Synovial Joint Formation during Mouse Limb Skeletogenesis: Roles of Indian Hedgehog Signaling

Abstract

Indian hedgehog (Ihh) has been previously found to regulate synovial joint formation. To analyze mechanisms, we carried out morphological, molecular and cell fate map analyses of interzone and joint development in wild type and Ihh^−/− mouse embryo long bones. We found that Ihh^−/− cartilaginous digit anlagen remained fused and lacked interzones or mature joints, whereas wrist skeletal elements were not fused but their joints were morphologically abnormal. E14.5 and E17.5 wild type digit and ankle anlagen expressed hedgehog target genes including Gli1 and Gli2 and interzone-associated genes including Gdf5, Erg and tenascin-C, but expression of all these genes was barely detectable in mutant joints. For cell fate map analysis of joint progenitor cells, we mated Gdf5-Cre^+/−/Rosa R26R^+/− double transgenic mice with heterozygous Ihh^+/− mice and monitored reporter β-galactosidase activity and gene expression in triple-transgenic progeny. In control Gdf5-Cre^+/−/R26R^+/−/Ihh^+/− limbs, reporter-positive cells were present in developing interzones, articulating layers and synovial lining tissue and absent from underlying growth plates. In mutant Gdf5-Cre^+/−/R26R^+/−/Ihh^−/− specimens, reporter-positive cells were present also. However, the cells were mostly located around the prospective and uninterrupted digit joint sites and, interestingly, still expressed Erg, tenascin-C and Gdf5. Topographical analysis revealed that interzone and associated cells were not uniformly distributed, but were much more numerous ventrally. A similar topographical bias was seen for cavitation process and capsule primordia formation. In sum, Ihh is a critical and possibly direct regulator of joint development. In its absence, distribution and function of Gdf5-expressing interzone-associated cells are abnormal, but their patterning at prospective joint sites still occurs. The joint-forming functions of the cells appear to normally involve a previously unsuspected asymmetric distribution along the ventral-to-dorsal plane of the developing joint.

INTRODUCTION

The biology and developmental biology of synovial joints continue to be the focus of intense research activity owing to the fact that joints are essential for skeletal function and quality of life and are susceptible to malfunction during natural aging and in congenital or acquired conditions including osteoarthritis. Classic embryological studies showed several years ago that the onset of limb skeletogenesis involves formation of an uninterrupted Y-shaped mesenchymal condensation made of a seemingly uniform population of cells.¹ The proximal arm of the Y-shaped condensation corresponds to the future femur or humerus and the two arms of the Y will give rise to tibia/fibula or radius/ulna, but the condensation initially displays no overt morphological sign of knee or elbow joint. Additional uninterrupted mesenchymal condensations appear in the autopod, are termed digital rays² and represent the primordia of tarsal/carpal and phalangeal elements. The first overt evidence of joint formation is the emergence of the interzone at each future joint location.^3-5 The interzone consists of closely-associated and flat-shaped mesenchymal cells and provides a clear demarcation between adjacent cartilaginous elements. The interzone is widely believed to be essential for joint formation,³ but its specific roles are not fully clear.⁶ To address this and related issues, genetic cell fate mapping studies were recently carried out in which Rosa R26R reporter mice⁷ were mated with mice expressing Cre-recombinase in incipient joints under the control of growth and differentiation factor-5 (Gdf5) regulatory sequences.⁸ The data indicated that Gdf5-expressing and interzone-associated progenitor cells are likely to give rise to several joint tissues, including articular cartilage and synovial lining.

Indian hedgehog (Ihh) is widely recognized as a critical regulator of long bone development and growth. Ihh is expressed in the pre-hypertrophic zone of growth plate^9-11 and regulates a number of processes. We found that Ihh promotes osteogenic cell differentiation, leading us to first propose that one key role for Ihh is to induce intramembranous bone collar formation around the developing diaphysis.^{10, 12} Other studies indicated that Ihh regulates the rates of chondrocyte proliferation and maturation in concert with peri-articular derived parathyroid hormone-related protein (PTHrP).^{11, 13} Studies on Ihh^−/− mouse embryos provided further and more conclusive support for such multiple Ihh roles in long bone development.^{14, 15} Strikingly, these studies led to the realization that Ihh also regulates synovial joint formation.¹⁴ The autopod region of Ihh^−/− mouse embryos was found to contain uninterrupted cartilaginous digital rays lacking obvious joints, when in fact joints were forming in the digits of wild type littermates. More moderate defects involving partial joint fusion or other defects were observed in more proximal mutant joints such as elbow and knee. Despite the remarkable nature of these observations and their potentially fundamental implications, it remains largely obscure how Ihh regulates joint formation and what specific roles it actually has. The present study was carried out to tackle these interesting and fundamental issues, using a combination of gene expression analyses and genetic cell fate map approaches in wild type and Ihh^−/− mouse embryos.

