m6A reader protein YTHDF2 regulates spermatogenesis by timely clearance of phase‐specific transcripts

Abstract Objectives Accumulating evidences show that the regulatory network of m6A modification is essential for mammalian spermatogenesis. However, as an m6A reader, the roles of YTHDF2 remain enigmatic due to the lack of a proper model. Here, we employed the germ cell conditional knockout mouse model and explored the function of YTHDF2 in spermatogenesis. Materials and methods Ythdf2 germ cell conditional knockout mice were obtained by crossing Ythdf2‐floxed mice with Vasa‐Cre and Stra8‐Cre mice. Haematoxylin and eosin (HE) staining, immunofluorescent staining and Western blotting were used for phenotyping. CASA, IVF and ICSI were applied for sperm function analysis. RNA‐seq, YTHDF2‐RIP‐seq and quantitative real‐time PCR were used to explore transcriptome changes and molecular mechanism analysis. Results Our results showed that YTHDF2 was highly expressed in spermatogenic cells. The germ cell conditional knockout males were sterile, and their sperm displayed malformation, impaired motility, and lost fertilization ability. During differentiated spermatogonia transiting to pachytene spermatocyte, most m6A‐modified YTHDF2 targets that were degraded in control germ cells persisted in pachytene spermatocytes of Ythdf2‐vKO mice. These delayed mRNAs were mainly enriched in pathways related to the regulation of transcription, and disturbed the transcriptome of round spermatid and elongated spermatid subsequently. Conclusion Our data demonstrate that YTHDF2 facilitates the timely turnover of phase‐specific transcripts to ensure the proper progression of spermatogenesis, which highlights a critical role of YTHDF2 in spermatogenesis.


| INTRODUC TI ON
N6-methyladenosine (m 6 A), the most abundant chemical modification in mRNA, plays crucial roles in multiple biological processes. 1,2 The mRNA m 6 A modification is dynamic and reversible. m 6 A is mainly installed by METTL3 and METTL14 core complex, together with several other key components, including WTAP, VIRMA and ZC3H13, [3][4][5][6] and this modification can also be removed by the 'eraser' proteins ALKBH5 and FTO. 7,8 Importantly, m 6 A exerts its regulatory functions primarily through various 'readers', including YTH domain-containing proteins (YTHDF1, YTHDF2, YTHDF3, YTHDC1 and YTHDC2), eIF3, HnRNP and IGF2BP2. [9][10][11][12] In different biological contexts, the reader proteins specifically decipher the m 6 Adecorated RNAs by recruiting distinct effectors, which are directly involved in regulating RNA stability, translation, splicing and subsequently determining the fates of m 6 A-containing transcripts. 9,[13][14][15][16] Accumulating evidences showed the regulatory network of m 6 A modification is essential for mammalian spermatogenesis. For instance, germ cell-specific inactivation of METTL3 or METTL14 induces progressive loss of spermatogonial stem cells (SSC). 17,18 Depletion of ALKBH5 results in abnormal spermatogenesis, due to aberrant metabolism of m 6 A-marked mRNAs. 7,19 Previous reports demonstrated that different 'reader' proteins are also critical for spermatogenesis by recognizing m 6 A marks on specific mRNAs and mediating various processes of RNA metabolism. Among the m 6 A reader proteins, YTHDC1 was uncovered to be essential for SSC survival, 20 and YTHDC2, proved to be an m 6 A 'reader' by our previous research, 21 targets meiosis-related genes and its loss induces defects in meiosis. [21][22][23][24] Besides YTHDC1 and YTHDC2, the abundant expression of YTHDF1, YTHDF2 and YTHDF3 can also be detected in testis, implying that these genes may be functional during spermatogenesis. However, ablation of Ythdf1 or Ythdf3 does not cause noticeable defects in spermatogenesis. 25 As to Ythdf2, its exact roles are controversial in different studies. An initial study claimed that knockout of Ythdf2 does not affect the fertility of male mice, 26 but a more recent study showed that deletion of Ythdf2 leads to male hypofertility. 25 To clarify the biological roles and underlying mechanism of YTHDF2 during spermatogenesis, in this study, we generated a conditional knockout mouse model with a deletion of the critical domain of YTHDF2, and revealed that YTHDF2 is indispensable for spermatogenesis and deletion of Ythdf2 in male germ cells resulted in sterility phenotype. Mechanistically, YTHDF2 facilitated the timely turnover of phase-specific transcripts to ensure the proper progression of spermatogenesis.

