Chapter 5:  Protein Function

Introduction

We begin by considering how the shape of a protein can be altered by the binding of another molecule, called a ligand. We then demonstrate the profound implications of this apparently simple phenomenon by describing a few of the many ways in which ligand-driven alterations in protein shape are exploited by cells.

The Binding of a Ligand Can Change the Shape of a Protein¹

The first example involves the enzyme hexokinase, which is present in nearly all cells. This enzyme catalyzes an early step in sugar metabolism - the transfer of the terminal phosphate of an ATP molecule to glucose, forming glucose 6-phosphate (discussed in Chapter 2). Hexokinase binds glucose tightly, and this greatly increases the affinity of the enzyme for ATP, which binds to a neighboring site on the protein. Specific amino acid side chains on the protein then catalyze the phosphate transfer, and the two products - glucose 6-phosphate and ADP - are released to finish the reaction cycle (Figure 5-1).

The hexokinase from yeast is composed of two domains. The binding sites for glucose and ATP lie in a cleft between these domains, and the domains move toward each other to narrow the cleft when glucose binds (Figure 5-2).

This type of domain movement in response to ligand binding is common and is easily explained. In the case of hexokinase there are binding sites for different parts of the glucose molecule on the inside face of each domain. The unfavorable change in the free energy of the protein that occurs when the domains move relative to each other to close the cleft is more than compensated for by the free energy released when the cleft clamps down on the glucose; in other words, the noncovalent bonds that glucose forms with the protein serve to "glue" the two domains together, causing the protein to shift from an open to a closed conformation.

Two Ligands That Bind to the Same Protein Often Affect Each Other's Binding²

The binding of glucose to hexokinase causes a fiftyfold increase in the affinity of the enzyme for ATP. The reason is easy to see. Like glucose, ATP can form noncovalent bonds with amino acids on the inside faces of the two domains if the cleft closes. When ATP alone binds to hexokinase, some of the binding energy must be used to close up the cleft; this energy is not required, however, if glucose binding has already induced this shape change (Figure 5-3). By the same reasoning, one would predict that glucose would bind more tightly to hexokinase when ATP is present than when it is absent, and this is what one observes (Figure 5-4).

ATP and glucose bind to neighboring sites in hexokinase. But the binding of one ligand to a protein's surface can sometimes affect the binding of a second ligand even if the two binding sites are far apart. Suppose, for example, that a protein that binds glucose in the same way as hexokinase also binds another molecule, X, at a distant site on the protein's surface. If the binding site for X changes shape as part of the large conformational change induced by glucose binding, one would say that the binding sites for X and for glucose are coupled. If the shift to the closed conformation, for example, causes the binding site for Xto fitXbetter, then glucose binding will increase the affinity of the protein for X, just as glucose binding increases the affinity of hexokinase for ATP (Figure 5-5).

As we discuss next, proteins in which conformational changes couple two widely separated binding sites have been selected in evolution because they enable a cell to link the fate of one molecule to the presence or absence of any other. This type of conformational coupling is known as allostery. A protein whose activity is regulated in this way is said to undergo an allosteric transition, and the protein is called an allosteric protein.

Two Ligands Whose Binding Sites Are Coupled Must Reciprocally Affect Each Other's Binding²

Whenever two ligands prefer to bind to the same conformation of an allosteric protein, it follows from basic thermodynamic considerations that each ligand must increase the affinity of the protein for the other. This concept is called linkage. It is well illustrated by the example already considered in Figure 5-5, where the binding of glucose to hexokinase increases the enzyme's affinity for molecule X and vice versa. The linkage relationship is quantitatively reciprocal, so that, for example, if glucose has a very large effect on the binding of X, X will have a very large effect on the binding of glucose.

Linkage will operate in a negative way if two ligands bind to different conformations of an allosteric protein. As a general rule, a ligand will act to stabilize the particular conformation of the protein to which it binds; if this is different from the conformation favored by a second ligand, the binding of the first will discourage the binding of the second. Thus, if a shape change caused by glucose binding reduces the affinity of a protein for molecule X, the binding of X must decrease the protein's affinity for glucose (Figure 5-6).

The relationships shown schematically in Figures 5-5 and 5-6 underlie all of cell biology. They seem so obvious in retrospect that we now take them for granted. But their discovery in the 1950s, followed by a general description of allostery in the early 1960s, was revolutionary at the time. Since the X in these examples binds at a site that is distinct from the site where catalysis occurs, it need have no chemical relationship to glucose or to any other ligand that binds at the active site. For enzymes that are regulated in this way, molecule X could either turn the enzyme on (see Figure 5-5) or turn it off (see Figure 5-6). By such a mechanism, allosteric proteins serve as general switches that allow one molecule in a cell to affect the fate of another.

Allosteric Transitions Help Regulate Metabolism³

As described in Chapter 2, the end product of a metabolic pathway often inhibits the enzyme that starts the pathway. Because of this negative feedback on the flux through a pathway, the intracellular concentration of the end product is kept approximately constant, despite large changes in the chemical conditions in the cell. Allosteric transitions are essential to this type of feedback regulation. Enzymes that act early in a pathway, for example, generally exist in two conformations. One is an active conformation that binds substrate at its active site and catalyzes its conversion to the next substance in the pathway. The other is an inactive conformation that binds the final product of the pathway at a different, regulatory site. As the final product accumulates, it binds to the enzymeand converts it to its inactive conformation (see Figure 2-38).

An enzyme involved in a metabolic pathway can also be activated by an allosteric transition induced by ligand binding. In this case the ligand is a molecule that accumulates when the cell is deficient in a product of the pathway; because the ligand binds preferentially to the active form of the protein, it drives the enzyme from an inactive to an active conformation. Examples of this type of positive feedback are provided by many of the enzymes involved in the catabolic pathways that produce ATP: they are stimulated by the rise in ADP concentration that occurs when ATP levels drop. For these enzymes the ADP has a purely regulatory role, in contrast to the substrate role played by ATP in the function of hexokinase.

