A single-cell atlas of the cycling murine ovary

  1. Mary E Morris
  2. Marie-Charlotte Meinsohn
  3. Maeva Chauvin
  4. Hatice D Saatcioglu
  5. Aki Kashiwagi
  6. Natalie A Sicher
  7. Ngoc Nguyen
  8. Selena Yuan
  9. Rhian Stavely
  10. Minsuk Hyun
  11. Patricia K Donahoe
  12. Bernardo L Sabatini
  13. David Pépin

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This study is of interest to the readership interested in the different cell types present in the mouse adult ovary and shows how cellular states change during the four phases of the estrous cycle. This is a valuable resource for the community.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The estrous cycle is regulated by rhythmic endocrine interactions of the nervous and reproductive systems, which coordinate the hormonal and ovulatory functions of the ovary. Folliculogenesis and follicle progression require the orchestrated response of a variety of cell types to allow the maturation of the follicle and its sequela, ovulation, corpus luteum formation, and ovulatory wound repair. Little is known about the cell state dynamics of the ovary during the estrous cycle and the paracrine factors that help coordinate this process. Herein, we used single-cell RNA sequencing to evaluate the transcriptome of >34,000 cells of the adult mouse ovary and describe the transcriptional changes that occur across the normal estrous cycle and other reproductive states to build a comprehensive dynamic atlas of murine ovarian cell types and states.

Article activity feed

  1. Version published to 10.7554/elife.77239 on eLife
  2. Version published to 10.1101/2022.02.08.479522v3 on bioRxiv
  3. Version published to 10.1101/2022.02.08.479522v2 on bioRxiv
  4. eLife

    Evaluation Summary:

    This study is of interest to the readership interested in the different cell types present in the mouse adult ovary and shows how cellular states change during the four phases of the estrous cycle. This is a valuable resource for the community.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

  5. eLife

    Reviewer #1 (Public Review):

    The manuscript provides a dataset of single-cell transcriptomics of several adult mice ovaries and performs computational analysis to determine the molecular signatures of the cells isolated.

    Strengths:
    - Provide data from different estrous stages and lactating.
    - Many markers are validated.
    - Several estrous cycle-specific biomarkers are revealed.

    Weaknesses:
    - It does not stratify or provided trajectories of the data according to the different estrous stages and lactation periods.
    - Only single markers are validated, making it difficult to see the population.
    - The population of peri-ovulatory GC could be better characterised.
    - There is no mention of specific populations or states in the lactation sample.
    - Monocle analysis could be made more robust.
    - Specific populations of theca cells (interna and externa) are not named.
    - Differences between stroma 1 and stroma 2 are not found.
    - OSE is only mentioned in the Discussion.

  6. eLife

    Reviewer #2 (Public Review):

    This manuscript by Morris et al., entitled "A single cell atlas of the cycling 1 murine ovary" presented an interesting dataset for understanding cellular and transcriptional dynamics during the estrous cycle in mice. By using single-cell RNA sequencing, the authors reported new marker genes for different cell types and validated some markers using In situ hybridization. However, I believe that the main problem of this paper lies in the interpretation of data.

    The major points include:

    1. The authors used tSNE for visualization of the generated scRNA seq dataset, which, according to my knowledge, is outdated for scRNA seq data visualization as its reproducibility has become an issue. Which version of the Seurat package does the author use? And also the other software information should be implemented. Therefore, I suggest the authors reanalyze their dataset using an updated Seurat pipeline, and also reanalyze all of their data using UMAP.

    2. The authors aimed to explore the unrecognized complexity in the cellular subtypes and their cyclic expression states. For cellular heterogeneity, the authors reported detailed cell percentages of different cell types and validated the new markers using in situ hybridization, however, how do these cells change during the estrous cycle? According to the current manuscript (Figures 3A and 4A), it's hard to interpret such changes. The authors should emphasize this aspect of the description, one possible solution is to add a stack bar chart to show the proportions of different cells at different stages.

    3. For trajectory analysis, which is the root state? according to line 233 and Fig. S3B, the root state should be state 3 in Fig. S3A, is it right? The authors should clarify them. Also for each branch, adding a piechart for each branch will be more informative for the readers.

    4. In lines 235-237, the authors demonstrated that "corpus luteum clusters (CL1-3) and the periovulatory cluster were ordered on another branch suggesting this latter branching fate represents a continuum of differentiation states corresponding to luteinization of granulosa cells.", which doesn't sound convincing enough to me because there is significant overlap for cells in CL1 and CL3, not ordering. To provide a more convincing interpretation of the scRNA dataset, I suggest the authors perform RNA velocity analysis and corporate RNA velocity vectors in the trajectory plots, which will greatly help to understand the scRNA dataset.

    5. For Fig. 4A, I suggest the authors add a barplot to show the number of different cells for the different stages as for scRNA dataset, it's sometimes common that some cell types were covered by the other in the tSNE plot.

    6. For Fig. 4D, what is the expression level of these genes according to the scRNA seq dataset? A comparison of such information will increase data reliability.

