• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Hum Mol Genet. Author manuscript; available in PMC Oct 31, 2008.
Published in final edited form as:
PMCID: PMC2576517

Loss of MMP-2 disrupts skeletal and craniofacial development, and results in decreased bone mineralization, joint erosion, and defects in osteoblast and osteoclast growth


The “vanishing bone” or inherited osteolysis/arthritis syndromes represent a heterogeneous group of skeletal disorders characterized by mineralization defects of affected bones and joints. Differing in anatomical distribution, severity, and associated syndromic features, gene identification in each “vanishing bone” disorder should provide unique insights into genetic/molecular pathways contributing to the overall control of skeletal growth and development. We previously described and then demonstrated that the novel autosomal recessive osteolysis/arthritis syndrome, Multicentric Osteolysis with Arthritis (MOA [MIM #605156]), was caused by inactivating mutations in the MMP2 gene (1). These in vivo results were counterintuitive and unexpected since previous in vitro studies suggested that MMP-2 overexpression and increased activity, not deficiency, would result in the bone and joint features of MOA. The apparent lack of a murine model (2) has hindered studies on disease pathogenesis and, more fundamentally, in addressing the paradox of how functional loss of a single proteolytic enzyme results in an apparent increase in bone loss. Here, we report that Mmp2-/- mice display attenuated features of human MOA including progressive loss of bone mineral density, articular cartilage destruction, and abnormal long bone and craniofacial development. Moreover, these changes are associated with markedly and developmentally-restricted decreases in osteoblast and osteoclast numbers in vivo. Mmp2-/- mice have ~50% fewer osteoblasts and osteoclasts than control littermates at 4 days of life but these differences have nearly resolved by 4 weeks of age. In addition, despite normal cell numbers in vivo at 8 weeks of life, Mmp2-/- bone marrow cells are unable to effectively support osteoblast and osteoclast growth and differentiation in culture. Targeted inhibition of MMP-2 using siRNA in human SaOS2 and murine MC3T3 osteoblast cell lines resulted in decreased cell proliferation rates. Taken together, our findings suggest that MMP-2 plays a direct role in early skeletal development and bone cell growth and proliferation. Thus, Mmp2-/- mice provide a valuable biologic resource for studying the pathophysiologic mechanisms underlying the human disease and defining the in vivo physiologic role of MMP-2.


The multicentric osteolyses or ‘vanishing bone’ syndromes are a group of hereditary autosomal dominant and recessive skeletal disorders characterized by marked and progressive bone loss and joint destruction resulting in skeletal deformities and significant functional impairment (3, 4). The first case of “disappearing bone” disease was originally described by Jackson in 1838 (5) and since then, a number of clinically distinguishable forms have been described under many different designations including idiopathic osteolysis, hereditary multicentric osteolysis, and carpal and tarsal agenesis (3, 4). These rare syndromes, although sharing many phenotypic characteristics, can differ from each other in mode of transmission and in clinical features, including the presence or absence of mental retardation and nephropathy and the extent and anatomic distribution of osteolysis. Based on clinical and radiographic features, and inheritance pattern, the International Skeletal Dysplasia Registry provided a useful comparative classification scheme of the multicentric disorders into four groups (6). Most notably, each of these rare Mendelian disorders provides the unique opportunity to define the genetic/molecular pathways contributing to normal osteoblast/osteoclast development and function.

We and another group previously described the novel autosomal recessive disorder multicentric osteolysis with arthropathy (MOA [OMIM 605156]) in unrelated consanguineous Saudi Arabian families (1, 7-9). Since then, a number of non-Arabian families with the disorder have been identified confirming the disease's pan-ethnic nature (10, 11, 13). Based on its clinical features, we placed MOA (1) into the group of diseases associated with carpal, tarsal, and interphalangeal destruction, namely Torg (MIM #259600), Winchester (MIM #277950), Francois (MIM #221800), and Whyte-Hemingway syndromes (6). However, MOA is clinically distinguished from the other inherited osteolyses/arthritis syndromes, and in particular the other subgroup of carpal, tarsal disorders by the time-dependent severity of osteolysis and osteopenia and resultant joint contractures and ankylosis (1, 7-11, 13). In addition, characteristic MOA facial features include proptosis, frontal bossing, hypertelorism, brachycephaly, and a bulbous nose with flattened nasal bridge. MOA affecteds usually present with symptoms in the first two years of life with the development and progression of painful swellings with decreased range of motion or deformity of the proximal interphalangeal joints of hands, feet, and wrists (1, 7-9). Also, while MOA affecteds may have normal birth weights they also have decreased growth rates and are in the bottom decile of height and weight during early childhood.

Using linkage analysis and positional cloning and then molecular modeling and biochemical analysis, we recently demonstrated that MOA was caused by inactivating mutations in MMP2 (12). This finding was unexpected for two reasons. First, previous in vitro studies suggested that MMP-2 overexpression and increased activity, not deficiency, would result in osteolysis and arthritis. Indeed, MMP-2 had originally been isolated from the culture media of rheumatoid arthritis synovial tissue (13). Second, the Mmp2-/- mouse was originally characterized as overtly normal (2, see below).

Our discovery that MMP-2 deficiency results in a disorder characterized by apparent tissue destruction in the form of osteolysis and arthritis suggested a more complex role for this gelatinase in normal bone development and joint maintenance processes than predicted based on previous in vitro studies. The MMPs, a group of structurally related, secreted and membrane-associated proteinases, are involved in degrading the structural components of the ECM as well as other extracellular and non-matrix proteins (14-23). In recent years, the view of this family of proteases, including MMP-2, as simply “destructive enzymes” has evolved to now appreciate that MMPs can alter the microenvironment, modulate the activity of bioactive molecules, and regulate the activity of other proteases, thereby playing roles in cell attachment, proliferation, differentiation, and apoptosis (14-25).

