The arginine methyltransferase Prmt1 coordinates the germline arginine methylome essential for spermatogonial homeostasis and male fertility

Abstract Arginine methylation, catalyzed by the protein arginine methyltransferases (PRMTs), is a common post-translational protein modification (PTM) that is engaged in a plethora of biological events. However, little is known about how the methylarginine-directed signaling functions in germline development. In this study, we discover that Prmt1 is predominantly distributed in the nuclei of spermatogonia but weakly in the spermatocytes throughout mouse spermatogenesis. By exploiting a combination of three Cre-mediated Prmt1 knockout mouse lines, we unravel that Prmt1 is essential for spermatogonial establishment and maintenance, and that Prmt1-catalyzed asymmetric methylarginine coordinates inherent transcriptional homeostasis within spermatogonial cells. In conjunction with high-throughput CUT&Tag profiling and modified mini-bulk Smart-seq2 analyses, we unveil that the Prmt1-deposited H4R3me2a mark is permissively enriched at promoter and exon/intron regions, and sculpts a distinctive transcriptomic landscape as well as the alternative splicing pattern, in the mouse spermatogonia. Collectively, our study provides the genetic and mechanistic evidence that connects the Prmt1-deposited methylarginine signaling to the establishment and maintenance of a high-fidelity transcriptomic identity in orchestrating spermatogonial development in the mammalian germline.


INTRODUCTION
In mammals, spermatogenesis is a spatiotemporally organized and complex developmental process that takes place in the epithelium of seminiferous tubules in the testis.In general, this process can be divided into three successi v e processes, including (i) mitotic proliferation of spermatogonial stem cells (SSCs) and progenitor spermatogonia, and subsequent dif ferentia tion giving rise to dif ferentia ted spermatogonia ( 1 ); (ii) meiotic division with one-time duplication of chromosomes followed by two times of nuclear divisions ( 2 ) and (iii) spermiogenesis, in which the haploid round spermatids undergo morphological transformation and nuclear condensation ( 3 ).The production of competent sperm necessitates rigorous gene expr ession r egulation at both the transcriptional and post-transcriptional le v els.A wide range of post-transcriptional RNA modifications and histone post-transla tional modifica tions (PTMs) was present in sper matogonia, sper matocytes and spermatids in a stage-specifically regulated manner.For instance, the most abundant N 6 -methyladenosine (m 6 A) modification in mRNA deposited by the Mettl3 / 14 complex is functionally important for the translation of m 6 A-carrying mRNAs required for SSC proliferation and spermatogonial differentia tion.Integra ti v e m ultiparametric anal yses re v ealed that a variety of histone lysine modification marks (e.g.histone H3 lysine 4 methylation, lysine 27 methylation and acetylation) occur in sperma togonia tha t epigenetically distinguish their sub-stages throughout spermatogonial de v elopment in developing mouse testes (4)(5)(6).
Arginine methylation is such one common PTM that is abundantly present in both histone and non-histone proteins.The initial high-performance liquid chromatography (HPLC)-based quantification estimated that ∼0.5-2% of total arginine residues are methylated in mammalian cells and tissues.A recent global profiling by label-free NMR spectr oscopy, independent of chr omato gra phy separation, has identified the methylated arginine proportion ranging from 1% to 3.4% in human cell lines ( 7 ).Arginine methylation is catalyzed by the family of protein arginine methyltr ansfer ases (PRMTs), which deposit three types (I ∼III) of methyl marks on arginine residues, including -N G , N G asymmetric dimethylarginine (ADMA), -N G , N G -symmetric dimethylarginine (SDMA) and -N Gmonomethylarginine (MMA), and accordingly, they belong to the Type I (PRMT1 / 2 / 3 / 4 / 6 / 8), Type II (PRMT5 / 9) and Type III (PRMT7) enzymes, respecti v ely ( 8 ).Of note, all three types of PRMTs could deposit the MMA mark on substrates and the MMA modification is generally regarded as an intermedia te sta te prior to further addition of a methyl mark by Type I or Type II PRMT enzymes, which thereby converts MMA to ADMA and SDMA, respecti v ely ( 9 ).Most PRMTs prefer to methylate glycine-and arginine-rich, consensus peptides, known as so-called GAR (R G / R GG / RX G) motifs ( 10 ).These motifs were found in both histone and non-histone protein substrates that support both ADMA and SDMA mar ks.Note worthily, PRMT4, also known as CARM1, displays unique substrate specificity within proline-, glycine-, and methionine-rich regions (termed the PGM motif) ( 11 ).On the other hand, PRMT5 can symmetrically dimethylate arginine residues within either GAR or PGM motifs ( 9 ), suggesting an intricate cross-talk of substrate methylation among three different types of PRMTs.Howe v er, owing to the presence of multiple neighboring arginine residues within the GAR or PGM motifs, it is currently unclear to what extent for the different types of methyl marks (MMA, ADMA and SDMA) to functionally compete with each other.
Mechanistically, the methyl mar ks serv e as 'docking sites' for recognition by 'reader' proteins, which specifically recognize this mark and, upon recognition / interaction, activate the downstream signaling directly or indirectly through the recruitment of protein readers / partners.Except for PRMT8, which is restricted to neuronal cells ( 12 ), most PRMTs are ubiquitously expressed in mammalian tissues.To date, arginine methylation has been widely implicated in a broad range of biological processes involving cell growth, prolifera tion, and dif ferentia tion, and the deficient methylarginine signaling is closely linked to cellular transformation or de v elopmental deficiency.For instance, we hav e shown that Carm1 is highly enriched in the nuclei of haploid spermatids in the mouse testis, and its loss leads to se v ere defects in the elongation and condensation of spermatids during spermiogenesis, resulting in male infertility ( 13 ).Prmt5 is abundantly expressed in the embryonic and postnatal germline, and deposits the symmetrical dimethylation of H4R3 (H4R3me2s), which is critical for the de v elopment of primordial germ cells (PGCs) and postnatal meiotic divisions of spermatocytes by maintaining the genome integrity likely through the Piwi-piRNA pathway in the mouse testis ( 14 ).Prmt7 is a unique member of the PRMT family in that it is solely responsible for the arginine monomethyla tion ( 15 ).W hile it is not r equir ed for postnatal germ cell de v elopment, it is crucial to pr omote germ cell pr oliferation in the embryonic testis through the BMP and TGF-␤ signaling pathway, and is implicated in paternal genomic imprinting.Prmt1 ( 16 ) and Prmt6 ( 17) hav e pre viously been implicated in spermatogenesis and male fertility.The mice with global loss of Prmt3 ( 18 ) or Prmt6 ( 19 ) showed normal gross morpholo gy, impl ying that Prmt3 / 6 are compatible with germ cell survival / development, or there might be redundant roles from other PRMT members.
In mammalian cells, Prmt1 is the primary Type I methyltr ansfer ase responsible for ADMA deposition.Herein, we showed that the Prmt1 protein is abundantly expressed in postnatal mouse testicular germ cells throughout spermatogenesis, with the highest le v els enriched in the nuclei of spermatogonia.The germline-specific Stra8-cre mediated inactivation of Prmt1 caused only the death of meiotic spermatocytes without any detectable defects in spermatogonia owing to the inefficiency of two Loxp-cassette allele excisions by germline Cre.By using two tamoxifen-inducible Cre lines [Ubc-CreERT2 ; Prmt1 lo xp / lo x (Prmt1-uKO), and Ddx4-CreERT2; Prmt1 lo xp / lo x (Prmt1-dKO)], we unexpectedly found that the substrate methylation le v els for ADMA, SDMA and MMA were to gether markedl y elevated in germ cells, which is different from those observed in somatic cells.Transcriptome-wide and CUT&Tag epigenomic profiling for fiv e histone methylarginine mar ks uncov ered that the pre-defined H4R3me2a by Prmt1 predominantly occupies both the gene promoter and intronic / exonic regions, which or chestrates the expr ession of a large number of splicing and splicing-related factors as well as maintains local chromatin signa ture, in sperma to gonial cells.To gether, our study reveals an essential role of methylarginine in coordinating spermatogonial de v elopment, and provides mechanistic e vidence that links the methylarginine signaling to shaping the spermatogonial transcriptome indispensable for male fertility.

