PDGFRα signaling regulates cartilage and fibrous tissue differentiation during synovial joint development

Woods, John P.; Rackley, Alex; Kwon, Hae Ryong; Olson, Lorin E.

doi:10.1038/s41467-025-59207-1

Download PDF

Article
Open access
Published: 30 April 2025

PDGFRα signaling regulates cartilage and fibrous tissue differentiation during synovial joint development

Nature Communications volume 16, Article number: 4041 (2025) Cite this article

3891 Accesses
2 Altmetric
Metrics details

Subjects

Abstract

Synovial joints develop from mesenchymal structures called interzones, with progenitor cells differentiating into specialized cartilaginous and fibrous tissues of the joint. Platelet-derived growth factor receptor-α (PDGFRα) is a tyrosine kinase expressed by cells of the limb bud, but its role in limb development is unknown. To investigate PDGFRα function, we generated mice expressing mutant PDGFRα with a point mutation (D842V) that increases receptor signaling. Mutant hindlimbs are immobile with knee joints fused by cartilage and lacking ligaments and menisci. The interzone marker Gdf5 is initially expressed at E12.5 but is downregulated thereafter, suggesting a defect in interzone maintenance. Omics analysis of the joint tissues identifies ectopic cartilage matrix expressing genes for cartilage and fibrotic tissue. Thus, elevated PDGFRα signaling corrupts joint development by downregulating Gdf5 and redirecting interzone progenitors into a fibrocartilage fate. This suggests that tight regulation of tyrosine kinase activity is necessary for the development of the mouse knee joint.

FGF receptor inhibitor BGJ398 partially rescues osteoarthritis-like phenotype in older high molecular weight FGF2 transgenic mice via multiple mechanisms

Article Open access 24 September 2022

The synovial microenvironment suppresses chondrocyte hypertrophy and promotes articular chondrocyte differentiation

Article Open access 16 September 2022

In major joint diseases the human synovium retains its potential to form repair cartilage

Article Open access 26 June 2023

Introduction

Synovial joints allow movement between skeletal elements and are made of multiple tissues that comprise a single organ. Each tissue in a synovial joint, including articular cartilage, menisci, and ligaments, is susceptible to diseases and injury that burden health and reduce quality of life^1,2,3,4. Greater understanding of how synovial joints develop will aid in identifying disease susceptibility genes and devising new cell-based treatments for joint injuries and disease.

Long bones and their synovial joints originate from buds of undifferentiated mesenchyme, or limb buds, that first emerge on mouse embryonic day (E) 9.5. By E10.5, the core of the limb bud condenses and expresses Sox9, a master regulator of chondrogenic genes including collagen type II and aggrecan^5,6. These condensations develop into the cartilage templates for long bones.

Each joint arises from an interzone, which is specified by the expression of Gdf5, a member of the bone morphogenic protein (BMP) family^7,8,9. Interzone cells are progenitors that differentiate into cell types that create mature joint tissues including the menisci, articular cartilage, and ligaments^10,11. Interzone specification occurs at E12.5–E13.5 in the mouse hindlimb and marks the site of the prospective knee. Interzone specification is followed by cavitation, a rapidly occurring process between E15 and E15.5, where the cartilage templates are separated by the synovial cavity. Of note, cell death is not a primary driver of this process, rather, cells are separated by the loss of ECM and the accumulation of hyaluronic acid¹². Concomitant with Gdf5 expression, Sox9 becomes downregulated in interzone progenitors, but later Sox9 is re-expressed in interzone cells that differentiate into articular cartilage and menisci¹³. Therefore, Sox9 has a dynamic expression pattern through joint development, being expressed in the cartilage template, then downregulated in the interzone, and finally re-expressed in select synovial joint tissues.

Delicate modulation of pro-chondrogenic BMP and TGFβ signaling activity is crucial for long bone formation, interzone specification, and normal joint development. Noggin is a BMP antagonist, and Nog^−/− limbs have expanded cartilage and fail to develop joints¹⁴. In humans, NOG mutations cause proximal symphalangism 1A (SYM1A, OMIM 185800) and multiple synostoses syndrome 1 (SYNS1, OMIM 186500), two conditions characterized by absence of certain joints¹⁵. In a highly similar fashion, human GDF5 mutations cause syndromes with joint tissue abnormalities or fusions, including multiple synostosis syndrome 2 (SYNS2, OMIM 610017), and proximal symphalangism 1B (SYM1B, OMIM 615298)^16,17. The brachypodism mouse, with an inactivating mutation in Gdf5, exhibits fusions in some distal joints of the limbs^8,18. They also have knee joint abnormalities including defects in bone shape, missing or severely underdeveloped cruciate ligaments, and underdeveloped menisci¹⁹.

TGFβ signaling also impacts long bone and joint development but in a slightly different way compared to BMP. Tgfbr2 binds all TGFβ isoforms to activate TGFβ signaling pathways, and deletion of Tgfbr2 in the mouse limb bud results in absence of interphalangeal joints^20,21. In these mice, the interzone is specified but it fails to maintain Gdf5 expression, resulting in excess cartilage differentiation. Because BMP signaling promotes chondrogenesis and defective TGFβ signaling leads to enhanced cartilage differentiation, it is possible that TGFβ downregulates BMP activity to maintain the process of joint formation²².

Platelet-derived growth factor receptor-α (PDGFRα) is a receptor tyrosine kinase that regulates the proliferation, migration, and survival of many mesenchymal cell types during embryogenesis²³. Deletion of Pdgfra causes defects in somite- and neural crest-derived skeletal progenitors, resulting in spina bifida, rib fusions, and midfacial clefting^24,25,26. Phosphatidylinositol 3-kinase (PI3K) signaling is a primary mechanism by which PDGFRα regulates development of the axial and craniofacial skeleton^27,28. Conversely, elevated PDGFRα signaling causes craniosynostosis and abnormal fusion of the sternal bars^29,30,31. In non-skeletal connective tissue, elevated PDGFRα signaling promotes fibrosis at the expense of adipose tissue expansion^29,32,33,34. Human PDGFRA mutations that affect skeletal development are not known, but PDGFRB gain-of-function mutations cause developmental syndromes with skeletal phenotypes³⁵. However, due to the lack of tissue specific PDGFRα manipulations, the role for PDGFRα signaling in developing limbs and joints has not been described.

In this study, we found that PDGFRα expression is largely the reciprocal of Sox9 expression in limb development, albeit with discrete sites of co-expression in developing knee joints. Intersectional lineage tracing showed that PDGFRα⁺ cells in the Prrx1⁺ limb bud give rise to the long bones and all components of the knee joint. To address whether PDGFRα signaling has a functional role in limb and joint development, we performed genetic gain-of-function and loss-of-function experiments using conditional Pdgfra alleles controlled by Prrx1-Cre. Limb and joint development were normal in the absence of Pdgfra. However, a constitutively active mutant form of PDGFRα disrupted joint development. Single-cell RNA sequencing and spatial transcriptomics demonstrated failure to maintain interzone Gdf5 expression, loss of a Creb5⁺Tppp3⁺Cd44⁺ subpopulation of synovial lining cells, dysregulation of TGF-β/BMP and Wnt signaling, and joint fusion resulting from joint progenitors differentiating into fibrocartilage. These results indicate that joint progenitors are highly sensitive to PDGFRα signaling, which must be tightly regulated for their differentiation into specialized fibrous and cartilaginous tissue types that compose a normal joint.

Results

Pdgfra expression in the developing limb bud and knee joint

Recent studies have explored murine joint development through single-cell RNA sequencing (scRNA-seq), but Pdgfra has not been a focus^13,36,37. scRNA-seq of the Gdf5-Cre lineage from E12.5 to E15.5 knees showed that cells at E12.5 could be divided into clusters of chondrogenic or fibrous/mesenchymal cells, with high Pdgfra expression in the fibrous/mesenchymal cells (Bian et al. ¹³). We queried Pdgfra expression in the data at E13.5-E15.5 and found that Pdgfra continued to be enriched in the Col1a2 expressing fibrogenic portion of the Gdf5-Cre lineage from E13.5 to E15.5 (Supplementary Fig. 1A–C). At E13.5 there was a small overlap between Pdgfra and chondrogenic markers Sox9 and Col2a1 (Supplementary Fig. 1D). However, from E14.5 to E15.5 there was a gradual expansion of low Pdgfra expression in the Sox9-expressing chondrogenic cluster (Supplementary Fig. 1E, F).

The proportion of fibrous and chondrogenic cells changed over time, reflecting the biological process of interzone progenitors increasingly differentiating into chondrogenic cells. From E13.5 to E15.5 time there was an increase in overall Pdgfra expression in the Gdf5-Cre lineage (Supplementary Fig. 1G–J). This shows that Pdgfra is initially expressed in limb bud mesenchyme in a fashion biased towards fibrous/mesenchymal cells, as marked by Col1a1 and Col1a2, but then gradually expands to include some chondrogenic cells marked by Sox9 and Col2a1.

Spatial expression of PDGFRα and Sox9 in joint development

To address the spatial expression of PDGFRα in relation to Sox9, we performed immunostaining. At E13.5 and E14.5, PDGFRα was expressed in the perichondral mesenchyme and interzone, and Sox9 was expressed in the femur and tibia cartilage, as expected. There were discrete areas of PDGFRα and Sox9 co-expression in joint progenitors of the interzone, as well as in the developing entheses, which are fibrocartilage structures that become attachment points for tendons and ligaments (Fig. 1A, arrows).