Prior to our original report demonstrating mutations in the human MMP2 gene, Mmp2-/- mice had been generated to examine the role of MMP-2 in amyloid precursor protein (APP) processing and secretion in vivo (2). In these mice, the wildtype promoter and first exon of the Mmp2 gene were replaced with a pgk-neo cassette by gene targeting. These mice had no differences in APP processing ability and were described as overtly normal. Curiously, Mmp2-/- mice were approximately 15% smaller at birth than control littermates (2). Given the genetic basis of the human disease, we sought to investigate the murine growth/skeletal phenotype in a targeted fashion by specifically focusing our studies on a possible “subclinical” skeletal phenotype. Our findings reveal that Mmp2-/- mice partially recapitulate the human disease, as they have marked age-related bone density loss, bone abnormalities, craniofacial defects, sclerotic cranial sutures, and articular cartilage destruction. Moreover, and potentially identifying the cellular basis of MMP-2-mediated skeletal defects, we show that these changes are associated with developmentally restricted changes in osteoblast and osteoclast numbers in vivo, defects in cultured Mmp2-/- bone marrow cells to support osteoblast and osteoblast differentiation in culture, and that targeted MMP-2 inhibition using siRNA in cultured cell lines suppresses osteoblast proliferation.


Mmp2-/- mice have early and persistent craniofacial defects

The Mmp2-/- mice used in these studies were originally generated and described by Itoh et al (3). In accord with those original studies, and similar to the human MOA affecteds, Mmp2-/- homozygous mice were smaller and had markedly decreased MMP-2 expression and zymographic activity, whereas Mmp2+/- mice had half normal levels and activity (data not shown). In addition, we noted that all Mmp2-/- mice had a distinctive facial appearance when compared to Mmp2+/- and Mmp2+/+ littermates. Mmp2-/- mice had shortened, broad snouts, and were hyperteloric with narrower, taller skulls (Fig 1A). These differences were noted as early as 10 days after birth and many persisted throughout life. Caliper measurements on 4 week old mice revealed the snouts of Mmp2-/- mice were ~15% shorter (p<0.05) while intercanthal distances were increased by ~10%, trending towards statistical significance (p=0.08, Fig 1).

Figure 1
Mmp2-/- mice have abnormal craniofacial development. (A) Frontal and dorsal views of 12 week old Mmp2-/-, Mmp2+/-, and Mmp2+/+ mice. (B-D) Lateral, superior and inferior views, respectively, of μCT images of 6 month old Mmp2-/-, Mmp2+/-, and ...

Direct examination of the craniofacial bones by micro computed tomography (μCT) confirmed these findings and identified a number of additional features (Figs 1B-E). First, and as shown in Figure 1C, coronal, sagittal and lambdoid sutures were prominent and sclerotic; similar to the radiologic finding shared with a majority of MOA affecteds (5). Second, upper and lower jaw lengths, nose length, and overall skull lengths were decreased in Mmp2-/- mice and all were approximately 10% shorter by 12 weeks. Third, and confirming visual inspection of the skulls, we noted that the area between the right and left frontal-squamosal intersection at the temporal crest, “mid-cranial width” in Figure 1E, was also wider in Mmp2-/- mice and this difference increased with age. Mid-cranial width differences continued to increase from ~10% at 4 weeks to nearly 25% at 24 weeks. Fourth, we also detected a significant decrease in bone volume (13.4%) present in 24-week-old Mmp2-/- mice. Finally, and providing the first suggestion of a transient nature to some of the skeletal features, increased intercanthal distances and skull heights of just below 10% noted at 4 weeks of age, were no longer statistically different by 12 weeks.

MMP-2 deficiency results in erosive articular cartilage defects

In addition to craniofacial changes, MOA is also characterized by early onset arthritis and joint contractures. These are generally first noted in carpal and tarsal joints and spread proximally as the disease progresses. Most strikingly, while Mmp2-/- mice displayed no obvious difficulties in ambulation and/or maintaining joint flexibility throughout growth and into adulthood, even to 2 years of age, direct histologic examination revealed clear evidence of joint abnormalities. The knee joints of all 12-week-old Mmp2-/- mice examined (n = 5) revealed articular cartilage destruction and erosion of the underlying bone surface (Fig 2B,C), resulting in the loss of the smooth tibial and femoral surfaces present in all age-matched controls (n = 6; Fig 3A). These findings suggest a similarity with the rheumatoid pannus present in rheumatoid arthritis, to which the arthritic changes in MOA have been compared (1, 7). In addition, these findings highlight a likely misinterpretation in the literature that suggested that articular cartilage destruction noted in Mmp2-/- mice was solely secondary to exogenous antibody induction and not the result of MMP-2 deficiency (26).

Figure 2
Photomicrograph of femoral condyles and proximal tibiae from 12 week old Mmp2+/+ and Mmp2-/- mice. Normal femoral condyle structure in Mmp2+/+ is shown in A, with a smooth articular cartilage surface. The condylar surface in Mmp2-/- mice (B) is arthritic ...
Figure 3
Bone changes resulting from MMP-2 deficiency. Long bone radiographs from 6 month old mice showing (A) slight shortening of long bones and lower density in Mmp2-/- bone compared to Mmp2+/+ bone; this lower apparent density was confirmed and shown to be ...

Bone mineral density decreases in a time-dependent manner

Given these findings and the fact that MOA affecteds also have marked osteopenia, we next investigated the quality of Mmp2-/- mouse bones. As shown in Figure 3, X-ray analysis suggested decreased radio-opacity in Mmp2-/- mouse skeletons. In order to quantify and better characterize the possible temporal changes in bone mineral density (BMD), we used dual energy x-ray absorptometry (DEXA) over a series of ages during growth and early adulthood. The mice demonstrated a similar pattern of progressive loss of BMD as that previously shown radiographically for MOA affecteds (4) (Fig 3B). Starting at 5 weeks of age, Mmp2-/- mice began lagging in bone mineral accrual. At 10 weeks, Mmp2-/- mice had a significant 20% reduction in BMD compared to controls and this reduced level persisted throughout early adulthood (ages 10-20 weeks). Concomitant with normal BMD loss associated with aging in mice (>20 weeks), knockouts lost mineral density at a more rapid pace than Mmp2+/+ littermates after 20 weeks (Fig 3C).

MMP-2 deficiency results in both transient and long-lasting bone remodeling defects

Based on the progressive BMD reduction in Mmp2-/- mice we next examined their bone histology. We selected three time points encompassed within the BMD analysis: 4 days, 4 weeks, and 12 weeks. Dramatic differences were present in all 4-day old Mmp2-/- mice (Fig 3C-H). We first focused our attention on the defects in trabecular and cortical bone. As shown in Figure 3, Mmp2-/- mice (n=11) had significant defects in trabecular and cortical bone formation compared to Mmp2+/+ controls (n=11, Fig 3C-H). Primary spongiosa and trabeculae of Mmp2-/- mice appeared less dense and more scant. Mmp2-/- bone was notable for an overall paucity and chaotic organization of trabecular bone, marked hypocellularity (discussed below) and ragged metaphyseal and diaphyseal cortical bone more reminiscent of earlier embryologic time points when compared to Mmp2+/+ bone.

