PTHrP and Indian hedgehog control differentiation of growth plate chondrocytes at multiple steps

Developing murine growth plates comprise at least three morphologically distinct groups of chondrocytes (see Fig. 3C): round periarticular chondrocytes, flat columnar chondrocytes and hypertrophic chondrocytes. Periarticular chondrocytes proliferate and differentiate into columnar chondrocytes that proliferate further and form orderly columns. Unlike the round periarticular chondrocytes, the flat columnar chondrocytes form columns with clearly defined polarity; this polarity is particularly important in the asymmetric lengthening of the long bones of the limbs. These cells stop proliferating and then differentiate into non-proliferating hypertrophic cells. This sequential and synchronized differentiation is tightly controlled during endochondral bone development; consequently, sharp borders separate the exclusive domains of these three types of cells. One of the major regulators of this differentiation is parathyroid hormone (PTH)-related protein (PTHrP) expressed in the periarticular region. Both PTHrP null (Pthrp^–/–; Pthlh – Mouse Genome Informatics) and PTH/PTHrP receptor (PPR) null (Ppr^–/–; Pthr – Mouse Genome Informatics) mice develop chondrodysplasia because of premature hypertrophic differentiation (Karaplis et al., 1994; Lanske et al., 1996). The columnar layer is short in Pthrp^–/– mice and virtually absent in Ppr^–/– mice.

PTHrP expression in the periarticular region depends upon and is upregulated by another factor that is crucial for cartilage development, Indian hedgehog (Ihh) (Vortkamp et al., 1996; Lanske et al., 1996; Chung et al., 1998; St-Jacque et al., 1999; Chung et al., 2001). In addition to stimulating PTHrP expression, Ihh expression is associated with PTHrP-independent chondrocyte proliferation (Karp et al., 2000).

Previous studies using chimeric growth plates comprising wild-type and PPR-null cells have demonstrated that loss of PPR signaling in proliferating chondrocytes caused premature hypertrophic differentiation (Chung et al., 1998). However, Ppr^–/– cells in the periarticular region of the chimeric growth plate are not morphologically distinct from wild-type cells. The possible influence of PPR signaling on the differentiation of periarticular chondrocytes to columnar chondrocytes is, therefore, unclear.

To understand better how PTHrP and Ihh signaling regulate chondrocyte differentiation, we have developed mice with PPR ablation in chondrocytes using the Cre-loxP system (Pluck, 1996). During generation of floxed mice, we also established a mouse line with reduced PPR expression. Through the analysis of these novel mouse models with abnormal PPR signaling, we show that loss or impairment of PPR signaling is associated with chondrocyte differentiation not only at the terminal step but also at an earlier step. In another model with mosaic ablation of the PPR in the growth plate, we show that the differentiation of early chondrocytes is correlated with Ihh action but is not directly regulated by PTHrP. Based on these findings, we propose a model in which PTHrP and Ihh control chondrocyte differentiation at multiple steps.

The targeting vector was designed to introduce two loxP sequences and a neomycin resistant gene (neo) to generate a floxed PPR allele for gene inactivation by Cre recombinase (Sauer, 1993). The loxP sites are placed in the introns flanking the essential E1 exon. A neo is placed adjacent to the first loxP site and an exogenous HindIII site is placed at the second loxP site (Fig. 1A). A thymidine kinase (tk) cassette is placed outside of the homologous sequence. The PPR genomic clones were isolated from a DashII 129/SvJ mouse liver genomic library using a rat PPR cDNA probe (Lanske et al., 1996). Two loxP sequences were inserted into a BamHI/KpnI PPR gene fragment containing exon E1 at the BglII and SacI sites. An MC-1 neo cassette (kindly provided by Dr S. Dymecki) was placed adjacent to the first loxP site. A XbaI/BamHI fragment and a KpnI/HindIII fragment, both adjacent to the BamHI/KpnI fragment in the PPR gene, were ligated to the modified BamHI/KpnI fragment to reconstitute the 7.5 kb of the PPR genomic sequence spanning the floxed exon E1 and exon E2. The tk cassette, derived from the pPNT vector (Karaplis et al., 1994) was placed at the 3′ end of the construct.

