Dynamic and regulated TAF gene expression during mouse embryonic germ cell development
Megan A. Gura aff001; Maria M. Mikedis aff002; Kimberly A. Seymour aff001; Dirk G. de Rooij aff002; David C. Page aff002; Richard N. Freiman aff001
Authors place of work:
Brown University, MCB Graduate Program and Department of Molecular Biology, Cell Biology and Biochemistry, Providence, RI, United States of America
aff001; Whitehead Institute, Cambridge, MA, United States of America
aff002; Department of Biology, Massachusetts Institute of Technology, Cambridge, MA, United States of America
aff003; Howard Hughes Medical Institute, Whitehead Institute, Cambridge, MA, United States of America
Published in the journal:
Dynamic and regulated TAF gene expression during mouse embryonic germ cell development. PLoS Genet 16(1): e1008515. doi:10.1371/journal.pgen.1008515
Germ cells undergo many developmental transitions before ultimately becoming either eggs or sperm, and during embryonic development these transitions include epigenetic reprogramming, quiescence, and meiosis. To begin understanding the transcriptional regulation underlying these complex processes, we examined the spatial and temporal expression of TAF4b, a variant TFIID subunit required for fertility, during embryonic germ cell development. By analyzing published datasets and using our own experimental system to validate these expression studies, we determined that both Taf4b mRNA and protein are highly germ cell-enriched and that Taf4b mRNA levels dramatically increase from embryonic day 12.5–18.5. Surprisingly, additional mRNAs encoding other TFIID subunits are coordinately upregulated through this time course, including Taf7l and Taf9b. The expression of several of these germ cell-enriched TFIID genes is dependent upon Dazl and/or Stra8, known regulators of germ cell development and meiosis. Together, these data suggest that germ cells employ a highly specialized and dynamic form of TFIID to drive the transcriptional programs that underlie mammalian germ cell development.
Healthy development and maintenance of germ cells is essential for the continuation of all sexually reproducing species. In the United States, approximately 10% of individuals face fertility issues, and worldwide, at least 1% of females experience primary ovarian insufficiency (POI), which is associated with infertility [1,2]. However, in most instances of POI and infertility, the underlying molecular causes are unknown . Work from our laboratory has shown that a TBP-Associated Factor 4b (Taf4b)-deficient mouse model recapitulates many aspects of POI, including a premature depletion of the ovarian reserve and female infertility [4,5]. Taf4b-deficiency in the mouse also leads to progressive male infertility in early adulthood that is associated with exhaustion of their adult spermatogonial stem cell (SSC) pool . Since human studies implicate TAF4B as important for fertility and oocyte quality [7–9], our efforts to understand the molecular mechanisms underlying the expression and function of TAF4b in mouse germ cell development may contribute to our increased understanding of human fertility and infertility.
TAF4b is a gonad-enriched subunit of the basal transcription factor TFIID, which is a complex of TATA-binding protein (TBP) and 13–14 TBP-associated factors (TAFs) [5,10]. TFIID, as part of a larger ensemble of the basal transcription machinery, recruits RNA Polymerase II (RNAPII) to the core promoters of genes. However, unlike its paralog Taf4a, which is expressed ubiquitously, Taf4b is more highly expressed in the mouse ovary and testis compared to other tissues . Most recently, we demonstrated that TAF4b is a critical regulator of female meiosis I in the mouse . Taf4b mRNA expression is also highly correlated with the expression of many important germline genes during human fetal ovary development, such as Deleted in Azoospermia-like (Dazl), which encodes a germ cell-specific RNA-binding protein that promotes translation and/or prevents the degradation of its target mRNAs [11–13]. In female mice as early as E13.5, when germ cells initiate meiosis, Taf4b-deficiency results in reduced expression of meiotic transcripts, including Stimulated by Retinoic Acid 8 (Stra8), which is a master regulator of meiosis . This reduced meiotic gene expression is followed by defective meiotic progression, and excessive perinatal oocyte loss . Stra8 mRNA has also been found to be greatly reduced in Taf4b-deficient male newborn testes compared to wild-type and Taf4b-heterozygous testes, indicating that this disruption of meiosis is likely shared between female and male Taf4b-deficient mice .
We previously detected TAF4b occupancy at the proximal promoters of Dazl and Stra8 in E18.5 embryonic ovaries, which suggests that TAF4b directly regulates these genes . Stra8 was recently identified as a major transcriptional regulator of meiotic initiation . However, it was postulated that Stra8 alone is not sufficient to induce meiosis, and other epigenetic and transcriptional regulators likely work together with Stra8 to orchestrate meiotic initiation. It is unknown what transcriptional program coordinates with Stra8 in germ cells to produce gametes.
While female germ cells initiate meiosis at E13.5, male germ cells remain mitotically active until ~E16.5 . During this time, Taf4b-deficient male mice display deficient embryonic gonocyte numbers, as well as defective postnatal germ cell proliferation and development . Therefore, during early postnatal development, a paucity of germ cells leads to a narrow window of fertility that may result from an inability to establish a healthy embryonic germ cell and/or postnatal SSC pool. One common theme in both adult females and males is that their Taf4b-dependent fertility deficits can be traced back to defects in embryonic germ cell development.
Despite the knowledge that Taf4b function is essential for embryonic germ cell development in both sexes, the precise expression profile and mechanism of action for this fertility transcription factor remains unknown. Thus, it is uncertain whether Taf4b uses similar molecular mechanisms in female and male germ cells. This gap in knowledge makes understanding the role of Taf4b in embryonic gametogenesis more challenging. To develop a better understanding of Taf4b, we implemented complementary computational and experimental tools to assess the sex-specific timing and localization of Taf4b expression in the embryonic mouse gonad. We found that Taf4b mRNA and protein are highly enriched in the germ cells of the embryonic gonad and that strikingly similar expression patterns were observed for other TFIID subunits, such as Taf7l, which is essential for male germ cell maturation and fertility [17,18]. In addition, we demonstrate that Taf4b is the only component of the TFIID and SAGA complexes that is directly regulated by both DAZL and STRA8 during meiosis. Based upon these data, Taf4b functions in an interdependent gene regulatory network with DAZL and STRA8 during meiosis.
Taf4b is significantly enriched and dynamically expressed in the germ cells of the embryonic ovary and testis
To evaluate Taf4b expression in the embryonic gonad, we reprocessed an RNA-sequencing (RNA-seq) dataset that used Oct4-EGFP mice covering 7 developmental time points, which range from embryonic day 9.5 (E9.5) to E18.5, wherein each time point had appropriate female and male replicates [19,20]. This sex-specific time course was further broken down into germ cell (GFP+) and gonadal somatic (GFP-) cells by performing fluorescence activated cell sorting (FACS). Principal component analysis (PCA) (Fig 1) indicates distinct grouping of germ cells and somatic cells on opposite sides of the PC1 dimension and further grouping of germ cells based on their developmental milestones.
We first examined in which embryonic gonadal cell type Taf4b mRNA is expressed since the embryonic ovary and testis are heterogeneous tissues composed of germ and somatic cells. In both tissues from E11.5 to E18.5, Taf4b mRNA is significantly and consistently enriched in the GFP+ germ cells (Fig 2A and 2B, log2FC > |0.25|, p-adj. < 0.05, S1 Table). Taf4b expression, shown in transcripts per million (TPM, S2 Table), remains relatively low in all gonadal somatic cells throughout the time course. The slight increase in Taf4b mRNA in male somatic samples at E18.5 (Fig 2B) is not significant. Thus, Taf4b mRNA expression is preferentially expressed in the germ cells of the embryonic gonad. We hypothesize that Taf4b plays a specialized role in germ cell development and fertility in contrast to its more ubiquitously expressed paralog Taf4a. We thus compared Taf4a mRNA to Taf4b expression levels in this dataset (Fig 2C and 2D). In contrast to Taf4b, Taf4a mRNA expression was in general not significantly different between germ cell and somatic cell types. Furthermore, Taf4a mRNA was detected at lower TPMs over the entire time course, which indicates that Taf4a mRNA is reduced compared to Taf4b mRNA.