Proteins Often Form Symmetrical Assemblies That Undergo Cooperative Allosteric Transitions⁴

An enzyme that is regulated by negative feedback and that consists of only one subunit with one regulatory site can at most decrease from 90% to about 10% activity in response to a 100-fold increase in the concentration of the inhibitory ligand (Figure 5-7, red line). Responses of this type are apparently not sharp enough for optimal cell regulation, and most enzymes that are turned on or off by ligand binding consist of symmetrical assemblies of identical subunits. With this arrangement the binding of a molecule of ligand to a single site on one subunit can trigger an allosteric change in the subunit that can be transmitted to the neighboring subunits, helping them to bind the same ligand. As a result of this cooperative allosteric transition, a relatively small change in ligand concentration in the cell can switch the whole assembly from an almost fully active to an almost fully inactive conformation or vice versa (Figure 5-7, blue line).

The principles involved in a cooperative "all-or-none" transition are easiest to visualize for an enzyme that forms a symmetrical dimer. In the example shown in Figure 5-8, the first molecule of an inhibitory ligand binds with great difficulty since its binding destroys an energetically favorable interaction between the two identical monomers in the dimer. A second ligand molecule now binds more easily, however, because its binding restores the monomer-monomer contacts of a symmetrical dimer (and also completely inactivates the enzyme). An even sharper response to a ligand can be obtained with larger assemblies, such as the enzyme formed from 12 polypeptide chains discussed next.

The Allosteric Transition in Aspartate Transcarbamoylase Is Understood in Atomic Detail⁵

One enzyme used in the early studies of negative feedback, allosteric regulation was aspartate transcarbamoylase from E. coli. It catalyzes the important reaction carbamoylphosphate + aspartate → N-carbamoylaspartate, which begins the synthesis of the pyrimidine ring of C, U, and T nucleotides. One of the final products of this pathway, cytosine triphosphate (CTP), binds to the enzyme to turn it off whenever CTP is plentiful.

Aspartate transcarbamoylase is a large complex of six regulatory and six catalytic subunits. The catalytic subunits are present as two trimers, each arranged like an equilateral triangle; the two trimers face each other and are held together by three regulatory dimers that form a bridge between them. The entire molecule is poised to undergo a concerted, all-or-none allosteric transition between two conformations, designated T ("tense") and R ("relaxed") states (Figure 5-9).

The binding of substrates (carbamoylphosphate and aspartate) to the catalytic trimers drives aspartate transcarbamoylase into its catalytically active R state, from which the regulatory CTP molecules dissociate. By contrast, the binding of CTP to the regulatory dimers converts the enzyme to the inactive T state, from which the substrates dissociate. This tug-of-war between CTP and substrates is identical in principle to that described previously in Figure 5-6 for a simpler allosteric protein. But because here the tug-of-war occurs in a symmetrical molecule with multiple binding sites, the effect is a cooperative allosteric transition that can either turn the enzyme on suddenly as substrates accumulate (forming the R state) or shut it off rapidly when CTP accumulates (forming the T state).

A combination of biochemistry and x-ray crystallography has revealed many fascinating details of this allosteric transition. Each regulatory subunit has two domains, and the binding of CTP causes the two domains to move relative to each other, so that they function like a lever that rotates the two catalytic trimers and pulls them closer together into the T state (see Figure 5-9). When this occurs, hydrogen bonds form between opposing catalytic subunits that help to widen the cleft that forms the active site within each catalytic subunit, thereby destroying the binding sites for the substrates (Figure 5-10). Adding large amounts of substrate has the opposite effect, favoring the R state by binding in the cleft of each catalytic subunit and opposing the above conformational change. Conformations that are intermediate between R and T are unstable, so that the enzyme mostly clicks back and forth between its R and T forms, producing a mixture of these two species, whose composition varies depending on the relative concentrations of CTP and substrates.

Protein Phosphorylation Is a Common Way of Driving Allosteric Transitions in Eucaryotic Cells⁶

The activity of proteins in a bacterium such as E. coli is regulated mainly by the myriad small molecules in the cell that bind to specific proteins to cause allo-steric transitions that control the protein's activity. Many of the proteins regulated in this way are enzymes that catalyze metabolic reactions; others transduce signals or turn genes on and off (see, for example, Figure 9-27). Some bacterial proteins are controlled in a different way, however - by the covalent attachment of a phosphate group to an amino acid side chain. Because each phosphate group carries two negative charges, its addition to a protein can cause a structural change, for example, by attracting a cluster of positively charged side chains (Figure 5-11). Such a change occurring at one site in a protein can in turn alter the protein's conformation elsewhere - to control allosterically the activity of a distant ligand-binding site, for instance.

Reversible protein phosphorylation is the predominant strategy used to control the activity of proteins in eucaryotic cells. More than 10% of the 10,000 proteins in a typical mammalian cell are thought to be phosphorylated. The phosphates are transferred from ATP molecules by protein kinases and are taken off by protein phosphatases. Eucaryotic cells contain a large variety of these enzymes, many of which play a central role in intracellular signaling (discussed in Chapter 15).

A Eucaryotic Cell Contains Many Protein Kinases and Phosphatases⁷

The protein kinases that phosphorylate proteins in eucaryotic cells belong to a large family of enzymes, which contain a similar 250 amino acid catalytic (kinase) domain (Figure 5-12). The various family members contain different amino acid sequences on either side of the kinase domain, and often have short amino acid sequences inserted into loops within it (see red arrowheads in Figure 5-12). Some of these additional amino acid sequences enable each kinase to recognize the specific set of proteins that it phosphorylates. Other unique sequences allow the activity of each enzyme to be tightly regulated, so that it can be turned on and off in response to different specific signals, as described below.

