Research Article

Developmental Biology

Hypertrophic chondrocytes serve as a reservoir for marrow-associated skeletal stem and progenitor cells, osteoblasts, and adipocytes during skeletal development

Department of Cell Biology, Duke University School of Medicine, United States
Department of Orthopaedic Surgery, Duke University School of Medicine, United States
Department of Pharmacology and Cancer Biology, Duke University School of Medicine, United States
Program of Developmental and Stem Cell Biology, Duke University School of Medicine, United States
Division of Laboratory Animal Resources, Duke University School of Medicine, United States
School of Biomedical Sciences, University of Hong Kong, Hong Kong

Feb 18, 2022

https://doi.org/10.7554/eLife.76932

Open access
Copyright information

Version of Record

Accepted for publication after peer review and revision.

Download
Cite
Share
CommentOpen annotations (there are currently 0 annotations on this page).

Version of Record published: March 3, 2022 (This version)
Accepted Manuscript published: February 18, 2022 (Go to version)
Accepted: February 13, 2022
Received: January 10, 2022
Preprint posted: May 18, 2021 (Go to version)

1. Of interest
Dynamic compartmentalization of the pro-invasive transcription factor NHR-67 reveals a role for Groucho in regulating a proliferative-invasive cellular switch in C. elegans

Taylor N Medwig-Kinney, Brian A Kinney ... David Q Matus

Research Article Dec 1, 2023
Further reading

Abstract
Editor's evaluation
Introduction
Results
Discussion
Materials and methods
Appendix 1
Data availability
References
Decision letter
Author response
Article and author information
Metrics

Abstract

Hypertrophic chondrocytes give rise to osteoblasts during skeletal development; however, the process by which these non-mitotic cells make this transition is not well understood. Prior studies have also suggested that skeletal stem and progenitor cells (SSPCs) localize to the surrounding periosteum and serve as a major source of marrow-associated SSPCs, osteoblasts, osteocytes, and adipocytes during skeletal development. To further understand the cell transition process by which hypertrophic chondrocytes contribute to osteoblasts or other marrow associated cells, we utilized inducible and constitutive hypertrophic chondrocyte lineage tracing and reporter mouse models (Col10a1CreERT2; Rosa26^fs-tdTomato and Col10a1Cre; Rosa26^fs-tdTomato) in combination with a PDGFRa^H2B-GFP transgenic line, single-cell RNA-sequencing, bulk RNA-sequencing, immunofluorescence staining, and cell transplantation assays. Our data demonstrate that hypertrophic chondrocytes undergo a process of dedifferentiation to generate marrow-associated SSPCs that serve as a primary source of osteoblasts during skeletal development. These hypertrophic chondrocyte-derived SSPCs commit to a CXCL12-abundant reticular (CAR) cell phenotype during skeletal development and demonstrate unique abilities to recruit vasculature and promote bone marrow establishment, while also contributing to the adipogenic lineage.

Editor's evaluation

The work reveals that a subpopulation of hypertrophic chondrocytes can form SSPCs that give rise to osteoblasts and adipocytes during skeletal development, providing new evidence to the current concept of chondrocyte-osteoprogenitors-osteoblast trans-differentiation.

https://doi.org/10.7554/eLife.76932.sa0

Introduction

Most skeletal elements arise from a cartilage template via the process of endochondral ossification (Long and Ornitz, 2013; Olsen et al., 2000; Ono et al., 2019). During skeletal development, mesenchymal stem and progenitor cells condense and differentiate to form chondrocytes or cartilage cells (Hirao et al., 2006; Maes et al., 2012; Olsen et al., 2000; Spencer et al., 2014). These chondrocytes undergo rapid proliferation while depositing an extracellular matrix (ECM) rich in type II collagen (COL2A1), forming the initial immature cartilage rudiments (Ono et al., 2019). As cartilage growth proceeds, the chondrocytes localized nearest the center of each rudiment exit the cell cycle and undergo a process of maturation and hypertrophic differentiation (Capasso et al., 1984; Kong et al., 1993; Kwan et al., 1997; Long and Ornitz, 2013; Olsen et al., 2000; Ono et al., 2019; Reginato et al., 1986; Schmid and Conrad, 1982). Hypertrophic chondrocytes, which are identified by their specific expression of type X collagen (COL10A1), are ultimately responsible for secreting additional proteins such as Vascular Endothelial Growth Factor alpha (VEGFα), Platelet Derived Growth Factor beta (PDGFβ), and Indian Hedgehog (IHH) among other factors to promote both vascular invasion and osteoblast differentiation of the surrounding perichondrial progenitors (Capasso et al., 1984; Kwan et al., 1997; Olsen et al., 2000; Reginato et al., 1986; Schmid and Conrad, 1982; St-Jacques et al., 1999; Zelzer et al., 2004). As vessels from the perichondrium invade the hypertrophic cartilage to establish the marrow cavity, newly formed osteoblasts and their progenitors are delivered by these vessels to the marrow space (Maes et al., 2010; Olsen et al., 2000). There, osteoblasts secrete a type I collagen (COL1A1)-rich matrix that ultimately mineralizes to replace the cartilage with bone, while the hypertrophic cartilage is remodeled and reduced with age (Olsen et al., 2000). Hypertrophic chondrocytes are essential for both establishment of the marrow cavity and bone formation; however, their precise fate and direct or indirect contribution to bone formation has been questioned (Tsang et al., 2015).

Recent genetic evidence suggests that hypertrophic chondrocytes undergo a process of transdifferentiation to directly generate osteoblasts and osteocytes that contribute to the formation of trabecular and endocortical bone (Park et al., 2015; Yang et al., 2014b; Yang et al., 2014a; Zhou et al., 2014a). To determine the fate of growth plate chondrocytes during endochondral bone formation, investigators have utilized mouse lines combining constitutive and inducible CRE recombinases under the control of the Col2a1, Acan, or Col10a1 promotors/enhancers with a recombination sensitive reporter. Utilization of the constitutive and inducible Col2a1Cre/Col2a1CreERT2 mouse lines have complicated our understanding of this lineage by the fact that in addition to their expression in immature chondrocytes, Col2a1 and these genetic tools have been noted to be expressed within and/or label directly perichondrial and marrow associated cell populations known to give rise to osteoblast lineage cells (Hilton et al., 2007; Long et al., 2001). Some of the more convincing data generated utilizing a Col2a1CreERT2 mouse line and recombination-sensitive reporter comes from Yang et al., 2014a, which established an approach for low-dose tamoxifen delivery to drive recombination in clones or clusters of immature chondrocytes that could be tracked over time. When the chase period was short (1–5 days post tamoxifen), reporter expression was largely restricted to immature chondrocyte clusters or extensions of these clusters into early hypertrophic chondrocytes. However, when the chase period was extended (10 days post tamoxifen), reporter expression could be observed throughout individual columns of proliferating to hypertrophic chondrocytes that also extended into Col1a1-expressing osteoblasts localized only to the regions of metaphyseal bone directly underneath the labeled chondrocytes; a result that is consistent with chondrocytes transitioning to an osteoblastic fate (Yang et al., 2014a). Similar pulse-chase experiments were performed utilizing the AcanCre^ERT2; Rosa26^fs-tdTomato reporter model at both embryonic and adult time points utilizing doses of tamoxifen that elicit high rates of recombination in chondrocytes. These data suggested that following short (1–3 days post tamoxifen) or long (2 weeks post tamoxifen) chase periods, non-chondrocytic cells associated with the primary spongiosa and trabecular bone were reporter positive with some cells labeling with the Col1a1(2.3 kb)-eGFP.(Zhou et al., 2014b) The most convincing data was generated utilizing a combination of constitutive and inducible Col10a1Cre/Col10a1CreERT2 mouse lines and recombination-sensitive reporters; models that are much more restricted in their expression to hypertrophic chondrocytes of the growth plate (Yang et al., 2014b). This study, as well as others solely utilizing a constitutive Col10a1Cre model, determined that hypertrophic chondrocytes can relatively quickly give rise to osteoblastic cells expressing OSTERIX (OSX), COL1A1, and eventually even OSTEOCALCIN (OCN) and/or SCLEROSTIN (SOST) expressing osteocytes within the trabecular and endocortical bone matrix (Park et al., 2015; Yang et al., 2014b; Yang et al., 2014a; Zhou et al., 2014b). Interestingly, Yang et al., 2014a utilized a Col10a1Cre; Rosa26^fs-eYFP reporter model and noted the presence of some YFP⁺, Perilipin-expressing adipocytes and YFP⁺, PDGFRβ⁺ pericytes within the bone marrow, suggesting the potential for alternate fates of hypertrophic chondrocytes beyond the osteoblast lineage. Utilizing similar techniques, Tan et al., 2020 also demonstrated the potential for mature adipocytes to arise from Col10a1-expressing hypertrophic chondrocytes.

Since the fates of mature osteoblasts, adipocytes, and pericytes are unlikely to arise from independent hypertrophic chondrocyte transdifferentiation events, we reasoned that the hypertrophic chondrocytes may undergo a dedifferentiation process to generate skeletal stem and progenitor cell (SSPC) population(s) that localize to the bone marrow and maintain the capacity of redifferentiating into more mature cell fates. To address this issue, we utilized hypertrophic chondrocyte genetic reporter mouse models combined with single cell RNA-sequencing, immunofluorescent staining validation assays, fluorescent activated cell sorting (FACS)/flow cytometry, and bulk RNA-sequencing approaches of clonal cell populations. Our data indicate that hypertrophic chondrocytes indeed undergo a process of dedifferentiation to generate bone-marrow-associated SSPCs embryonically and postnatally. These SSPC likely serve as a primary source of osteogenic cells during skeletal development and also contribute to the adipogenic lineage, but not the pericyte lineage. Further, these studies indicate that the hypertrophic chondrocyte derived SSPCs are strikingly similar at the transcriptomic level to marrow-associated SSPCs derived from other sources, but behave quite differently upon kidney capsule transplantation to test their multipotency. Hypertrophic chondrocyte derived SSPCs were able to give rise to a fully mature bone ossicle, complete with marrow cavity and adipocytes, while marrow SSPCs from other sources could only generate bone.

Results

Hypertrophic chondrocytes give rise to osteoblasts and other marrow-associated cells

To validate the osteogenic fate of hypertrophic chondrocytes, we first utilized the Col10a1CreERT2 mouse line crossed with a Rosa26^fs-tdTomato reporter (Col10a1CreERT2; Rosa26^fs-tdTomato). These mice were harvested at E16.5 following tamoxifen injections from E13.5-E15.5. Hypertrophic chondrocytes and their descendants found within the newly formed bone marrow cavity were tdTOMATO⁺ (Figure 1a). A number of these descendants co-expressed the osteoblast protein OCN (Figure 1b–c, white arrow); however, OCN-, tdTOMATO⁺ cells were also apparent within the marrow (Figure 1b–c, orange arrow). Additionally, we injected Col10a1CreERT2; Rosa26^fs-tdTomato mice with a single dose of tamoxifen (TM) at postnatal day 6 (P6) and harvested at 12 hr, 24 hr, and 2 weeks following TM (Figure 1d–g). At 12 hr following TM administration, specific expression of the tdTOMATO reporter was observed within hypertrophic chondrocytes and no cells within the marrow or surrounding tissues (Figure 1d). At 24 hr following TM administration, the tdTOMATO reporter could now be observed in a limited number of marrow-associated cells directly below the cartilage growth plate, as well as within hypertrophic chondrocytes (Figure 1e). At 2 weeks following TM administration, the tdTOMATO reporter had cleared from the hypertrophic chondrocytes and was now exclusively present in a number marrow-associated cells; some of those being at a significant distance from the growth plate (Figure 1f). In this distal region of the marrow, we also observed trabecular bone lining cells that co-labeled for tdTOMATO and OCN (Figure 1g; white arrows), suggesting that these cells had transitioned from hypertrophic chondrocytes to osteoblasts. Interestingly, other tdTOMATO-positive marrow-associated cells were not directly in contact with bone matrix and did not co-label with OCN (Figure 1g; orange arrow), suggesting the presence of another cell type or cell stage within the bone marrow that may be derived from hypertrophic chondrocytes. Additionally, by inducing recombination at both 2 months and 4 months of age and chasing for 2 and 4 weeks, respectively, we continued to observe contribution of tdTOMATO⁺ cells to the bone marrow (Figure 1—figure supplement 1a, b).

Figure 1 with 1 supplement see all

Download asset Open asset

Hypertrophic chondrocytes are the primary source of osteoblasts/osteocytes in trabecular bone.

Tibia sections of *Col10a1CreERT2; Rosa26^fs-tdTomato* mice injected with tamoxifen at E13.5, 14.5, and 15.5 and subsequently sectioned at E16.5 and stained for (a) DAPI/RFP. (**b–c**) Higher magnification with OCN⁺/tdTOMATO⁺ descendants marked with white arrows and tdTOMTATO⁺ descendants not associated with bone (OCN^-) are indicated by orange arrows. Tibia sections of *Col10a1CreERT2; Rosa26^fs-tdTomato* mice injected with tamoxifen at P6 and subsequently sectioned at (d) 12 hour chase, (e) 24 hr chase, and (f) 2 week chase. (g) OCN immunostaining of *Col10a1CreERT2; Rosa26^fs-tdTomato* bone sections following 2 week chase of TM. OCN⁺/tdTOMATO⁺ descendants marked with white arrows. tdTOMATO⁺ descendants not associated with bone (OCN^-) are indicated by orange arrows. (h) Section of 2 M old *Col10a1Cre; Rosa26^fs-tdTomato* tibia with higher magnifications of (i) hypertrophic zone (green box) and (j) bone marrow cavity (orange box). Non-bone lining TOMATO⁺ descendants marked with white arrows and potential vessel associated tdTOMATO⁺ descendants indicated by yellow arrow. (k) OCN and RFP immunostaining of 2 M *Col10a1Cre; Rosa26^fs-tdTomato* tibia. OCN⁺/tdTOMATO⁺ osteoblasts represented by green arrows and tdTOMATO⁺ osteocytes with white arrows with quantifications shown in (l – Figure 1—source data 1). Scale bars = 100 um. N = 3 slides from three biologic replicates, SD ±9.7% for OCN stain, SD ±14.8% for osteocytes. Dotted line demarks the chondro-osseous junction.

Figure 1—source data 1 Osteoblast and osteocyte quantification on Col10a1Cre;Rosa26fs-tdTomato 2M old mice.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig1-data1-v2.xlsx
Download elife-76932-fig1-data1-v2.xlsx

To get a better understanding of the proportionality of these hypertrophic chondrocyte-derived marrow and osteogenic cells, we performed similar analyses on constitutively active Col10a1Cre; Rosa26^fs-tdTomato mice. While tdTOMATO reporter expression was observed within hypertrophic chondrocytes and osteogenic cells associated with trabecular and endocortical bone of mice at P0 (Figure 1—figure supplement 1d-f) and 2 months (2 M) of age (Figure 1h–j), CRE recombinase could only be detected within hypertrophic chondrocytes (Figure 1—figure supplement 1c). The vast majority of the trabecular and endocortical bone was lined with tdTOMATO reporter-positive cells, and 69.4% ± 14.8% of osteocytes were reporter positive (Figure 1k and l – white arrows). We also observed that 83.3% ± 9.7% of all OCN-positive osteoblasts were tdTOMATO reporter positive, confirming that hypertrophic chondrocytes can give rise to the majority of intramedullary osteoblasts during bone development (Figure 1k and l – green arrows). Of note, a number of tdTOMATO negative and OCN positive cells were also identified lining portions of the trabecular and endocortical bone, indicating that some of these osteoblasts originate from an alternate source such as the perichondrium/periosteum (Figure 1k). Finally, utilizing this model, we were able to confirm the existence of tdTOMATO reporter expressing cells within the marrow that were not associated with bone matrix (white arrow); many of which were localized to vessel-like structures (yellow arrow) (Figure 1j).

Single-cell transcriptomics identifies several cell populations, including osteoblasts that are derived from hypertrophic chondrocytes during embryonic and postnatal development

Following the observation that tdTOMATO reporter-positive cells were both marrow- and bone-associated (Figure 1), we hypothesized that the hypertrophic chondrocytes may give rise to multiple cell fates. We therefore took an unbiased approach by performing single cell RNA-sequencing (scRNA-seq) on cells derived from Col10a1Cre; Rosa26^fs-tdTomato mice at E16.5, a timepoint coincident with bone marrow cavity initiation and the potential transition of hypertrophic chondrocytes to alternative fates. To isolate the hypertrophic chondrocytes and their descendants we liberated the femur, tibia, and fibula from soft tissues and performed collagenase digestions of the skeletal rudiments. Following matrix digestion, red blood cells were lysed and the remaining cell populations were subjected to FACS. We selected tdTOMATO reporter positive cells that demonstrated high levels of tdTOMATO fluorescence for further analysis (Figure 2—figure supplement 1a). Using the 10 x Genomics platform, we performed scRNA-seq and further analyzed 2,273 tdTOMATO reporter positive sorted cells following quality control processing and elimination of a small number of cells lacking tdTomato transcripts. We then performed a Seurat-based unsupervised clustering of gene expression profiles from the tdTOMATO positive cells, which ultimately yielded eight distinct cell clusters (Figure 2a; see Materials and methods for details). Initial observations demonstrated that both Cre and Col10a1 transcripts were detected in a portion of the cells, but we also observed cells expressing tdTomato without expression of either Cre or Col10a1, indicating the presence of both hypertrophic chondrocytes and their descendants within these clusters (Figure 2—figure supplement 2a, i).

Figure 2 with 2 supplements see all

Download asset Open asset

Single cell transcriptomics at E16.5 captures hypertrophic chondrocytes and osteoblasts.

(a) UMAP shown in two-dimensional space produced using Seurat 4 package of R from single-cell RNA-sequencing of cleaned skeletal rudiments from *Col10Cre; Rosa26^fs-tdTomato* mice at E16.5. (**b–d**) Feature plots of hypertrophic chondrocyte associated genes identified in clusters 1–7. (**e–h**) Feature plots of osteoprogenitor/osteoblast specific genes identified in clusters 7–8. (i) Violin plot representing the relative level of chondrocyte and osteoblast associated gene expression (**b–h**).

Next we analyzed expression of chondrogenic genes. Differential gene expression analysis revealed the presence of Ihh and Col10a1 expressing chondrocytes (most notably in clusters 1, 3, and 5), identifying cells in transition from pre-hypertrophy to hypertrophy (Figure 2b and i; St-Jacques et al., 1999). Col10a1 expression was observed throughout many clusters; however was most highly enriched in clusters 1–5 and 7 (Figure 2c and i). Moderate expression of Sox9 and Aggrecan were observed throughout all clusters, but primarily enriched in clusters 1–5 (Figure 2—figure supplement 2b-c, i). Mmp13, noted for its role in terminal hypertrophic chondrocytes, reached highest expression in clusters 4, 5, and 6 (Figure 3d and i). The osteogenic genes, Sp7, Clec11a, Col1a1, Bglap, Runx2, Spp1, Alpl, and Ibsp, were most highly expressed in clusters 7 and 8, which represent two potential osteoblast clusters (Figure 2e–h and i and Figure 2—figure supplement 2d-i), while some were also detected at much lower levels in pre-hypertrophic/hypertrophic chondrocyte clusters.

Figure 3 with 1 supplement see all

Download asset Open asset

Single-cell transcriptomics of E16.5 hypertrophic chondrocytes and descendants reveals an intermediate SSPC upstream of osteoblasts.