**Fig. 1: Dynamic spatial expression of PDGFRα and Sox9 in limb and joint development.**

To investigate the contribution of Pdgfra-lineage cells to limb and joint development, we performed PDGFRα-CreER x R26-LSL-tdTomato lineage tracing with tamoxifen (TMX) given at E9.5 or E10.5 and analysis at E13.5 or E16.5. When TMX was given at E9.5, the Sox9^neg perichondral mesenchyme and interzone, as well as Sox9⁺ cartilage, were robustly labeled at E13.5 (Fig. 1B). This indicates that Pdgfra expression included most or all limb bud mesenchyme at E9.5, which is prior to formation of Sox9⁺ cartilage (Fig. 1B). However, when TMX was given at E10.5, Tomato-labeling of Sox9⁺ cartilage was reduced (Fig. 1B, C). This, as well as the restricted expression of PDGFRα at later time points (Fig. 1A), indicates that from E9.5 to E10.5 Pdgfra was downregulated in limb mesenchyme that condensed to form the cartilage template.

These data showed that PDGFRα and Sox9 are largely anti-correlated, but clearly there is some co-expression at E13.5-E14.5 in joint and enthesis progenitors. As summarized in Fig. 1D, the early limb bud is highly PDGFRα⁺, evidenced by E9.5 lineage tracing (Fig. 1B). PDGFRα is downregulated when cells condense into the Sox9⁺ cartilage template at E10.5, shown by E10.5 lineage tracing (Fig. 1B). Between E12.5 and E14.5, most interzone cells lose Sox9 and re-express PDGFRα, but discrete populations co-express PDGFRα and Sox9, as shown by antibody staining (Fig. 1A). These data are consistent with scRNA-seq data¹³ (Supplementary Fig. 1).

PDGFRα gain-of-function causes joint fusion

To identify a functional role for PDGFRα signaling in appendicular skeletal development, we first performed genetic loss-of-function experiments using Pdgfra^flox alleles and Prrx1-Cre, which is expressed in the limb bud starting at E9.5³⁸. However, an earlier study investigating PDGF signaling in hair follicle development performed double-deletion of Pdgfra and Pdgfrb with Prrx1-Cre, and limb development appeared normal until at least E14.5³⁹. When we deleted Pdgfra with Prrx1-Cre, the mice exhibited normal ambulation and joint structure at 3 months of age (Supplementary Fig. 2). This suggests that PDGFRα signaling is not required for limb and joint development.

Next, we performed gain-of-function experiments using a Pdgfra knock-in allele engineered for Cre/lox inducible expression of mutant PDGFRα (denoted as PDGFRα^K) with a kinase domain point mutation (D842V) that increases receptor signaling²⁹. After Cre/lox recombination mediated by Prrx1-Cre, the allele co-expresses PDGFRα^K and Flp^o recombinase under control of the endogenous Pdgfra promoter³⁴ (Fig. 2A). At birth, heterozygous Prrx1-Pdgfrα^K/+ mutants were similar in size to littermate controls, but mutant hindlimbs were consistently extended and immobile at the hips and knees (Fig. 2B, C).

**Fig. 2: PDGFRα gain-of-function causes joint fusion at birth.**

Using whole mount skeletal preparations, we found that Prrx1-Pdgfrα^K/+ hip joints and knee joints were fused (Fig. 2D, Supplementary Fig. 3A). Mutants also exhibited fusion at the glenohumeral (shoulder) joint (Supplementary Fig. 3B). Whole mount knees appeared to have bridges of cartilage on the lateral and medial aspects, with cavitated joint space only in the center (Fig. 2D). Fusions were not seen in other joints.

Upon histological examination, excess cartilage was found in the lateral knee joint, connecting the femur and tibia (Fig. 2E). Coronal sections of the knee at P7 confirmed that the femur and tibia were bridged by ectopic cartilage only on the lateral and medial aspects of the knee (Supplementary Fig. 3C). It was not possible to analyze later time points because Prrx1-Pdgfrα^K/+ mice died by 1 week for reasons that remain unclear. Every mutant knee failed to develop cruciate ligaments (Fig. 2E1). However, some mutants formed rudimentary menisci, and the patellar fat pad was present (Supplementary Fig. 3C). Mutant knees also exhibited an elongated enthesis, or attachment point, of the patellar tendon (Fig. 2E2).

In normal joint development, the perichondrium is a fibrous tissue that forms a sharp, unbroken border with cartilage (Fig. 2F2 white arrows). Safranin O staining of Prrx1-Pdgfra^K/+ mutants revealed blebs of cartilaginous matrix in the perichondrium around the femoral enthesis and articular cartilage (Fig. 2F1, F2 arrows), and an enlarged and misshapen patellar tendon enthesis (Fig. 2F3). Because PDGFRα is highly expressed in the perichondrium (Fig. 1), this phenotype suggested that the border function of the perichondrium was compromised by elevated PDGFRα signaling.

Prrx1-Pdgfrα ^K/+ mutants fail to maintain Gdf5 expression

We sought to determine when joint development defects first appeared in Prrx1-Pdgfrα^K/+ mutants. To explore whether interzone specification was normal, we performed in situ hybridization for Gdf5 at E12.5 and E13.5. Whole mount in situ hybridization at E12.5 showed similar Gdf5 expression in the hindlimb interzone of both genotypes, indicating that specification had occurred (Fig. 3A). However, sections at E13.5 showed that Gdf5 expression was significantly decreased in the mutant, even while Sox9 was properly downregulated in interzone (Fig. 3B, D). By EdU staining, proliferation was unchanged in mutant cartilage and interzone at E13.5 (Supplementary Fig. 4). TUNEL+ cells could not be detected in the mutant or control interzone at E13.5, arguing against loss of Gdf5-expressing cells via apoptosis. Therefore, the interzone was properly specified, but subsequent loss of Gdf5 suggests defective maintenance of the interzone signaling environment.

**Fig. 3: *Prrx1-Pdgfra*^K/+ mutants fail to maintain the interzone.**

Sections at E14.5 stained for Sox9 and PDGFRα showed expanded PDGFRα⁺ area into the Sox9⁺ femoral and tibial cartilage (Fig. 3C arrows, Fig. 3E). Importantly, PDGFRα^K is controlled by the endogenous Pdgfra promoter, and RNAseq has shown that this mutant allele does not lead to Pdgfra overexpression³², so expanded PDGFRα area in the mutant interzone is not merely a misexpression artifact.

Sagittal sections through the middle of the knee joint at E14.5 showed that control knees had clearly demarcated perichondrium between the interzone and flanking cartilage, but the perichondrium of mutant knees was poorly defined with a gradual transition from interzone to cartilage (Fig. 3F). By E15.5, cruciate ligaments had formed in the control, but they were absent in mutants and the mutant patellar tendon was smaller (Fig. 3F). At P1, picrosirius red staining for aligned collagen fibers verified the absence of cruciate ligaments in the mutant (Fig. 3G). These data show that elevated PDGFRα signaling interferes with interzone maintenance by downregulating Gdf5 by E13.5, leading to expanded expression of PDGFRα in Sox9⁺ cartilage by E14.5, and poorly defined cartilage borders and failure to generate cruciate ligaments by P1.

Normal joint development in Gdf5- and Sox9-Pdgfrα ^K/+ mice

To better define the cells that give rise to the knee joint phenotype, we tested two Cre drivers with more selective activity than Prrx1-Cre. First, we tested Gdf5-Cre, which targets interzone progenitor cells beginning at E12.5⁴⁰. However, all Gdf5-Pdgfrα^K/+ mutants (n = 5) exhibited normal knee development when analyzed at E16.5 (Fig. 4A) and adult mice were ambulatory and seemingly normal. This was surprising because Gdf5 and Pdgfra expression appear to overlap in the interzone (Supplementary Fig. 1 and Fig. 1A).

**Fig. 4: *Gdf5-Pdgfra*^K/+ and *Sox9-Pdgfra*^K/+ mice have normal knee joint development.**

We used an intersectional reporter system to verify expression of PDGFRα^K in Gdf5-lineage cells. This approach relies on Gdf5-Cre to induce co-expression of PDGFRα^K and Flp^o recombinase from the Pdgfra^K knock-in allele (Fig. 2A), with Flp^o acting on a R26-frtSTOPfrt-tdTomato reporter to label cells with intersecting Gdf5 and PDGFRα^K expression (Fig. 4B). This demonstrated that all tissues of the synovial joint were labeled, which means PDGFRα^K was indeed expressed in Gdf5-lineage cells (Fig. 4C). These results suggest that disruption of joint development by PDGFRα^K involves elevated tyrosine kinase signaling before the onset of Gdf5-Cre activity at E12.5. Alternatively, disrupted joint development could arise from PDGFRα^K signaling in cells outside the Gdf5-Cre lineage acting in a paracrine manner.

We also tested Sox9-CreER where an IRES-CreER cassette was inserted into the 3’UTR of the Sox9 gene without disrupting endogenous gene expression⁴¹. We verified Cre activity by crossing with a R26-loxSTOPlox-tdTomato reporter and administered TMX daily from E9.5 to E11.5. At E16.5, there was widespread labeling of skeletal tissues including the knee joint (Fig. 4D, E). However, Sox9-Pdgfra^K/+ knee development was indistinguishable from Sox9-CreER control littermates at E16.5 (Fig. 4F).

