RESEARCH ARTICLE

Precise spatial restriction of BMP signaling is essential for articularcartilage differentiationAyan Ray1,*, Pratik Narendra Pratap Singh1,*, Michael L. Sohaskey2, Richard M. Harland2 andAmitabha Bandyopadhyay1,‡

ABSTRACTThe articular cartilage, which lines the joints of the limb skeleton, isdistinct from the adjoining transient cartilage, and yet, it differentiatesas a unique population within a contiguous cartilage element.Current literature suggests that articular cartilage and transientcartilage originate from different cell populations. Using acombination of lineage tracing and pulse-chase of activelyproliferating chondrocytes, we here demonstrate that, similar totransient cartilage, embryonic articular cartilage cells also originatefrom the proliferating chondrocytes situated near the distal ends ofskeletal anlagen.We show that nascent cartilage cells are capable ofdifferentiating as articular or transient cartilage, depending onexposure to Wnt or BMP signaling, respectively. The spatialorganization of the articular cartilage results from a band of Nog-expressing cells, which insulates these proliferating chondrocytesfrom BMP signaling and allows them to differentiate as articularcartilage under the influence of Wnt signaling emanating from theinterzone. Through experiments conducted in both chick andmouseembryos we have developed a model explaining simultaneousgrowth and differentiation of transient and articular cartilage injuxtaposed domains.

KEY WORDS: Articular cartilage, BMP, Nog, Wnt, Proliferatingchondrocytes

INTRODUCTIONLong bones of vertebrate limbs are formed by endochondralossification. In this process, a cartilage scaffold pre-figures thedeveloping skeleton (Akiyama et al., 2005). Chondrocytes at thecenter of this cartilage undergo hypertrophic differentiation and arereferred to as transient cartilage. Eventually, bone forms withinthe domain of hypertrophic differentiation. However, a smallpopulation of cartilage cells on either side, at the site ofsegmentation, resists this change. This remnant cartilage, whichlines the ends of skeletal elements, is referred to as articular cartilageor permanent cartilage (Karsenty and Wagner, 2002; Archer et al.,2003; Kronenberg, 2003; Pacifici et al., 2005).Current literature suggests that a Sox9-expressing, contiguous

cartilage condensation forms most of the skeletal elements of thedeveloping limb by branching and segmentation (Shubin andAlberch, 1986; Bi et al., 1999). The first sign of segmentationbecomes apparent with the formation of the interzone, characterized

by a region of densely packed flattened cells. The cells of theinterzone turn off the expression of Col2a1, a marker of cartilagecells, while turning on expression of growth differentiation factor 5(Gdf5), Wnt9a and autotaxin (Atx; Enpp2 – Mouse GenomeInformatics Database) (Hartmann and Tabin, 2001; Karsenty andWagner, 2002; Pacifici et al., 2005).

It is currently not well understood how two spatially andmolecularly distinct cell populations are formed within a uniformcell population at the time of interzone induction. However, theinterzone and the Wnt/BMP signaling pathways play importantroles in this process. BMP signaling has been particularly wellstudied in the context of joint development. It has been reportedthat any molecular manipulation leading to ectopic activation ofBMP signaling through (1) overexpression of BMP ligands orGdf5 in developing chick or mouse cartilage (Duprez et al., 1996;Tsumaki et al., 1999); (2) misexpression of constitutively activeBMP receptor in developing chick cartilage (Zou et al., 1997); or(3) knockout of Nog, a BMP signaling inhibitor (Brunet et al.,1998), results in transient cartilage differentiation throughout thecartilage anlagen, including in the cells of the presumptiveinterzone.

Nevertheless,Wnt ligands secreted from the interzone are cruciallyrequired for embryonic articular cartilage development (Hartmannand Tabin, 2001; Guo et al., 2004; Später et al., 2006a, b). Gain-of-function experiments with BMP and Wnt signaling indicate that thecells of the nascent cartilage anlagen are competent to differentiateinto transient or articular cartilage in response to one signal or theother. Therefore, the central issue that remains to be solved in thecontext of cartilage development is how two distinct populations,namely transient and articular cartilage cells, are formed adjacent toeach other from one nascent population of cartilage cells uponinterzone induction.

Proliferating chondrocytes near the ends of developing skeletalanlagen contribute to longitudinal growth of transient cartilage(Colnot, 2005); here, we show that these cells also contribute tothe appositional growth of articular cartilage. Further, wedemonstrate that a band of noggin-expressing cells insulates aregion of these proliferative cells from the influence of BMPsignaling, and allows them to differentiate as articular cartilageupon exposure to Wnt signaling emanating from the interzone.Thus, we propose a model that explains the concomitantlongitudinal growth of transient cartilage and the appositionalgrowth of articular cartilage.

RESULTSIncrease in cell number in the interzone and articularcartilage during early embryonic developmentGrowth of chicken articular cartilage during early embryonicdevelopment has not been systematically characterized before. Wefirst investigated whether the growth in volume is associated with aReceived 4 April 2014; Accepted 27 January 2015

1Department of Biological Sciences and Bioengineering, Indian Institute ofTechnology, Kanpur, U.P. 208016, India. 2Department of Molecular and CellBiology, University of California, Berkeley, CA 94720, USA.*These authors contributed equally to this work

‡Author for correspondence (abandopa@iitk.ac.in)

1169

© 2015. Published by The Company of Biologists Ltd | Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

concomitant increase in cell number in this tissue. Throughout thisarticle we use the term ‘articular cartilage’ only when cavitation ofthe synovial joint is histologically obvious. For developmentalstages prior to cavitation we use the term ‘interzone’. We measuredthe increase in cell number between successive stages in theinterzone/articular cartilage, as well as transient cartilage of themetatarsophalangeal (MTP) joint of hindlimb digit 3, starting fromHamburger–Hamilton (HH) stage 28, when the interzone firstbecomes apparent, up to HH38, the stage by which cavitation hasalready taken place at this joint. We observed a nearly 12-foldincrease in cell number in the developing articular cartilage duringthis period (Table 1; supplementary material Fig. S1A,E). It shouldalso be noted that there was a steady increase in the number ofcells per unit area in the interzone/articular cartilage region(supplementary material Fig. S1D).Our data (Table 1) suggest a significant increase in the number of

cells in the interzone/articular cartilage during early chickdevelopment. However, the developing interzone has beenreported to be mitotically deficient (Mankin, 1962). Therefore, weinvestigated whether this increase in cell number of articularcartilage cells results from proliferation of interzone cells.

Interzone is proliferation deficientWe used three independent criteria to identify proliferating cells indeveloping chicken embryonic cartilage: (1) expression of PCNAmRNA (a marker for S-phase cells) (Köhler et al., 2005); (2)detection of phosphorylated histone 3 (pH3) immunoreactivity; and(3) incorporation of 5-ethynyl-2′-deoxyuridine (EdU), a thymidineanalogue. Using any of these methods, we could rarely detectproliferating cells in the interzone of developing interphalangealjoints and MTP joints in any of the stages examined (Fig. 1A-I;supplementary material Fig. S2C). This suggests that the interzonecells proliferate at a very slow rate, if at all, which cannot explain theobserved rate of increase in cell number. However, we observedmany proliferating cells on either side of the interzone (Fig. 1A-I;supplementary material Fig. S2C). This zone of proliferating cells issituated at the distal ends of the developing skeletal anlagen and willbe referred to as ‘distal proliferative zone’ or DPZ throughout themanuscript. Likewise, in mouse embryos the interzone cells also didnot incorporate EdU, whereas the DPZ cells did (Fig. 1J-L). Furthermolecular characterization of the DPZ is described below. Incontrast to our observations, two previous reports (Kahn et al., 2009;Gunnell et al., 2010) showed BrdU-labeled cells in the interzone ofmouse forelimb humeroulnar or humeroradial joints at 14.5 dayspost coitum (dpc). This discrepancy might be due to the differencein joints analyzed.