NotI-linearlized vector was transfected into J1 ES cells by electroporation. ES cells were selected with G418 (250 μg/ml) and gancyclovir (2 μM). Doubly resistant colonies were subjected to Southern blot analysis. Homologous recombinants were diagnosed by Southern hybridization using the 5′-external probe A (Fig. 1A) after HindIII digestion of the ES cell genomic DNA. Homologous recombinants showed a mutant 8.5 kb band and a 10 kb band corresponding to the wild-type allele. Homologous recombination at the 3′ side of the targeting vector was also confirmed by Southern hybridization using 3′-external probe B after XhoI digestion. Homologous recombinants had a 6 kb band that corresponded to the mutant allele. Several ES cell lines were injected into C57BL/6 blastocysts for generation of chimeric mice. Chimeric mice were crossed with C57BL/6 mice to establish F₁ lines. Three independent lines were established. One of them (line FL23) exhibited an unexpected mutation in the PPR locus, as described in the text.

Col2-Cre transgenic mice where Cre recombinase (O’Gorman et al., 1997) is expressed under the rat collagen type II promoter (Yamada, 1990) were generated as described elsewhere (Schipani et al., 2001). Ost-Cre transgenic mice bear a fusion gene composed of a 1.3 kb fragment of the mouse OG2 promoter (Ferendo et al., 1998) fused to Cre recombinase and polyA signal excised from a pCBM-9 vector (Saur and Henderson, 1989). After removing vector sequence, DNA was subjected to pronuclear injection. Cre activity was assessed using Rosa26-R (R26R) reporter mice (Soriano, 1999): Cre transgenic mice were crossed with R26R mice. Embryos were collected, fixed and stained with X-gal, as described elsewhere (Chung et al., 1998).

The floxed PPR allele and the unanticipated d allele were analyzed by PCR using primers P1 (5′-TGGACGCAGACGATGTCTTTACCA-3′) and P2 (5′-ACATGGCCATGCCTGGGTCTGAGA-3′), which recognize the sequences spanning the 3′ loxP site. Wild-type and mutant alleles give 450 bp and 490 bp PCR products, respectively. The PPR null locus was detected using PCR primers, P3 (5′-CCACCAATGTGAGTTCCTACAGAAA-3′) for an intronic sequence between exons E2 and E3, and P4 (5′-TCCAGACTGCCTTGGGAAAAGCGC-3′) for the PGK promoter used for the neomycin resistant marker. The mutant allele with a retained PGK promoter sequence gives a 500 bp band. Cre sequences were detected by PCR using primers recognizing internal sequence of the transgene (P5, 5′-CGCGGTCTGGCAGTAAAAACTATC-3′; P6, 5′-CCCACCGTCAGTACGTGAGATATC-3′). All PCR reactions used the following program: 94°C for 10 minutes, followed by 35 cycles of 95°C for 30 seconds, 68°C for 30 seconds, and 72°C for 1 minute.

DNA was prepared from the livers of mice homozygous for the wild-type PPR, the floxed PPR and the mutated PPR described in the text. After overnight enzyme digestion, DNA was separated in 0.7% agarose gels, denatured and transferred onto nylon or nitrocellulose membranes.

Total RNA was extracted from the kidneys of 3-week-old mice. RNA was separated in a 1% agarose gel and transferred onto a nitrocellulose membrane. The probe for the mouse PPR was generated by a random priming method (Megaprime, Amersham), using the DNA template amplified from mouse kidneys by RT-PCR using primers P7 (5′-ACCAACTACTACTGGATTCTGGTGG-3′) and P8 (5′-CGGCTCCAAGACTTCCTAATCTCTG-3′). The probes used for Southern analysis are indicated in Fig. 1A. RNA and DNA hybridization was performed using the QuickHyb Kit (Stratagene) according to the manufacturer’s instruction. Signals were visualized on X-ray films or by the Cyclone storage phosphor system (Packard). PPR and GAPDH mRNA were quantified using the Cyclone storage phosphor system. PPR mRNA levels normalized for GAPDH mRNA were compared. The band intensity of the Southern analysis was similarly quantified and normalized with the Gapdh gene. The normalized band intensities obtained from at least three independent blots for each set were used for the comparisons.