Unexpectedly, given the chronic nature of the human syndrome, many of these differences were no longer apparent in bones from 4 week old Mmp2-/- mice (Figure 3I). This apparent resolution occurred during a developmental period in mice normally correlating with rapid growth and comparable to the age when human MOA affecteds experience rapid bone loss. Overall, the cortical, trabecular and cell density differences between Mmp2-/- (n=7) and Mmp2+/+ (n=7) marrow cavities at this time point had by now mostly resolved. Nonetheless, stigmata of the abnormal Mmp2-/- cortical growth remained and appearing as streaky, hypostaining infiltrating regions enhanced under polarizing light. At 12 weeks of age, these histologic differences had almost completely disappeared (Fig 3J).

Early Mmp2-/- marrow has transiently decreased osteoblast and osteoclast numbers in vivo

To investigate the cellular basis of these findings, we analyzed specific cellular components of the marrow cavity by immunohistochemistry and directly quantitated cellular amounts by Bioquant image analysis. We used procollagen I as a marker of cells actively producing ECM, which corresponded to active osteoblasts, and Cathepsin K as a marker of osteoclasts. Osteoblast and osteoclast numbers were dramatically and transiently reduced in the immediate post-natal period in Mmp2-/- mice (Fig 4). Specifically, at 4 days the Mmp2-/- osteoblast (Fig4B,C,H; p<0.005) and osteoclast (Fig4D,E,H; p<0.005) numbers per trabecular bone surface area were decreased by more than half (Fig 4H). At 4 and 12 weeks, no significant differences in numbers of either cell type were identified (data not shown). Consistent with these immunohistochemical findings, proliferating cell nuclear antigen (PCNA) staining, a general marker of cell proliferation status, revealed an ~60% decrease in the number of proliferating cells in Mmp2-/- marrow at 4 days (Fig 4F-H; p<0.05). No significant changes at either 4 or 12 weeks were seen (data not shown).

Figure 4
Osteoblast, osteoclast, and proliferating cell numbers are decreased in Mmp2-/- early postnatal bones. (A) Low power micrograph showing distal femur head (fh), growth plate (gp), and metaphysis (met) of Mmp2+/+ 4 day old mouse. Box indicates approximate ...

Mmp2-/- bone marrow stromal cells and calvaria are ineffective in supporting osteoblast and osteoclast growth ex vivo

Given these in vivo findings, we next analyzed the effect of MMP-2 loss on the formation of osteoblastic and osteoclastic cells in explanted bone marrow stromal cell (BMC) cultures. In each of the assays used, Mmp2-/- BMC was ineffective in supporting osteoblast or osteoclast growth despite the fact that comparable numbers of cells were isolated and plated from both marrow sources. In the first set of assays, analyzing osteoblast growth from marrow of 8 week old mice, Mmp2+/+ cells formed numerous, large, alkaline phosphatase-positive osteoblast colonies. In marked contrast, only relatively infrequent and small colonies were identified in Mmp2-/- cultures (Fig 5A). When we modified the culturing procedure to reduce the washing stringency following plating of BMCs, to account for possible differences in adherence, Mmp2-/- alkaline phosphatase positive colonies were now identified (Fig. 5B), but these smaller colonies still represented less than 20% of control (p<0.0005, Fig 5C)

Figure 5
MMP-2 deficiency results in decreased osteoblast- and osteoclastogenesis ex vivo. (A-B) Mmp2-/- bone marrow derived alkaline phosphatase positive colonies are fewer, smaller, and sparser than those derived from Mmp2+/+ BMC under either (A) normal or (B) ...

In agreement with these findings revealing defects in Mmp2-/- osteoblastic potential from BMC, we also consistently obtained smaller cell yields and observed lower in vitro growth rates using calvarial cell populations isolated from 4 day old Mmp2-/- mice. Notably, Mmp2+/+ osteoblasts reached confluence in 8 days, undergoing 1.5 population doublings, whereas Mmp2-/- cells plated at the same density increased their numbers by only 50% over the same time period (Fig 5D,E).

To assess osteoclast growth potential, Mmp2+/+ and Mmp2-/- BMC or splenic cells were harvested and cultured in the presence of receptor activator of NFκB ligand (RANKL) and M-CSF. Osteoclasts, identified as multinucleated, tartrate resistant acid phosphatase (TRAP) positive cells, were then counted (Fig 5F-H). The number of osteoclasts detected in cultures derived from Mmp2-/- BMC (p<0.00005) and splenic cells (p< 0.0005) were reduced by ~60% compared to Mmp2+/+ osteoclasts despite being similarly harvested, cultured, and assayed (Fig 5H).

Targeted siRNA-mediated MMP2 knockdown decreases osteoblast growth in cultured cell lines

Having identified these differences in MMP2-/- osteoblasts both in vivo and ex vivo from mice which were globally deficient in MMP-2, we wondered whether targeted loss of this enzyme in cultured wild type cells, devoid of supporting stromal cells and environment, would also affect their growth. Therefore, we experimentally inhibited MMP-2 expression in two well-characterized osteoblastic cell lines, human SaOS2 and mouse MC3T3, by siRNA-mediated gene silencing. Independent pools of siRNA oligos specific to human MMP2 or mouse Mmp2 (siMMP2) were used to target multiple regions of the mRNA transcript. Using transient transfection in SaOS2 cells, MMP2 mRNA levels were maximally reduced 80% at 48 hours post-transfection and 65% by 72 hours (Fig 6A). MMP-2 protein activity levels in conditioned media were reduced by greater than 70% at both these time points (Fig 6B). The effect of targeted downregulation of MMP-2 was an ~30% reduction of cellular proliferation at 48 hours after transfection (p<0.05), and ~20% reduction at 72 hours (p<0.05, Fig 6C).

Figure 6
MMP-2 inhibition by siRNA decreases proliferation of osteoblast cell lines. (A & C) MMP-2 mRNA expression levels in (A) SaOS2 and (D) MC3T3 cells transfected with pools of siRNA targeting MMP-2 (siMMP-2) are reduced by up to 80% compared to cells ...