We again used the intersectional reporter system to verify expression of PDGFRα^K. Based on single-cell analysis and immunostaining showing relatively little overlapping Sox9 and Pdgfra expression (Supplementary Fig. 1 and Fig. 1A), we expected the reporter to label only a subset of joint tissues. Indeed, when TMX was given at E9.5-E11.5, or just E11.5, there was selective labeling of the elbows and knees at E16.5 (Fig. 4G arrowheads). This is consistent with Sox9⁺ cells re-expressing PDGFRα upon forming the interzone (Fig. 1D). However, sections of embryonic knee showed mosaic labeling of joint tissues (Fig. 4F), which suggests that Sox9-CreER did not generate high expression of PDGFRα^K in joint progenitors. Thus, the combination of Sox9-CreER and PDGFRα^K was probably too mosaic to be informative.

Disrupted interzone progenitor populations in mutant joints

Loss of cruciate ligaments and menisci, disruption of the sharp perichondral border and expansion of entheses suggested a shift in joint progenitor populations. To characterize changes in cell populations, we performed scRNA-sequencing on developing knee joints at E13.5, before the phenotype was histologically apparent, and at E15.5, when the joint was clearly dysmorphogenetic. Cells were harvested from the whole E13.5 hindlimbs after removing the paw. At E15.5, cells were harvested from dissected knee joints. We used 2 controls (Prrx1-Pdgfra^+/+) and 2 mutants (Prrx1-Pdgfrα^K/+) for each time point without sorting.

In total, we analyzed 18,207 cells after quality control. We clustered the cells using Seurat’s SNN algorithm, resulting in 13 clusters (Supplementary Fig. 5A, B). UMAP comparisons showed that similar populations recovered between time points or genotypes (Supplementary Fig. 5C, D). Cell-cycle phase analysis showed no changes between control and mutant (Supplementary Fig. 5E), in agreement with EdU analysis suggesting that cell proliferation was not affected (Supplementary Fig. 4).

We examined expression of PDGF receptors and ligands and found that Pdgfra was more highly expressed than Pdgfrb in the main cluster of skeletal cells. There was a low expression of the ligands Pdgfa, Pdgfb, and Pdgfc within the main cluster, with higher expression in muscle (Pdgfa, Pdgfc), immune cells (Pdgfa, Pdgfb), and endothelial cells (Pdgfb) (Supplementary Fig. 5F).

We used differential gene expression analysis to determine cell identities within the main cluster, resulting in eight skeletal cell clusters, C0–C7 (Fig. 5A). Skeletal cell clusters included Cxcl12-, Meis2-high mesenchyme-1 (C0), Postn-, Shox2-high mesenchyme-2 (C6), Tnmd- and Scx-high mesenchyme-3 (C4) and mesenchyme-4 (C3), Runx2-high osteoprogenitors (C5), Sox9-, Gdf5-high interzone progenitors (C1), Sox9-, Acan-high chondroprogenitors (C2), and Mki67-high proliferating cells (C7) (Fig. 5B). Interestingly, skeletal cell clusters expressed different skeletal stem cell markers (Supplementary Fig. 5G), with Hic1-high mesenchyme-1 (C0), Hmgn3-high mesenchyme-3 (C4), Ctsk-, Twist1-, Prrx1-high osteoprogenitors and mesenchyme-2 (C5 and C6), and Sox5-, Sox6-high interzone progenitors and chondroprogenitors (C1 and C2). Pdgfra was highest in mesenchyme-1 (C0), osteoprogenitors (C5), and interzone progenitors (C1) (Supplementary Fig. 5G). From E13.5 to E15.5, Pdgfra expression in control and mutant was increased in whole skeletal clusters and in most individual clusters (Supplementary Fig. 5H, I), but Pdgfrb was maintained or downregulated.

**Fig. 5: Disrupted interzone progenitor populations in *Prrx1-Pdgfra*^K/+ mutant joints.**

When comparing cluster size between Prrx1-Pdgfra^K/+ mutants and controls, mutant interzone progenitor cluster (C1) was less abundant than controls at both E13.5 (−0.10 for differential abundance (DA) score) and E15.5 (−0.18 DA score) (Supplementary Fig. 5J, K), suggesting a loss of interzone progenitors in the mutant. As joint development progresses from E13.5 to E15.5, interzone progenitors should downregulate chondrogenic gene expression (Supplementary Fig. 1). Indeed, E13.5 control interzone progenitors highly expressed chondrogenic collagens (Col2a1, Col9a3), but not fibrogenic collagens (Col1a1, Col3a1). Then, at E15.5, control interzone progenitors downregulated Col2a1 and Col9a3 and upregulated Col1a1 and Col3a1 (Fig. 5C). In sharp contrast, the Prrx1-Pdgfrα^K/+ interzone progenitors did not shift to fibrogenic collagen dominance but maintained high Col2a1 and Col9a3 alongside upregulated Col1a1 and Col3a1 (Fig. 5C). A scatter plot showing relative expression of Col1a1 and Col2a1 in individual interzone progenitors showed a high proportion of co-expressors among E15.5 Prrx1-Pdgfrα^K/+ cells (Fig. 5D). This supports the notion that Prrx1-Pdgfrα^K/+ interzone progenitors are defective in specification and differentiation of joint structures. In agreement with this notion, and with the in situ hybridization at E13.5 (Fig. 3B), Gdf5 expression was downregulated in Prrx1-Pdgfrα^K/+ interzone progenitors at both time points (Fig. 5C).

To more specifically define the cells impacted by PDGFRα^K, we sub-grouped the interzone progenitors (C1) and chondroprogenitors (C2) and re-clustered them to give eight subclusters (SC0–SC7) (Fig. 5E). These clusters included Fzd8-high intermediate chondrogenic progenitors (SC0), Sfrp2-high ligament precursors (SC1), Sox9-high chondrogenic progenitors (SC2), Cxcl12-high perichondral progenitors-1 (SC3), Postn-high perichondral progenitors-2 (SC4), Col1a1-high meniscus progenitors (SC5), an unidentified cell type (SC6), and Creb5-high synovial lining cells (SC7) (Fig. 5F, Supplementary Fig. 6A). By cluster size comparison, mutants had fewer ligament precursors (SC1) than controls (−0.19 DA score at E15.5). Another difference was in the small cluster of synovial lining precursors (SC7) that appeared in the controls at E15.5 but not in Prrx1-Pdgfra^k/+ mutants (−1.00 DA score) (Supplementary Fig. 6B, C). SC7 exhibited a contractile fibroblast-like signature that aligns well with the criteria for synovial lining cells (Supplementary Fig. 6D). Among its highly expressed genes were Tppp3, a marker of synovial lining progenitors as well as tendons and ligaments⁴², and Creb5, a synovial lining marker required for joint development⁴³. Gene ontology analysis identified enrichment for collagen-containing extracellular matrix, muscle contraction, and cytoskeleton (Supplementary Fig. 6E, F). Therefore, loss of synovial lining cells (SC7) is consistent with the joint phenotype in Prrx1-Pdgfra^K/+ mutants.

To infer cell trajectories that may be altered in Prrx1-Pdgfra^K/+ mutants, we performed RNA velocity analysis. At E13.5, control ligament progenitors (SC1) were sourced from chondrogenic progenitors (SC2) via intermediate chondrogenic progenitors (SC0). At E15.5, there was a shift such that control ligament progenitors (SC1) were sourced from perichondral progenitors (SC3) (Fig. 5G). Overall, E15.5 Prrx1-Pdgfra^k/+ mutant trajectories resembled controls, but at E13.5 they were different, with ligament progenitors (SC1) primarily sourced from the two perichondral progenitor clusters (SC3 and SC4) (Fig. 5G). Furthermore, in E15.5 Prrx1-Pdgfra^K/+ mutants, SC4 was increased chondrogenic ECM expression (Col2a1) (Fig. 5H). The transitions inferred by this analysis suggests that PDGFRα^K signaling alters the influx of cells from chondrogenic and perichondral origin.

To better understand changes in SC4, we performed gene set enrichment analysis of differentially expressed genes (DEGs) between mutants and controls. This showed mutant enrichment of anterior-posterior pattern specification, appendage morphogenesis/development, and organ morphogenesis gene sets, but de-enrichment for immune response, amino acids transport, actin filament, and branching morphogenesis (Supplementary Fig. 6G). Among the signaling target gene sets, there was mutant enrichment of BMP2, TP53, RB1, and TGFB1 targets, but de-enrichment of antigen response, metabolic syndrome, RHOA, and CTF21 targets (Supplementary Fig. 6H).

To summarize, scRNA-seq analysis revealed altered cell trajectories contributing to mutant ligament progenitors (SC1) at E13.5 (Fig. 5G), retention of chondrogenic characteristics by mutant interzone progenitors (C1) at E15.5 (Fig. 5C, D), loss of the mutant synovial lining cell cluster (SC7) at E15.5 (Supplementary Fig. 6B, C), and changes in patterning and signaling gene expression in perichondral progenitors (SC4) at E15.5 (Fig. 5H, Supplementary Fig. 6G, H). Although the emergence of different joint structures from a common pool of progenitors is still not fully understood, these analyses point to impaired specification and differentiation of interzone progenitors in Prrx1-Pdgfra^k/+ mutants.