Total RNA was prepared from the kidney of homozygous d/d mutant mice. Reverse transcription was performed using Superscript reverse transcriptase (Gibco/BRL) followed by PCR using primers P9 (5′-CCGAGGGACGCGGCCCTAG-3′) and P10 (5′-AGTCCTGAATAGACAGCCAGCCAAA-3′) to amplify the entire coding sequence of the PPR. DNA sequence was determined by direct sequencing bi-directionally using the following sequencing primers: forward primers, P9, P11 (5′-GCTGCTCAAGGAAGTTCTGCACACA-3′), P12 (5′-GATCTACACCGTGGGATATTCCATG-3′), P13 (5′-ACCAACTACTACTGGATTCTGGTGG-3′), P14 (5′-CACTGTGGCAGATCCAGATGCACTA-3′); backward primers, P10, P14 (5′-CGGCTCCAAGACTTCCTAATCTCTG-3′), P15 (5′-AGTGTTGGCCAAGGTTGCTCTGACA-3′), P16 (5′-GCAGCATAAACGACAGGAACATGTG-3′), P17 (5′-CAGCAAACGATGTTGTCCCACTCTG-3′).

Tissues were fixed in 4% paraformaldehyde/PBS overnight at 4°C, processed, embedded in paraffin wax and cut. Sections were stained with H&E or nuclear Fast Red (Vector laboratories). In situ hybridization was performed as described previously (Lee et al., 1995) by using complimentary ³⁵S-labeled riboprobes. The probes for mouse type X collagen, mouse patched 1, mouse Ihh and rat PTHrP were obtained from Dr Bjorn Olsen (Harvard Medical School), Dr Ron Johnson (Stanford University, Stanford) Dr Benoit St-Jacques (Harvard University) and Dr Andrew C. Karaplis (McGill University), respectively. Rat full-length PPR cDNA probe, R15B has been described previously (Calvi et al., 2001). An exon E1 specific PPR probe was generated by PCR using primers: F, 5′-GTGGACGCAGACGATGTCTTTACC-3′; and R, 5′-CTGCTGTGTGCAGAACTTCCTTGA-3′. PCR product was subcloned into pGEMT Easy vector (Promega).

Pregnant mice received intra-peritoneal injections of 50 μg BrdU/g of body weight and were sacrificed 1 or 24 hours later. Limbs were dissected and fixed in 4% paraformaldehyde overnight at 4°C. Tissues were processed, embedded and sectioned using standard procedures. BrdU was detected using a BrdU Staining Kit (Zymed Laboratories). The BrdU-positive and -negative nuclei were counted in the periarticular region and the columnar region separately. The border between the periarticular and columnar regions was defined as the line separating these two morphologically distinct groups of chondrocytes.

For counting of BrdU-positive cells in the periarticular region, using sections with smaller periarticular regions, we first determined an area for BrdU counting by drawing a closed line that made the area as large as possible, while avoiding the border containing ambiguous cells. Then, we applied the same area on the control sections as it was placed in the center of the corresponding region. We confirmed that the area only included cells with the typical morphological appearance of periarticular chondrocytes. For counting of BrdU-positive cells in the columnar region, as Ppr^d/– growth plates lack sharp transition of between columnar and hypertrophic regions, we first chose top one third of the columnar region of Ppr^d/– mice not to include any hypertrophic cells. We set a rectangular area excluding periarticular and perichondrial cells for BrdU counting. The same rectangle was applied onto control sections. Exclusion of other types of cells was similarly confirmed.

Mutants and controls used in this study were littermates. Nine sections from at least three independent mice per group were counted. Statistical analysis was done by ANOVA.

For the generation of mice with insertion of loxP sequences (floxed mice), a targeting vector was designed with loxP sequences in introns flanking the essential E1 exon of the PPR gene (Fig. 1A). Floxed mice were crossed with Ppr^+/– mice to generate compound heterozygous mice, in which one PPR allele was disrupted and the other was mutated by neo-loxP targeting. The compound heterozygous mice from two independent lines were phenotypically normal. However, mice from one floxed line developed growth retardation and deformity of the limbs postnatally. Homozygous mice for this mutated allele appeared grossly normal. Because the growth retardation over multiple matings and generations invariably appeared in mice carrying the mutated PPR gene opposite an ablated PPR gene, we concluded that the mutant PPR gene was closely linked to and probably responsible for the phenotype. We, hereafter, designate this mutant allele ‘d’ for damaged PPR allele. In situ hybridization for the PPR mRNA in the cartilage of 3-week-old mice revealed decreased expression of PPR mRNA, but the expression pattern was preserved (Fig. 1B). PPR transcripts of normal length were produced in the mutant mice with reduced expression level, determined by northern blot analysis using kidney RNA prepared from homozygous mice (Fig. 1C). Direct sequence of RT-PCR products from homozygous mouse kidney RNA revealed that the transcripts had a normal coding sequence (data not shown). A series of Southern analyses were performed using homozygous mouse DNA to analyze the abnormality of the Ppr gene in the d allele (Fig. 1D-K). Three copies of the targeting vector are inserted in the targeted locus through homologous recombination. This was accompanied by replacement of the endogenous sequence (determined by PCR using P1 and P2 primers; data not shown), an amplification of endogenous sequence immediately 5′ of the targeted region and an insertion of a tk cassette. A possible model for the d Ppr gene is shown in Fig. 1L.