Animal use and care
All of the mice utilized in this study were in the C57BL / 6J genetic background.To study the effect of Prmt1 abrogation on germ cells, we generated floxed Prmt1 ( Prmt1 lo x / lo x ) alleles (GemPharmatech Co., Ltd).Based on the Ensemble database, the Prmt1 locus encodes 14 transcripts in total.Exons 4 and 5 are conserved across all transcripts and were thus chosen for Loxp cassette targeting through the CRISPR / Cas9 strategy.Briefly, a pair of sgRNAs were transcribed in vitro and subsequently injected into mouse zygotes along with a donor vector to generate F0 chimera offspring.The loxp-positi v e line was confirmed via PCR genotyping and crossed with C57BL / 6J mice to attain the F1 generation of Prmt1 lox allele.The Stra8-GFPCre knock-in (KI) mouse strain (gift from Ming-Han Tong's lab at the Chinese Academy of Sciences, Shanghai), the Zp3-Cre line (purchased from Jackson Laboratory) and the Ubc-CreERT2 KI mice (purchased from GemPharmatech Co., Ltd) were crossed with Prmt1 lo x / lo x to attain Prmt1-sKO, Prmt1-zKO and Prmt1-uKO progenies, respecti v ely.To bypass the embryonic lethality and to manipulate the gene deletion in a stage-dependent manner, the Ddx4-CreERT2 KI mouse strain ( C57BL / 6J-Ddx4 tm1(5 ×HA-P2A-EGFP-T2A-CreERT2)Bao ), wherein the Ddx4 promoter dri v es the e xpression of C-terminally 5 × HAtagged Ddx4 fusion protein, along with EGFP and CreERT2 fusion enzyme in the germline, sim ultaneousl y, was generated and characterized in-house (manuscript in preparation).Upon the manual injection of tamoxifen (Tam), the released CreERT2 fusion protein enters the nucleus and excises the gene fragment flanked by a pair of Loxp cassettes, leading to the production of inducible Pr mt1 KO (Pr mt1-dKO).All mice were bred in a 12h light / dar k cy cle and with free access to food and water in a specific pathogen-free facility.All animal experiments were carried out in compliance with and authorized by the Animal Care and Use Committee of the Uni v ersity of Science and Technology of China (USTC).

Fertility test
Starting from 8 weeks of age, the fertility of wild-type (WT) and Prmt1-cKO male mice housed with fertility-proven adult WT females was examined as previously described.One male mouse was allowed to breed with at most two females in each cage.The pregnant females that had been plugged were separated and placed in isolated cages.The fertility test was performed for at least 6 months for each breeding pair.