| Ythdf2 cKO mouse generation
Ythdf2-floxed mice were generated before by our laboratory. 27,28 We generated germ cell-specific knockout mice by crossing Ythdf2floxed mice with Vasa-Cre or Stra8-Cre mice, 18,29 respectively. All mice used in this study were kept at C57BL/6 genetic background, and housed under specific pathogen-free (SPF) conditions.

| Histological and immunofluorescent staining analysis
For haematoxylin and eosin (H&E) staining, testes were fixed in modified Davidson's fluid (MDF), dehydrated with increasing concentration of ethanol, embedded in paraffin and cut into 5μm-thick sections. Then, the sections were deparaffinized, rehydrated and stained with haematoxylin and eosin.
For immunofluorescent staining, the sections were deparaffinized, rehydrated and boiled in sodium citrate buffer (Maxim, MVS-0066) for 10 min. After washing three times by 0.1% Triton ×-100 in PBS, the sections were blocked with 5% BSA and then incubated with primary antibody overnight at 4°C. Primary antibodies used in this study were as follows: anit-YTHDF2 (Abcam, ab220163), anti-SYCP3 (Abcam, ab15093), and anti-γH2AX (Abcam, ab26350). On the next day, the slides were washed three times in PBST (0.1% Tween-20 in PBS) and incubated with the secondary antibody and related to the regulation of transcription, and disturbed the transcriptome of round spermatid and elongated spermatid subsequently.

Conclusion:
Our data demonstrate that YTHDF2 facilitates the timely turnover of phase-specific transcripts to ensure the proper progression of spermatogenesis, which highlights a critical role of YTHDF2 in spermatogenesis.

| TUNEL assay
TUNEL assay was performed with TUNEL BrightRed Apoptosis Detection Kit (Vazyme, A113-01) according to the manufacturer's instructions. Briefly, the paraffin sections were deparaffinized, rehydrated and incubated with PBS containing Proteinase K at room temperature for 20 min. After washing two times by PBS, the sections were incubated with equilibration buffer at room temperature for 30 min, and TdT buffer at 37°C for 1 h. Then, the sections were washed three times by PBS and incubated with DAPI at room temperature for 5 min. The slides finally were washed by PBS and mounted with 50% glycerol.

| Computer-assisted sperm analysis
Sperm was collected from cauda epididymides and incubated with HTF medium (Irvine Scientific, 90125) for 5 min at 37°C. 10 μl of sperm suspension was placed in slide chamber and analysed by IVOS II system (Hamilton Thorne) with default parameters.

| In vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI)
For IVF, sperm was harvested from cauda epididymides of control and vKO mice aged 10-20 weeks and incubated in HTF medium for 1 h.
Cumulus-oocyte complexes (COCs) were collected from oviduct ampullae of wild-type mice and incubated in HTF drops. Then, the sperm was added to HTF drops containing COCs and incubated at 37°C for 5 h. Finally, the presumptive zygotes were washed several times until the cumulus cells and excess sperm were removed. Then, the zygotes were transferred into KSOM medium (Millipore, MR-020P-D) and cultured in a humidified atmosphere at 37°C with 5% CO 2 .
For ICSI, sperm was isolated from cauda epididymides of control and vKO mice aged 10-20 weeks and incubated in HTF medium for 1 h. COCs were isolated from oviduct ampullae of wild-type mice, followed by removing the cumulus cells in medium containing 0.5 mg/ml hyaluronidase (Sigma, H3506) at 37°C. Then, the dispersed sperm was injected into the MII oocyte by a microinjector.

| Western blot analysis
Total protein lysates were extracted with RIPA lysis buffer. Then, the protein samples were loaded and separated in a SDS-PAGE gel, transferred onto the PVDF membranes (Millipore, IPVH00010). The membranes were blocked with 5% skimmed milk for 2 h, and then incubated with primary antibodies overnight at 4°C. On the next day, the membranes were washed four times using TBST and incubated with secondary antibody at room temperature for 2 h. Primary antibodies were as follows: anti-GAPDH (Santa Cruz, sc-32233) and anti-YTHDF2 (Abcam, ab220163).