    7. How did the authors identify the secreted markers using their scRNA seq dataset? An explanation should be added.

  7. eLife

    Reviewer #3 (Public Review):

    The authors sought to identify transcriptional changes that occur in the various somatic cell populations of the adult mouse ovary during different reproductive states using single-cell RNA sequencing. The ovaries for the analysis were harvested from mice during the four stages of the normal estrus cycle (proestrus, estrus, metestrus and diestrus), from lactating or non-lactating 10 days postpartum mice, and from randomly cycling mice. They identified the major cell subtypes of the adult ovary but focused their analysis on the mesenchyme (stromal and theca) and granulosa cells. They identified novel markers for stromal, theca and granulosa cell subpopulations and validated these by RNA in situ hybridization. They used trajectory analysis to infer differentiation lineages within the stromal and granulosa cell subtypes. Finally, from their data set they identify four secreted factors that could serve as biomarkers for staging estrus cycle progression.

    Strengths - This is the first study to profile ovarian somatic gonad cells at different stages of the reproductive cycle.

    Weaknesses - Enthusiasm for the current manuscript is lessened because it does not employ state-of-the-art scRNA-seq analysis. For example, once general cell populations have been determined by clustering with all cells, it is best to individually re-cluster these cell populations to identify more refined and accurate subpopulations. The PC used for the initial clustering is very useful for distinguishing different general cell populations (e.g. mesenchyme vs. granulosa vs. endothelial) but may not be as useful for distinguishing biologically relevant subpopulations (e.g. stromal subpopulations). Finally, certain cell subpopulations were excluded from the trajectory analysis without justification - specifically, the mitotic and atretic granulosa cells - calling into question what conclusions can be drawn from this analysis.

  8. eLife

    Author Response

    Reviewer 1

    The manuscript provides a dataset of single-cell transcriptomics of several adult mice ovaries and performs computational analysis to determine the molecular signatures of the cells isolated.

    Strengths:

    • Provide data from different estrous stages and lactating.
    • Many markers are validated.
    • Several estrous cycle-specific biomarkers are revealed.

    We thank the reviewer for the positive assessment of our efforts to comprehensively validate cell and estrous-cycle specific biomarkers.

    Weaknesses:

    • It does not stratify or provided trajectories of the data according to the different estrous stages and lactation periods.

    We have now added stratification of data sources in figures 1B, 5A, and 5B.

    • Only single markers are validated, making it difficult to see the population.

    While we show representative RNAish of single markers for the identification of the cellular populations, we provide heatmaps with complete signatures in Figures 1C, 2B, 3B, 4B, Dotplots in figures 4D, 5D, 6A, Figure 2 – supplement 1D,E, feature plots in Figure 3-supplement 1A,B, as well as fully referenced tables of validated markers in Supplementary Files 2, 3, and 4.

    • The population of peri-ovulatory GC could be better characterized.

    We now provide Oxtr as a more specific marker of peri-ovulatory Gc (see figure 3C and figure 3- supplement 1D), in addition to the validated markers presented in Supplementary File 4.

    • There is no mention of specific populations or states in the lactation sample.

    We now provide cluster composition by sample type in Figure 1B, as well as represent cell state differences in lactating samples in S3E, which are discussed in lines 196-197 and 272-275.

    • Monocle analysis could be made more robust.

    Since the graphical representation of lineage pseudotime trajectories of granulosa cells was counter intuitive, we have removed this analysis in favor or more concise explanation of differentiation and cell states amongst the mural and cumulus granulosa cells and their response to LH in lines 378-382.

    • Specific populations of theca cells (interna and externa) are not named.

    Specific populations of theca are now shown in figures 2 and figure 2 – supplement 1 and more extensively described in the results (lines 230-244) and discussion (lines 397-417).

    • Differences between stroma 1 and stroma 2 are not found.

    After reanalyzing the data, we concluded that these two interstitial stromal cell clusters could not be differentiated by specific dichotomous markers. Nevertheless, we noted that the expression of Ectonucleotide Pyrophosphatase/Phosphoiestrase 2 (Ennp2) was specific and limited to one of the clusters. Given that this same cluster expressed markers such as Col1a1, Lum, Loxl1, shown in the literature to be characteristic of fibroblast, we named them fibroblast-like stroma. The second subcluster, Enpp2 negative, was shown to be enriched in expression of steroidogenic markers such as Cyp11a1, and Cyp17a1 and was named steroidogenic stroma cluster. These results are now presented in figure 2 - supplement 1E,F.

    • OSE is only mentioned in the Discussion.

    The findings in OSE are now presented in Figure 4 and discussed in the results (lines 287-294) and discussion (lines 418-429).

    Reviewer 2

    This manuscript by Morris et al., entitled "A single cell atlas of the cycling 1 murine ovary" presented an interesting dataset for understanding cellular and transcriptional dynamics during the estrous cycle in mice. By using single-cell RNA sequencing, the authors reported new marker genes for different cell types and validated some markers using In situ hybridization. However, I believe that the main problem of this paper lies in the interpretation of data.