Prrx1-Pdgfrα ^K/+ knee joints develop ectopic fibrocartilage

Spatial information is missing in scRNA-sequencing. To localize gene expression differences in the developing joint tissue, we used NanoString GeoMx (GeoMx) for spatial transcriptomics. This approach provided full gene expression data for specific regions of interest (ROI), with each ROI consisting of 40–50 cells. We used coronal knee sections from two controls and two Prrx1-Pdgfrα^K/+ mutants that were E14.5 littermates. Sections were stained for nuclei and Sox9 to assist in ROI selection. We selected ROIs from three main tissue types from controls and mutants: Sox9-high cartilage from the femur or tibia, Sox9-low interzone, and perichondral border consisting of cells at the interface of cartilage and interzone. For reference, we also selected one ROI of non-skeletal mesenchyme from outside the skeleton of each control and mutant. We also selected a fifth type of ROI from mutants only: Sox9-high interzone representing the ectopic tissue found exclusively in mutants (Fig. 6A). After quality control, there were 179 ROIs (86 controls and 93 mutants, Supplementary Table 2).

**Fig. 6: *Prrx1-Pdgfra*^K/+ knee joints develop ectopic fibrocartilage.**

After sequencing each ROI, normalized expression data of control and mutant were used for principal component analysis (PCA). This showed that control and mutant ROIs grouped separately with minimal overlap (Fig. 6B). Tomato expression served as an internal control for mutant ROIs (Fig. 6C). ROIs from the same tissue type also grouped together (Fig. 6D). Mutant and control cartilage ROIs grouped separately but were still adjacent. Non-skeletal mesenchyme ROIs grouped together regardless of genotype. On the other hand, mutant and control perichondral border ROIs grouped separately, and Sox9-low interzone ROIs were the most distant between genotypes. Therefore, PCA detected genotype-dependent transcriptional differences for the three main tissue types, with the greatest difference found in Sox9-low interzone. For non-skeletal mesenchyme with only one ROI per embryo, the small sample number was probably insufficient for them to cluster separately. Finally, the Sox9-high interzone ROIs from Prrx1-Pdgfrα^K/+ were distributed among the cartilage and perichondral border ROIs rather than among the Sox9-low interzone ROIs (Fig. 6D). UMAP analysis of higher dimension data showed similar patterns as PCA (Supplementary Fig. 7A–C). Taken together, these results suggest that Sox9-high interzone is an ectopic tissue with cartilage-like characteristics instead of interzone tissue.

We examined the expression of joint development genes in the ROIs (Fig. 6E, F). Pdgfra was highest in Sox9-low interzone, intermediate in perichondral border, and lowest in cartilage ROIs as expected. Sox9 showed a graded expression pattern in the opposite direction from Pdgfra. The ectopic Sox9-high interzone tissue, found only in Prrx1-Pdgfrα^K/+, was exceptional for having high expression of both Pdgfra and Sox9 (Fig. 6F). Interzone markers Gdf5 and Sfrp2 were highly expressed in perichondral border and Sox9-low interzone in controls, but their expression was depressed in mutant tissues (Fig. 6E, F). Consistent with scRNA-seq, interzone markers Tppp3, Creb5, and Cd44 were reduced in mutant (Supplementary Fig. 7D, E). Fibrogenic Col1a1 and chondrogenic Col2a1 were slightly decreased in mutant tissues, but in Sox9-high interzone tissue they were co-expressed, much like Pdgfra and Sox9 (Fig. 6F).

To identify DEGs potentially involved in the joint phenotype, we combined the mutant Sox9-low and Sox9-high interzone ROIs and compared them to control interzone ROIs. This revealed 630 upregulated genes and 901 downregulated genes (Fig. 6G, H, Supplementary Data 1). Functional enrichment analysis of upregulated genes, using gene ontology (GO) annotations⁴⁴, highlighted pathways associated with skeletal differentiation, ribosomal protein genes, and protein tyrosine phosphatase receptors (PTPR) (Fig. 6G). GO analysis of downregulated genes showed pathways related to BMP/TGFβ (Gdf5, Gdf6, Tgfb2, Bmp6, Bmpr1a, Ltbp1) and WNT signaling (Fzd6, Lgr5, Dc44, Sfrp2, Wnt2b, Wnt7b, Wnt9a), which are critical for specification of joint progenitors and differentiation of specialized joint cell types (Fig. 6H). Interestingly, specific to mutant Sox9-high interzone ROIs relative to mutant/control Sox9-low interzone, there was upregulation of chondrocyte differentiation, collagen, and a selection of WNT signaling genes (Ctnnb1, Fzd9, Frzb, Lrp5, Lef1, Gpc3, Wnt4) (Supplementary Fig. 7F). The tenocyte markers Htra1, Mkx, and Tnmd were downregulated in the mutant interzone (Supplementary Fig. 7G)^45,46,47.

To further characterize the mutant ectopic Sox9-high interzone, we compared this tissue to other mutant tissues with more normal characteristics. Thus, a volcano plot of the Sox9-high interzone vs Sox9-low interzone DEGs displayed 1241 upregulated genes and 1030 downregulated genes. There was significantly higher expression of chondrogenic genes (Col2a1, Col9a2, Col11a2, Acan, Sox9) in Sox9-high interzone (Fig. 6I, Supplementary Data 2). A volcano plot of the Sox9-high interzone vs cartilage DEGs identified 982 upregulated genes and 453 downregulated genes, including higher expression of fibrogenic genes (Postn, Col3a1, Col1a1, Lox, Pdgfra) in Sox9-high interzone (Fig. 6J, Supplementary Data 3). Therefore, ectopic Sox9-high interzone is a fibrocartilage-like tissue with elevated cartilage genes compared to other interzone tissue and elevated fibrogenic genes compared to other cartilage. In normal joint development, fibrocartilage develops in the mature outer meniscus^48,49 and in the patellar tendon enthesis, which was enlarged in mutant knees (Fig. 2D). Therefore, elevated PDGFRα signaling redirects progenitor cells from early ligament and meniscus fates into a fibrocartilage mass that causes joint fusion.

Altered signaling among interzone progenitor populations

To increase the resolution of the NanoString analysis, we integrated our single-cell and spatial transcriptomic data using SpatialDecon. This algorithm assigns cell identities, based on scRNA-seq clusters, to the sequences generated for each NanoString-defined ROI⁵⁰. Heat maps of the integrated control and Prrx1-Pdgfrα^K/+ data are shown in Fig. 7A, with the eight scRNA-seq clusters (C0–C7) arranged along the y-axis and the 175 NanoString ROIs arranged along the x-axis. The four mesenchyme ROIs were excluded due to limited sample numbers. More condensed visualization of the same data is shown as bar graphs with the proportion of each scRNA-seq cluster within each tissue-type ROI (Fig. 7B).

**Fig. 7: Spatiotemporally impaired interzone-resident cells and signaling in *Prrx1-Pdgfra*^K/+ mutant joints.**

These displays show that control cartilage ROIs were predominantly composed of chondroprogenitors (C2), interzone progenitors (C1), and proliferating cells (C7), with a small representation of mesenchyme-2 (C6). Control perichondral border ROIs were composed of interzone progenitors (C1), chondroprogenitors (C2), and proliferating cells (C7), with small representation of mesenchyme-2 (C6) and osteoprogenitors (C5). Control Sox9-low interzone ROIs were dominated by interzone progenitors (C1) but were also the most heterogeneous with proliferating cells (C7), chondroprogenitors (C2), osteoprogenitors (C5), and minor populations of mesenchyme 1–3 (C0, C6, C4). Compared to control tissues, all mutant tissues included a large increase in mesenchyme-2 (C6), characterized as Prrx1-, Twist1-, Ctsk-, Postn-, Col1a1-high (Fig. 5B, Supplementary Fig. 5G). Whereas mesenchyme-2 (C6) represented less than 1% of any control tissue, this cluster represented between 7.9% and 13.8% of each mutant tissue. Mesenchyme-1 (C0), characterized as Pdgfra-, Hic1-, Meis2-, Cxcl12-high, was also increased in mutant perichondral border and Sox9-low interzone tissues.

The mutant Sox9-high interzone ROIs were the most heterogeneous (Fig. 7B). Hierarchical clustering segregated these ROIs into two main groups (Supplementary Fig. 8A). One group was predominantly interzone progenitor (C1), and the other was composed of mainly chondroprogenitor (C2), osteoprogenitor (C5) and mesenchyme-2 (C6). This heterogeneity likely reflects the disorganized character of the progenitors that give rise to ectopic fibrocartilage.

To infer signals in interzone progenitors and neighboring cells that could contribute to the Prrx1-Pdgfrα^K/+ joint phenotype, we performed CellChat. This approach uses a curated iterative list of ligand-receptor complexes and multiple co-factors to identify cell-cell communication networks between scRNA-seq clusters⁵¹. To identify interzone-resident cell-specific interaction signals, we specified interzone progenitor (C1), chondroprogenitor (C2), and mesenchyme-2 (C6) clusters as source and target cells. In the mutant, midkine (MK, heparin binding growth factor) signaling was increasingly delivered from C6 to C1 and C2, and from C1 to C6, as well as in autocrine manner from C6 to itself (Fig. 7C, Supplementary Fig. 8B). In addition, collagen signaling was increased from C1 and C2 to C6 (Fig. 7C, Supplementary Fig. 8C). Signaling pathways that were absent in control but emerged in mutant interzone-resident clusters included GAP, LAMININ, NOTCH, and noncanonical WNT (ncWNT) (Fig. 7C, D).