| Isolation of spermatogenic cells
The STA-PUT method was used to isolate pachytene spermatocyte, round spermatid and elongated spermatid as described previously. 30 Briefly, mouse testes were decapsulated and digested with collagenase IV (1 mg/ml) until the seminiferous tubules were

| RNA extraction and qRT-PCR
Total RNA was extracted using RNAiso Plus (TaKaRa, #9109) according to the manufacturer's instructions. cDNA synthesis was carried out using the PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa, RR047B). Real-time RT-PCR analysis was performed using AceQ qPCR SYBR Green Master Mix (Vazyme, Q141-03). Primer sequences are listed in Table S1.  Table S2.

| YTHDF2-RIP-seq and data analysis
Whole testes or purified germ cells were lysed with lysis buffer  Table S3.
The intersection of the following four data sets was identified as delayed-decay RNAs upon YTHDF2 depletion: (1) the significant down-regulated genes (adjusted p values <0.05 and fold change >1.5) between two development stages of wild-type germ cells (such as SG to P period or P to RS period); (2) the m6A-modified genes in wild-type germ cells of previous stage (these data were acquired from the published data 18 ); (3) the YTHDF2-targeted genes; and (4) the genes with an up-regulated tendency between two stages in Ythdf2-null mice (fold change >1.2).

| YTHDF2 in germ cells is essential for male fertility
Considering that Ythdf2 is highly and dynamically expressed in male germ cells 29 ( Figure S1A), we postulated that YTHDF2 likely plays functional roles during spermatogenesis. Immunofluorescent staining showed that YTHDF2 was mainly located in the cytoplasm of germ cells and its protein level was highest in spermatocytes, followed by spermatogonia, and lowest in haploid germ cells ( Figure 1A).
To determine the function of YTHDF2 in spermatogenesis, we gen-  Figure 1C,D). In addition, the onset of high expression of YTHDF2 in P18 testis also indicated that it was mostly enriched in spermatocytes ( Figure 1C).
We then sought to investigate the fertility of Ythdf2-vKO male mice by mating them with wild-type females. Although copulation plugs could be observed normally, no offspring of Ythdf2-vKO male mice was born ( Figure 1E), suggesting that knockout of Ythdf2 in germ cells caused loss of male fertility.

| Ythdf2 deficiency leads to sperm defects and loss of fertilization capacity
To examine the male infertility of Ythdf2-vKO mice in detail, we first analysed the gross morphology of testes. No noticeable difference was observed in testis size and weight between Ythdf2-vKO and control mice (Figure 2A,B). Then, we performed computer-assisted sperm analysis (CASA) and found that the count of epididymal sperm significantly reduced in Ythdf2-vKO mice ( Figure 2C). CASA result showed that the motility and progressive motility of Ythdf2-vKO sperm were markedly impaired ( Figure 2D,E). Histological analysis of sperm collected from cauda epididymides indicated that more than 40% of sperms were deformed with various structural abnormalities ( Figure 2F,G and Figure S2A). TUNEL assay revealed that the apoptotic cells increased significantly in Ythdf2-vKO testes compared with the control (Figure 2H,I), and the apoptosis signal was mainly detected in elongated spermatids, suggesting that these sperm cells may be retained and phagocytosed by Sertoli cells ( Figure 2H).
To investigate whether the Ythdf2-vKO sperm is functionally normal in fertilization, in vitro fertilization (IVF) was carried out and these sperm failed to fertilize the wild-type oocytes ( Figure S2B).
Interestingly, intracytoplasmic sperm injection (ICSI) of Ythdf2-vKO sperm showed that fertilized eggs were able to develop to blastocytes normally, in spite of lower fertilization rates of Ythdf2-null sperm than that of wild-type sperm ( Figure S2C, D). These results suggested that Ythdf2 deficiency impaired the natural fertilization ability of sperm.
Since YTHDF2 is highly expressed in spermatogonia and spermatocytes, we reasoned that Ythdf2-vKO mice might display abnormalities in the early stage of spermatogenesis. Unexpectedly, Ythdf2-vKO mice showed normal spermatogenesis cycle ( Figure S3A).  Ythdf2-vKO and control mice ( Figure S4A, B), implying that YTHDF2 is unlikely to be essential for spermatogonia. To further narrow down the exact stage when YTHDF2 regulates spermatogenesis, we crossed Ythdf2-floxed mice with Stra8-Cre mice, which induces recombination starting from type A1 spermatogonia, and Data are presented as means ± SD (B, n = 4 for each group; C-E, G and I, n = 3 for each group). Significance was calculated with unpaired two-tailed Student's t-test (n.s., not significant, * p < 0.05, ** p < 0.01) obtained a Ythdf2-null model before meiosis initiation (Ythdf2-sKO).
As expected, immunofluorescent staining result demonstrated that YTHDF2 was still readily detected in spermatogonia, while it disappeared thereafter in Ythdf2-sKO testes ( Figure S5A). Similar to Ythdf2-vKO mice, although the overall morphology of Ythdf2-sKO testis was comparable with the control (Figure S5B, C), the sperm count, motility and progressive motility in Ythdf2-sKO were all substantially decreased ( Figure S5D-F), along with the increased number of deformed sperm and loss of fertilizing capacity ( Figure S5G-I).
The TUNEL result also revealed that the apoptotic sperm in the testes of Ythdf2-sKO mice increased significantly ( Figure S5J).
Taken together, these results demonstrate that Ythdf2 deficiency apparently does not affect the early stage of spermatogenesis, but leads to sperm defects and loss of fertilization capacity.

