Single-cell RNA sequencing reveals cellular and molecular heterogeneity in fibrocartilaginous enthesis formation

Curation statements for this article:
  • Curated by eLife

    eLife logo

    eLife assessment

    This paper represents a valuable single-cell level analysis of tendon enthesis development. It will allow further understanding of this specific process with clinical implications. Specifically, the authors provided convincing evidence for the heterogeneity of postnatal enthesis growth and the molecular dynamics and signaling networks during enthesis formation.

This article has been Reviewed by the following groups

Read the full article See related articles

Abstract

The attachment site of the rotator cuff (RC) is a classic fibrocartilaginous enthesis, which is the junction between bone and tendon with typical characteristics of a fibrocartilage transition zone. Enthesis development has historically been studied with lineage tracing of individual genes selected a priori, which does not allow for the determination of single-cell landscapes yielding mature cell types and tissues. Here, in together with open-source GSE182997 datasets (three samples) provided by Fang et al., we applied Single-cell RNA sequencing (scRNA-seq) to delineate the comprehensive postnatal RC enthesis growth and the temporal atlas from as early as postnatal day 1 up to postnatal week 8. And, we furtherly performed single-cell spatial transcriptomic sequencing on postnatal day 1 mouse enthesis, in order to deconvolute bone-tendon junction (BTJ) chondrocytes onto spatial spots. In summary, we deciphered the cellular heterogeneity and the molecular dynamics during fibrocartilage differentiation. Combined with current spatial transcriptomic data, our results provide a transcriptional resource that will support future investigations of enthesis development at the mechanistic level and may shed light on the strategies for enhanced RC healing outcomes.

Article activity feed

  1. Author Response

    Reviewer #1 (Public Review)

    The authors present a scRNAseq study describing the transcriptomes of the tendon enthesis during postnatal development. This is an important topic that has major implication for the care of common clinical problems such as rotator cuff repair. The results are a valuable addition to the literature, providing a descriptive data set reinforcing other, more comprehensive studies. There are weaknesses, however, in the scRNAseq analyses.

    1)The authors should provide additional rationale for the PCA analysis shown in Fig 1d. It is uncommon to use PCA for histomorphologic parameters. These results do not convincingly demonstrate that P7 is as a critical developmental timepoint.

    1. According to the methods, it appears that the entire humeral head-supraspinatus tendon was used for cell isolation for scRNAseq. This results in the inclusion of cells from a variety of tissues, including bone, growth plate, enthesis and tendon. As such, only a very small percentage of cells in the analysis came from the enthesis. Inclusion of such a wide range of cells makes interpretation of enthesis cells difficult.
    1. The differentiationpseudotime analysis described in Fig 3 is difficult to follow. This map includes cell transcriptomes from vastly different tissues. Presumably, embedded in these maps are trajectories for osteoblast differentiation, chondrocyte differentiation, tenocyte differentiation, etc. With so many layers of overlapping information, it is difficult to (algorithmically) deduce a differentiation path of a particular cell type.
    1. The authors uses the term function throughout the paper (e.g., functional definition of fibrocartilage subpopulations). However, this is a descriptive scRNAseq study, and function can therefore only theoretically be inferred from the algorithms used to analyze the data. A functional role for any of the identified pathways or processes can only be defined with gain- andor loss-of-function studies.
    1. C2 highly expressed biomineralization-related genes (Clec3a, Tnn, Acan). The three example genes are not related to biomineralization.
    1. The functional characterization of the three enthesis cell clusters is not convincing. For example, activation of metabolism-related processes can mean a lot of things (including changes in differentiation), yet the authors interpret it very specifically as role in postnatal fibrochondrocyte formation and growth.
    1. The pseudotime analysis of the enthesis cell clusters is not convincing. The three clusters are quite close and overlapping on the UMAP. Furthermore, the authors focus on Tnn as a novel and unique gene, yet the expression pattern shown in Fig 5g implies even expression of this gene across all three clusters.
    1. The TC1 markers (Ly6a, Dlk3, Clec3b) imply a non-tendon-specific cell population. Perhaps a tendon progenitor pool or an endothelial cell phenotype is more appropriate.
    1. Pseudotime analyses assume that your data set includes cells from progenitor through mature cell populations. It is unclear that the timepoints studied here included cells from early progenitor states.
    1. The CellChat analysis is difficult to follow, as the authors included 18 cell types. The number of possible interactions among so many cell types is enormous, and deducing valid connections between any two cell types in this case should be justified. Is the algorithm robust to so many possible interactions

    Thank you very much for your comments and suggestions. According to your suggestions, we carefully revised the paper. We integrated our dataset with open source GSE182997 datasets and re-performed the downstream analysis. On the other hand, we added immunofluorescence tests to validate the results came from single-cell datasets. And we hope all the mentioned issues in prior version to be well addressed.

    Reviewer #2 (Public Review)

    To reveals cellular and molecular heterogeneity in enthesis, the authors established a single-cell temporal atlas during development. This study provides a transcriptional resource for further investigation of fibrocartilage development.

    Thank you very much for your kind suggestions. According to your suggestions, we integrated our dataset with open source GSE182997 datasets and re-performed the downstream analysis. On the other hand, we added immunofluorescence tests to validate the results came from sinlge-cell datasets. And we hope the mentioned issues in prior version to be well addressed.

  2. eLife assessment

    This paper represents a valuable single-cell level analysis of tendon enthesis development. It will allow further understanding of this specific process with clinical implications. Specifically, the authors provided convincing evidence for the heterogeneity of postnatal enthesis growth and the molecular dynamics and signaling networks during enthesis formation.

