PRC1 directs PRC2-H3K27me3 deposition to shield adult spermatogonial stem cells from differentiation

Abstract Spermatogonial stem cells functionality reside in the slow-cycling and heterogeneous undifferentiated spermatogonia cell population. This pool of cells supports lifelong fertility in adult males by balancing self-renewal and differentiation to produce haploid gametes. However, the molecular mechanisms underpinning long-term stemness of undifferentiated spermatogonia during adulthood remain unclear. Here, we discover that an epigenetic regulator, Polycomb repressive complex 1 (PRC1), shields adult undifferentiated spermatogonia from differentiation, maintains slow cycling, and directs commitment to differentiation during steady-state spermatogenesis in adults. We show that PRC2-mediated H3K27me3 is an epigenetic hallmark of adult undifferentiated spermatogonia. Indeed, spermatogonial differentiation is accompanied by a global loss of H3K27me3. Disruption of PRC1 impairs global H3K27me3 deposition, leading to precocious spermatogonial differentiation. Therefore, PRC1 directs PRC2-H3K27me3 deposition to maintain the self-renewing state of undifferentiated spermatogonia. Importantly, in contrast to its role in other tissue stem cells, PRC1 negatively regulates the cell cycle to maintain slow cycling of undifferentiated spermatogonia. Our findings have implications for how epigenetic regulators can be tuned to regulate the stem cell potential, cell cycle and differentiation to ensure lifelong fertility in adult males.


Introduction
Mammalian males sustain lifelong fertility relying on balanced self-renewal and differentiation of spermatogonial stem cells (SSCs), which are destined to produce a massive number of haploid spermatozoa; ∼1000 sperm per second for decades in average human males ( 1 ).In adult mice, stem cell activity resides in a heterogenous population of slow-cycling undifferentiated type A spermatogonia (A undiff , including A single , A paired and A aligned spermatogonia), whose cell states are maintained at a delicate equilibrium ( 2 ,3 ).A fraction of A undiff that are positive for the cell surface receptor GFR α1 comprises the vast self-renewing compartment in homeostatic, steady-state spermatogenesis.While A undiff expressing NGN3 (Neurog3), which are GFR α1 -, are destined for differentiation but retain the capacity to revert to the GFR α1 + state and repopulate the self-renewing pool (Figure 1 A) ( 4 ,5 ).In response to retinoic acid (RA) signaling, the NGN3 + fraction of A undiff undergo irreversible commitment to become fast-cycling KIT + differentiating spermatogonia (designated A 1 , A 2 , A 3 , A 4 , Intermediate (In) and B spermatogonia), which are devoid of stem cell potential (Figure 1 A) ( 6 ) and undergo a series of mitotic divisions before entering meiosis ( 7 ,8 ).Therefore, the transition from A undiff (KIT -) to KIT + differentiating spermatogonia is accompanied by the irreversible loss of self-renewing stem cell potential and the acceleration of their cell cycle.
However, while numerous factors have been linked to the formation and maintenance of undifferentiated spermatogonia, major questions remain regarding the molecular underpinnings that shield and maintain robust stemness in these cells.In particular, it is unclear how slow cycling is regulated while lifelong differentiation potential is maintained in adult undifferentiated spermatogonia.This is primarily due to the fact that most previous studies have focused on juvenile spermatogenesis before the establishment of steady-state adult spermatogenesis ( 37 ,38 ) or used cultured germline stem cells that do not fully represent in vivo features of spermatogonial stem cells ( 39 ,40 ).
Epigenetic mechanisms that generate long-lasting chromatin states and define gene expression programs are critical for the development and regulation of stem cells ( 41 ).Polycomb group proteins (PcGs) mediate such mechanisms by establishing repressive chromatin and defining cell-type specific gene expression programs ( 42 ,43 ).Two major functionallyrelated Polycomb repressive complexes-PRC1 and PRC2catalyze the formation of two major repressive histone marks, monoubiquitination of H2A at lysine 119 (H2AK119ub) and trimethylation of H3 at lysine 27 (H3K27me3) ( 44 ), respectively.Analyses of PcG-mediated modifications have provided a picture of the epigenomic landscape of male germ cells, including undifferentiated spermatogonia ( 37 , 38 , 45-47 ).Of note, undifferentiated spermatogonia contain bivalent chromatin; the repressive H3K27me3 modification cooccurs with an active mark, trimethylation of H3 at lysine 4 (H3K4me3) ( 37 ,45 ), a feature associated with developmental potential in the germline.Nevertheless, the functions of these marks remain largely speculative due to the lack of quantitative analysis combined with genetic approaches.
In this study, by quantitative epigenomic analysis, we define H3K27me3 as a major barrier to spermatogonial differentiation, arguing that this is a hallmark of the undifferentiated state in adult undifferentiated spermatogonia.Indeed, spermatogonial differentiation is accompanied by the global loss of H3K27me3.By using a PRC1 conditional lossof-function model, we show that PRC1 directs deposition of PRC2-mediated H3K27me3 to shield the undifferentiated state and control spermatogonial differentiation.Unlike other tissue stem cell systems where PcG proteins promote cell cycle progression, we demonstrate that PRC1 acts as a negative cell cycle regulator to ensure the slow cycling property of undifferentiated spermatogonia.Therefore, our study uncovers that key features of undifferentiated spermatogonia, including the undifferentiated state, slow cycling property and the potential for commitment to differentiation, are molecularly linked by PRC1.

Animals
Generation of conditionally deficient Rnf2 mice in a Ring1 −/ − background was performed as described previously ( 30 ).Briefly, PRC1cKO mice Ring1 −/ − ; Rnf2 F / − ; Ddx4-Cre were generated from Ring1 −/ − ; Rnf2 F / F females crossed with Ring1 −/ − ; Rnf2 F / + ; Ddx4-Cre males and PRC1ctrl mice used in experiments were Ring1 −/ − ; Rnf2 F / + ; Ddx4-Cre littermate males.Mice were maintained on a mixed genetic background of FVB and C57BL / 6J.Generation of mutant Ring1 ( Ring1 −/ − ) and Rnf2 floxed alleles ( Rnf2 F / F ) was reported previously ( 48 ,49 ).Ring1 −/ − and Rnf2 F / F mice were obtained from Dr. Haruhiko Koseki.Ddx4-Cre transgenic mice and wild-type (WT) C57BL / 6J mice were purchased from the Jackson Laboratory ( 50 ).For each experiment, a minimum of three independent mice were analyzed.Mice were maintained on a 12:12 hour light: dark cycle in a temperature and humidity-controlled vivarium (22 ± 2 • C; 40-50% humidity) with free access to food and water in the pathogen-free animal care facility.Mice were used according to the guidelines of the Institutional Animal Care and Use Committee (IACUC: protocol no.IACUC2018-0040 and 21931) at Cincinnati Children's Hospital Medical Center and the University of California, Davis.