Tamoxifen injection scheme for inducible Cre-driver lines
For in vivo gene deletion (Ubc-CreERT2 and Ddx4-CreERT2) experiments, mice (Ubc-CreERT2-Prmt lo x / lo x and Ddx4-CreERT2-Prmt lo x / lo x ) at various time intervals were injected with tamoxifen intra peritoneall y (75mg / kg body weight).Tamoxifen (Sigma-Aldrich) was dissolved in corn oil at a concentration of 20 mg / ml by sonication at 60 • C for 10 min or by shaking at 37 • C overnight with protection by aluminum foil from the light, and stored at 4 • C over the period of injection ( −20 • C for long-term storage).By tracking the Cr e-mediated Loxp-r ecombination efficiency, tamoxifen injection was performed for three or four consecuti v e days as needed.During the time of injection and post-injection, the mice were scrutinized for adverse reactions.

Histological analysis
The mice were sacrificed by cervical dislocation after isoflurane inhalation in a closed chamber.Se v enty percent ethanol was spr ay ed on the ventral abdomen, and a Vshaped opening was made in the abdominopelvic cavity using sterile scissors and forceps.The epidid ymal fa t was pulled out to locate the testes, dissected by using scissors and placed on a petri dish containing phospha te-buf fered saline (PBS).The testes were isolated and fixed in Bouin's solution overnight on a rotator.The tissue was carefully positioned and embedded in paraffin and was subsequently sectioned into 5 m thick slices using a Leica vibratome machine.The sections were deparaffinized and rehydrated sequentially in gradient concentrations of ethanol series, followed by hematoxylin and eosin (H&E) staining.After mounting the sections with neutral resins, images were captured by using a light microscope (MShot paired with MSX2 camera).

Immunofluorescence
For immunofluorescence staining of paraffin-embedded tissues, the sections were dewaxed and rehydrated, followed by antigen retrieval through heat-induced epitope retrieval (HIER) ( 20 , 21 ).The rack of slides was placed in a vessel that contained boiled sodium citrate buffer (pH 6.0).The slides were exposed to HIER treatment for 30 min.Thereafter, the slides were subjected to immunofluorescence staining, starting by washing with 1 × PBS (5 min each for three times), followed by permeabilization (0.2% Triton X-100 in 1 × PBS) for 10 min at RT.After blocking, the sections were incubated with diluted (50% 1 × PBS and 50% blocking buffer) primary antibody at 4 • C overnight.The next day, the samples were rinsed in 1 × PBST (containing 0.1% Tween) three times, followed by a 60 min incubation at 37 • C with a secondary antibody for visualization.The slides were dried down after DAPI staining, mounted using an antifade mounting medium and images were acquired on an inverted confocal fluorescence microscope (LSM980, Zeiss).

Whole-mount immunofluorescence
Testes wer e car efull y isolated and deca psulated using finetipped scissors by cutting an incision in the tunica albuginea (fibrous sheet).The tubules were forced out with forceps by pressing, and the tunica was discarded.The tubules were pipetted up and down se v er al times, then tr ansferred to a 5 ml tube containing cold 1 × PBS and allowed to sediment on ice.During sedimentation, the supernatant consisting of the cell debris and interstitial cells was removed carefully without dislodging the tubules, and fresh ice-cold 1 × PBS was added and mixed by inversion.Subsequently, the supernatant was removed, and 4% PFA was added to the tubules for fixation for 4 hours on a rotator.After fixa tion, the fixa ti v e buffer was removed and the tubules were washed twice with 1 × PBS, followed by dehydration using a methanol series.Subsequently, the tubules were hydrated in a methanol series and washed, followed by blocking for 1 h at RT.After blocking, the tubules were incubated in diluted primary antibodies at 4 • C ov ernight.The ne xt day, the tubules were rinsed in 1 × PBST three times and incubated with secondary antibodies using the recommended dilutions for 1-2 h at RT in a black staining box.After incubation, the tubules were washed with 1 × PBST twice and spread onto a microscopic slide by draining off the excess b uffer.The tub ules were mounted using an antifade mounting medium and covered with glass coverslips.

RN A e xtraction and mRN A e xpression analysis using RT-qPCR
Total RNA was extracted from whole testes or somatic organs following standard protocols as described previously ( 22 ).In brief, the fresh tissue was homogenized in TRIzol reagent (1 ml per 50-100 mg tissue) supplemented with 3 mm RNase-free beads (Servicebio, Cat no G0203) using a homogenizer.The lysate was centrifuged and the supernatant was transferred to a fresh tube, followed by centrifugation upon chloroform addition.The upper phase (clear-aqueous) was transferred to a fresh tube and an equal volume of 70% ethanol was added.Subsequently, a kit (Qiagen-74104) protocol was followed and treated by DNase I (Invitrogen ™-18068015).The quality and yield were assessed through NanoPhotometer ® N50 (Implen) and Qubit assay (Thermo Fisher Scientific).First-strand cDNA re v erse transcription was carried out using a Prime-Script RT reagent kit (Vazyme, R302-1) with gDNA Eraser to eradicate DNA contamination.Real-time RT-PCR was performed using a Hieff ® qPCR SYBR Green Master Mix (No Rox) on a Q2000B Real-Time PCR System.Relati v e gene expression was analysed by the method of 2 − CT as described by Kenneth J. Livak ( 23 ).All qPCR primers are listed in Supplementary Table S1.Data are presented as mean ± SEM unless otherwise stated.Student's t-test was employed to examine group differences.P < 0.05 was deemed statistically significant.