To test if the enrichment of Taf4b mRNA within germ cells is reflected at the protein level, we performed western blots on whole cell protein extracts of E13.5 germ and somatic cells derived from a similar Oct4-EGFP mouse line. At E13.5, the signal for mouse vasa homolog (MVH), a germ cell-specific marker, is only found in the GFP+ protein samples of female and male cells (Fig 2E). This marker expression indicates that we successfully separated the germ cells from the somatic cells. Based on quantification of western blot signal from 3 biological replicates, TAF4b protein expression is predominantly detected in the GFP+ germ cells from both female and male samples, whereas TAF4a is expressed in both GFP+ germ cells and GFP- somatic cells (Fig 2E and 2F, S1A–S1D Fig). Even at higher GFP- somatic cell numbers, a TAF4b protein signal is barely detectable (S1E Fig), indicating that TAF4b protein is strongly enriched in the germ cells of E13.5 mouse gonads and suggesting that the expression of both Taf4b mRNA and protein during embryonic reproductive development is primarily within the germ cells.
Given its expression in germ cells, we more closely examined the mRNA expression of Taf4b versus Taf4a in the E9.5-E18.5 time course. Taf4b dramatically increases at E15.5 in both female and male germ cells while Taf4a expression is relatively constant and much lower (Fig 2A–2D, dark gray bars). However, the female expression of Taf4b reaches greater TPM than the males, as indicated by the differences in the scale of the y-axes. Similar Taf4b and Taf4a mRNA expression profiles can be reliably found in independent RNA-seq datasets that isolate germ cells from the embryonic gonad (S2 Fig). These data suggest that as germ cells progress through development, and enter meiosis in female germ cells, Taf4b mRNA and protein are available to play a role in these processes. In addition, when the human embryonic gonad is sorted into germ and somatic cell populations, TAF4B mRNA expression is similarly germ cell-enriched (Fig 3A and 3B), and increases over comparable time points in the available human datasets (Fig 3C and 3D, S3–S5 Tables), indicating that mouse and human TAF4B may play analogous roles during embryonic germ cell development [21–23].
While we identified many differentially expressed genes in pairwise comparisons of germ versus somatic cell gene expression across time (S1 Table), we used the program ImpluseDE2  for time course differential expression analysis in female and male germ cells. ImpulseDE2 determined that Taf4b, but not Taf4a mRNA expression was significant (p-adj. < 0.05) across both female and male germ cell timelines, referred to as “significantly dynamic” (S6 Table). Together these data suggest that Taf4b expression is nearly exclusive to the germ cells of the embryonic gonad and in the female this corresponds to our time period of interest, when meiosis and primordial follicle development have initiated.
Many TFIID subunits are dynamically expressed in embryonic germ cells
Using these ImpulseDE2 data, we asked whether other subunits of the TFIID complex were co-regulated with Taf4b. To our surprise, many other TFIID subunits were also found to be significantly dynamic in the female timeline, male timeline, or both (Fig 4A). One especially notable TAF variant that was found to be significantly dynamic in both female and male embryonic germ cells was Taf7l. Taf7l is X-linked and known to be essential for male fertility [17,18]. Interestingly, Taf7l is a paralog of Taf7 and has an autosomal retrotransposed copy called Taf7l2 (previously known as “4933416C03Rik” ). ImpulseDE2 identified Taf7l2 and Taf7 as significantly dynamic in the female and male germ cells, respectively. In female germ cells, Taf7l mRNA expression drastically increases starting at E14.5 and remains high (Fig 4B). Taf7l2 appears to mimic the female gene expression trajectory of Taf7l albeit at a lower TPM level (S3A Fig). In male germ cells, Taf7 is the most abundant of the three, with Taf7l increasing in expression at E14.5, like its female germ cell expression (Fig 4C). When looking more closely at only Taf7l and Taf7l2, we see that Taf7l begins to increase in expression at E14.5, like in females, but Taf7l2 appears to increase in expression around E18.5 and does not mimic the male trajectory of Taf7l (S3B Fig). These data suggest that both embryonic female and male germ cells share Taf7l expression but express Taf7l2 and Taf7 in female- and male-specific ways, respectively.
Furthermore, the germ cell enrichment or specificity of Taf7l, Taf7l2, and Taf7 depends on the sex of the germ cells. In females, Taf7l and Taf7l2 are significantly and consistently germ cell-specific starting at E14.5 whereas in males it is Taf7 and Taf7l that are significantly and consistently germ cell-enriched (S3C–S3H Fig). This implicates Taf7l as a shared TFIID component of importance between female and male germ cells while Taf7l2 and Taf7 display a female or male germ cell-specific preference, respectively. However, in the available human data, TAF7L is not detectable in the embryonic gonad 8–20 weeks post fertilization, TAF7 shows a non-significant preference for somatic cell expression at 7 weeks post-fertilization, and Taf7l2 is a rodent-exclusive retrogene which altogether might indicate that this trio is less important in human prenatal germ cell development (S3I–S3L Fig).
Since both Taf4b and Taf7l are germ cell-enriched, and several TFIID subunits were identified as significantly dynamic by ImpulseDE2 in the mouse, we asked if any other components of TFIID are also germ cell-enriched like Taf4b and Taf7l. If individual TFIID components had significantly (log2FC > |0.25|, p-adj. < 0.05) higher mRNA expression in the germ cells over the somatic cell samples for at least 3 of the 7 measured time points in either sex, it was considered “germ cell-enriched”. We found that 14 additional TFIID-encoding genes were germ cell-enriched (S3M Fig), demonstrating that germ-cell enrichment of TFIID complex genes is not unique to Taf4b and Taf7l.
Taf4b, Taf7l and Taf9b cluster with critical germ cell development and meiotic genes
The enrichment and dynamic expression of TFIID subunits in both female and male germ cells led us to question if these TFIID components were coordinately expressed. We performed k-means clustering on the top 10,000 most variable genes across all germ cell samples in the reprocessed mouse time course dataset (Fig 5). Nine clusters were chosen based on the elbow method of determining k clusters (S4 Fig). Using this clustering method, 11 TAFs were found in our 9 clusters (Table 1). Cluster D (Fig 5, red box) is particularly interesting because 5 TAFs (Taf4b, Taf5, Taf7l, Taf7l2, and Taf9b) clustered with genes highly relevant to overall germ cell development (Dazl, Ddx4, Ybx2) and meiosis I (Stra8, Meioc, Ythdc2). Furthermore, gene ontology of Cluster D indicates that there is significant enrichment of meiosis I and germ cell development genes in this group (Table 1, S7 Table).
It is very interesting to find that Taf4b, a gene with known connections to meiotic genes, clusters with Taf5, Taf7l, Taf7l2, and Taf9b, which have not been previously associated with meiosis. Furthermore, all but Taf5 are alternative TAFs, meaning that their place in the TFIID complex is not guaranteed but could instead be occupied by their counterpart. For example, both TAF4a and TAF4b are capable of replacing each other in TFIID.
Loss of Dazl and Stra8 regulation affect proper expression of many TFIID components, including Taf4b
To further examine the potential relevance of TFIID in germ cells and what could be connecting their expression patterns, we reprocessed three RNA-seq datasets from Dazl- and Stra8-KO female and male mice [13,14,25]. Both Dazl and Stra8 are germ cell-specific regulators of gene expression that are essential for meiosis and fertility. Dazl encodes an RNA-binding protein that may promote translation of its target mRNAs and/or stabilize transcripts to prevent their degradation [12,13]. Stra8 encodes a sequence-specific transcriptional activator that amplifies meiotic and cell cycle genes to initiate meiosis I .