By comparing the numbers of amino acid sequence differences between the members of a protein family, one can construct an "evolutionary tree" that is thought to reflect the pattern of gene duplication and divergence that gave rise to the family (see Figure 8-76). An evolutionary tree of protein kinases is shown in Figure 5-13. Not surprisingly, kinases with related functions are often located on nearby branches of the tree: the protein kinases involved in cell signaling that phosphorylate tyrosine side chains, for example, are all clustered at the upper left corner of the tree. The other kinases shown phosphorylate either a serine or a threonine side chain, and many are organized into clusters that seem to reflect their function - in transmembrane signaling, intracellular amplification of signals, cell-cycle control, and so on.

The basic reaction catalyzed by a protein kinase is illustrated in Figure 5-14. A phosphate group is transferred from an ATP molecule to a hydroxyl group on a serine, threonine, or tyrosine side chain of a protein. This reaction is essentially unidirectional because of the large amount of free energy released when the phosphate-phosphate bond in ATP is broken to produce ADP (see Figure 2-28). The phosphorylations catalyzed by protein kinases can nevertheless be reversed by a second group of enzymes, called protein phosphatases, which remove the phosphate (see Figure 5-14). There are several families of protein phosphatases: some are highly specific and remove phosphate groups from only one or a few proteins, while others are relatively nonspecific and act on a broad range of proteins. The extent of phosphorylation of a particular protein in a cell at a particular time depends on the relative activities of the protein kinases and phosphatases that act on it.

The Structure of Cdk Protein Kinase Shows How a Protein Can Function as a Microchip⁸

The hundreds of different protein kinases in a eucaryotic cell are organized into complex networks of signaling pathways that help coordinate the cell's activities, drive the cell cycle, and relay signals into the cell from the cell's environment. Many of the signals involved need to be both integrated and amplified. Individual protein kinases (and other signaling proteins) serve as processing devices, or "microchips," in the integration process. An important part of the input to these proteins comes from the control that is exerted by phosphates added to them by other protein kinases in the network: specific sets of phosphate groups serve to activate the protein, while other sets inactivate it.

A cyclin-dependent protein kinase (Cdk) represents a good example of such a processing device. Kinases in this class are central components of the cell-division-cycle control system in eucaryotic cells (discussed in Chapter 17). In a vertebrate cell, individual Cdk enzymes turn on and off in succession as a cell proceeds through the different phases of its division cycle, and when they are on, they influence various aspects of cell behavior through their effects on the proteins they phosphorylate. The three-dimensional structure of this important class of protein kinases is now known, and we shall use it to demonstrate how a protein can function as a microchip.

A Cdk protein is active as a protein kinase only when it is bound to a second protein called a cyclin. But, as illustrated in Figure 5-15, the binding of cyclin is only one of three distinct "inputs" required to activate the Cdk: in addition, a phosphate must be added to a specific threonine side chain and a phosphate elsewhere in the protein (covalently bound to a specific tyrosine side chain) must be removed. Cdk thus monitors a specific set of cell components - a cyclin, a protein kinase, and a protein phosphataseand turns on if, and only if, each of these components has attained its appropriate activity state. Some cyclins, for example, rise and fall in concentration in step with the cell cycle, increasing gradually in amount until they are suddenly destroyed at a particular point in the cycle. The sudden destruction of a cyclin (by targeted proteolysis) will immediately shut off its partner Cdk enzyme, and this is an important way of controlling intracellular events such as mitosis.

The three-dimensional structure of Cdk (Figure 5-16A) suggests a likely molecular explanation for the regulation of this enzyme. The Cdk protein on its own is inactive for two reasons: its ATP-binding site is distorted, and a flexible loop of about 20 amino acids blocks access of the protein substrate to the active site. Cyclin binding both removes the distortion and permits the addition of the activating phosphate group to the tip of the flexible loop; this phosphate is then thought to be attracted to a pocket formed by positively charged amino acids, pulling down the loop so as to permit access to the active site (Figure 5-16B). Cyclin binding also allows the rapid addition of the inhibitory phosphate, however, which interferes with the ATP site, and this keeps the Cdk protein in an inactive state. The kinase is finally activated when a specific phosphatase removes the inhibiting phosphate (Figure 5-17).

Proteins That Bind and Hydrolyze GTP Are Ubiquitous Cellular Regulators⁹

We have described how the addition or removal of phosphate groups on a protein can be used by a cell to control the protein's activity. In the examples discussed so far, the phosphate is transferred from an ATP molecule to an amino acid side chain of the protein in a reaction that is catalyzed by a specific protein kinase. Eucaryotic cells also use another way to control protein activity by phosphate addition and removal. In this case the phosphate is not attached directly to the protein; instead, it is a part of the guanine nucleotide GTP, which binds tightly to the protein. With GTP bound the protein is active. The loss of a phosphate group occurs when the bound GTP is hydrolyzed to GDP in a reaction that is catalyzed by the protein itself; with GDP bound the protein is inactive.

GTP-binding proteins (also called GTPases because of the GTP hydrolysis that they catalyze) constitute a large family of proteins that all have a similar GTP-binding globular domain. When its bound GTP is hydrolyzed to GDP, this domain undergoes a conformational change that inactivates the protein. The three-dimensional structure of a small GTP-binding protein called Ras is illustrated in Figure 5-18.