RESULTS

Anatomical and gene expression studies

Whole mount inspection of mouse embryo limbs was used to further analyze whether development of interphalangeal joints is more affected by lack of hedgehog signaling than that of more proximal limb joints. Staining with alcian blue did show that the cartilaginous anlagen of E18.5 Ihh^−/− mouse embryo toes were uninterrupted, lacked overt interzones and stained uniformly with alcian blue (Fig. 1B). Instead, skeletal elements in neighboring mutant ankle were distinct and separated from each other, as revealed by presence of alcian blue-negative mesenchymal tissue (Fig. 1B, arrows). Standard joint formation processes were well under way in wild type littermate toes and ankle that displayed obvious alcian blue-negative joint-associated tissues (Fig. 1A).

FIGURE 1. Anatomical examination of E18.5 wild type and Ihh^−/− mouse embryo hindlimbs.

Specimens were stained with alcian blue to distinguish cartilaginous from non-cartilaginous structures. (A) Wild type specimen in which the alcian blue-positive phalangeal and ankle skeletal elements are clearly interrupted at prescribed locations by alcian blue-negative joints. (B) Ihh^−/− specimen in which the digit primordia are uniformly stained and uninterrupted and lack obvious joints. Arrows point to alcian blue-negative tissue surrounding the mutant ankle elements.

To examine hedgehog signaling, we determined the gene expression patterns of hedgehog target and effector genes by in situ hybridization. In early E14.5 wild type joints, interzones were apparent between adjacent cartilaginous anlagen and displayed clear levels of Gli1 and Gli3 transcripts (Fig. 2A, C, E, arrows). Gli2 expression was not as prominent (Fig. 2D), expression of hedgehog receptor Patched-1 demarcated the cartilaginous anlagen (Fig. 2B) and expression of hedgehog signaling receptor Smoothened was widespread (Fig. 2F). At E17.5 when the cartilaginous anlagen become physically separated, expression of Gli1, Gli3 and Patched-1 was still appreciable and seemed to be more prominent on the concave side of the developing joins (Fig. 2G, H, I, K, arrows). In E17.5 Ihh^−/− limbs, however, Patched-1, Gli1 and Gli2 transcripts were essentially undetectable (Fig. 2M, N-P); interestingly, Gli3 and Smoothened transcripts were present, but largely occupied the mesenchymal tissues flanking the prospective and uninterrupted joint site (Fig. 2M, R, arrowheads).

FIGURE 2. Gene expression patterns of hedgehog target and effector genes.

(A-F) E14.5 wild type forelimb digit early joints in which nascent interzone cells clearly display Gli1 and Gli3 transcripts (arrow). Patched-1 expression characterizes contiguous cartilaginous tissue (B), Gli2 expression is low (D), and Smoothened (Smo) expression is widespread (F). (G-L) E17.5 wild type forelimb digit joints in which the cartilaginous elements are now well separated from each other. Note that transcripts for Patched-1, Gli1 and Gli3 seem to be slightly more abundant on the convex side of the developing joint (H, I, K, arrow). (M-R) E17.5 Ihh^−/− forelimb specimen from littermates in which joints are absent and prospective joint sites are uninterrupted. Note that Patched-1, Gli2 and Gli2 transcripts are essentially undetectable (M-P), while Gli3 and Smoothened transcripts are present but flank the prospective joint site (Q-R, arrowhead).

To corroborate these latter findings, we examined expression of Gdf5 and also Erg and tenascin-C. Erg is an ets transcription factor family member that we found to be associated with joint and articular cartilage formation and function,^16-20 and tenascin-C is a pericellular matrix protein abundant in developing and adult articular cartilage.^{16, 17, 21} In both E15.5 and E17.5 wild type digits, Gdf5, Erg and tenascin-C were all expressed in the interzones and associated articular layers (Fig. 3A-F). However, in Ihh^−/− digits, expression of the three genes was essentially restricted to the mesenchymal tissues flanking the prospective and uninterrupted joint sites (Fig. 3G-M, arrowheads), thus resembling the gene expression pattern of Gli3 and Smoothened.

FIGURE 3. Gene expression patterns of joint marker genes.