Flow cytometry and cell sorting
Flow cytometric experiments and cell sorting were performed using SH800S (SONY), with antibody-stained testicular single-cell suspensions prepared as described previously ( 51 ).Data were analyzed with SH800S software (SONY) and FCS Express 7 (De Novo Software).
Cells were analyzed after removing small and large debris in FSC-A versus SSC-A gating, doublets in FSC-W versus FSC-H gating, and 7AAD + dead cells.Then the desired cell population was collected in gates determined based on antibody staining ( Supplementary Figure S1 A).

RNA-seq library generation and sequencing
RNA-seq libraries from PRC1ctrl and PRC1cKO A undiff were prepared as follows using the Smart-Seq2 method.∼7000 A undiff cells were isolated from 2-month-old mice and pooled as one replicate.Two independent biological replicates were used for RNA-seq library generation.Total RNA was extracted using the RNeasy Plus Micro Kit (QIAGEN, Cat #74034) according to the manufacturer's instructions.Libraries were constructed using NEBNext ® Single Cell / Low Input RNA Library Prep Kit for Illumina ® (NEB, E6420S) according to the manufacturer's instructions.Prepared RNA-seq libraries were sequenced on the Novaseq 6000 system (Illumina) with 100-bp paired-end reads.
Stranded RNA-seq libraries were prepared for (i) WT A undiff and WT A diff ; (ii) PRC1ctrl A undiff and PRC1cKO A undiff after Win18,446 treatment.∼100 000-150 000 cells isolated from adult mice were pooled as one replicate, and two independent biological replicates were used for RNA-seq library generation.Total RNA was extracted using the RNeasy Plus Mini Kit (QIAGEN, Cat #74134) according to the manufacturer's instructions.Library construction was performed by the CCHMC DNA Sequencing and Genotyping Core (Cincin-nati, Ohio, USA) using the Illumina TruSeq stranded mRNA kit after polyA enrichment.Prepared RNA-seq libraries were sequenced on the Novaseq 6000 system (Illumina) with 100bp paired-end reads.
To identify differentially expressed genes in RNA-seq, raw read counts for each gene were generated using the htseq-count function, part of the HTSeq package ( 57 ) (version 1.6.0)based on mouse gene annotations (gencode.vM24.annotation.gtf,GRCm38 / mm10).DESeq2 ( 58 ) (version 1.28.1) was used for differential gene expression analyses with cutoffs log 2 FoldChange > 1 and FDR values ( P adjusted: P adj values) < 0.05.In DESeq2, P values attained by the Wald test were corrected for multiple testing using the Benjamini and Hochberg method by default (FDR values).FDR values were used to determine significantly dysregulated genes.The TPM values of each gene were generated using RSEM ( 59 ) (version 1.3.3)for comparative expression analyses and computing the Pearson correlation coefficient between biological replicates.
To perform GO analyses, we used the online functional annotation clustering tool Metascape ( 60 ) ( http:// metascape.org).Bioinformatics analyses were visualized as heatmaps using Morpheus ( https://software.broadinstitute.org/ morpheus/ ).The TPM values of each gene were first transformed by the Robust Z -Score Method, also known as the Median Absolute Deviation method, or log-transformed, then used as input for plotting.PCA plot was generated using TPM values by the web tool ClustVis ( https:// biit.cs.ut.ee/ clustvis/ ).

CUT&RUN and ChIP-seq data reanalysis
Processed bigwig files of CBX2, PHC2 and BMI1 CUT&RUN in adult spermatogonia and bedGraph files of RAR ChIPseq in GS cells were downloaded from Gene Expression Omnibus (GEO) under accession no.GSE210367 and GSE116798, respectively, and used for plotting directly.Raw STRA8 ChIP-seq data of mouse testes were downloaded from the DDBJ Sequence Read Archive (DRA) under accession number DRA007778 and aligned to the mouse genome (GRCm38 / mm10) using Bowtie2 with default options.Aligned bam files were converted to bigwig files using deeptools by normalizing with RPKM and then used for plotting.

Statistics
Statistical methods and P values for each plot are listed in the figure legends and / or in the Methods.Statistical significance for pairwise comparisons was determined using two-tailed unpaired t -tests.Next-generation sequencing data (RNA-seq, CUT&RUN / Tag) were based on two independent replicates.No statistical methods were used to predetermine sample size in these experiments.Experiments were not randomized, and investigators were not blinded to allocation during experiments and outcome assessments.

Dynamic gene expression changes during spermatogonial differentiation
To elucidate how A undiff are maintained and how they lose stem cell potential and commit to differentiation during adult spermatogenesis, we first isolated A undiff and early differentiating spermatogonia and performed gene expression analysis using bulk RNA-seq.We isolated these fractions from WT adult male C57BL / 6 mice at 2-3 months (mo) of age using our previously established fluorescence-activated cell sorting (FACS) method (Figure 1 B, Supplementary Figure S1 A) ( 2 ).We utilized the cell surface protein E-Cadherin to isolate A undiff and early differentiating spermatogonia ( 72 ,73 ), and KIT expression to distinguish differentiating spermatogonia (Figure 1 A).In addition, another cell surface protein, CD9, enabled the isolation of a highly pure spermatogonial population while excluding somatic contaminants ( 73 ).Using this method, A undiff were collected based on E-Cadherin expression and high levels of CD9, as well as the absence of KIT (E-Cadherin + CD9 high KIT -); early differentiating spermatogonia were isolated based on the expression of KIT and E-Cadherin, and a medium level of CD9 (E-Cadherin + CD9 medium KIT + ).The E-Cadherin + CD9 medium KIT + fraction contains highly enriched early stages of differentiating spermatogonia, most likely A 1 through A 4 spermatogonia, rather than In and B spermatogonia that are about to enter meiosis ( 2 , 72 , 74 ) ( Supplementary Figure S1 A).Therefore, we hereafter abbreviate the E-Cadherin + CD9 medium KIT + fraction as A diff .We confirmed the high purity of these cell populations ( Supplementary Figure S1 B, C) and high reproducibility of RNA-seq data between biological replicates ( Supplementary Figure S1 D).
Comparing gene expression between A undiff and A diff revealed a dynamic change in the transcriptome during spermatogonial differentiation, marked by the upregulation of 1458 genes and downregulation of 2070 genes (Figure 1 C, Supplementary Data 1 ).GO term analysis revealed that genes involved in cell morphogenesis, cell cycle, cell fate commitment and meiosis were upregulated upon differentiation.In contrast, genes functioning in cytokine production, cell adhesion and especially stem cell population maintenance were downregulated ( Supplementary Figure S1 E).Consistent with this analysis, A undiff markers ( Id4 , Gfra1 , Zbtb16 ( PLZF ), Neurog3 ( Ngn3 ) and Tspan8 ) and pluripotency markers ( Oct4 , Nanos3 and Lin28a ) were downregulated upon differentiation, while Kit was upregulated in A diff ( Figure 1 C).In addition, Stra8 and retinoic-acid-binding-protein 1 Crabp1 were significantly upregulated in A diff (Figure 1 C), consistent with the roles of RA signaling in spermatogonial differentiation.Further, cell cycle factors ( Ccnd1 , Ccna2 , Ccne1 and Ccne2 ) were markedly upregulated in A diff , in line with the rapid proliferative state of this population, and other recentlyidentified markers for spermatogonial differentiation (such as Esx1 and Tcea3 ) ( 75 ) were also upregulated in A diff (Figure 1 C, Supplementary Figure S1 F).Taken together, these analyses confirm that distinct gene expression programs are active in A undiff compared to A diff populations in adult testes.

