But we did not know whether a clinically realistic PTH treatment can promote or enhance the effectiveness of real carcinogens.

We were briefly reassured by the fact that none of 12,000 chronically hyperparathyroid Swedes has developed osteosarcoma (Center for Drug Evaluation and Research, 2001). But they didn't look hard enough—as of the first half of 2003, there were reports of four women with parathyroid adenomas and high blood concentrations of PTH, two of whom are Swedes! Wiig et al. (1971) reported two cases of hyperparathyroid Swedish women, a 46-year-old with tibial dedifferentiated chondrosarcoma and a 34-year-old with a high-grade mandibular osteosarcoma.Smith et al. (1997) described the case of a 49-year-old woman with a surgically removed parathyroid adenoma and multiple malignant bone tumors. The latest is a recent report by Betancourt et al. (2003) of a 69 year-old black woman who had a parathyroid adenoma that had raised the circulating PTH concentration above 1000 pg/ml (a normal concentration would have been 10-65 pg/ml), and along with this, she had developed a high-grade fibroblastic osteosarcoma. These osteosarcomatous women are very significant because they were all much older than than the usual osteosarcoma patient, who is 15 or less years old.

But there could be another danger—colon cancer! The proliferatively active colon cells, like osteoblasts, also have PTHR1 receptors (Gentili et al., 2001, 2002; Li et al., 1995). Signals from these receptors in rat duodenal cells rapidly (within just 5 minutes) cause a c-Src-kinase-mediated phosphorylation and activation of phosphatidylinositol-3 kinase, which in turn triggers a mitogenic ERK1•ERK2 cascade (see Pearson et al., 2001 for a review of these proliferation-stimulating enzymes) in rat intestinal cells (Gentili et al., 2001, 2002) just as the signals from PTHR1 receptors attached to NHERF-2 scaffolds do in PS120 fibroblasts (Mahon and Segre, 2002). Therefore, boluses of PTH could stimulate colon cell proliferation and at least promote colon carcinogenesis. Indeed, there are scattered reports of primary hyperparathyroidism with its 2-10-fold higher level of circulating PTH-(1-84) being associated with increased colon cancer incidence in humans (Artru et al., 2001; Conteas et al., 1988; Farr, 1976; Feig and Gottesman, 1987; Kawamura et al., 1999; Strodel et al., 1988).

It seems that colon cancer may not be a problem in rats. The lifetime treatment of Lilly's Fisher 344 rats with huge doses of hPTH-(1-34) (Forteo™) seems to have only induced osteosarcomas which could spread to soft tissue (Vahle et al., 2002). But we still had to find out whether a realistic PTH treatment, i.e., one that was just long enough to stimulate bone growth but no longer could either by itself trigger colon carcinogenesis or enhance the triggering of carcinogenesis by a colon carcinogen. In collaboration with Dr R.P. Bird of the University of Manitoba, we (Whitfield et al., 2003) have tested the ability of a strongly osteogenic 6-week treatment (i.e., once-daily injections of 20 nmoles/kg of body weight) of Sprague-Dawley rats with two PTH fragments, hPTH-(1-34) and [Leu²⁷]cyclo(Glu²²-Lys²⁶)hPTH-(1-31)NH₂ (Ostabolin C™), starting 9 weeks after OVX, to initiate colon carcinogenesis or increase the initiatory activity of two pretreatment injections (15 mg/kg of body weight) of a potent colon carcinogen—AOM (azoxymethane). The initiation of colon carcinogenesis by AOM is indicated by the appearance within just 6 weeks of as many as 263-273 of strikingly visible and very easily counted ACFs (aberrant crypt foci) per colon, a few of which if we had waited long enough would have sprouted tumors (Bird and Good, 2000; Whitfield et al., 2003). (In other words, one does not have to wait very long to find out whether an agent can trigger colon carcinogenesis, but of course it would be necessary to wait for 2 years to know how many of the aberrant crypt foci could generate tumors (Bird and Good, 2000)). While the daily PTH injections stimulated femoral bone formation, they did not cause the appearance of ACFs or affect the induction of ACFs by AOM or accelerate the progression of primary single ACFs to multiple ACFs (Whitfield et al., 2003). Nor did AOM affect the PTHs' abilities to stimulate bone formation (Whitfield et al., 2003).