(A-F) E15.5 and E17.5 wild type forelimb digit joints in which interzone and associated cells strongly express Gdf5, Erg and tenascin-C (Tn). (G-M) E15.5 and E17.5 Ihh^−/− forelimb digits in which transcripts for Gdf5, Erg and tenascin-C (Tn) are present in tissue flanking the prospective joint sites (arrowheads).

As pointed out in Fig. 1B, mesenchymal tissue separates the various skeletal elements in Ihh^−/− ankles and wrists. To clarify its nature, we processed control and mutant specimens for in situ hybridization. The tissue present in control E15.5 wrists was conspicuous and well-organized (Fig. 4F) and strongly expressed Gdf5 and Patched-1 (Fig. 4G-H). Instead, the tissue present in the Ihh^−/− wrists was thin and inconspicuous (Fig. 4M) and expressed barely detectable levels of Patched-1 and Gdf5 (Fig. 4N-O).

FIGURE 4. Anatomical and histochemical analyses of Gdf5-Cre/ROSA R26R control and Ihh^−/− autopods.

(A-E) Control E15.5 autopods showing obvious presence of reporter activity (blue color) at prescribed interphalangeal joint locations (A). Histochemical analysis (B) shows that reporter-positive cells represent interzone cells, adjacent most-epiphyseal chondrocytes and joint lining cells, but the cells are absent in underlying growth plate and shaft. Joint cross sections (C-E) corresponding to location I and II in panel (A) reveal that the ventral half of the joints contain more numerous reporter-positive cells (arrowhead), large ventro-lateral mesenchymal cell masses (double arrowhead) and ongoing cavitation process (arrow). (F-H) Sections of control E15.5 wrists reveal peri-cartilaginous distribution of reporter-positive cells and well-contoured and reciprocally-shaped developing joints (F) that express Patched-1 and Gdf5 (G-H). (I-L) E15.5 Ihh^−/− autopods showing surprising presence of reporter activity at prospective joint site (arrow in I), but histochemical analysis reveal that the reporter-positive cells are located around the uninterrupted site (arrow in J). Cross sections corresponding to location III in panel (I) show some crowding of reporter-positive cells in the ventral aspect of the uninterrupted joint site (arrowhead in L) and absence of mesenchymal masses (double arrowhead in L) and cavitation (arrow in L). (M-O) Sections of E15.5 Ihh^−/− wrists showing that there are few reporter-positive cells and that the prospective joints are completely mis-shaped and lack reciprocal concave-convex contours (M) and express barely detectable levels of Patched-1 and Gdf5 (N-O).

Genetic cell fate map analysis

Next, we carried out genetic cell fate map analysis of joint progenitor cells in wild type and Ihh^−/− littermates by mating Gdf5-Cre^+/−/Rosa R26R^+/− double transgenic mice with heterozygous Ihh^+/− mice and analyzing their triple-transgenic progenies. Resulting embryonic limbs were processed for detection of reporter β-galactosidase activity by whole mount staining and histochemistry.²² As shown in Fig. 4A, reporter activity was clearly prominent and restricted to the developing interphalangeal and wrist joints in control Gdf5-Cre^+/−/R26R^+/−/Ihh^+/− heterozygous embryos. Identical patterns were seen in Ihh wild type (Gdf5-Cre^+/−/R26R^+/−/Ihh^+/+) mice (not shown). Histochemical analysis showed that reporter-positive cells were present in interzone, associated articulating layers and developing lining tissue but were absent in underlying shaft and growth plate cartilage (Fig. 4B) in agreement with previous findings.⁸ A similar distribution of reporter-positive cells were seen in the wrists (Fig. 4F). When we examined mutant Gdf5-Cre^+/−/R26R^+/−/Ihh^−/− specimens by whole mount, we were surprised to observe a band of reporter-positive cells located at the prospective location of interphalangeal joints (Fig. 4I, arrow), a finding seemingly to indicate that digit joints were actually forming in the Ihh^−/− embryos. However, histochemical analysis revealed that the reporter positive cells were mostly located outside and around the prospective joint site and were largely absent from the uninterrupted cartilaginous tissue (Fig. 4J, arrow). A moderate number of reporter-positive cells characterized the mutant wrist joints (Fig. 4M).