Proliferative chondrocytes of the DPZ contribute to thegrowth of developing interzone and articular cartilageAs the cells of the interzone are non-proliferative, we hypothesizedthat cells from outside the interzone must be contributing to thegrowth of this tissue.

A previous study found that articular cartilage cells aredescendants of Gdf5-expressing cells (Koyama et al., 2008). Toinvestigate whether there are Gdf5 mRNA-expressing cells in theinterzone that are proliferative, we co-labeled BrdU+ proliferatingandGdf5mRNA-expressing cells in developing digits at HH31 inchick and 14.5 dpc in mice (Fig. 2). We observed that, in a distalphalangeal joint, some of the Gdf5-expressing cells flanking theinterzone are also BrdU immunoreactive (Fig. 2B,E). However,within proximal joints of the same digit, cells expressingGdf5 areconfined within the interzone and are not BrdU immunoreactive(Fig. 2C,F). These observations suggest that flanking cells of thenewly formed interzone are proliferative and lie within thedomain of Gdf5 expression, whereas, in more mature interzone/articular cartilage, Gdf5-expressing cells are primarily non-proliferative. Therefore, the further interstitial growth ofmatured interzone cannot be accounted for by proliferation ofcells within the interzone.

To investigate whether flanking cells are incorporated into theinterzone, we conducted a pulse-chase DNA-labeling experiment(Fu et al., 2012). Such pulse-chase experiments are not possiblewith developing chick embryos because the label is neither diluted(as is possible in cell culture experiments) nor can it be removedby the circulatory system (Hämmerle and Tejedor, 2002).Therefore, we used developing mouse embryos and pulse-labeled with EdU and/or BrdU. Interzones of mouse 13.5 and15.5 dpc harvested 2 h post EdU injection were largely devoid ofEdU+ cells, whereas proliferating cells were detected within theregions immediately flanking the interzone (Fig. 3B,D). Bycontrast, embryos harvested at 14.5 and 16.5 dpc, 24 h post EdUinjection, showed many EdU+ cells in the interzone and very fewEdU+ cells in the flanking regions (Fig. 3C,E). Col2a1 and Gdf5mRNA expression domains on sections adjacent to the ones usedin Fig. 3D,E are presented in supplementary material Fig. S3A-D.Intermediate chases conducted for up to 5 h post BrdU injectionfailed to detect any BrdU+ cells in the interzone (supplementarymaterial Fig. S3G-J).

Finally, to rule out the possibility that the EdU-labeled cellswhich appear in the interzone after a 24-h chase might be a resultof slow proliferation, we conducted a competition experiment.We established that, if a fivefold molar excess of unlabeled dTTPis injected along with EdU, no EdU+ cells could be detected inthe hindlimb of these animals, demonstrating that the unlabeleddTTP can successfully outcompete EdU for incorporation(Fig. 3G). On the other hand, when unlabeled dTTP is injected2 h post EdU injection, many EdU+ cells could still be detectedin the interzone 24 h post EdU injection (Fig. 3H), thusdemonstrating that cells which incorporated EdU within 2 h ofits injection, and before dTTP administration, are eventuallyincorporated in the interzone.

Furthermore, we injected pregnant females with EdU at15.5 dpc, followed by BrdU injection after 22 h, and harvested2 h post BrdU injection, i.e. 24 h post EdU injection (Fig. 3I-M).Most of the interzone cells were labeled with EdU (Fig. 3J,L). Onthe other hand, many chondrocytes in the DPZ, flanking theinterzone, were labeled with BrdU (Fig. 3J,K). This domain ismarked by Col2a1 mRNA expression (supplementary materialFig. S3F,F′). Taken together, these data suggest that proliferative

Table 1. Increase in volume and cell number in the interzone/articularcartilage of metatarsophalangeal joint of digit 3 in chick

Stage

Fold increase involume ofinterzone/articularcartilagebetweensuccessivestages

Fold increase incell number ininterzone/articularcartilagebetweensuccessivestages

Fold increase incell numberin interzone/articularcartilagecompared withHH28

HH28 – – –

HH31 2 1.67 1.67HH34 1.6 2.01 3.36HH36 1.5 2.56 8.61HH38 1.5 1.41 12.14

1170

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

chondrocytes in the DPZ are labeled with EdU at 15.5 dpc butbecome incorporated in the interzone 24 h later. However, cells inthe DPZ remain proliferative at 16.5 dpc (as they are also labeledwith BrdU), whereas interzone cells remain non-proliferative atboth 15.5 and 16.5 dpc.Our data suggest that the cells that contribute to the growth of

the interzone during embryonic development are the proliferativeDPZ cells located within the Col2a1 mRNA expression domain(supplementary material Fig. S3F′). However, pulse-chase labelingwith thymidine analogs cannot completely rule out the possibilitythat the labeled cells that are incorporated in the interzone are notcoming from some other nearby pool of proliferative cells. To testwhether the proliferative cells that are incorporated into the interzonehave a history of Col2a1 expression, we traced the lineage ofCol2a1-expressing cells in the developing cartilage (Fig. 3N).Col2a1mRNA-expressing cells were marked at 15.5 dpc inCol2a1-CreERT2;RosamT/mG embryos (Nakamura et al., 2006); at this stage,there is no expression of Col2a1 mRNA in the interzone (Fig. 3O).Nevertheless, marked cells were found in the newly cavitatedarticular cartilage at 18.5 dpc (Fig. 3P). Some of these GFP-expressing cells also have incorporated EdU (Fig. 3Q,R),demonstrating that at least some of the cells that contribute to thegrowth of interzone/embryonic articular cartilage come from theCol2a1-expressing, proliferating cells of the DPZ.

To test the possibility that the proliferative DPZ cells contributeto postnatal articular cartilage after incorporation into theinterzone, we conducted a long-term chase of EdU- or BrdU-labeled cells. We monitored the presence of EdU+ (P0, P3 and P7)or BrdU+ (P21 or P42) cells in the articular cartilage of the MTPjoint in postnatal stages of mouse development after threeconsecutive pulses of EdU or BrdU on 13.5, 14.5 and 15.5 dpc.We observed a number of EdU+ or BrdU+ cells in the articularcartilage of P0, P3, P7, P21 and P42 MTP joints of mousehindlimbs (Fig. 3S-W; supplementary material Fig. S3P,Q). Itshould be noted that the number of BrdU+ cells detected in thearticular cartilage is lower in P21 or P42 compared with the EdU+

cells at earlier time points. Whereas most of the BrdU-labeled cellsdetected in P21 or P42 articular cartilage are located in the deeperlayers, some of the EdU-labeled cells detected at P0, P3 and P7 arealso located in the superficial layers of articular cartilage andhave a flattened morphology (Fig. 3T,U; supplementary materialFig. S3Q).

Based on our observations (Figs 1-3), we define DPZ as the zoneof proliferative cells, which is mostly confined to the distal zone ofthe Col2a1 mRNA expression domain (supplementary materialFig. S3F′). Our observations suggest that cells of the DPZcontribute to the growth of both transient as well as articularcartilage.