While a PTH treatment that is no longer than needed to strongly stimulate bone formation does not by itself initiate colon carcinogenesis or enhance or accelerate the initiation of colon carcinogenesis by a potent carcinogen in rats, we can't lower our guard yet. There is a remote danger that a human (or primate)-type colon may respond differently to PTHs, as indicated by the seemingly increased incidence of colon cancer due to the continuous PTH stimulation in hyperparathyroid humans. Clearly, we must find out whether PTHs stimulate ACFs and ultimately tumor formation, at least in the colons of nonhuman primates.

Statins Reduce Fracturing

Bauer et al.(1999) reported the results of an analysis of the data from a study of osteoporotic fractures (8412 women 65 years and older) and a fracture intervention trial (6459 women aged 55 to 80 years). Most of the 599 statin users among the 14871 women in the two studies had been given lovastatin. The relative risk of hip fracture was only 0.30 (95% confidence interval (CI), 0.08-1.18), and the overall relative risk of nonspine fractures was 0.83 (95% CI, 0.61-1.15).

Chan et al.(2000) reported that according to the records from six U.S. HMOs (health maintenance organizations), the relative risk of fracturing for patients who had had more than 13 dispensings of statins was 0.48 (95% CI, 0.27-0.83). Meier et al. (2000) found the same from 300 practices in the UK-based General Practice Research Database. Their base population was 91,611 patients who were at least 50 years old. Of these, 28,340 were taking lipid-lowering drugs. The adjusted odds ratio for fracturing in statin users relative to controls was 0.55 (95% CI, 0.44-0.69). Other lipid-lowering drugs did not affect the odds ratio. L. Wang et al. (2000) found similarly lowered fractured risks in the records of 6,110 New Jersey residents aged 65 years or more. Statin users had a relative hip fracture risk of 0.54 (95% CI, 0.36-0.82). Chung et al. (2000) examined the records of 69 type 2 diabetics, 33 of whom had not been given statins and 36 of whom had been given lovastatin, pravastatin (which would not have affected bone cells), and simvastatin. The spinal BMD rose 2.9% in the statin-treated group during 14 months. During the same time, the BMDs in the different regions of the femur and in the total hip increased by a significant 0.88-2.32% in the statin-treated men but by only 1.8% in the statin-treated women. Finally, M.H. Chan et al. (2001) have reported that giving simvastatin (20 mg/day for 4 weeks) to 17 hypercholesterolemic, nonosteoporotic patients increased one indicator of bone formation, the serum osteocalcin concentration. But it did not increase another formation indicator, the bone-specific alkaline phosphatase activity.

Pasco et al. (2002) have reported the results of the small-scale "Geelong" (Geelong, southeastern Australia) osteoporosis study. In this study, there were 16 statin users in the fracture group and 53 users in the nonfracture group with the fracture odds for the statin users (adjusted for BMD) at the femoral neck, spine, and whole body being, respectively, 0.45(95% CI, 0.25-0.80), 0.42 (95% CI, 0.24-0.75), 0.43 (95% CI, 0.24-0.78) of the values in the nonusers. Adjusting for age, weight, medications, and "life-style" factors did not "substantially" change these facture odds. Surprisingly, in view of the statin users' large, 60% reduction in fracture risk, there were at most only marginal (in fact, nonsignificant) increases in BMD in the femoral neck, spine, and whole body.

Montagnani et al. (2003) have treated 30 postmenopausal, hypercholesterolemic women (of average age, 61.2 ± 4.9 years) with simvastatin for one year. They found small increases in the femoral and lumbar BMDs by the end of the treatment, which was significantly higher than in the postmenopausal control women with normal blood cholesterol levels.

It seems from these reports that the statins are indeed safe, effective bone anabolics or pseudo-anabolics that can reduce the risk of hip and spine fractures by as much as 71%, depending on the extent of use. However, there are other reports, which just as convincingly indicate that lipophilic statins do not stimulate bone growth or affect the fracture risk.

Statins Don't Reduce Fracturing

A problem with retrospectively assessed effects of giving statins on fracturing is that one statin, the popular hydrophilic pravastatin, does not affect cultured bone cells or stimulate bone growth in rodents (Garrett et al., 2001). This explains the findings of Reid et al. (2001) on the effects of long-term pravastatin treatment on fracturing in 9,014 patients with ischemic heart disease. There were 107 pravastatin-treated (40 mg/day) patients admitted to hospital for fractures compared with 101 in the placebo-treated group (fracture risk, 1.05 (95% CI 0.80-1.16)). When fractures not requiring hospital admission were also included, the fracture risk was 0.94 (95% CI, 0.77-1.16).