(**a–d**) Feature plot of SSPC-associated genes identified between hypertrophic chondrocyte and osteoblast clusters (e) Violin plot representing the relative level of SSPC associated gene expression (**a–d**). (f) Monocle three trajectory analysis throughout pseudotime. (g) E16.5 tibia sections of *Col10a1CreERT2; Rosa26^fs-tdTomato;PDGFRa^H2B-GFP* mice injected with tamoxifen at E13.5, 14.5, and 15.5 and quantification (Figure 3—source data 1). Orange arrows represent tdTOMATO⁺/PDGFRa^H2B-GFP+ cells that are OCN^-. White arrows represent tdTOMATO⁺/ PDGFRa^H2B-GFP+ cells that co-express OCN. Scale bar = 100 um. N = 1/2 slides for three biological replicates, Average = 36.7%, SD ±20.7%. Dotted line demarks the chondro-osseous junction.

Figure 3—source data 1 PDGFRa+ TOMATO+ quantification at e16.5 of Col10a1Cre; Rosa26-fs-tdTomato; PDGFRa-H2B-GFP.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig3-data1-v2.xlsx
Download elife-76932-fig3-data1-v2.xlsx

Differential gene expression analysis also identified the SSPC-associated genes, Pdgfra, Ly6a (encoding SCA1), Lepr, Cxcl12, to be expressed within clusters 4, 5, and 6, with some expression in cluster 8 (Figure 3a–e). Each of these genes and/or their regulatory elements genetically identify SSPCs capable of becoming osteoblasts in vivo (Crisan et al., 2008; Morikawa et al., 2009; Omatsu et al., 2010; Worthley et al., 2015; Zhou et al., 2014a). We did not observe Adipoq expression in any clusters, another gene previously shown to be co-expressed with Lepr in some SSPC populations (Figure 2—figure supplement 2h-i; Zhong et al., 2020). To further understand the cluster hierarchy of these hypertrophic chondrocytes and their descendants, we performed pseudotime analysis of the scRNA-seq data. This analysis indicated a singular trajectory linking hypertrophic chondrocytes (clusters 1–5) to osteoblasts (clusters 7 and 8) by way of an intermediate SSPC-like population (cluster 6) (Figure 3f). To validate this transition of hypertrophic chondrocytes to a marrow-associated SSPC fate, we generated Col10a1CreERT2; Rosa26^fs-tdTomato; PDGFRa^H2B-GFP mice and performed additional in vivo lineage-tracing analysis at E16.5. Fluorescent reporter analysis identified a number of tdTOMATO and PDGFRa^H2B-GFP double positive cells within the marrow space that were either OCN⁺ or OCN^- (Figure 3g–h), indicating that tdTOMATO⁺, GFP⁺, and OCN^- cells likely represent a more primitive hypertrophic chondrocyte-derived SSPC. To further validate that hypertrophic chondrocytes are capable of giving rise to marrow-associated SSPCs expressing Lepr, Pdgfra, and Pdgfrb in vivo throughout development, we performed IF co-labeling assays on bone sections from Col10a1CreERT2; Rosa26-tdTomato ^f/+ mice at 3 W of age (TM delivered at P6). Co-labeling of cells with tdTOMATO fluorescence and antibodies for LEPR, PDGFRα, and PDGFRβ demonstrate that hypertrophic chondrocytes indeed give rise to marrow associated SSPCs during skeletal development (Figure 3—figure supplement 1a).

After identifying Col10a1CreERT2; Rosa26^fs-tdTomato reporter positive SSPCs at early stages of development, we hypothesized that these cells may still arise from hypertrophic chondrocytes during later stages of skeletal development. Therefore, we performed scRNA-seq of tdTOMATO⁺ sorted marrow cells from Col10a1Cre; Rosa26^fs-tdTomato mice at 2 months of age (2 M). To isolate and specifically enrich hypertrophic chondrocyte-derived marrow-associated cells, we performed centrifugation of the marrow followed by a gentle flush of femurs and tibias and subjected these marrow plugs to a brief collagenase digestion. This short digestion allowed for potential isolation of a limited number of osteoblasts, since our focus was centered on the analysis of novel non-osteoblast/osteocyte populations. Following the digest, red blood cells were lysed, and the remaining cell populations were subjected to FACS. We selected tdTOMATO reporter positive cells that demonstrated high levels of unambiguous tdTOMATO fluorescence for further analysis (Figure 2—figure supplement 1b). Using the 10 x Genomics platform, we performed scRNA-seq and further analyzed 1,066 tdTOMATO reporter positive sorted cells following quality control processing and elimination of any cells lacking tdTomato transcripts. Since neither Cre nor Col10a1 expression could be detected in any of these cells, they clearly represented descendants of Col10a1Cre-expressing hypertrophic chondrocytes. Finally, we performed a Seurat-based unsupervised clustering of gene expression profiles from the tdTOMATO-positive cells, which ultimately yielded five distinct cell clusters (Figure 4a; see Materials and methods for details).

Figure 4 with 4 supplements see all

Download asset Open asset

Single cell transcriptomics of hypertrophic chondrocyte descendants following FACS and 10 X Genomics sequencing at 2 M of age.

(a) UMAP shown in two-dimensional space produced using Seurat 3 package of R from single-cell RNA-sequencing of bone marrow digest from *Col10Cre; Rosa26^fs-tdTomato* mice at 2 M. (**b–f**) Feature plots of osteoblast specific genes identified in cluster 1 (**b–d**) and cluster 2 (**e–f**). (g) Feature plot of the osteoprogenitor associated gene, *Sp7*. (h) Violin plot representing the relative level of osteoblast-associated gene expression (**b–g**).

Differential gene expression analyses further demonstrated that cells associated with clusters 1 and 2 expressed several known osteogenic genes (Figure 4b–h). Many cells in cluster one expressed genes such as Mepe, Dmp1, and Fam20c (Figure 4b–d and h), known for their roles in osteocytes and/or the bone mineralization process (Robling and Bonewald, 2020; Sapir-Koren and Livshits, 2014). Some cells in cluster 2 also expressed Dmp1 and Fam20c; however, their expression was detected at much lower levels. Most cells in cluster two expressed other osteoblast associated genes, such as Bglap and Col1a1 (Figure 4e–f and h), critical components of the organic bone matrix. In addition to the few genes highlighted here, additional osteoblast-associated genes were also observed in cells within clusters 1 and 2 (Figure 4—figure supplement 1a-h). Expression analyses further determined that the master transcriptional regulator of osteoblast differentiation and known osteoprogenitor cell marker, Sp7 (encoding the protein OSTERIX) (Maes et al., 2010; Mizoguchi et al., 2014), not only exhibited expression within osteoblast clusters 1 and 2, but also within a limited number of cells associated with clusters 3, 4, and 5 (Figure 4g–h). These data suggest that the majority of marrow-associated cells derived from the hypertrophic chondrocytes localized to clusters 3, 4, and 5 are not osteoblasts or osteoprogenitors, but rather may be more primitive cells within the osteoblast lineage or distinct cell types altogether. We next utilized differential gene expression analyses to identify genes that were uniquely and broadly expressed across the cells of clusters 3, 4, 5, with limited expression in the more differentiated osteoblasts of clusters 1 and 2. This analysis identified a number of genes utilized for identifying SSPC populations. Specifically, Lepr, Grem1, Pdgfra (but not Ly6a – SCA1), Cxcl12, and Pdgfrb were highly expressed within cells across all three clusters (Figure 5a–f and k). Each gene and/or their regulatory elements genetically identify or can be utilized to label SSPCs with the potential of becoming osteoblasts in vivo (Crisan et al., 2008; Morikawa et al., 2009; Omatsu et al., 2010; Worthley et al., 2015; Zhou et al., 2014a). To confirm that hypertrophic chondrocytes are capable of giving rise to marrow associated SSPCs expressing Lepr, Pdgfra, and Pdgfrb in vivo at this later developmental time point, we performed IF co-labeling assays on bone sections from Col10a1Cre; Rosa26^fs-tdTomato mice at 2months of age. Co-labeling of cells with tdTOMATO fluorescence and antibodies for LEPR, PDGFRα, and PDGFRβ further demonstrate and validate that hypertrophic chondrocytes indeed give rise to marrow-associated SSPCs during postnatal skeletal development (Figure 3—figure supplement 1b). Quantification of Col10a1Cre; Rosa26^fs-tdTomato for LEPR demonstrates that 74.21%±13.92% of all tdTOMATO+ cells were LEPR+ (Figure 3—figure supplement 1c). We additionally performed flow cytometry of marrow cells from Col10a1Cre; Rosa26^fs-tdTomato; PDGFRa^H2B-GFP mice at 1 month of age and observed 73.6% ± 2.83% of tdTOMATO⁺ cells were PDGFRa-GFP+ (Figure 3—figure supplement 1c). At 2 M of age, flow cytometry of marrow cells from Col10a1Cre; Rosa26^fs-tdTomato mice demonstrated that 64.33% ± 3.48% of tdTOMATO⁺ cells were PDGFRB+ (Figure 3—figure supplement 1c) and that 37.4% ± 6.2% of the total LEPR⁺ population were also tdTOMATO⁺ as per quantitation from stained bone sections (Figure 5—figure supplement 1a). This was further corroborated by utilizing flow cytometry of marrow cells from Col10a1Cre; Rosa26^fs-tdTomato; PDGFRa^H2B-GFP mice, which demonstrates that a similar 26.93%±3.79% of all PDGFRa^H2B-GFP+ cells were also tdTOMATO reporter positive (Figure 5—figure supplement 1b). These data indicate that hypertrophic chondrocyte-derived SSPCs constitute nearly 25–45% of all marrow SSPCs at these stages, and that a significant portion of the marrow-associated SSPCs are likely derived from other origins such as the perichondrium and/or periosteum in mice 2 months of age and younger.

Figure 5 with 1 supplement see all

Download asset Open asset

Many hypertrophic chondrocyte derived cells express genes associated with SSPCs.

(**a–f**) Feature plots of genes previously identified as SSPC markers in genetic mouse models. (**g–i**) Feature plots of genes previously identified as SSPC markers for use in flow cytometry and FACS. (j) Monocle three trajectory analysis with clusters noted throughout pseudotime. (k) Violin plots of SSPC-associated genes in (**a–i**).

FACS has been highly utilized to identify, isolate, and enrich populations of SSPCs from bone marrow and periosteal tissues in mice and/or humans based on the expression of specific cell surface proteins (Chan et al., 2015; Chan et al., 2018; Zhou et al., 2014a). FACS assays utilizing antibodies against Integrin Subunit Alpha V (ITGAV) have been used for the identification, isolation, and enrichment of bone marrow stromal cell populations in mice, which are noted to be largely quiescent and capable of generating most of the colonies within a colony-forming unit – fibroblastic (CFU-F) assay; a technique utilized to identify heterogeneous, non-hematopoietic, marrow stromal cells with progenitor cell characteristics (Boulais et al., 2018; Robey et al., 2014). While we identified Itgav expression sporadically throughout cells of all clusters, it is enriched in the SSPC clusters 3, 4, and 5 (Figure 5g and k). Recently, CD200, ITGAV, and CD105 antibodies have been used in conjunction with antibodies detecting the hematopoietic and endothelial lineages (CD45, CD31, and Ter119) for flow cytometric evaluation and identification of unique subsets of SSPCs (Gulati et al., 2018). These studies indicate that lineage negative (CD45^-,CD31^-,Ter119^-) and CD200⁺,ITGAV⁺,CD105^- cells represent the most primitive self-renewing and multipotent skeletal stem cell (SSC) population, while lineage negative and ITGAV⁺,CD105⁺ cells represent a more committed SSPC population (Gulati et al., 2018). Our gene expression analyses observed broad expression of Cd200 across all clusters including the osteoblasts; however, Eng (encoding CD105) was more restricted to SSPC clusters 3, 4, 5 and enriched specifically within cluster 4 (Figure 5h–i and k). When assessing the precise cellular overlap of these factors, we observed 128 of 953 SSPCs to be Itgav+, Cd200+, Cd105-. While this most primitive SSC population exists among our SSPCs analyzed via scRNA-seq, they were relatively evenly distributed among cells within clusters 3, 4, and 5 (Figure 5g–I and k). These data suggest that hypertrophic chondrocytes likely undergo a process of dedifferentiation to generate even the most primitive SSCs; however, these cells themselves may be heterogeneous due to the differential expression of genes that allow for their segregation into multiple distinct clusters within our Seurat analysis.

In an attempt to further delineate each of the SSPC clusters in our Seurat analysis, we analyzed the data to identify genes restricted in their expression to a single SSPC cluster. A limited number of genes were ultimately found to be uniquely expressed within solitary SSPC clusters; however, much more commonly we observed genes to be enriched within specific clusters with limited expression within the other one or two SSPC clusters (Figure 4—figure supplement 2).

SSPCs are known to reside within specific niches often associated with blood vessels (Zhou et al., 2014a). Previously, we noted that some of the hypertrophic chondrocyte derived tdTOMATO reporter cells appeared to associate with vascular structures. To assess this directly, we performed immunofluorescent staining for ENDOMUCIN (EMCN) on bone sections of Col10a1Cre; Rosa26^fs-tdTomato mice at 2 months of age and identified a number of tdTOMATO+ reporter cells directly adjacent to EMCN expressing endothelial cells (Figure 4—figure supplement 3a). To ensure that these tdTOMATO+ cells were not pericytes or vascular smooth muscle cells (vSMCs), we assessed our differential gene expression data for genes that serve as common markers for these specific cell types, including Acta2, Nes, Kcnj8, Rgs5, Cspg4, and Mcam (Chasseigneaux et al., 2018; Yamazaki and Mukouyama, 2018). We observed limited or no gene expression for these markers in SSPCs of clusters 3, 4, or 5 (Figure 4—figure supplement 3b-g).

To finally parse the relatedness or relationship between SSPC and osteoblast cell clusters, we performed Monocle three pseudotime cell trajectory analysis of the Seurat data. These data indicate a clear hierarchy of cellular differentiation and relatedness between all cell clusters in an unbranched or continuous trajectory. Clusters 4 and 5 shared a common trajectory, with cluster 4 being slightly earlier in pseudotime. These clusters then give rise to cluster 3, which further differentiates into the osteoblast clusters 2 and then 1 (Figure 5j). Taken together these data indicate that cluster 3, 4, and 5 are hypertrophic chondrocyte-derived SSPCs that are capable of becoming osteoblasts and have unique niche localizations with at least some of these SSPCs being vessel associated.

Hypertrophic chondrocyte-derived SSPCs are dually primed for both osteoblast and adipocyte differentiation

Since most bone-marrow-associated SSPCs have the potential to become either osteoblasts or adipocytes during development, aging, regeneration, and/or disease, we further examined our differential gene expression data for indicators of this competency. While we observed limited expression of the osteoprogenitor transcription factor Sp7 in the SSPC clusters 3, 4, and 5, we noticed much broader and robust expression of Runx2, the earliest master transcriptional regulator of osteoblast specification, throughout all clusters and especially concentrated in the SSPC clusters 3, 4, and 5 (Figure 6a). RUNX2 has been observed in LEPR⁺ cells and appears to prime cells for osteoblastic specification and differentiation (Yang et al., 2017). We have already demonstrated in Figure 1 the efficiency by which hypertrophic chondrocytes can contribute to the osteoblast lineage, and that these osteoblasts are part of a differentiation trajectory that can originate from hypertrophic chondrocyte-derived SSPCs.

Figure 6

Download asset Open asset

Hypertrophic chondrocyte derived SSPCs exhibit osteogenic and adipogenic differentiation capacities.

(a) Feature plot and violin plot of the osteoblast specification gene, Runx2. (**b–c**) Feature plots and violin plots of the adipogenic specification gene, Pparg, and adipogenic associated gene, Adipoq. (d) Feature plot indicating a lack of expression of the mature lipid laden adipocyte gene, Perilipin. (e) Immunostaining for PERILIPIN in *Col10a1Cre;R26-tdTomato* mice at 2 months and 13 months of age. PERILIPIN⁺, tdTOMATO⁺ adipocytes noted with white arrows. Scale bar = 100 µm (f) Quantification of PERILIPIN⁺, tdTOMATO⁺ adipocytes at 13 months of age in (e) Average = 27.59% (Figure 6—source data 1). N = 3 slides from three biologic replicates, SD ±16.2%.

Figure 6—source data 1 PERILIPIN+ TOMATO+ quantifications on Col10a1Cre;Rosa26fs-tdTomato at 2M.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig6-data1-v2.xlsx
Download elife-76932-fig6-data1-v2.xlsx

Recent data has shown that some LEPR⁺ SSPCs also express PPARγ (encoded by Pparg) and ADIPONECTIN (encoded by Adipoq) (Zhong et al., 2020). Similar to this report we observed Pparg and Adipoq expression within Lepr-expressing SSPCs of clusters 3, 4, and 5 (Figure 6b–c). While PPAR-gamma is necessary and sufficient to induce adipogenesis of SSPCs, ADIPONECTIN enhances osteogenesis but is primarily secreted by adipocytes (Tencerova and Kassem, 2016). Given the expression of these adipogenic regulators and marker genes, we asked if hypertrophic chondrocyte-derived cells in clusters 3, 4, and 5 represent mature adipocytes or whether these cells could ultimately give rise to mature adipocytes in vivo. Our differential gene expression analysis did not indicate the presence of Perilipin (Plin1), a mature lipid laden adipocyte marker, in SSPC clusters 3, 4, or 5, suggesting that none of these cells isolated utilizing our procedures were representative of mature adipocytes (Figure 6d). Providing evidence for an ultimate adipogenic fate for some of these SSPCs, Col10a1Cre; Rosa26^fs-tdTomato mice at 2months and 13months of age displayed co-localization of the tdTOMATO reporter and PERILIPIN antibody staining (Figure 6e). By 13 M of age, approximately 25% of adipocytes labeled with PERILIPIN were also tdTOMATO reporter positive (Figure 6f). None of these double-labeled cells are either known to express Col10a1, nor could we detect CRE expression within any of these cells. Therefore, our results suggest that hypertrophic chondrocyte-derived SSPCs are also a significant contributor to the adipogenic fate; however, these mature cells likely eluded our scRNA-seq analyses due to the technical nature for isolating and sorting lipid laden adipocytes and their limited availability in the bone marrow at 2 months of age.

To assess SSPC contribution to the osteoblast and adipocyte lineages, we also performed reporter studies utilizing LeprCre; Rosa26^fs-tdTomato and AdipoqCre; Rosa26^fs-tdTomato mice. Utilizing bone sections of P0 LeprCre; Rosa26^fs-tdTomato reporter mice, we observed reporter positive signals in both-marrow-associated cells and some bone lining cells; however, OCN⁺, tdTOMATO⁺ osteoblasts were not apparent (Figure 4—figure supplement 4a). By 2 M of age we observed bone lining cells, OCN-expressing osteoblasts, and osteocytes that were tdTOMATO reporter positive (Figure 4—figure supplement 4a – white arrows), similar to data previously reported (Zhou et al., 2014a). As expected, a limited number of PERILIPIN⁺, tdTOMATO⁺ adipocytes were also observed within the bone marrow of 2 M old LeprCre; Rosa26^fs-tdTomato mice (Figure 4—figure supplement 4a). Similarly, P0 AdipoqCre; Rosa26^fs-tdTomato bone sections exhibited marrow-associated and bone lining reporter-positive cells that not yet contributed to OCN+ osteoblasts; however, by 2 M of age tdTOMATO⁺ and OCN⁺ osteoblasts were observed, as well as tdTOMATO⁺ osteocytes and adipocytes (PLIN1+) were apparent in Adipoq Cre; Rosa26^fs-tdTomato bone sections (Figure 4—figure supplement 4b – white arrows). In contrast with these studies, Col10a1Cre; Rosa26^fs-tdTomato descendants demonstrate a relatively quick commitment to the osteoblast lineage that is observed at P0 and earlier, while Col10a1CreERT2; Rosa26^fs-tdTomato mice document that bone lining cells and mature OCN-expressing osteoblasts can be generated from hypertrophic chondrocytes within 1 day to 2 weeks (Figure 1a–g). This highlights the potential for functional differences between sub-populations of LEPR and ADIPOQ expressing SSPCs based on their particular origin and/or niche.