To examine the ability of exogenous MMP-2 to compensate for reduced endogenous expression, stable siRNA expressing SaOS2 cell lines were generated and two independent clones characterized. Stable siMMP2 clones A and B showed ~70% reduction of the MMP2 RNA transcript and protein, and cellular proliferation was reduced in each by greater than 35% (p< 0.05, Fig 6G) regardless of the presence or absence of FBS, which contains high levels of MMP-2. We next analyzed the effect of exchanging the conditioned media between siMMP2 cells and control, empty vector cells on proliferation. Again, no effect was seen on proliferation rates as siMMP2 clones still proliferated at rates ~50% less than empty vector control cells (p<0.05, Fig 6G).

The general nature of this effect on cellular proliferation was then shown in the second cell line. In siMMP2 transfected MC3T3 cells, Mmp2 RNA levels were reduced by ~70% at 24 hours and ~50% at 48 hours (Fig 6D) with a greater than 50% decrease in protein activity at both time points (Fig 6E). Similar to the transient transfection results obtained in the SaOS2 cell line, MMP-2 inhibition in MC3T3 cells reduced proliferation rates by 25% at both 24 and 48 hour time points (p<0.05, Fig 6F).


A targeted approach to the molecular basis of skeletal development can begin with the genetic analysis of a monogenic syndrome whose major phenotype derails the normal patterns of skeletal growth and homeostasis. Each monogenic trait can theoretically illuminate one component in the complex, multifactorial pathway underlying normal and pathologic bone biology. Using this paradigm, a number of highly relevant and completely unexpected targets have been identified from rare inherited skeletal diseases. For example, familial expansile polyostotic osteolysis (MIM #174810), results from mutations in receptor activator of NFκB (RANK), a molecule essential in osteoclast formation (27). Osteoporosis-pseudoglioma syndrome (MIM #259770) and the “high bone mass trait” (MIM #601884) are caused by inactivating and activating mutations, respectively, in the low-density lipoprotein receptor-related protein 5 (LRP5), a Wnt co-receptor critical to Wnt-mediated osteoblastic proliferation and differentiation (28, 29). Moreover, the initial identification of many genes now known to regulate osteoblast and osteoclast differentiation, growth and function – Runx2 (CBFA, OSF2)[cleidocranial dysplasia], Msx2 [cleidocranial dysplasia, Boston-type], Sox9 [campomelic dysplasia], FGFRs 1-3 [Pfeiffer syndrome, Crouzon, and achondroplasia, respectively], and cathepsin K [Pycnodysostosis] - were first defined from the analysis of rare genetic syndromes (reviewed in (30-34)).

MOA represents the unique intersection of a rare genetic disease and an a priori candidate gene, albeit, by a completely counterintuitive mechanism (35, 36). Based on their substrate specificity and biologic plausibility, MMPs would be expected to play an important role in skeletogenesis and since our initial genetic studies, evidence now exists for three other MMPs, MMP-9, MMP-13 and MT1-MMP, in bone development (37-44). Taken together, our current results demonstrate that MMP-2 deficiency results in a unique bone phenotype in mice comparable in many respects to that seen in the human disease and, that in part, this results from a previously unappreciated role for MMP-2 activity on osteoblast and osteoclast behavior and development.

First and foremost, these studies begin to unravel the paradox generated by our initial gene discovery studies in MOA, namely, “How does functional loss of a single proteolytic enzyme result in a “vanishing bone” disorder?” Taken together, the in vivo and ex vivo defects observed in Mmp2-/- osteoblasts and osteoclasts and in cultured osteoblast cell lines following siRNA-mediated MMP-2 inhibition, supports the hypothesis that MMP-2 directly affects osteoblast and osteoclast growth. Most intriguingly, this deficit is not recovered with the replacement of MMP-2 in the growth media. A question which therefore arises from these studies and is consonant with the increasing recognition of the importance of the MMP family in complex cellular activities (beyond their initially defined roles in matrix degradation) including cell attachment, proliferation, differentiation, and apoptosis (30-34) is the identity of the physiologically relevant MMP-2 substrates that can induce these effects on osteoblasts and osteoclasts. Ultimately, the availability of a murine model and MMP-2 deficient cell lines will provide the framework resources for addressing these questions.

As we have shown, many of the clinical features distinctive of MOA are replicated and paralleled by Mmp2-/-mice. These include craniofacial defects, osteopenia, and joint defects consistent with an underlying arthritis. Interestingly, histologic evidence of arthritis in Mmp2-/- mice had been previously noted but was believed to be the result of an increased response to arthritogenic insult, and was therefore not appreciated as a direct result of enzymatic deficiency (26). In these studies, Mmp2-/- mice were injected with an arthritogenic monoclonal antibody (mAB) cocktail and then clinically and histologically graded for evidence of arthritic changes. Curiously, only mAB treated Mmp2-/- mice were examined but not untreated Mmp2-/- control mice (26). Thus, given our present findings that Mmp2-/- mice suffer from arthritic erosions, the previous results and their original interpretation must be re-examined.

Similarly, MT1-MMP deficient mice possess defects in both endochondral and intramembranous bone formation and, and as we originally noted (12), share a number of features consistent with MOA (41, 42). Notwithstanding that MT1-MMP deficient mice die within a few weeks of age, these mice develop craniofacial dysmorphia, osteopenia, arthritis, dwarfism and soft-tissue fibrosis. These analogous findings are particularly important since MT1-MMP is known to be a critical activator of MMP-2. Physiologic activation is regulated by a cell surface interaction and achieved following the formation of a tri-molecular complex between MMP-2, MT1-MMP, and tissue inhibitor of metalloproteinase (TIMP)-2 (43, 44). As would be expected, cultured fibroblasts from these mice were unable to fully activate pro-MMP-2 (41, 42). Since it was believed that MMP-2 knockouts lacked a phenotype, the skeletal consequences of MT1-MMP deletion have been ascribed solely to the loss of MT1-MMP activity. In light of our MMP-2 human and mouse results, this interpretation should be re-examined. Of note, TIMP-2 knockout mice, also deficient in pro-MMP-2 activation, appear phenotypically normal and develop and procreate indistinguishably from wildtype littermates (45, 46). To date, no skeletal-investigations have been reported.