Isolation of spermatogonia by fluor escence-activ ated cell sorting (FACS)
The testes from the heterozygous Ddx4-GFP-CreERT2-Pr mt1 + / lox and Pr mt1-dKO (Ddx4-GFP-CreERT2-Prmt1 lo x / lo x male mice at P7 were decapsulated in a petri dish containing Dulbecco's Modified Eagle Medium (DMEM).Tubules were dispersed using sterile forceps and digested into a single-cell suspension by collagenase and trypsin enzymes in DMEM medium.Briefly, the tubules wer e transferr ed to fr esh colla genase b uffer and incuba ted a t 37 • C f or 5 min, f ollowed by rinsing twice or three times with 1 × Krebs-Henseleit solution (Krebs) to remove interstitial somatic cells.The sedimented tubules wer e transferr ed to a 60 mm dish and delicately chopped using sterile forceps, which were further digested into single cells in 10 ml trypsin buffer by incubation at 37 • C for 10 min.The digestion was quenched by 2% FBS.
For flow sorting of spermatogonial cells, a new template for FACSCalibur ™ was documented with a density plot to set up the machine and gather data.To reduce e v ents on the axes that result in a single cell population, a dot plot with FSC (forward scatter) for the X-axis and SSC (side scatter) for the Y-axis was adjusted.The fluorophore signal from GFP-positi v e sperma togonia was ga ted to determine the fluorophore laser settings.

Modified Smart-seq2 library construction for low-input cell samples
Total RNA was extracted from a limited amount of FACS-sorted spermatogonial cells ( ∼1 × 10 4 in total for each sample) using a MACHEREY-NAGEL microkit (MACHEREY-N AGEL, Germany).RN A quality and concentration were monitored by a Qubit dsDNA assay kit (Thermo Fisher Scientific, USA) and qPCR analyses.The full-length mRNA library was pr epar ed using an inhouse optimized Smart-seq2 protocol ( 24 ).Briefly, ∼10 ng of total RNA for each sample was utilized for first-strand cDNA re v erse transcription in a 10 l RT buffer containing SuperScript III RTase (100 U), RNase inhibitor (10 U), dNTP mix (10 mM each), dCTP (10 mM), SS III firststrand buffer (1 ×), DTT (5 mM), betaine (1 M), MgCl 2 (6 mM), TSO (1 M) and oligodT30VN (10 M).The RT reaction was performed at 60 • C for 50 min.The fulllength cDNA was subsequently amplified through semisuppressi v e PCR for 15 cycles in a buffer containing 10 l of first-strand cDNA, 12.5 l KAPA HiFi HotStart ReadyMix (1 ×), ISPCR primers (0.1 M), and nucleasefree water.The amplified full-length DNA library was purified to get rid of < 500 bp fragments and other contaminants using Hieff NGS ® DNA Selection Beads (Yeasen) following the manufacturer's protocol.1 ng of purified DNA was exploited for library preparation using TruePrep ® DNA Library Prep Kit V2 for Illumina (Vazyme).The final DNA library cleanup was carried out by double-side bead selection (0.6 ×/ 0.3 ×) to extract library DNA in sizes ranging from 350 to 550 bp as measured by Bioanalyzer 2100 (Agilent), and was ultimately subjected to library sequencing by Novaseq 6000 (PE150 mode).

RNA-seq library preparation
Total RNA was extracted from whole testis using the E.Z.N.A. ® MicroElute ® Total RNA Kit following the manufactur er's protocol.DNase I-tr eated total RNA samples were subjected to pol yA mRN A isolation, followed by hea t-induced mRNA fragmenta tion prior to first-strand cDNA re v erse transcription using randomized hexamer primers.After second-strand DNA synthesis, the doublestrand DNA was end-r epair ed prior to dA-tailing, followed by Y adaptor ligation, according to standard protocols ( 24 ).

RNA-seq / modified Smart-seq2 data analysis
The raw data were subjected to quality-control using FastQC (version 0.11.9),followed by fastp (version 0.23.2) to remove the unwanted and low-quality reads.The pairedend clean reads were mapped to the mouse reference genome (mm10) using STAR (version 2.7.6a).The mapping files containing the matching reads were run through RSEM to calculate the expression values for genes and transcripts, as well as the raw read counts.Genes with averaged TPM ≥ 1 were deemed as expressed genes ( ∼16 000 genes detected per sample).The DESeq2R package was employed to analyze the differential expression differences between the contr ol gr oup and the experimental gr oup (cutoff: fold change ≥ 2; adj-P -value ≤ 0.05).GO enrichment was performed by the DAVID database ( 25 ).
For alternati v e splicing analyses, the rMATS (v ersion 4.1.2)software was run for the fastq file with the following parameters: -Read length 150 -variable-read-length.The false discovery rate (FDR) ≤0.05 was categorized as differentially alternati v e splicing e v ents.The rmats2sashimiplot was used to convert the rMATS output into a Sashimi plot.