Hypertrophic chondrocyte-derived SSPCs may be uniquely primed for developmental bone growth in vivo

Col10a1Cre-derived marrow-associated SSPCs constitute only a portion of LEPR⁺ SSPCs and LeprCre-derived marrow-associated SSPCs contribute to nearly 100% of all CFU-F colonies, therefore we reasoned that Col10a1Cre-derived marrow-associated SSPCs would contribute to only a fraction of total CFU-Fs and that this may provide a method for separating hypertrophic chondrocyte-derived progenitors from those derived from alternative sources such as the perichondrium/periosteum (Zhou et al., 2014a). Following plating at clonal density and 10 days of growth, 50–60% of CFU-F colonies derived from Col10a1Cre; Rosa26^fs-tdTomato mice exhibited relatively homogeneous tdTOMATO reporter expression while the other colonies were largely devoid of tdTOMATO positivity at both 2 months and 8 months of age (Figure 7a–c). To achieve a better understanding of the similarities and differences between progenitors derived from these different sources at the transcriptomic level, we isolated three tdTOMATO⁺ and three tdTOMATO^- CFU-F colonies and performed individual RNA-isolations and submitted for bulk RNA-sequencing. Of the differentially expressed genes reaching significance (p < 0.05, LogFC >1), the three tdTOMATO⁺ CFU-Fs clustered separately from three tdTOMATO^- CFU-Fs on the basis of only 363 differentially expressed genes (227 enriched in tdTOMATO⁺ CFU-Fs and 136 enriched in tdTOMATO^- CFU-Fs) out of approximately 12,000 genes with an average expression level above biological thresholds (Figure 7d and Methods for threshold calculations). We observed Lepr, Grem1, Pdgfra, Adipoq, Pparg, Itgav, Cd200, and Cxcl12, all highly expressed above biological thresholds in all CFU-Fs, indicating a general maintenance of SSPC phenotype while in culture and a striking similarity between progenitors from disparate sources.

Figure 7

Download asset Open asset

CFU-Fs derived from hypertrophic chondrocytes are similar to CFU-Fs derived from other cell sources, but contain SSPCs with enhanced proliferative capacities.

(**a–c**) CFU-Fs derived from hypertrophic chondrocytes (tdTOMATO⁺) (a) and those derived from other cell sources (tdTOMATO^-)(b) are established and develop at similar frequencies at 2- and 8 months of age (c – Figure 7—source data 1). Scale bar = 500 µm (d) Heat-map from bulk RNA-seq data from three tdTOMATO⁺ CFU-Fs and 3 TOMATO^- CFU-Fs (Figure 7—source data 2). (e) Gene ontology terms associated with tdTOMATO⁺ CFU-Fs (Figure 7—source data 3 and Figure 7—source data 4). (f) EdU incorporation within tdTOMATO⁺ and tdTOMATO^- CFU-Fs (top - Figure 7—source data 5) and differences in cell numbers between tdTOMATO⁺ and tdTOMATO^- CFU-Fs (bottom - Figure 7—source data 6). N = 2 technical replicates of three biologic replicates, SD tdTOMATO^- ± 16.5%, SD tdTOMATO⁺ ± 22.3%, p-value = 0.044. (g) Cell cycle analysis of the SSPCs associated with clusters 1–5 using Seurat 3.

Figure 7—source data 1 Tomato+ colony unit formation quantifications at 2 and 8M of Col10a1Cre;Rosa26fs-tdTomato.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig7-data1-v2.xlsx
Download elife-76932-fig7-data1-v2.xlsx
Figure 7—source data 2 Bulk RNA-sequencing total gene list of 3 TOMATO+ and 3 TOMATO- colonies from Col10a1Cre;Rosa26fs-tdTomato.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig7-data2-v2.xlsx
Download elife-76932-fig7-data2-v2.xlsx
Figure 7—source data 3 Gene ontology analysis of genes enriched in TOMATO+ colonies.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig7-data3-v2.xlsx
Download elife-76932-fig7-data3-v2.xlsx
Figure 7—source data 4 Gene ontology analysis of genes enriched in TOMATO- colonies.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig7-data4-v2.xlsx
Download elife-76932-fig7-data4-v2.xlsx
Figure 7—source data 5 EDU+ quantifications of TOMATO+ and TOMATO- colonies.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig7-data5-v2.xlsx
Download elife-76932-fig7-data5-v2.xlsx
Figure 7—source data 6 Colony size of TOMATO+ and TOMATO- colonies.: https://cdn.elifesciences.org/articles/76932/elife-76932-fig7-data6-v2.xlsx
Download elife-76932-fig7-data6-v2.xlsx

To more readily identify differences between tdTOMATO⁺ and tdTOMATO^- CFU-Fs, we performed gene ontology analysis for biological processes and found two interesting terms arise due to enriched expression of genes within the tdTOMATO⁺ or hypertrophic chondrocyte derived CFU-Fs; cell population proliferation and regulation of vascular development (Figure 7e; Mi et al., 2019; Gene Ontology Consortium, 2021; Ashburner et al., 2000). Regulation of vascular development was interesting, although not a surprising finding, since we know that terminal hypertrophic chondrocytes secrete factors to initiate vessel invasion from the periosteum and we had identified a number of marrow-associated hypertrophic chondrocyte derived SSPCs localized near blood vessels. The fact that the tdTOMATO⁺ or hypertrophic chondrocyte derived CFU-Fs were enriched in genes associated with cellular proliferation was unanticipated. To this end, we investigated EdU incorporation into all CFU-Fs from Col10a1Cre; Rosa26^fs-tdTomato mice at day 5 in culture that consisted of at least 5 cells. We observed a greater percentage of cells in tdTOMATO⁺ CFU-Fs that had incorporated EdU as compared to tdTOMATO^- CFU-Fs (Figure 7f - top). By day 10 in culture, tdTOMATO⁺ CFU-Fs on average were larger and displayed more colonies with 200+ cells as compared to tdTOMATO^- CFU-Fs (Figure 7f - bottom). To determine whether this proliferative advantage could be observed in SSPCs immediately harvested from Col10a1Cre; Rosa26^fs-tdTomato bone marrow, we performed cell cycle analysis on our scRNA-seq dataset and found that as a percentage of the population there was an equal or greater percentage of tdTOMATO⁺ SSPCs (cluster 3, 4, and 5) in a proliferative state (S, G2, or M phase) as compared to osteoblasts (clusters 1 and 2) (Figure 7g). This unexpectedly high proliferative capacity for hypertrophic chondrocyte-derived SSPCs may be reflective of the fact that hypertrophic chondrocytes eventually re-enter the cell cycle as they transition fates within the marrow space (Park et al., 2015; Yang et al., 2014a).

To determine whether tdTOMATO⁺ and tdTOMATO^- CFUs contained inherent differences in their functional capacities to undergo osteogenic and adipogenic differentiation in vitro, we performed CFU-osteoblastic (CFU-Ob) and CFU-adipogenic (CFU-Ad) assays. Colonies composed of cells from either hypertrophic or non-hypertrophic origins were able to differentiate at similar capacities in vitro when measured by Alkaline Phosphatase, Von Kossa, and Oil Red O staining (Figure 8—figure supplement 1). However, when we performed kidney capsule transplantations to generate bone ossicles in vivo, there was a much more apparent functional difference between tdTOMATO⁺ and tdTOMATO^- cells. The tdTOMATO⁺ cells were able to generate a full ossicle with bone marrow, vessels, and adipocytes (Figure 8a -top and Figure 8b). Conversely, the tdTOMATO^- cells could only generate bone with no vascular invasion or adipocyte generation (Figure 8 -bottom and Figure 8b). These apparent differences when transplanted suggest that hypertrophic chondrocyte-derived SSPCs are much more supportive of vascular invasion and marrow formation than SSPCs derived from other cell sources.

Figure 8 with 1 supplement see all

Download asset Open asset

Kidney capsule transplantations of tdTOMATO⁺ cells exhibit complete ossicle formation with bone, adipocytes, and bone marrow compared to tdTOMATO^- resulting in only bone formation.

(a) Immunofluroescent stain for bone (osteopontin), adipocytes (perilipin), and vessels (endomucin) for tomato⁺ transplant (top) and tomato^- transplant (bottom). (b) Hematoxylin and Eosin stain for marrow establishment in tdTomato⁺ transplant and tdTOMATO^- transplant.

Discussion

Recent and historic data have suggested that hypertrophic chondrocytes are capable of becoming osteoblasts that directly contribute to trabecular bone formation (Giovannone et al., 2019; Park et al., 2015; Tsang et al., 2015; Yang et al., 2014b; Yang et al., 2014a; Zhou et al., 2014b); however, the process by which hypertrophic chondrocytes generate osteoblasts remains unknown. Current genetic lineage tracing studies have suggested a transdifferentiation mechanism by which hypertrophic chondrocytes differentiate directly into an osteoprogenitor or mature osteoblast without transiting through an intermediate or more primitive state (Park et al., 2015). Here, we have provided genetic and transcriptomic evidence, consistent with our hypothesis, that hypertrophic chondrocytes undergo a process of dedifferentiation to form SSPC populations ultimately giving rise to osteoblastic and adipogenic cell fates within the marrow compartment of embryonic, postnatal, and adult bones. A number of genetic and flow cytometric approaches have been used to identify and isolate multipotent SSPC populations within both the bone marrow and periosteal compartments. Many of these efforts have used alternative genetic models (Gremlin1Cre, LeprCre, Col2a1Cre, Cxcl12Cre) with lineage tracing and/or RNA-seq analyses or have utilized flow cytometric-based selection methods with antibodies against PDGFRα, CD200, and ITGAV among others followed by differentiation assays (Boulais et al., 2018; Chan et al., 2015; Farahani and Xaymardan, 2015; Gulati et al., 2018; Mizuhashi et al., 2019; Ono et al., 2014; Worthley et al., 2015; Zhou et al., 2014a). Generally, these studies have identified the presence of non-hematopoietic, non-endothelial multipotent mesenchymal cell populations with at least the capacity for osteogenic and/or adipogenic differentiation (Robey et al., 2014). It has also largely been assumed that these marrow-associated SSPCs are derived from the perichondrial/periosteal cells during skeletal development. Our study demonstrates that hypertrophic chondrocytes of the growth plate serve as a reservoir and alternative source of marrow-associated SSPCs likely utilized for rapid bone growth during skeletal development (Figure 9).

Figure 9

Download asset Open asset

Revised model for the generation of trabecular/endocortical bone associated osteoblasts/osteocytes from multiple sources.

(a) Perichondrial or periosteal SSPCs/osteoprogenitors (pSSPC/pOP) migrate into the marrow utilizing blood vessels and possess the ability to further differentiate into osteoblasts/osteocytes (OB/OC). (**b–c**) Hypertrophic chondrocytes (HC) dedifferentiate into chondrocyte derived SSPCs (cSSPC) with the capacity to generate osteoblasts/osteocytes (b) and adipocytes (Adipo) (c). (d) Data presented does not rule out the potential for the transdifferentiation of hypertrophic chondrocytes directly into osteoblasts/osteocytes. HSC = hematopoietic stem cells.

Specifically, our analyses of Col10a1CreERT2; Rosa26^fs-tdTomato and Col10a1Cre; Rosa26^fs-tdTomato skeletal rudiments at E16.5 identified not only hypertrophic chondrocytes (Col10a1 expressing cells) and their descendant osteoblasts (Bglap expressing cells), but also hypertrophic chondrocyte derived SSPCs expressing Pdgfra, Ly6a, Lepr, and Cxcl12. A population of lineage negative, CXCL12-, PDGFRα+, SCA1+ SSPCs (PαS cells) have been identified as a quiescent population associated with vessels near the endosteum (Morikawa et al., 2009). These PαS cells are also capable of forming CFU-Fs, can differentiate into osteoblasts, chondrocytes, and adipocytes, while also being important for the migration and maintenance of HSCs (Morikawa et al., 2009; Nusspaumer et al., 2017). The appearance of these cells at E16.5 (but not 2 months) suggests that these might be either a more transient population derived from hypertrophic chondrocytes or are of an initial finite number and contribute early in development and are lost with age, as opposed to hypertrophic chondrocyte derived CXCL12-abundant reticular (CAR) cells (Cxcl12⁺; Ly6a1^-) that continue to be present in our data at 2 months of age.

Our analyses of 2-month Col10a1Cre; Rosa26^fs-tdTomato bone marrow identified not only hypertrophic chondrocyte-derived osteoprogenitors (Sp7 expressing) (Park et al., 2015) and osteoblast populations, but also hypertrophic chondrocyte-derived marrow and vessel-associated cells expressing Lepr, Pdgfra, Pdgfrb, Grem1, Cxcl12, with several expressing Itgav, Cd200, and Eng (CD105) while lacking the expression of Col10a1, Cre, and most osteogenic genes. Since Runx2, Pparg, and Adipoq are also expressed throughout many of these cells at 2 months of age, we postulate that these hypertrophic chondrocyte-derived cells encompass populations of SSPCs with the capability of osteogenic and adipogenic differentiation. This indeed was confirmed via our in vitro and in vivo differentiation assays. Prior work has demonstrated that Runx2 expressing and LEPR⁺ progenitors are primed for osteoblastic differentiation (Yang et al., 2017), while LEPR⁺ progenitors expressing Adipoq and Pparg are associated with adipogenic differentiation (Zhong et al., 2020). Priming of the SSPCs may allow for their rapid differentiation when in the proper supportive osteogenic/adipogenic niche or outside a niche that supports SSPC maintenance.

Since SSPC populations, including CAR cells, were thought to originate from perichondrial/periosteal cell sources and since LEPR+ cells constitute ~100% of CFU-Fs, we attempted to compare SSPCs/CAR cells derived from hypertrophic chondrocytes to those of other origins by utilizing Col10a1Cre; Rosa26^fs-tdTomato bone marrow cells plated at clonal density (Zhou et al., 2014a). Nearly 50% of the CFU-Fs were originally derived from hypertrophic chondrocytes (tdTOMATO+), while the other half of CFU-Fs (tdTOMATO-) were likely derived from a perichondrial/periosteal origin. Additionally, when compared to SSPCs derived from other cell sources, the hypertrophic chondrocyte derived SSPCs were much more conducive to the formation of a bone ossicle with adipocytes and marrow upon transplantation, highlighting the supportive nature of these cells to vascular invasion and hematopoietic cell populations. Collectively, these data indicate for first time that growth plate hypertrophic chondrocytes may serve as a reservoir for the generation of SSPCs (including PαS cells) and osteoblasts during early embryonic development, but then primarily shift to CAR cell populations as the mice proceed through postnatal development and homeostasis.

Assessments of 2 M bulk RNA-seq data and EdU incorporation assays from CFU-F cultures, as well as cell cycle analysis of our scRNA-seq data, indicate an enhanced proliferative capacity for hypertrophic chondrocyte derived CFU-Fs and SSPCs/CAR cells. This data may be reflective of the fact that Yang et al. also identified the presence of both hypertrophic chondrocyte-derived Col1a1⁺ and Col1a1^- marrow-associated cells that were BrdU⁺ proliferating cells, indicating that hypertrophic chondrocytes eventually re-enter the cell cycle during this cell fate transition (Park et al., 2015; Yang et al., 2014a). Interestingly, our scRNA-seq dataset identified similar proportions of cycling cells in hypertrophic chondrocyte derived SSPCs/CAR cells and osteoblasts. This is in stark contrast to the more quiescent phenotype observed in most primitive SSPC populations that have been highlighted across numerous studies (Tikhonova et al., 2019; Zhong et al., 2020). However, recent work utilizing similar scRNA-seq approaches assessed proliferation rates of chondrocyte and perichondrial-derived SSPCs and osteoblast populations in the Col2a1Cre; Rosa26^fs-tdTomato bone marrow and also indicated similar proliferative capacities between later committed progenitors and osteoblasts, which was greater than the most primitive of SSPCs (Zhong et al., 2020). Zhong et al., also utilized the AdipoqCre^ERT2; Rosa26^fs-tdTomato model to investigate colony forming and proliferative potential of broad populations of Adipoq-expressing cells and descendants. These tdTOMATO⁺ cells contributed very little to CFU-Fs and exhibited minimal proliferation in vitro (Zhong et al., 2020). Tikhonova et al. also utilized scRNA-seq and identified a subset of cells from the LeprCre; Rosa26^fs-tdTomato lineage that cluster into a high cycling group (Tikhonova et al., 2019), which may represent a more proliferative LEPR⁺ SSPC population derived from the cartilage growth plate. Finally, it is worth noting again that the in vivo contribution of AdipoqCre; Rosa26^fs-tdTomato and LeprCre; Rosa26^fs-tdTomato lineages to osteoblasts and/or osteocytes is something that occurs over developmental time postnatally, while the Col10a1Cre; Rosa26^fs-tdTomato lineage contributes to the osteoblast and osteocyte fates rapidly; as early as the embryonic period of skeletal development. This may also be attributed to the rapid or transient nature by which hypertrophic chondrocytes transit through the SSPC phase during these periods, such that the LeprCre or AdipoqCre do not have sufficient time or levels of Cre expression to induce recombination within this population. Also note that we do not observe Adipoq expression in the transitional cells at E16.5. Therefore, LeprCre and AdipoqCre may only be efficient at recombining floxed alleles within the less transient SSPC populations postnatally that continue to be maintained as SSPCs and subsequently slowly/weakly contribute to the osteogenic and adipogenic lineages during late postnatal/adult skeletal development and homeostasis.

While the data presented here demonstrates hypertrophic chondrocyte contribution to SSPCs, we note the importance of further studies to elucidate the mechanisms driving this process. Recent publications have suggested roles for pathways like WNT/β-catenin in balancing osteogenesis and adipogenesis from hypertrophic chondrocytes (Tan et al., 2020). Additionally, Sox9 was shown to be important for maintaining chondrocytes before changing fates toward the osteoblast lineage (Haseeb et al., 2021). Future studies will need to take into account the potential different signaling pathways that drive the dedifferentiation of hypertrophic chondrocytes to a SSPC fate versus the re-differentiation of hypertrophic chondrocyte derived SSPCs toward the osteogenic and adipogenic lineages. Additional studies have also demonstrated a potential transient embryonic and postnatal contribution of hypertrophic chondrocytes to developmental bone formation, since genetic deletion of Runx2 within hypertrophic chondrocytes (and their descendants) initially reduces osteoblastic bone formation that ultimately fully recovers around 3–6 weeks of age (Qin et al., 2020). While it is possible that the late postnatal formation is ‘recovered’ from osteoblasts derived purely from an alternate source (such as the periosteum), it is also possible that removal of Runx2 within the hypertrophic chondrocyte lineages induces some hypertrophic chondrocyte cell death, alters angiogenic/osteogenic factors (VEGFa among others) secreted from hypertrophic chondrocytes, and/or delays the re-differentiation of hypertrophic chondrocyte-derived SSPCs toward the osteoblast lineage; thus simply delaying postnatal bone formation by reducing the contribution of hypertrophic chondrocytes and their descendants to the earliest stages of bone development. Further studies are needed to more thoroughly understand these lineage contributions over time, as we demonstrate with the Col10a1CreERT2; Rosa26^fs-tdTomato mouse line that there is a continued contribution of osteoblastic cells from hypertrophic chondrocytes throughout embryonic development to at least approximately 4–5 months of age or skeletal maturity.