To determine which TAFs, if any, were deregulated when Dazl expression is disrupted, we reprocessed two Dazl-KO RNA-seq datasets. For the female Dazl-KO, Soh et al. performed RNA-seq on E14.5 whole ovaries with a global disruption of the Dazl gene (S8 Table). Dazl is expressed in a germ cell-specific manner, so noise from the somatic cells of the ovary is minimized. For the male Dazl-KO, Zagore et al. used the same Dazl mutation but combined its usage with a Stra8-Cre; IRG+ system, which expresses RFP before Cre recombination and EGFP after, so that EGFP+ spermatogonia could be collected via FACS at PND6 despite low germ cell numbers (S9 Table) . Even with different female and male biological contexts in these RNA-seq datasets, we found many components of the TFIID complex to be significantly affected (log2FC > |0.25|, p-adj < 0.05), with Taf4b, Taf7l, and Taf9b significantly reduced in both sexes of the Dazl-KO mouse (Fig 6A). Closer examination of these data indicates that Taf4b, but not Taf4a, is significantly decreased in the female and male Dazl-KO mouse (Fig 6B and 6C). Additionally, Taf7l and Taf7l2 are significantly decreased in the female Dazl-KO data while Taf7 and Taf7l are significantly decreased in the male Dazl-KO (Fig 6D–6E).
To determine which TAFs, if any, might be dependent on Stra8 expression, we reanalyzed two Stra8-KO RNA-seq datasets. For the female Stra8-KO, Soh et al. also performed RNA-seq on E14.5 whole ovaries in a global disruption in the Stra8 gene (S8 Table). For the male Stra8-KO, Kojima et al. used the same Stra8-KO crossed to a Ddx4-Cre; Rosa26-tdTomato mouse line and isolated preleptotene male germ cells (S10 Table). Like in the Dazl-KO data, Taf4b and Taf7l were significantly reduced in the female and male Stra8-KO gonads along with many other TAFs, including Taf9b in the females (Fig 7A).
The mRNA dynamics of Taf4a and Taf4b in the Stra8-KO data are much like the Dazl-KO data, with Taf4b being significantly decreased in the Stra8-KO female and male samples (Fig 7B and 7C). Interestingly, Taf4a is significantly increased in the Stra8-KO preleptotene germ cells. Both Taf7l and Taf7l2 are significantly decreased in the Stra8-KO but Taf7 is unchanged (Fig 7D and 7E). These data suggest that not only do DAZL and STRA8 regulate several TFIID components not traditionally thought of as part of the germ cell development gene program, but also that Taf4b and Taf7l are consistent, active members of both the female and male germ cell-specific programs of gene expression.
DAZL and STRA8 directly bind to many of the TFIID components improperly expressed in Dazl- and Stra8-KO mice
To inquire as to whether DAZL may directly regulate any TFIID components, we identified DAZL-bound transcripts via individual-nucleotide resolution cross-linking and immunoprecipitation (iCLIP) (S5A Fig) . Given that we have previously shown that Taf4b impacts meiotic prophase I in females and males , we sought to understand whether Taf4b is regulated by DAZL in a meiotic context, focusing on the leptotene stage of meiotic prophase I. To obtain the large number of leptotene spermatocytes needed for iCLIP, we developmentally synchronized spermatogenesis by chemically regulating the levels of retinoic acid, which is required for spermatogonial differentiation [28,29]. This synchronized development allowed us to collect testes enriched for leptotene spermatocytes. The successful accumulation of leptotene spermatocytes, without contamination from other spermatocyte stages, was verified in a testis biopsy via histological analysis .
DAZL iCLIP peaks in 3 biological replicates were identified in 3’ UTRs (Fig 8A and 8B), where DAZL binds to facilitate post-transcriptional regulation [12,13,31–34]. Defining DAZL binding sites as those iCLIP peaks in 3’ UTRs present in at least 2 of 3 biological replicates, we identified a total of 1,652 DAZL binding sites corresponding to 1,281 genes (Fig 8B). This is far fewer binding sites and targets than identified by Zagore et al. and Li et al. [12,13]. These differences may be due to the tissue used, as these studies used whole testes which contain multiple spermatogenic stages that express DAZL [12,13]. The 3’ UTR binding sites identified here were strongly enriched for the GUU motif with which DAZL preferentially interacts (Fig 8C) [12,13,35]. Replicated peaks were also identified in other genomic regions (S5B Fig, S11 Table), but the number of replicated peaks in these genomic regions was small and/or these peaks showed only moderate enrichment for DAZL’s GUU motif relative to the 3’ UTR peaks (Fig 8B and 8C, S5B and S5C Fig), suggesting that peaks outside of 3’ UTRs originate from nonspecific DAZL:RNA interactions.
Based on the 3’ UTR binding sites, DAZL targets 6 TFIID component genes, including Taf4b but not Taf4a (Fig 8D–8J, S11 Table). The observation that all of these genes exhibit reduced expression in Dazl-KO spermatogonia may reflect that DAZL protects these transcripts from degradation. Alternatively, the reduced TFIID gene expression in the Dazl-KO may reflect indirect regulation by DAZL, which may increase the protein expression of other transcription factors that are required for the expression of TFIID component genes.
To determine whether STRA8 directly promotes the gene expression of TFIID components, we reprocessed the ChIP-seq data of Kojima et al.  on preleptotene synchronized whole testis expressing a FLAG-tagged STRA8 protein. We found peaks in two of the three replicates at many of the same genes with reduced expression in Stra8-KO RNA-seq experiments, including Taf4b, Taf7l, and Taf7l2. By contrast, STRA8 does not bind to Taf4a or Taf7, which are not differentially expressed in the RNA-seq analysis (Fig 9A–9E, S6A Fig, S12 Table). Interestingly, STRA8-FLAG ChIP signal reliably localized to the 5’ UTR regions of Taf4b and Taf7l2 (Fig 9A and 9E), but localized after the first exon of Taf7l (Fig 9D). In addition to ChIP-seq data, in in vitro differentiated primordial germ cell-like cells (PGCLCs), treatment of PGCLCs with retinoic acid and bone morphogenetic protein 2 (BMP2) induces Taf4b, but not Taf4a, expression on culture day 9 (S6B and S6C Fig) . However, this induction of Taf4b fails to occur if Stra8 is knocked down in the cell culture system, which is also the case for Taf7l and Taf7l2, but not Taf7 (S6D–S6F Fig). These combined lines of evidence including RNA-seq, ChIP-seq, and in vitro Stra8 knockdown demonstrate a direct link between Stra8 and several TFIID components in and around the time of meiotic initiation.
As TFIID and Spt-Ada-Gcn5-acetyltransferase (SAGA) are closely related complexes, even sharing TAF9, TAF10, and TAF12 as subunits, we investigated whether these trends in germ cells were unique to TFIID or also applied to SAGA [10,37]. In the mouse time course data, we found four SAGA subunits (including Taf12) to be germ cell-enriched (Fig 10A) by the criteria of higher mRNA expression in the germ cells over the somatic cell samples for at least 3 of the 7 measured time points in either sex. ImpulseDE2 identified eleven SAGA subunits as significantly dynamic in female and/or male germ cells (Fig 10B) and Atxn7, Tada2b, and Supt3 were found in Cluster D of our k-means clustering (Fig 10C, red box). When examining our data regarding Dazl, we found eleven subunits differentially expressed between WT and Dazl-KO samples in female and/or male mice, but only two of these genes (Usp22 and Trrap) were identified as DAZL-bound in our iCLIP data and neither were found to be differentially expressed (Fig 10D, S11 Table). Similarly, we found ten SAGA genes to be differentially expressed between WT and Stra8-KO samples in female and/or male mice, but in the ChIP-seq data, only three of these genes display STRA8-FLAG peaks in their promoter regions (Fig 10E) and three more (Taf5l, Taf9, Trrap) were not differentially expressed. Altogether, only expression of Supt3 is most similar to Taf4b, but its mRNA is not found to be DAZL-bound (S11 Table), which suggests that there are differences in how they are regulated during embryonic germ cell development.