The Ras protein plays a crucial role in cell signaling (as discussed in Chapter 15). In its GTP-bound form it is active and stimulates a cascade of protein phosphorylations in the cell. Most of the time, however, the protein is in its inactive, GDP-bound form. It is activated when it exchanges its GDP for a GTP molecule in response to extracellular signals, such as growth factors, that bind to receptors in the plasma membrane (see Figure 15-53). Thus the Ras protein acts as an on-off switch whose activity is determined by the presence or absence of an additional phosphate on a bound GDP molecule, just as the activity of a Cdk protein is controlled by the presence of one or more phosphate groups on amino acid side chains (see Figure 5-17).

Other Proteins Control the Activity of GTP-binding Proteins by Determining Whether GTP or GDP Is Bound¹⁰

The activity of Ras and other GTP-binding proteins is controlled by regulatory proteins that determine whether GTP or GDP is bound, just as the activity of a Cdk protein is controlled by cyclins, protein kinases, and protein phosphatases. Ras is inactivated by a GTPase-activating protein (or GAP), which binds to the Ras protein and induces it to hydrolyze its bound GTP molecule to GDP - which remains tightly bound - and inorganic phosphate (P_i), which is rapidly released. The Ras protein will stay in its inactive, GDP-bound conformation until it encounters a guanine nucleotide releasing protein (GNRP), which binds to GDP-Ras and causes it to release its GDP. Because the empty nucleotide-binding site is immediately filled by a GTP molecule (GTP is present in large excess over GDP in cells), the GNRP activates Ras by indirectly adding back the phosphate removed by GTP hydrolysis. Thus, in a sense, the roles of GAP and GNRP are analogous to those of a protein phosphatase and a protein kinase, respectively (Figure 5-19).

The Allosteric Transition in EF-Tu Protein Shows How Large Movements Can Be Generated from Small Ones¹¹

The Ras protein is a member of a family of monomeric regulatory GTPases, each of which consists of a single GTP-binding domain of about 200 amino acids. During the course of evolution this domain has also become joined to other protein domains to create a large family of GTP-binding proteins, whose members include the receptor-associated trimeric G proteins (discussed in Chapter 15), proteins regulating the traffic of vesicles between intracellular compartments (discussed in Chapter 13), and proteins that bind to transfer RNA and are required for protein synthesis on the ribosome (discussed in Chapter 6). In each case, an important biological activity is controlled by a change in the protein's conformation caused by GTP hydrolysis in a Ras-like domain.

The EF-Tu protein provides a good example of how this family of proteins works. EF-Tu is an abundant molecule in bacterial cells, where it serves as an elongation factor in protein synthesis, loading each amino-acyl tRNA molecule onto the ribosome. The tRNA molecule forms a tight complex with the GTP-bound form of EF-Tu. In this complex, the amino acid attached to the tRNA is masked; its unmasking, which is required for protein synthesis, occurs on the ribosome when the tRNA is released following hydrolysis of the GTP bound to EF-Tu (see Figure 6-31 for an illustration of the clock-like function of EF-Tu).

The three-dimensional structure of EF-Tu, in both its GTP- and GDP-bound forms, has been determined by x-ray crystallography. These studies reveal how the unmasking of the tRNA occurs. The dissociation of the inorganic phosphate group (P_i), which follows the reaction GTP → GDP + P_i, causes a shift of a few tenths of a nanometer at the GTP-binding site, just as it does in the Ras protein. This tiny movement, equivalent to a few times the diameter of a hydrogen atom, causes a conformational change to propagate along a crucial piece of alpha helix, called the switch helix, in the Ras-like domain of the protein. The switch helix seems to serve as a latch that adheres to a specific site in another domain of the molecule, holding the protein in a "shut" conformation. The conformational change triggered by GTP hydrolysis causes the switch helix to detach, allowing separate domains of the protein to swing apart, through a distance of about 4 nanometers, thereby releasing the bound tRNA (Figure 5-20).

One can see from this example how cells can exploit simple chemical changes that occur on the surface of a small protein domain to evolve larger proteins with sophisticated functions. In the transition from Ras to EF-Tu we have entered a world that begins to feel like biology.

Proteins That Hydrolyze ATP Do Mechanical Work in Cells¹²

Allosteric shape changes can be used to generate orderly movements in cells as well as to regulate chemical reactions. Suppose, for example, that a protein is required that can "walk" along a narrow thread, such as a DNA molecule. Figure 5-21 shows schematically how an allosteric protein might do this by undergoing a series of conformational changes. With nothing to drive these changes in an orderly sequence, however, they will be perfectly reversible, and the protein will wander randomly back and forth along the thread.

We can look at this situation another way. Since the directional movement of a protein does work, the laws of thermodynamics demand that such movement depletes free energy from some other source (otherwise the protein could be used to make a perpetual motion machine). Therefore, no matter what modifications we make to the model shown in Figure 5-21, such as adding ligands that favor particular conformations, without an input of energy the protein molecule shown could only wander aimlessly.

How can one make the series of conformational changes unidirectional? To make the entire cycle proceed in one direction, it is enough to make any one of the steps irreversible. One way to do this is to use the mechanism just discussed for driving allosteric changes in a protein molecule by GTP hydrolysis. For example, because a great deal of free energy is released when GTP is hydrolyzed, it is very unlikely that the EF-Tu protein will directly add a phosphate molecule to GDP to reverse the hydrolysis of its GTP. Precisely the same principle applies to ATP hydrolysis, and most proteins that are able to walk in one direction for long distances (the so-called motor proteins) do so by hydrolyzing ATP.

In the highly schematic model shown in Figure 5-22, ATP binding shifts a motor protein from conformation 1 to conformation 2. The bound ATP is then hydrolyzed to produce ADP and inorganic phosphate (P_i), causing a change from conformation 2 to conformation 3. Finally, the release of the bound ADP and P_i drives the protein back to conformation 1. Because the transitions 1 → 2 → 3 → 1 are driven by the energy provided by ATP hydrolysis, this series of conformational changes will be effectively irreversible under physiological conditions (that is, the probability that ADP will recombine with P_i to form ATP by the route 1 → 3 → 2 → 1 is extremely low). Thus the entire cycle will go in only one direction, causing the protein molecule to move continuously to the right in this example. Many proteins generate directional movement in this way, including DNA helicase enzymes that propel themselves along DNA at rates as high as 1000 nucleotides per second.