Fig. 1. Chick and mouse interzone cells are proliferationdeficient. (A-D) mRNA in situ hybridization (ISH) of PCNA in chickhindlimb digit from HH31 (A) and HH36 (C). (B,D) Highermagnification views of the boxed regions in A and C, respectively.(E-I) EdU incorporation in chick digits after 2 h of EdU injection.(E) Digit 3 of HH36 chick hindlimb. (F-I) MTP joints of digit 3 atHH28 (F), HH31 (G), HH34 (H) and HH38 (I). (J) EdU incorporationin 16.5-dpc mouse hindlimb digits. (K,L) Higher magnificationviews of the boxed regions in J. Dashed lines represent themargins of the skeletal elements in the developing digits.Red asterisks denote interzone, green arrowheads indicatedistal proliferative zone (DPZ) and yellow arrowheads highlighthypertrophic region. MTP, metatarsophalangeal joint.Scale bars: 100 μm.

1171

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

Dynamic changes in gene expression during cartilagedifferentiationIn order to understand how both transient and articular cartilage candifferentiate from the same cell population within the DPZ, weexamined expression of markers for both transient cartilage(Col2a1) and interzone/articular cartilage (Atx and Gdf5)(Hartmann and Tabin, 2001; Ikeda et al., 2004; Hyde et al., 2008)during the course of early cartilage development.Col2a1 and Atx mRNAs are co-expressed in a broad domain,

spanning the interzone and the adjoining proliferative chondrocytesin the MTP joints of the chick hindlimb at early stages (HH28 andHH31) (Fig. 4A,B). There is a graded expression of Atx extendingfrom the interzone towards the center of the element, and anopposite graded expression of Col2a1, extending from the centertowards the interzone. At later stages (HH34 and HH38), theinterzone cells continue to express high levels of Atx but turn offCol2a1, whereas the opposite happens outside the interzone.Eventually, a few layers of cells emerge between these twodomains, in which both Col2a1 and Atx expression aredownregulated (Fig. 4A-D). Thus, it seems that there is an overlapof Atx and Col2a1 mRNA expression domains in the nascentcartilage cells, but this is resolved over time to two separate domainsin more mature proximal joints.It is possible that early chondrocytes express bothCol2a1 and Atx

and, as development progresses, the cells of the transient cartilagemaintain expression of Col2a1, whereas the cells of the interzonemaintain expression of Atx or Gdf5. To examine such a possibility,we took advantage of the temporal gradient of differentiation fromthe proximal to the distal tip of the digit, with proximal being at amore advanced stage of differentiation than the distal phalanx. Wecarried out fluorescent double mRNA in situ hybridization for Atxand Col2a1 on a HH36 chick hindlimb digit 3 (Fig. 4E-I). Manycells in the two distal phalanges co-express Atx and Col2a1(Fig. 4E-G), whereas, in the proximal phalanges, cells expressingAtx are distinct from those expressing Col2a1 (Fig. 4E,H,I).Furthermore, the cells flanking the proximal interzone havedownregulated both Atx and Col2a1 expression (compare theregions marked by white asterisks in Fig. 4F and G with those inFig. 4H and I, respectively).We have also examined the relative expression patterns ofCol2a1

andGdf5 in the digit 3 of chick hind limb. Our results with Gdf5 are

similar to those obtained with Atx, i.e. nascent cartilage cells expressCol2a1 and Gdf5 in an overlapping manner, whereas, in moredeveloped cartilage, transient cartilage cells express Col2a1 mRNAand interzone/articular cartilage cells expressGdf5mRNA (Fig. 4J-L;supplementary material Fig. S4A). We observed a similardynamic resolution of gene expression patterns in mice as inchick (Fig. 4M-O).

Dynamic expression of noggin and pSMAD1/5/8To understand the molecular basis of such dynamic changes in geneexpression, we surveyed mRNA expression of many signalingmolecules that are expressed in the articular cartilage between HH28and HH38 (supplementary material Table S1).We found the mRNAexpression of Nog, a secreted inhibitor of BMP signaling, to beparticularly informative. In agreement with earlier reports (Brunetet al., 1998; Francis-West et al., 1999), we found that, at earlierstages, Nog mRNA is expressed in a broad domain throughout thecartilage, other than in the interzone (HH28, supplementarymaterial Fig. S5A; and HH31, Fig. 5A,B). In more developedcartilage, Nog mRNA is expressed in a tight band of cells with adistinguishable boundary (Fig. 5C-G; supplementary material Fig.S5B) towards the interzone of an MTP joint. This results in a fewlayers of cells on either side of the interzone that are devoid of NogmRNA expression (Fig. 5C-G; supplementary material Fig. S5B).The distinct domain of Nog mRNA expression is clearly visiblearound a more mature joint, e.g. the tibiotarsal joint at HH36(Fig. 5E).

These observations prompted us to investigate the spatial domainsof active BMP signaling in the time window when Nog mRNAexpression undergoes these dynamic changes. Thus, we assayedBMP signaling using pSMAD1/5/8 indirect immunofluorescence. Inearly stages (HH28), we find that pSMAD1/5/8 activity is primarilyrestricted to the perichondrium and absent from the interzone(supplementary material Fig. S5C,D). Later (HH31, HH34, HH36),pSMAD1/5/8 activity is present in a precisely restricted manner in asubpopulation of cells in the epiphysis but is undetectable in theinterzone or the cells flanking it (Fig. 5H-J; supplementary materialFig. S5E-G). To directly compare noggin protein expression andthe domain of active BMP signaling we performed doubleimmunohistochemistry in the MTP joint of HH34 (supplementarymaterial Fig. S5H-K). We observed that, in some cells of the

Fig. 2. Gdf5 mRNA-expressing cells in a matureinterzone are proliferation deficient. (A) Panorama ofdeveloping chick HH31 hindlimb digits; pseudocolored ISHof cGdf5 mRNA (green) and incorporation of BrdU (red)overlapped using panels in supplementary materialFig. S2D. (B,C) Higher magnification views of the boxedregions in A. Similarly, D shows panorama of digits from14.5-dpc mouse embryos, generated using images insupplementary material Fig. S2E. (E,F) Highermagnification views of boxed regions in D. Purplearrowheads denote distal-most phalangeal interzone,white arrowheads indicate proximal matured interzones.Scale bars: 100 μm.