    The major points include:

    1. The authors used tSNE for visualization of the generated scRNA seq dataset, which, according to my knowledge, is outdated for scRNA seq data visualization as its reproducibility has become an issue. Which version of the Seurat package does the author use? And also the other software information should be implemented. Therefore, I suggest the authors reanalyze their dataset using an updated Seurat pipeline, and also reanalyze all of their data using UMAP.

    The data have been reanalyzed with UMAP instead of tSNE. The R version is 4.1.3, it is now indicated in the Material and Methods section.

    1. The authors aimed to explore the unrecognized complexity in the cellular subtypes and their cyclic expression states. For cellular heterogeneity, the authors reported detailed cell percentages of different cell types and validated the new markers using in situ hybridization, however, how do these cells change during the estrous cycle? According to the current manuscript (Figures 3A and 4A), it's hard to interpret such changes. The authors should emphasize this aspect of the description, one possible solution is to add a stack bar chart to show the proportions of different cells at different stages.

    We added a plot showing the composition of each cluster by stage of the estrous cycle in figure 1B, 5A, and 4C. The percentage of cells in each cluster has been added to each feature plot.

    1. For trajectory analysis, which is the root state? according to line 233 and Fig. S3B, the root state should be state 3 in Fig. S3A, is it right? The authors should clarify them. Also for each branch, adding a piechart for each branch will be more informative for the readers.

    Given that the monocle pseudotime trajectories of granulosa cells were difficult to interpret, we have removed this analysis from the manuscript.

    1. In lines 235-237, the authors demonstrated that "corpus luteum clusters (CL1-3) and the periovulatory cluster were ordered on another branch suggesting this latter branching fate represents a continuum of differentiation states corresponding to luteinization of granulosa cells.", which doesn't sound convincing enough to me because there is significant overlap for cells in CL1 and CL3, not ordering. To provide a more convincing interpretation of the scRNA dataset, I suggest the authors perform RNA velocity analysis and corporate RNA velocity vectors in the trajectory plots, which will greatly help to understand the scRNA dataset.

    Unfortunately, the inDROP pipeline was not compatible with RNAvelocity. Other pseudotime analyses gave similar representation of trajectories. We have therefore elected to remove pseudotime analysis from the manuscript.

    1. For Fig. 4A, I suggest the authors add a barplot to show the number of different cells for the different stages as for scRNA dataset, it's sometimes common that some cell types were covered by the other in the tSNE plot.

    We thank the reviewer for this suggestion, we have now added the percentage of clusters by sample type in every UMAP dimplot.

    1. For Fig. 4D, what is the expression level of these genes according to the scRNA seq dataset? A comparison of such information will increase data reliability.

    We thank the reviewer for this suggestion, we have added a dotplot representing the level of expression of each gene in the scRNAseq dataset beside the qPCR graphs (now Fig 5D and S5E).

    1. How did the authors identify the secreted markers using their scRNA seq dataset? An explanation should be added.

    We screened the significantly differentially expressed genes by estrous stage for proteins predicted to be secreted according to UniProt annotation (Table S5), and prioritized genes with the highest fold change. The text has been modified to include this information (lines 320-324).

    Reviewer 3

    The authors sought to identify transcriptional changes that occur in the various somatic cell populations of the adult mouse ovary during different reproductive states using single-cell RNA sequencing. The ovaries for the analysis were harvested from mice during the four stages of the normal estrus cycle (proestrus, estrus, metestrus and diestrus), from lactating or non-lactating 10 days postpartum mice, and from randomly cycling mice. They identified the major cell subtypes of the adult ovary but focused their analysis on the mesenchyme (stromal and theca) and granulosa cells. They identified novel markers for stromal, theca and granulosa cell subpopulations and validated these by RNA in situ hybridization. They used trajectory analysis to infer differentiation lineages within the stromal and granulosa cell subtypes. Finally, from their data set they identify four secreted factors that could serve as biomarkers for staging estrus cycle progression.

    Strengths - This is the first study to profile ovarian somatic gonad cells at different stages of the reproductive cycle.

    We thank the reviewer for the positive assessment of our efforts to profile ovarian gonadal somatic cells comprehensively across the estrous cycle

    Weaknesses - Enthusiasm for the current manuscript is lessened because it does not employ stateof-the-art scRNA-seq analysis. For example, once general cell populations have been determined by clustering with all cells, it is best to individually re-cluster these cell populations to identify more refined and accurate subpopulations. The PC used for the initial clustering is very useful for distinguishing different general cell populations (e.g. mesenchyme vs. granulosa vs. endothelial) but may not be as useful for distinguishing biologically relevant subpopulations (e.g. stromal subpopulations). Finally, certain cell subpopulations were excluded from the trajectory analysis without justification - specifically, the mitotic and atretic granulosa cells - calling into question what conclusions can be drawn from this analysis.

    We have re-analyzed our dataset using the most up to date version of Seurat and R, including reclustering, and changed our dimensional reduction to UMAP. We have also removed pseudotime analysis from the manuscript, given the difficulties in interpreting and representing trajectories.

  13. Version published to 10.1101/2022.02.08.479522v1 on bioRxiv