The CellChat database includes joint development signaling networks such as TGFβ, BMP, and WNT, but analysis only identified differences between mutants and controls at E15.5 (Supplementary Fig. 8D). Therefore, we searched for individual TGFβ, BMP, and WNT DEGs in the C1, C2, and C6 clusters at E13.5 and E15.5. The BMP inhibitor Nog and the receptor Bmpr1b were downregulated in control interzone progenitors (C1) from E13.5 to E15.5, but both were missing from mutant at E13.5 and Bmpr1b was upregulated at E15.5 (Supplementary Fig. 8E). Wnt4 and the WNT inhibitor Ndrg2 were downregulated in control C1 from E13.5 to E15.5, but Wnt4 was upregulated in mutant C1 and Ndgr2 was missing (Supplementary Fig. 8F). On the other hand, Wnt2 was upregulated in control mesenchyme-2 (C6) from E13.5 to E15.5, but it was missing in mutant (Supplementary Fig. 8G).

Taken together, the integrated analysis of spatial transcriptomics and scRNA-seq suggests expansion of mesenchyme-2 (C6) in the interzone due to elevated PDGFRα signaling. We inferred disturbed MK, collagen, gap junction, cell-cell adhesion, NOTCH, and noncanonical WNT signaling at E13.5, and specific TGFβ, BMP, and WNT signaling genes were affected between E13.5 and E15.5. BMP signaling promotes chondrogenesis and is essential for joint development, but it must be carefully regulated by antagonists like Nog to prevent aberrant cartilage formation. WNT signaling also modulates BMP to maintain interzone progenitors. These signaling changes suggest that elevated PDGFRα signaling disrupts the interzone microenvironment, impairing the maintenance of the interzone progenitor pool, heterogeneity, and cell-cell interaction, ultimately skewing cell specification and maturation.

Discussion

PDGFRα expression has been documented in the developing limb skeleton and connective tissue, but whether PDGFRα signaling regulates any aspect of limb development is unknown. We show that PDGFRα-lineage cells give rise to all skeletal elements of the knee joint. Then, PDGFRα expression is downregulated in the mesenchymal cells that condense into cartilage and express Sox9, but it continues to be expressed in the perichondrium. PDGFRα is re-expressed in condensed mesenchymal cells when they downregulate Sox9 to form the interzone. This creates a dynamic and largely reciprocal expression pattern of the two factors: Sox9⁺ cartilage condensations form within PDGFRα⁺ mesenchyme, and then PDGFRα⁺ joint progenitors form within Sox9⁺ cartilage condensations. Co-expression of Sox9 and PDGFRα does occur, however, in some joint progenitors between E13.5 and E14.5 and possibly earlier. By analyzing embryos with a D842V mutation that generates constitutive PDGFRα signaling, we provide evidence that PDGFRα regulates the differentiation of joint progenitors between chondrogenic and fibrogenic fates, which is crucial for proper joint development. These mutant embryos exhibit a loss of cruciate ligaments, defective perichondral border formation, and a gain of Sox9-high fibrocartilage within the developing joint. Our sequencing results suggest that elevated PDGFRα signaling perturbs the fine-tuning of BMP, Wnt, and ECM signals, leading to an expansion of fibrocartilage differentiation at the expense of specialized synovial joint tissues.

The earliest evidence of defective joint development in Prrx1-Pdgfra^K/+ embryos was poor expression of Gdf5 at E13.5, which marks the specified joint anlage within the Sox9⁺ cartilage condensation¹³. However, Gdf5 expression was normal at E12.5, suggesting that elevated PDGFRα signaling disrupts the maintenance of specified joint progenitors rather than a defect in initial specification. This was supported by the loss of interzone progenitors when analyzed by scRNA-seq. The mutant interzone failed to differentiate Creb5-, Tppp3-, and Cd44-high synovial lining progenitors and ligament progenitors were reduced. Spatial transcriptomics also showed that mutant perichondral border tissue had lost the gene signature of synovial lining cells. The functional importance of several of these genes has been recently demonstrated in mice. Creb5 mutants exhibit joint fusion and fail to form articular cartilage⁴³. Deletion of Has2, which produces hyaluronic acid and binds to CD44, leads to delayed interzone cavitation¹². Taken together, our results suggest that PDGFRα signaling plays a role in establishing or maintaining the border function of the perichondrium, which restricts pro-chondrogenic factors to the cartilage template.

As a result of elevated PDGFRα signaling there were several joint anomalies by E14.5, including a poorly defined perichondrium and increased co-expression of Sox9 and PDGFRα in spaces where the articular cartilage and patellar enthesis should form. Ectopic cartilage subsequently formed in the joint space as well as in the perichondrium around the enthesis and articular cartilage, suggesting a defect in the perichondral border function. Ectopic cartilage and joint fusions have been uncovered in embryos deficient for Ext1, which encodes an enzyme that produces heparan sulfate in the extracellular matrix to modulate the availability of BMP⁵².

How does elevated PDGFRα signaling lead to ectopic fibrocartilage? It has been proposed that chondrogenic vs non-chondrogenic progenitors determine interzone specification and patterning⁵³. It is possible that a certain ratio of cartilage-lineage to non-cartilage-lineage interzone progenitors is required for proper synovial tissue development. The non-cartilage-lineage cells could have a greater propensity to differentiate into fibrous terminal tissue types. Our spatial transcriptomics data indicated that fibrous cell types (i.e., Mesenchyme-2 (C6), Mesenchyme-1 (C0), and osteoprogenitor (C5)) were expanded and distributed widely throughout the mutant joint space. Thus, we found upregulated chondrogenesis factors such as Sox9, Col2a1, and Acan, and downregulated anti-chondrogenic factors such as Nog and Wnt9a in the mutant interzone^7,54. Our results suggest that elevated PDGFRα signaling perturbs the balance of chondrogenic and anti-chondrogenic signaling, leading to fibrocartilage differentiation and expansion at the expense of specialized synovial joint tissues. This is reminiscent of our previous finding that elevated PDGFRα signaling disrupts the balance of fibrous connective tissue cells and adipogenic cells during fat development³³.

Interestingly, early molecular changes in the mutant interzone seem to be related to structural proteins, such as collagens, cytoskeleton, adhesion, and junctional proteins, rather than extracellular growth factors and ligands. The mutant Prrx1-Pdgfra^K/+ interzone may undergo microenvironmental perturbations that disrupt regional heterogeneity and juxtacrine communication. For example, ECM remodeling and mechanical signaling can regulate stem cell fate and development^55,56. In the current study, the mutant interzone maintained the expression of chondrogenic genes and matrix proteins, which were normally discontinued in wild type.

From RNA velocity analysis we inferred a shift in populations that contribute to the mutant interzone at E13.5. Therefore, our findings suggest that PDGFRα regulates chondrogenic and fibrogenic cell fate balance at the developing synovial joint. This agrees with previous work in cranial development using the same PDGFRα-D849V mutation, where elevated PDGFRα signaling caused ectopic cranial chondrogenesis via a loss of Wnt9a signaling⁵⁷. In the current study, spatial transcriptomics at E14.5 also found downregulation of WNT signaling, including reduced Wnt9a. However, additional work is needed to establish this as a cause of joint defects in Prrx1-Pdgfra^K/+ mice.

We found that elevated PDGFRα signaling specifically affected the knee, hip, and shoulder joints. Most genetic mutations that cause joint fusions do not cause global joint phenotypes^{20,21,58,59,60,61,62}. The reason for joint-selective phenotypes, however, is unclear. Furthermore, many joint fusions described in the literature involve the whole interzone because cavitation does not occur. For example, Seo and Serra showed that mouse embryos with Prrx1-Cre mediated deletion of Tgfbr2 were able to form the interzone in the phalanges, but they were not maintained²⁰. Instead, interzone cells were replaced with cartilage. Furthermore, Spater et al. showed that disruption of Wnt signaling, through loss of either β-catenin or Wnt9a and Wnt4, caused fusions of some carpal bones as well as chondroid metaplasia within the knees and elbows⁵⁹. They concluded that Wnt signaling is not required for interzone specification, but it supports interzone maintenance and terminal joint tissue differentiation. Our model is unique because the Prrx1-Pdgfra^K/+ knee joint underwent cavitation in the center but retained fibrocartilaginous bridges of tissue on the lateral aspects. The resulting phenotype highlights the need for tightly regulated tyrosine kinase signaling during joint formation.

Somewhat surprisingly, PDGFRα signaling was not required for limb or joint development despite its high expression in the early limb bud. It is possible that the essential signaling pathways regulated by PDGFRα are compensated by other receptor tyrosine kinases in limb mesenchyme, including those of the fibroblasts growth factor (FGF), insulin-like growth factor (IGF), discoidin domain receptor (DDR), and Eph receptor families.

Methods

Animal models

All procedures performed on mice were approved by the Institutional Animal Care and Use Committee (IACUC) of the Oklahoma Medical Research Foundation. Mouse strains Pdgfra^k.Flp;R26-fsf-Tomato and PDGFRa^Flp;R26-fsf-Tomato were made by the Olson lab. Mouses strain Pdgfra^k.Flp;R26-fsf-Tomato is available at the Jackson Laboratory (JAX#038627). Mouse strains R26-LSL-Tomato (JAX#007914), Prrx1-Cre (JAX#033173), Gdf5-Cre (JAX#031327), and Sox9-CreER (JAX#035092) were obtained from the Jackson Laboratory. Mice were maintained on a mixed C57BL/6;129 genetic background. The mice were fed a standard mouse chow diet (5053 Purina), and were maintained at 20–25 °C with a 12 h light/dark cycle. Mutant Pdgfra^K.Flp and Pdgfra^Flp/Flp mice were compared with age-matched littermate controls. Due to embryonic time points, sex was not considered in the study design and analysis. For timed pregnancies, females were added to the male’s cage and checked daily, before 11:00 AM, for plugs. Plugged females were removed from the breeder cage, and this was considered embryonic day 0.5. Tamoxifen in corn oil (50 mg/kg body weight) was administered via intraperitoneal injection for inducible Cres.