The Structure of Myosin Reveals How Muscles Exert Force¹³

In Chapter 16 we discuss how various cell movements are produced by motor proteins that move rapidly along protein filaments, driven by energy derived from repeated cycles of ATP hydrolysis (see Figure 5-22). The best understood of these motor proteins is myosin, whose directed movement along actin filaments causes both intracellular movements and muscle contraction. The three-dimensional structures of myosin (and actin) have been determined by x-ray diffraction analyses, providing a glimpse of the inner workings of a biological motor. The structure of the myosin head domain (Figure 5-23) suggests how ATP hydrolysis may be coupled to force generation. ATP binding and hydrolysis are thought to cause an ordered series of conformational changes that move the tip of the head by about 5 nanometers, as illustrated schematically in Figure 5-24. This movement, coupled to the making and breaking of interactions with actin and repeated with each round of ATP hydrolysis, propels the myosin molecule unidirectionally along an actin filament (see Figure 16-91). Thus in myosin, as in the EF-Tu protein discussed earlier, a small perturbation in the nucleotide-binding site is translated, via allosteric transitions that magnify the effect, to create the much more extensive, orderly protein motions that underlie much of cell biology.

ATP-driven Membrane-bound Allosteric Proteins Can Either Act as Ion Pumps or Work in Reverse to Synthesize ATP¹⁴

Besides generating mechanical force, allosteric proteins can use the energy of ATP hydrolysis to do other forms of work, such as pumping specific ions into or out of the cell. An important example is the Na⁺-K⁺ ATPase found in the plasma membrane of all animal cells, which pumps 3 Na⁺ out of the cell and 2 K⁺ in during each cycle of conformational changes driven by ATP hydrolysis (see Figure 11-11). This ATP-driven pump consumes more than 30% of the total energy requirement of most cells. By continuously pumping Na⁺ out and K⁺ in, it keeps the Na⁺ concentration much lower inside the cell than outside and the K⁺ concentration much higher inside than outside, thereby generating two ion gradients (in opposite directions) across the plasma membrane. These and other ion gradients across various cell membranes can store energy, just as the differences of water pressure on either side of a dam can. The energy is used to drive conformational changes in a variety of membrane-bound allosteric proteins that do useful work. The large Na⁺ gradient across the plasma membrane, for example, drives many other plasma-membrane-bound protein pumps that transport glucose or specific amino acids into the cell; the glucose and amino acids are dragged in by the simultaneous influx of Na⁺ that occurs as Na⁺ moves down its concentration gradient.

The membrane-bound allosteric pumps that are driven by ATP hydrolysis can also work in reverse and employ the energy in the ion gradient to synthesize ATP. In fact, the energy available in the H⁺ gradient across the inner mitochondrial membrane is used in this way by the membrane-bound allosteric protein complex, ATP synthase, which synthesizes most of the ATP required by animal cells, as we discuss in Chapter 14.

Energy-coupled Allosteric Transitions in Proteins Allow the Proteins to Function as Motors, Clocks, Assembly Factors, or Transducers of Information¹⁵

Many proteins undergo ordered conformational changes that are coupled to the energy released when a nucleoside triphosphate (either ATP or GTP) is hydrolyzed to a nucleoside diphosphate (ADP or GDP, respectively). Some of these changes involve the covalent attachment of a phosphate group to the protein (protein phosphorylation), but many others, as for myosin, do not. Each change is generally triggered by a specific event (the binding of myosin to an actin filament, for example, triggers ATP hydrolysis by myosin), imparting directionality and order to the interactions of macromolecules in the cell.

The ability to harness the energy in nucleoside triphosphates to drive allosteric changes in proteins has been crucial for the evolution of cells in much the same way that the ability to harness electrical energy has been crucial for the development of modern technology. In both cases rich opportunities have opened up for the development of useful devices. Proteins like Cdk, for example, act as sophisticated integrating switches (see Figure 5-15), receiving information about a cell's environment and the stage of the cell cycle and using it to coordinate the behavior of the cell. Motor proteins like myosin move unidirectionally along filaments to generate various movements and create order inside the cell. Proteins such as EF-Tu serve as timing devices that improve the fidelity of important biological reactions (see Figure 6-31). Other proteins use the energy released by nucleoside triphosphate hydrolysis to catalyze the assembly of specific protein complexes. A summary is presented in Figure 5-25.

Proteins Often Form Large Complexes That Function as Protein Machines¹⁶

As one progresses from small proteins to large proteins formed from many domains, the functions that a protein can perform become more elaborate. The most impressive tasks, however, are carried out by large protein assemblies formed from multiple individual subunits. Now that it is possible to reconstruct most biological processes in cell-free systems in a test tube, one can see that each central process in a cell - such as DNA replication, RNA or protein synthesis, vesicle budding, or transmembrane signaling - is catalyzed by a complex of 10 or more proteins. In such protein machines the hydrolysis of bound nucleoside triphosphate molecules (ATP or GTP) drives ordered conformational changes in the individual proteins, enabling the ensemble of proteins to move coordinately. In this way, for example, the appropriate enzymes are moved directly into the positions where they are needed to carry out each reaction in a series instead of waiting for the random collision of each separate component that would otherwise be required. A simple mechanical analogy is illustrated in Figure 5-26.

Cells have evolved protein machines for the same reason that humans have invented mechanical and electronic machines: manipulations that are spatially and temporally coordinated through linked processes are much more efficient for accomplishing almost any task than is the sequential use of individual tools.