| YTHDF2 depletion in germ cells disturbs transcriptome during spermatogenesis
To explore the underlying mechanism of sperm defects upon YTHDF2 depletion, we systematically illuminated genome-wide gene expression changes induced by Ythdf2 deletion in germ cells.  Figure 3E). Together, the transcriptome of Ythdf2-vKO germ cells changed mainly thereafter pachytene stage, and the DEGs were coincident with the sperm malformation, impaired motility and loss of fertilization capacity in Ythdf2-vKO mice.

| YTHDF2 is required for m 6 A-modified mRNA clearance
Previous studies have shown that YTHDF2 preferentially recognizes m 6 A-modified mRNA and mediates their rapid degradation in many biological contexts. 27,[42][43][44][45] Based on the expression pattern of YTHDF2 and the phenotype of Ythdf2-vKO mice, we fo-  Figure 4B and Figure S8A). Remarkably, GO analysis showed that these delayed mRNAs were mainly enriched in pathways related to the regulation of transcription ( Figure 4C), and the transcription-associated genes showed a higher level in Ythdf2-vKO than that in control mice ( Figure 4D). Among the aberrant degradation of transcriptionassociated genes, such as Brwd1, Jarid2, Egr1 and Tsc22d3, all of these genes were m6A-modified and YTHDF2-targeted, and they could be independently validated by real-time qPCR (Figure 4E-G).
Interestingly, BRWD1 contains a Bromodomain and is closely related to chromatin remodelling 46 ; JARID2 is a transcription repressor that recruits Polycomb repressive complex 2 (PRC2) to regulate gene expression. 47 The delayed degradation of these transcriptionassociated genes may cause abnormal transcription in the subsequent stages via a transcriptional cascade effect. 48 Indeed, from pachytene spermatocyte to elongated spermatid, we observed a 10.8-fold increase (286 to 3365) in the number of DEGs ( Figure 3B).
In the transition from pachytene spermatocytes to round spermatid, 1481 m 6 A-modified YTHDF2 targets were degraded ( Figure S8B), and 389 of them showed delayed decay in Ythdf2-vKO round spermatid ( Figure S8C, D). GO analysis showed that most of these delayed mRNAs were engaged in cell cycle and transcription ( Figure S8E). Although YTHDF2 is more likely involved in this process Data of (E-G) are presented as means ± SEM (n = 3 for each group). Significance was calculated with unpaired two-tailed Student's t-test (n.s., not significant, *p < 0.05, **p < 0.01, ***p < 0.001) based on its expression pattern and the phenotype of Ythdf2-vKO mice, we cannot rule out the possibility that these alterations were caused indirectly, because a large number of transcription-related factors have been abnormally expressed in the previous stage.
Considering that YTHDF2 is still low expressed in round spermatid, we intended to explore whether the DEGs in round spermatid are directly regulated by YTHDF2 via m 6 A modification. Only 8.27% (79/955) down-regulated genes and 4.48% (36/804) up-regulated genes were both YTHDF2-targeted and m6A-modified ( Figure S9A, B), and these genes had no obvious relationship with round spermatid development, suggesting YTHDF2 may have no direct effect on round spermatid development.
In summary, the improper accumulation of mRNA induced by YTHDF2 depletion may be the cause of abnormalities in the later spermatogenesis stages, ultimately leading to male sterility.