1172

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

Fig. 3. Proliferating cells from the DPZ contribute to the growth of embryonic articular cartilage. (A-E) EdU pulse-chase experiments were carried out inpregnant mice 13.5 and 15.5 dpc. (A) Schematic of the experimental design. (B,D) Mouse hindlimbs 2 h post EdU injection. (C,E) Mouse hindlimbs 24 h post EdUinjection. Red asterisks show EdU-incorporated cells of DPZ and white asterisks indicate DPZ cells devoid of EdU. White arrowheads denote interzone cellsdevoid of EdU and red arrowheads highlight EdU-incorporated cells in the interzone. For D,E, correspondingCol2a1 andGdf5 expression domains are presentedin supplementary material Fig. S3A-D. (F) Schematic showing the experimental design for EdU-dTTP competition in mouse embryos. (G) Simultaneous injectionof EdU and excess dTTP abolished any selective uptake of EdU. (H) Many EdU+ cells are detected in the interzone when excess dTTP is administered 2 h postEdU injection. White arrowhead indicates interzone. MT, metatarsus; PH, phalangeal element. (I) Schematic of EdU/BrdU double pulse-chase experiment.(J-M) J is merged image of K-M. Many cells in the interzone of MTP incorporated EdU (J,L, white arrowheads), but did not incorporate BrdU. However, cellsflanking the interzone displayed BrdU immunoreactivity (J,K, asterisks). (M) DAPI-stained section. For J-M, correspondingCol2a1 andGdf5 expression domainsare presented in supplementary material Fig. S3E,F. (N-R) Genetic lineage tracing usingCol2a1-CreERT;RosamT/mG. (N) Schematic for the experimental design.(O) At 15.5 dpc, Col2a1 is not expressed at the interzone of the MTP joint (black asterisk). (P,Q) Adjacent sections from 18.5 dpc Col2a1-CreERT2;RosamT/mG

MTP joint. (P) Col2a1 mRNA ISH (pseudocolored in red) and anti-GFP antibody (green). (Q) EdU (red) and anti-GFP antibody (green). Individual frames areprovided in supplementary material Fig. S3L-O. (R) Higher magnification view of the region marked in Q shows multiple EdU+ and GFP+ double-positive cells(arrowheads). (S) Schematic showing long term pulse-chase of thymidine analogs EdU or BrdU in mice. (T-W) EdU and BrdU incorporation (arrowheads) weredetected in theMTP joint articular cartilage at P0, P7, P21 and P42. The corresponding data for P3, P21 (higher magnification view) and P42 (higher magnificationview) are presented in supplementary material Fig. S3P-S. Dashed lines represent the margins of the skeletal elements in the developing digits.

1173

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

epiphysis, both Nog protein and pSMAD1/5/8 are detectable,whereas cells flanking the interzone on either side express onlyNog (supplementary material Fig. S5H-K).It is important to note that the region which shows

downregulation of BMP activity, flanking the interzone, is thesame region that upregulates Wnt signaling, as can be seen fromβ-catenin immunostaining (Fig. 5K).

Restricted domains of BMP and Wnt signaling are crucial forproper transient and articular cartilage differentiationOur expression analysis reveals that the cells of the DPZ are flankedby separate BMP and Wnt signaling domains (Fig. 6A). This led usto predict that ectopic activation of BMP or Wnt signaling in theDPZ would result in ectopic transient or articular cartilagedifferentiation of these cells, respectively.To activate ectopic BMP signaling, HH18 hindlimbs were

infected with viral particles expressing BMP4 (RCAS-BMP4) andconstitutively active BMPRIB (RCAS-ca-BMPRIB).We observed that, upon overexpression of the BMP ligand using

RCAS-BMP4 (Duprez et al., 1996),Col2a1mRNA is expressed ina broad domain encompassing the domain of presumptiveinterzone (Fig. 6B). We further observed that the cells thatectopically expressed Col2a1 mRNA retained neither β-cateninimmunoreactivity (compare Fig. 6B and D, white asterisks) norendogenous Gdf5 mRNA expression (Fig. 6C, white asterisk). Bycontrast, the interzone cells that lacked ectopic Col2a1 mRNAexpression retained β-catenin immunoreactivity (Fig. 6B,D) as

well as expression of Gdf5 mRNA (Fig. 6C). Thus, our datasuggest that active BMP signaling suppresses Wnt signaling indeveloping cartilage cells.

As reported earlier (Zou et al., 1997), we observed loss ofsegmentation of skeletal elements upon ectopic expression ofcaBMPRIB. Furthermore, virally infected patches (as determined by3C2 immunoreactivity, Fig. 6E) of HH36 digit cartilage ectopicallyturned on Col2a1 mRNA expression in the cells of both thepresumptive interzone and the adjoining region (Fig. 6F). Bycontrast, the cells of the interzone that were not infected did notexpress Col2a1, and instead continued to express Gdf5 (Fig. 6G)and Atx (supplementary material Fig. S5L). It is also apparent thatthe infected cells of the interzone not only ectopically turned onCol2a1 mRNA expression but also have altered morphology toresemble the rounded cells of the epiphysis, as opposed to theflattened cells of the interzone (compare Fig. 6H and I).

In the viral misexpression experiments, it is possible that the limbmesenchymal cells experienced ectopic activation of BMP signalingeven before the formation of interzone or establishment of the cellpopulations flanking the interzone. We therefore used an organculture system (see Materials and Methods) to test the effect oflater administration of BMP or Wnt, after the interzone has beenestablished. Following administration of recombinant BMP4 protein,ectopic pSMAD1/5/8 immunoreactivity (Fig. 6J) and ectopicCol2a1 mRNA expression (Fig. 6L) were detected in the cells ofboth the interzone and the adjoining region, in contrast to the sham-injected joint (Fig. 6K,M). Similarly, as opposed to a sham-injected

Fig. 4. Double-labeling of digit skeletal elements withmarkers of articular cartilage (Atx and Gdf5) andtransient cartilage (Col2a1) show that nascentchondrocytes co-express markers of both these celltypes in chick and mice. Double-fluorescent mRNAin situ hybridization of chickenAtx (green) andCol2a1 (red)in the MTP joints at HH28 (A), HH31 (B), HH34 (C) andHH38 (D). (E) Panorama of digit 3 of HH36 chick hindlimb,following double-fluorescent mRNA in situ hybridization forCol2a1 (red, top panel), Atx (green, second panel from thetop), DAPI (blue, third panel from the top) and merge(bottom panel). (F-I) Magnified views of the correspondingregions labeled in E. (J-O) Relative domains of Col2a1(red) and Gdf5 (green) mRNA expression in chick HH36hindlimb digit (J); in mouse 16.5 dpc hindlimb digit (M) andin 16.5 dpc metatarsus (N). K and L are highermagnification views of the boxed regions in J; O is thehigher magnification view of the boxed region in N. Col2a1andGdf5mRNA ISH (using chromogenic detection, shownin supplementary material Fig. S4) from adjacent sectionswere pseudocolored and superposed to generate panelsJ-O. White arrowheads indicate domains of articular andtransient cartilage marker co-expression. White asterisksshow regions where markers of articular and transientcartilage are downregulated in matured joints comparedwith nascent interzones. Scale bars: 100 μm.

1174

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

control, ectopic application of exogenous mWnt3a induced Wntsignaling in a broader population of cells around the interzone, asdemonstrated by β-catenin immunostaining (compare Fig. 6N andO). Notably, Atx mRNA expression could also be detected in abroader domain in theWnt protein-injected interzone compared witha sham-injected control interzone (compare Fig. 6P and Q). It isimportant to note that, upon ectopic BMP activation at the joint site,the interzone cells start to express markers of transient cartilage butdo not lose their characteristic flattened shape. Taken together, ourdata demonstrate that ectopic exposure to Wnt induces articularcartilage markers, and that BMP signaling induces transient cartilagemarkers in the chondrocytes of the developing cartilage anlagen.