Materials and methods

Mouse strains

Request a detailed protocol

The Col10a1Cre mouse line was a generous gift from Dr. Klaus von der Mark and has been previously described (Park et al., 2015). The Col10a1CreERT2 mouse line was a generous gift from Dr. Kathryn Cheah (University of Hong Kong). The Rosa26^fs-tdTomato, Pdgfra^H2B-GFP, LeprCre, and AdipoqCre mouse lines were obtained from Jackson Laboratories (007909, 007669, 008320, and 028020, respectively). All animal work was approved by Duke University Institutional Animal Care and Use Committees (IACUC). Both sexes of mice were utilized for experiments.

Tissue analysis

Request a detailed protocol

Harvest and sectioning: Animals were harvested at embryonic day 16.5 and postnatal day 0, 2–4 months, and 13 months of age. Tissues were washed in 1 x PBS and then fixed overnight with 4% paraformaldehyde (PFA - Sigma). P0 animals were treated overnight with 14% EDTA (Sigma), while 2 M old animals were treated for 10 days. Animals used for PERILIPIN labeling were not decalcified. Tissues were then transferred to 30% sucrose (Sigma) for 2 days. Samples were imbedded in Shandon Cryomatrix (ThermoFisher Scientific) and cryosectioned at 10 um, and slides were stored at –20 °C until staining.

For ENDOMUCIN and LEPR staining: Samples were prepared for staining as published (Kusumbe et al., 2015) 2 M animals were harvested and fixed overnight in 4% PFA. Samples were decalcified overnight in 14% EDTA. Tissues were transferred to 20% sucrose and 2% polyvinylpyrrolidone (PVP - Sigma) overnight. Samples were embedded in 8% gelatin (Sigma), 2% PVP, and 20% sucrose and cryosectioned at 30–200 um with slides stored at –20°C until staining.

Staining: Slides were brought to room temperature and rehydrated in PBS. Antigen retrieval was performed as shown in Appendix 1. Blocking was performed in 1% goat serum or 3% BSA. Slides were incubated with antibodies at concentrations in Appendix 1 overnight at 4 °C in a humidified dark box. Slides were washed and incubated with secondary antibodies (Invitrogen catalog number in Appendix 1 1:200) for 30–60 min at room temperature in the dark. Slides were imaged on a Leica DMI3000B microscope using a Leica DFC3000G camera with the latest Leica imaging suite, a Zeiss 710 or Zeiss 780 inverted confocal microscope running the latest ZEN software. Images were reconstructed and pseudocolored on Fiji and quantified using built-in quantification measures.

Single-cell RNA sequencing

Request a detailed protocol

Cell isolation E16.5: Single cells were isolated by removing soft tissue on mechanical dissociation using scissors and then placed in 15 mL conical tubes with 2 mg/mL collagenase type II (Gibco 17101–015) in aMEM +20% FBS rotating at 37 °C. A pipet was used for further mechanical dissociation at 1- and 2 hr. Cells were removed at 2 hrs and subjected to red blood cell lysis for 30 s. Cells were stained with DAPI for 30 min on ice. Cells were washed with PBS and placed in PBS + 2% FBS for fluorescent activated cell sorting (FACS) on a B-C Astrios cell sorter. Gating was performed for single cells, DAPI, and tdTOMATO as shown in Figure 2—figure supplement 1a (Figure generated using FlowJo version 10). Cells were sorted into PBS + 10% FBS.

Cell isolation 2M-old: Single cells were isolated from bisected tibia and femur of 2 M Col10a1Cre; Rosa26^fs-tdTomato mice centrifuged at 4 °C twice pulsed up to 5000 x g. A 27 G needle attached to a syringe was then used to flush through the centrifuged bones. Bones and marrow were placed in 3 mg/mL of Collagenase D (Roche 11088866001) for 30 min in hypoxia. Red blood cells were lysed and strained through a 70 µm filter. Cells were stained with DAPI for 30 min on ice. Cells were washed with PBS and placed in PBS + 2% FBS for fFACS on a B-C Astrios cell sorter. Gating was performed for single cells, DAPI, and tdTOMATO as shown in Figure 2—figure supplement 1b (Figure generated using FlowJo version 10). Cells were sorted into PBS + 10% FBS.

10 x Transcriptome library prep: Cells were transferred to the Duke Molecular Physiology Institute Molecular Genomics core where they were combined with a master mix that contained reverse transcription reagents. The gel beads carrying the Illumina TruSeq Read one sequencing primer, a 16 bp 10 x barcode, a 12 bp unique molecular identifier (UMI) and a poly-dT primer were loaded onto the chip, together with oil for the emulsion reaction. The Chromium Controller partitions the cells into nanoliter-scale gel beads in emulsion (GEMS) within which reverse-transcription occurs. All cDNAs within a GEM, that is from one cell, share a common barcode. After the RT reaction, the GEMs were broken and the full length cDNAs cleaned with both Silane Dynabeads and SPRI beads. After purification, the cDNAs were assayed on an Agilent 4,200 TapeStation High Sensitivity D5000 ScreenTape for qualitative and quantitative analysis.

Enzymatic fragmentation and size selection were used to optimize the cDNA amplicon size. Illumina P5 and P7 sequences, a sample index, and TruSeq read two primer sequence were added via End Repair, A-tailing, Adaptor Ligation, and PCR. The final libraries contained P5 and P7 primers used in Illumina bridge amplification. Sequence was generated using paired end sequencing (one end to generate cell specific, barcoded sequence and the other to generate sequence of the expressed poly-A tailed mRNA) on an Illumina NextSeq500 with 400 M read per run.

Analysis: The primary analytical pipeline for the SC analysis followed the recommended protocols from 10 X Genomics. Briefly, we demultiplexed raw base call (BCL) files generated by Illumina sequencers into FASTQ files, upon which alignment to the mouse reference transcriptome, filtering, barcode counting, and UMI counting were performed using the most current version of 10 X’s Cell Ranger software. We used the Chromium cell barcode to generate feature-barcode matrices encompassing all cells captured in each library.

E16.5 timepoint: We sequenced a median of 4,868 genes per cell, a mean of 107,918 reads per cell, and reached a sequencing saturation of 44.6%. The secondary statistical analysis was performed using the latest R package of Seurat 4 (Hao et al., 2020). In Seurat, data was first normalized and scaled after basic filtering for minimum gene and cell observance frequency cut-offs ( > 200 genes, 15% mitochondrial genes, genes expressed in >5 cells). We thresholded tdTomato expression at two or higher to exclude a minimal number of cells that were separated via FACS but were tdTomato negative by gene expression. We observed some separation of clusters based on cell cycle and regressed out the cell cycle score during default SCTransform (Hafemeister and Satija, 2019).

After quality control procedures were complete, we performed linear dimensional reduction calculating principal components using the most variably expressed genes in our dataset (2000 variable genes, npcs = 35). Significant principal components for downstream analyses are determined through methods mirroring those implemented by Macosko et al, and these principal components were carried forward for two main purposes: to perform cell clustering and to enhance visualization (Macosko et al., 2015). Cells were grouped into an optimal number of clusters for de novo cell type discovery using Seurat’s FindNeighbors and FindClusters functions (resolution = 0.5) graph-based clustering approach with visualization of cells being achieved through the use of UMAP, which reduces the information captured in the selected significant principal components to two dimensions. Differential expression of relevant genes was visualized on UMAP plots to reveal specific individual cell types. Additional downstream analyses included examining the cellular distribution and differential gene expression of a priori genes of interest and closer examination of genes associated within the identified 8 cell clusters. Monocle analysis was performed using the Monocle3 package after cds was generated from Seurat using SeuratWrappers (Cao et al., 2019; Qiu et al., 2017a; Qiu et al., 2017b; Satija et al., 2020; Trapnell et al., 2014). GEO accession number GSE190616.

Two-month timepoint: We sequenced a median of 2,820 genes per cell, a mean of 261,899 reads per cell, and a sequencing saturation of 92.3%. The secondary statistical analysis was performed using the last R package of Seurat 3 (Butler et al., 2018; Konopka, 2019; R Development Core Team, 2013; Stuart et al., 2019; Wickham et al., 2020). In Seurat, data was first normalized and scaled after basic filtering for minimum gene and cell observance frequency cut-offs (200–4000 genes, 7% mitochondrial genes, genes expressed in >5 cells). We then closely examined the data and performed further filtering based on a range of metrics in an attempt to identify and exclude possible multiplets (i.e. instances where more than one cell was present and sequenced in a single emulsified gel bead). We thresholded tdTomato expression at two or higher to exclude a minimal number of cells that were separated via FACS but were tdTomato negative by gene expression.

After quality control procedures were complete, we performed linear dimensional reduction calculating principal components using the most variably expressed genes in our dataset (2000 variable genes, npcs = 25). The genes underlying the resulting principal components are examined in order to confirm they are not enriched in genes involved in cell division or other standard cellular processes (subsetting out percent.mito). Significant principal components for downstream analyses are determined through methods mirroring those implemented by Macosko et al, and these principal components were carried forward for two main purposes: to perform cell clustering and to enhance visualization (Macosko et al., 2015). Cells were grouped into an optimal number of clusters for de novo cell type discovery using Seurat’s FindNeighbors and FindClusters functions (resolution = 0.4) graph-based clustering approach with visualization of cells being achieved through the use of UMAP, which reduces the information captured in the selected significant principal components to two dimensions. Differential expression of relevant genes was visualized on UMAP plots to reveal specific individual cell types. Additional downstream analyses included examining the cellular distribution and differential gene expression of a priori genes of interest and closer examination of genes associated within the identified 5 cell clusters. Cell cycle analysis of the scRNA-seq dataset was preformed using a gene list and commands packaged in Seurat (Tirosh et al., 2016; Wickham, 2019). Monocle analysis was performed using the Monocle3 package (num_dim = 25) (Cao et al., 2019; Qiu et al., 2017a; Qiu et al., 2017b; Trapnell et al., 2014). GEO accession number GSE179148.

Colony-forming unit assays

Request a detailed protocol

The Colony Forming Unit – Fibroblastic (CFU-F) assay was performed as published, with the exception of placing cells in hypoxia based on a recent publication demonstrating reduction of contaminating cells (Guo et al., 2020; Robey et al., 2014). In short, cells were isolated from bisected tibia and femur of 2-month-old Col10a1Cre; R26-tdTomato mice centrifuged at 4°C twice pulsed up to 5 K xG. Following red blood cell lysis (Roche), cells were plated at 750 K cells per 15 cm tissue culture plate in complete media (a-MEM, 20% FBS, 1% penicillin-streptomycin) and glutamax (Gibco) after straining through a 70 µm filter. After incubation in hypoxia (2% oxygen, 5% CO2, 37 C) for 3 hr, plates were washed three times with Hanks Balanced Salt Solution (Gibco) and replaced with complete media and glutamax. Media was changed on days 3 and 6. Colonies were utilized for analysis using a Leica DMI3000B microscope using a Leica DFC3000G camera with the latest Leica imaging suite, bulk RNA sequencing, or EdU labeling. TOMATO fluorescence was used to identify positive and negative colonies.

For CFU-OB, osteogenic differentiation was initiated after day 10 colony growth by replacing media with aMEM supplemented 10% FBS, 1% penicillin-streptomycin, 50 µg/mL L-ascorbic acid (Sigma-A4544), and 10 mM B-glycerophosphate (Sigma-G9422). Alkaline phosphatase and Von Kossa (Ricca) staining were performed per general histological procedures. For CFU-AD, adipogenic diffeentiation was performed as published; replacing media with aMEM supplemented with 10% FBS, 1% penicillin-streptomycin, 1.7 uM Insulin (Sigma), 1 µM Dexamethasone (Sigma), and 0.5 mM IBMX (Sigma) at day 10 colony growth (Yu et al., 2019). Oil red O (Sigma) staining was performed per general histological procedures.

Bulk RNA sequencing

Request a detailed protocol

Colonies were imaged using a Leica DMI3000B microscope using a Leica DFC3000G camera with the latest Leica imaging suite to identify TOMATO positive and negative colonies grown as in the CFU-F assay section to day 10. Colonies were prepared using Corning 10 mm cloning cylinders and vacuum grease to provide single colony isolation. RLT buffer was pipetted directly into the cloning cylinder and then transferred to ethanol for processing using the RNeasy Micro kit (QIAGEN). RNA was provided to the Duke Center for Genomic and Computational Biology core for quality assessment and sequenced 50 bp single-end read on an Illumina HiSeq4000.

RNA-seq data was processed using the TrimGalore (https://github.com/FelixKrueger/TrimGalore) toolkit which employs Cutadapt to trim low-quality bases and Illumina sequencing adapters from the 3’ end of the reads (Martin, 2011). Only reads that were 20nt or longer after trimming were kept for further analysis. Reads were mapped to the GRCm38v73 version of the mouse genome and transcriptome using the STAR RNA-seq alignment tool (Dobin et al., 2013; Kersey et al., 2012). Reads were kept for subsequent analysis if they mapped to a single genomic location. Gene counts were compiled using the HTSeq tool (Anders et al., 2015). Only genes that had at least 10 reads in any given library were used in subsequent analysis. Normalization and differential gene expression was carried out using the DESeq2 Bioconductor package with the R statistical programming environment (Huber et al., 2015; Love et al., 2014; R Development Core Team, 2013). The false discovery rate was calculated to control for multiple hypothesis testing. Biological thresholds were determined by taking the log2 expression value plotted on a histogram. A bimodal distribution is observed for low expressed genes followed by a larger Guassian distribution. The split point between these was used as a value cutoff for biological relevance. GEO accession number GSE179174.

Gene ontology analysis

Request a detailed protocol

Gene ontology analysis was performed using geneontology.org Analysis Type: PANTHER Overrepresentation test released 02-24-2021. Annotation version and release date: release date 01-01-2021 and version 10.5281/zenodo.4437524 for biological processes using Mus musculus as the reference. Bulk RNA sequencing data was organized into log fold changes > 1 and p-value < 0.05 for input into the enrichment analysis. A negative log fold change in the dataset represents higher expression in TOMATO positive colonies while positive log fold change represents higher expression in TOMATO-negative colonies. Test type: Fisher’s exact. Correction: Calculate False Discovery Rate (Mi et al., 2019; Gene Ontology Consortium, 2021; Ashburner et al., 2000).

EdU incorporation assay

Request a detailed protocol

Colonies were grown as in CFU-F assays (in a six-well plate at 500 K cells per well instead of 15 cm plate) until day 5. EdU staining was performed following the protocol of Click-iT Plus EdU Cell Proliferation Kit (Invitrogen – C10637). In short, EdU was added to growth media to allow a half media change of the cells. The cells were incubated at in hypoxia at 37 °C for 4 hours. Cells were fixed, permeabilized, and treated with staining solution. Cells were imaged using a Zeiss Axio Observer Z1 microscope at ×5 magnification tiled across 70% of each well and stitched together using the latest Zen Blue software. Number of cells per colony calculations were performed at day 10 of plates grown in parallel to the EdU day 5 plate.

Cell transplantations and ossicle formation assays

Request a detailed protocol

Cells from Col10a1Cre; Rosa26^fs-tdTomato mice at age 2 months were isolated as in Colony-Forming Unit Assays above. Cells were plated at high density on 10 cm plates for 6 days at 37 °C in hypoxia. Cells were trypsinized from the plates and were stained with DAPI for 30 min on ice. Cells were washed with PBS and placed in PBS + 2% FBS for FACS. Gating was performed for single cells, DAPI, and tdTOMATO similar to Figure 2—figure supplement 1. Cells that were tdTOMATO⁺ and tdTOMATO^- were isolated from each of 5 plates (each from an individual mouse) and transplanted into the kidney capsule (one transplant per mouse – two male/3 female) after being resuspended and absorbed into a Gelfoam matrix at 200–300 k cells per transplant. Transplants were isolated at 8 weeks post transplantation, washed in 1 x PBS, and then fixed overnight with 4% paraformaldehyde (PFA - Sigma). The samples were then decalcified overnight with 14% EDTA (Sigma). Tissues were then transferred to 30% sucrose (Sigma) for 2 days. Samples were imbedded in Shandon Cryomatrix (ThermoFisher Scientific) and cryosectioned at 20 µm, and slides were stored at –20 °C until staining. Staining was performed as above in Tissue Analysis.

Flow cytometry

Request a detailed protocol

Following isolation techniques noted in the Colony Forming Unit Assay methods above, cells were resuspended in complete media and stained for 30–45 min in conjugated antibodies (noted in above table). Cells were spun down following staining and resuspended in PBS + 2% FBS. Flow cytometry was performed on a BD FACSCanto II machine.

Statistical analysis

Request a detailed protocol

Statistical analyses were performed using two-tailed, unpaired Student’s t-test. A p-value < 0.05 was considered significant.