Homozygous floxed, Ppr^fl/fl mice and compound heterozygous Ppr^fl/– mice have no abnormality in growth plate cartilage (data not shown). Chondrocyte-specific Ppr gene ablation was carried out by mating these floxed mice with Cre transgenic mice expressing Cre under the control of the rat type II collagen promoter (Col2-Cre). Cre activity was determined using R26R reporter mice, and was present in growth plate chondrocytes but limited to the chondrocytes and a part of the perichondrium, intra-joint tissues, ligaments and tendons (Fig. 2A). PPR is predominantly expressed in the prehypertrophic region and weakly in columnar chondrocytes (Fig. 2J). PPR expression determined by an exon E1-specific probe was lost in the growth plate of double mutant mice, Col2-Cre:Ppr^fl/fl (Fig. 2K). Col2-Cre:Ppr^fl/fl mice develop chondrodysplasia that resembles that of Ppr^–/– (Fig. 2B,C): the tibial growth plate is shortened and lacks most of the columnar chondrocytes. Reduction on the proliferating chondrocytes with preservation of a fairly normal hypertrophic layer was confirmed by the expression patterns of type II and type X collagens (Fig. 2D-I). The periarticular region is flattened (Fig. 2C). At E16.5, the mutant sternum is mostly occupied by hypertrophic cells, whereas there are few hypertrophic cells in the control (Fig. 2L,M). However, unlike Ppr^–/– mice in some genetic backgrounds, mutant mice survive until birth and they are not as small as Ppr^–/– mice (data not shown). From these observations, we conclude that the Ppr^–/– growth plate phenotype is primarily caused by loss of PPR signaling in the chondrocytes themselves.

In contrast to Ppr^–/– mice, Ppr^d/– mice are born with a reasonably normal appearance and size. Around 7 days of age, they start to show growth retardation and limb deformities (data not shown). To clarify the nature of bone abnormality of Ppr^d/– mice, we examined fetal growth plates. One of the most striking features of the tibial growth plates of the E17.5 Ppr^d/– embryo is the expansion of the hypertrophic layer (Fig. 3A,C). The total length of the tibia is slightly greater than that of the control littermates, unlike the tibial length of Ppr-null mice. The nature of the expanded layer was characterized by in situ hybridization (Fig. 3B): the reduction of PPR expression in Ppr^d/– mice was also present at this age. The type II collagen-expressing domain and the type X collagen-expressing domain have a minimal overlap in the wild-type growth plate, whereas both of the type II and type X collagen-expressing domains are broadened, with an expanded overlap zone in the Ppr^d/– growth plate. Similarly, the prehypertrophic domain marked by Ihh expression is expanded. The upper part of the expanded hypertrophic layer comprises mostly unmineralized, relatively small hypertrophic cells with abundant extracellular matrix (Fig. 3C). These hypertrophic cells do not express osteopontin, a marker for mineralized hypertrophic cells and osteoblasts (Fig. 3B). Therefore, the expansion of the hypertrophic layer is mostly due to an increase in early hypertrophic cells. The columnar region, as well as the periarticular region of Ppr^d/– growth plates, is slightly, but consistently, smaller than that of controls (indicated by brackets in Fig. 3C).