To more clearly understand the topography of distribution of reporter-positive cells, E15.5 control and mutant triple-transgenic autopods were cross-sectioned that the dorso-ventral distribution of the cells could be appreciated and analyzed. Interestingly, we found that the reporter-positive cells were not distributed uniformly along such dorso-ventral plane, but appeared to be more numerous and concentrated in the ventral half in control joints (Fig. 4C-E, arrowheads). The ventral side displayed also conspicuous bilateral masses of mesenchymal cells (Fig. 4E, double arrowhead) and, interestingly, the cavitation process was well underway ventrally but not dorsally (Fig. 4E, arrow). The primordia of extensor (et) and flexor (ft) tendons were clearly visible at their characteristic dorsal and ventral locations (Fig. 4C-E). When we examined Ihh^−/− prospective digit joint sites, we found that the reporter-positive cells also appeared to be slightly more numerous on the ventral aspect of the uninterrupted site (Fig. 4K-L). However, there was no overt sign of cavitation or presence of ventral-lateral mesenchymal masses (Fig. 4K-L, arrow and double arrowhead, respectively

DISCUSSION

Since it was discovered about 10 years ago and found to be expressed by pre-hypertrophic chondrocytes in the growth plate, Ihh has received a great deal of well-deserved attention. It can safely be said that Ihh is a true master regulator of limb skeletogenesis. Its expression in the pre-hypertrophic zone places it in a developmentally strategic position. We theorized long ago that the pre-hypertrophic zone is critical for the overall orchestration, coordination and temporo-spatial unfolding of the chondrocyte maturation program and long bone anlaga development.²³ Ihh likely represents a key element in such pre-hypertrophic chondrocytes' function, and it is now well established and recognized that Ihh regulates central processes and functions in developing long bones that include intramembranous bone collar formation, chondrocyte proliferation and maturation rates, expression of PTHrP, and endochondral ossification.^{24, 25}

The data in our present study now provide further insights into the fact that Ihh is also important for synovial joint formation and that joint initiation and morphogenesis are disrupted in Ihh^−/− limbs. We find that wild type interzones and early joints express Gli1 and Gli3, and Patched-1 expression characterizes the contiguous cartilaginous tissue in line with previous studies.^26-28 Given that these genes are mediators of hedgehog signaling, the findings indicate that hedgehog signaling is locally involved in joint formation. Indeed, Gli1 and Patched-1 are not expressed at appreciable levels in mutant Ihh^−/− autopods, and this is associated with, and possibly causally linked to, absence or defects in joint formation. It remains to be clarified whether Ihh action in joint formation is direct or indirect, the former implying that Ihh itself diffuses and reaches the joint-forming sites. Ihh, and hedgehog proteins in general, are known to be able to travel from their site of synthesis to far-away targets.^{29, 30} This task is rather demanding and difficult since hedgehog proteins contain C-terminal and N-terminal-bound cholesterol and palmitoylate tails and tend to stay close to their synthetic site.³¹ One diffusion and traveling strategy characterizing hedgehog proteins is the possible formation of micelles by which the hydrophobic tails would be buried on the inside and the more hydrophilic protein moiety would face the extracellular fluids.³² An additional strategy is likely provided by the fact that hedgehog proteins contain a heparin-binding domain, and interaction with heparan sulfate proteoglycans modulates their diffusion.³³ Previous work from our group and others has shown that growth plate chondrocytes express a number of heparan sulfate proteoglycans. Syndecan-3 and –4 are particularly evident in the proliferative zone^{34, 35} and syndecan-2 characterizes the perichondrium.³⁶ Given their binding characteristics and location, these macromolecules could attract Ihh and regulate its diffusion along specific spatial directions. Indeed, we and others have shown that Ihh diffuses away from its site of synthesis and is present in the proliferative zone and surrounding perichondrial tissue^{29, 30} and that defects in heparan sulfate proteoglycan synthesis or expression are accompanied by aberrant distribution of Ihh.^{37, 38} It is thus possible that mechanisms such as these could allow Ihh to influence joint development.

Our cell fate tracking data using Gdf5-Cre/Rosa R26R mice now reveal that reporter-positive cells are present at prospective joint sites in mutant Ihh^−/− digits and are distributed outside of the uninterrupted cartilaginous tissue. Gene expression analysis shows that the cells express joint marker genes including Erg, tenascin-C and Gdf5. Thus, these cells exhibit interzone-like characteristics and may represent interzone precursors or full-fledged interzone cells, implying that the cells become developmentally determined and patterned even in the absence of overt joint formation. Why then would the cells be largely located around the periphery of the prospective mutant joint site? One possibility is that in the absence of Ihh signaling, the cells may not be able to participate in interzone formation because chondrocytes occupying the site do not undergo de-differentiation and maintain the skeletal anlaga uninterrupted. If so, this would suggest that interzone formation may normally involve participation of peri-joint precursor cells migrating into the joint site under the auspices of hedgehog signaling, a possibility also raised in previous studies.^{39, 40}