Reduction ofNogactivity leads toectopic transient cartilagedifferentiationEarlier, it has been demonstrated that inactivation of Nog in miceresulted in ectopic transient cartilage differentiation in the cells ofthe putative interzone (Brunet et al., 1998). We examined whetherthe band of noggin-expressing cells, uncovered in this study,protects the DPZ cells from BMP signaling. For this purpose, wereduced noggin activity by injecting anti-noggin neutralizingantibody at multiple joint sites of a HH34 chick hindlimb digitand cultured for 24 h. Anti-noggin antibody was spread around theMTP joint of digit 3 (Fig. 7A). BMP signaling, measured bypSMAD1/5/8 immunoreactivity, was found in the interzone and thecells flanking the interzone of the anti-noggin antibody-injecteddigit (Fig. 7B), whereas the control antibody-injected jointexhibited no pSMAD1/5/8 immunoreactivity in the same location(Fig. 7E). Moreover, as compared with the control antibody-injected digit, in the anti-noggin antibody-injected digit ectopicCol2a1 mRNA expression was detected in the interzone as well asin the flanking cells (Fig. 7C,F)

Furthermore, we knocked out Nog in the transient cartilage cellsof mice, using a previously described transgenicCol2a1-CRE driverline and aNog-floxed allele line (Terpstra et al., 2003; Stafford et al.,2011) (supplementary material Fig. S5N,O).

The cells of the interzone of a 15.5 dpc interphalangeal joint andthose flanking it on either side in a control (Col2a1Cre;Nogfl/+) mousehindlimb digit are devoid of pSMAD1/5/8 activity (Fig. 7I). Onthe other hand, the cells in the analogous region in a mutant (Col2a1-CRE;Nogfl/fl) are pSMAD1/5/8 immunoreactive (Fig. 7L), indicatingthat these cells are exposed to active BMP signaling due to the removalof Nog from the zone of transient cartilage. This ectopic activation ofBMP signaling leads to ectopic expression ofCol2a1mRNA (comparepanels Fig. 7H and K) and downregulation of Gdf5 mRNA in theinterzone (compare panels Fig. 7G and J). These molecular changes ingene expression by abolition of Nog from the domain of transientcartilage region led to loss of joint formation. This became apparenthistologically by Alcian Blue staining of section of an autopod at15.5 dpc (compareFig. 7MandN).At a later developmental time point,such as 18.5 dpc, we observed complete absence of joint structures inthe digits and the metatarsus of the mutant hindlimb (compare Fig. 7Oand P). These data suggest that the domain of noggin expression isnecessary to insulate the cells around the interzone fromBMPsignaling,and in turn allow the development of embryonic articular cartilage.

DISCUSSIONArticular cartilage cells are derived from proliferativechondrocytes present at the DPZ of embryonic skeletalelementsBased on lineage tracing of Gdf5-expressing cells, it had beenconcluded previously that articular cartilage originates from aGdf5-expressing population (Koyama et al., 2008). The observationsmade in that report have been subsequently interpreted as interzone

Fig. 5. Domains of Nog mRNA expression and BMPsignaling activity in developing chick cartilage.(A-G) Nog mRNA ISH at HH31 (A,B), HH34 (C,D), HH36(E,G) in chicken hindlimb. A,C and F are hindlimb digits.B,D and G are higher magnification views of the boxedregions in A,C and F, respectively. (E) Panoramic view ofNogmRNA ISH of HH36 tibiotarsal joint. Black arrowheadsindicate extent of NogmRNA expression domains towardsthe interzone/articular cartilage. Asterisks indicate regiondevoid of Nog mRNA expression. (H-J) pSMAD 1/5/8immunoreactivity (red) in HH34 tibiotarsal joint (H,I) andHH36 MTP joint (J). (I) Higher magnification view of theboxed region in H. Asterisks in H-J denote domain devoidof pSMAD 1/5/8 immunoreactivity. (K) Nuclear β-cateninimmunoreactivity (red) in the MTP joint of HH36 chickhindlimb. Asterisks mark the active Wnt/β-catenin domain.Dashed lines represent the margins of the skeletalelements in the developing digits. Scale bars: 100 μm.

1175

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

cells giving rise to articular cartilage (Kahn et al., 2009; Candela et al.,2014). By contrast, here we present data strongly indicating that theinterzone cells are largely proliferation deficient and hence cannotsupport the growth of articular cartilage; instead, proliferativechondrocytes of the DPZ become part of the articular cartilage.Thus, our data demonstrate for the first time that these cells not onlycontribute to transient cartilage growth, but also contribute toembryonic articular cartilage growth. Our data also provide themechanistic framework that allows contrasting differentiation events,e.g. embryonic transient and articular cartilage differentiation, to takeplace in juxtaposed cell populations.To support the conclusion that embryonic articular cartilage must

be derived from the DPZ, we first showed that, in developing chickembryos, there is a significant increase in cell number in the interzone/articular cartilage (Table 1); yet, the interzone cells of either chick ormouse do not proliferate significantly during the time window ofarticular cartilage specification (Fig. 1). These results suggest thatcells from other source(s) contribute to the growth of interzone/embryonic articular cartilage. EdU/BrdU pulse-chase of proliferatingchondrocytes in conjunction with lineage tracing of Col2a1 mRNA-expressing cells, presented in this study, show that cells of the DPZcontribute to the embryonic articular cartilage. Although we cannot

rule out the possibility that other proliferating cell populations do notcontribute to the growth of interzone, DPZ certainly contributes to thegrowth of embryonal articular cartilage. It should be noted that, at15.5 dpc, theDPZcells expressCol2a1mRNA,but at 18.5, these cellsare detected in the interzone, where cells expressGdf5mRNAbut notCol2a1. Thus, the DPZ cells must undergo a transition before beingincorporated in the interzone. In fact, we have identified such a zonebetween the DPZ and the interzone, where the cells express neitherCol2a1 nor Gdf5 mRNA (Fig. 4). Furthermore, we could detectBrdU+ cells in the articular cartilage of six-week-old mice that werelabeled at 13.5-15.5 dpc (Fig. 3), suggesting that the DPZ cellseventually become part of adult articular cartilage. Some of the BrdU-labeled cells at P21 or P42 are located in the superficial layer ofarticular cartilage having a flattened morphology (Fig. 3V,W). Wesuggest that these cells act as reservoir that might support postnatalarticular cartilage growth (Dowthwaite et al., 2004). Our data is inagreement with recently published results (Candela et al., 2014).

Model for simultaneous differentiation of transient andembryonic articular cartilageEctopic activation of BMP signaling in developing cartilage resultedin the entire limb cartilage differentiating as transient cartilage, with

Fig. 6. Ectopic activation of BMP or Wnt signaling affects domain of articular cartilage differentiation. (A) Schematic summarizing different proliferationand expression domains identified. Cells of DPZ (blue), flanked by domains of BMP (orange) andWnt (green) signaling, contribute to both transient (red arrows)and articular (blue arrows) cartilage. A domain of noggin expression (gray) lies between the BMP signaling domain and DPZ. (B-D) Alternate sectionsthrough putative phalangeal joint of chick hindlimb infected with RCAS-BMP4 at HH36. (B) Col2a1 mRNA ISH, (C) Gdf5 mRNA ISH (black arrowhead) and(D) β-catenin immunoreactivity. White asterisks denote ectopic Col2a1 mRNA expression domain devoid of Gdf5 and β-catenin immunoreactivity; whitearrowheads show absence of ectopic Col2a1mRNA expression but β-catenin immunoreactivity is retained. (E-G) Alternate sections through putative phalangealjoint of chick hindlimb infected with RCAS-caBMPRIB at HH36. (E) Retroviral infection detected by 3C2 immunohistochemistry (white arrowhead denotes theinfected cells in the interzone). (F) Col2a1 mRNA ISH (black arrowhead indicates putative interzone cells ectopically expressing Col2a1 mRNA) and (G) Gdf5mRNA ISH (black arrowhead denotes uninfected interzone cells expressing Gdf5 mRNA). (H,I) Higher magnification views of the boxed regions in F and G,respectively. Black asterisks mark cells with flattened morphology, white asterisks mark cells with rounded morphology. (J-Q) Interzone and the cells flanking itdifferentiate along the transient or articular cartilage fate under BMPorWnt signaling influence, respectively. pSMAD1/5/8 indirect immunofluorescence (J,K) andCol2a1 mRNA ISH (L,M) in control (K,M) and in BMP4-administered (J,L) HH34 chick MTP joint of hindlimb after 24 h of organ culture. β-catenin indirectimmunofluorescence (N,O) and AtxmRNA ISH (P,Q) in control (O,Q) andWnt3a-administered (N,P) HH34 chick MTP joint of hindlimb after 24 h of organ culture.Dashed lines represent the margins of the skeletal elements in the developing digits.