Appendix 1

Appendix 1—key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Strain, strain background (Mus musculus)	C57BL/6 J	Jackson Laboratory	RRID:IMSR_JAX:000664
Genetic reagent (Mus musculus)	Col10a1Cre	DOI: 10.1242/bio.201411031		Dr. Klaus von der Mark
Genetic reagent (Mus musculus)	Col10a1CreERT2	This paper		Dr. Kathryn Cheah
Genetic reagent (Mus musculus)	Rosa26^fs-tdTomato	Jackson Laboratory	RRID:IMSR_JAX:007909
Genetic reagent (Mus musculus)	Pdgfra^H2B-GFP	Jackson Laboratory	RRID:IMSR_JAX:007669
Genetic reagent (Mus musculus)	LeprCre	Jackson Laboratory	RRID:IMSR_JAX:008320
Genetic reagent (Mus musculus)	AdipoqCre	Jackson Laboratory	RRID:IMSR_JAX:028020
Sequence-based reagent	Col10a1Cre/ ERT2_F	DOI: 10.1242/bio.201411031		TTTAGAGCATTATT TCAAGGCA GTTTCCADr. Klaus von der Mark
Sequence-based reagent	Col10a1Cre/ ERT2_R	DOI: 10.1242/bio.201411031		AGGCAAATCTT GGTGTACGGDr. Klaus von der Mark
Antibody	ENDOMUCIN (rat mAb)	Abcam	Cat# ab106100, RRID:AB_10859306	IF (1:100), 0.3% Triton X-100 in 1 x PBS
Antibody	PERILIPIN (rabbit mAb)	Cell Signaling	Cat# 9349, RRID:AB_10829911	IF (1:100)
Antibody	OSTEOCALCIN (rabbit pAb)	Millipore	Cat# AB10911, RRID:AB_1587337	IF (1:200), 10 ug/mL Proteinase K
Antibody	OSTERIX(Rabbit pAb)	Abcam	Cat# ab22552, RRID:AB_2194492	IF (1:400)
Antibody	CRE(Rabbit mAb)	Cell Signaling	Cat# 15036, RRID:AB_2798694	IF (1:100), 0.3% Triton X-100 in 1 x PBS
Antibody	RFP (rabbit pAb)	Abcam	Cat# ab62341, RRID:AB_945213	IF (1:100)
Antibody	LEPR (goat pAb)	R&D Systems	Cat# AF497, RRID:AB_2281270	IF (1:50), 0.3% Triton X-100 in 1 x PBS
Antibody	PDGFRa (goat mAb)	R&D Systems	Cat# AF1062, RRID:AB_2236897	IF (1:50), 0.3% Triton X-100 in 1 x PBS
Antibody	PDGFRb (rat mAb)	Invitrogen	Cat# 14-1402-81, RRID:AB_467492	IF (1:50), 0.3% Triton X-100 in 1 x PBS
Antibody	Osteopontin (goat pAb)	R&D Systems	Cat# AF808, RRID:AB_2194992	IF (1:100), 0.3% Triton X-100 in 1 x PBS
Antibody	PDGFRb (rat mAb)	Invitrogen	Cat# 17-1402-80, RRID:AB_1548752	Flow Cytometry (1:25)
Antibody	CD45-FITC (rat mAb)	Biolegend	Cat# 103107, RRID:AB_312972	Flow Cytometry (1:1000)
Antibody	CD45-APC (rat mAb)	Biolegend	Cat# 103111, RRID:AB_312976	Flow Cytometry (1:1000)
Antibody	CD31-FITC (rat mAb)	Biolegend	Cat# 102405, RRID:AB_312900	Flow Cytometry (1:1000)
Antibody	CD31-APC (rat mAb)	Biolegend	Cat# 102409, RRID:AB_312904	Flow Cytometry (1:1000)
Antibody	TER119-FITC (rat mAb)	BD Biosciences	Cat# 561032, RRID:AB_10563083	Flow Cytometry (1:500)
Antibody	TER119-APC (rat mAb)	Biolegend	Cat# 116211, RRID:AB_313712	Flow Cytometry (1:250)
Antibody	DAPI (FLOW)	ThermoFisher	Cat# D1306, RRID:AB_2629482	Flow Cytometry (1:1000)
Antibody	Goat anti-Mouse Alexa Fluor 488 pAb	Invitrogen	Cat# A-11001, RRID:AB_2534069	IF (1:200)
Antibody	Donkey anti-Goat Alexa Fluor 488 pAb	Invitrogen	Cat# A-11055, RRID:AB_2534102	IF (1:200)
Antibody	Goat anti-Rat Alexa Fluor 488 pAb	Invitrogen	Cat# A-11006, RRID:AB_2534074	IF (1:200)
Antibody	Goat anti-Rabbit Alexa Fluor 594 pAb	Invitrogen	Cat# A-11037, RRID:AB_2534095	IF (1:200)
Antibody	Goat anti-Rabbit Alexa Fluor 488 pAb	Invitrogen	Cat# A-11034, RRID:AB_2576217	IF (1:200)
Antibody	Goat anti-Rat Alexa Fluor 647 pAb	Invitrogen	Cat# A-21247, RRID:AB_141778	IF (1:200)
Antibody	Goat anti-Rabbit Alexa Fluor 488 pAb	Invitrogen	Cat# A-11008, RRID:AB_143165	For LepRCre;tdTomato stained PERILIPIN only - IF (1:200)
Chemical compound, drug	Collagenase II	Gibco	17101–015
Chemical compound, drug	Collagenase D	Roche	11088866001
Chemical compound, drug	Polyvinylpyrrolidone	Sigma	P5288
Chemical compound, drug	Glutamax 100 x	Gibco	35050–061
Chemical compound, drug	L-Ascorbic Acid	Sigma	A4544
Chemical compound, drug	B-glycerophosphate	Sigma	G9422
Chemical compound, drug	Insulin Human	Sigma	I3536
Chemical compound, drug	IBMX	Sigma	I7018
Chemical compound, drug	Dexamethasone	Sigma	D2915
Chemical compound, drug	Antigen Unmasking solution, Citrate Acid Based	Vector	H3300
Commercial assay, kit	Click-iT Plus EdU Cell Proliferation Kit	Invitrogen	C10637
Software, algorithm	FIJI	doi:10.1038/ nmeth.2019	Fiji, RRID:SCR_002285
Software, algorithm	Seurat	CRAN	Seurat, RRID:SCR_007322
Software, algorithm	Monocle3	Github	Monocle3, RRID:SCR_018685
Software, algorithm	Trim Galore	Babraham Bioinformatics	Trim Galore,RRID:SCR_011847	Felix Krueger
Software, algorithem	Cutadapt	Github	cutadapt, RRID:SCR_011841
Software, algorithm	STAR RNA-seq Alignment	Github	STAR, RRID:SCR_004463
Software, algorithm	HTSeq tool	Github	HTSeq, RRID:SCR_005514
Software, algorithm	DESeq2	Bioconductor	DESeq2, RRID:SCR_015687
Software, algorithm	Gene Ontology	Geneontoloy.org	Gene Ontology, RRID:SCR_002811
Other	Silver Nitrate 10%	Ricca	6830–4
Other	Oil Red O	Sigma	O0625
Other	Alizarin Red	Sigma	A5533
Other	Mayer’s Hematoxylin solution	Electron Microscopy Sciences	26043–06
Other	Eosin Y with Phloxine staining solution	Electron Microscopy Sciences	26051–21

Data availability

All raw data has been made available as source data files within the manuscript. All sequencing datasets are available via the Gene Expression Omnibus (GEO) under the accession numbers: GSE179174, GSE190616, and GSE179148.

The following data sets were generated

1. Long JT
2. Leinroth A
3. Liao Y
4. Ren Y
5. Mirando AJ
6. Nguyen T
7. Guo W
8. Sharma D
9. Rouse D
10. Wu C
11. Cheah KS
12. Karner CM
13. Hilton MJ
(2022) NCBI Gene Expression Omnibus
ID GSE179174. Collagen10a1-Cre;Rosa26-tdTomato Bone marrow colony forming units.

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=gse179174
1. Long JT
2. Leinroth A
3. Liao Y
4. Ren Y
5. Mirando AJ
6. Nguyen T
7. Guo W
8. Sharma D
9. Rouse D
10. Wu C
11. Cheah KS
12. Karner CM
13. Hilton MJ
(2022) NCBI Gene Expression Omnibus
ID GSE190616. Collagen10-Cre;Rosa26-tdTomato e16.5 single cell RNA-sequencing.

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE190616
1. Long JT
2. Leinroth A
3. Liao Y
4. Ren Y
5. Mirando AJ
6. Nguyen T
7. Guo W
8. Sharma D
9. Rouse D
10. Wu C
11. Cheah KS
12. Karner CM
13. Hilton MJ
(2022) NCBI Gene Expression Omnibus
ID GSE179148. Collagen10-Cre;Rosa26-tdTomato bone marrow single cell RNA-sequencing.

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE179148

References

1. Anders S
2. Pyl PT
3. Huber W
(2015) HTSeq--a Python framework to work with high-throughput sequencing data
Bioinformatics (Oxford, England) 31:166–169.
https://doi.org/10.1093/bioinformatics/btu638
- PubMed
- Google Scholar
1. Ashburner M
2. Ball CA
3. Blake JA
4. Botstein D
5. Butler H
6. Cherry JM
7. Davis AP
8. Dolinski K
9. Dwight SS
10. Eppig JT
11. Harris MA
12. Hill DP
13. Issel-Tarver L
14. Kasarskis A
15. Lewis S
16. Matese JC
17. Richardson JE
18. Ringwald M
19. Rubin GM
20. Sherlock G
(2000) Gene Ontology: tool for the unification of biology
Nature Genetics 25:25–29.
https://doi.org/10.1038/75556
- Google Scholar
(2018) The Majority of CD45– Ter119– CD31– Bone Marrow Cell Fraction Is of Hematopoietic Origin and Contains Erythroid and Lymphoid Progenitors
Immunity 49:627–639.
https://doi.org/10.1016/j.immuni.2018.08.019
- Google Scholar
1. Butler A
2. Hoffman P
3. Smibert P
4. Papalexi E
5. Satija R
(2018) Integrating single-cell transcriptomic data across different conditions, technologies, and species
Nature Biotechnology 36:411–420.
https://doi.org/10.1038/nbt.4096
- PubMed
- Google Scholar
1. Cao J
2. Spielmann M
3. Qiu X
4. Huang X
5. Ibrahim DM
6. Hill AJ
7. Zhang F
8. Mundlos S
9. Christiansen L
10. Steemers FJ
11. Trapnell C
12. Shendure J
(2019) The single-cell transcriptional landscape of mammalian organogenesis
Nature 566:496–502.
https://doi.org/10.1038/s41586-019-0969-x
- PubMed
- Google Scholar
(1984) Location of 64K collagen producer chondrocytes in developing chicken embryo tibiae
Molecular and Cellular Biology 4:1163–1168.
https://doi.org/10.1128/mcb.4.6.1163-1168.1984
- PubMed
- Google Scholar
1. Chan CKF
2. Seo EY
3. Chen JY
4. Lo D
5. McArdle A
6. Sinha R
7. Tevlin R
8. Seita J
9. Vincent-Tompkins J
10. Wearda T
11. Lu W-J
12. Senarath-Yapa K
13. Chung MT
14. Marecic O
15. Tran M
16. Yan KS
17. Upton R
18. Walmsley GG
19. Lee AS
20. Sahoo D
21. Kuo CJ
22. Weissman IL
23. Longaker MT
(2015) Identification and specification of the mouse skeletal stem cell
Cell 160:285–298.
https://doi.org/10.1016/j.cell.2014.12.002
- PubMed
- Google Scholar
1. Chan CKF
2. Gulati GS
3. Sinha R
4. Tompkins JV
5. Lopez M
6. Carter AC
7. Ransom RC
8. Reinisch A
9. Wearda T
10. Murphy M
11. Brewer RE
12. Koepke LS
13. Marecic O
14. Manjunath A
15. Seo EY
16. Leavitt T
17. Lu W-J
18. Nguyen A
19. Conley SD
20. Salhotra A
21. Ambrosi TH
22. Borrelli MR
23. Siebel T
24. Chan K
25. Schallmoser K
26. Seita J
27. Sahoo D
28. Goodnough H
29. Bishop J
30. Gardner M
31. Majeti R
32. Wan DC
33. Goodman S
34. Weissman IL
35. Chang HY
36. Longaker MT
(2018) Identification of the Human Skeletal Stem Cell
Cell 175:43–56.
https://doi.org/10.1016/j.cell.2018.07.029
- PubMed
- Google Scholar
(2018) Isolation and differential transcriptome of vascular smooth muscle cells and mid-capillary pericytes from the rat brain
Scientific Reports 8:12272.
https://doi.org/10.1038/s41598-018-30739-5
- PubMed
- Google Scholar
1. Crisan M
2. Yap S
3. Casteilla L
4. Chen C-W
5. Corselli M
6. Park TS
7. Andriolo G
8. Sun B
9. Zheng B
10. Zhang L
11. Norotte C
12. Teng P-N
13. Traas J
14. Schugar R
15. Deasy BM
16. Badylak S
17. Buhring H-J
18. Giacobino J-P
19. Lazzari L
20. Huard J
21. Péault B
(2008) A perivascular origin for mesenchymal stem cells in multiple human organs
Cell Stem Cell 3:301–313.
https://doi.org/10.1016/j.stem.2008.07.003
- PubMed
- Google Scholar
1. Dobin A
2. Davis CA
3. Schlesinger F
4. Drenkow J
5. Zaleski C
6. Jha S
7. Batut P
8. Chaisson M
9. Gingeras TR
(2013) STAR: ultrafast universal RNA-seq aligner
Bioinformatics (Oxford, England) 29:15–21.
https://doi.org/10.1093/bioinformatics/bts635
- PubMed
- Google Scholar
1. Farahani RM
2. Xaymardan M
(2015) Platelet-Derived Growth Factor Receptor Alpha as a Marker of Mesenchymal Stem Cells in Development and Stem Cell Biology
Stem Cells International 2015:1–8.
https://doi.org/10.1155/2015/362753
- PubMed
- Google Scholar
1. Gene Ontology Consortium
(2021) The Gene Ontology resource: enriching a GOld mine
Nucleic Acids Research 49:D325–D334.
https://doi.org/10.1093/nar/gkaa1113
- PubMed
- Google Scholar
1. Giovannone D
2. Paul S
3. Schindler S
4. Arata C
5. Farmer DT
6. Patel P
7. Smeeton J
8. Crump JG
(2019) Programmed conversion of hypertrophic chondrocytes into osteoblasts and marrow adipocytes within zebrafish bones
eLife 8:e42736.
https://doi.org/10.7554/eLife.42736
- Google Scholar
1. Gulati GS
2. Murphy MP
3. Marecic O
4. Lopez M
5. Brewer RE
6. Koepke LS
7. Manjunath A
8. Ransom RC
9. Salhotra A
10. Weissman IL
11. Longaker MT
12. Chan CKF
(2018) Isolation and functional assessment of mouse skeletal stem cell lineage
Nature Protocols 13:1294–1309.
https://doi.org/10.1038/nprot.2018.041
- PubMed
- Google Scholar
1. Guo W
2. Spiller KV
3. Tang J
4. Karner CM
5. Hilton MJ
6. Wu C
(2020) Hypoxia Depletes Contaminating CD45+ Hematopoietic Cells from Bone Marrow Stromal Cell (BMSC) Cultures: Methods for BMSC Culture Purification
Cell Biology 53:102317.
https://doi.org/10.1016/j.scr.2021.102317
- Google Scholar
1. Hafemeister C
2. Satija R
(2019) Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression
Genome Biology 20:296.
https://doi.org/10.1186/s13059-019-1874-1
- Google Scholar
1. Hao Y
2. Hao S
3. Andersen-Nissen E
4. Mauck WM
5. Zheng S
6. Butler A
7. Lee MJ
8. Wilk AJ
9. Darby C
10. Zagar M
11. Hoffman P
12. Stoeckius M
13. Papalexi E
14. Mimitou EP
15. Jain J
16. Srivastava A
17. Stuart T
18. Fleming LB
19. Yeung B
20. Rogers AJ
21. McElrath JM
22. Blish CA
23. Gottardo R
24. Smibert P
25. Satija R
(2020) Integrated Analysis of Multimodal Single-Cell Data
Genomics 184:3573–3587.
https://doi.org/10.1016/j.cell.2021.04.048
- Google Scholar
1. Haseeb A
2. Kc R
3. Angelozzi M
4. de Charleroy C
5. Rux D
6. Tower RJ
7. Yao L
8. Pellegrino da Silva R
9. Pacifici M
10. Qin L
11. Lefebvre V
(2021) SOX9 keeps growth plates and articular cartilage healthy by inhibiting chondrocyte dedifferentiation/osteoblastic redifferentiation
PNAS 118:e2019152118.
https://doi.org/10.1073/pnas.2019152118
- PubMed
- Google Scholar
1. Hilton MJ
2. Tu X
3. Long F
(2007) Tamoxifen-inducible gene deletion reveals a distinct cell type associated with trabecular bone, and direct regulation of PTHrP expression and chondrocyte morphology by Ihh in growth region cartilage
Developmental Biology 308:93–105.
https://doi.org/10.1016/j.ydbio.2007.05.011
- PubMed
- Google Scholar
1. Hirao M
2. Tamai N
3. Tsumaki N
4. Yoshikawa H
5. Myoui A
(2006) Oxygen tension regulates chondrocyte differentiation and function during endochondral ossification
The Journal of Biological Chemistry 281:31079–31092.
https://doi.org/10.1074/jbc.M602296200
- PubMed
- Google Scholar
1. Huber W
2. Carey VJ
3. Gentleman R
4. Anders S
5. Carlson M
6. Carvalho BS
7. Bravo HC
8. Davis S
9. Gatto L
10. Girke T
11. Gottardo R
12. Hahne F
13. Hansen KD
14. Irizarry RA
15. Lawrence M
16. Love MI
17. MacDonald J
18. Obenchain V
19. Oleś AK
20. Pagès H
21. Reyes A
22. Shannon P
23. Smyth GK
24. Tenenbaum D
25. Waldron L
26. Morgan M
(2015) Orchestrating high-throughput genomic analysis with Bioconductor
Nature Methods 12:115–121.
https://doi.org/10.1038/nmeth.3252
- PubMed
- Google Scholar
1. Kersey PJ
2. Staines DM
3. Lawson D
4. Kulesha E
5. Derwent P
6. Humphrey JC
7. Hughes DST
8. Keenan S
9. Kerhornou A
10. Koscielny G
11. Langridge N
12. McDowall MD
13. Megy K
14. Maheswari U
15. Nuhn M
16. Paulini M
17. Pedro H
18. Toneva I
19. Wilson D
20. Yates A
21. Birney E
(2012) Ensembl Genomes: an integrative resource for genome-scale data from non-vertebrate species
Nucleic Acids Research 40:D91–D97.
https://doi.org/10.1093/nar/gkr895
- PubMed
- Google Scholar
1. Kong RY
2. Kwan KM
3. Lau ET
4. Thomas JT
5. Boot-Handford RP
6. Grant ME
7. Cheah KS
(1993) Intron-exon structure, alternative use of promoter and expression of the mouse collagen X gene, Col10a-1
European Journal of Biochemistry 213:99–111.
https://doi.org/10.1111/j.1432-1033.1993.tb17739.x
- PubMed
- Google Scholar
Software
1. Konopka T
(2019)
umap: Uniform Manifold Approximation and Projection