The underlying transcriptional programs that drive germ cell development during mammalian embryogenesis are just coming into focus. In this study, we reprocessed publicly available RNA-seq datasets to understand the dynamics and potential regulators of the TFIID complex. We discovered that Taf4b is germ cell-enriched, significantly dynamic, clusters with meiotic genes, and is Dazl- and Stra8-regulated in female and male mice. Furthermore, our DAZL iCLIP data demonstrates that the 3’ UTRs of six TFIID components (Taf1, Taf2, Taf4b, Taf7, Taf9b, and Tbp) are directly bound by DAZL in leptotene spermatocytes. Moreover, reprocessing of STRA8-FLAG ChIP-seq data indicates that six TAFs (Taf3, Taf4b, Taf7l, Taf7l2, Taf9, and Taf11) have STRA8 bound to their promoters in preleptotene spermatocytes [13,14]. Of all these genes that were either DAZL- or STRA8-regulated, only Taf4b, Taf7l, and Taf9b were also germ cell-enriched, significantly dynamic in their expression, and clustered with meiotic genes. Taken together, these data indicate that several TAFs may participate in germ cell development, particularly during early meiosis, and potentially implicate a germ cell-specific TFIID complex that is preferentially stabilized to integrate proper germ cell-specific transcription and development.
Although basal transcription components such as TFIID are not widely studied in the context of reproduction, Sisakhtnezhad and Heshmati  found that Taf7l (with Dazl, Mael, etc.) was one of the 30 most significantly SSC-enriched genes in comparison to mesenchymal stem cells in PND7 mice. In addition, Taf1, Taf2, Taf4b, Taf7, Taf7l2, and Taf9b were also significantly enriched in SSCs along with other basal transcription machinery such as RNAPII subunits . Soh et al. found that Taf4b, Taf5, Taf7l, Taf7l2, and Taf9b were “ovarian germ cell enriched”, of which Taf7l, Taf7l2, and Taf9b are also highlighted as “meiotic prophase” genes . It is already known that Taf7l is essential for male fertility: Taf7l-KO male mice are infertile due to defects in spermiogenesis . Taf7l2 was only recently uncovered as a rodent-specific Taf7l retrogene  and it is not known if there is a fertility defect in a Taf7l2 mutant. Taf7l interacts with Trf2 (also known as Tbpl1) to bind to promoters of genes in mouse testis. Interestingly, Trf2/Tbpl1 was found to be significantly dynamic in female germ cells, but not germ cell-enriched nor significantly affected by Dazl or Stra8 in our data. It might be that Taf7l interacts with different proteins to perform a different role in the embryonic ovary or postnatal testis. E13.5 germ cells have been found to have much higher transcription levels than their somatic cell counterparts, and this has been proposed to be a phase of hypertranscription in the germ cells . However, this phase was found to end at E15.5, which contrasts with many of the TAFs we observed that increase in gene expression at E15.5.
Taf9b participating in the Cluster D in the k-means clustering was surprising out of the five TAFs found in that cluster. Taf9b is known to play a critical role in proper differentiation of motor neurons, but it has no known role in germline development, as Taf9b-KO mice are viable and fertile . Taf9b is not the only gene to be associated with cell type-specific differentiation. Taf7l has also been found to play a role in adipocyte formation [41,42]. When the germ cell samples were split into female- or male-only for individual analysis, Taf4b, Taf7l, Taf7l2, and Taf9b remained clustered together and with other meiotic genes in female germ cells. However, in male germ cells, these four TAFs did not cluster together, but Taf4b and Taf7l2 did, as well as Taf7l and Taf9b (S7 Table). Hill et al. also grouped Taf7l, Taf7l2, and Taf9b together as germline reprogramming responsive (GRR) genes, which is a group of 45 genes that includes Dazl, Hormad1, and thirteen other genes that were also found in Cluster D . These genes were designated GRR because their promoters were highly methylated at E10.5 but this DNA methylation was reduced during germline epigenetic programming, and the genes became progressively more transcribed in both sexes. This consistent association with essential gamete generation and meiotic genes in both our clustering analysis and other published research strongly implicates Taf4b, Taf7l, Taf7l2, and Taf9b as genes that warrant further study. Only TAF4b and TAF7l are being studied for their germ cell-specific function. It is unknown to what extent, if any, these other TAFs regulate the expression of each other. Future research should investigate if genes such as Taf7l are differentially expressed in the Taf4b-deficient mouse. Our finding that, among the SAGA complex genes, only Supt3 (also known as Spt3) is germ-cell enriched, significantly dynamic, in Cluster D, and STRA8-regulated suggests that this gene may function in germline-specific gene expression. Despite not being DAZL-regulated, it is possible that the SAGA complex could play a role in germ cell development in a similar way to Taf4b.
Our findings also reinforce the role of Taf4b in germ cell development, but now as part of a larger, coordinated effort with additional TFIID components. Given that previous research has shown that TAF4b functions during meiosis and localizes to the promoters of Dazl and Stra8 at E18.5, it is possible that TAF4b, DAZL, STRA8, and perhaps other factors work interdependently. Identifying DAZL and STRA8 as regulators of Taf4b is an important step forward in understanding Taf4b, but nevertheless, there are likely more regulators yet to be found. This is hinted at in our data, where male Taf4b expression rises at E15.5 (Fig 2A and 2B) in the absence of STRA8. It is plausible that another transcription factor facilitates this increase. As STRA8 is considered a transcriptional amplifier rather than an activator, it is likely that this unknown transcription factor(s) is shared between female and male germ cells. Future work in embryonic germ cells, particularly in males, should add more context to the role TAF4b may be playing there.
The specialized role for TAF4b as a germ cell relevant TFIID component is bolstered by single-cell RNA-seq analyses in mice and humans that associate Taf4b with essential developmental steps in Id4-GFP adult mouse germ cells and adult human spermatogonia . Alternative Taf4 subunits have been identified in both Drosophila melanogaster (known as nht) and Arabidopsis thaliana (known as taf4b) [45,46]. The alternative Taf4 genes in vertebrates, Drosophila, and Arabidopsis arose independently . Interestingly, Drosophila nht is testis-specific and required for male fertility, and Arabidopsis Taf4b is directly implicated in regulating meiotic crossover rates and germline transcription, similar to mouse Taf4b . This mechanism of an alternative Taf4 in TFIID, particularly for the purpose of germ cell development and function, is an intriguing instance of convergent evolution across plants, invertebrates, and vertebrates.
TFIID has been typically viewed as static and generic in composition as a basal transcription factor. These dynamic gene expression data, particularly for Taf4b, challenge this view and suggest that there are subunits in TFIID that are preferred and regulated over others in mammalian germ cells. This may also be the case for other cell type-specific differentiation programs, as some of the same variant TFIID subunits shown to be expressed in a germ cell-specific manner here have also been shown to regulate somatic cells, including neurons and adipocytes. Though initially unexpected, such research indicates that there is more information to be garnered from studying TFIID in the context of germ cells and other specialized cell types.
Materials & methods
This study was approved by Brown IACUC protocol #1803000344. For the iCLIP experiment, the MIT IACUC approved this research (#067-059-20). The primary method of euthanasia is CO2 inhalation and the secondary method used is cervical dislocation both as per AVMA guidelines on euthanasia.