LaCroix et al. (2003) searched the huge database in the Women's Health Initiative Observational Study (WHI-OS) and found no reducton of fracturing in statin (atorvastatin, lovastatin, pravastatin, fluvastatin)-treated women. In a study funded by Proctor and Gamble, van Staa et al. (2000) also found that cholesterol-lowering doses of atorvastatin, pravastatin, and simvastatin did not significantly lower the fracture risk. This was a much more extensive analysis of the UK General Practice Research database than that of Meier et al., which indicated that statins nearly halved the fracture risk. van Staa et al.'s conclusion that were was no reduction was based on a population of 216,062 fracture patients from 686 general practices, while Meier et al.'s conclusion was based on 91,611 fracture patients from "only" 300 practices. Sirola et al. (2001) and Solomon et al. (2001) could also find no evidence in the clinical databases that statins significantly lower fracture risk. Finally, Cauley et al. (2000) used the WHI-OS database to find out whether statins could enhance the BMDs of post-menopausal women. They found that although statin treatment did not affect fracture risk, treatment with a high-potency (based on cholesterol-lowering ability) lipophilic statin, such as atorvastatin or simvastatin for more than 3 years, only modestly protected hip and lumber vertebral BMD. Edwards et al. (2000) also found that treatment of 41 women with various statins (27 received the high potency simvastatin or atorvastatin) did not reduce fracturing, although there were 8.5 to 12% increases in the BMDs of hip and spine.

Cosman et al. (2001), like M. H. Chan et al. (2001), have carried out a small, short-term experiment on 14 postmenopausal women (mean age of 58 years) designed to find out what a 12-week treatment with 0.4 mg of cerivastatin/day might do to bone instead of serum cholesterol levels. Bone formation markers (type I procollagen propeptide and osteocalcin) did not change, but the resorption markers (urinary N- and C-terminal telopeptides) did drop slightly (<20%) within 6 weeks in the cerivastatin-treated group. They concluded that this statin did not detectably stimulate bone formation, but it might have had a modest bisphosphonate-like antiresorptive action. Moreover, the very small increase Montagnani et al. (2003) found in BMD and bone alkaline phosphatase activity in 30 hypercholesterolemic postmenopausal women treated for 1 year with simvastatin could have been due to as much to osteoclast inhibition as to direct stimulation of bone growth.

Rejnmark et al. (2004) have reported the results of an experiment to determine whether simvaststin can stimulate bone formation. Eighty-two otherwise healthy, osteopenic or osteoporotic postmenopausal women were randomized for daily treatment for 1 year (52 weeks) with 40 mg of simvastatin/day or placebo. None of the women were hyperlipidemic, but simvastatin reduced the serum cholesterol by an average 27%, but by 26 weeks after the end of the treatment, the cholesterol level had returned to the baseline value. This obviously effective dose of the statin did not affect indicators of bone formation (bone-specific alkaline phosphatase or osteocalcin) or indicators of bone resorption. Nor did it significantly affect the BMD of lumbar spine or hip. There was, however, a small increase in the forearm BMD in the treated group. The authors conclude that "our results do not support a general beneficial effect of simvastatin on bone".

However, statins are rather effective anti-osteoclast agents because they, like N-bisphosphonates such as alendronate (Fosamax™), can prevent the synthesis of the so-called isoprenoids farnesyl diphosphate and geranygeranyl diphosphate. Indeed, statins are better than N-BPs at inhibiting bone resorption by rabbit osteoclasts and by osteoclasts in mouse calvarial cultures (Rogers, 2003). These lipids are tacked by appropriate transferases onto the C-terminal cysteine residue of each member of the small GTPases of the Ras, Rho, and Rab families (Coxon and Rogers, 2003). This enables the "isoprenylated" GTPases to drive various osteoclast functions such as podosome formation, formation of the sealing ring, the transport to the apical ruffled border of key components, the fusion of the acid pumps to the ruffled border membrane, and the prevention of apoptosis (Coxon and Rogers, 2003). Obviously, the prevention of the GTPases' isoprenylation when a statin inhibits an osteoclasts's HMG-CoA reductase will throw a very large "monkey wrench" into the digger's machinery with disastrous consequences, including death for the digger but happily respite for the bone.