Umap.
(2015) Sample preparation for high-resolution 3D confocal imaging of mouse skeletal tissue
Nature Protocols 10:1904–1914.
https://doi.org/10.1038/nprot.2015.125
- PubMed
- Google Scholar
1. Kwan KM
2. Pang MKM
3. Zhou S
4. Cowan SK
5. Kong RYC
6. Pfordte T
7. Olsen BR
8. Sillence DO
9. Tam PPL
10. Cheah KSE
(1997) Abnormal compartmentalization of cartilage matrix components in mice lacking collagen X: implications for function
The Journal of Cell Biology 136:459–471.
https://doi.org/10.1083/jcb.136.2.459
- PubMed
- Google Scholar
1. Long F
2. Zhang XM
3. Karp S
4. Yang Y
5. McMahon AP
(2001) Genetic manipulation of hedgehog signaling in the endochondral skeleton reveals a direct role in the regulation of chondrocyte proliferation
Development (Cambridge, England) 128:5099–5108.
https://doi.org/10.1242/dev.128.24.5099
- PubMed
- Google Scholar
1. Long F
2. Ornitz DM
(2013) Development of the endochondral skeleton
Cold Spring Harbor Perspectives in Biology 5:a008334.
https://doi.org/10.1101/cshperspect.a008334
- PubMed
- Google Scholar
1. Love MI
2. Huber W
3. Anders S
(2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
Genome Biology 15:550.
https://doi.org/10.1186/s13059-014-0550-8
- PubMed
- Google Scholar
1. Macosko EZ
2. Basu A
3. Satija R
4. Nemesh J
5. Shekhar K
6. Goldman M
7. Tirosh I
8. Bialas AR
9. Kamitaki N
10. Martersteck EM
11. Trombetta JJ
12. Weitz DA
13. Sanes JR
14. Shalek AK
15. Regev A
16. McCarroll SA
(2015) Highly Parallel Genome-wide Expression Profiling of Individual Cells Using Nanoliter Droplets
Cell 161:1202–1214.
https://doi.org/10.1016/j.cell.2015.05.002
- PubMed
- Google Scholar
1. Maes C
2. Kobayashi T
3. Selig MK
4. Torrekens S
5. Roth SI
6. Mackem S
7. Carmeliet G
8. Kronenberg HM
(2010) Osteoblast precursors, but not mature osteoblasts, move into developing and fractured bones along with invading blood vessels
Developmental Cell 19:329–344.
https://doi.org/10.1016/j.devcel.2010.07.010
- PubMed
- Google Scholar
(2012) Hypoxia-driven pathways in bone development, regeneration and disease
Nature Reviews. Rheumatology 8:358–366.
https://doi.org/10.1038/nrrheum.2012.36
- PubMed
- Google Scholar
1. Martin M
(2011) Cutadapt removes adapter sequences from high-throughput sequencing reads
EMBnet.Journal 17:200.
https://doi.org/10.14806/ej.17.1.200
- Google Scholar
1. Mi H
2. Muruganujan A
3. Ebert D
4. Huang X
5. Thomas PD
(2019) PANTHER version 14: more genomes, a new PANTHER GO-slim and improvements in enrichment analysis tools
Nucleic Acids Research 47:D419–D426.
https://doi.org/10.1093/nar/gky1038
- PubMed
- Google Scholar
1. Mizoguchi T
2. Pinho S
3. Ahmed J
4. Kunisaki Y
5. Hanoun M
6. Mendelson A
7. Ono N
8. Kronenberg HM
9. Frenette PS
(2014) Osterix marks distinct waves of primitive and definitive stromal progenitors during bone marrow development
Developmental Cell 29:340–349.
https://doi.org/10.1016/j.devcel.2014.03.013
- PubMed
- Google Scholar
1. Mizuhashi K
2. Nagata M
3. Matsushita Y
4. Ono W
5. Ono N
(2019) Growth Plate Borderline Chondrocytes Behave as Transient Mesenchymal Precursor Cells
Journal of Bone and Mineral Research 34:1387–1392.
https://doi.org/10.1002/jbmr.3719
- PubMed
- Google Scholar
1. Morikawa S
2. Mabuchi Y
3. Kubota Y
4. Nagai Y
5. Niibe K
6. Hiratsu E
7. Suzuki S
8. Miyauchi-Hara C
9. Nagoshi N
10. Sunabori T
11. Shimmura S
12. Miyawaki A
13. Nakagawa T
14. Suda T
15. Okano H
16. Matsuzaki Y
(2009) Prospective identification, isolation, and systemic transplantation of multipotent mesenchymal stem cells in murine bone marrow
The Journal of Experimental Medicine 206:2483–2496.
https://doi.org/10.1084/jem.20091046
- PubMed
- Google Scholar
1. Nusspaumer G
2. Jaiswal S
3. Barbero A
4. Reinhardt R
5. Ishay Ronen D
6. Haumer A
7. Lufkin T
8. Martin I
9. Zeller R
(2017) Ontogenic Identification and Analysis of Mesenchymal Stromal Cell Populations during Mouse Limb and Long Bone Development
Stem Cell Reports 9:1124–1138.
https://doi.org/10.1016/j.stemcr.2017.08.007
- PubMed
- Google Scholar
(2000) Bone development
Annual Review of Cell and Developmental Biology 16:191–220.
https://doi.org/10.1146/annurev.cellbio.16.1.191
- PubMed
- Google Scholar
1. Omatsu Y
2. Sugiyama T
3. Kohara H
4. Kondoh G
5. Fujii N
6. Kohno K
7. Nagasawa T
(2010) The essential functions of adipo-osteogenic progenitors as the hematopoietic stem and progenitor cell niche
Immunity 33:387–399.
https://doi.org/10.1016/j.immuni.2010.08.017
- PubMed
- Google Scholar
1. Ono N
2. Ono W
3. Nagasawa T
4. Kronenberg HM
(2014) A subset of chondrogenic cells provides early mesenchymal progenitors in growing bones
Nature Cell Biology 16:1157–1167.
https://doi.org/10.1038/ncb3067
- PubMed
- Google Scholar
(2019) Stem and progenitor cells in skeletal development
Current Topics in Developmental Biology 133:1–24.
https://doi.org/10.1016/bs.ctdb.2019.01.006
- PubMed
- Google Scholar
(2015) Dual pathways to endochondral osteoblasts: a novel chondrocyte-derived osteoprogenitor cell identified in hypertrophic cartilage
Biology Open 4:608–621.
https://doi.org/10.1242/bio.201411031
- PubMed
- Google Scholar
1. Qin X
2. Jiang Q
3. Nagano K
4. Moriishi T
5. Miyazaki T
6. Komori H
7. Ito K
8. Mark K
9. Sakane C
10. Kaneko H
11. Komori T
(2020) Runx2 is essential for the transdifferentiation of chondrocytes into osteoblasts
PLOS Genetics 16:e1009169.
https://doi.org/10.1371/journal.pgen.1009169
- PubMed
- Google Scholar
1. Qiu X
2. Hill A
3. Packer J
4. Lin D
5. Ma YA
6. Trapnell C
(2017a) Single-cell mRNA quantification and differential analysis with Census
Nature Methods 14:309–315.
https://doi.org/10.1038/nmeth.4150
- PubMed
- Google Scholar
1. Qiu X
2. Mao Q
3. Tang Y
4. Wang L
5. Chawla R
6. Pliner HA
7. Trapnell C
(2017b) Reversed graph embedding resolves complex single-cell trajectories
Nature Methods 14:979–982.
https://doi.org/10.1038/nmeth.4402
- PubMed
- Google Scholar
Software
1. R Development Core Team
(2013) R: A language and environment for statistical computing
R Foundation for Statistical Computing, Vienna, Austria.

http://www.r-project.org
(1986) Biosynthetic expression of type X collagen in embryonic chick sternum cartilage during development
The Journal of Biological Chemistry 261:2897–2904.
https://doi.org/10.1016/S0021-9258(17)35871-4
- PubMed
- Google Scholar
Book
(2014) Bone Marrow Stromal Cell Assays
In: Hilton MJ, editors. In Skeletal Development and Repair: Methods and Protocols. Springer. pp. 279–293.
https://doi.org/10.1007/978-1-62703-989-5
- Google Scholar
1. Robling AG
2. Bonewald LF
(2020) The Osteocyte: New Insights
Annual Review of Physiology 82:485–506.
https://doi.org/10.1146/annurev-physiol-021119-034332
- PubMed
- Google Scholar
1. Sapir-Koren R
2. Livshits G
(2014) Bone mineralization is regulated by signaling cross talk between molecular factors of local and systemic origin: the role of fibroblast growth factor 23
BioFactors (Oxford, England) 40:555–568.
https://doi.org/10.1002/biof.1186
- PubMed
- Google Scholar
Software
1. Satija R
2. Butler A
3. Hoffman P
4. Stuart T
(2020)
SeuratWrappers: Community-Provided Methods and Extensions for the Seurat Object

SeuratWrappers.
1. Schmid TM
2. Conrad HE
(1982)
A unique low molecular weight collagen secreted by cultured chick embryo chondrocytes

The Journal of Biological Chemistry 257:12444–12450.
- PubMed
- Google Scholar
1. Spencer JA
2. Ferraro F
3. Roussakis E
4. Klein A
5. Wu J
6. Runnels JM
7. Zaher W
8. Mortensen LJ
9. Alt C
10. Turcotte R
11. Yusuf R
12. Côté D
13. Vinogradov SA
14. Scadden DT
15. Lin CP
(2014) Direct measurement of local oxygen concentration in the bone marrow of live animals
Nature 508:269–273.
https://doi.org/10.1038/nature13034
- PubMed
- Google Scholar
(1999) Indian hedgehog signaling regulates proliferation and differentiation of chondrocytes and is essential for bone formation
Genes & Development 13:2072–2086.
https://doi.org/10.1101/gad.13.16.2072
- PubMed
- Google Scholar
1. Stuart T
2. Butler A
3. Hoffman P
4. Hafemeister C
5. Papalexi E
6. Mauck WM
7. Hao Y
8. Stoeckius M
9. Smibert P
10. Satija R
(2019) Comprehensive Integration of Single-Cell Data
Cell 177:1888–1902.
https://doi.org/10.1016/j.cell.2019.05.031
- PubMed
- Google Scholar
1. Tan Z
2. Kong M
3. Wen S
4. Tsang KY
5. Niu B
6. Hartmann C
7. Chan D
8. Hui C-C
9. Cheah KSE
(2020) IRX3 and IRX5 Inhibit Adipogenic Differentiation of Hypertrophic Chondrocytes and Promote Osteogenesis
Journal of Bone and Mineral Research 35:2444–2457.
https://doi.org/10.1002/jbmr.4132
- PubMed
- Google Scholar
1. Tencerova M
2. Kassem M
(2016) The Bone Marrow-Derived Stromal Cells: Commitment and Regulation of Adipogenesis
Frontiers in Endocrinology 7:127.
https://doi.org/10.3389/fendo.2016.00127
- PubMed
- Google Scholar
1. Tikhonova AN
2. Dolgalev I
3. Hu H
4. Sivaraj KK
5. Hoxha E
6. Cuesta-Domínguez Á
7. Pinho S
8. Akhmetzyanova I
9. Gao J
10. Witkowski M
11. Guillamot M
12. Gutkin MC
13. Zhang Y
14. Marier C
15. Diefenbach C
16. Kousteni S
17. Heguy A
18. Zhong H
19. Fooksman DR
20. Butler JM
21. Economides A
22. Frenette PS
23. Adams RH
24. Satija R
25. Tsirigos A
26. Aifantis I
(2019) The bone marrow microenvironment at single-cell resolution
Nature 569:222–228.
https://doi.org/10.1038/s41586-019-1104-8
- PubMed
- Google Scholar
1. Tirosh I
2. Izar B
3. Prakadan SM
4. Wadsworth MH
5. Treacy D
6. Trombetta JJ
7. Rotem A
8. Rodman C
9. Lian C
10. Murphy G
11. Fallahi-Sichani M
12. Dutton-Regester K
13. Lin JR
14. Cohen O
15. Shah P
16. Lu D
17. Genshaft AS
18. Hughes TK
19. Ziegler CGK
20. Kazer SW
21. Gaillard A
22. Kolb KE
23. Villani AC
24. Johannessen CM
25. Andreev AY
26. Van Allen EM
27. Bertagnolli M
28. Sorger PK
29. Sullivan RJ
30. Flaherty KT
31. Frederick DT
32. Jané-Valbuena J
33. Yoon CH
34. Rozenblatt-Rosen O
35. Shalek AK
36. Regev A
37. Garraway LA
(2016) Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq
Science (New York, N.Y.) 352:189–196.
https://doi.org/10.1126/science.aad0501
- PubMed
- Google Scholar
1. Trapnell C
2. Cacchiarelli D
3. Grimsby J
4. Pokharel P
5. Li S
6. Morse M
7. Lennon NJ
8. Livak KJ
9. Mikkelsen TS
10. Rinn JL
(2014) The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells
Nature Biotechnology 32:381–386.
https://doi.org/10.1038/nbt.2859
- PubMed
- Google Scholar
(2015) Fate of growth plate hypertrophic chondrocytes: death or lineage extension?
Development, Growth & Differentiation 57:179–192.
https://doi.org/10.1111/dgd.12203
- PubMed
- Google Scholar
Software
1. Wickham H
(2019)
stringr: Simple, Consistent Wrappers for Common String Operationsitle

Stringr.
Software
(2020)
dplyr: A Grammar of Data Manipulation

Dplyr.
1. Worthley DL
2. Churchill M
3. Compton JT
4. Tailor Y
5. Rao M
6. Si Y
7. Levin D
8. Schwartz MG
9. Uygur A
10. Hayakawa Y
(2015) Gremlin 1 Identifies a Skeletal Stem Cell with Bone
Cartilage, and Reticular Stromal Potential. Cell 160:269–284.
https://doi.org/10.1016/j.cell.2014.11.042
- Google Scholar
1. Yamazaki T
2. Mukouyama YS
(2018) Tissue Specific Origin, Development, and Pathological Perspectives of Pericytes
Frontiers in Cardiovascular Medicine 5:78.
https://doi.org/10.3389/fcvm.2018.00078
- PubMed
- Google Scholar
1. Yang L
2. Tsang KY
3. Tang HC
4. Chan D
5. Cheah KSE
(2014a) Hypertrophic chondrocytes can become osteoblasts and osteocytes in endochondral bone formation
PNAS 111:12097–12102.
https://doi.org/10.1073/pnas.1302703111
- PubMed
- Google Scholar
1. Yang G
2. Zhu L
3. Hou N
4. Lan Y
5. Wu XM
6. Zhou B
7. Teng Y
8. Yang X
(2014b) Osteogenic fate of hypertrophic chondrocytes
Cell Research 24:1266–1269.
https://doi.org/10.1038/cr.2014.111
- PubMed
- Google Scholar
1. Yang M
2. Arai A
3. Udagawa N
4. Hiraga T
5. Lijuan Z
6. Ito S
7. Komori T
8. Moriishi T
9. Matsuo K
10. Shimoda K
11. Zahalka AH
12. Kobayashi Y
13. Takahashi N
14. Mizoguchi T
(2017) Osteogenic Factor Runx2 Marks a Subset of Leptin Receptor-Positive Cells that Sit Atop the Bone Marrow Stromal Cell Hierarchy
Scientific Reports 7:4928.
https://doi.org/10.1038/s41598-017-05401-1
- PubMed
- Google Scholar
1. Yu Y
2. Newman H
3. Shen L
4. Sharma D
5. Hu G
6. Mirando AJ
7. Zhang H
8. Knudsen E
9. Zhang G-F
10. Hilton MJ
11. Karner CM
(2019) Glutamine Metabolism Regulates Proliferation and Lineage Allocation in Skeletal Stem Cells
Cell Metabolism 29:966–978.
https://doi.org/10.1016/j.cmet.2019.01.016
- PubMed
- Google Scholar
1. Zelzer E
2. Mamluk R
3. Ferrara N
4. Johnson RS
5. Schipani E
6. Olsen BR
(2004) VEGFA is necessary for chondrocyte survival during bone development
Development (Cambridge, England) 131:2161–2171.
https://doi.org/10.1242/dev.01053
- PubMed
- Google Scholar
1. Zhong L
2. Yao L
3. Tower RJ
4. Wei Y
5. Miao Z
6. Park J
7. Shrestha R
8. Wang L
9. Yu W
10. Holdreith N
11. Huang X
12. Zhang Y
13. Tong W
14. Gong Y
15. Ahn J
16. Susztak K
17. Dyment N
18. Li M
19. Long F
20. Chen C
21. Seale P
22. Qin L
(2020) Single cell transcriptomics identifies a unique adipose lineage cell population that regulates bone marrow environment
eLife 9:e54695.
https://doi.org/10.7554/eLife.54695
- PubMed
- Google Scholar
(2014a) Chondrocytes transdifferentiate into osteoblasts in endochondral bone during development, postnatal growth and fracture healing in mice
PLOS Genetics 10:e1004820.
https://doi.org/10.1371/journal.pgen.1004820
- PubMed
- Google Scholar
1. Zhou BO
2. Yue R
3. Murphy MM
4. Peyer JG
5. Morrison SJ
(2014b) Leptin-receptor-expressing mesenchymal stromal cells represent the main source of bone formed by adult bone marrow
Cell Stem Cell 15:154–168.
https://doi.org/10.1016/j.stem.2014.06.008
- PubMed
- Google Scholar

Decision letter

Mei Wan

Reviewing Editor; Johns Hopkins University School of Medicine, United States
Mone Zaidi

Senior Editor; Icahn School of Medicine at Mount Sinai, United States
Mei Wan

Reviewer; Johns Hopkins University School of Medicine, United States
Noriaki Ono

Reviewer; University of Texas Health Science Center at Houston, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Hypertrophic Chondrocytes Serve as a Reservoir for Unique Marrow Associated Skeletal Stem and Progenitor Cells, Osteoblasts, and Adipocytes During Skeletal Development" for consideration at eLife. Your initial submission has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. Although the work is of interest, we regret to inform you that the findings at this stage are too preliminary for further consideration at eLife.

Reviewer #1:

Several prior studies showed that hypertrophic chondrocytes can give rise to osteoblast lineage cells during skeletal development and growth. In the present study, the authors, by conducted a characterization on the localization and lineage fate of the tdTomator+ cells using Col10a1CreERT2 ; R26-tdTomato lineage-tracing system, confirmed that hypertrophic chondrocytes gave rise to osteoblasts and other bone marrow associated cells. The authors then performed single cell RNA-sequencing on cells derived from 2-month-old Col10a1Cre; R26-tdTomato mice. They found that a subset of bone marrow tdTomato+ cells are skeletal stem and progenitor cells (SSPC) that contribute to osteogenic cells and the adipogenic lineage cells. Finally, the authors performed bulk RNA-sequencing on the tdTomoto+ and tdTomoto- CFU-Fs, and they found enriched expression of cell proliferation and vascular development genes within the tdTomoto+ CFU-Fs. These findings add new evidence to the current concept of chondrocyte-osteoprogenitors-osteoblast transdifferentiation and suggest hypertrophic chondrocytes may undergo dedifferentiate into SSPC. I have the following issues for the authors to address:

1. The lineage tracing study was conducted only in postnatal mice (P6) and chased for 2 weeks. Lineage tracing at different age groups (pre-natal or adult mice) would be useful. Moreover, more evidence is needed to claim that hypertrophic chondrocytes-derived cells undergo a process of dedifferentiation into SSPCs. For example, it is necessary to conduct fluorescence imaging to assess whether stem/progenitor markers (preferentially use another fluorescence reporter mouse strain) colocalize with tdTomato in bone/bone marrow tissue sections from Col10a1CreERT2 ; R26-tdTomato mice at different time points after tamoxifen treatment. Flow cytometry analyses of the expression of the stem/progenitor markers in sorted tdTomato+ cells would also be helpful.

2. Could the authors explain why the single cell RNA-seq were conducted using bone/bone marrow cells derived from 2-month-old non-inducible Col10a1Cre; R26-tdTomato mice? The Tamoxifen-inducible strain at early postnatal stage may make more sense.

3. In Figure 1l, 70-80% of OCN was tdTomato+ in Col10a1Cre; R26-tdTomato mice. However, in Figure 5 and Figure S10, tdTomoto+ and tdTomoto- colonies showed similar ability of osteogenesis. Please explain the discrepancy.

4. In Figure 5, the authors show that nearly 50% of CFU-F colonies derived from Col10a1Cre;R26-tdTomato mice. What is the age of the mice used? Is this age-associated?

5. In Figure S10, the authors assessed the in vitro osteogenic and adipogenic differentiation capacities of the colonies. The images of different staining results should be presented.

Recommendations for the authors:

1. I suggest that Figure S3, especially the staining data in Col10a1CreERT2;Rosa26-tdTomato mice, could be moved to the main text because it provides important evidence to support the dedifferentiation of the labeled tdTomato+ cells.

2. The mechanisms underlying the dedifferentiation of hypertrophic chondrocytes are interesting. This may be elaborated in the Discussion section.

Reviewer #2:

Strengths:

1. This study utilized a powerful combination of hypertrophic chondrocyte-specific lineage tracing models and single cell RNA-seq analyses, to uncover previously unknown identities of hypertrophic chondrocyte-derived bone marrow stromal cell populations.

2. This study took a closer look at Col10a1-creER-based lineage-tracing models to confirm that hypertrophic chondrocytes can transition into bone marrow stromal cells.

3. This study discovers heterogeneity of hypertrophic chondrocyte-derived cells in bone marrow, including osteoblasts and undifferentiated stromal cells that may overlap with skeletal stem and progenitor cells.

Weaknesses:

1. The major conclusion of this study on de-differentiation is mostly supported by single cell RNA-seq analyses of flushed and sorted bone marrow stromal cells, which does not capture the transition of hypertrophic chondrocytes.

2. The authors used Col10a1-CreER and Col10a1-Cre models interchangeably without much considering the important temporal factor. SSPCs that they defined in scRNA-seq experiments may not represent those generated by a recent transformation event. The transformation may have occurred at a much earlier development time point. Essentially all SSPCs in the endochondroal pathway are, after all, derived from chondrocytes. Therefore, these SSPCs may not represent a unique cell population.

3. The authors did not consider the heterogeneity of chondrocytes that are marked by Col10a1-Cre, which include hypertrophic chondrocytes at various locations of the growth plate.

4. This study did not directly show that hypertrophic chondrocyte-derived SSPC populations actually contribute to osteoblasts. These cells could be just bystanders, whereas direct transformation of hypertrophic chondrocytes to osteoblasts may be more significant.