Histone isolation
For histone isolation, we followed a standard protocol ( 26 ).In short, the tissue was transferred to a Dounce homogenizer after weighing and homogenized into a single-cell suspension (smaller clumps) in TEB buffer (1 ml / 200 mg tissue) with 50-60 strokes.The tissue pellet was resuspended in 3 volumes of acidic buffer (0.5N HCl + 10% glycerol) and incubated for 30 min on ice with rotation at 4 • C, followed by centrifuga tion a t 12 000 rpm for 5 min.The superna tant was transferred to a fresh tube and processed for Trichloroacetic acid (T CA) pr ecipitation.The histones wer e r ecover ed by centrifugation for 10 min at 12 000 rpm.After supernatant removal, the pellet was sequentially washed with ice-cold acetone containing 0.05% HCl, and three times with icecold acetone.After successi v e washing the pellet was dried at −20 • C overnight, then dissolved in water and stored at −20 • C.

Immunoblot and protein band quantification
The tissues were freshly collected in liquid nitrogen for snap freezing.Samples were homogenized in ice-cold RIPA buffer using an electric homogenizer (Servicebio-G0203).The homogenized sample was centrifuged and the supernatant was transferred to a fresh tube.Pierce ™ BCA Protein Assay Kit (Thermo Scientific ™) was used to quantify the protein concentration.Both semi-dry and wet-transfer methods were used and the membranes were blocked (5% skim milk) for 60 min at RT, followed by primary antibody (diluted in 5% milk) incubation at 4 • C overnight.The images were obtained from the membrane with the clinixscience instrument coupled with a CCD camera.When needed, the membrane was stripped in stripping buffer (pH 2.2) before probing with another antibody.ImageJ (NIH) was used to measure the intensity of protein bands for densitometric analysis.In a nutshell, the whole protein band's ROI was chosen, and integrated intensity was assessed using the set measurement menu.The integrated intensities values for the same size selection were also applied to areas below and above the specific band used as background.

Cleavage under targets & tagmentation (CUT&Tag) assay
To decipher the chromatin localization landscape of histone modification markers, cleavage under targets & tagmentation (CUT&Tag) was carried out as previously described ( 27 ).In brief, the testes from two time points were collected, including spermatogonia-enriched (P7) and spermatocytes (P14).After removing the tunica albuginea, the seminiferous tubules were digested with Collagenase IV briefly to eliminate the somatic interstitial cells.The pure tubules were digested by trypsin into homogeneous single-cell suspension, with filtering through 40 m cell strainers to eliminate Sertoli cell contamination, as described previously ( 28 , 29 ).For CUT&Tag, the single-cell capture, antibody incubation and Tn5 transposon activation were carried out according to the procedures as described in Hieff NGS ® G-Type In-Situ DNA Binding Profiling Library Prep Kit (Yeasen) following the manufacturer's instructions.For each sample, briefly, an averaged total of ∼50 000 single cells were allowed to bind the ConA beads.The primary antibody was incubated with the beads-cell suspension on a rotator at 4 • C overnight.The next day, the secondary antibody was added and incubated for 60 min at RT, followed by incubation with pA / G-Tn5 transposome and tagmenta tion in activa ting buf fer supplemented with magnesium.The ada ptor-ligated DN A fr agments were extr acted and PCR-amplified using full-length index primers.The amplified product was purified using Hieff NGS ® DNA Selection Beads (Yeasen), and the quantity and quality of the DNA library were examined through Qubit ™ 4 Fluorometer and Bioanalyzer 2100 (Invitrogen), respecti v ely.The libraries were sequenced by NovaSeq 6000 (PE150).

CUT&Tag data processing and alternative splicing analysis
The peaks were called using MACS2 (version 2.1.0).The bamCoverage command in deeptools (version 3.5.1)was used to convert the bam file to a bw file that has been normalized by CPM.CUT&Tag densities were visualized by Integrati v e Genomic vie wer (IGV).The deeptools commands multiBigwigSummary and plotCorrelation were used to calculate the sample correla tion.ComputeMa trix and plotProfile (plotHeatmap) commands were used for assessing the peak distributions of histone modifications at TSS , TES , and CGI.
To visualize CUT&Tag signal pattern for both constituti v e and skipped exons, we first acquired the coordinates (e xonStart 0base, e xonEnd) of the target skipped exons.In such case, the adjacent up and downstream exons were termed constituti v e.The surrounding region of the splice junction was extended 100 bp to the 5 and 3 ends following normalization, the CUT&Tag signal intensity was calculated, and the Student's t-test was used to compare the sta tistic dif ference for both sides of exons, as well as the regions between alternati v e and constituti v e e xons.To account for the difference in the signal distribution of histones on exons and introns, we excluded the influence of TSS by removing first exon and setting the minimum length for intron 500 bp, whereas for exon we set minimum length of 200 bp.

TUNEL assay
The TUNEL assay was performed using the Vazyme FITC apoptosis kit (A111) following the manufacturer's instructions with some minor modifications.Briefly, the sections were permeabilized using 0.3% Triton X-100.After equilibrating the sections, the TdT reaction mixture was added and incubated for 60 min a t 37 • C .Afterward, the sections were incubated for 10 min with DAPI (1:1000) for nucleus counterstaining.Following the final wash, the slides were dried with Kimwipes and mounted with antifade mounting media before storage at 4 • C in a black box until microscopic imaging.

Statistical analyses
All experiments in this study were conducted at least in biological triplicates unless otherwise stated.All statistical analyses were performed using Prism 8 (GraphPad software).Unless otherwise sta ted, sta tistical significance was determined using the Student's t -test.P values of < 0.05 were deemed statistically significant.*, **, *** and **** r epr esent P < 0.05, P < 0.01, P < 0.001 and P < 0.0001, respecti v ely.