Whatever the explanation, it is noteworthy that our data provide novel and unexpected insights into interzone and joint organization and functioning. We show that reporter-positive cells are not uniformly distributed across the dorso-ventral axis of developing control digit joints and display a gradient-like distribution, with more numerous cells present in the ventral half. Our data also show that the cavitation process appears to follow a similar ventral-to-dorsal pattern and is evident in the ventral side already at E15.5 but unappreciable on the dorsal side. Interestingly, there appears to be some ventral crowding of reporter-positive cells around prospective (uninterrupted) Ihh^−/− digit joint sites, suggesting that such distribution is independent of hedgehog signaling and overt joint formation. Several previous studies have focused on the important question of how the dorsal-ventral axis of limb development and symmetry is regulated, and factors such as Wnt7a and Lmx1 proteins are critical players.^41-43 Thus, it is possible that interzone precursors and associated cells are patterned by similar mechanisms and attain the differential distribution revealed by our data. Why should the cells be differentially distributed along the ventral-to-dorsal axis? One obvious possibility is that this asymmetric distribution is linked to, and needed for, digit joint formation. The digits can only bend in one direction (toward the ventral side) and thus their joints must develop accordingly to permit and direct such uni-directional movement. Presence of more numerous reporter-positive cells in the ventral half, presence of conspicuous ventro-lateral cell masses, and the ventral-to-dorsal direction of the cavitation process may all be needed for appropriate formation, organization and functioning of digit joints.

As previously realized and reiterated here, developing limb joints are differentially sensitive to the absence of Ihh. Interphalangeal joints fail to form in Ihh^−/− embryos and the intervening cartilaginous tissue remains uninterrupted. Instead, the mutant wrist and ankle sites contain cartilaginous elements separated by mesenchymal tissue (Fig. 1) and, as shown previously, elbow and knee sites contain separated or partially separated cartilaginous elements.¹⁴ It is been known for a while that though limb synovial joints are comparable biomechanical instruments, their development and organization are responsive to, and directed by, somewhat distinct mechanisms. For instance, major defects are seen in interphalangeal joints in mice lacking Gdf5, whereas the more proximal joints are minimally affected.⁴⁴ Proximal joints are significantly affected only in double Gdf5/Gdf6-null mice,⁴⁵ strongly indicating that Gdf5 and Gdf6 have redundant function in development of proximal joints, whereas development of interphalangeal joints largely relies on Gdf5. It is possible then that the different limb joints have distinct compensatory mechanisms or redundant pathways that would render them differentially sensitive to Ihh deficiency. Clearly, much more work is needed to sort out this important issue. In any case, it is interesting to note that even though they are not fully ablated by Ihh absence, the developing mutant wrist (and ankle) joints are profoundly mis-shaped and disorganized. The reciprocal concave-convex contours that characterize normal wrist elements and sculpt the two sides into functional and inter-locking instruments are absent in the mutants and both sides display a convex shape; in addition, the intervening mesenchymal tissue is poorly organized and thin (see Fig. 4M). Thus, Ihh signaling appears to be essential for joint morphogenesis, and this suggestion is reiterated by our observation that the opposing sides of wild type digit joints display small but appreciable differences in gene expression of hedgehog target and receptor genes (see Fig. 2G-L).

A major future task is to clarify how articular cartilage, ligaments, synovial lining and other differentiated joint tissues form during joint development and to what extent they do or do not derive from interzone cells. Based on matrillin-1 gene expression patterns and genetic tracking, it has been proposed recently that articular chondrocytes do not derive from interzone cells but are descendants of the original mesenchymal cell condensations.⁴⁶ Interzone cells would instead give rise to the other joint tissues such as synovial lining. Our Gdf5-Cre/Rosa R26R-based cell fate studies do not allow us to confirm or refute such possibility, given that both interzone cells and most-epiphyseal early chondrocytes express Gdf5 and would become reporter-positive. In other studies, we have shown that Erg is likely to have an important role in establishing the stable permanent status of articular chondrocytes.²⁰ Important roles are also found to be played by TGFβ signaling mechanisms.^{47, 48} These and related studies emphasize the fact that joint formation is very complex and requires a multitude of mechanisms regulating: joint site patterning and determination; interzone formation and joint morphogenesis; and formation of differentiated joint tissues. This multitude of mechanisms and processes needs to be coordinated and orchestrated spatio-temporally such that appropriate joints form at each specific location along the longitudinal axis of the limb. The challenge now is to figure out exactly how this feat is accomplished and whether disturbances in these fundamental mechanisms underlie congenital joint defects.