Vitamin Ds

In 1998, Sicinski et al. reported the synthesis of new 1α,25-(OH)₂-19-nor-vitamin D₃ analogs such as 2MD (2-methylene-19-nor-(20S)-1α,25(OH)₂D₃), which are super-potent derivatives of the natural 1α,25-(OH)₂-vitamin D₃. Four years later, Shevde et al. (2002) reported that 2MD is a far more potent stimulator of osteoblast activities than 1α,25-(OH)₂-vitamin D₃. Indeed, Shevde et al. (2002) have written "…there is little evidence that vitamin D plays a direct role in new bone formation, apart from its action to maintain calcium and phosphorus levels in the blood ". But this osteoblast stimulation is compromised by 2MD being a 100-fold better stimulator of the RANKL gene (which has a vitamin D-responsive site in its promoter (Kitazawa et al., 2003)) and down-regulator of osteoprotegrin expression by presumably immature osteoblastic cells than the natural hormone. The result of this is the formation of more, bigger, and better bone-digging osteoclasts, which make 2MD a 30-fold more effective mobilizer of bone Ca²⁺.

As might be expected from this, orally giving 18.7 pmoles of 2MD/kg of body weight to old OVXed retired breeder rats for 7 days a week for 23 weeks increased total bone mineral density to a mean value 7% higher than in sham-operated rats and 9% higher than in untreated OVXed rats (Shevde et al., 2002). The steroid actually increased femoral trabecular bone volume. And, most important, it didn't do this by decreasing bone resorption, but by actually increasing the rate of bone formation which 1α,25-(OH)₂-vitamin D₃ can't do (Shevde et al., 2002). However, Shevde et al. did not say, by how much, 2MD increased the formation rate, which would be a surprising omission if the stimulation had been significantly large.

How might 2MD stimulate osteoblast activity? 1α,25-(OH)₂-vitamin D₃ activates two differently sited receptors (Boland et al., 2002; Farach-Carson, 2001; Farach-Carson and Xu, 2002; Nemere and Farach-Carson, 1998). One of these is the slowly acting, classical nVDR which operates in the nucleus and stimulates various genes and requires 1α,25-(OH)₂-vitamin D₃'s 1α OH group to be activated while the other, somehow related receptor—the almost instantly responding, fast-acting mVDR—is plugged into the osteoblast's cell membrane and does not need the 1α,25-(OH)₂-vitamin D₃'s 1a OH group to be activated. The rapidly triggered signals from the mVDR are similar, if not identical, to those from activated PTHR1—they include G-protein-coupled activations of c-Src protein-tyrosine kinase, adenylyl cyclase and phospholipases A, C, and D, which open Ca²⁺ channels and stimulate the release of Ca²⁺ from internal stores and activate PKCs, and set off MAP kinase cascades that stimulate various genes (Boland et al., 2002; Nemere and Farach-Carson, 1998). The fact that 2MD does not have a greater affinity for the nVDR (Shevde et al., 2002) suggests that the analog might have a greater affinitiy for the mVDR than 1α,25-(OH)₂-vitamin D₃. This would mean that 2MD is actually a PTH mimic! This also suggests that other analogs, such as 25(OH)-16ene-23yne-D₃ (Farach-Carson, 2001), which cannot activate nVDR because it lacks 1α (OH), should also be PTH-like bone anabolics.

2MD certainly looks like a good candidate for admission to the exclusive Bone Anabolics Club. It has the great marketing advantage of being an oral drug. But it is also unfortunately a super osteoclast generator in rats, which raises the worrisome specter of hypercalcemia in osteoporotic humans, which has prevented worldwide approval of other vitamin Ds for treating osteoporosis. Indeed, as I said above. the osteoclast-stimulating RANKL gene in osteoblasts has a vitamin D-responsive site in its promoter (Kitazawa et al., 2003). It is important now to find out whether 2MD is a direct stimulator of osteoblast generation and activity like PTH, or whether the anabolic response is just a reaction to strong osteoclast activity. Can 2MD stimulate bone formation when given along with osteoprotegerin to prevent osteoclastogenesis?