To support the hypothesis further that reduced PPR signaling in chondrocytes is responsible for the cartilage phenotype of Ppr^d/– mice, we crossed Ppr^d/– mice with transgenic mice that express a constitutively active Ppr mutant gene in chondrocytes using rat type II collagen promoter (caPpr). This gene has previously been shown to rescue the growth plate abnormality of Pthrp^–/– mice (Schipani et al., 1997). The proximal tibias of E17.5 caPpr mice, Ppr^d/– mice and caPpr:Ppr^d/– mice were compared (Fig. 3D). The tibia of the caPpr mouse is characterized by an expansion of fully differentiated hypertrophic chondrocytes that express osteopontin and a decreased amount of type X collagen mRNA. Although both the caPpr and Ppr^d/– mice show expansion of the hypertrophic region, the hypertrophic region of Ppr^d/– mice expresses high levels of type X collagen mRNA and little osteopontin mRNA. This characteristic abnormality of the Ppr^d/– growth plate disappears when the caPpr gene is introduced, and the growth plate of the double mutant mice, caPpr:Ppr^d/– is indistinguishable from that of caPpr.

The number of cells in each chondrocyte layer represents the balance between the number of cells entering and leaving the layer as well as proliferation within the layer. To analyze these steps, first, we performed BrdU labeling (Fig. 4A): when E17.5 mice were sacrificed 1 hour after BrdU administration, no BrdU-positive cells were found in the hypertrophic region either in Ppr^d/– or in wild-type controls. The fraction of cells labeled with BrdU in the columnar region was unchanged in Ppr^d/– mice. It was, however, significantly increased in the periarticular region of Ppr^d/– mice, even though the region was smaller. This was also confirmed in E16.5 and E18.5 mice (data not shown). This suggests that periarticular chondrocytes differentiate into columnar chondrocytes and therefore leave the periarticular region at a greater rate than that of controls. Then, we performed BrdU pulse-chase assay to see whether hypertrophic chondrocytes are also generated at a greater rate in Ppr^d/– mice. Proliferating chondrocytes of E17.5 embryos were pulse labeled with BrdU. Mice were sacrificed 24 hours after BrdU labeling. BrdU-positive hypertrophic chondrocytes are generated during the period. We found that those BrdU-positive hypertrophic chondrocyte did not reach chondro-osseous junction at this condition; therefore, we did not lose BrdU-positive hypertrophic chondrocytes by the replacement of cartilage by bone cells. As hypertrophic chondrocytes do not incorporate BrdU (Fig. 4A), these BrdU-positive hypertrophic chondrocytes are generated by differentiation of columnar chondrocytes labeled with BrdU 24 hours before sacrifice. The number of BrdU-positive hypertrophic chondrocytes is determined by the number and proliferation of BrdU-labeled proliferating chondrocytes and the rate of their hypertrophic differentiation. The number of BrdU-positive hypertrophic cells produced in Ppr^d/– mice during this time was greater than that of controls, and the area encompassed by BrdU-positive hypertrophic cells was greater (Fig. 4B). As the number and the proliferation rate of columnar chondrocytes of Ppr^d/– mice were not greater than the control (Fig. 4A), we concluded that the increased number of BrdU-positive hypertrophic chondrocytes was due to an increase in the rate of hypertrophic differentiation. The observations that osteopontin is only expressed in the terminal end of the hypertrophic region of Ppr^d/– mice with no expansion in number of osteopontin-positive cells suggest that this acceleration does not continue in the step of further hypertrophic maturation. Thus, acceleration of the differentiation rate of periarticular into columnar chondrocytes as well as of columnar to hypertrophic chondrocytes appear to have caused an accumulation of early hypertrophic chondrocytes in Ppr^d/– mice. The acceleration of periarticular to columnar differentiation was also observed in mice with PPR ablation in chondrocytes of E17.5 mice (Fig. 4C). Despite the decreased size of the periarticular region, chondrocyte proliferation was increased.

The PPR is little, if at all, expressed in periarticular chondrocytes. We, therefore, considered other possible mediators of the accelerated differentiation of periarticular chondrocytes in PPR-defective mice. We found that morphological changes of the growth plate in these PPR-defective mice also altered the Ihh expression domain (Fig. 5). As Ihh has PTHrP-independent roles in cartilage development, and Ihh^–/– mice show marked reduction in proliferation (Karp et al., 2000), we hypothesized that possible upregulation of Ihh action may have caused this acceleration of early chondrocyte differentiation. To test this possibility, we examined expression of patched 1 (Ptc; Ptch – Mouse Genome Informatics), a marker of Ihh action (Goodrich et al., 1996), and Pprp expression (Fig. 5). Ptc expression is strongest in the domain adjacent to the Ihh domain, fading away towards the articular surface in the Ppr^+/+ growth plate, whereas its expression in the periarticular region is upregulated in Col-2Cre: Ppr^fl/fl and Ppr^d/– mice. Upregulation of PTHrP expression in the periarticular region also suggests increased Ihh activity in this region.