Prmt1 is highly expressed and predominantly localized to the nuclei in spermatogonia of mouse testes
We first assessed the multi-organ mRNA expression levels of Prmt1 in mice.qPCR assays showed that the highest Prmt1 mRNA le v els were detected in mouse lung and testis (Figure 1 A).Howe v er, at the protein le v el, Prmt1 is most abundantly present in the testis compared with other somatic tissues at postnatal day 7 (P7) (spermatogonia) or P42 (when the first wave of spermatogenesis is completed) (Figure 1 B, C and Supplementary Figure S1A, B), suggesting Prmt1 is likely subject to post-transcriptional regulation.The male germline de v elopment is strictly time-dependent, i.e. the specific type / stage of germ cells proceeds at defined time points ( 30 ).We thus examined the expression of Prmt1 during postnatal testicular de v elopment.The qPCR assay showed relati v ely comparab le mRNA e xpression le v els for Prmt1 (Figure 1 D), whereas Prmt1 protein displayed higher le v els around P5 ∼P10 and decr eased appar ently ther eafter (Figure 1 E, F), presumabl y impl ying that Prmt1 mRNA is subject to post-transcriptional regulation as well in the ger mline, and that Pr mt1 is most likely expressed in spermatogonial cells, since the highest proportion of spermatogonial cells are present in the younger testis ( < P10).Therefore, we next conducted immunofluorescent staining (IF) on cryosections of mouse testes at various de v elopmental time points (Figure 1 G and Supplementary Figure S1C) as well as the immunohistochemistry (IHC) staining on paraffinembedded testicular sections (Supplementary Figure S1D).Co-staining with Sertoli cell-specific marker (Sox9) and the germline-specific Ddx4 marker, demonstrated that the Prmt1 protein is pr edominantly expr essed in the spermatogonial popula tion (undif ferentia ted & dif ferentia ted), with markedly declined levels detectable at later stages of germ cells (e.g.meiotic spermatocytes or haploid spermatids) (Supplementary Figure S1C-E).Noticeably, the Prmt1 protein is largely distributed within the nuclei of spermatogonia during postnatal germ cell de v elopment (Figure 1 G-H and Supplementary Figure S1C-E).We further evaluated Prmt1 mRNA expression at the single-cell level using published testicular single-cell RNA-seq datasets (Supplementary Figure S1F, G).In agreement with findings above, the Prmt1 mRNA le v els are highest in spermato gonia w hen compared with other types of germ cells in juvenile or adult mouse testes.Noteworthily, Prmt1 displays the highest expression levels among the PRMTs (Prmt1 ∼9) based on the single-cell RN A-seq anal yses in P15 testes (Figure 1 I), presumably implying that it has non-redundant function with other PRMT family members during germline dev elopment.Together, this e vidence validated that Prmt1 is strongly enriched in the nuclei of spermatogonial cells, but weakly in the nuclei of spermatocytes, within mouse testes.

Prmt1 is indispensable for spermatogenesis and male fertility
To study the role of Prmt1 in germline in vivo , we generated the Cre-inducible Prmt1 loxp mouse allele since global ablation of Prmt1 led to embryonic lethality ( 31 ).There is a total of 14 predicted Prmt1 transcripts (Prmt1-201 ∼214) in the Ensembl mouse database (GRCm39, GCA 000001635.9), of which the Prmt1-202 (ENSMUST00000107843.11) encodes the longest protein isoform consisting of closely juxtaposed exons 4 and 5 conserved across all predicted transcripts.Moreov er, ab lation of exons 4 and 5 presumably elicits frame-shift mutation causing aborted function of Prmt1 in vivo .Ther efor e, exons 4 and 5 were selected for Loxp cassette flanking to produce the Prmt1 loxp allele (Figure 2 A).Unlike somatic cells, germ cell de v elopment is not synchronized, but rather, is genetically programmed and spa tiotemporally regula ted.We thus exploited three Cre lines to cross with Prmt1 lo xp / lo x mice, including Stra8-Cr e (specifically activa ted in sperma togonia starting from  P3 testis), Ubc-CreERT2 (Tamoxifen [Tam]-inducible in all cell types), and Ddx4-CreERT2 , which is in-house generated, tamoxifen-inducib le specifically dri v en by Ddx4 promoter in germ cells (see Materials and Methods).These lines thereby provided powerful genetic tools to dissect the time-dependent, and the stage-specific roles of Prmt1 in the mouse germline.Through two generations of crossings between Cre lines and Prmt1 lo xp / lo x alleles, we obtained the Str a8-Cr e; Prmt1 lo xp / lo x (her eafter r eferr ed to as Prmt1-sK O), Ubc-CreERT2; Prmt1 lo xp / lo x (Prmt1-uK O, following Tam injection) and Ddx4-CreERT2; Prmt1 lo xp / lo x (Prmt1-dKO, following Tam injection), respecti v ely (Figure 2 B).
To examine the KO efficiency of Loxp cassettes, we injected Tam into both WT and Prmt1-uKO males for three consecuti v e days starting from P12, and sacrificed them at day 7 of post-tam injection.Immunoblotting assays on the testicular lysates from the seminiferous tubules with antibodies against Prmt1 or its known histone methyl substra te --H4R3me2a, verified tha t the le v els of both Prmt1 protein and H4R3me2a were significantly declined in the Prmt1-uKO testes, as compared with WT testes (Figure 2 C, D).Next, we characterized the phenotypic outcome upon Prmt1 inactivation.As seen in Figure 2 E, F, the testicular sizes in all Prmt1-sK O / uK O / dK O males apparently dropped down to by less than one-fifth of those in WT males.In agreement with this, > 5-month fertility test demonstra ted tha t all KO males were completely sterile, signifying that spermatogenesis was disrupted in Prmt1null mouse testes (Figure 2 G).Consistentl y, imm unofluorescent staining re v ealed that the signals for both Prmt1 and H4R3me2a were largely demolished from Prmt1-dKO and Prmt1-sKO testes (Figure 2 H, I).Taken together, these data suggest Prmt1 is indispensable for testicular germ cell de v elopment and male fertility.