Mice that were homozygous for an Oct4-EGFP transgene (The Jackson Laboratory: B6;129S4-Pou5f1tm2Jae/J) were mated for mRNA and protein collections. Timed matings were estimated to begin at day 0.5 by evidence of a copulatory plug. To confirm the sex of E12.5 and younger embryos, genotyping for the presence or absence of the Sry gene was completed. Male and female embryos were identified at E13.5 and older by confirming the presence or absence of testicular cords. Genomic DNA from tails was isolated using ThermoFisher PureLink Genomic DNA isolation kit (Cat #: K182001) for PCR genotyping assays. All animal protocols were reviewed and approved by Brown University Institutional Animal Care and Use Committee and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Gonads were dissected out of embryos into cold PBS. Wildtype C57BL/6N mice (Taconic Biosciences) were mated to produce postnatal males for iCLIP collection. All experiments involving mice were performed in accordance with the guidelines of the Massachusetts Institute of Technology (MIT) Division of Comparative Medicine, which is overseen by MIT’s Institutional Animal Care and Use Committee (IACUC). The animal care program at MIT/Whitehead Institute is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International (AAALAC), and meets or exceeds the standards of AAALAC as detailed in the Guide for the Care and Use of Laboratory Animals. The MIT IACUC approved this research (no. 0617-059-20).
Embryonic gonad dissociation and fluorescence-activated cell sorting
To dissociate gonadal tissue into a single-cell suspension, embryonic gonads at specified times were harvested and placed in 0.25% Trypsin/EDTA and incubated at 37°C for 15–20 minutes. Eppendorf tubes were flicked to dissociate tissue halfway through and again at the end of the incubation. Trypsin was neutralized with 75 μL of FBS. Cells were pelleted, the supernatant was removed, and cells were resuspended in 100 μL PBS. The cell suspension was strained through a 35 μm mesh cap into a FACS tube (Gibco REF# 352235). Propidium iodide (1:500) was added to the cell suspension as a live/dead distinguishing stain. Fluorescence-activated cell sorting (FACS) was performed using a Becton Dickinson FACSAria III in the Flow Cytometry and Cell Sorting Core Facility at Brown. A negative control of a non-transgenic mouse gonad was used for each experiment to establish an appropriate GFP signal baseline. Dead cells were discarded based on propidium iodide signal and then cells were sorted at 4°C in PBS based on GFP signal into GFP+ or GFP- samples. Cells for western blots were pelleted, PBS was removed, flash-frozen with liquid nitrogen, and stored at -80°C.
Protein to be loaded for western blots was estimated based on cell numbers reported during FACS. Thawed cells were mixed with laemmli sample buffer and β-mercaptoethanol, pipetted thoroughly, incubated on ice for 1 hour, and stored at -80°C. Western blotting was performed by loading roughly similar amounts of protein (estimated based on cell number) into a BioRad Mini-PROTEAN TGX 4–15% gel (Cat. #456–1083). Protein was transferred to nitrocellulose membrane (0.45 μm). Nitrocellulose was cut, blocked in 5% milk in PBS with 0.1% Tween (PBS-T) or blocked in 5% Amersham ECL Prime Blocking Reagent (RPN418) for 30 minutes at room temperature. Incubation in primary antibody in 1% milk in PBS-T or only PBS-T was performed in 4°C overnight and secondary antibody in 1% milk in PBS-T or only PBS-T for 1 hour at room temperature. In between the block/primary and primary/secondary incubations, nitrocellulose was washed in PBS-T at least three times for 5 minutes. β-Actin or GAPDH was used as a loading control. GE Healthcare Amersham ECL Prime Western Blotting Detection Reagent (RPN2232) was used for horseradish peroxidase-tagged (HRP) secondary antibody detection. For stripping antibodies, Restore Western Blot Stripping Buffer (ThermoFisher: 21059) was used for 10 minutes at room temperature. For relative protein quantification, ImageJ was used to normalize protein signal to β-Actin in each cell type.
Antibodies used were as follows: mouse monoclonal anti-TAFII135 (BD Transduction Laboratories: 612054), polyclonal rabbit anti-mouse TAF4B (as described previously ), mouse monoclonal TAF7 (Novus Biologicals: H00006879-M01), rabbit polyclonal Taf7l (a generous gift from Dr. Haiying Zhou and Dr. Robert Tjian, rabbit polyclonal MVH (Abcam: ab13840); mouse monoclonal β-Actin (Ambion: AM4302).
RNA-seq data analysis
All computational scripts regarding RNA-seq used in this publication are available to the public: https://github.com/mg859337/TAF_Manuscript_MG/tree/master/RNAseq. Datasets SRP059601, SRP059599, SRP057098, SRP045294, SRP049981, SRP128645, and SRP150721 were from NCBI SRA, and dataset E-MTAB-4616 was obtained from ArrayExpress. Kojima et al. performed RNA-seq on four Stra8-WT samples, two labeled “high-Stra8” and two “low-Stra8”, but only the “high-Stra8” were used as WT in these data because the “low-Stra8” were in early preleptotene and not representative of full meiotic initiation in male germ cells. All accessed raw fastq files were initially processed on Brown University’s high-performance computing cluster at the Center for Computation and Visualization. They were analyzed using FastQC (v0.11.4 or 0.11.5) for quality and then aligned to the mm10 or hg38 genomes using HiSat2 (v2.1.0) with the optional -dta setting enabled (S12 Table) [48,49]. Resulting sam files were converted to bam files using Samtools (v1.9) .
To obtain transcripts per million (TPMs) for each sample, StringTie (v1.3.3b) was used with the optional parameters -A and -e . A gtf file for each sample was downloaded and, using RStudio (R v3.5.1), TPMs of all samples were aggregated into one comma separated (csv) file using a custom R script. To create interactive Microsoft Excel files for exploring the TPMs of each dataset: the csv of aggregated TPMs was saved as an Excel spreadsheet, colored tabs were added to set up different comparisons, and a flexible Excel function was created to adjust to new gene name inputs. To explore the Excel files, please find the appropriate tab and type in the gene name of interest into the highlighted yellow boxes. There is an Excel file for each dataset analyzed in the supplementary tables.
To obtain count tables, HTSeq (v0.9.1) and the count tables for each dataset were merged using a custom RStudio script . Metadata files for dataset were created manually in Excel and saved as a csv. These count tables were used to create PCA plots by variance-stabilizing transformation (vst) of the data in DESeq2 (v1.22.2) and plotting by ggplot2 (v3.1.0) [52,53]. DESeq2 was also used for differential gene expression analysis, where count tables and metadata files were used as input.
ImpulseDE2 (v1.6.1), a Bioconductor R package that can identify differentially expressed genes in longitudinal count datasets, was used to identify “significantly dynamic” genes from E9.5 –E18.5 female and male Oct4-EGFP+ germ versus Oct4-EGFP- somatic cell time course datasets . ImpulseDE2 was used in case-only mode and no batch effects settings were added to the parameters.