Tissue histology and immunostaining

Tissue was fixed overnight in 10% neutral buffered formalin or 4% paraformaldehyde. Postnatal bones were decalcified in 0.5 M EDTA solution (pH 7.4) for 1 week at 4 °C. For histology, fixed tissues were embedded in paraffin and sectioned at 8 µm thickness followed by Masson’s trichrome staining (Hematoxylin, Fuchsin, and Analine Blue) or Safranin O staining (Fast Green and Hematoxylin as counter stain). For immunostaining, fixed tissues were cryosectioned at 8–12 µm. Slides were stained with primary antibody for Sox9 (Millipore AB5535, 1:500 dilution) or PDGFRα (R&D AF1062, 1:250 dilution), conjugated secondary antibody (Jackson ImmunoResearch), and DAPI (Sigma) as a nuclear stain. Imaging was performed with either a Nikon Eclipse 80i microscope with a digital camera and NIS-Elements D 3.22 software or Nikon AX R confocal microscope system with NIS-Elements AR 6.10.01 software. ImageJ (version 2.14) was used for microscopic image quantification and Prism 10 was used for statistical analysis.

In situ hybridization by hybridization chain reaction (HCR)

Gdf5 probe, HCR amplifiers, and buffers were purchased from Molecular Instruments. For whole mount, embryos were removed from the yolk sac in PBS and fixed overnight with 10% neutral buffered formalin (NBF, Sigma-Aldrich). The embryos were then processed as per the whole mount mouse protocol at www.molecularinstruments.com. In short, samples were dehydrated through graded MeOH/PBST washes on ice and stored in 100% MeOH at −20 °C until use (at least 16 h). Samples were rehydrated to 100% PBST, immersed in 10 μg/mL proteinase K, postfixed in 5% PFA, and pre-hybridized for 30 min at 37 °C. Samples were then incubated in 2 pmol of probe set in probe hybridization buffer overnight at 37 °C. Following successive probe wash buffer washes and 5× SSCT washes, samples were incubated overnight at room temperature in amplification buffer containing 30 pmol of hairpins h1 and h2. Samples were then washed and stored in 5× SSCT. For clearing, samples were dehydrated 25%, 50%, 75%, and 100% methanol in PBS for 15 min each at room temp. Samples were then moved to 100% methanol overnight. Samples were cleared in freshly prepared BABB (1:2 v/v mixture of benzyl alcohol (Sigma, #305197) and benzyl benzoate (Sigma, #B6630)) at 4 °C for overnight.

Single-cell RNA sequencing and annotation

Previously published single-cell data was downloaded from NCBI’s Gene Expression Omnibus (GEO) from the following accession number: GSE151985¹³. The data was normalized and integrated to correct for batch effect. Processing was performed similarly to the protocol below. For our in-house data, tissue samples were harvested using a modified version of a previously described protocol^13,63. At E13.5, hindlimbs were harvested in ice-cold PBS and the paws were removed. The samples were digested in 0.25% trypsin-EDTA for 1 h at 37 °C. For E15.5, knees were dissected in ice-cold PBS and digested in 3 mg/mL collagenase D in DMEM for 45 min at 37 °C. They were then digested in 1 mg/mL collagenase D in DMEM for 3 h at 37 °C. Following digestion, samples from both time points were gently dissociated by pipetting, filtered through a 40 μm strainer, and centrifuged for 10 min at 400 × g. The cells were resuspended in 0.4% FBS in PBS at a concentration of 1000/μL.

Samples (maximum 10,000 cells per sample) were loaded on the 10X Genomics Chromium platform, barcoded, and sequenced on the Illumina NovaSeq SP flowcell platform at a depth of 500 million reads per sample. See Supplementary Table 1 for batch information and actual cell numbers sequenced and analyzed. The 10X Genomics Cellranger software (version 6.0.2) pipeline was run for demultiplexing and sequence alignment to the murine mm10 genome reference. Quality control, data integration, clustering, and gene expression analysis were performed with Seurat 4.4.0 for R (version 4.41)^64,65. Samples were individually filtered to remove cells with low gene numbers. This was done because 2 of the E13.5 samples (1 control and 1 mutant in Batch 3) required a lower cutoff. For these samples, cells were removed that contained <250 genes. For the remaining samples, cells were removed that contained <1500 genes. The remaining quality control was performed equally across the samples. Cells with >8000 genes or >8% mitochondrial-associated genes were removed. Cells were then categorized into cell-cycle phases based on cell-cycle gene expression⁶⁶. Samples were individually normalized using Seurat’s SCTransform with cell-cycle regression⁶⁷. Normalized samples were then integrated to remove batch effect using Seurat’s IntegrateData function.

Seurat’s SNN algorithm generated 13 clusters. For skeletal cluster analysis, non-skeletal clusters were excluded, and the remaining clusters were re-clustered with a resolution of 0.3 using Clustree R package⁶⁸, resulting in 8 clusters (C0–C7). Chondrogenic/interzone subclusters were generated by subsetting chondroprogenitors (C2) and interzone progenitors (C1) and regrouping them as eight subclusters (SC0–SC7) with a cluster resolution of 0.5 using Clustree. Seurat’s FindAllMarkers determined unique genes of both main skeletal clusters and chondrogenic/interzone subclusters and their tentative cell types. Figures were generated by Seurat functions (DimPlot for UMAP for clusters, DotPlot for dot plots, VlnPlot for violin plots, FeaturePlot for UMAP for individual gene expression, and FeatureScatter for visualizing for two genes’ expression) and scCustomize functions (Stacked_VlnPlot and FeaturePlot_scCustom) (https://doi.org/10.5281/zenodo.14529706). RNA velocity analysis was performed by focusing on the chondrogenic subclusters using scVelo⁶⁹. Downstream analysis of DEGs was performed using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) knowledgebase v2024q4 with gene ontology and KEGG pathway annotation and gene set enrichment analysis (GSEA) version 4.3.3 with MSigDB mouse M2 (curated gene sets) and M5 (gene ontology) database^70,71. Cell-cell interaction analysis was conducted separately for E13.5 and E15.5 using normalized data matrix of scRNA-seq clusters (C0 through C7) and CellChat (version 2)⁵¹. CellChat’s rankNet function was used to find informational flow between C1, C2, and C6. CellChat’s netVisual_aggregate function with “circle” layout parameter was used to visualize differentially regulated signaling between scRNA-seq clusters (cell types) by genotype. Differential abundance analysis was performed using DA-seq⁷².

Spatial transcriptomics and spatial deconvolution

Tissues were fixed in 10% NBF, embedded in paraffin, and sectioned at 5 μm. For GeoMx WTA profiling, four slides containing multiple sections of either Prrx1-CreER or Prrx1-Pdgfra^K/+ tissue were generated and processed according to protocols listed in the GeoMx DSP Manual Slide Preparation guide (MAN-10150-01). Slides were deparaffinized and rehydrated, underwent antigen retrieval and Proteinase K digestion, then were hybridized overnight with UV-photocleavable, barcoded oligonucleotide probes from the Mouse Whole Transcriptome Atlas (WTA) RNA probe set. After washing, slides were incubated with Sox9 (Millipore AB5535, 1:500) and SYTOX13 (NanoString) to allow for visualization of tissue morphology for ROI selection. Slides were placed in the GeoMx Digital Spatial Profiler instrument and ROIs were selected. These selected areas then underwent barcode cleavage from the hybridized probes, with the barcodes collected into individual wells of the collection plate based on ROI of origination. PCR was used to uniquely index each ROI’s barcodes with specific Illumina i5/i7 dual-indexing primers. These library products were then pooled, purified, and assessed for quality. Paired-end sequencing was performed on a NextSeq 2K P1 PE50 (Illumina), targeting the NanoString recommended read depth of 86 million read-pairs (based on ROI summed areas). Data was then transferred back to the GeoMx DSP Control Center software (version 2.4.0.421) for further processing and analysis. 179 ROIs (86 control and 93 mutant) were selected quality control (Supplementary Table 2) and normalized using Nanostring GeoMx Q3 normalization. For genotype comparison between Prrx1-Pdgfrα^+/+ and Prrx1-Pdgfrα^K/+, a Seurat object was created with normalized data and ROI annotation metadata of 179 ROIs. Principle component analysis (PCA) was performed and ROIs were plotted along PC1 and PC2. The top 30 PCs were selected to performed UMAP. By doing this we could match each ROI with their associated tags (i.e., genotypes and spatial ROIs). Figures were generated by Seurat functions (DimPlot for PCA and UMAP for clusters, VlnPlot, and FeaturePlot for PCA and UMAP for individual gene expression). DEGs were determined by the comparison of spatial ROI types of interest, using a linear mixed model with Benjamini-Hochberg correction, with the following cutoffs: |Log2FC| ≥ 0.25 and adjusted p-value ≤ 0.05. Volcano plots were visualized with highlighting DEGs based on the cutoffs. Heat maps were generated with genes selected from DEGs after DAVID analysis with gene ontology and KEGG pathway, using pheatmap package in R (https://github.com/raivokolde/pheatmap). Spatial deconvolution was performed using NanoString’s SpatialDecon-plugin⁵⁰. A cell profile matrix was made using our scRNA-seq data (C0 through C7 clusters of both E13.5 and E15.5 time points) and uploaded to the GeoMx DSP instrument to compare with 175 NanoString ROIs (4 mesenchyme ROIs were excluded). Heat maps were used to visualize spatial deconvolution results by genotype or for selected ROIs, using pheatmap package. We summarized the proportion of each scRNA-seq cluster within each tissue-type ROI, presented as bar graphs using Prism 10 (Error bars with standard error of the mean (SEM)).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Source data are provided with this paper. Data supporting the findings of this study are available in the article, its Supplementary Information, the source data file, and from the corresponding author upon request. Single-cell RNA sequencing data and NanoString GeoMx data are available at the NCBI Gene Expression Omnibus under accession numbers GSE289874 and GSE289873. Source data are provided with this paper.