Summary

Allosteric proteins reversibly change their shape when ligands bind to their surface. The changes produced by one ligand often affect the binding of a second ligand, and this type of linkage between two ligand-binding sites provides a crucial mechanism for regulating cell processes. Metabolic pathways, for example, are controlled by feedback regulation: some small molecules will inhibit and other small molecules activate enzymes early in a pathway. Enzymes regulated in this way generally form symmetrical assemblies, allowing cooperative conformational changes to create a steep response to ligands.

Changes in protein shape can be driven in a unidirectional manner by the expenditure of chemical energy. By coupling allosteric shape changes to ATP hydrolysis, for example, proteins can do useful work, such as generating a mechanical force or pumping ions across a membrane. The three-dimensional structures of several proteins, determined by x-ray crystallography, have revealed how a small local change caused by nucleoside triphosphate hydrolysis is amplified to create major changes elsewhere in the protein; by such means these proteins are able to serve as transducers of information, motors, clocks, or assembly factors. Highly efficient "protein machines" are formed by incorporating many different protein subunits into larger assemblies in which allosteric movements of the individual components are coordinated to carry out many, if not most, biological reactions.

Introduction

Having described some of the remarkable devices that cells make out of proteins, we now consider how these devices are produced and how they are destroyed. The mechanism of protein synthesis is discussed elsewhere. We begin by considering how a protein folds and assembles once it leaves the ribosome as a finished polypeptide chain.

Proteins Are Thought to Fold Through a Molten Globule Intermediate¹⁷

Because many purified proteins will refold properly on their own after being unfolded in vitro, for many years it was thought that a protein will try out every conceivable conformation as it folds until it attains the one conformation with the lowest free energy, which was assumed to be its correctly folded state. We now know that this view is incorrect: despite the high speed of molecular motions in a protein (see p. 97), there are vastly more possible conformations for any large protein than can be explored in the few seconds that are typically required for folding. Moreover, the existence of mutant proteins that have specific defects in folding indicates that a protein's amino acid sequence has been selected during evolution, not only for the properties of its final structure, but also for the ability to fold rapidly into its native conformation.

The ability of pure, denatured proteins to reform their native structures on their own has made it possible to dissect the process of protein folding experimentally. Proteins appear to fold rapidly into a structure in which most (but not all) of the final secondary structure (α helices and β sheets) has formed and in which these elements of structure are aligned in roughly the right way (Figure 5-27). This unusually open and flexible conformation, which is called a molten globule (Figure 5-28), is the starting point for a relatively slow process in which many side-chain adjustments occur in order to form the correct tertiary structure. In the latter process a variety of pathways can be taken toward the final conformation. Some of these may be nonproductive dead ends without the help of a molecular chaperone, special proteins in cells whose function is to help other proteins fold and assemble into stable, active structures (see Figure 5-27).

Molecular Chaperones Facilitate Protein Folding¹⁸

Molecular chaperones were first identified in bacteria when E. coli mutants that failed to allow bacteriophage lambda to replicate in them were studied. These mutants produce slightly altered versions of two components of the chaperone machinery, related to heat-shock proteins 60 and 70 (hsp60 and hsp70), and as a result are defective in specific steps in the assembly of the viral proteins.

Eucaryotic cells have families of hsp60 and hsp70 proteins, and different family members function in different organelles. Thus, as discussed in Chapter 12, mitochondria contain their own hsp60 and hsp70 molecules that are distinct from those that function in the cytosol, and a special hsp70 (called BIP) helps to fold proteins in the endoplasmic reticulum.

Both hsp60-like and hsp70 proteins work with a small set of associated proteins when they help other proteins to fold. They share an affinity for the exposed hydrophobic patches on incompletely folded proteins, and they hydrolyze ATP, possibly binding and releasing their protein with each cycle of ATP hydrolysis. Originally, molecular chaperones were thought to act only by preventing the promiscuous aggregation of still unfolded proteins (hence their name). It is now thought, however, that they also interact more intimately with their clients, producing effects that can be likened to a "protein massage." By binding to exposed hydrophobic regions, the chaperone massages those regions of a protein that are likely to have misfolded from the molten globule state, changing their structure in a way that gives the protein another chance to fold (see Figure 5-27).

In some other respects the two types of hsp proteins function differently. The hsp70 machinery is thought to act early in the life of a protein, binding to a string of about seven hydrophobic amino acids before the protein leaves the ribosome (Figure 5-29). In contrast, hsp60-like proteins form a large barrel-shaped structure (Figure 5-30) that acts later in a protein's life; this chaperone is thought to form an "isolation chamber" into which misfolded proteins are fed, providing them with a favorable environment in which to attempt to refold (see Figure 5-29).

These molecular chaperones are called heat-shock proteins because they are synthesized in dramatically increased amounts following a brief exposure of cells to an elevated temperature (for example, 42°C). This seems to reflect the operation of a feedback system that responds to any increase in misfolded proteins (such as those produced by elevated temperatures) by boosting the synthesis of the chaperones that help the protein refold.

Many Proteins Contain a Series of Independently Folded Modules¹⁹

The folding of a newly synthesized protein often begins with the formation of a number of distinct structurally stable domains that correspond to functional units, which seem to have ancient evolutionary origins. Elsewhere we discuss the pathways by which proteins are thought to have evolved, emphasizing how new proteins have been created by the shuffling of exons that code for conserved domains with useful properties (see pp. 386-394). Evolution has preserved some of these domains as folding units that retain their structure even when cut out of the protein - either by selected proteolysis or, more efficiently, by genetic engineering techniques. Protein domains of this type that are very frequently involved in evolutionary exon shuffling are called modules; their importance has become clear now that DNA sequences are available for thousands of genes.