Prmt1 modulates the protein substrate arginine methylome in the germline distinct from somatic cells
As the primary Type-I arginine methyltr ansfer ase, loss of Prmt1 activity spurred an eminent increase of global levels for both MMA and SDMA by other PRMT members ( 32 ).Since the Prmt1 protein is present in the nuclei of both the mitotic spermatogonia and meiotic spermatocytes, as shown above (Figure 1 ), we next interrogated how the substrate methylarginine cross-talks upon Prmt1 depletion in testicular germ cells.To this end, we performed the immunoblotting on the Prmt1-sKO testicular lysates collected a t P8 (domina ted by sperma togonia) and P17 (sperma tocytes) with three pan-methylarginine antibodies --MMA, ADMA and SDMA.These antibodies were generated using a mixture of short stretches of peptides containing methylated arginine residues within consensus motifs, and immune-react with the corresponding methylarginine ( 33 ).Consistent with prior studies in somatic cells, we found a dramatic increase in MMA le v els and a moderate enhancement of SDMA methylation (Figure 3 A-D), suggesting that there is a substrate cross-talk between Prmt1 and other PRMT members in testes.Howe v er, surprisingly, we discovered that the ADMA le v els were significantly up-regulated as well, especially in P8 testes.Notably, theh increased signal comprises novel bands that were not seen in the WT testes, in addition to the shared bands present in both WT and Prmt1-sKO testes, hinting that Prmt1 likely orchestrates a distincti v e networ k of substrate ADMA methylation in the germline.
To test whether the ele vated ADMA le v els ar e r estricted to testes as opposed to other primary tissues, we next compared the e xpression le v els of all three methylarginine marks (MMA, ADMA, and SDMA) between testes and a select panel of somatic organs (heart, spleen, kidney, li v er and lung) by employing the ubiquitous expression of the Ubc-Cre mouse line (Prmt1-uKO).We carried out the Tam injection for three consecuti v e days and collected the tissues on the fourth and se v enth day post-Tam injection (Supplementary Figure S2A).In line with the high le v els of Prmt1 protein in the spleen and the testis, both organs exhibited higher le v els of all three types of substrate arginine methylation than other tissues (Figure 3 E-G).Consistently, we observed that the MMA levels were most drastically upregulated upon loss of Prmt1 activity in all the testis and somatic tissues (Figure 3 E and H).The SDMA le v els were also markedly enhanced upon Prmt1 KO in testis (Figure 3 F and I).Not surprisingly, the ADMA le v els dropped significantly in the somatic organs upon Prmt1 loss (Figure 3 G and J).Ne v ertheless, in agreement with observation in Prmt1-sKO testes, the ADMA marks were unambiguously elevated in Prmt1-null testes, which is opposite to the observation in soma (Figure 3 C and J).Therefore, Prmt1 KO reshaped the global substrate methylation dynamics for all three types of methylarginine, which is in sharp contrast to the phenomenon observed in somatic organs.
Prmt1 is ubiquitously expressed in all tissues.We further found that e v en in the adult mice, global KO caused rapid mouse lethality (9-15 days following Tam injection) in the Prmt1-uKO line (Supplementary Figure S2B).To track the dynamics of substrate methylation as well as to exclude the impact of interstitial cells seen in Prmt1-uKO, we conducted a time-course measurement of all three methylarginines in the Tam-inducible Prmt1-dKO testes, at four post-Tam injection time points: P16, P18, P20 and P22 (Supplementary Figure S2C).In support of the findings above, while the substrate SDMA le v els were less strong compared with the other MMA and ADMA marks, all three methylarginine le v els were unambiguously enhanced upon tamoxifeninduced Prmt1 deletion in the Prmt1-dKO testes (Supplementary Figure S2D-G).Of note, at P22, all three types of methylation le v els wer e slightly decr eased compar ed with those at P20, which pr esumably r esulted from the loss of germ cells during long-term Tam-induced Prmt1 KO.To explore this hypothesis, we performed immunofluorescence staining with the germline-specific marker, GCNA, and counted the various types of germ cells in mouse testes.This re v ealed that, whereas there appeared to be no loss of germ cells at P16, the numbers of meiotic and haploid spermatids were significantly reduced at P22 (Supplementary Figure S2H-K), which accounted for the reduced substra te methyla tion, in Prmt1-dKO testes a t P22.Altogether, these data suggest that Prmt1 activity alter an intricate substrate cross-talk both among three methylarginine types and within Type I asymmetric dimethylation in the germline.

