CRISPR loss of function screening to identify genes involved in human primordial germ cell-like cell development

  1. Young Sun Hwang
  2. Yasunari Seita
  3. M. Andrés Blanco
  4. Kotaro Sasaki

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Despite our increasing knowledge of molecular mechanisms guiding various aspects of human reproduction, those underlying human primordial germ cell (PGC) development remain largely unknown. Here, we conducted custom CRISPR screening in an in vitro system of human PGC-like cells (hPGCLCs) to identify genes required for acquisition and maintenance of PGC fate. Amongst our candidates, we identified TCL1A , an AKT coactivator. Functional assessment in our in vitro hPGCLCs system revealed that TCL1A played a critical role in later stages of hPGCLC development. Moreover, we found that TCL1A loss reduced AKT-mTOR signaling, downregulated expression of genes related to translational control, and subsequently led to a reduction in global protein synthesis and proliferation. Together, our study highlights the utility of CRISPR screening for human in vitro-derived germ cells and identifies novel translational regulators critical for hPGCLC development.

Article activity feed

  1. Version published to 10.1371/journal.pgen.1011080
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    The authors do not wish to provide a response at this time.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Summary:

    In this study, Hwang et al. develop an inducible Cas9 hiPSC line and perform with it a pooled CRISPR knockout screen using a custom sgRNA library to identify novel genes involved in human primordial germ cell like cell (hPGCLC) differentiation in vitro. Thereby they find the AKT coactivator TCL1A to be important in the proliferation/survival of hPGCLCs after specification.

    Specific Comments:

    1.) p.7-p.8: "Using the MAGeCK algorithm (Li et al, 2014) to call hits on merged replicates, 25 genes scored as significantly depleted from the AG+ population at p < 0.05. Among the top hits was SOX17, and near-hits included TFAP2C, both of which are well-known drivers of the hPGCLC state (Fig. S2E)."

    When looking at Table S3, it appears that only 2 genes were significantly depleted (P < 0.05) in replicate 1, 23 in replicate 2 and 10 genes when rep 1 and rep 2 were analyzed together. The essential germ cell genes SOX17 and TFAP2C were not significantly depleted in replicate 1 and only TFAP2C but not SOX17 was depleted significantly in replicate 2. Also the main hits discussed in this paper, METTL7 and TCL1A were not significantly depleted in replicate 1 and only METTL7 but not TCL1A was significantly depleted in replicate 2. This indicates that replicate 1 might not have been robust enough to reliably detect depleted genes and that TCL1A was not among the significant hits. A potential explanation could be that not enough cells have been used to ensure a sufficient representation of sgRNAs to provide significant results in a depletion screen. Ideally the screen would need to be repeated to provide another informative replicate, or the authors should at least correct the sentences above and openly state that their hits are only based on one replicate of the screen and that the list of their hits might therefore not be fully reliable. Also the statement on page 8 that "the screen was both technically and biologically successful" might need to be toned down.

    2.) p.8 and Fig. 2E: The authors do not clearly describe, what are the 25 top hits in PGCLC(+) and PGCLC(-) cells and how they were chosen (Score, p-value or fold change), which they compared in the gene set enrichment analysis to the RNA-Seq data.

    3.) Fig. S3I-K: The authors mention in the text on p. 9 a significant reduction in hPGCLC induction efficiency for both TCL1A and METTL7A KO cells, but they do not provide statistics and do not mention, how many biological replicates have been used. As hiPSCs generally show a high clone to clone variability in hPGCLC induction efficiency, results from a single KO clone can not be considered as a reliable result. The authors should provide results from additional wt and KO clones (they are showing in S3E multiple for each gene) to ensure reliable effects, especially for METTL7A KO cells, where the reduction in PGCLC induction efficiency is more modest and might not be significant (Fig. S3I). Another way of validating the phenotype would be to use individual control, TCL1A and METTL7A sgRNAS from the screen and compare induction efficiencies with or without DOX-induced Cas9 expression.

    4.) Fig. 3F, Supp Tables S4, S5, S6: It is not clearly described, what was the criteria to define DEGs for the GO term analysis of TCL1A KO cells, as more genes have been used than the relatively few significant DEGs reported in Table S4. Furthermore, only FDRs or otherwise adjusted p-values (not raw p-values as done in the figure and tables) should be used to determine significantly enriched GO terms. Also no representative gene names are displayed in Fig.3F, as stated in the figure legend.

    5.) Fig.4: p-AKT and p-mTOR signaling are represented as the mechanism, by which TCLA1 KO affects hPGCLC maintenance/proliferation. It is not clear from the presented data, what is meant with biological triplicates (different germ cell inductions, different subclones) mentioned in the figure legend. As some of the effects observed are quite small (e.g. p-mTOR differences in 4E, cell cycle differences in 4J), biological replicates with a different KO clone should be performed (see point 3). Otherwise it is hard to judge how robust the data and conclusions derived are.