A particularly intriguing aspect of the Mmp2-/- mouse phenotype is the apparent time-dependent resolution of certain aspects of the skeletal changes. Indeed, a recent study by Inoue et al (47) of aged Mmp2-/- mice failed to detect a statistically significant difference in osteoblast and osteocyte activity, and instead identified specific failure of osteocytic canalicular formation. (47). Persistent changes, such as the craniofacial and skeletal anomalies, were similar to those reported here. While unexpected, developmentally-limited tissue and cellular phenotypes are not unprecedented in studies of MMP deficiency. For example, MMP-9, along with MMP-2, is one of two known gelatinases in the MMP family and shares extensive structural similarities to the MMP2 gene. The only developmentally apparent abnormalities noted in Mmp9-/- mice were a 10% shorter long bone length and alteration of endochondral bone formation (48). In these mice, the growth abnormality resulted from lengthened zones of hypertrophic cartilage in chondrocytes owing to impaired vascular invasion and delayed apoptosis of terminal hypertrophic chondrocytes. In addition, MMP-9 is required for invasion of osteoclasts and may be mediated through the ability of MMP-9 to make VEGF bioavailable from the ECM (49, 50). Beyond this example, mutations in MMP13 result in the human disease spondoepimetaphyseal dysplasia, Missouri type (SEMD-(MO) [OMIM 602111]) and engineered deletion of Mmp13 in mice results in similar skeletal defects (51). Interestingly, in humans, the radiographic changes seen in metaphyses, epiphyses, and vertebrae during childhood resolve in adults (52) and defects in murine Mmp13-/- growth plates resolve by 12 weeks while increases in trabecular bone persist for longer but are then also lost by one year (51). In both cases, it was postulated that upregulation of compensatory pathways allowed this resolution in the absence of critical MMPs.

The counterintuitive finding that loss of a proteolytic enzyme causes a “vanishing bone” and arthritic syndrome in humans (12) began our investigation into a more complex role for MMP-2 in bone biology. Knowledge of the human disease, coupled with findings in a species as evolutionally distant as zebrafish, wherein loss of MMP-2 leads to defects during embryogenesis (53), made the apparent lack of a mouse phenotype particularly enigmatic. In this study of the Mmp2-/- mouse, we demonstrate that despite an attenuation of the MOA phenotype, many aspects of the disease, including abnormal craniofacial development, bone and joint growth, and bone mineral density deficiency, are recapitulated and maintained in the mice. Postnatal bone growth and development is retarded, due at least in part to a proliferation defect of the marrow precursor cells, osteoblasts, and osteoclasts both in vivo and in vitro. Interestingly, a previous report had also demonstrated that MMP-2 inhibition could regulate proliferation, albeit in glomerular mesangial cells (54). To our knowledge, ours is the first report demonstrating a direct effect of MMP-2 loss on osteoblasts and osteoclasts and suggest that in turn, these findings may ultimately provide novel insight into the mechanisms coupling osteoblast bone formation and osteoclast resorption that underlies normal and pathologic bone remodeling.

Materials and Methods


Generation of the Mmp2-/- mice used in these studies was described by Itoh et al. (3) by targeted deletion of the promoter, 5′UTR, and exon 1 of the Mmp2 gene. Genotyping was done by PCR of genomic DNA extracted from tail snips. Primers amplifying exon 1 were used to confirm the presence of the wildtype allele, while neo cassette primers targeted the null allele. Exon 1 forward: CAGACTTCCCTGGTGGCT; Exon 1 reverse: CACCATCGCCCATCATCAAG; neomycin forward : AGGATCTCCTGTCATCTCACCTTGCTCC; neomycin reverse: AAGAACTCGTCACGAAGGCGATAGAAGG. The reaction conditions for Exon 1 primers were: 95°C for 10 minutes, followed by 35 cycles of 95°C for 30 sec, 55°C for 30 sec, 72°C for 45sec, and a final extension at 72°C. Reaction conditions for neomycin primers were identical except for an annealing temperature of 62°C.

All procedures were performed in accordance with an institutionally approved protocol for the use of animals in research.

Micro-computed tomography (μCT)

Skulls were scanned on a General Electric Explore Locus μCT. The scans were done at 80 kilovolts peak and 500 μamps with an exposure time of 3000 milliseconds. 200 views were done in a 1° increment with a frame average of 2 over 180°. Scans were reconstructed using GE reconstruction software yielding a post reconstruction resolution of 20 microns.

Histology and immunohistochemistry

Bones were isolated from age-matched Mmp2-/- and Mmp2+/+ mice and fixed in 10% phosphate buffered formalin overnight and decalcified in cold EDTA. Four-micrometer thick sagittal sections were stained with standard H&E. Sections were stained with antibodies to procollagen I, cathepsin K, or PCNA (Santa Cruz Biotechnology, Inc) and counterstained with Hematoxylin. Positive cells were counted using Bioquant TCW software (Nashville, TN) and normalized to the trabecular bone surface area.

Cell Culture and siRNA silencing


Marrow from femurs and tibias of 8-week mice was extracted in αMEM with FBS and Penicillin/Streptomycin antibiotic. Cells were allowed to adhere and the non-adherent population removed. After 2 days in culture, cells were re-plated in osteoblast differentiation media contain 1μM ascorbic acid and 5μM dexamethasone at 30,000 cells/ml. Cell colonies formed were stained for alkaline phosphatase using the SigmaFas AlkPhos staining kit (Sigma).

Calvarial osteoblastic cells were prepared by serial digestion of calvaria of 4-day-old Mmp2+/+ or Mmp2-/- mice with 200U/ml collagenase type I. Calvaria were isolated, rinse with PBS and briefly decalcified in EDTA. After serial digestions with collagenase, fractions 2-5 were pooled, washed in αMEM, and plated at 8×104 cells/ml in osteoblast differentiation media contain 1μM ascorbic acid and 5μM dexamethasone.

SiMMP2 oligo pools targeting either human MMP2 or mouse Mmp2 were obtained from Dharmacon, Inc. Transfections were done in serum free media, using lipofectamine 2000 (Invitrogen) and 100pmol of oligo per well of a 12 well plate as previously described (55).