Protein modules are typically 40 to 100 amino acids in length. Their small size and ability to fold independently has made it possible to determine many of their three-dimensional structures in solution by high-resolution NMR techniques, which is a convenient alternative to x-ray crystallography. Some typical modules are illustrated in Figure 5-31. Each of these modules has a stable core structure formed from strands of β sheet, from which less-ordered loops of polypeptide chain protrude (shown in green). The loops are ideally situated to form binding sites for other molecules, as well demonstrated for the immunoglobulin fold, which was first recognized in antibody molecules (see Figure 23-35). The evolutionary success of β-sheet-based modules is likely to have been due to their forming a convenient framework for the generation of new binding sites for ligands through changes to these protruding loops.

Modules Confer Versatility and Often Mediate Protein-Protein Interactions^{19, 20}

A second feature of protein modules that explains their utility is the ease with which they can be integrated into other proteins. Five of the six modules illustrated in Figure 5-31 have their N- and C-terminal ends (marked with red balls) at opposite ends of the module. This "in-line" arrangement means that when the DNA encoding such a module undergoes tandem duplication, which is not unusual in the evolution of genomes (discussed in Chapter 8), the duplicated modules can be readily accommodated in the protein. In this way such modules can become linked in series to form extended structures, either with themselves (Figure 5-32) or with other in-line modules. Stiff extended structures composed of a series of modules are commonly found both in extracellular matrix molecules and in the extracellular portions of cell surface receptor proteins.

Other modules, like the kringle module in Figure 5-31, are of a "plug-in" type. After genomic rearrangements, they can be easily accommodated as an insertion into a loop region of a second protein. Some of these modules act as specific binding sites for other proteins or structures in the cell. An important example is the SH2 domain, which can bind tightly to a region of polypeptide chain that contains a phosphorylated tyrosine side chain. Because each SH2 domain also recognizes other features of the polypeptide, it binds only to a subset of proteins that contains phosphorylated tyrosines. The presence of an SH2 domain in a protein allows it to form complexes with proteins that become phosphorylated on tyrosines in response to cell-signaling events (Figure 5-33). Such protein complexes that form and break up as a result of changes in protein phosphorylation play a central part in transducing extracellular signals into intracellular ones, as described in Chapter 15.

Proteins Can Bind to Each Other Through Several Types of Interfaces

Proteins can bind to other proteins in at least three ways. In many cases a portion of the surface of one protein contacts an extended loop of polypeptide chain (a "string") on a second protein (Figure 5-34A). Such a surface-string interaction, for example, allows the SH2 domain to recognize a phosphorylated loop of another protein, and it also enables a protein kinase to recognize the proteins that it will phosphorylate (see Figure 5-16B).

A second type of protein-protein interface is formed when two alpha helices, one from each protein, pair together to form a coiled-coil (Figure 5-34B). This type of protein interface is found in several families of gene regulatory proteins, as discussed in Chapter 9.

The most common way for proteins to interact, however, is by the precise matching of one rigid surface with that of another (Figure 5-34C). Such interactions can be very tight, since a large number of weak bonds can form between two surfaces that match well. For the same reason such surface-surface interactions can be extremely specific, allowing one protein to select a specific partner from the many thousands of different proteins found in a higher eucaryotic cell.

Linkage and Selective Proteolysis Ensure All-or-None Assembly

Many proteins are present in large complexes with other proteins. This requires that the protein bind to several other proteins at the same time. It is crucial for the cell that each protein complex form efficiently and that the formation of partial complexes, which can interfere with the function of complete complexes, be kept to a minimum. There must be mechanisms, therefore, for ensuring that assembly is an all-or-none process.

One important mechanism relies on the phenomenon of linkage, which we described earlier. Because of linkage, if a ligand changes the shape of an allo-steric protein so that the protein binds a second ligand more tightly, the second ligand must similarly increase the affinity of the protein for the first ligand (see Figure 5-5). The same principle applies to protein-protein interactions. When two proteins bind to each other, they often increase the affinity of one of the partners for a third protein. Because of linkage, the complex of all three proteins will be much more stable than a complex containing only two. A mechanism of this type can produce all-or-none assembly (Figure 5-35).

Even if an all-or-none assembly mechanism drives the formation of complete protein complexes, unless the cell contains exactly the right proportions of each protein in the complex, unassembled proteins will be left over. In fact, cells do not always produce their components in precise amounts and are instead able to degrade selectively any protein component that is left unassembled (Figure 5-36). Cells therefore require a sophisticated system to identify abnormally assembled proteins and destroy them. Indeed, the eucaryotic cell contains an elaborate set of proteins that enables such incomplete assemblies to be selectively directed to its protein-degradation machinery, as we now discuss.

Ubiquitin-dependent Proteolytic Pathways Are Largely Responsible for Selective Protein Turnover in Eucaryotes²¹

One function of intracellular proteolytic mechanisms is to recognize and eliminate unassembled proteins, as just described. Another is to dispose of damaged or misfolded proteins (see Figure 5-27). Yet another is to confer short half-lives on certain normal proteins whose concentrations must change promptly withalterations in the state of a cell; many of these short-lived proteins are degraded rapidly at all times, while others, most notably the cyclins, are stable until they are suddenly degraded at one particular point in the cell cycle. Although here we mainly discuss how proteins are degraded in the cytosol, important degradation pathways also operate in the endoplasmic reticulum (ER) and, as discussed in Chapter 13, in lysosomes.

Most of the proteins that are degraded in the cytosol are delivered to large protein complexes called proteasomes, which are present in many copies and are dispersed throughout the cell. Each proteasome consists of a central cylinder formed from multiple distinct proteases, whose active sites are thought to face an inner chamber. Each end of the cylinder is "stoppered" by a large protein complex formed from at least 10 types of polypeptides, some of which hydrolyze ATP (Figure 5-37). These protein stoppers are thought to select the proteins for destruction by binding to them and feeding them into the inner chamber of the cylinder, where multiple proteases degrade the proteins to short peptides that are then released.