For k-means clustering, iDEP (v.9) was used. It is a webpage hosted by South Dakota State University that is a simple and fast resource to perform typical RNA-seq data analysis . The Sangrithi et al. (2018) mouse time course merged counts table was uploaded to iDEP. Counts were filtered out by the criteria of at least 0.5 counts per million in one of the samples. The data was then transformed using vst, in the “Heatmap” module the blue-white-red color scheme was chosen and in the “k-Means” clustering module the top 10,000 most variable genes across the time course were selected for clustering. The setting of 9 clusters was chosen using the within-sum-of-squares (wss) elbow method displayed under the “How many clusters?” button. The heatmap of gene clustering and associated list of clustered genes were downloaded. Gene ontology is another part of the “k-Means” module, the GO category of Biological Process was chosen, and the enrichment details were downloaded.
iCLIP library construction, sequencing, and computational analysis
To obtain testes enriched for the leptotene stage of meiotic prophase I, spermatogenesis was chemically synchronized via a protocol originally developed by Hogarth et al. and modified by Romer et al. [55,56]. Briefly, on PND2-8, male mice were injected daily subcutaneously with WIN 18,446 (Santa Cruz Biotechnology) at 0.1 mg/gram body weight. On PND9, the mice were injected once subcutaneously with retinoic acid (RA; MilliporeSigma) at 0.0125 mg/gram body weight. Mice were euthanized 8.0 days after the RA injection to obtain testes enriched for leptotene spermatocytes. From each pup, one half of a testis was collected as a biopsy for histological analysis to verify enrichment of leptotene spermatocytes, and the remaining 1.5 testes were collected for iCLIP. For histological analysis, the testis biopsy was fixed in Bouin’s solution for 3h at room temperature. Testis sections were stained with anti-STRA8 (rabbit polyclonal; Abcam ab49405) and then counterstained with hematoxylin, as previously described . At least 40 seminiferous tubule cross-sections per mouse were analyzed for morphology  and STRA8 expression, found in preleptotenes and early leptotenes [57,58]. For all samples analyzed via iCLIP, 100% of the tubule cross-sections contained leptotene spermatocytes but not preleptotene or zygotene spermatocytes.
After synchronization of spermatogenesis, testis tubules were dissociated by pipetting in ice-cold PBS and irradiated three times at 200 mJ/cm2 at 254 nm in a Stratalinker 2400 before pelleted, flash frozen, and stored at -80°C until library preparation. DAZL and IgG iCLIP libraries were prepared as previously described . For each biological replicate, 1.5 testes from a single mouse were lysed in 640 μl lysis buffer. Immunoprecipitations were carried out via anti-DAZL (rabbit polyclonal; Abcam ab34139; validated for CLIP in studies [12,13]) or polyclonal rabbit IgG (Abcam ab27478) using 12 μg antibody bound to 120 μl Dynabeads Protein G (ThermoFisher Scientific 10003D) with 600 μl lysate. iCLIP libraries were prepared in two batches, with each batch containing an equal number of DAZL iCLIP and IgG iCLIP libraries. A total of 3 DAZL iCLIP and 3 IgG iCLIP libraries were generated. The libraries were pooled and sequenced with 50bp single-end reads on the Illumina HiSeq 2500 machine. The 5’ end of each iCLIP read contained (in 5’ to 3’ orientation), a 3-nt random barcode, a 4-nt sample-specific barcode, and a 2-nt random barcode.
All computational scripts used in this publication for the iCLIP analysis are available at: https://github.com/mmikedis/Gura_et_al_2019_PLOS_Genetics. Reads were quality-trimmed with cutadapt v1.8 (options: -q 20 -m 24) . PCR duplicates were collapsed using the FASTX-Toolkit v0.0.14 (http://hannonlab.cshl.edu/fastx_toolkit/index.html), and the 5'-most 3-nt random barcode was trimmed from the 5' end of each read with the fastx_trimmer tool from FASTX-Toolkit v0.0.14 (options: -f 4). Reads were demultiplexed using the fastx_barcode_splitter.pl tool from FASTX-Toolkit v0.0.14 (options:—bol), and the sample-specific barcodes with remaining random barcodes were trimmed from the 5’ end of each read with the fastx_trimmer tool (options: -f 7). iCLIP libraries were mapped to the mouse genome (mm10 assembly) via STAR v2.5.4b  (options:—outFilterMultimapNmax 1—alignEndsType Extend5pOfRead1—outFilterMismatchNmax 2—outSAMattributes None—outReadsUnmapped Fastx). All other parameters were set to default. The iCLIP mapped reads were then converted to crosslinked nucleotides (i.e., the nucleotide immediately preceding the first nucleotide of the mapped read) using a custom script . Within each library batch, DAZL crosslinked peaks were called using the DAZL iCLIP crosslinked nucleotides via ASPeak v2.0.0  with the IgG iCLIP crosslinked nucleotides for the -control parameter. All other parameters were set to default. The UCSC RefSeq transcript annotations and Retrogenes V6 annotations for the mm10 assembly were used to call peaks. Given that DAZL binds the 3’ UTR of coding transcripts, peaks were called via the following hierarchy: 3' UTR exon > 5' UTR exon > coding exon > ncRNA > retrogene > intron > intergenic region. Peaks were filtered for p < 0.001, and then replicated peaks present in at least 2 out of 3 biological replicates were identified. Replicated peaks identified in consecutive nucleotides were merged into a single peak. iCLIP sequencing data and analysis have been deposited in NCBI GEO under accession number GSE139005 and NCBI SRA under accession number SRP226023.
Enrichment of the GUU motif at replicated peaks from each genomic region was assessed via the MEME Suite’s AME v4.11.2  (options:—scoring avg—method ranksum) using the replicated crosslinked nucleotides ±2 nt and shuffled control sequences.
These trimmed reads were then aligned to the mm10 genome using Bowtie2 (v 2.3.0) (S12 Table) . Resulting sam files were converted to bam files, then unmapped and duplicated reads were removed using Samtools (v1.9) . Bedtools (v2.26.0) was used to removed regions in bam files that mapped to ChIP-seq blacklisted regions in the mm10 genome . For visualization, Samtools was used to merge replicates and Integrative Genomics Viewer (IGV) was used to visualize the gene tracks . For calling peaks MACS (v 2.1.1) was used, comparing the FLAG replicate to its input control .
S1 Fig [a]
Replicates of western blots and protein signal of TAF4b in cells sorted from female E13.5 Oct4-EGFP gonads.
S2 Fig [a]
Independent RNA-seq datasets replicate increase in mRNA expression after E13.5 in both female and male mouse germ cells and low expression of .
S3 Fig [a]
Germ cell-enriched TFIID subunits and closer examination of , , and .
S4 Fig [pdf]
Within sum of squares (wss) graph for evaluating how many clusters to set in k-means clustering.
S5 Fig [a]
DAZL iCLIP in testes synchronized for leptotene spermatocytes.
S6 Fig [a]
STRA8 peaks in preleptotene germ cells and knockdown experiments in primordial germ cell-like cells (PGCLCs).
S1 Table [xlsx]
DESeq2 output for comparisons between germ and somatic cells at different time points.
S2 Table [xlsx]
Transcripts per million (TPM) of all reprocessed time course data from Sangrithi et al. with tabs for asking specific questions about the data.
S3 Table [xlsx]
Transcripts per million (TPM) of all reprocessed time course data from Gkountela et al. with tabs for asking specific questions about the data.
S4 Table [xlsx]
Transcripts per million (TPM) of all reprocessed time course data from Tang et al. with tabs for asking specific questions about the data.
S5 Table [xlsx]
Transcripts per million (TPM) of all reprocessed time course data from Irie et al. with tabs for asking specific questions about the data.
S8 Table [xlsx]
Transcripts per million (TPM) of all reprocessed RNA-seq data from Soh et al. E14.5 females with tabs for asking specific questions about the data.
S9 Table [xlsx]
Transcripts per million (TPM) of all reprocessed RNA-seq data from Zagore et al. PND6 spermatogonia with tabs for asking specific questions about the data.
S10 Table [xlsx]
Transcripts per million (TPM) of all reprocessed RNA-seq data from Kojima et al. male preleptotene germ cells with tabs for asking specific questions about the data.
S11 Table [xlsx]
Information regarding iCLIP libraries and replicated peaks identified in DAZL iCLIP.
S12 Table [xlsx]
Narrowpeaks called from all reprocessed ChIP-seq data from Kojima et al. male preleptotene-synchronized testis germ cells.
S13 Table [xlsx]
Information regarding the sequencing samples reprocessed in this manuscript.