References

Fallon, E. A. et al. Prevalence of diagnosed arthritis—United States, 2019-2021. MMWR Morb. Mortal. Wkly. Rep. 72, 1101–1107 (2023).
Article PubMed PubMed Central Google Scholar
Loeser, R. F., Goldring, S. R., Scanzello, C. R. & Goldring, M. B. Osteoarthritis: a disease of the joint as an organ. Arthritis Rheum. 64, 1697–1707 (2012).
Article PubMed PubMed Central Google Scholar
Fox, A. J., Wanivenhaus, F., Burge, A. J., Warren, R. F. & Rodeo, S. A. The human meniscus: a review of anatomy, function, injury, and advances in treatment. Clin. Anat. 28, 269–287 (2015).
Article PubMed Google Scholar
Goldring, S. R. Pathogenesis of bone and cartilage destruction in rheumatoid arthritis. Rheumatology 42, 11–16 (2003).
Article Google Scholar
Lefebvre, V. & de Crombrugghe, B. Toward understanding SOX9 function in chondrocyte differentiation. Matrix Biol. 16, 529–540 (1998).
Article PubMed CAS Google Scholar
Akiyama, H. et al. Osteo-chondroprogenitor cells are derived from Sox9 expressing precursors. Proc. Natl. Acad. Sci. USA 102, 14665–14670 (2005).
Article ADS PubMed CAS Google Scholar
Hartmann, C. & Tabin, C. J. Wnt-14 plays a pivotal role in inducing synovial joint formation in the developing appendicular skeleton. Cell 104, 341–351 (2001).
Article PubMed CAS Google Scholar
Storm, E. E. et al. Limb alterations in brachypodism mice due to mutations in a new member of the TGF beta-superfamily. Nature 368, 639–643 (1994).
Article ADS PubMed CAS Google Scholar
Chen, H. et al. Heads, shoulders, elbows, knees, and toes: modular Gdf5 enhancers control different joints in the vertebrate skeleton. PLoS Genet. 12, e1006454 (2016).
Article PubMed PubMed Central Google Scholar
Rux, D., Decker, R. S., Koyama, E. & Pacifici, M. Joints in the appendicular skeleton: developmental mechanisms and evolutionary influences. Curr. Top. Dev. Biol. 133, 119–151 (2019).
Article PubMed CAS Google Scholar
Shwartz, Y., Viukov, S., Krief, S. & Zelzer, E. Joint development involves a continuous influx of Gdf5-positive cells. Cell Rep. 15, 2577–2587 (2016).
Article PubMed PubMed Central CAS Google Scholar
Kim, M. et al. Synovial joint cavitation initiates with microcavities in interzone and is coupled to skeletal flexion and elongation in developing mouse embryo limbs. Biol. Open 11, https://doi.org/10.1242/bio.059381 (2022).
Bian, Q. et al. A single cell transcriptional atlas of early synovial joint development. Development 147, https://doi.org/10.1242/dev.185777 (2020).
Brunet, L. J., McMahon, J. A., McMahon, A. P. & Harland, R. M. Noggin, cartilage morphogenesis, and joint formation in the mammalian skeleton. Science 280, 1455–1457 (1998).
Article ADS PubMed CAS Google Scholar
Gong, Y. et al. Heterozygous mutations in the gene encoding noggin affect human joint morphogenesis. Nat. Genet. 21, 302–304 (1999).
Article PubMed CAS Google Scholar
Seemann, P. et al. Activating and deactivating mutations in the receptor interaction site of GDF5 cause symphalangism or brachydactyly type A2. J. Clin. Investig. 115, 2373–2381 (2005).
Article PubMed PubMed Central CAS Google Scholar
Dawson, K. et al. GDF5 is a second locus for multiple-synostosis syndrome. Am. J. Hum. Genet. 78, 708–712 (2006).
Article PubMed PubMed Central CAS Google Scholar
Storm, E. E. & Kingsley, D. M. Joint patterning defects caused by single and double mutations in members of the bone morphogenetic protein (BMP) family. Development 122, 3969–3979 (1996).
Article PubMed CAS Google Scholar
Pregizer, S. K. et al. Impact of broad regulatory regions on Gdf5 expression and function in knee development and susceptibility to osteoarthritis. Ann. Rheum. Dis. 77, 450 (2018).
Article PubMed CAS Google Scholar
Seo, H. S. & Serra, R. Deletion of Tgfbr2 in Prx1-cre expressing mesenchyme results in defects in development of the long bones and joints. Dev. Biol. 310, 304–316 (2007).
Article PubMed PubMed Central CAS Google Scholar
Spagnoli, A. et al. TGF-beta signaling is essential for joint morphogenesis. J. Cell Biol. 177, 1105–1117 (2007).
Article PubMed PubMed Central CAS Google Scholar
Lyons, K. M. & Rosen, V. BMPs, TGFbeta, and border security at the interzone. Curr. Top. Dev. Biol. 133, 153–170 (2019).
Article PubMed CAS Google Scholar
Hoch, R. V. & Soriano, P. Roles of PDGF in animal development. Development 130, 4769–4784 (2003).
Article PubMed CAS Google Scholar
Tallquist, M. D. & Soriano, P. Cell autonomous requirement for PDGFRalpha in populations of cranial and cardiac neural crest cells. Development 130, 507–518 (2003).
Article PubMed CAS Google Scholar
Soriano, P. The PDGF alpha receptor is required for neural crest cell development and for normal patterning of the somites. Development 124, 2691–2700 (1997).
Article PubMed CAS Google Scholar
Tallquist, M. D., Weismann, K. E., Hellstrom, M. & Soriano, P. Early myotome specification regulates PDGFA expression and axial skeleton development. Development 127, 5059–5070 (2000).
Article PubMed CAS Google Scholar
Fantauzzo, K. A. & Soriano, P. PI3K-mediated PDGFRalpha signaling regulates survival and proliferation in skeletal development through p53-dependent intracellular pathways. Genes Dev. 28, 1005–1017 (2014).
Article PubMed PubMed Central CAS Google Scholar
Pickett, E. A., Olsen, G. S. & Tallquist, M. D. Disruption of PDGFRalpha-initiated PI3K activation and migration of somite derivatives leads to spina bifida. Development 135, 589–598 (2008).
Article PubMed CAS Google Scholar
Olson, L. E. & Soriano, P. Increased PDGFRalpha activation disrupts connective tissue development and drives systemic fibrosis. Dev. Cell 16, 303–313 (2009).
Article PubMed PubMed Central CAS Google Scholar
Moenning, A. et al. Sustained platelet-derived growth factor receptor alpha signaling in osteoblasts results in craniosynostosis by overactivating the phospholipase C-gamma pathway. Mol. Cell. Biol. 29, 881–891 (2009).
Article PubMed CAS Google Scholar
He, F. & Soriano, P. Dysregulated PDGFRalpha signaling alters coronal suture morphogenesis and leads to craniosynostosis through endochondral ossification. Development 144, 4026–4036 (2017).
PubMed PubMed Central CAS Google Scholar
Iwayama, T. et al. PDGFRalpha signaling drives adipose tissue fibrosis by targeting progenitor cell plasticity. Genes Dev. 29, 1106–1119 (2015).
Article PubMed PubMed Central CAS Google Scholar
Sun, C., Berry, W. L. & Olson, L. E. PDGFRalpha controls the balance of stromal and adipogenic cells during adipose tissue organogenesis. Development 144, 83–94 (2017).
Article PubMed PubMed Central CAS Google Scholar
Sun, C. et al. Mosaic mutant analysis identifies PDGFRalpha/PDGFRbeta as negative regulators of adipogenesis. Cell Stem Cell 26, 707–721.e705 (2020).
Article PubMed PubMed Central CAS Google Scholar
Guerit, E., Arts, F., Dachy, G., Boulouadnine, B. & Demoulin, J. B. PDGF receptor mutations in human diseases. Cell Mol. Life Sci. 78, 3867–3881 (2021).
Article PubMed PubMed Central CAS Google Scholar
Feng, C. et al. Lgr5 and Col22a1 mark progenitor cells in the lineage toward juvenile articular chondrocytes. Stem Cell Rep. 13, 713–729 (2019).
Article CAS Google Scholar
Markman, S. et al. A single-cell census of mouse limb development identifies complex spatiotemporal dynamics of skeleton formation. Dev. Cell 58, 565–581.e564 (2023).
Article PubMed CAS Google Scholar
Logan, M. et al. Expression of Cre Recombinase in the developing mouse limb bud driven by a Prxl enhancer. Genesis 33, 77–80 (2002).
Article PubMed CAS Google Scholar
Rezza, A., Sennett, R., Tanguy, M., Clavel, C. & Rendl, M. PDGF signalling in the dermis and in dermal condensates is dispensable for hair follicle induction and formation. Exp. Dermatol. 24, 468–470 (2015).
Article PubMed PubMed Central CAS Google Scholar
Rountree, R. B. et al. BMP receptor signaling is required for postnatal maintenance of articular cartilage. PLoS Biol. 2, e355 (2004).
Article PubMed PubMed Central Google Scholar
Xu, Z. et al. Embryonic attenuated Wnt/beta-catenin signaling defines niche location and long-term stem cell fate in hair follicle. eLife 4, e10567 (2015).
Article PubMed PubMed Central Google Scholar
Harvey, T., Flamenco, S. & Fan, C. M. A Tppp3(+)Pdgfra(+) tendon stem cell population contributes to regeneration and reveals a shared role for PDGF signalling in regeneration and fibrosis. Nat. Cell Biol. 21, 1490–1503 (2019).
Article PubMed PubMed Central CAS Google Scholar
Zhang, C. H. et al. Creb5 coordinates synovial joint formation with the genesis of articular cartilage. Nat. Commun. 13, 7295 (2022).
Article ADS PubMed PubMed Central CAS Google Scholar
Wu, T. et al. clusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation 2, 100141 (2021).
PubMed PubMed Central CAS Google Scholar
Havis, E. et al. Transcriptomic analysis of mouse limb tendon cells during development. Development 141, 3683–3696 (2014).
Article PubMed CAS Google Scholar
Ito, Y. et al. The Mohawk homeobox gene is a critical regulator of tendon differentiation. Proc. Natl. Acad. Sci. USA 107, 10538–10542 (2010).
Article ADS PubMed PubMed Central CAS Google Scholar
Asahara, H., Inui, M. & Lotz, M. K. Tendons and ligaments: connecting developmental biology to musculoskeletal disease pathogenesis. J. Bone Miner. Res. 32, 1773–1782 (2017).
Article PubMed Google Scholar
Makris, E. A., Hadidi, P. & Athanasiou, K. A. The knee meniscus: structure-function, pathophysiology, current repair techniques, and prospects for regeneration. Biomaterials 32, 7411–7431 (2011).
Article PubMed PubMed Central CAS Google Scholar
Gamer, L. W., Xiang, L. & Rosen, V. Formation and maturation of the murine meniscus. J. Orthop. Res. 35, 1683–1689 (2017).
Article PubMed CAS Google Scholar
Danaher, P. et al. Advances in mixed cell deconvolution enable quantification of cell types in spatial transcriptomic data. Nat. Commun. 13, 385 (2022).
Article ADS PubMed PubMed Central CAS Google Scholar
Jin, S. et al. Inference and analysis of cell-cell communication using CellChat. Nat. Commun. 12, 1088 (2021).
Article ADS PubMed PubMed Central CAS Google Scholar
Huegel, J. et al. Perichondrium phenotype and border function are regulated by Ext1 and heparan sulfate in developing long bones: a mechanism likely deranged in Hereditary Multiple Exostoses. Dev. Biol. 377, 100–112 (2013).
Article PubMed PubMed Central CAS Google Scholar
Koyama, E. et al. A distinct cohort of progenitor cells participates in synovial joint and articular cartilage formation during mouse limb skeletogenesis. Dev. Biol. 316, 62–73 (2008).
Article PubMed PubMed Central CAS Google Scholar
Nifuji, A., Kellermann, O. & Noda, M. Noggin inhibits chondrogenic but not osteogenic differentiation in mesodermal stem cell line C1 and skeletal cells. Endocrinology 145, 3434–3442 (2004).
Article PubMed CAS Google Scholar
Bonnans, C., Chou, J. & Werb, Z. Remodelling the extracellular matrix in development and disease. Nat. Rev. Mol. Cell Biol. 15, 786–801 (2014).
Article PubMed PubMed Central CAS Google Scholar
De Belly, H., Paluch, E. K. & Chalut, K. J. Interplay between mechanics and signalling in regulating cell fate. Nat. Rev. Mol. Cell Biol. 23, 465–480 (2022).
Article PubMed Google Scholar
Bartoletti, G., Dong, C., Umar, M. & He, F. Pdgfra regulates multipotent cell differentiation towards chondrocytes via inhibiting Wnt9a/beta-catenin pathway during chondrocranial cartilage development. Dev. Biol. 466, 36–46 (2020).
Article PubMed PubMed Central CAS Google Scholar
Tsumaki, N. et al. Bone morphogenetic protein signals are required for cartilage formation and differently regulate joint development during skeletogenesis. J. Bone Miner. Res. 17, 898–906 (2002).
Article PubMed CAS Google Scholar
Spater, D., Hill, T. P., Gruber, M. & Hartmann, C. Role of canonical Wnt-signalling in joint formation. Eur. Cell Mater. 12, 71–80 (2006).
Article PubMed CAS Google Scholar
Kahn, J. et al. Muscle contraction is necessary to maintain joint progenitor cell fate. Dev. Cell 16, 734–743 (2009).
Article PubMed CAS Google Scholar
Shim, J. H. et al. TAK1 is an essential regulator of BMP signalling in cartilage. EMBO J. 28, 2028–2041 (2009).
Article PubMed PubMed Central CAS Google Scholar
Gunnell, L. M. et al. TAK1 regulates cartilage and joint development via the MAPK and BMP signaling pathways. J. Bone Miner. Res. 25, 1784–1797 (2010).
Article PubMed PubMed Central CAS Google Scholar
Gosset, M., Berenbaum, F., Thirion, S. & Jacques, C. Primary culture and phenotyping of murine chondrocytes. Nat. Protoc. 3, 1253–1260 (2008).
Article PubMed CAS Google Scholar
Stuart, T. et al. Comprehensive integration of single-cell data. Cell 177, 1888–1902.e1821 (2019).
Article PubMed PubMed Central CAS Google Scholar
Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).
Article PubMed PubMed Central CAS Google Scholar
Kowalczyk, M. S. et al. Single-cell RNA-seq reveals changes in cell cycle and differentiation programs upon aging of hematopoietic stem cells. Genome Res. 25, 1860–1872 (2015).
Article PubMed PubMed Central CAS Google Scholar
Choudhary, S. & Satija, R. Comparison and evaluation of statistical error models for scRNA-seq. Genome Biol. 23, 27 (2022).
Article PubMed PubMed Central CAS Google Scholar
Zappia, L. & Oshlack, A. Clustering trees: a visualization for evaluating clusterings at multiple resolutions. Gigascience 7, https://doi.org/10.1093/gigascience/giy083 (2018).
Bergen, V., Lange, M., Peidli, S., Wolf, F. A. & Theis, F. J. Generalizing RNA velocity to transient cell states through dynamical modeling. Nat. Biotechnol. 38, 1408–1414 (2020).
Article PubMed CAS Google Scholar
Sherman, B. T. et al. DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update). Nucleic Acids Res. 50, W216–W221 (2022).
Article ADS PubMed PubMed Central CAS Google Scholar
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550 (2005).
Article ADS PubMed PubMed Central CAS Google Scholar
Zhao, J. et al. Detection of differentially abundant cell subpopulations in scRNA-seq data. Proc. Natl. Acad. Sci. USA 118, https://doi.org/10.1073/pnas.2100293118 (2021).