Proteasomes act on proteins that have been specifically marked for destruction by the covalent attachment of a small protein called ubiquitin (Figure 5-38). Ubiquitin exists in cells either free or covalently linked to proteins. Most ubiquinated proteins have been tagged for degradation. (Some long-lived proteins such as histones are also ubiquinated, but in these cases the function of ubiquitin is not understood.) Different ubiquitin-dependent proteolytic pathways employ structurally similar but distinct ubiquitin-conjugating enzymes that are associated with recognition subunits that direct them to proteins carrying a particular degradation signal. The conjugating enzyme adds ubiquitin to a lysine residue of a target protein and thereafter adds a series of additional ubiquitin moieties, forming a multiubiquitin chain (Figure 5-39) that is thought to be recognized by a specific receptor protein in the proteasome.

Denatured or misfolded proteins, as well as proteins containing oxidized or otherwise abnormal amino acids, are recognized and degraded by ubiquitin-dependent proteolytic systems. The ubiquitin-conjugating enzymes presumably recognize signals that are exposed on these proteins as a result of their misfolding or chemical damage; such signals are likely to include amino acid sequences or conformational motifs that are buried and therefore inaccessible in the normal counterparts of these proteins.

A proteolytic pathway that recognizes and destroys abnormal proteins must be able to distinguish between completed proteins that have "wrong" conformations and the many growing polypeptides on ribosomes (as well as polypeptides just released from ribosomes) that have not yet achieved their normal folded conformation. That this is not a trivial problem can be demonstrated experimentally: if puromycin - an inhibitor of protein synthesis - is added to cells, the prematurely terminated proteins that are formed are rapidly degraded by a ubiquitin-dependent pathway. One possibility is that the normally forming proteins are temporarily protected by the translation machinery or by chaperone molecules. Another is that nascent and newly completed proteins are actually vulnerable to proteolysis but manage to fold up into their native conformations fast enough to escape being targeted for destruction by proteolysis.

The Lifetime of a Protein Can Be Determined by Enzymes That Alter Its N-Terminus²²

One feature that has an important influence on the stability of a protein is the nature of the first (N-terminal) amino acid in the polypeptide chain. There is a strong relation, called the N-end rule, between the in vivo half-life of a protein and the identity of its N-terminal amino acid. Distinct versions of the N-end rule operate in all organisms examined, from bacteria to mammals. The amino acids Met, Ser, Thr, Ala, Val, Cys, Gly, or Pro, for example, protect proteins in the yeast S. cerevisiae when present at the N-terminus; these amino acids are not recognized by targeting components of the N-end rule pathway, while the remaining 12 amino acids attract a proteolytic attack. Most of the proteins that are rapidly degraded by the N-end rule pathway (which operates in both the cytosol and the nucleus) remain to be identified. Since destabilizing amino acids, however, are rare at the N-termini of cytosolic proteins but are frequently present at the N-terminus of proteins that have been transported to other compartments, one hypothetical function of the N-end rule pathway is to degrade proteins that normally function in the ER, the Golgi apparatus, or another membrane-bounded compartment but for some reason have leaked back into the cytosol.

It is not known how destabilizing amino acids become exposed at the N-terminus of a newly formed protein. As discussed in Chapter 6, all proteins are initially synthesized with methionine (or formyl-methionine in bacteria) as their N-terminal amino acid. This methionine, which is a stabilizing amino acid in the N-end rule, is often removed by a specific aminopeptidase. The presently known methionine aminopeptidases, however, will remove the N-terminal methionine if and only if the second amino acid is also stabilizing in the N-end rule. The proteases that produce physiological substrates of the N-end rule pathway, and the sequences they recognize as signals for cleavage, remain to be discovered.

Certain destabilizing N-terminal amino acids, such as aspartate and glutamate, are not recognized directly by the targeting component of the N-end rule pathway. Instead, they are modified by the enzyme arginyl-tRNA-protein transferase, which links arginine, one of the directly recognized destabilizing amino acids, to the N-terminus of proteins bearing N-terminal aspartate or glutamate. Arginine is thus one of the primary destabilizing amino acids in the N-end rule, while aspartate and glutamate are secondary destabilizing amino acids. In eucaryotes there are also tertiary destabilizing N-terminal amino acids - asparagine and glutamine - which are destabilizing through their conversion, by a specific amidase, into the secondary destabilizing amino acids aspartate and glutamate.

The N-terminal amino acid of a protein is often found to be resistant to hydrolysis by the reagents used in protein sequenators. Such proteins have a chemically modified ("blocked") N-terminus, the most frequent modification being acetylation. This modification was believed to play a role in protecting long-lived proteins from degradation. However, recent experiments with yeast mutants that lack the major species of N-terminal acetylase, so that the bulk of the normally acetylated proteins are unacetylated, show that most of these unacetylated proteins remain long-lived. The function of N-terminal acetylation in these proteins remains to be deciphered.

Summary

From the moment of its birth on a ribosome to its death by targeted proteolysis, a protein is accompanied by molecular chaperones and other surveying devices whose purpose is to massage it into shape, repair it, or eliminate it. Misfolded proteins are first induced to refold correctly by hsp70 or hsp60 chaperone molecules; if this fails, they are coupled to ubiquitin and thereby targeted for digestion in proteasomes.

Proteins are often composed of discrete modular domains that have been juxtaposed during evolution by duplication and shuffling of the DNA sequences that encode the modules. The modules often contain specific binding sites for other molecules, including other proteins, and they often enable proteins to assemble into large complexes. The principle of linkage explains how cells manage to use allosteric transitions to assemble such protein complexes in an all-or-none fashion.