5. The section pertaining to support of hematopoiesis was mostly irrelevant.

Recommendations for the authors:

1. Title:

It is unclear what the authors mean by "unique" marrow associated skeletal stem and progenitor cells. They defined these cells based on previously available markers as described in the main text; therefore, they are not "unique".

2. Abstract:

Generally speaking, this abstract is oversimplified and overstated, not capturing the essence of this study. The distinction between two models of lineage tracing (constitutive active vs. inducible) should be delineated.

Line 38: I am not sure which reference(s) the authors mean by saying that "SSPCs localized to the surrounding periosteum serve as the primary source of marrow associated SSPCs, osteoblasts, osteocytes, and adipocytes during skeletal development". The periosteum does not provide a precursor for marrow cell populations. The perichondrium is not a primary source of marrow cells.

Line 40: The authors should specify which genetic reporter mouse models they used in this study.

Line 42: What do the authors mean by this: "…unique marrow associated SSPCs, previously characterized as a CXCL12-abundant reticular (CAR) cells"? This sentence is contradictory. Moreover, the definition of CAR cells is well-established in the field.

3. Introduction:

Line 47-63: Olsen et al. 2000 is cited five times in the first paragraph. Although this is a valid book chapter to cite, they should identify the primary literature and also cite these papers accordingly.

Line 83-85: It is not true that the Acan-CreER model is more restricted to chondrocytes within the growth plate. It shows more recombination in the periosteum and other marrow components.

Line 109-110: It is not surprising that hypertrophic chondrocyte derived SSPCs are striking similar to marrow associated SSPCs, because, after all, most SSPCs are derived from chondrocytes in the endochondral pathway.

4. Results:

Line 128: IF staining in Figure 1d is not convincing.

Line 131-132: This point is not new and has been already shown in preceding studies. Although I appreciate their efforts to reproduce, these results are confirmatory and not entirely novel.

Line 153: Singe cell RNA-seq is not the right approach to reveal "multiple cell fates" of hypertrophic chondrocytes.

Line 160-161: To support this statement, the authors should include FACS plots in the paper.

Line 162: Why is the number of cells analyzed here so small (~1,000 cells)? This would limit the power of scRNA-seq analysis.

Line 217: Although interesting, the authors did not provide concrete evidence that de-differentiation generates "the most primitive SSCs". This is largely speculative.

Line 242-251: I am not sure how these discussions on support of hematopoiesis fit into the lines of discussion.

5. Discussion:

Line 390: I don't think that the authors showed here with genetic evidence that "hypertrophic chondrocytes undergo a process of dedifferentiation to form unique SSPCs populations". This piece of data would be critical to sustain the major conclusion of this study.

Line 403 (and the title): Why do the authors think that hypertrophic chondrocytes serve as a "reservoir and alternative source" of SSPCs? Their data clearly support the now well-accepted notion that these cells are a "primary" source of SSPCs.

Line 436: The authors are not the first group demonstrating that growth plate chondrocytes serve as a reservoir for CAR cells.

Line 466-469: The last conclusions appear to be vastly overstated.

Reviewer #3:

Authors examined the process of transdifferentiation of hypertrophic chondrocytes to osteoblasts or other marrow associated cells using the lineage tracing reporter mice (Col10a1 Cre-Rosa Tomato mice), which express Tomato in the hypertrophic chondrocytes and their descendants. They showed that more than 80% of bone marrow osteoblasts was derived from hypertrophic chondrocytes. They performed scRNA sequencing of Tomato-positive cells, and the gene expression profiles of most of the Tomato-positive cells are like those in skeletal stem and progenitor cells (SSPCs) and a niche cell for hematopoietic stem cells. It was confirmed by the staining of the Tamato-positive cells with immunohistochemistry using antibodies against PDGFRb, LEPR, and PDGFRa, which are expressed in SSPCs. Further, they showed that 25% of adipocytes and 37% of LepR+ cells were derived from Tamato-positive cells by immunohistochemistry. They also showed that half of the CFU-F was derived from Tamato-positive cells, and they had more proliferative activity than that derived from Tamato-negative cells. They concluded that hypertrophic chondrocytes undergo a process of dedifferentiation to generate SSPCs, which serve as a primary source of bone marrow osteoblasts and some source of adipocytes.

Strengths:

The scRNA sequencing of the transdifferentiated cells enabled the characterization of SSPCs derived from hypertrophic chondrocytes in detail.

Weaknesses:

1. They used immunohistochemistry for the quantitative analysis, including the frequencies of transdifferentiated osteoblasts and LepR+ cells. The immunohistochemistry using adult mouse samples is not easy because of the process of decalcification, which reduces antigen reactivity. They should use reporter mice driven by 2.3 kb Col1a1 promoter for the frequency of transdifferentiated osteoblasts, and LepR Cre Rosa mice for the frequency of LepR+ cells. Further, they prepared the samples by immersion fixation but not perfusion fixation. It is important to prepare the samples for immunohistochemistry by perfusion fixation to keep the antigen reactivity. They performed immunohistochemistry using OCN antibody to detect osteoblasts in Figure 1 and Supplementary Figure 8. The number of OCN-positive cells are too low, indicating that the immunohistochemical analysis is not working. Further, the pictures with higher magnification are required for the evaluation of the staining in Figure 1 and Supplementary Figures3, 6, 8, and the locations of the pictures with high magnification should be shown in the pictures with low magnification. In Supplementary Figure 3, most of the cells look Tomato single positive cells and PDGFRb, LEPR, or PDGFRa single positive cells. It is contradictory to Figure 3.

2. 37 % of LepR+ cells were derived from hypertrophic chondrocytes. However, more than 80% of OCN-positive cells were derived from hypertrophic chondrocytes. The discrepancy also seems to have been caused by the quantitative analysis using immunohistochemistry of LepR and OCN.

3. In Figure 2, the populations of osteocytes (cluster 1) and osteoblasts (cluster 2) were too small, indicating that the samples were not appropriately collected. It skews the understanding of dedifferentiation and transdifferentiation. It is recommended to perform scRNA sequencing at embryonic stage (E16.5-18.5) to confirm the appearance of dedifferentiated cells.

4. The frequency of osteoblasts derived from hypertrophic chondrocytes at 2 months of age was 83% in this paper, and it was described as consistent to the previous reports. However, it was 19-31% at 3 weeks of age in Yang et al. (Cell Res 24:1266, 2014), 62-63% at 1 month of age in Zhou et al. (PLoS Genet 10: e1004820, 2014), and 15% at 3 weeks of age in Qin et al. (PLoS Genet 16: e1009169, 2020). The frequencies were examined using reporter mice in the previous three papers. As the OCN immunostaining was not working, however, the frequency of this paper is not reliable. Even if it is reliable, the data is not consistent to the previous papers. Further, the last paper, which was not cited in this paper, also showed that bone volume becomes normal by 6 weeks of age in the absence of transdifferentiation of hypertrophic chondrocytes, indicating the transient requirement of the transdifferentiation at embryonic and neonatal stage. Referring the previous findings, precise discussion should be done.

5. The presence of the dedifferentiated cells does not mean that hypertrophic chondrocytes transdifferentiate into osteoblasts through the dedifferentiated cell stage. The data in this paper is consistent with the previous findings, which indicated that hypertrophic chondrocytes become osteoblasts, adipocytes, and stromal cells. Further, the differentiation of hypertrophic chondrocytes into Cxcl12-positive stromal cells was already shown in the previous paper (Nature 563: 254, 2018), although in detailed expression profiles of the stromal cells were provided in this paper.

6. In Figure 5G, are more than half of the cells in cluster 1 (osteocytes) S/G2/M phase?

Recommendations for the authors:

1. In line 131 on page 6, what are other osteogenic markers?

2. In line 147-148 on page 7, what does "loosely associated with the marrow" mean?

https://doi.org/10.7554/eLife.76932.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

Several prior studies showed that hypertrophic chondrocytes can give rise to osteoblast lineage cells during skeletal development and growth. In the present study, the authors, by conducted a characterization on the localization and lineage fate of the tdTomator+ cells using Col10a1CreERT2 ; R26-tdTomato lineage-tracing system, confirmed that hypertrophic chondrocytes gave rise to osteoblasts and other bone marrow associated cells. The authors then performed single cell RNA-sequencing on cells derived from 2-month-old Col10a1Cre; R26-tdTomato mice. They found that a subset of bone marrow tdTomato+ cells are skeletal stem and progenitor cells (SSPC) that contribute to osteogenic cells and the adipogenic lineage cells. Finally, the authors performed bulk RNA-sequencing on the tdTomoto+ and tdTomoto- CFU-Fs, and they found enriched expression of cell proliferation and vascular development genes within the tdTomoto+ CFU-Fs. These findings add new evidence to the current concept of chondrocyte-osteoprogenitors-osteoblast transdifferentiation and suggest hypertrophic chondrocytes may undergo dedifferentiate into SSPC. I have the following issues for the authors to address:

1. The lineage tracing study was conducted only in postnatal mice (P6) and chased for 2 weeks. Lineage tracing at different age groups (pre-natal or adult mice) would be useful. Moreover, more evidence is needed to claim that hypertrophic chondrocytes-derived cells undergo a process of dedifferentiation into SSPCs. For example, it is necessary to conduct fluorescence imaging to assess whether stem/progenitor markers (preferentially use another fluorescence reporter mouse strain) colocalize with tdTomato in bone/bone marrow tissue sections from Col10a1CreERT2 ; R26-tdTomato mice at different time points after tamoxifen treatment. Flow cytometry analyses of the expression of the stem/progenitor markers in sorted tdTomato+ cells would also be helpful.

To address these concerns, we have provided new inducible lineage tracing data at both embryonic (Figure 1a-c, 3g; E16.5) and postnatal/adult (Figure S1a-b; 2-months and 4-months) time points. In Figure 3g, we demonstrate that hypertrophic chondrocytes initially labeled at E13.5-15.5 transition into a marrow associated Pdgfra-expressing cells during bone development. These data were achieved by generating a Col10a1Cre^ERT2; R26-tdTomato mouse line that incorporates a PDGFRa-H2B-GFP knock-in allele. This PDGFRa-H2B- GFP allele was also utilized with our Col10a1Cre; R26-tdTomato mouse line for flow cytometry of marrow associated cells at 2-months of age and demonstrated that greater than 70% of all of hypertrophic chondrocyte derived marrow associated cells or SSPCs are PDGFRa/GFP positive, which is consistent with our scRNA-seq data at this timepoint (Figure S4c).

2. Could the authors explain why the single cell RNA-seq were conducted using bone/bone marrow cells derived from 2-month-old non-inducible Col10a1Cre; R26-tdTomato mice? The Tamoxifen-inducible strain at early postnatal stage may make more sense.

Due to the transient nature of the hypertrophic chondrocytes following single tamoxifen injections in Col10a1Cre^ERT2; R26-tdTomato mice and their limited contribution to the marrow utilizing this strategy, we opted to use the Col10a1Cre^ERT2; R26-tdTomato mouse line. Original inducible experiments did not yield sufficient cell numbers following tissue digestion, FACS, and downstream quality check/removals. We chose 2-months of age as a postnatal developmental time point in which the hypertrophic chondrocytes would have likely produced a potential variety of de-differentiated and or differentiated marrow associated cell types. We thought this would be ideal for capturing the full breadth of what hypertrophic chondroyctes were capable of becoming during skeletal development and homeostasis, in addition to the bone associated osteoblasts and osteocytes. To further address reviewers’ concerns, we have now performed an additional scRNA-seq experiment and analysis of E16.5 Col10a1Cre; R26-tdTomato long bones (cartilage, bone, and marrow tdTomato+ cells) and have confirmed prior 2-month scRNA-seq data that hypertrophic chondrocytes undergo a dedifferentiation process to generate SSPCs prior to becoming cells of the osteoblast lineage (Figure 2, 3, S2, and S3).

3. In Figure 1l, 70-80% of OCN was tdTomato+ in Col10a1Cre; R26-tdTomato mice. However, in Figure 5 and Figure S10, tdTomoto+ and tdTomoto- colonies showed similar ability of osteogenesis. Please explain the discrepancy.

The reviewer notes that 70-80% of OCN+ osteoblasts were tdTomato+, and that the tdTomato+ and tdTomato- colonies were roughly equal in numbers and osteogenic potential. So then why are most osteoblasts and osteocytes derived from hypertrophic chondrocytes? We attribute this to the rapid growth that is required during early skeletal development and the quick turnover of hypertrophic chondrocytes from the growth plate. This likely allows for faster contribution of the hypertrophic chondrocyte derived cells as compared to the invading periosteal associated osteo- progenitors.

4. In Figure 5, the authors show that nearly 50% of CFU-F colonies derived from Col10a1Cre;R26-tdTomato mice. What is the age of the mice used? Is this age-associated?

The 50% CFU ratio was performed at 2-months of age (the same timepoint as our adult scRNA-seq experiment). We have now performed an additional CFU analysis at 8-months of age (Figure 7) and observed relatively similar numbers.

5. In Figure S10, the authors assessed the in vitro osteogenic and adipogenic differentiation capacities of the colonies. The images of different staining results should be presented.

We performed Oil Red O and Von Kossa staining to quantify the in vitro differentiation capacity. These were first analyzed under a fluorescence microscope and marked on the plate for tdTomato positive and tdTomato negative. After the stain is performed that tdTomato signal is much weaker and in the case of Von Kossa is completely opaque because of the stain. In preparation for this re-submission we instead performed in vivo kidney capsule transplantation assays to truly assess their multilineage differentiation capacity. These in vivo data demonstrate that both SSPC populations are capable of forming osteoblasts that generate bone; however, only the hypertrophic chondrocyte derived SSPCs are capable of forming a complete vascularized bone organ with host derived marrow, and donor derived osteoblasts, osteocytes, some marrow associated cells, and adipocytes (Figure 8).

6. I suggest that Figure S3, especially the staining data in Col10a1CreERT2;Rosa26-tdTomato mice, could be moved to the main text because it provides important evidence to support the dedifferentiation of the labeled tdTomato+ cells.

We have now included new Col10a1Cre^ERT2; R26tdTomato lineage tracing data in Figure 1 and 3. We have also now provided additional scRNA-seq data in Figure 2 and 3 that clearly identifies the transition of hypertrophic chondrocytes into a SSPC cell population prior to progressing to the osteoblast lineage. The additional supportive data previously supplied remains in Figure S3.

7. The mechanisms underlying the dedifferentiation of hypertrophic chondrocytes are interesting. This may be elaborated in the Discussion section.

A discussion of the potential mechanisms underlying the dedifferentiation of hypertrophic chondrocytes has been expanded within the Discussion section of the manuscript.

Reviewer #2:

Strengths:

1. This study utilized a powerful combination of hypertrophic chondrocyte-specific lineage tracing models and single cell RNA-seq analyses, to uncover previously unknown identities of hypertrophic chondrocyte-derived bone marrow stromal cell populations.

2. This study took a closer look at Col10a1-creER-based lineage-tracing models to confirm that hypertrophic chondrocytes can transition into bone marrow stromal cells.

3. This study discovers heterogeneity of hypertrophic chondrocyte-derived cells in bone marrow, including osteoblasts and undifferentiated stromal cells that may overlap with skeletal stem and progenitor cells.

Weaknesses:

1. The major conclusion of this study on de-differentiation is mostly supported by single cell RNA-seq analyses of flushed and sorted bone marrow stromal cells, which does not capture the transition of hypertrophic chondrocytes.

We have now performed E16.5 scRNA-sequencing that captures various populations of hypertrophic chondrocytes, as well as their transition through a SSPC stage prior to becoming osteoblast lineage cells (Figure 2 and 3). Imaging of long bone sections from E16.5 Col10a1Cre^ERT2; R26-tdTomatof/+; Pdgfra-H2B-GFP mice (tamoxifen E13.5-15.5) have also demonstrated tdTomato/GFP double positive cells within the marrow space that were OCN-, indicating the presence of Pdgfra (SSPC marker) expressing cells that are derived from hypertrophic chondrocytes and are not osteoblasts (Figure 3g).

2. The authors used Col10a1-CreER and Col10a1-Cre models interchangeably without much considering the important temporal factor. SSPCs that they defined in scRNA-seq experiments may not represent those generated by a recent transformation event. The transformation may have occurred at a much earlier development time point. Essentially all SSPCs in the endochondroal pathway are, after all, derived from chondrocytes. Therefore, these SSPCs may not represent a unique cell population.

We disagree that all SSPCs are derived from chondrocytes.

SSPCs have been identified in the periosteum which originate only from common osteo-chondral or mesenchymal progenitors and not chondrocytes specifically during early skeletal development. We do; however, agree that there might be subclasses of chondrocytes that potentially contribute to SSPCs differently. A recent study by Mizuhashi, et al. (2019) found that borderline chondrocytes exist and rarely undergo hypertrophy or express Col10a1. Differing sources of SSPCs is further supported by our CFU analysis that indicate than nearly half of colonies or progenitors do not originate from hypertrophic chondrocytes and that hypertrophic chondrocyte derived SSPCs appear to have different multilineage capabilities when transplanted in vivo. With this in mind, we also performed scRNA-seq of tdTomato+ cells from E16.5 Col10a1Cre; R26-tdTomato long bones which captures the specific transitional period as hypertrophic chondrocytes give way for the formation of the nascent marrow cavity and endochondral bone, allowing us to demonstrate that some of the SSPCs are certainly derived from hypertrophic chondrocytes during skeletal development (Figure 2, 3, and S3). Some of the SSPCs identified at 2months of age may indeed be cells that were generated at earlier time-points and remain as progenitors within the marrow space, however new lineage tracing data at 2- and 4-months of age suggest that their contribution to marrow associated cells and the osteoblast lineage persists well into adulthood (Figure S1).

3. The authors did not consider the heterogeneity of chondrocytes that are marked by Col10a1-Cre, which include hypertrophic chondrocytes at various locations of the growth plate.

We believe the reviewer is asking if there might be regional differences between hypertrophic chondrocytes. We cannot exclude this possibility, but note that we see SSPCs derived from hypertrophic chondrocytes along the length of the metaphyseal/trabecular region, as well as, throughout the marrow cavity. The expression of the Col10a1-driven CRE is specific to cells only within the hypertrophic zone of the growth plate, and therefore demonstrate that the marrow derived SSPCs, osteoblast lineage cells, and adipocytes are derived from cells within this hypertrophic zone. We have now referenced this in our discussion.

4. This study did not directly show that hypertrophic chondrocyte-derived SSPC populations actually contribute to osteoblasts. These cells could be just bystanders, whereas direct transformation of hypertrophic chondrocytes to osteoblasts may be more significant.

While it is difficult to conclude the exact percentage of cells actively contributing to the osteoblasts from hypertrophic chondrocyte derived SSPCs, we observed in our E16.5 scRNA-seq analysis only a small population of Bglap+ cells and trajectory analysis places them downstream of Pdgfra, Lepr, and Cxcl12 expressing SSPCs (Figure 2 and 3). We also demonstrated that hypertrophic chondrocyte derived SSPCs are fully capable of generating marrow associated cells, osteoblasts, osteocytes, and adipocytes when transplanted under the kidney capsule (Figure 8). We have also added more detail to the Discussion section to acknowledge this potential bystander impact during normal skeletal development in vivo.

5. The section pertaining to support of hematopoiesis was mostly irrelevant.

This section has been removed.

Recommendations for the authors:

1. Title:

It is unclear what the authors mean by "unique" marrow associated skeletal stem and progenitor cells. They defined these cells based on previously available markers as described in the main text; therefore, they are not "unique".

We have removed the word unique.