Polycomb-mediated chromatin remodeling underlies spermatogonial differentiation
To determine the chromatin states underlying the gene expression changes during spermatogonial differentiation, we performed CUT&Tag (Cleavage Under Targets and Tagmentation) ( 53 ,54 ) or CUT&RUN (cleavage under targets and release using nuclease) ( 52 ,76 ) analyses of A undiff and A diff in adult testes.Using cell-number normalized spike-in controls, we quantitatively determined the genomic distributions of (i) PRC1-mediated H2AK119ub (by CUT&Tag), (ii) PRC2mediated H3K27me3 (by CUT&Tag); and (iii) an active mark H3K4me3 (by CUT&RUN) in both isolated A undiff and A diff fractions (Figure 1 D, Supplementary Figure S2 ).This spike-in approach enabled quantitative evaluation of changes in histone marks during spermatogonial differentiation and overcame the major limitation of previous epigenomic profiling studies that were not quantitative ( 37 , 38 , 45-47 ).We confirmed the high reproducibility between biological replicates ( Supplementary Figure S2 A), and these biological replicates were pooled for downstream analyses.
We found that H2AK119ub peaks are abundant throughout the genome in both A undiff and A diff (63182 peaks in A undiff and 47409 peaks in A diff : Figure 1 D, Supplementary Figure S2 B), and are partially associated with H3K27me3, H3K4me3, or both ( Supplementary Figure S2 C-F).A notable change upon differentiation is a marked reduction of H3K27me3 enrichment and the number of genome-wide H3K27me3 peaks (Figure 1 D).Among all H3K27me3 peaks detected across both A undiff and A diff (25077 peaks), more than 50% (13597 peaks) were identified as differential peaks (FDR < 0.05), and most of these H3K27me3 peaks were lost upon differentiation (Figure 1 E).Thus, our quantitative analysis uncovers a marked reduction of H3K27me3 as a characteristic of spermatogonial differentiation not noted in previous profiling studies ( 45 ,46 ).
Because PRCs bind gene promoters to regulate transcription ( 77-80 ), we next focused on the analysis of gene promoters.Among all Refseq gene promoters, 3295 promoters showed differential enrichment of H3K27me3 from A undiff to A diff , and 2911 of them showed a significant loss of H3K27me3 (Figure 2 A).To examine the relationship between chromatin state and gene regulation, we analyzed the enrichment of these marks at promoter regions of the differentially expressed genes (DEGs) during spermatogonial differentiation (identified in Figure 1 C).Upregulated genes in A diff were associated with a significant loss of H3K27me3 and a modest gain of H3K4me3 while maintaining stable H2AK119ub levels (Figure 2 B, C).In contrast, downregulated genes in A diff were associated with a modest gain of H2AK119ub on promoters without changes in H3K27me3 and H3K4me3 (Figure 2 B, C).These features were evident at each representative locus: at the Kit gene locus, which is significantly induced in A diff , H3K27me3 was lost while H2AK119ub was maintained at its promoter during spermatogonial differentiation (Figure 2 D).In contrast, at the Lin28a gene locus, a stem cell marker which is repressed in A diff , we saw increased H2AK119ub enrichment without apparent changes in H3K27me3 and H3K4me3 during spermatogonial differentiation (Figure 2 E).The gain of H2AK119ub is in line with the recently reported function of a canonical PRC1 (cPRC1) component CBX2 in spermatogonia ( 81 ).
To capture genome-wide changes of these marks at promoters, we performed a k -means clustering analysis.In agreement with its genome-wide abundance, H2AK119ub was present on a broad range of promoters in clusters 1-4 (Figure 2 F).H2AK119ub and H3K27me3 were similarly abundant in Cluster 1 and 2, which was also marked by weak H3K4me3 (Figure 2 F).Cluster 1-2 genes are common targets of various Polycomb proteins (SUZ12, RNF2, MTF2, JARID2 and EZH2) in embryonic stem cells (ESCs; Supplementary Figure S3 A) and are associated with transcription and developmental processes ( Supplementary Figure S3 B).Cluster 2 genes showed a marked reduction in H3K27me3 levels upon spermatogonial differentiation (Figure 2 F).The presence of both H3K27me3 and H3K4me3 are characteristic of bivalent promoters, a chromatin context thought to maintain the potential for expression of developmental regulator genes in the germline ( 37 , 47 , 82 , 83 ).Thus, spermatogonial differentiation is accompanied by a reduction of H3K27me3 in a subset of bivalent promoters.
Notably, H2AK119ub was also enriched at promoters of active genes marked by high H3K4me3 level (Cluster 3 and 4) (Figure 2 F), suggesting that H2AK119ub has a broader role than H3K27me3 in transcriptional regulation.These Cluster 3 and 4 genes are targets of various transcription factors, such as FOXO3, MYC, E2F1 and YY1 ( Supplementary Figure S3 A), and function in a broad range of biological processes, including RNA processing, DNA metabolic process, DNA repair, and cell cycle ( Supplementary Figure S3 B), which are constitutively active during spermatogonial differentiation (Figure 2 F).
Together, these results demonstrate that H3K27me3 reduction is a prominent change that occurs during spermatogonial differentiation, raising the possibility that H3K27me3 shields the undifferentiated state of adult A undiff to antagonize commitment to differentiation.