1. Chandra A, Copen CE, Stephen EH. Infertility and impaired fecundity in the United States, 1982–2010: data from the National Survey of Family Growth. Natl Health Stat Report. 2013; doi: 10.1093/humrep/des207
2. Nelson LM. Clinical practice. Primary ovarian insufficiency. N Engl J Med. 2009;360: 606–14. doi: 10.1056/NEJMcp0808697 19196677
3. Rossetti R, Ferrari I, Bonomi M, Persani L. Genetics of primary ovarian insufficiency. Clinical Genetics. 2017. pp. 183–198. doi: 10.1111/cge.12921 27861765
4. Grive KJ, Seymour K a., Mehta R, Freiman RN. TAF4b promotes mouse primordial follicle assembly and oocyte survival. Dev Biol. Elsevier; 2014;392: 42–51. doi: 10.1016/j.ydbio.2014.05.001 24836512
5. Freiman RN, Albright SR, Zheng S, Sha WC, Hammer RE, Tjian R. Requirement of tissue-selective TBP-associated factor TAFII105 in ovarian development. Science. 2001;293: 2084–2087. doi: 10.1126/science.1061935 11557891
6. Lovasco LA, Gustafson EA, Seymour KA, De Rooij DG, Freiman RN. TAF4b is required for mouse spermatogonial stem cell development. Stem Cells. 2015;33: 1267–1276. doi: 10.1002/stem.1914 25727968
7. Knauff EAH, Franke L, van Es MA, van den Berg LH, van der Schouw YT, Laven JSE, et al. Genome-wide association study in premature ovarian failure patients suggests ADAMTS19 as a possible candidate gene. Hum Reprod. 2009;24: 2372–8. doi: 10.1093/humrep/dep197 19508998
8. Di Pietro C, Vento M, Ragusa M, Barbagallo D, Guglielmino MR, Maniscalchi T, et al. Expression analysis of TFIID in single human oocytes: new potential molecular markers of oocyte quality. Reprod Biomed Online. 2008;17: 338–49. doi: 10.1016/s1472-6483(10)60217-9 18765004
9. Ayhan Ö, Balkan M, Guven A, Hazan R, Atar M, Tok A, et al. Truncating mutations in TAF4B and ZMYND15 causing recessive azoospermia. J Med Genet. 2014; doi: 10.1136/jmedgenet-2013-102102 24431330
10. Antonova S V., Boeren J, Timmers HTM, Snel B. Epigenetics and transcription regulation during eukaryotic diversification: the saga of TFIID. Genes Dev. 2019;33: 1–15. doi: 10.1101/gad.322990.118
11. Grive KJ, Gustafson EA, Seymour KA, Baddoo M, Schorl C, Golnoski K, et al. TAF4b Regulates Oocyte-Specific Genes Essential for Meiosis. PLoS Genet. 2016;12: 1–18. doi: 10.1371/journal.pgen.1006128 27341508
12. Li H, Liang Z, Yang J, Wang D, Wang H, Zhu M, et al. DAZL is a master translational regulator of murine spermatogenesis. Natl Sci Rev. 2019; doi: 10.1093/nsr/nwy163 31355046
13. Zagore LL, Sweet TJ, Hannigan MM, Weyn-Vanhentenryck SM, Jobava R, Hatzoglou M, et al. DAZL Regulates Germ Cell Survival through a Network of PolyA-Proximal mRNA Interactions. Cell Rep. 2018; doi: 10.1016/j.celrep.2018.10.012 30380414
14. Kojima ML, de Rooij DG, Page DC. Amplification of a broad transcriptional program by a common factor triggers the meiotic cell cycle in mice. Elife. 2019; doi: 10.7554/elife.43738 30810530
15. Falender AE, Freiman RN, Geles KG, Lo KC, Hwang KS, Lamb DJ, et al. Maintenance of spermatogenesis requires TAF4b, a gonad-specific subunit of TFIID. Genes Dev. 2005; doi: 10.1101/gad.1290105 15774719
17. Cheng Y, Buffone MG, Kouadio M, Goodheart M, Page DC, Gerton GL, et al. Abnormal Sperm in Mice Lacking the Taf7l Gene. Mol Cell Biol. 2007; doi: 10.1128/mcb.01722-06 17242199
18. Zhou H, Grubisic I, Zheng K, He Y, Wang PJ, Kaplan T, et al. Taf7l cooperates with Trf2 to regulate spermiogenesis. Proc Natl Acad Sci. 2013; doi: 10.1073/pnas.1317034110 24082143
19. Sangrithi MN, Royo H, Mahadevaiah SK, Ojarikre O, Bhaw L, Sesay A, et al. Non-Canonical and Sexually Dimorphic X Dosage Compensation States in the Mouse and Human Germline. Dev Cell. 2017; doi: 10.1016/j.devcel.2016.12.023 28132849
20. Yoshimizu T, Sugiyama N, Felice M De, Ii Y, Ohbo K, Masuko K, et al. Germline-specific expression of the Oct-4/green fluorescent protein (GFP) transgene in mice. Dev Growth, Differ. 1999;41: 675–684. Available: https://doi.org/10.1046/j.1440-169x.1999.00474.x
21. Gkountela S, Zhang KX, Shafiq TA, Liao WW, Hargan-Calvopiña J, Chen PY, et al. DNA demethylation dynamics in the human prenatal germline. Cell. 2015;161: 1425–1436. doi: 10.1016/j.cell.2015.05.012 26004067
22. Irie N, Weinberger L, Tang WWC, Kobayashi T, Viukov S, Manor YS, et al. SOX17 is a critical specifier of human primordial germ cell fate. Cell. 2015; doi: 10.1016/j.cell.2014.12.013 25543152
23. Tang WWC, Dietmann S, Irie N, Leitch HG, Floros VI, Bradshaw CR, et al. A Unique Gene Regulatory Network Resets the Human Germline Epigenome for Development. Cell. 2015; doi: 10.1016/j.cell.2015.04.053 26046444
24. Fischer DS, Theis FJ, Yosef N. Impulse model-based differential expression analysis of time course sequencing data. Nucleic Acids Res. 2018; doi: 10.1093/nar/gky675 30102402
25. Soh YQS, Junker JP, Gill ME, Mueller JL, van Oudenaarden A, Page DC. A Gene Regulatory Program for Meiotic Prophase in the Fetal Ovary. PLoS Genet. 2015;11. doi: 10.1371/journal.pgen.1005531 26378784
26. De Gasperi R, Rocher AB, Sosa MAG, Wearne SL, Perez GM, Friedrich VL, et al. The IRG mouse: A two-color fluorescent reporter for assessing Cre-mediated recombination and imaging complex cellular relationships in situ. Genesis. 2008; doi: 10.1002/dvg.20400 18543298
27. Huppertz I, Attig J, D’Ambrogio A, Easton LE, Sibley CR, Sugimoto Y, et al. iCLIP: Protein-RNA interactions at nucleotide resolution. Methods. 2014;65: 274–287. doi: 10.1016/j.ymeth.2013.10.011 24184352
28. van Pelt AM, de Rooij DG. Synchronization of the seminiferous epithelium after vitamin A replacement in vitamin A-deficient mice. Biol Reprod. 1990;43: 363–67. doi: 10.1095/biolreprod43.3.363 2271719
29. Endo T, Romer KA, Anderson EL, Baltus AE, de Rooij DG, Page DC. Periodic retinoic acid-STRA8 signaling intersects with periodic germ-cell competencies to regulate spermatogenesis. Proc Natl Acad Sci U S A. 2015;112: E2347–56. doi: 10.1073/pnas.1505683112 25902548
30. Russell LD, Ettlin RA, Hikim APS, Clegg ED. Histological and Histopathological Evaluation of the Testis. Int J Androl. 1993; doi: 10.1111/j.1365-2605.1993.tb01156.x
31. Rosario R, Smith RWP, Adams IR, Anderson RA. RNA immunoprecipitation identifies novel targets of DAZL in human foetal ovary. Mol Hum Reprod. 2017; doi: 10.1093/molehr/gax004 28364521