1176

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

no morphologically distinct articular joint visible (this study andDuprez et al., 1996; Zou et al., 1997; Brunet et al., 1998). We havealso demonstrated that ectopic activation of BMP signalingsuppresses Wnt signaling in the cells flanking the interzone.Therefore, for simultaneous differentiation of transient and articularcartilage in juxtaposed populations, Wnt-responsive cells must beinsulated from BMP influence. We have demonstrated that a band ofNog-expressing cells (Fig. 5) serves this purpose. Thus, the absence ofjoint structures upon ectopic activation of BMP signaling should notbe interpreted as fusion of cartilage elements (Duprez et al., 1996;Zou et al., 1997; Guo et al., 2004; Rountree et al., 2004). Rather, itshould be interpreted as failure of articular cartilage formation andconsequent failure to segment the cartilage anlagen. This is in keepingwith the postulation of Brunet et al. (1998). We have demonstratedthat ectopic activation of BMP or of Wnt even at late stage (HH34)(i.e. after specification/differentiation of interzone) turns on ectopicexpression of transient or of articular cartilage markers, respectively(Fig. 6J-Q). Therefore, the domains of BMP and Wnt signaling mustbe kept spatially distinct to allow for simultaneous differentiation ofembryonic articular and transient cartilage.Based on our experimental data, we propose a model integrating

concurrent longitudinal growth of transient cartilage andappositional growth of articular cartilage from the proliferating

cells of the embryonic cartilage anlagen (Fig. 8). The proposedmechanism is likely to be conserved among vertebrates, as amajority of the postulates have been validated both in mouse and inchick.

MATERIALS AND METHODSTissueMouse experiments were conducted as per protocol approved by theInstitute Animal Ethics Committee (registration number 810/03/ac/CPCSEA, dated 15 October 2003). A Nog loss-of-function mutant(Col2a1-CRE;Nogfl/fl) was generated by mating Col2a1-CRE transgenicmice (Terpstra et al., 2003) with conditional Nogfl/fl mice (Stafford et al.,2011). Col2a1-CreERT2;RosamT/mG was generated by crossing Col2a1-CreERT2 (Nakamura et al., 2006) and RosamT/mG (Muzumdar et al., 2007)(Jackson Laboratories) mouse strains. A single tamoxifen injection (Sigma-Aldrich, 2.5 mg/20 g body weight) was administered intraperitoneally topregnant females at 15.5 dpc. Embryos were harvested at 18.5 dpc forfurther analysis.

Processing of tissue, histology and in situ hybridizationPost harvesting, tissues were fixed in 4% paraformaldehyde overnight at4°C. Mouse tissues harvested postnatally were decalcified in 8% EDTAsolution (pH 7.5) before further processing. Fixed tissues were embedded inparaffin and sectioned at 5 μm thickness.

Fig. 7. Abrogation of noggin function in theinterzone leads to abolition of joint formation.(A-F) Neutralizing noggin activity with anti-nogginantibody results in expansion of BMP signaling activityin the interzone and the regions flanking it. (A-C) HH34chick hindlimb injected with anti-noggin antibody.(D-F) HH34 chick hindlimb injected with control anti-WGA antibody. Anti-noggin and anti-WGA antibodies,green in the extracellular space. (B,E) pSMAD1/5/8indirect immunofluorescence (red). (C,F) Col2a1mRNA expression in sections adjacent to those usedin B and E. (G-P) Deletion ofNog in mouse embryos at15.5 and 18.5 dpc causes complete loss of molecularand histological structures of interzone/articularcartilage. In contrast to a control mouse digit (G),Gdf5mRNA expression (G,J) is not detected in the putativeinterzone region of a noggin-deleted mutant digit (J).Ectopic Col2a1 mRNA is detected in the putativeinterzone region of the mutant digit (compare panels Hand K). (I,L) Similarly, compared with the control (I),expansion of BMP signaling activity is detected in theinterzone (yellow arrowhead) and in the cells flankingthe interzone of the mutant (L). (M,N) 15.5-dpc control(M) and mutant (N) Alcian Blue-stained skeletal prepsshow absence of joint structures in the MTP region ofthe mutant. (O,P) Alcian Blue/Alizarin Red-stained18.5-dpc mouse whole-mount skeletal prep revealsabsence of joint structures in the mutant (P) comparedwith the control (black arrowheads, O). Dashed linesrepresent the margins of the skeletal elements in thedeveloping digits. Scale bars: 100 μm.

1177

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

Skeletal sections were stained with Alcian Blue 8GX (Sigma-Aldrich)solution (pH 2.5) for 2 min and washed.

Labeled antisense-RNA probes were synthesized using ChEST clones(Geneservice) for Atx (ChEST520a5) and PCNA (ChEST631f8). Thetemplates for Gdf5 and Col2a1 (Hartmann and Tabin, 2001) and for Nog(Pathi et al., 1999) have been described previously. Digoxigenin- or biotin-labeled anti-sense probes were generated as described, and mRNA in situhybridization on paraffin sections for chromogenic detection was performedessentially as described (Murtaugh et al., 1999). Fluorescent double mRNAin situ hybridization was carried out as described (Roy et al., 2013).

Cell proliferation assay200 μg of EdU (E10415)/BrdU (B23151) (Invitrogen) in 200 μl of PBS wasinjectedon topof chickenembryos, irrespectiveof the stageofdevelopment.ForEdU/BrdU labeling in mice, pregnant females were injected intraperitoneallywith 100 mg/kg body weight of EdU and/or BrdU. In the EdU competitionassay, five times molar excess of dTTP (Thermo Fisher Scientific) was used.

EdU detection was carried out as per manufacturer’s instruction (C10083,Invitrogen). BrdU or double-labeling of EdU-BrdU was performed as perZeng et al. (2010)

For postnatal stages, we could not get either EdU or DAPI to work on P21and P42 samples. Moreover, we observed non-specific staining influorescence detection of BrdU. Therefore, we developed BrdU withDAB reagents to detect it chromogenically for these stages.

ImmunohistochemistrySections were deparaffinized and were rehydrated in PBS, followed bypost-fixation in 4% PFA. Sections were incubated overnight at 4°C with thefollowing primary antibodies: anti-pH3 (Sigma-Aldrich, H0412; 1:100),anti-BrdU (Sigma-Aldrich, B8434; 1:150), anti-β-catenin-Cy3 (Sigma-Aldrich, C7738; 1:200), anti-GFP (Invitrogen, A6455; 1:300), anti-GAG(3C2; Potts et al., 1987), anti-pSMAD1/5/8 (Cell Signaling, 9511; 1:100)and anti-noggin (R&D Systems, AF719; 1:100). Following this step, tissueswere washed in PBT and incubated with respective secondary antibodies,such as DyLight 549 AffiniPure goat anti-rabbit IgG (JacksonImmunoResearch Laboratories, 111-505-003; 1:200); Alexa Fluor 594-conjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories,115-585-003; 1:200); anti-goat HRP antibody (Vector Labs, PI-9500;1:100) for noggin. These secondary antibodies were later detected using aTSA kit (T30955, T20932, Invitrogen) and a Vectastain ABCKit (PK6100).