2. Abstract:

Generally speaking, this abstract is oversimplified and overstated, not capturing the essence of this study. The distinction between two models of lineage tracing (constitutive active vs. inducible) should be delineated.

The abstract has been reworked to address this issue.

Line 38: I am not sure which reference(s) the authors mean by saying that "SSPCs localized to the surrounding periosteum serve as the primary source of marrow associated SSPCs, osteoblasts, osteocytes, and adipocytes during skeletal development". The periosteum does not provide a precursor for marrow cell populations. The perichondrium is not a primary source of marrow cells.

We disagree with the reviewer’s assessment that the periosteum does not provide precursors for marrow cell populations, since previous literature has shown that osteoblastic progenitors migrate into the marrow with blood vessels. Specific citations have been provided in the manuscript.

Line 40: The authors should specify which genetic reporter mouse models they used in this study.

Details regarding the genetic reporters have been added.

Line 42: What do the authors mean by this: "…unique marrow associated SSPCs, previously characterized as a CXCL12-abundant reticular (CAR) cells"? This sentence is contradictory. Moreover, the definition of CAR cells is well-established in the field.

We have fixed this contradiction.

3. Introduction:

Line 47-63: Olsen et al. 2000 is cited five times in the first paragraph. Although this is a valid book chapter to cite, they should identify the primary literature and also cite these papers accordingly.

Citations have been added to cite primary literature of the original data.

Line 83-85: It is not true that the Acan-CreER model is more restricted to chondrocytes within the growth plate. It shows more recombination in the periosteum and other marrow components.

The original publication that documented the creation of the AcanCre^ERT2 mouse line (Henry, et al. Genesis 2009) demonstrates fairly restricted X-gal staining to chondrocytes when observed by a short chase during postnatal skeletal development.

Line 109-110: It is not surprising that hypertrophic chondrocyte derived SSPCs are striking similar to marrow associated SSPCs, because, after all, most SSPCs are derived from chondrocytes in the endochondral pathway.

We disagree with the reviewer’s assessment that the periosteum does not provide precursors for marrow cell populations, since prior literature has shown that osteoblastic progenitors migrate into the marrow with blood vessels. Further, our manuscript demonstrates that slightly more than half of the SSPCs are derived from the hypertrophic chondrocytes and interestingly our in vivo transplantation assays indicate important functional differences between these SSPC populations derived from various sources.

4. Results:

Line 128: IF staining in Figure 1d is not convincing.

We have replaced this image with a more representative stained image.

Line 131-132: This point is not new and has been already shown in preceding studies. Although I appreciate their efforts to reproduce, these results are confirmatory and not entirely novel.

We understand that we are not providing novel findings with this replication of the data, but we wanted to replicate the data for context of the utility of the Cre lines and to draw the attention to the non-osteoblast associated cells. Additionally, the specific Col10a1Cre^ERT2 mouse line utilized in this study has not been previously published and we wanted to validate similar results with this model.

Line 153: Singe cell RNA-seq is not the right approach to reveal "multiple cell fates" of hypertrophic chondrocytes.

We utilize scRNA-seq as one method to confirm the multiple cell fates of hypertrophic chondrocytes, in addition to lineage tracing, immunostaining (IF), and flow cytometry.

Line 160-161: To support this statement, the authors should include FACS plots in the paper.

These have now been added to Figure S2.

Line 162: Why is the number of cells analyzed here so small (~1,000 cells)? This would limit the power of scRNA-seq analysis.

The number of cells analyzed was small because only approximately 1/2 of all SSPCs are derived from hypertrophic chondrocytes, which have been previously shown to be a very small population. Further with the smaller number we were able to increase the depth of sequencing to assess potential variation within the SSPC populations. The additional E16.5 scRNA-seq was performed with >2000 cells and provides further confirmation of this transition.

Line 217: Although interesting, the authors did not provide concrete evidence that de-differentiation generates "the most primitive SSCs". This is largely speculative.

This has been corrected.

Line 242-251: I am not sure how these discussions on support of hematopoiesis fit into the lines of discussion.

This section has been moved as we believe it is consistent with the transplantation data that revealed marrow formation ability of the hypertrophic derived SSPCs, consistent with previous literature of the importance of these hematopoietic regulatory factors.

5. Discussion:

Line 390: I don't think that the authors showed here with genetic evidence that "hypertrophic chondrocytes undergo a process of dedifferentiation to form unique SSPCs populations". This piece of data would be critical to sustain the major conclusion of this study.

We have corrected (eliminated) the use of “unique” as the data is consistent with previously identified SSPCs, such as the CAR cells.

Line 403 (and the title): Why do the authors think that hypertrophic chondrocytes serve as a "reservoir and alternative source" of SSPCs? Their data clearly support the now well-accepted notion that these cells are a "primary" source of SSPCs.

Line 436: The authors are not the first group demonstrating that growth plate chondrocytes serve as a reservoir for CAR cells.

We do not believe it to be a well-accepted notion that chondrocytes, specifically Col10a1-expressing hypertrophic chondrocytes give rise to SSPCs. Previous data to target chondrocytes (Col2a1Cre is expressed in cells outside of the growth plate and AcanCre^ERT2 can only provide a portion of the information as it is inducible). Pthrp labeled cells, appear to be a more progenitor population that can either give rise to columns of chondrocytes (closer to our model) or to borderline chondrocytes (largely Col10a1-negative cells). The columns of chondrocytes have been shown to give rise to some CXCL12+ cells; however, this data was limited in that it took months for contribution to a full column and then to CXCL12+ cells. Our data is the first to more completely demonstrate that Col10a1-expressing hypertrophic chondrocytes give rise to not only CXCL12+ cells but also Lepr+ cells, and PαS cells embryonically. Furthermore, we have provided a more robust transcriptomic signature of the SSPCs that are generated by hypertrophic chondrocytes at multiple developmental time points.

Line 466-469: The last conclusions appear to be vastly overstated.

We have reworked the conclusions of the manuscript based on the both the previous and new data.

Reviewer #3:

Authors examined the process of transdifferentiation of hypertrophic chondrocytes to osteoblasts or other marrow associated cells using the lineage tracing reporter mice (Col10a1 Cre-Rosa Tomato mice), which express Tomato in the hypertrophic chondrocytes and their descendants. They showed that more than 80% of bone marrow osteoblasts was derived from hypertrophic chondrocytes. They performed scRNA sequencing of Tomato-positive cells, and the gene expression profiles of most of the Tomato-positive cells are like those in skeletal stem and progenitor cells (SSPCs) and a niche cell for hematopoietic stem cells. It was confirmed by the staining of the Tamato-positive cells with immunohistochemistry using antibodies against PDGFRb, LEPR, and PDGFRa, which are expressed in SSPCs. Further, they showed that 25% of adipocytes and 37% of LepR+ cells were derived from Tamato-positive cells by immunohistochemistry. They also showed that half of the CFU-F was derived from Tamato-positive cells, and they had more proliferative activity than that derived from Tamato-negative cells. They concluded that hypertrophic chondrocytes undergo a process of dedifferentiation to generate SSPCs, which serve as a primary source of bone marrow osteoblasts and some source of adipocytes.

Strengths:

The scRNA sequencing of the transdifferentiated cells enabled the characterization of SSPCs derived from hypertrophic chondrocytes in detail.

Weaknesses:

1. They used immunohistochemistry for the quantitative analysis, including the frequencies of transdifferentiated osteoblasts and LepR+ cells. The immunohistochemistry using adult mouse samples is not easy because of the process of decalcification, which reduces antigen reactivity. They should use reporter mice driven by 2.3 kb Col1a1 promoter for the frequency of transdifferentiated osteoblasts, and LepR Cre Rosa mice for the frequency of LepR+ cells. Further, they prepared the samples by immersion fixation but not perfusion fixation. It is important to prepare the samples for immunohistochemistry by perfusion fixation to keep the antigen reactivity. They performed immunohistochemistry using OCN antibody to detect osteoblasts in Figure 1 and Supplementary Figure 8. The number of OCN-positive cells are too low, indicating that the immunohistochemical analysis is not working. Further, the pictures with higher magnification are required for the evaluation of the staining in Figure 1 and Supplementary Figures3, 6, 8, and the locations of the pictures with high magnification should be shown in the pictures with low magnification. In Supplementary Figure 3, most of the cells look Tomato single positive cells and PDGFRb, LEPR, or PDGFRa single positive cells. It is contradictory to Figure 3.

We have performed flow analysis for PDGFRa utilizing another reporter line (PDGFRa-H2B-GFP) and antibody for PDGFRB and found a majority of tdTomato+ cells in the marrow are SSPCs (Figure S4c). We cannot fully understand the contribution to direct transdifferentiation and dedifferentiation with the currently available tools. We disagree that using a Col1a1(2.3kb) reporter would only demonstrate transdifferentiated but could also include osteoblasts that went through a SSPC intermediate. Additionally, a LepRCre combined with a genetic reporter would also not be usable as the CRE recombinase would affect the reporter already in the Col10a1Cre;R26-tdTomato line. Previous publications cited in this paper have utilized immersion fixation for immuno-staining and we disagree that perfusion fixation is absolutely required for preserving antigen reactivity. We have provided better images further demonstrating the co-labeling of cells.

2. 37 % of LepR+ cells were derived from hypertrophic chondrocytes. However, more than 80% of OCN-positive cells were derived from hypertrophic chondrocytes. The discrepancy also seems to have been caused by the quantitative analysis using immunohistochemistry of LepR and OCN.

The difference in rates of generating the LEPR+ (SSPCs) and OCN+ (osteoblasts) are likely independent processes. The rate at which LEPR+ populations differentiate, proliferate, and/or become quiescent may not be the same and therefore the contributions to the OCN+ osteoblasts could change over time. Additionally, we cannot rule out some transdifferentiation of hypertrophic chondrocytes which would increase the OCN+ population independent of LEPR+ population.

3. In Figure 2, the populations of osteocytes (cluster 1) and osteoblasts (cluster 2) were too small, indicating that the samples were not appropriately collected. It skews the understanding of dedifferentiation and transdifferentiation. It is recommended to perform scRNA sequencing at embryonic stage (E16.5-18.5) to confirm the appearance of dedifferentiated cells.

We purposefully performed only a short digest on the bones as to not obtain as many osteoblasts. We were interested more in the question of whether hypertrophic chondrocytes could dedifferentiate into an SSPC-type population. The reviewer is correct that we cannot draw a direct connection between the rate of dedifferentiation and transdifferentiation from this dataset. Additionally, we have performed E16.5 scRNA-sequencing, as noted by multiple reviewers.

4. The frequency of osteoblasts derived from hypertrophic chondrocytes at 2 months of age was 83% in this paper, and it was described as consistent to the previous reports. However, it was 19-31% at 3 weeks of age in Yang et al. (Cell Res 24:1266, 2014), 62-63% at 1 month of age in Zhou et al. (PLoS Genet 10: e1004820, 2014), and 15% at 3 weeks of age in Qin et al. (PLoS Genet 16: e1009169, 2020). The frequencies were examined using reporter mice in the previous three papers. As the OCN immunostaining was not working, however, the frequency of this paper is not reliable. Even if it is reliable, the data is not consistent to the previous papers. Further, the last paper, which was not cited in this paper, also showed that bone volume becomes normal by 6 weeks of age in the absence of transdifferentiation of hypertrophic chondrocytes, indicating the transient requirement of the transdifferentiation at embryonic and neonatal stage. Referring the previous findings, precise discussion should be done.

We have reworded the conclusions of this study in comparison to the previous studies noted by the reviewer. Yang, et al. utilized a different Cre line as well as used in situ hybridization to analyze overlap. Zhou, et al. utilized the same Cre as this study but looked at 1 month of age which could be different than our 2-month time point because of continued turnover of the hypertrophic cells and proliferation of osteoblasts although their numbers are consistent with our osteocyte quantifications. Qin, et al. provided numbers that are not comparable as the control for these quantifications have one allele of Runx2 deleted (controls were Col10a1Cre;R26- tdTomato;Runx2fl/+). The authors did not provide evidence that the heterozygous allele did not cause a phenotype of its own and also confounds the conclusion that transdifferentiation is only an embryonic/neonatal phenomenon.

5. The presence of the dedifferentiated cells does not mean that hypertrophic chondrocytes transdifferentiate into osteoblasts through the dedifferentiated cell stage. The data in this paper is consistent with the previous findings, which indicated that hypertrophic chondrocytes become osteoblasts, adipocytes, and stromal cells. Further, the differentiation of hypertrophic chondrocytes into Cxcl12-positive stromal cells was already shown in the previous paper (Nature 563: 254, 2018), although in detailed expression profiles of the stromal cells were provided in this paper.

See Reviewer 2 comment 4 for the transdifferentiation versus dedifferentiation conclusions. Additionally, we disagree that this paper demonstrated that hypertrophic chondrocytes were the source of CAR cells. The authors utilized a Pthrp-Cre^ER model that targeted mSSC (based on their flow data) and no evidence of them being chondrocytes (only showing a much earlier state). The authors saw contribution to chondrocyte columns and what appear to be hypertrophic chondrocytes, but do not investigate this further. Additionally, their later paper 2019 JBMR showed that this Pthrp-Cre^ER can label a borderline chondrocyte (Col10a1-negative) cell population associated with some contribution to CAR cells, albeit very minimally at an early postnatal time point. Furthermore, since this tool is not a constitutive CRE, the total contribution over time is very limited. When quantifications were performed the authors did not calculate the percentage of contribution, but rather simply noted that Pthrp-Cre^ER derived cells are capable of generating CXCL12+ marrow associated cells. Finally, the Pthrp-Cre^ER derived cells were very restricted in their potential as they could never give rise to adipocytes. We leave open the possibility that if these cells do contribute to hypertrophic chondrocytes later in development, they likely are not the only contributing cells as we have demonstrated that hypertrophic chondrocyte derived SSPCs can give rise to adipocytes.

6. In Figure 5G, are more than half of the cells in cluster 1 (osteocytes) S/G2/M phase?

We do observe some distribution across the cell cycle in cluster 1. Cluster 1 does not appear to be an osteocyte as osteoblasts can express the genes highlighted in Figure 4 and we do not observe Sost or other osteocyte associated genes in cluster 1, suggesting this is likely not an osteocyte population but rather a second osteoblast population, potentially on a differentiation path toward an osteocyte. Our digestion protocol was not designed to dissociate osteocytes, but rather focus on the marrow associated hypertrophic chondrocyte derived cell populations.

In line 131 on page 6, what are other osteogenic markers?

This has been reworded.

In line 147-148 on page 7, what does “loosely associated with the marrow” mean?

This has been reworded.

https://doi.org/10.7554/eLife.76932.sa2

Article and author information

Author details

Jason T Long
1. Department of Cell Biology, Duke University School of Medicine, Durham, United States
2. Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States
Contribution
Data curation, Formal analysis, Investigation, Writing - original draft, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-6006-0932
Abigail Leinroth
1. Department of Cell Biology, Duke University School of Medicine, Durham, United States
2. Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States
Contribution
Data curation, Investigation, Writing - review and editing

Competing interests
No competing interests declared
Yihan Liao
1. Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States
2. Department of Pharmacology and Cancer Biology, Duke University School of Medicine, Durham, United States
Contribution
Data curation, Investigation, Writing - review and editing

Competing interests
No competing interests declared
Yinshi Ren

Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States

Contribution
Data curation, Investigation, Writing - review and editing

Competing interests
No competing interests declared
Anthony J Mirando

Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States

Contribution
Data curation, Investigation, Methodology, Writing - review and editing

Competing interests
No competing interests declared
Tuyet Nguyen

Program of Developmental and Stem Cell Biology, Duke University School of Medicine, Durham, United States

Contribution
Data curation, Investigation, Writing - review and editing

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0002-8769-9955
Wendi Guo
1. Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States
2. Department of Pharmacology and Cancer Biology, Duke University School of Medicine, Durham, United States
Contribution
Data curation, Investigation, Writing - review and editing

Competing interests
No competing interests declared
Deepika Sharma

Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States

Contribution
Data curation, Investigation, Writing - review and editing

Competing interests
No competing interests declared
Douglas Rouse

Division of Laboratory Animal Resources, Duke University School of Medicine, Durham, United States

Contribution
Data curation, Investigation

Competing interests
No competing interests declared
Colleen Wu
1. Department of Cell Biology, Duke University School of Medicine, Durham, United States
2. Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States
3. Department of Pharmacology and Cancer Biology, Duke University School of Medicine, Durham, United States
Contribution
Resources, Supervision

Competing interests
No competing interests declared
Kathryn Song Eng Cheah

School of Biomedical Sciences, University of Hong Kong, Hong Kong, Hong Kong

Contribution
Resources

Competing interests
Reviewing editor, eLife

"This ORCID iD identifies the author of this article:" 0000-0003-0802-8799
Courtney M Karner
1. Department of Cell Biology, Duke University School of Medicine, Durham, United States
2. Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States
Contribution
Resources, Supervision

Competing interests
No competing interests declared
Matthew J Hilton
1. Department of Cell Biology, Duke University School of Medicine, Durham, United States
2. Department of Orthopaedic Surgery, Duke University School of Medicine, Durham, United States
Contribution
Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing - original draft, Writing - review and editing

For correspondence
matthew.hilton@duke.edu

Competing interests
No competing interests declared

"This ORCID iD identifies the author of this article:" 0000-0003-3165-267X

Funding

NIH/NIAMS (R01AR071722)

Matthew J Hilton

NIH/NIAMS (R01AR063071)

Matthew J Hilton

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We acknowledge the assistance of the Duke Molecular Physiology Institute Molecular Genomics core for the generation of single cell RNA-seq libraries and the Duke Center for Genomic and Computational Biology for the bulk RNA-seq and sequencing of the scRNA-seq libraries. We also acknowledge Dr. Klaus von der Mark for generously providing the Col10a1Cre mouse line. Further, we acknowledge the Duke Light Microscopy Core Facility for their support with imaging and analysis, as well as, the Flow Cytometry Shared Resource Core at Duke for FACS and flow cytometric analysis. This work was supported in large part by R01 grants from the National Institutes of Health (NIH) and the National Institute of Arthritis and Musculoskeletal and Skin (NIAMS) Diseases to MJH (R01AR071722 and R01AR063071) and CMK (AR076325 and AR071967), as well as, by grants from the Research Grants Council and University Grants Council of Hong Kong: AoE/M-04/04, T12-708/12 N to KSEC.

Ethics

This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to and approved by the Duke University Institutional Animal Care and Use Committees (IACUC) (A068-20-03).

Senior Editor

Mone Zaidi, Icahn School of Medicine at Mount Sinai, United States

Reviewing Editor

Mei Wan, Johns Hopkins University School of Medicine, United States

Reviewers

Mei Wan, Johns Hopkins University School of Medicine, United States
Noriaki Ono, University of Texas Health Science Center at Houston, United States

Version history

Preprint posted: May 18, 2021 (view preprint)
Received: January 10, 2022
Accepted: February 13, 2022
Accepted Manuscript published: February 18, 2022 (version 1)
Version of Record published: March 3, 2022 (version 2)

Copyright

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

2,791

Page views
554

Downloads
17

Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Mendeley

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

Jason T Long
Abigail Leinroth
Yihan Liao
Yinshi Ren
Anthony J Mirando
Tuyet Nguyen
Wendi Guo
Deepika Sharma
Douglas Rouse
Colleen Wu
Kathryn Song Eng Cheah
Courtney M Karner
Matthew J Hilton

(2022)

Hypertrophic chondrocytes serve as a reservoir for marrow-associated skeletal stem and progenitor cells, osteoblasts, and adipocytes during skeletal development

eLife 11:e76932.

https://doi.org/10.7554/eLife.76932