PRC1 safeguards A undiff from over-proliferation and differentiation
PRC1 and PRC2 are both required for the maintenance of undifferentiated spermatogonia ( 30 ,31 ).However, how PRCs function in steady-state spermatogenesis to sustain male fertility in the long term remains unknown.To elucidate this, we evaluated how the loss of PRC1 function affects steadystate spermatogenesis using a previously established conditional knockout mouse model ( 30 ,84 ).In brief, PRC1 function was eliminated after embryonic day 15 (E15) using Ddx4 -Cre ( 50 ) to inactivate the E3 ubiquitin ligase RNF2 in a background lacking its partially redundant paralog, RING1 (RING1A) ( 30 , 48 , 84-86 ).These PRC1 conditional knockout (PRC1cKO) mice show nearly complete depletion of H2AK119ub in germ cells ( 30 ,84 ), representing a highly efficient loss-of-function PRC1 model.
Although PRC1 inactivation leads to gradual loss of the stem cell population and depletion of differentiated germ cells in postnatal spermatogenesis ( 30 ), A undiff (detected based on the expression of PLZF, a marker for undifferentiated spermatogonia) were still present at 2 mo in PRC1cKO males (Figure 3 A).By focusing on the analysis of PRC1cKO A undiff at 2 mo, we sought to determine the role of PRC1 in steady-state spermatogenesis during which A undiff slowly divide to maintain the stem cell pool.A previous study estimated that the doubling time of A undiff is 3.9 days (93.6 h), which becomes shorter once A undiff commit to spermatogonial differentiation (28.5-34 h) ( 87 ).Therefore, we first evaluated the proportion of A undiff with an active cell cycle at 2 mo by immunostaining with Ki67, a proliferation marker present in G1, S, G2 and M, but absent in G0 ( 88 ).We compared PRC1cKO males to littermate control mice (PRC1ctrl) which show normal spermatogenesis despite being Ring1 −/ − ( 30 ).Consistent with the slow cycling of A undiff , only 17.4% of A undiff in PRC1ctrl mice were Ki67 + (Figure 3 A, B).However, 70.4% of A undiff in the PRC1cKO testes were Ki67 + , a ∼3-fold increase compared to PRC1ctrl males (Figure 3 A, B), suggesting that PRC1cKO A undiff are highly proliferative at 2 mo.However, the proportion of Ki67 + A undiff was comparable between PRC1cKO and PRC1ctrl in juvenile testes at postnatal day 7 (P7) ( 30 ).Thus, the increased proliferation of A undiff is a unique feature of steady-state spermatogenesis in PRC1cK O .
To examine the cellular states of spermatogonia populations, we performed flow cytometry analyses of PRC1ctrl and PRC1cKO testicular cells at 2 mo.Like WT ( Supplementary Figure S1 ), the CD9 + E-Cadherin + spermatogonia in PRC1ctrls consisted of two distinct populations (A undiff and A diff ; Figure 3 C).By contrast, PRC1cKO spermatogonia did not show a clear separation between these two populations.Instead, intermediate state spermatogonia, that are transitioning to A diff , were abundant (CD9 + E-Cadherin + KIT medium cells, abbreviated as 'inter', Figure 3 C).43.2% of total CD9 + E-Cadherin + spermatogonia in PRC1cKO males were transitioning, significantly higher than the 12.2% present in PRC1ctrls (Figure 3 D).These results raise the possibility that PRC1 has cellintrinsic roles in safeguarding A undiff from over-proliferation and suppressing precocious commitment to differentiation.
Nevertheless, an alternative possibility is that the effects are cell-extrinsic: the above phenotypes could be indirectly caused by depletion of differentiated germ cells in PRC1cKO (Figure 3 A) ( 30 ).As a result, A undiff would become overcommitted to differentiation / proliferation in order to repopulate the depleted tubules.To distinguish between a cell-intrinsic and a cell-extrinsic effect resulting from PRC1 deletion, we blocked spermatogonial differentiation using Win18,446, an inhibitor of the enzyme aldehyde dehydrogenase family 1, subfamily A2 (ALDH1a2).This enzyme catalyzes endogenous RA biosynthesis from retinol (vitamin A) thus promoting RAsignaling which is essential for spermatogonial differentiation and meiosis initiation (89)(90)(91)(92).
We reasoned that treatment with Win18,446 would block differentiation and enable evaluation of A undiff without the influence of extrinsic differentiation signals.Pubertal PRC1ctrl and PRC1cKO males (1-month-old mice) were treated with Win18,446 injection daily until they fully matured at 2 months.After the treatment, both PRC1ctrl and PRC1cKO testes contained 'empty' tubules devoid of differentiated germ cells, and only PLZF + A undiff were present, confirming the successful blockage of spermatogenesis after treatment with Win18,446 (Figure 3 E).Due to RA-signaling inhibition, the proliferation rates of A undiff were largely decreased in both PR C1ctrl and PR C1cK O , but PRC1cK O A undiff were more susceptible to the RA-signaling inhibition (3.1-fold change from 17.4% to 5.7% in PRC1ctrl; versus 4.5-fold from 70.7% to 15.7% in PRC1cKO; Figure 3 B, F).Overall, these results suggest that the RA-mediated differentiation cue extrinsically promotes the proliferation of A undiff, and the effect of RA is more pronounced in the absence of differentiated germ cells in PRC1cKO mice.However, the proliferative activity of A undiff in the PRC1cKO was still significantly higher than in PRC1ctrl testes (15.7% versus 5.7% of all A undiff ; Figure 3 F), suggesting that PRC1 cell-intrinsically suppresses the proliferation of A undiff .In addition, enrichment of intermediate state spermatogonia was still observed after treatment with Win18,446 (Figure 3 G, H).Altogether, we conclude that PRC1 cell-intrinsically safeguards A undiff from overproliferation / differentiation.