Forward and reverse sequences for MMP-2 shRNA construct were 5′-GATCCCCGACAGTGGATGATGCCTTTTTCAAGAGAAAAGGCATCATCCACTGTCTTTTTGGAAA-3′ and 5′AGCTTTTCCAAAAAGACAGTGGATGATGCCTTTTCTCTTGAAAAAGGCATCATCCACTGTCGGG-3′, respectively. Oligonucleotides were cloned into pSUPER.retro vector (Oligoengine, Seattle, WA, USA) according to manufacturer's recommendations. Briefly, forward and reverse oligonucleotides were annealed and then ligated into the BglII-HindIII cleavage site within the pSUPER.retro vector prelinearized with the same restriction enzymes. pSUPER.retro-siMMP2 vector or empty vector was transfected into Phi-NX (‘Phoenix’) packaging cell line to produce ecotropic retroviral supernatants. Phoenix cells were cultured in DMEM supplemented with 10% FBS. The day before transfection, Phoenix cells were seeded in 10 cm dishes (3×106 cells/dish) in order to reach 60% confluence at the time of transfection. Cells were transfected with 10 μg of viral vector DNA using Lipofectamine 2000 according to manufacturer's protocol (Invitrogen, San Diego, CA, USA). At 48h after transfection, culture medium was filtered through a 0.45 μm filter and the viral supernatant was used for SaOS2 cell infection after addition of 4 μg/ml of polybrene (Sigma, USA). After infection, SaOS2 cells were incubated at 37°C in 5% CO2 for 5h, followed by a second round of infection. After a second 5h incubation, medium was changed with fresh medium and SaOS2 cells were allowed to recover for 48h at 37°C in 5% CO2. Infected cells were selected by adding puromycin (1 μg/ml) and clones were picked and grown under further selection. MMP2 expression in empty vector and siMMP2 infected SaOS2 cells was analyzed by real-time polymerase chain reaction (RT-PCR) and zymography as described below.


Bone marrow cells and splenic cells were harvested and plated at 6.25×104 cells/cm2 in the presence of 30ng/ml M-CSF and 100ng/ml RANKL to induce osteoclast format. After 9 days, cells were fixed in 4% paraformaldehyde and stained for TRAP activity (Sigma).

RNA isolation and quantitative RT-PCR

RNA was collected (Qiagen RNeasy Mini Kit) and treated with DNase (Qiagen). A total of 0.5 μg of RNA was reverse-transcribed per reaction using first-strand complementary DNA synthesis with random primers (Iscript, Bio-Rad Laboratories). Quantitative real-time PCR was performed on an ABI PRISM 7900HT sequence detection system (Applied Biosystems). Cycle number values were normalized with both GAPDH and β-actin values. Values were analyzed as fold change compared to siNTC or empty vector control values. Data shown is the average of three separate experiments done in triplicate. Statistical significance was determined by comparing fold change using the unpaired, two-tailed, students t test assuming equal variances.

Gelatin Zymography

5μg of conditioned media was run on 10% gelatin zymograms gels (Novex), which were developed and stained according to manufacturers instructions as described previously (2).

Proliferation Assay

One μCi/mL [3H]thymidine (Amersham) was added to transiently transfected or stable cell lines with either siNTC or siMMP2 and incubated for 8hrs. Cells were washed four times with ice-cold PBS and solubilized in 0.25% sodium hydroxide/0.25% SDS. After neutralization with hydrochloric acid (1N), disintegrations per minute were estimated by liquid scintillation counting.

For conditioned media replacement assays of siMMP2 stable lines, cells were plated at equal densities. The following morning, media was removed and cells washed 2 times with serum-free media. Serum-free conditioned media from control empty vector clone (E) was added to siMMP2 clones (A and B), and serum-free conditioned media from siMMP2 clones added to control clones. Proliferation assay was performed over 4 hours as described above after a 30 minute pre-incubation in the “replaced” media.


The work reported here was supported by NIH award AR47074 (to S. Apte). Micro-CT was facilitated by the Imaging Core established through award of a Musculoskeletal Research Core Center to the Cleveland Clinic (NIH AR50953 to V. Hascall and S. Apte).


Multicentric Osteolysis and Arthropathy
amyloid precursor protein
micro computed tomography
bone mineral density
dual energy X-ray absorptometry
proliferating cell nuclear antigen
bone marrow cell
RANK ligand
tartrate resistant acid phosphatase
receptor activator of NFκB
low-density lipoprotein receptor-related protein 5


Conflicts of Interest: The Authors state no conflicts of interest.