Download references

Acknowledgements

This work was supported by funding from the Oklahoma Medical Research Foundation’s (OMRF) predoctoral fellowship to J.P.W. The Olson lab is funded by NIH-NIAMS R01-AR073828 and AR080896; the Oklahoma Center for Adult Stem Cell Research—a program of TSET; and the Oklahoma City-based Presbyterian Health Foundation. H.R.K. is funded by NIH-NHLBI F32-HL142222. Sequencing was performed by OMRF’s Clinical Genomics core facility. Data processing and analysis were supported by the OMRF Center for Biomedical Data Sciences. The authors thank Grace Morrison for graphical art and the OMRF Imaging core facility (associated with P30-GM149376) for assistance with spatial transcriptomics.

Author information

Authors and Affiliations

Cardiovascular Biology Research Program, Oklahoma Medical Research Foundation, Oklahoma City, OK, USA
John P. Woods, Alex Rackley, Hae Ryong Kwon & Lorin E. Olson
Department of Cell Biology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, USA
John P. Woods & Lorin E. Olson

Authors

John P. Woods
View author publications
Search author on:PubMed Google Scholar
Alex Rackley
View author publications
Search author on:PubMed Google Scholar
Hae Ryong Kwon
View author publications
Search author on:PubMed Google Scholar
Lorin E. Olson
View author publications
Search author on:PubMed Google Scholar

Contributions

J.P.W. and L.E.O. conceived and designed the study. J.P.W. performed all embryo analysis, carried out the single-cell and spatial transcriptomics experiments. J.P.W. and H.R.K. performed bioinformatics analysis. J.P.W. and A.R. maintained mouse lines and performed histology. H.R.K. and L.E.O. supervised the research. J.P.W. wrote the paper. H.R.K. and L.E.O. edited the paper.

Corresponding authors

Correspondence to Hae Ryong Kwon or Lorin E. Olson.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review file

Description of Additional Supplementary Files

Supplementary Data 1–3

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Woods, J.P., Rackley, A., Kwon, H.R. et al. PDGFRα signaling regulates cartilage and fibrous tissue differentiation during synovial joint development. Nat Commun 16, 4041 (2025). https://doi.org/10.1038/s41467-025-59207-1

Download citation

Received: 05 June 2024
Accepted: 14 April 2025
Published: 30 April 2025
DOI: https://doi.org/10.1038/s41467-025-59207-1