Single-nucleus transcriptional and chromatin accessibility analyses of maturing mouse Achilles tendon uncover the molecular landscape of tendon stem/progenitor cells

Reviewer #1 (Public review):

This study is focused on identifying unique, innovative surface markers for mature Achilles tendons by combining the latest multi-omics approaches and in vitro evaluation, which would address the knowledge gap of the controversial identity of TPSCs with unspecific surface markers. The use of multi-omics technologies, in vivo characterization, in vitro standard assays of stem cells, and in vitro tissue formation is a strength of this work and could be applied for other stem cell quantification in musculoskeletal research. The evaluation and identification of Cd55 and Cd248 in TPSCs have not been conducted in tendons, which is considered innovative. Additionally, the study provided solid sequencing data to confirm co-expressions of Cd55 and Cd248 with other well-described surface markers such as Ly6a, Tpp3, Pdgfra, and Cd34. Generally, the data shown in the manuscript support the claims that the identified surface antigens mark TPSCs in juvenile tendons.

However, there are missing links between scientific questions aimed to be addressed in Introduction and Methodology/Results. If the study focuses on unsatisfactory healing responses of mature tendons and understanding of mature TPSCs, at least mature Achilles tendons from more than 12-week-old mice and their comparison with tendons from juvenile/neonatal mice should be conducted. However, either 2-week or 6-week-old mice, used for characterization here, are not skeletally mature, Additionally, there is a lack of complete comparison of TPSCs between 2-week and 6-week-old mice in the transcriptional and epigenetic levels.

In order to distinguish TPSCs and characterize their epigenetic activities, the authors used scRNA-seq, snRNA-seq, and snATAC-seq approaches. The integration, analysis, and comparison of sequencing data across assays and/or time points is confusing and incomplete. For example, it should be more comprehensive to integrate both scRNA-seq and snRNA-seq data (if not, why both assays were used for Achilles tendons of both 2-week and 6-week timepoints). snRNA-seq and snATAC-seq data of 6-week-old mice were separately analyzed. No comparison of difference and similarity of TPSCs of 2-week and 6-week-old mice was conducted.

Given the goal of this work to identify specific TPSC markers, the specificity of Cd55 and Cd248 for TPSCs is not clear. First, based on the data shown here, Cd55 and Cd248 mark the same cell population which is identified by Ly6a, TPPP3, and Pdgfra. Although, for instance, Cd34 is expressed by other tissues as discussed here, no data/evidence is provided by this work showing that Cd55 and Cd248 are not expressed by other musculoskeletal tissues/cells. Second, the immunostaining of Cd55 and Cd248 doesn't support their specificity. What is the advantage of using Cd55 and Cd248 for TPSCs compared to using other markers?

Reviewer #2 (Public review):

Summary:

The molecular signature of tendon stem cells is not fully identified. The endogenous location of tendon stem cells within the native tendon is also not fully elucidated. Several molecular markers have been identified to isolate tendon stem cells but they lack tendon specificity. Using the declining tendon repair capacity of mature mice, the authors compared the transcriptome landscape and activity of juvenile (2 weeks) and mature (6 weeks) tendon cells of mouse Achilles tendons and identified CD55 and CD248 as novel surface markers for tendon stem cells. CD55+ CD248+ FACS-sorted cells display a preferential tendency to differentiate into tendon cells compared to CD55neg CD248neg cells.

Strengths:

The authors generated a lot of data on juvenile and mature Achilles tendons, using scRNAseq, snRNAseq, and ATACseq strategies. This constitutes a resource dataset.

Weaknesses:

The analyses and validation of identified genes are not complete and could be pushed further. The endogenous expression of newly identified genes in native tendons would be informative. The comparison of scRNAseq and snRNAseq datasets for tendon cell populations would strengthen the identification of tendon cell populations.

Reviewer #3 (Public review):

Summary:

In their report, Tsutsumi et al., use single nucleus transcriptional and chromatin accessibility analyses of mouse achilles tendon in an attempt to uncover new markers of tendon stem/progenitor cells. They propose CD55 and CD248 as novel markers of tendon stem/progenitor cells.

Strengths:

This is an interesting and important research area. The paper is overall well written.

Weaknesses:

Major problems:

(1) It is not clear what tissue exactly is being analyzed. The authors build a story on tendons, but there is little description of the dissection. The authors claim to detect MTJ and cartilage cells, but not bone or muscle cells. The tendon sheath is known to express CD55, so the population of "progenitors" may not be of tendon origin.

(2) Cluster annotations are seemingly done with a single gene. Names are given to cells without functional or spatial validation. For example, MTJ cells are annotated based on Postn, but it is never shown that Postn is only expressed at the MTJ, and not in other anatomical locations in the tendon.

(3) The authors compare their data to public data based on interrogating single genes in their dataset. It is now standard practice to integrate datasets (eg, using harmony), or at a minimum using gene signatures built into Seurat (eg AddModuleScore).

(4) Progenitor populations (SP1, SP2). The authors claim these are progenitors but show very clearly that they express macrophage genes. What are they, macrophages or fibroblasts?

(5) All omics analysis is done on single data points (from many mice pooled). The authors make many claims on n=1 per group for readouts dependent on sample number (eg frequency of clusters).

(6) The scRNAseq atlas in Figure 1 is made by analyzing 2W and 6W tendons at the same time. The snRNAseq and ATACseq atlas are built first on 2W data, after which the 6W data is compared. Why use the 2W data as a reference? Why not analyze the two-time points together as done with the scRNAseq?

(7) Figure 5: The authors should show the gating strategy for FACS. Were non-fibroblasts excluded (eg, immune cells, endothelia...etc). Was a dead cell marker used? If not, it is not surprising that fibroblasts form colonies and express fibroblast genes when compared to CD55-CD248- immune cells, dead cells, or debris. Can control genes such as Ptprc or Pecam1 be tested to rule out contamination with other cell types?

Minor problems:

(1) Report the important tissue processing details: type of collagenase used. Viability before loading into 10x machine.