1. Al Aqeel A, Al Sewairi W, Edress B, Gorlin RJ, Desnick RJ, Martignetti JA. Inherited multicentric osteolysis with arthritis: A variant resembling Torg syndrome in a Saudi family. Am J Med Genet. 2000;93:11–18. [PubMed]
2. Itoh T, Ikeda T, Gomi H, Nakao S, Suzuki T, Itohara S. Unaltered secretion of beta-amyloid precursor protein in gelatinase A (matrix metalloproteinase 2)-deficient mice. J Biol Chem. 1997;272:22389–22392. [PubMed]
3. Torg JS, Steel HH. Essential osteolysis with nephropathy. A review of the literature and case report of an unusual syndrome. J Bone Joint Surg Am. 1968;50:1629–1638. [PubMed]
4. Tyler T, Rosenbaum HD. Idiopathic multicentric osteolysis. AJR Am J Roentgenol. 1976;126:23–31. [PubMed]
5. Jackson J. A boneless arm. Boston Med Surg J. 1838;18:368–369.
6. Lachman RS. International nomenclature and classification of the osteochondrodysplasias (1997) Pediatr Radiol. 1998;28:737–744. [PubMed]
7. Al-Mayouf SM, Majeed M, Hugosson C, Bahabri S. New form of idiopathic osteolysis: Nodulosis, arthropathy and osteolysis (NAO) syndrome. Am J Med Genet. 2000;93:5–10. [PubMed]
8. Al-Otaibi L, Al-Mayouf SM, Majeed M, Al-Eid W, Bahabri S, Hugosson CO. Radiological findings in NAO syndrome. Pediatr Radiol. 2002;32:523–528. [PubMed]
9. Dowling Oonagh, John A Martignetti. Human MMP-2 Deficiency: The Multicentric Osteolysis with Nodulosis and Arthritis (MONA) and Winchester Syndromes. In: Epstein CJ, Erickson RP, Wynshaw-Boris A, editors. Inborn Errors of Development: The Molecular Basis of Clinical Disorders of Morphogenesis. Oxford University Press; New York: 2004. pp. 977–986.
10. Zankl A, Bonafe L, Calcaterra V, Di Rocco M, Superti-Furga A. Winchester syndrome caused by a homozygous mutation affecting the active site of matrix metalloproteinase 2. Clin Genet. 2005;67:261–266. [PubMed]
11. Rouzier C, Vanatka R, Bannwarth S, Philip N, Coussement A, Paquis-Flucklinger V, Lambert JC. A novel homozygous MMP2 mutation in a family with winchester syndrome. Clin Genet. 2006;69:271–276. [PubMed]
12. Martignetti JA, Aqeel AA, Sewairi WA, Boumah CE, Kambouris M, Mayouf SA, Sheth KV, Eid WA, Dowling O, Harris J, et al. Mutation of the matrix metalloproteinase 2 gene (MMP2) causes a multicentric osteolysis and arthritis syndrome. Nat Genet. 2001;28:261–265. [PubMed]
13. Sorsa T, Konttinen YT, Lindy O, Ritchlin C, Saari H, Suomalainen K, Eklund KK, Santavirta S. Collagenase in synovitis of rheumatoid arthritis. Semin Arthritis Rheum. 1992;22:44–53. [PubMed]
14. Murphy G, Cawston TE, Galloway WA, Barnes MJ, Bunning RA, Mercer E, Reynolds JJ, Burgeson RE. Metalloproteinases from rabbit bone culture medium degrade types IV and V collagens, laminin and fibronectin. Biochem J. 1981;199:807–811. [PMC free article] [PubMed]
15. Agnihotri R, Crawford HC, Haro H, Matrisian LM, Havrda MC, Liaw L. Osteopontin, a novel substrate for matrix metalloproteinase-3 (stromelysin-1) and matrix metalloproteinase-7 (matrilysin) J Biol Chem. 2001;276:28261–28267. [PubMed]
16. McQuibban GA, Gong JH, Tam EM, McCulloch CA, Clark-Lewis I, Overall CM. Inflammation dampened by gelatinase A cleavage of monocyte chemoattractant protein-3. Science. 2000;289:1202–1206. [PubMed]
17. Stamenkovic I. Extracellular matrix remodeling: The role of matrix metalloproteinases. J Pathol. 2003;200:448–464. [PubMed]
18. Shubayev VI, Myers RR. Endoneurial remodeling by TNFalpha- and TNFalpha-releasing proteases. A spatial and temporal co-localization study in painful neuropathy. J Peripher Nerv Syst. 2002;7:28–36. [PubMed]
19. Schonbeck U, Mach F, Libby P. Generation of biologically active IL-1 beta by matrix metalloproteinases: A novel caspase-1-independent pathway of IL-1 beta processing. J Immunol. 1998;161:3340–3346. [PubMed]
20. Fowlkes JL, Thrailkill KM, Serra DM, Suzuki K, Nagase H. Matrix metalloproteinases as insulin-like growth factor binding protein-degrading proteinases. Prog Growth Factor Res. 1995;6:255–263. [PubMed]
21. Levi E, Fridman R, Miao HQ, Ma YS, Yayon A, Vlodavsky I. Matrix metalloproteinase 2 releases active soluble ectodomain of fibroblast growth factor receptor 1. Proc Natl Acad Sci U S A. 1996;93:7069–7074. [PMC free article] [PubMed]
22. Martinez A, Oh HR, Unsworth EJ, Bregonzio C, Saavedra JM, Stetler-Stevenson WG, Cuttitta F. Matrix metalloproteinase-2 cleavage of adrenomedullin produces a vasoconstrictor out of a vasodilator. Biochem J. 2004;383:413–418. [PMC free article] [PubMed]
23. Sternlicht MD, Bissell MJ, Werb Z. The matrix metalloproteinase stromelysin-1 acts as a natural mammary tumor promoter. Oncogene. 2000;19:1102–1113. [PMC free article] [PubMed]
24. McCawley LJ, Matrisian LM. Matrix metalloproteinases: They're not just for matrix anymore! Curr Opin Cell Biol. 2001;13:534–540. [PubMed]
25. Mott JD, Werb Z. Regulation of matrix biology by matrix metalloproteinases. Curr Opin Cell Biol. 2004;16:558–564. [PMC free article] [PubMed]
26. Itoh T, Matsuda H, Tanioka M, Kuwabara K, Itohara S, Suzuki R. The role of matrix metalloproteinase-2 and matrix metalloproteinase-9 in antibody-induced arthritis. J Immunol. 2002;169:2643–2647. [PubMed]
27. Hughes AE, Ralston SH, Marken J, Bell C, MacPherson H, Wallace RG, van Hul W, Whyte MP, Nakatsuka K, Hovy L, et al. Mutations in TNFRSF11A, affecting the signal peptide of RANK, cause familial expansile osteolysis. Nat Genet. 2000;24:45–48. [PubMed]
28. Little RD, Carulli JP, Del Mastro RG, Dupuis J, Osborne M, Folz C, Manning SP, Swain PM, Zhao SC, Eustace B, et al. A mutation in the LDL receptor-related protein 5 gene results in the autosomal dominant high-bone-mass trait. Am J Hum Genet. 2002;70:11–19. [PMC free article] [PubMed]
29. Gong Y, Slee RB, Fukai N, Rawadi G, Roman-Roman S, Reginato AM, Wang H, Cundy T, Glorieux FH, Lev D, et al. LDL receptor-related protein 5 (LRP5) affects bone accrual and eye development. Cell. 2001;107:513–523. [PubMed]