To determine whether Ihh action stimulates differentiation of periarticular to columnar chondrocytes, we introduced ectopic Ihh expression near the periarticular region. For this purpose, we took advantage of a line of Osteocalcin-Cre (Ost-Cre) transgenic mice. A reporter experiment showed that 30% of proliferating chondrocytes had Cre activity (Fig. 6A). The activity is primarily limited to the columnar region and barely present in the periarticular region. Ost-Cre: Ppr^fl/fl mice develop ectopic hypertrophy in the columnar region due to premature hypertrophic differentiation of the cells that have lost the PPR gene (Fig. 6C). These cells express Ihh close to the periarticular region (Fig. 6E), with evidence of increased Ihh signaling in the periarticular region, as indicated by Ptc and Pthrp upregulation (Fig. 6G,I). The BrdU labeling ratio of the periarticular chondrocytes of E17.5 mice was significantly increased without enlargement of the periarticular region, suggesting acceleration of periarticular chondrocyte differentiation into columnar cells (Fig. 6J). This increase in BrdU labeling ratio of periarticular chondrocytes was also confirmed in E16.5 and E18.5 embryos (data not shown). This acceleration occurs in the presence of increased PTHrP expression (Fig. 6I) with presumably little disruption of the Ppr gene in the periarticular region, as Cre activity in this region is minimal. Hence, we conclude that the acceleration of periarticular to columnar cells is not caused directly by loss of PPR of the periarticular chondrocytes.

The role of PTHrP signaling through the PPR in chondrocyte differentiation has been demonstrated by analysis of mutant mice missing the genes for these proteins (Karaplis et al., 1994; Lanske et al., 1996) as well as though the use of chimeric mice (Chung et al., 1998; Chung et al., 2001).

The observation that Col2-Cre:Ppr^fl/fl mice showed a growth plate abnormality similar to that of the Ppr^–/– mice is consistent with the previous finding that loss of PPR signaling in columnar chondrocytes per se accelerates their terminal differentiation (Chung et al., 1998). However, chondrocyte-specific Ppr-null mice differ from Ppr^–/– mice in body size and embryonic lethality. Further, in the Ppr^–/– growth plate, the initial hypertrophic differentiation is delayed (Lanske et al., 1999), whereas it is accelerated in chondrocyte-specific Ppr-null mice (e.g. Fig. 2J,K). Thus, the loss of the PPR in the cartilage as well as in the other tissues contribute to the Ppr^–/– growth plate phenotype. Nevertheless, the strong similarities of the cartilage phenotypes seen among Ppr^–/–, Pthrp^–/– and Col2-Cre:Ppr^fl/fl mice suggests that these phenotypes are primarily caused by the loss of PPR signaling in chondrocytes.