  3. Reviewer #1 (Public Review):

    The authors present a scRNAseq study describing the transcriptomes of the tendon enthesis during postnatal development. This is an important topic that has major implication for the care of common clinical problems such as rotator cuff repair. The results are a valuable addition to the literature, providing a descriptive data set reinforcing other, more comprehensive studies. There are weaknesses, however, in the scRNAseq analyses.

    1.The authors should provide additional rationale for the PCA analysis shown in Fig 1d. It is uncommon to use PCA for histomorphologic parameters. These results do not convincingly demonstrate that P7 is as a critical developmental timepoint.

    2. According to the methods, it appears that the entire humeral head-supraspinatus tendon was used for cell isolation for scRNAseq. This results in the inclusion of cells from a variety of tissues, including bone, growth plate, enthesis and tendon. As such, only a very small percentage of cells in the analysis came from the enthesis. Inclusion of such a wide range of cells makes interpretation of enthesis cells difficult.

    3. The differentiation/pseudotime analysis described in Fig 3 is difficult to follow. This map includes cell transcriptomes from vastly different tissues. Presumably, embedded in these maps are trajectories for osteoblast differentiation, chondrocyte differentiation, tenocyte differentiation, etc. With so many layers of overlapping information, it is difficult to (algorithmically) deduce a differentiation path of a particular cell type.

    4. The authors uses the term "function" throughout the paper (e.g., "functional definition of fibrocartilage subpopulations"). However, this is a descriptive scRNAseq study, and "function" can therefore only theoretically be inferred from the algorithms used to analyze the data. A functional role for any of the identified pathways or processes can only be defined with gain- and/or loss-of-function studies.

    5. "C2 highly expressed biomineralization-related genes (Clec3a, Tnn, Acan)". The three example genes are not related to biomineralization.

    6. The functional characterization of the three enthesis cell clusters is not convincing. For example, activation of metabolism-related processes can mean a lot of things (including changes in differentiation), yet the authors interpret it very specifically as "role in postnatal fibrochondrocyte formation and growth".

    7. The pseudotime analysis of the enthesis cell clusters is not convincing. The three clusters are quite close and overlapping on the UMAP. Furthermore, the authors focus on Tnn as a novel and unique gene, yet the expression pattern shown in Fig 5g implies even expression of this gene across all three clusters.

    8. The TC1 markers (Ly6a, Dlk3, Clec3b) imply a non-tendon-specific cell population. Perhaps a tendon progenitor pool or an endothelial cell phenotype is more appropriate.

    9. Pseudotime analyses assume that your data set includes cells from progenitor through mature cell populations. It is unclear that the timepoints studied here included cells from early progenitor states.

    10. The CellChat analysis is difficult to follow, as the authors included 18 cell types. The number of possible interactions among so many cell types is enormous, and deducing valid connections between any two cell types in this case should be justified. Is the algorithm robust to so many possible interactions?

  4. Reviewer #2 (Public Review):

    To reveals cellular and molecular heterogeneity in enthesis, the authors established a single-cell temporal atlas during development. This study provides a transcriptional resource for further investigation of fibrocartilage development.

    Reviewer #2 (Recommendations for the authors):

    1. As known, Fei Fang et al. have established single-cell transcriptomes of mouse supraspinatus tendon enthesis cells (Cell Stem Cell, 2022). It is suggested that the authors introduced Fei Fang et al.'s work in Introduction and emphasize the significant novelty compared with Fei Fang et al.'s work.
    2. In Fig1, the authors highlighted P7 was a critical stage for enthesis differentiation. But this section was less associated with the following content. The authors should link these results with the scRNASeq data. Is there any time-dependent change/signaling in scRNASeq data at this critical time point?
    3. In the H&E staining of Fig1a, the tendon structure was separated and random. It is suggested that the authors provide high-quality staining figures.
    4. Fig2 showed that the Scx+ or Sox9+ cells was decreased in enthesis over time. At least it should be co-staining to show the distribution and frequency of double positive and single positive cell populations. However, a previous study has demonstrated this finding (PLOS ONE, 2020). It is suggested to verify some new findings by IF or IHC staining.
    5. There are some conflicts about trajectory analysis. In Fig3c, RNA velocity showed that the arrow flowed from BTJ to MTJ and CTFb. However, in Fig3d, PAGA plot indicated that BTJ cells is independent of other cells. Furthermore, in supplementary figure S3, RNA velocity showed that the trajectory flowed from TC to BTJ. These figures were inconsistent with the described results. Please provide detailed explanation to avoid misleading readers.
    6. Fig5 showed that C1 was the original cluster, and whether C1 cluster expressed canonical progenic/stem cell markers.
    7. The authors performed cell-cell interaction based on cellchat analysis. But the cell-cell interaction was not actively examined.

  5. Reviewer #3 (Public Review):

    This manuscript describes the use of scRNA-seq to decipher the cellular heterogeneity, molecular dynamics and signaling interactions during fibrocartilaginous enthesis formation. They delineate the enthesis growth and the temporal atlas from embryonic stage to postnatal stage by scRNA-seq, compared the development pattern of enthesis origins with tendon and articular cartilage, then demonstrated the cellular complexity and heterogeneity of postnatal enthesis growth and revealed the molecular dynamics and signaling networks during enthesis formation.

    This manuscript used appropriate and validated methodology in line with current state-of-the-art, and the conclusions of this paper are mostly well supported by data, more in vitro or in vivo experiments are encouraged to verify the key molecular dynamics and signaling networks revealed by scRNA-seq during enthesis formation.

    This manuscript facilitates better understand of the enthesis development, which will benefit the important field of enthesis research.