Viral misexpressionRCAS-caBMPR1B (Zou et al., 1997) and RCAS-mBMP4 (Duprez et al.,1996) virus particles were generated as described (Logan and Tabin, 1998).

Organ cultureH34 chick embryonic hindlimbs were cultured for 24 h (Bandyopadhyayet al., 2008). rhBMP4 protein (peproTech, 120-05) was dissolved in DMEMand 2% BSA to a working concentration of 100 ng/μl, along with the vitaldye Fast Green, and injected in HH34MTP joints prior to organ culture. Thecontrol included the buffer without protein. Wnt3a (R&D Systems, 1324-WN-002) protein injection was carried out in a similar manner at a workingconcentration of 100 ng/μl.

For injection, the anti-noggin antibody (R&D Systems, AF719) wasdissolved at 0.1 mg/ml concentration in glycerol. Control was an isotype-matched antibody, i.e. an anti-WGA antibody (Vector Labs, AS2024) wasused as an isotype-matched control.

AcknowledgementsWe are immensely grateful to Prof. Andrew Lassar (HMS, Harvard University, MA,USA) for sharing unpublished data with us.We sincerely thank Prof. K. Subramaniam(IITK, India), Dr Jonaki Sen (IITK, India), Dr Jennifer Mansfield (Barnard College, NY,USA) and Dr Kimberly Cooper (UCSD, CA, USA) for their critical comments on themanuscript. We acknowledge kind help of Prof. Jyotsna Dhawan (InStem, Bangalore,India), Dr Christine Hartmann (IMP, Vienna, Austria) and Prof. Clifford Tabin (HarvardUniversity, MA, USA) for valuable reagents used in thiswork.We thankPremSwaroopYadav for technical help.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsA.R., P.N.P.S., M.L.S., R.M.H. and A.B. designed the experiments. A.R., P.N.P.S.and M.L.S. performed experiments and interpreted data. R.M.H., A.R., P.N.P.S. andA.B. wrote the manuscript.

FundingThis work was supported by grants from the Department of Biotechnology, India(DBT) Department of Biotechnology, India (DBT) [BT/PR11202/MED/32/46/2008andBT/IN/DENMARK/02/PDN/2011 toA.B.]; theDanishStrategicResearchCouncil[10-093757 to A.B.] and a National Institutes of Health grant [GM 49346 to R.M.H.].Fellowships to A.R. and P.N.P.S. are supported by the Indian Ministry of HumanResource Development (MHRD) and the Indian Council of Scientific & IndustrialResearch (CSIR), respectively. Deposited in PMC for release after 12 months.

Supplementary materialSupplementary material available online athttp://dev.biologists.org/lookup/suppl/doi:10.1242/dev.110940/-/DC1

Fig. 8. Model for articular cartilage development. (A) Schematic oflandmark events in cartilage differentiation, resulting in simultaneous formationof transient and articular cartilage in adjacent populations. (B,C) According toour model, all newly formed chondrocytes are capable of differentiating alongthe transient as well as articular cartilage fates. When a morphologicallydistinct interzone is first visible, Nog mRNA is expressed in a broad domain inthe developing cartilage outside the interzone. During this early cartilagedevelopment, BMP signaling is restricted to the perichondrium (I). Asdevelopment progresses, the Nog domain becomes restricted to theepiphyseal ends of the skeletal elements, a few cell layers away from theinterzone. As a result, the BMP signaling domain expands but the cells of theDPZ (blue area in B,C) remain insulated. Wnt ligands that had been producedin the interzone signal the cells flanking the interzone to promote articularcartilage fate (II). During the even later developmental period, theseexpression domains are further consolidated. The progeny of the DPZ cellsthat expand towards the interzone turn off Col2a1 mRNA expression butcontinue to express the prechondrogenic marker Sox9. Theseprechondrogenic cells, sandwiched between the BMP and Wnt signalingdomains (III), are capable of undergoing transient as well as articular cartilagedifferentiation. When these cells come under the influence of Wnt signaling,they differentiate as articular cartilage. On the other hand, the progeny of DPZcells that expand towards the hypertrophic region comewithin the zone of BMPsignaling influence and differentiate as transient cartilage.

1178

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT

ReferencesAkiyama, H., Kim, J.-E., Nakashima, K., Balmes, G., Iwai, N., Deng, J. M., Zhang,Z., Martin, J. F., Behringer, R. R., Nakamura, T. et al. (2005). Osteo-chondroprogenitor cells are derived from Sox9 expressing precursors. Proc.Natl. Acad. Sci. USA 102, 14665-14670.

Archer, C. W., Dowthwaite, G. P. and Francis-West, P. (2003). Development ofsynovial joints. Birth Defects Res. C Embryo Today 69, 144-155.

Bandyopadhyay, A., Kubilus, J. K., Crochiere, M. L., Linsenmayer, T. F. andTabin, C. J. (2008). Identification of unique molecular subdomains in theperichondrium and periosteum and their role in regulating gene expression in theunderlying chondrocytes. Dev. Biol. 321, 162-174.

Bi, W., Deng, J. M., Zhang, Z., Behringer, R. R. and de Crombrugghe, B. (1999).Sox9 is required for cartilage formation. Nat. Genet. 22, 85-89.

Brunet, L. J., McMahon, J. A., McMahon, A. P. andHarland, R.M. (1998). Noggin,cartilage morphogenesis, and joint formation in the mammalian skeleton. Science280, 1455-1457.

Candela, M. E., Cantley, L., Yasuaha, R., Iwamoto, M., Pacifici, M. andEnomoto-Iwamoto, M. (2014). Distribution of slow-cycling cells in epiphyseal cartilage andrequirement of beta-catenin signaling for their maintenance in growth plate.J. Orthop. Res. 32, 661-668.

Colnot, C. (2005). Cellular and molecular interactions regulating skeletogenesis.J. Cell. Biochem. 95, 688-697.

Dowthwaite, G. P., Bishop, J. C., Redman, S. N., Khan, I. M., Rooney, P., Evans,D. J. R., Haughton, L., Bayram, Z., Boyer, S., Thomson, B. et al. (2004). Thesurface of articular cartilage contains a progenitor cell population. J. Cell Sci. 117,889-897.

Duprez, D., Bell, E. J. d. H., Richardson, M. K., Archer, C. W., Wolpert, L.,Brickell, P. M. and Francis-West, P. H. (1996). Overexpression of BMP-2 andBMP-4 alters the size and shape of developing skeletal elements in the chick limb.Mech. Dev. 57, 145-157.

Francis-West, P. H., Parish, J., Lee, K. and Archer, C. W. (1999). BMP/GDF-signalling interactions during synovial joint development. Cell Tissue Res. 296,111-119.

Fu, Q., Gremeaux, L., Luque, R. M., Liekens, D., Chen, J., Buch, T.,Waisman, A.,Kineman, R. and Vankelecom, H. (2012). The adult pituitary shows stem/progenitor cell activation in response to injury and is capable of regeneration.Endocrinology 153, 3224-3235.