PRC1 suppresses the spermatogonial differentiation program in A undiff
To determine the function of PRC1 in A undiff , using the FACS method described above, we isolated A undiff from 2-month-old PRC1ctrl and PRC1cKO testes (identical E-Cadherin + CD9 high KIT -population) and performed bulk RNA-seq ( Supplementary Figure S4 A).We found that, in PRC1cK O , 230 genes were significantly upregulated, and 303 genes were significantly downregulated compared to PRC1ctrls (Figure 4 A and Supplementary Data 2 ).Notably, upregulated genes included those known to be related to spermatogonial differentiation ( Kit, Esx1 and Crabp1 ) as well as meiotic processes ( Prdm9, M1ap and Meioc ), whereas genes required for A undiff maintenance ( Sall4, Lin28a, Nanos3 and Sall1 ) were downregulated (Figure 4 A).We further performed an RNA-seq analysis of A undiff after treatment with Win18,446, to minimize cell-extrinsic effects (Figure 4B; Supplementary Figure S4 B).Under this condition, more genes were misregulated in PRC1cKO compared to PRC1ctrl (570 genes upregulated and 836 genes downregulated), and, of note, spermatogonial differentiation genes (e.g.Kit, Dmrt1, Dmrtb1, Esx1 and Tcea3 ) were still significantly upregulated after the treatment (Figure 4 B, Supplementary Data 2 ).
It is worth noting that the Cdkn2a gene, which encodes two major cell cycle inhibitors (p16 (INK4A) and p14 (ARF)) and is a major Polycomb target for suppression through which Polycomb promotes proliferation in many other tissue stem cells and somatic cells, was not upregulated in the PRC1cKO A undiff ( Supplementary Figure S4 C), consistent with previous studies using Polycomb mutants in spermatogenesis ( 30 ,31 ).Therefore, Cdkn2a is not involved in the PRC1-mediated regulation of cell cycle in spermatogonia.GO analysis revealed that upregulated genes in both PRC1cKO A undiff samples were enriched for cell-cell adhesion and cell fate commitment, processes associated with development / morphogenesis ( Supplementary Figure S4 D, E).In turn, downregulated genes were over-represented with diverse molecular functions, including stem cell population maintenance, spermatogenesis and chromatin organization implying compromised spermatogenic programs in PRC1cKO A undiff ( Supplementary Figure S4 D, E).
To clarify the role of PRC1 in spermatogonial differentiation, we compared all RNA-seq data in this study by Principalcomponent analysis (PCA) to capture overall transcriptomic changes among all pairs of samples (Figure 4 C).A transcriptome change during spermatogonial differentiation in WT (from WT A undiff to WT A diff ) was accompanied by increased PC2 loading.Similar directional changes (PC2 increase) were observed in PRC1ctrl A undiff and PRC1cKO A undiff both in non-treated and Win18,446-treated conditions.Notably, the genes contributing to the positive PC2 values were enriched with those upregulated during spermatogonial differentiation, involved in mitotic cell cycle, retinoic acid receptor signaling pathway, and spermatogenesis.Examples include Ccna2, Mki67, Stra8 and Kit (Figure 4 D).
To validate this observation, we next examined the expression level of previously identified genes that are upregulated during spermatogonial differentiation ( 75 ).These differentiating spermatogonia marker genes tended to show increased expression levels in the A undiff of PRC1cKO males both in non-treated and Win18,446-treated conditions compared to PR C1ctrls (Figure 4 E).W e also tested how the DEGs detected in WT spermatogonial differentiation (shown in Figure 1 C) are regulated by PRC1.Genes upregulated in WT A diff showed significantly higher expression in the A undiff of PRC1cKO than PRC1ctrl (Figure 4 F), suggesting they were precociously activated in mutant A undiff .By contrast, genes active in WT A undiff showed decreased expression in PRC1cKO A undiff compared to PRC1ctrls.
Collectively, genes linked to spermatogonial differentiation were precociously activated in PRC1cKO A undiff whereas those linked to A undiff identity were present at decreased levels.Therefore, we conclude that PRC1 shields the undifferentiated state of adult A undiff by suppressing the spermatogonial differentiation program.

Removal of PRC1 impairs the global deposition of H3K27me3 in A undiff
To determine the mechanism by which PRC1 regulates spermatogonial differentiation at the chromatin level, we applied quantitative CUT&RUN to profile the genomic distributions of H3K27me3 and H3K4me3 in FACS-purified A undiff from PRC1ctrl and PRC1cKO at 2 months of age (Figure 5 A and Supplementary Figure S5 A).Since our data show that H3K27me3 levels are a major difference between A undiff and A diff in WT (Figure 2 ), we sought to determine how PR C1 regulates PR C2-mediated H3K27me3.Remarkably, the H3K27me3 landscape was greatly impaired in PRC1cKO A undiff with a dramatic reduction of both peak numbers and intensity ( Supplementary Figure S5 B, C, D).H3K27me3 peak distribution changed in various genomic regions, including gene bodies and intergenic regions ( Supplementary Figure S5 C), and a global loss of H3K27me3 at gene promoters was observed in PRC1cKO A undiff (Figure 5 A).The active mark, H3K4me3, was instead slightly increased at gene promoters (Figure 5 A).These features are exemplified by track views of the Kit, Stra8 and Tcea3 loci (Figure 5 B), where H3K27me3 nearly completely disappeared at promoters, consistent with transcriptional de-repression.These results suggest that PRC1 regulates the distribution of H3K27me3 and counteracts H3K4me3 in A undiff .After treatment with Win18,446, genome-wide H3K27me3 deposition was still largely impaired in PRC1cKO A undiff ( Supplementary Figure S5 B-D), and there was a prominent loss of H3K27me3 at gene promoters (Figure 5 A), indicating that the PRC1 regulation of H3K27me3 is a cell-intrinsic feature of A undiff .By contrast, after treatment with Win18,446, H3K4me3 slightly decreased at promoters of PRC1cKO A undiff , as opposed to the slight increase in non-treated PRC1cKO A undiff , suggesting that H3K4me3 changes were secondary consequences of PRC1 deletion.
We further identified 3006 promoters with significant changes in H3K27me3 occupancy in PRC1cKO A undiff , and 2999 of them lost H3K27me3 (Figure 5 C), resembling changes that occur during WT spermatogonial differentiation.Remarkably, genes showing H3K27me3 loss at promoters in PRC1cKO A undiff significantly overlapped with genes that lost H3K27me3 in WT A diff (Figure 5 D); 1893 overlapping genes included those involved in cell / tube morphogenesis, cell fate commitment and MAPK signaling (Figure 5 E, Supplementary Data 3 ).A similar overlap was observed in PRC1cKO after treatment with Win18,446 ( Supplementary Figure S6 A-C, Supplementary Data 3 ).These results suggest that the overall chromatin state of PRC1cKO A undiff resembles that of WT A diff , raising the possibility that PRC1 regulates PRC2-mediated H3K27me3 to shield the chromatin state in A undiff and prevent precocious differentiation.

