Lineage-specific differences and regulatory networks governing human chondrocyte development

Reviewer #2 (Recommendations for the authors):

1. Although the study is aiming to provide an in vitro model for studying human cartilage development. Only samples of one time point were analyzed for both hESC-derived cartilage and human fetal cartilage, including terminal differentiated hESC-derived cartilage and E67 fetal donor femur cartilage. The information can hardly be defined as a developmental biology analysis.

We thank this Reviewer for this terrific suggestion and we have now included a new set of experiments in which we performed bulk RNA sequencing analysis on early, mid, and late stage cartilage tissues during differentiation. The late stage cartilage tissues showed similar gene expression differences to those we identified in the initial transcriptomic dataset (Results, Page 6). We present the biological processes associated with DEGs identified at early and mid-stages of differentiation, with an additional focus on the transcription factors both unique to each developmental stage, and those that are shared at all stages during articular or growth plate cartilage differentiation in vitro. These new data can be found in Figure 1—figure supplement 3 and supplementary file 2.

2. Single cells RNA sequencing is becoming the current standard for most sequencing-related studies. The bulk RNA sequencing analysis used in this study provided much less information and is below the current standard. The sequencing results are, therefore, less valuable as a resource for other researchers in the field.

We very much appreciate this suggestion, as we also believe that single cell transcriptomic data, particularly of human cartilage, is highly valuable as a resource for other researchers. This work was initiated by a senior fellow in the Craft laboratory. However, we must respectfully disagree on the bulk RNA sequencing providing less information, particularly for our goal of defining differential transcription factor expression. Single cell RNA sequencing is less powered to detect statistically significant differences in low copy number transcripts, notably including transcription factors, due to significantly lower read depth than that which can be obtained by bulk RNA sequencing. Methods for integrating bulk RNA-seq and ATAC-seq were available when the present work was performed, while methodology to obtain matched RNA- and ATAC-seq from individual cells, especially those that must be isolated from dense extra-cellular matrix environments, is currently being optimized. We think the bulk data are interesting, important, and appropriate to publish now. However, we agree with this Reviewer about the importance of adding single cell data when we are confident those data are as robust and reproducible as our bulk data.

3. The validation evidence provided in the study is too weak. Transgenic mouse models are strongly recommended to validate the transcript regulation target identified in the analysis.

We agree that the function of many of our newly identified hits should be validated in vivo using transgenic animal models, and we highlight this limitation in the Discussion (Page 13). We have referred to published studies for several of the transcription factors we found in our analyses. Our discovery of known chondrogenic factors in our analyses suggests that the new targets we identified may also have biological functions in chondrocyte biology. We strongly believe that the resource we are providing to the community will lead to such investigations.