Then there is ED-71 (1α,25-(OH)₂-2β-(3-hydroxypropoxy)vitamin D₃) (Kubodera et al., 2003). This is an excellent example of a pseudo-anabolic agent. In OVXed Wistar-Imamichi rats, ED-71 lowered bone resorption without affecting ongoing bone formation and thus stopped and reversed the OVX-induced BMD drop (Kubodera et al., 2003). This differential effect resulted in a large increase (double the value in sham-operated rats) in the concentration of osteocalcin from the unaffected osteoblasts in the blood of OVXed rats given 0.2 μg/kg of body weight twice a week for 3 months. However, the authors of this report did not tell us what ED-71 did to bone in sham-operated rats.

ED-71 also reduces osteoclast activity and increases the BMD of the L2-L4 vertebrae of osteoporotic men and women without causing hypercalcemia (Kubodera et al., 2003).

Unisex Steroids

It is possible to make politically correct unisex steroid analogs that can activate cell surface receptors in caveolar scaffolds but have no genomic (or genotropic) actions (Kousteni et al., 2001, 2002a, 2002b). One of these gender-nonspecific ligands is estren—4-estren-3α,17β-diol—which does not stimulate the transcription of genes with promoters having EREs (estrogen-response elements) (Kousteni et al., 2002a, 2002b). Remarkably, it can target the ligand-binding domains of androgen as well as estrogen receptors.

Starting continuous estren infusion (by implanting it in slow-release pellets) immediately after OVX into female Webster mice or after orchidectomy into male Webster mice significantly reduces or prevents any post-operative bone loss and prevents frequency of apopotic osteoblasts from rising above the sham level (Kousteni et al., 2002a; Manolagas et al., 1999). As expected from its lack of genotropic activity, the analog cannot affect uterine or seminal vesicle weight or stimulate the proliferation MCF-7 mammary cells (Kousteni et al., 2002a; Manolagas et al., 1999). However, and this is a very important "however", the continuous pounding with estren reduces, but by no means stops, as can estradiol, the loss of femoral trabeculae or raise the bone mass above the sham value as do the PTHs (Kousteni et al., 2002a; Manolagas et al., 1999). Furthermore, the analog even cannot hold the vertebral bone formation rate at the sham-operated value. In the experiment of Kousteni et al. (2002a), the bone formation rate/bone area (%/day) in the estren-treated OVXed mice was only 0.158 compared to 0.457 in the sham-OVXed mice. Nor does estren increase the number of osteoblasts per bone perimeter above the sham value. By contrast, an optimal injection of any of the anabolic PTHs starting immediately after OVX or several weeks after OVX can raise the trabecular thickness far above the sham level in rats (Whitfield et al., 1998c). Indeed, it would be interesting to find out what this unisex steroid analog does to bone loss in OVXed rats as well as mice and directly compare its actions to those of the PTHs.

A potent anabolic agent like a PTH stimulates bone growth in three ways—cause the reversion of lining cells to active osteoblasts, the secretion by PTHR1-expressing osteoblastic cells of missionary factors such as FGF-2, IGF-I, and TGF-β1 that stimulate the proliferation of osteoprogenitor cells, and prevent osteoblast and osteocyte apoptosis. It appears that estren signaling at best prevents osteoblast apoptosis, which is obviously not enough to cause supra-sham bone growth as do the PTHs. However, it is no more effective than dihydrotestosterone or estradiol in preventing osteoblast apopotosis. It seems that the anti-apoptotic action of estrogens on osteoblasts is due to the activation of the membrane-associated ERα, ERβ, or androgen receptors' ligand-binding domain, which triggers a Src/Shc/ERK1/2 signal cascade (Kousteni et al., 2002a, 2002b). Indeed, an analog such as 1,2,5-tris(4-hydroxylphenyl)-4-propylpyrazole that only has a genotropic action does not reduce osteoblastic apoptosis or increase osteoclast apoptosis as do the native sex steroids (Kousteni et al., 2002a).

All that can be said at the time of writing about this journalist-hyped unisex steroid estren is that it is only a weak pseudo-anabolic antiresorptive like the native sex steroids. But it could be a safe osteopenia/osteoporosis preventative or maintainer of new PTH-bone because it has none of the potential breast, seminal vesicle, or uterine cancer-promoting activities of the native steroids (Kousteni et al., 2002a). However, it has clearly shown how little an increase in bone growth can be produced by just blocking apoptosis.