| DISCUSS ION
m 6 A is the most prevalent modification of eukaryotic mRNA. 49 It has been reported to be involved in almost every aspect of RNA metabolism, including mRNA splicing, transport, stability and translation. 9,13,50 An increasing body of studies have demonstrated that various biological processes are tightly regulated by m 6 A modification, such as haematopoiesis, neurodevelopment and spermatogenesis. 18,19,51,52 Spermatogenesis is a complex and multi-stage biological process involving self-renewal and differentiation of spermatogonia, meiosis and spermiogenesis. 53,54 During spermatogenesis, the germ cells undergo dramatic changes in transcriptomic, epigenomic, proteomic and metabolic levels, with the biogenesis and removal of various biomolecules in distinct stages. 55 Single-cell RNA-seq has shown that spermatogenic cells of different stages have their own unique transcriptomic characteristics, which are sophisticatedly regulated. 29 YTHDF2, an m 6 A binding protein, is reported to be critical for mediating mRNA decay. 14,45 By deleting exon 2 of Ythdf2, an initial study found these conventional knockout males on a mixed genetic background possess normal reproductive capability, 26 while in a later study, the authors used CRISRP/Cas9 to delete exon 4 of Ythdf2, and found that the surviving KO males on a mixed genetic background are subfertile. 25 These findings indicate that different targeting strategies and genetic background might lead to distinct outcomes. In this study, we generated the germ cell-specific knockout mice on a C57BL/6 genetic background to dissect the exact role of YTHDF2 during spermatogenesis, and uncovered that the global transcriptional dysregulation may be the major cause that contributes to sperm deformity and infertility upon Ythdf2 deletion. Mechanistically, we demonstrated that the proper and timely mRNA clearance mediated by YTHDF2 is essential for spermatogenesis.
As m 6 A 'reader' proteins, YTH family members have been shown to be crucial for mRNA processing, stability and translation.
Interestingly, their expressions are quite dynamic in germ cells.
During spermatogenesis, Ythdc1 deletion results in progressive loss of spermatogonia. 20 Ythdc2 KO mice are infertile and exhibit meiotic arrest due to inefficient translation and degradation of YTHDC2targeted mRNA. 21,22 In our study, we showed that Ythdf2 deletion has no effect on the self-renew and differentiation of spermatogonia, but it is indispensable in later developmental stages of spermatogenesis. Our results showed that YTHDF2 is mainly expressed in spermatogonia and spermatocytes, and its expression is decreased rapidly in the later stages. We collected different stages of germ cells to perform RNA-seq analysis, which showed that the dysregulated genes are significantly increased at spermatid. Given YTHDF2 expression is very low in spermatid, most likely, the transcriptomic Our analysis of YTHDF2-RIP-seq at different developmental stages indicated that some of the above-mentioned transcripts are also YTHDF2 targets at different stages, such as Brwd1 and Jarid2. They are all expressed at both spermatogonia and pachytene spermatocyte, meanwhile, they are also targeted by YTHDF2 at both stages.
It raises an open question concerning how YTHDF2 is switched on at the right time and place. An attractive hypothesis is that YTHDF2 may accelerate mRNA clearance only when it becomes redundant, which deserves further study.

ACK N OWLED G M ENTS
We would like to thank Dr. Ming-Han Tong at the CAS Center for Excellence in Molecular Cell Science, Chinese Academy of

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.