32. Reynolds N, Collier B, Maratou K, Bingham V, Speed RM, Taggart M, et al. Dazl binds in vivo to specific transcripts and can regulate the pre-meiotic translation of Mvh in germ cells. Hum Mol Genet. 2005; doi: 10.1093/hmg/ddi414 16278232
33. Reynolds N, Collier B, Bingham V, Gray NK, Cooke HJ. Translation of the synaptonemal complex component Sycp3 is enhanced in vivo by the germ cell specific regulator Dazl. RNA. 2007; doi: 10.1261/rna.465507 17526644
34. Martins JPS, Liu X, Oke A, Arora R, Franciosi F, Viville S, et al. DAZL and CPEB1 regulate mRNA translation synergistically during oocyte maturation. J Cell Sci. 2016; doi: 10.1242/jcs.179218 26826184
35. Jenkins HT, Malkova B, Edwards TA. Kinked β-strands mediate high-affinity recognition of mRNA targets by the germ-cell regulator DAZL. Proc Natl Acad Sci U S A. 2011; doi: 10.1073/pnas.1105211108 22021443
36. Miyauchi H, Ohta H, Nagaoka S, Nakaki F, Sasaki K, Hayashi K, et al. Bone morphogenetic protein and retinoic acid synergistically specify female germ‐cell fate in mice. EMBO J. 2017; doi: 10.15252/embj.201796875 28928204
37. Koutelou E, Hirsch CL, Dent SYR. Multiple faces of the SAGA complex. Current Opinion in Cell Biology. 2010. doi: 10.1016/j.ceb.2010.03.005 20363118
38. Sisakhtnezhad S, Heshmati P. Comparative analysis of single-cell RNA sequencing data from mouse spermatogonial and mesenchymal stem cells to identify differentially expressed genes and transcriptional regulators of germline cells. J Cell Physiol. 2018; doi: 10.1002/jcp.26303 29194616
39. Percharde M, Wong P, Ramalho-Santos M. Global Hypertranscription in the Mouse Embryonic Germline. Cell Rep. 2017; doi: 10.1016/j.celrep.2017.05.036 28591571
41. Zhou H, Kaplan T, Li Y, Grubisic I, Zhang Z, Wang PJ, et al. Dual functions of TAF7L in adipocyte differentiation. Elife. 2013; doi: 10.7554/eLife.00170 23326641
42. Zhou H, Wan B, Grubisic I, Kaplan T, Tjian R. TAF7L modulates brown adipose tissue formation. Elife. 2014; doi: 10.7554/eLife.02811 24876128
43. Hill PWS, Leitch HG, Requena CE, Sun Z, Amouroux R, Roman-Trufero M, et al. Epigenetic reprogramming enables the transition from primordial germ cell to gonocyte. Nature. 2018; doi: 10.1038/nature25964 29513657
44. Hermann BP, Cheng K, Singh A, Roa-De La Cruz L, Mutoji KN, Chen IC, et al. The Mammalian Spermatogenesis Single-Cell Transcriptome, from Spermatogonial Stem Cells to Spermatids. Cell Rep. 2018; doi: 10.1016/j.celrep.2018.10.026 30404016
45. Hiller M. Testis-specific TAF homologs collaborate to control a tissue-specific transcription program. Development. 2004; doi: 10.1242/dev.01314 15456720
46. Lago C, Clerici E, Mizzi L, Colombo L, Kater MM. TBP-associated factors in Arabidopsis. Gene. 2004; doi: 10.1016/j.gene.2004.08.023 15527982
47. Lawrence EJ, Gao H, Tock AJ, Lambing C, Blackwell AR, Feng X, et al. Natural Variation in TBP-ASSOCIATED FACTOR 4b Controls Meiotic Crossover and Germline Transcription in Arabidopsis. Curr Biol. 2019; doi: 10.1016/j.cub.2019.06.084 31378616
48. Andrews S. FastQC: A quality control tool for high throughput sequence data. In: Http://Www.Bioinformatics.Babraham.Ac.Uk/Projects/Fastqc/. 2010 p. http://www.bioinformatics.babraham.ac.uk/projects/. citeulike-article-id:11583827
49. Pertea M, Kim D, Pertea GM, Leek JT, Salzberg SL. Transcript-level expression analysis of RNA-seq experiments with HISAT, StringTie and Ballgown. Nat Protoc. 2016; doi: 10.1038/nprot.2016.095 27560171
50. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25: 2078–2079. doi: 10.1093/bioinformatics/btp352 19505943
51. Anders S, Pyl PT, Huber W. HTSeq-A Python framework to work with high-throughput sequencing data. Bioinformatics. 2015; doi: 10.1093/bioinformatics/btu638 25260700
52. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15: 550. doi: 10.1186/s13059-014-0550-8 25516281
53. Wickham H. ggplot2: Elegant Graphics for Data Analysis [Internet]. 2nd ed. Wiley Interdisciplinary Reviews: Computational Statistics. Springer-Verlag; 2016. Available: ggplot.org doi: 10.1002/wics.1383 29657666
54. Ge SX, Son EW, Yao R. iDEP: An integrated web application for differential expression and pathway analysis of RNA-Seq data. BMC Bioinformatics. 2018; doi: 10.1186/s12859-018-2486-6 30567491
55. Hogarth CA, Evanoff R, Mitchell D, Kent T, Small C, Amory JK, et al. Turning a Spermatogenic Wave into a Tsunami: Synchronizing Murine Spermatogenesis Using WIN 18,4461. Biol Reprod. 2013; doi: 10.1095/biolreprod.112.105346 23284139
56. Romer KA, de Rooij DG, Kojima ML, Page DC. Isolating mitotic and meiotic germ cells from male mice by developmental synchronization, staging, and sorting. Dev Biol. 2018; doi: 10.1016/j.ydbio.2018.08.009 30149006
57. Oulad-Abdelghani M, Bouillet P, Décimo D, Gansmuller A, Heyberger S, Dollé P, et al. Characterization of a premeiotic germ cell-specific cytoplasmic protein encoded by Stra8, a novel retinoic acid-responsive gene. J Cell Biol. 1996;135: 469–477. doi: 10.1083/jcb.135.2.469 8896602
58. Zhou Q, Nie R, Li Y, Friel P, Mitchell D, Hess RA, et al. Expression of Stimulated by Retinoic Acid Gene 8 (Stra8) in Spermatogenic Cells Induced by Retinoic Acid: An In Vivo Study in Vitamin A-Sufficient Postnatal Murine Testes1. Biol Reprod. 2008; doi: 10.1095/biolreprod.107.066795 18322276
59. Martin M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal. 2011; doi: 10.14806/ej.17.1.200
60. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29: 15–21. doi: 10.1093/bioinformatics/bts635 23104886
61. König J, Zarnack K, Rot G, Curk T, Kayikci M, Zupan B, et al. iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution. Nat Struct Mol Biol. 2010;17: 909–915. doi: 10.1038/nsmb.1838 20601959
62. Kucukural A, Ozadam H, Singh G, Moore MJ, Cenik C. ASPeak: an abundance sensitive peak detection algorithm for RIP-Seq. Bioinformatics. 2013;29: 2485–2486. doi: 10.1093/bioinformatics/btt428 23929032
63. McLeay RC, Bailey TL. Motif Enrichment Analysis: a unified framework and an evaluation on ChIP data. BMC Bioinformatics. 2010;11: 165. doi: 10.1186/1471-2105-11-165 20356413
64. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods. 2012;9: 357–359. doi: 10.1038/nmeth.1923 22388286
65. Quinlan AR, Hall IM. BEDTools: A flexible suite of utilities for comparing genomic features. Bioinformatics. 2010; doi: 10.1093/bioinformatics/btq033 20110278