Gunnell, L. M., Jonason, J. H., Loiselle, A. E., Kohn, A., Schwarz, E. M., Hilton,M. J. and O’Keefe, R. J. (2010). TAK1 regulates cartilage and joint developmentvia the MAPK and BMP signaling pathways. J. Bone Miner. Res. 25, 1784-1797.

Guo, X., Day, T. F., Jiang, X., Garrett-Beal, L., Topol, L. and Yang, Y. (2004). Wnt/beta-catenin signaling is sufficient and necessary for synovial joint formation.Genes Dev. 18, 2404-2417.

Hammerle, B. and Tejedor, F. J. (2002). A method for pulse and chase BrdU-labeling of early chick embryos. J. Neurosci. Methods 122, 59-64.

Hartmann, C. and Tabin, C. J. (2001). Wnt-14 plays a pivotal role in inducingsynovial joint formation in the developing appendicular skeleton. Cell 104,341-351.

Hyde, G., Boot-Handford, R. P. and Wallis, G. A. (2008). Col2a1 lineage tracingreveals that the meniscus of the knee joint has a complex cellular origin. J. Anat.213, 531-538.

Ikeda, T., Kamekura, S., Mabuchi, A., Kou, I., Seki, S., Takato, T., Nakamura, K.,Kawaguchi, H., Ikegawa, S. and Chung, U.-i. (2004). The combination of SOX5,SOX6, and SOX9 (the SOX trio) provides signals sufficient for induction ofpermanent cartilage. Arthritis Rheum. 50, 3561-3573.

Kahn, J., Shwartz, Y., Blitz, E., Krief, S., Sharir, A., Breitel, D. A., Rattenbach, R.,Relaix, F., Maire, P. and Rountree, R. B. (2009). Muscle contraction is necessaryto maintain joint progenitor cell fate. Dev. Cell 16, 734-743.

Karsenty, G. and Wagner, E. F. (2002). Reaching a genetic and molecularunderstanding of skeletal development. Dev. Cell 2, 389-406.

Kohler, T., Prols, F. and Brand-Saberi, B. (2005). PCNA in situ hybridization: anovel and reliable tool for detection of dynamic changes in proliferative activity.Histochem. Cell Biol. 123, 315-327.

Koyama, E., Shibukawa, Y., Nagayama, M., Sugito, H., Young, B., Yuasa, T.,Okabe, T., Ochiai, T., Kamiya, N., Rountree, R. B. et al. (2008). A distinct cohortof progenitor cells participates in synovial joint and articular cartilage formationduring mouse limb skeletogenesis. Dev. Biol. 316, 62-73.

Kronenberg, H. M. (2003). Developmental regulation of the growth plate. Nature423, 332-336.

Logan, M. and Tabin, C. (1998). Targeted gene misexpression in chick limb budsusing avian replication-competent retroviruses. Methods 14, 407-420.

Mankin, H. J. (1962). Localization of tritiated thymidine in articular cartilage ofrabbits I. Growth in immature cartilage. J. Bone Joint Surg. 44, 682-688.

Murtaugh, L. C., Chyung, J. H. and Lassar, A. B. (1999). Sonic hedgehogpromotes somitic chondrogenesis by altering the cellular response to BMPsignaling. Genes Dev. 13, 225-237.

Muzumdar, M. D., Tasic, B., Miyamichi, K., Li, L. and Luo, L. (2007). A globaldouble-fluorescent Cre reporter mouse. Genesis 45, 593-605.

Nakamura, E., Nguyen, M.-T. and Mackem, S. (2006). Kinetics of tamoxifen-regulated Cre activity in mice using a cartilage-specific CreERT to assay temporalactivity windows along the proximodistal limb skeleton. Dev. Dyn. 235,2603-2612.

Pacifici, M., Koyama, E. and Iwamoto,M. (2005). Mechanisms of synovial joint andarticular cartilage formation: recent advances, but many lingering mysteries. BirthDefects Res. C Embryo Today 75, 237-248.

Pathi, S., Rutenberg, J. B., Johnson, R. L. and Vortkamp, A. (1999). Interaction ofIhh and BMP/Noggin signaling during cartilage differentiation. Dev. Biol. 209,239-253.

Potts, W. M., Olsen, M., Boettiger, D. and Vogt, V. M. (1987). Epitope mapping ofmonoclonal antibodies to gag protein p19 of avian sarcoma and leukaemiaviruses. J. Gen. Virol. 68, 3177-3182.

Rountree, R. B., Schoor, M., Chen, H., Marks, M. E., Harley, V., Mishina, Y. andKingsley, D. M. (2004). BMP receptor signaling is required for postnatalmaintenance of articular cartilage. PLoS Biol. 2, e355.

Roy, P., Kumar, B., Shende, A., Singh, A., Meena, A., Ghosal, R., Ranganathan,M. and Bandyopadhyay, A. (2013). A genome-wide screen indicates correlationbetween differentiation and expression of metabolism related genes. PLoS ONE8, e63670.

Shubin, N. and Alberch, P. (1986). A morphogenetic approach to the origin andbasic organization of the tetrapod limb. In Evolutionary Biology (ed. M. Hecht, B.Wallace and G. Prance). pp. 319-387. US: Springer.

Spater, D., Hill, T. P., Gruber, M. and Hartmann, C. (2006a). Role of canonicalWnt-signalling in joint formation. Eur. Cell. Mater. 12, 71-80.

Spater, D., Hill, T. P., O’Sullivan, R. J., Gruber, M., Conner, D. A. and Hartmann,C. (2006b). Wnt9a signaling is required for joint integrity and regulation of Ihhduring chondrogenesis. Development 133, 3039-3049.

Stafford, D. A., Brunet, L. J., Khokha, M. K., Economides, A. N. and Harland,R. M. (2011). Cooperative activity of noggin and gremlin 1 in axial skeletondevelopment. Development 138, 1005-1014.

Terpstra, L., Prud’homme, J., Arabian, A., Takeda, S., Karsenty, G., Dedhar, S.and St-Arnaud, R. (2003). Reduced chondrocyte proliferation andchondrodysplasia in mice lacking the integrin-linked kinase in chondrocytes.J. Cell. Biol. 162, 139-148.

Tsumaki, N., Tanaka, K., Arikawa-Hirasawa, E., Nakase, T., Kimura, T., Thomas,J. T., Ochi, T., Luyten, F. P. and Yamada, Y. (1999). Role of CDMP-1 in skeletalmorphogenesis: promotion of mesenchymal cell recruitment and chondrocytedifferentiation. J. Cell. Biol. 144, 161-173.

Zeng, C., Pan, F., Jones, L. A., Lim, M. M., Griffin, E. A., Sheline, Y. I., Mintun,M. A., Holtzman, D. M. and Mach, R. H. (2010). Evaluation of 5-ethynyl-2′-deoxyuridine staining as a sensitive and reliable method for studying cellproliferation in the adult nervous system. Brain Res. 1319, 21-32.

Zou, H.,Wieser, R., Massague, J. andNiswander, L. (1997). Distinct roles of type Ibone morphogenetic protein receptors in the formation and differentiation ofcartilage. Genes Dev. 11, 2191-2203.

1179

RESEARCH ARTICLE Development (2015) 142, 1169-1179 doi:10.1242/dev.110940

DEVELO

PM

ENT