PRC1 suppresses precocious spermatogonial differentiation and ectopic somatic program in A undiff
Because PcG proteins regulate a broad range of developmental genes that are critical for pattern formation and cell fate specification in development, we next sought to determine how PRC1 regulates developmental genes in A undiff .To this end, we used a list of canonical PcG target genes that are known to be suppressed to maintain developmental potential in ESCs ( 93 ).We reasoned that PRC1 shields the undifferentiated state by defining an A undiff -specific gene expression program.Indeed, canonical PcG target genes largely overlapped with genes that lose H3K27me3 at promoters in WT spermatogonial differentiation and genes that carry PRC1-dependent H3K27me3 at promoters ( Supplementary Figure S6 D, E).
We first categorized the canonical PcG target genes into three major groups based on their expression dynamics dur-ing spermatogonial differentiation from A undiff to A diff in WT: (i) 608 upregulated PcG target genes (PcG up); (ii) 618 downregulated PcG target genes (PcG down); (iii) 2373 PcG targets that did not show significant changes (PcG stable: Figure 6 A).A notable feature was decreased H3K27me3 at entire regions of 'PcG up' genes (TSSs, gene bodies, TESs (transcription end sites)) during WT spermatogonial differentiation (Figure 6 B).On these 'PcG up' genes, H3K27me3 was largely PRC1-dependent as shown in PRC1cKO A undiff with or without Win18,446 (Figure 6 C), indicating that the loss of PRC1-dependent H3K27me3 occurred upon spermatogonial differentiation.Quantification of H3K27me3 at promoters further confirmed this conclusion (Figure 6 D).These 'PcG up' genes were upregulated in PRC1cKO A undiff (Figure 6 E).Therefore, in A undiff , PRC1 prevents precocious activation of developmental genes linked to spermatogonial differentiation.
These results also suggest that PRC1 plays a major role in restricting transcription of a large number of PcG targets through the regulation of PRC2 during spermatogonial differentiation.In addition, decreased H3K27me3 was also observed on 'PcG stable' genes during spermatogonial differentiation (Figure 6 B), and, in A undiff , this is largely PRC1-dependent (Figure 6 C, D).The 'PcG stable' genes were ectopically upregulated in PRC1cKO A undiff (Figure 6 E).Therefore, in A undiff , PRC1 suppresses ectopic activation of developmental genes that are required for somatic development.Furthermore, increased H2AK119ub was observed on PcG down genes (Figure 6 D), suggesting that PRC1 directly represses developmental genes during spermatogonial differentiation.Although this PRC1-dependent regulation of H3K27me3 was consistent regardless of Win18,446 treatment, H3K4me3 showed distinct changes in the presence or absence of Win18,446 ( Supplementary Figure S6 F, G), further demonstrating an indirect role for PRC1 in H3K4me3 regulation.
Misregulated gene expression at the Hoxa cluster, which includes classic Polycomb-targeted developmental genes, exemplifies the changes observed in PRC1cKO A undiff ( Supplementary Figure S7 ).These Hoxa genes were regulated differently during WT spermatogonial differentiation, e.g. the Hoxa1, Hoxa2 and Hoxa3 gene loci substantially lost H3K27me3, resulting in transcriptional upregulation; however, the Hoxa9, Hoxa10 and Hoxa11 genes gained H2AK119ub and were downregulated.In PRC1cKO A undiff , H3K27me3 was depleted across the entire Hoxa cluster, leading to ectopic de-repression of Hoxa genes.These results demonstrate that PRC1 has two major functions in A undiff : preventing precocious activation of developmental genes that are activated later during spermatogonial differentiation and suppressing ectopic expression of the somatic program.Together these functions shield the undifferentiated state by defining an A undiff -specific gene expression program.

PRC1 is a negative regulator of the cell cycle in A undiff
Because we found increased proliferation of the PRC1cKO A undiff that leads to A undiff exhaustion, we next examined the PRC1-dependent regulation of cell cycle-related genes.We found that H2AK119ub accumulated on genes that positively regulate the cell cycle, e.g.cyclins, including Ccna1, Ccnd1, Ccne1 and Ccne2 ; the cyclin-dependent kinases, Cdk2 and Cdk5 ; cell division regulators, Cdc6 and Cdc25a ; as well as the cell cycle inhibitor, Cdkn1b ; suggesting direct regulation of these genes by PRC1 (Figure 7 A, Supplementary Figure S8 A, B).Consistent with the over-proliferation phenotype of PRC1cKO A undiff , increased expression of these pro-cell cycle regulators and decreased Cdkn1b expression in PRC1cKO A undiff were observed ( Supplementary Figure S8 A, B).Among these genes, H3K27me3 only accumulated on the Ccna1 gene , indicating that PRC1 regulates the other genes independent of PRC2 (Figure 7 A).Together with the over-proliferation phenotype of PRC1cKO cells, these results demonstrate that PRC1 is a negative regulator of the cell cycle in A undiff that maintains the characteristic slow cycling of A undiff .Therefore, the slow cycling of A undiff is part of a PRC1-directed germline program that protects the undifferentiated state of adult A undiff and regulates spermatogonial differentiation.