Although universal gene ablation is a powerful method for analyzing the roles of genes in vivo, the extreme nature of the phenotypes can limit conclusions, especially when gene ablation causes early embryonic lethality. Partial knockout (knock-down) of gene expression can, therefore, be revealing. To date, several different methods have been reported to generate mice that have phenotypes due to reduced gene expression or function: transgenic animals that express antisense RNA (Nemir et al., 2000) or dominant negative proteins (Go et al., 2000), introduction of hypomorphic mutation into proteins (Tang et al., 1997) and insertion of foreign sequences into intronic sequences to disturb efficient RNA splicing (Nagy et al., 1998; Meyers et al., 1998; Mohn et al., 1999). The mice with reduced PPR expression presented here were unintentionally obtained during an attempt to generate floxed PPR mice. Although the genetic structure of the d allele is not completely understood, the following findings indicate that the phenotypes of Ppr^d/– mice are caused by reduction of normal PPR mRNA expression: (1) PPR mRNA expression was reduced, as determined by RNA blot and in situ hybridization analysis; (2) the RT-PCR product had the normal coding sequence of PPR and the mRNA was of normal size; (3) homozygous Ppr^d/d mice have milder phenotypes than do Ppr^d/– mice (data not shown), a result expected if the d allele generates a smaller amount of a normal mRNA; (4) heterozygous Ppr^d/+ mice are virtually normal (data not shown); and (5) overactivity of the PPR caused by a constitutively active PPR mutation completely reverses the Ppr^d/– mutant phenotype. Based on northern blot analysis (Fig. 1B,C and data not shown) and the fact that the Ppr^d/d mice have slightly abnormal growth plates while Ppr^+/– mice do not have apparent morphologic abnormalities, PPR expression level in Ppr^d/– mice is estimated to be less than 50% of that of heterozygous Ppr^+/– mice. The growth plate phenotype of Ppr^d/– mice is superficially very different from that of Ppr^–/– mice. Close observations, however, revealed relative reductions of the periarticular and the columnar regions in Ppr^d/– mice, which are also seen in Ppr^–/– and Col2-Cre:Ppr^fl/fl mice in extreme forms. The diminished extent of the periarticular region, despite increased proliferation in this region in both Ppr^d/– and Col2-Cre:Ppr^fl/fl mice, suggests that loss or impairment of PPR signaling accelerates the differentiation of periarticular cells to columnar cells. The BrdU study also excluded the possibility that an increase in proliferation of columnar cells might have caused the early hypertrophic expansion in the Ppr^d/– growth plate. Thus, the expansion of the hypertrophic layer in the Ppr^d/– mice is probably caused both by acceleration of differentiation of periarticular to columnar cells and acceleration of differentiation of columnar cells into hypertrophic cells. The former process supplies cells to the pool with the highest proliferation rate; therefore, modest acceleration of this step may cause a substantial difference in the production of hypertrophic cells, further abetted by early conversion of proliferating cells to hypertrophic cells. This combination of cell transformation causes the Ppr^d/– mouse to have a tibia longer than normal at E17.5.

PTHrP expression in the periarticular region is dependent on Ihh expressed predominantly in the prehypertrophic chondrocytes. PTHrP, in turn, blocks premature hypertrophic differentiation of columnar chondrocytes (Kronenberg and Chung, 2001). However, Ihh clearly has roles in cartilage development independent of PTHrP, as Ihh^–/– mice show a growth plate phenotype distinct from that of Ppr^–/– mice, with marked reduction of chondrocyte proliferation. Expression of constitutively active PPR in the cartilage is able to reverse only acceleration of terminal differentiation of Ihh-null chondrocytes and is not able to rescue the reduced proliferation (Karp et al., 2000). The positive association between cellular proliferation and Ihh activity is also observed in periarticular chondrocytes of the different models presented here. The increased proliferation accompanies increased rates of differentiation of periarticular cells to columnar/hypertrophic cells.

Based on the observations above, we propose that alteration of PPR signaling in chondrocytes changes the rates of differentiation of periarticular to the columnar chondrocytes (arrow 1) as well as the generation of hypertrophic chondrocytes (arrow 2) (Fig. 7A). This model can explain the diversified phenotypes in various PPR mutant growth plates (Fig. 7B). However, the data from mosaic PPR ablation demonstrate that PPR signaling itself does not directly regulate the first step. Conversely, Ihh activity is positively correlated with the acceleration of the first step, along with an increase in proliferation, suggesting that Ihh action may positively control this step (Fig. 7C). There remains a possibility that another factor secreted from the ectopic hypertrophic chondrocytes might be responsible for this step. This possibility, however, appears unlikely because a previous study of chimeric growth plates composed of wild-type cells and Ppr^–/–;Ihh^–/– doubly mutant cells showed that Ihh in the ectopic hypertrophic chondrocytes was responsible for the characteristic elongation of the columnar layer of the growth plate (Chung et al., 2001).

Our study demonstrates that impairment of PPR signaling creates a novel growth plate phenotype. Through analysis of these mice, we have found that chondrocyte differentiation is controlled at multiple steps by the feedback loop of PTHrP and Ihh, and that Ihh probably stimulates differentiation of early chondrocytes. This process thus may increases a chondrocyte population with a high rate of proliferation.

We thank Dr En Li for providing us ES cells and Melissa Knight for technical assistance. This work was funded in part by NIH grants DK56246 and AR44855.