    CROSS-CONSULTATION COMMENTS

    Apart from my points regarding the weakness in statistical confidence I agree with the other reviewers that it needs to be shown whether the effect of TCL1A KO is based on a general proliferation defect of the entire aggregation body or if the effect is really hPGCLC-specific.

    Significance

    The study is the first CRISPR screen performed during hPGCLC differentiation and provides a proof of principle for a useful tool to allow dissection of gene regulatory networks during human germ cell development. Overall this is a technical advancement and an ambitious study and will generate interest in the human germ cell field. Generally it is easy to follow but could be improved in the description of some of the methodology (points 2+4). Overall the study suffers from weaknesses in the statistical robustness, as the screen itself did not provide many significant hits (major point 1) and the follow up was only performed using a single KO clone (points 3+5). Therefore adding replicates would be necessary to strengthen the confidence in the drawn conclusions.

    Reviewers' relevant expertise: in vitro germ cell differentiation, pluripotency, CRISPR screening

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    The authors conducted a custom CRISPR screening including 422 coding-genes in hPGC-like cells (hPGCLCs) to identify genes important to hPGCLC. Based on the screen, they found two candidates, TCL1A and METTL7A, that regulate hPGCLC specification. They concluded that TCL1A, an AKT coactivator, is critical for hPGCLC specification through regulating AKT-mTOR signaling. Unfortunately, we found that the evidences for the key conclusions are not quite convincing. Reasons are as below:

    1. The results to demonstrate the key role of TCL1A on hPGCLC specification is not convincing. Fig. 3B, C & D. the cell number per aggregate is also significantly reduced in TCL1A KO (2843/20.7% =13734) compared to that in WT cells (545/16.4%=3323). Despite that, the percentage of AG+ cells per aggregate is significantly, but not dramatically decreased in TCL1A KO (20.7%) vs WT (16.4%) cells. Thus, the effect of TCL1A KO may not be specific on the AG+ cells, but on the whole aggregate.
    2. The overall effect of METTL7A KO on hPGCLC development is too moderate to conclude it as a key regulator for hPGCLC specification.

    Significance

    a CRISPR screening for key regulator for human germline cell apecification was not reported before.

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    In this manuscript, the authors performed CRISPR/Cas9-based screening to identify genes involved in human germ cell development. Using human PGCLC system, the authors found that sgRNAs targeting TCL1A and METTL7A genes were enriched in non-PGCLC population. Gene-disruption of each gene resulted in a significant reduction in PGCLC differentiation. Moreover, TCL1A-knockout PGCLCs failed to proliferation during PGCLC induction, possibly due to attenuation of AKT signaling. Further analyses showed that protein synthesis and cell-cycle progression, especially in S-phase, were impaired in TCL1A-knockout PGCLCs. Thus, the authors provided remarks on successful identification of genes functionally important for human PGCLC differentiation and the importance of translational control in human germ line.

    This manuscript demonstrates a genetic screening for genes involved in hPGCLC differentiation. Although the number of genes targeted by sgRNAs were limited (422 genes), it is still valuable to show such genetic screening can be applied to the hPGCLC system. Overall, statements and data in the manuscript are convincing, except for following points, and the novelty is sufficient for publication. It would be further improved, considering following points with additional experiment, if feasible.

    1. A major concern in this manuscript is whether TCL1A function is specifically involved in hPGC development or generally important for other cell type. As AKT signaling plays multiple roles on many cell contexts, this is important to verify the author's conclusion. For example, is there any defect in proliferation of TCL1A-knockout iPS lines? The authors should quantify the doubling rate of TCL1A-knockout iPS cells, the number of iMeLCs yielded, and the cell number included in the aggregates. Looking at Figure S3K and K, the total number of the cells in the aggregates seems lower in TCL1A-knockout aggregates than in WT.

    2. Related to the comment above, the author should add a statement describing expression pattern and level of TCL1A and METTL7A in tissue. Are they preferentially expressed in the germ line, or generally expressed in a broad range of tissue?

    3. The quantification of pAKT and p-mTOR is vague. The authors should quantify in a different way. Although the author claimed that Western blot analysis was not able to detect pAKT and p-mTOR in PGCLCs, there are a number of reports that detect these proteins. As an advantage of PGCLC system is to handle a large number of cells in culture, the authors should perform rigorously the quantification.

    4. In the abstract, there is a statement "demonstrate the importance of translational control in human reproduction". Isn't this too general? Is it a new finding? With critical examination for cell-specificity of the translational control by TCL1A as described above, this should be refined.

    Significance

    As authors performed CRISPR/Cas9-based screening to identify genes involved in human germ cell development using human PGCLC system, and then successfully isolated TCL1A and METTL7A genes that are previously known as drives for PGCLC induction. Therefore, from both technological and scientific viewpoints, there is a significant advance shown in this manuscript.

  6. Version published to 10.1101/2022.05.23.493064v1 on bioRxiv