Discussion
In this study, we find that PRC1 plays a central role in defining the A undiff -specific developmental program and maintaining a long-lasting chromatin state in A undiff during adult steady-state spermatogenesis.PRC1 shields the undifferentiated state of A undiff from spermatogonial differentiation.Furthermore, PRC1 is a negative cell cycle regulator that maintains slow cycling of A undiff .In the absence of PRC1, cell proliferation increased in A undiff both cell-intrinsically and cellextrinsically, and A undiff committed to abnormal differentiation.Our quantitative epigenomic analysis clarified the dynamics of chromatin changes during spermatogonial differentiation, an aspect not quantitatively evaluated in previous studies ( 37 , 38 , 45-47 ).Importantly, we show that H3K27me3 is a major epigenetic hallmark of A undiff maintenance.Indeed, spermatogonial differentiation is accompanied by a global resolution of bivalent domains through the loss of H3K27me3 on genes required for spermatogonial differentiation.Our data argue that PRC1 directs PRC2-H3K27me3 deposition, thus supporting A undiff maintenance (Figure 7 B, C).Overall, we have determined a molecular logic underlying spermatogonial differentiation: the loss of H3K27me3 at promoters activates the differentiation program (Figure 7 B, C).
In WT spermatogonial differentiation, an extensive loss of H3K27me3 at target promoters does not necessarily lead to massive gene de-repression in normal A diff , presumably due to constant genome-wide occupancy of H2AK119ub.Hence, H3K27me3 does not maintain constitutive repression at bivalent promoters alone; instead, H2AK119ub acts as an additional safety lock to ensure developmental gene regulation.Further, spermatogonial differentiation occurs with the suppression of A undiff genes by PRC1 independent of PRC2.Without PRC1, both spermatogonial differentiation genes and somatic genes were ectopically activated, leading to abnormal differentiation (Figure 7 D).Our study demonstrates extensive direct regulation of PR C2-H3K27me3 by PR C1-H2AK119ub in vivo .Although early studies showed that cPRC1 acts downstream of PRC2-deposited H3K27me3 ( 94 ,95 ), recent studies have instead demonstrated that variant PRC1 (vPRC1) mediates H2AK119ub-dependent PRC2 recruitment (96)(97)(98).We reanalyzed recent cPRC1 CUT&RUN data in adult spermatogonia ( 81 ) and found that key cPRC1 components, including CBX2, PHC2 and BMI1, bind a small portion of PRC2-H3K27me3 target genes specifically in KIT + differentiating spermatogonia ( Supplementary Figure S9 A).This analysis supports the hypothesis that vPRC1 plays a major role in regulating PRC2-H3K27me3 in A undiff .We also note that cPRC1 and vPRC1 are regulated differently during spermatogonial differentiation.Several genes encoding cPRC1 components, such as Pcgf2 , Cbx2 / 4 and Phc2 genes, show increased expression as spermatogonia move from the A undiff to A diff state ( Supplementary Figure S9 B).However, the regulation of vPRC1 seems complicated; while the expression of the Pcgf3 / 5 genes (encoding the components of vPRC1.3 and 1.5) is increased, the expression of the Pcgf6 gene (encoding the components of vPRC1.6) is decreased, raising the possibility that vPRC1 subcomplexes are reorganized upon spermatogonial differentiation.Further studies are needed to clarify the regulation of PRC1 subcomplexes in spermatogonial differentiation.
Major questions raised in this study are how H3K27me3 loss occurs at the onset of spermatogonial differentiation and what triggers this process.There are three possible mechanisms that could account for H3K27me3 loss.One is decreased PRC2 activity, the second is H3K27 demethylation, and the third is histone H3 replacement with a testis-specific histone variant, H3T ( 99 ).We found that the expression levels of PRC2 components are largely consistent; however, the H3K27 demethylase KDM6B / JMJD3 is significantly upregulated from A undiff to A diff ( Supplementary Figure S9 B).In line with this observation, a germline-specific deletion of KDM6B leads to increased and prolonged maintenance of undifferentiated spermatogonia ( 100 ).Therefore, we speculate the KDM6B actively demethylates H3K27me3 at specific target loci at the onset of spermatogonial differentiation.As for the trigger of H3K27me3 loss, we suspect that RA signaling plays a role.This is because our reanalysis of a previous study ( 101 ) suggests that RA receptor (RAR) and STRA8 bind to the promoters of the Rnf2 , Eed (PRC2 component), and Kdm6b genes; this RAR binding increases in response to RA ( Supplementary Figure S9 C).These data suggest that the RA signaling may trigger the expression of the Kdm6b gene, leading to H3K27me3 loss as A undiff differentiate.Intriguingly, the RA receptor gene loci ( Rara , Rarb and Rarg ) and the Stra8 gene locus are also targeted by PR C1 / PR C2 ( Supplementary Figure S9 D), suggesting that the Polycomb system and RA-STRA8 signaling may form a regulatory loop to coordinate each other's expression.
We note that H2AK119ub is pervasively enriched at highly active genes (Figure 2F; Clusters 3 and 4) during spermatogonial differentiation, a phenomenon that has been reported in many other contexts, including growing oocytes ( 102 ), skin stem cells ( 103 ) and leukemic cells ( 104 ).Dual roles for PRC1 in gene activation and repression were also reported in juvenile spermatogonia ( 30 ).In adult spermatogonia, in a small portion of H2AK119ub / H3K4me3 co-enriched active genes, we found both upregulated and downregulated DEGs in WT A diff or PRC1cKO A undiff ( Supplementary Figure S9 E), supporting a mechanism where PRC1 fine-tunes gene repression and activation.However, it is unlikely that local H2AK119ub at active gene promoters plays a critical role in gene regulation because these genes are not greatly dysregulated in PRC1cKO A undiff .
Notably, these gene regulatory mechanisms are tightly coupled with cell cycle regulation by PRC1.This feature is unique to A undiff because PRC1 positively regulates the cell cycle in somatic cells by suppressing the major cell cycle inhibitor Cdkn2a gene, and the loss of PcG-related functions leads to cell cycle inhibition ( 105 ,106 ).In contrast, a Cdkn2aindependent function of PRC1 regulates the slow cycling property of A undiff prior to the increased proliferation observed upon spermatogonial differentiation.We found that PRC1 directly suppresses genes that activate the cell cycle and promotes expression of the cell cycle inhibitor Cdkn1b , thereby functioning as a negative regulator of the cell cycle in A undiff .In PRC1cKO A undiff, increased expression of these procell cycle regulators and decreased Cdkn1b expression likely account for their over-proliferation phenotype.Alternatively, increased expression of pro-apoptotic genes, e.g.Bok, Bax and Casp8, resulting from loss of PRC1 activity at these loci could also contribute to the gradual loss of A undiff in PRC1cKO testes ( Supplementary Figure S8 C).
Our study uncovers PRC1-directed mechanisms for the maintenance and differentiation of undifferentiated spermatogonia in adults.These newly defined Polycomb-regulated mechanisms are likely to be related to the unidirectional differentiation program of spermatogenesis, in which the chromatin states of progenitor cells are predetermined to direct spermatogenic differentiation ( 47 ,107 ).Similarly, PRC2-H3K27me3 functions in human pluripotent cells to shield the potency of stem cells ( 108 ,109 ); thus, a parallel mechanism may underlie stem cell maintenance in general.Given the persistent nature of H3K27me3 in the germline ( 37 , 45 , 47 , 110 ), PRC1-directed regulation of PRC2-H3K27me3 could be a pivotal mechanism to sustain germline potential and to recover totipotency and somatic gene regulation after fertilization.

Figure 1 .
Figure 1.Dynamic transcriptomic and epigenomic changes during spermatogonial differentiation in adult wild-type mice.( A ) Schematic of spermatogonial hierarchy and marker gene expression (created with BioRender.com).Based on nuclear morphology, spermatogonia can be classified as type A, intermediate, or type B. Type A includes the undifferentiated spermatogonia (A undiff , including A single (A s ), A paired (A pr ) and A aligned (A al )), and early differentiating spermatogonia (A diff , including A 1 , A 2 , A 3 and A 4 ), whereas intermediate (In) and type B encompass the later differentiating spermatogonia.( B ) Ov ervie w of the e xperimental steps to perf orm bulk RNA-Seq and chromatin profiling using CUT&R UN / T ag on F ACS-isolated spermatogonia (created with BioRender.com).( C ) MA plot showing a comparison of transcriptomes between A undiff and A diff in adult WT mice (2-3 mo).All genes with False Discovery Rate (FDR) values are plotted.Differentially expressed genes (DEGs: Log 2 FoldChange > 1, FDR < 0.05) are color-coded (orange: upregulated genes; green: downregulated genes).Two independent biological replicates were examined for RNA-seq.( D ) Heatmaps and a v erage tag density plots showing peaks of H2AK119ub, H3K27me3 and H3K4me3 in WT A undiff and A diff .The numbers of peaks are indicated ( n ).The color k e y s represent signal intensity, and the numbers represent spik e-in normaliz ed reads counts.( E ) Pie c hart and bar c hart showing the H3K27me3 peak change and the genomic distribution of gained and lost H3K27me3 peaks during spermatogonial differentiation.The numbers of peaks are shown (n).