30. Zelzer E, Olsen BR. The genetic basis for skeletal diseases. Nature. 2003;423:343–348. [PubMed]
31. Harada S, Rodan GA. Control of osteoblast function and regulation of bone mass. Nature. 2003;423:349–355. [PubMed]
32. Baldock PA, Eisman JA. Genetic determinants of bone mass. Curr Opin Rheumatol. 2004;16:450–456. [PubMed]
33. Karsenty G, Wagner EF. Reaching a genetic and molecular understanding of skeletal development. Dev Cell. 2002;2:389–406. [PubMed]
34. Kornak U, Mundlos S. Genetic disorders of the skeleton: A developmental approach. Am J Hum Genet. 2003;73:447–474. [PMC free article] [PubMed]
35. Vu TH. Don't mess with the matrix. Nat Genet. 2001;28:202–203. [PubMed]
36. Krane SM. Genetic control of bone remodeling--insights from a rare disease. N Engl J Med. 2002;347:210–212. [PubMed]
37. Holmbeck K, Bianco P, Pidoux I, Inoue S, Billinghurst RC, Wu W, Chrysovergis K, Yamada S, Birkedal-Hansen H, Poole AR. The metalloproteinase MT1-MMP is required for normal development and maintenance of osteocyte processes in bone. J Cell Sci. 2005;118:147–156. [PubMed]
38. Karsdal MA, Andersen TA, Bonewald L, Christiansen C. Matrix metalloproteinases (MMPs) safeguard osteoblasts from apoptosis during transdifferentiation into osteocytes: MT1-MMP maintains osteocyte viability. DNA Cell Biol. 2004;23:155–165. [PubMed]
39. Holmbeck K, Bianco P, Chrysovergis K, Yamada S, Birkedal-Hansen H. MT1-MMP-dependent, apoptotic remodeling of unmineralized cartilage: A critical process in skeletal growth. J Cell Biol. 2003;163:661–671. [PMC free article] [PubMed]
40. Karsdal MA, Larsen L, Engsig MT, Lou H, Ferreras M, Lochter A, Delaisse JM, Foged NT. Matrix metalloproteinase-dependent activation of latent transforming growth factor-beta controls the conversion of osteoblasts into osteocytes by blocking osteoblast apoptosis. J Biol Chem. 2002;277:44061–44067. [PubMed]
41. Holmbeck K, Bianco P, Caterina J, Yamada S, Kromer M, Kuznetsov SA, Mankani M, Robey PG, Poole AR, Pidoux I, et al. MT1-MMP-deficient mice develop dwarfism, osteopenia, arthritis, and connective tissue disease due to inadequate collagen turnover. Cell. 1999;99:81–92. [PubMed]
42. Zhou Z, Apte SS, Soininen R, Cao R, Baaklini GY, Rauser RW, Wang J, Cao Y, Tryggvason K. Impaired endochondral ossification and angiogenesis in mice deficient in membrane-type matrix metalloproteinase I. Proc Natl Acad Sci U S A. 2000;97:4052–4057. [PMC free article] [PubMed]
43. Strongin AY, Collier I, Bannikov G, Marmer BL, Grant GA, Goldberg GI. Mechanism of cell surface activation of 72-kDa type IV collagenase. isolation of the activated form of the membrane metalloprotease. J Biol Chem. 1995;270:5331–5338. [PubMed]
44. Butler GS, Butler MJ, Atkinson SJ, Will H, Tamura T, Schade van Westrum S, Crabbe T, Clements J, d'Ortho MP, Murphy G. The TIMP2 membrane type 1 metalloproteinase “receptor” regulates the concentration and efficient activation of progelatinase A. A kinetic study. J Biol Chem. 1998;273:871–880. [PubMed]
45. Caterina JJ, Yamada S, Caterina NC, Longenecker G, Holmbeck K, Shi J, Yermovsky AE, Engler JA, Birkedal-Hansen H. Inactivating mutation of the mouse tissue inhibitor of metalloproteinases-2(timp-2) gene alters proMMP-2 activation. J Biol Chem. 2000;275:26416–26422. [PubMed]
46. Caterina J, Caterina N, Yamada S, Holmback K, Longenecker G, Shi J, Netzel-Arnett S, Engler J, Yermovski A, Windsor J, et al. Murine TIMP-2 gene-targeted mutation. Ann N Y Acad Sci. 1999;878:528–530. [PubMed]
47. Inoue K, Mikuni-Takagaki Y, Oikawa K, Itoh T, Inada M, Noguchi T, Park JS, Onodera T, Krane SM, Noda M, et al. A crucial role for matrix metalloproteinase 2 in osteocytic canalicular formation and bone metabolism. J Biol Chem. 2006;281:33814–33824. [PubMed]
48. Vu TH, Shipley JM, Bergers G, Berger JE, Helms JA, Hanahan D, Shapiro SD, Senior RM, Werb Z. MMP-9/gelatinase B is a key regulator of growth plate angiogenesis and apoptosis of hypertrophic chondrocytes. Cell. 1998;93:411–422. [PMC free article] [PubMed]
49. Engsig MT, Chen QJ, Vu TH, Pedersen AC, Therkidsen B, Lund LR, Henriksen K, Lenhard T, Foged NT, Werb Z, et al. Matrix metalloproteinase 9 and vascular endothelial growth factor are essential for osteoclast recruitment into developing long bones. J Cell Biol. 2000;151:879–889. [PMC free article] [PubMed]
50. Bergers G, Brekken R, McMahon G, Vu TH, Itoh T, Tamaki K, Tanzawa K, Thorpe P, Itohara S, Werb Z, et al. Matrix metalloproteinase-9 triggers the angiogenic switch during carcinogenesis. Nat Cell Biol. 2000;2:737–744. [PMC free article] [PubMed]
51. Stickens D, Behonick DJ, Ortega N, Heyer B, Hartenstein B, Yu Y, Fosang AJ, Schorpp-Kistner M, Angel P, Werb Z. Altered endochondral bone development in matrix metalloproteinase 13-deficient mice. Development. 2004;131:5883–5895. [PMC free article] [PubMed]
52. Patel AC, McAlister WH, Whyte MP. Spondyloepimetaphyseal dysplasia: Clinical and radiologic investigation of a large kindred manifesting autosomal dominant inheritance, and a review of the literature. Medicine (Baltimore) 1993;72:326–342. [PubMed]
53. Zhang J, Bai S, Zhang X, Nagase H, Sarras MP., Jr The expression of gelatinase A (MMP-2) is required for normal development of zebrafish embryos. Dev Genes Evol. 2003;213:456–463. [PubMed]
54. Turck J, Pollock AS, Lee LK, Marti HP, Lovett DH. Matrix metalloproteinase 2 (gelatinase A) regulates glomerular mesangial cell proliferation and differentiation. J Biol Chem. 1996;271:15074–15083. [PubMed]
55. Narla G, DiFeo A, Yao S, Banno A, Hod E, Reeves HL, Qiao RF, Camacho-Vanegas O, Levine A, Kirschenbaum A, et al. Targeted inhibition of the KLF6 splice variant, KLF6 SV1, suppresses prostate cancer cell growth and spread. Cancer Res. 2005;65:5761–5768. [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...