Figure 2 .
Figure 2. Spermatogonial differentiation is accompanied by a global loss of H3K27me3.( A ) Scatter plot showing the comparison of H3K27me3 enrichment at promoters (TSS ± 2 kb) between A undiff and A diff in adult WT mice.Red dots represent genes with differential H3K27me3 enrichment at promoters (FDR < 0.05).( B ) Average tag density plots showing H2AK119ub, H3K27me3 and H3K4me3 enrichment at promoter regions (TSS ± 2 kb) of DEGs during spermatogonial differentiation in adult WT mice (defined in Figure 1 C).( C ) B o x-and-whisk er plots showing H2AK119ub, H3K27me3 and H3K4me3 enrichment around TSSs in the corresponding gene groups shown in (B).Central bars represent medians, the boxes encompass 50% of data points, and the whiskers include 90% of data points.n.s., not significant, *** P < 0.001, **** P < 5.1 × 10 −19 ; t wo-t ailed unpaired t -tests.( D and E ) Track views of the Kit and Lin28a gene loci showing differential H2AK119ub, H3K27me3 and H3K4me3 occupancy in WT A undiff and A diff .Data ranges in the upper right represent spike-in normalized reads counts from combined replicates.( F ) Heatmaps and average tag density plots of k -means clusters showing H2AK119ub, H3K27me3 and H3K4me3 enrichment at promoter regions (TSS ± 2.5 kb) in WT A undiff and A diff .The color keys represent signal intensity, and the numbers represent spike-in normalized reads counts.Right: Heat map showing RNA-seq expression (log 2 -transformed TPM) of corresponding k -means clusters.

Figure 3 .
Figure 3. PRC1 deficiency causes over-proliferation of A undiff .( A and E ) Schematics of the experimental workflow and immunostaining of testicular sections in PRC1cKO and a control littermate at 2 mo (A) or after win18,446 treatment (E).Ki67 (red) is a proliferation marker, and PLZF (green) is an undifferentiated spermatogonia marker.White arrows indicate the quiescent A undiff , and yellow arrows indicate A undiff in the active cell cycle.Scale bars, 50 μm (10 μm in the bo x ed areas).( B and F ) Quantitative analysis of immunostaining (A) and (E).Percentages representing numbers of PLZF + ; Ki67 + cells per number of PLZF + cells.n indicates the number of cells used for analysis.Data are represented as mean ± SD.Two-tailed unpaired Student's t -tests.Three independent biological replicates were analyzed for each genotype.( C and G ) Flow Cytometry profiles of all CD9 + cells from adult mouse testes in PRC1ctrl and PRC1cKO mice at 2 mo (C) or after win18,446 treatment (G).Percentages of each fraction relative to total cells are indicated.At least three independent biological replicates were analyzed for each genotype and representative profiles are shown.( D and H ) Quantitative analysis of Flow Cytometry data (C) and (G).Percentages of each spermatogonial subpopulation per total CD9 + E-Cadherin + spermatogonia populations.Data are represented as mean ± SD.Two-tailed unpaired Student's t -tests.Individual data points are also plotted.

Figure 4 .
Figure 4. PRC1 deficiency in A undiff leads to precocious activation of the spermatogonial differentiation program.( A and B ) Comparison of PRC1ctrl and PRC1cKO A undiff transcriptomes at 2 mo (A) or after Win18,446 treatment (B).Two independent biological replicates were examined for RNA-seq.All genes with false discovery rate (FDR) values are plotted.Differentially expressed genes (DEGs: 2 FoldChange > 1, FDR < 0.05, TPM > 1) are colored (red: upregulated in PRC1cKO A undiff ; blue: downregulated in PRC1cKO A undiff ), and numbers are shown.( C ) PCA plot showing transcriptome relationships between A undiff and A diff in WT, and A undiff between PRC1ctrl PRC1cKO at 2 mo with or without Win18,446.Arrows indicate increased PC2 loading.( D ) GO enrichments and representative genes contributing to high PC2 loading (top 500 genes) in PCA (C) are listed.( E ) Heatmap showing gene expression of representative differentiating spermatogonia markers in PRC1ctrl and PRC1cKO A undiff at 2 mo with or without Win18,446 treatment, as well as in A undiff and A diff of WT. *, FDR < 0.05.( ) Violin plots the expression levels of WT spermatogonia DEG groups detected in Figure 1 C in PRC1ctrl and PRC1cKO A at 2 Two-tailed unpaired t -tests.

Figure 5 .
Figure 5. R emo v al of PR C1 impairs the global deposition of H3K27me3 in A undiff genomes.( A ) Heatmaps and a v erage t ag densit y plots showing H3K27me3 and H3K4me3 enrichment at promoter regions (TSS ± 3 kb) in PRC1ctrl and PRC1cKO A undiff at 2 mo with or without Win18,446 treatment.The bars below the heatmaps represent signal intensity, and the numbers represent spike-in normalized reads counts.( B ) Track views of the Kit , Stra8 and Tcea3 gene loci showing H3K27me3 and H3K4me3 CUT&RUN peaks in A undiff of indicated genot ypes.Dat a ranges are shown in the upper right.( C ) Scatter plot showing comparison of H3K27me3 enrichment at promoters (TSS ± 2 kb) between PRC1ctrl and PRC1cKO A undiff at 2 mo.Red dots represent genes with differential H3K27me3 enrichment (FDR < 0.05).( D ) Overlap between H3K27me3-lost promoters in PRC1cKO A undiff and H3K27me3-lost promoters in WT A diff .**** P = 2.8 × 10 −1207 , two-sided hypergeometric test.( E ) Key GO enrichments in the overlapping genes detected in (D).

Figure 6 .
Figure 6.Polycomb shields the A undiff -specific developmental program in adult spermatogonia.( A ) Heatmap showing RNA-seq expression (log 2 -transformed TPM) of corresponding Polycomb-targets groups.( B and C ) Heatmaps and average tag density plots of H2AK119ub, H3K27me3 and H3K4me3 enrichment at gene regions (TSS-TES ± 2 kb) of different Polycomb-target groups in WT A undiff and A diff (B) or PRC1 A undiff of indicated genotypes (C).The bars below the heatmaps represent signal intensity, the numbers represent spike-in normalized reads counts.( D ) Box-and-whisker plots showing H2AK119ub and H3K27me3 enrichment at promoters (TSS ± 2 kb, log 2 -transformed FPKM) in the corresponding gene groups shown in (B) and (C).Central bars represent medians, the bo x es encompass 50% of the data points, and the whiskers indicate 90% of the data points.n.s, not significant, **** P < 4.1 × 10 −6 ; t wo-t ailed unpaired t -tests.( E ) Violin plots showing the expression level of different Polycomb-targets groups in 2 mo A undiff of PRC1ctrl and PRC1cKO.

Figure 7 .
Figure 7. Models of functions in adult A undiff and spermatogonial differentiation.( A ) Summary of cell cycle regulation by PRC1 and PRC2.( B -D ) Models of Polycomb functions that shield the undifferentiated state of A undiff (B), regulate spermatogonial differentiation (C) and defective spermatogonial differentiation in the absence of PRC1 (D).