Prmt1 depletion ev ok es an intrinsic interplay with Prmt2, Prmt5 and Prmt6 responsible for the enhanced SDMA and ADMA methylation in the testis
The significantly increased le v els of ADMA upon loss of Prmt1 activity cannot be explained by the traditional notion that Prmt1 is primarily for ADMA deposition (Figure 3 G), which led us to speculate that there might be compensa tory ef fects a ttempted by other Type-I PRMTs when the pre-defined ADMA marks by Prmt1 are unmasked upon Prmt1 ablation.To test this hypothesis, we performed the qPCR analyses for PRMT family members using the testis and spleen tissues deri v ed from WT and Prmt1-uKO mice (Figure 4 A and Supplementary Figure S3A).While the mRNA le v els of Prmt1 were significantly down-regulated as expected, the levels for asymmetric arginine methyltransferase Prmt2 / 6 / 7, as well as the symmetric arginine methyltr ansfer ase Prmt5, were all markedly up-regulated, albeit with varied le v els, in the Prmt1-uKO testes (Figure 4 A).By comparison, in the spleen tissue, this phenomenon was not observed except for Prmt5 (Figure 4 B).Notably, Carm1 mRNA le v els were not altered in the testis, but rather decreased in the spleen.At the protein le v el, immunob lot analysis showed that the Prmt2 / 5 / 6 members were overtly upregulated except for Carm1 in the Prmt1-dKO testes during the time-course of Tam induction (Figure 4 C-H and Supplementary Figure S2C).Fluorescent immunostaining with Prmt5 re v ealed a sharp difference between Prmt1-dKO and WT testes (Supplementary Figure S3B).The dynamic expression of Prmt5 was abolished by Prmt1 loss.When Prmt1 was deleted, Prmt5 is predominantly restricted in the nuclei as opposed to that in P22-WT testes, where Prmt5 is distributed in the cytoplasm of spermatocytes and round spermatids (Supplementary Figure S3B).At the same exposure time, higher Prmt5 expression le v els were observed in the Prmt1-dKO testes (Figure 4 E, F).We also examined the expression levels of the well-known Prmt6 substrate, H3R2me2a, using acid-purified histone proteins, and found that this methyl mark was indeed prominently upregulated in the Prmt1-uKO testes (Figure 4 I, J), but not in the Prmt1-uKO spleen.In contrast, although there were slight increases for both Prmt4 (Carm1) and Prmt7 upon Prmt1 loss, statistical analyses ov erall re v ealed comparab le le v els between the WT and Prmt1-dKO testes (Supplementary Figure S3C-H).Therefore, the enhanced le v els for MMA, SDMA and ADMA in the Prmt1-null germ cells were likely achie v ed through Pr mt2, Pr mt5 and Prmt6.Inter estingly, pr evious studies demonstra ted tha t in the MEF cells, Prmt1 loss induced the compensatory elevation of Prmt6 / 7, concomitant with ADMA declining during early days of treatment, which is opposed to the phenomenon in germ cells ( 32 ).Together, these data prompt that there exists a cell-autonomous cross-talk among Type I and II enzymes (Prmt1 / 2 / 5 / 6) uniquely in germ cells, but not present in soma.

Prmt1 is r equir ed to maintain the developmental tr ajectory of the spermatogonial population through the first wave of spermatogenesis and in adult mouse testes
As aforementioned, deletion of Prmt1 caused eminent testicular atrophy and the resultant male infertility.To de-cipher at what stage the germ cells were impacted in the Prmt1-null testis during the first wave of spermatogenesis, we examined the histology of the testes from Prmt1-sKO mice by hematoxylin and eosin (H&E) staining.We collected the testes at four time points, which correspond to the occurrence of representati v e germ cell types in testis, including P7 (spermatogonia), P14 / 17 (early and late sperma tocytes), and P21 (sperma tids).We found tha t while the cellular composition and morphology resembled each other between WT and Prmt1-sKO testes at P7, the spermatocytes at P14 / 17 exhibited aberrant nuclei, reminiscent of cell dying (Figure 5 A).At P21, a large number of haploid spermatids present in WT testicular tubules were not seen in the Prmt1-sKO tubules, and the majority of spermatocytes were largely depleted.Detailed counting of the germ cells at each time point indicated that the germline development was arrested at the spermatogonial stage in the Prmt1-sKO testes (Figure 5 B).To further verify the findings from the H&E staining, we carried out the immunofluorescence staining on testicular cryosections with two antibodies against Plzf (spermatogonial stem and progenitor cell population) and Ki67 (proliferating spermatogonial cells) respecti v ely.In agr eement with the H&E r esults, at P7, the average numbers of the spermatogonial sub-populations r epr esented by either Plzf or Ki67 staining were comparable in WT and Prmt1-sKO testes (Supplementary Figure S4A-C).Howe v er, at P21, the av erage number of Ki67-positi v e cells was significantly reduced, in contrast to the similar numbers of Plzf-positi v e sperma togonial cell popula tion (Supplementary Figure S4D-F).The Ki67 marker has been generally regarded as being representati v e of the differentia ted sperma togonial cell popula tion in the mouse testis.Since the c-Kit marker for dif ferentia ted sperma togonia was not effecti v e for immunofluorescence staining, we thus conducted immunoblot to compare the expression levels of c-Kit upon Prmt1 deletion.We discovered that the le v els of c-Kit protein were evidently decreased in the Prmt1-sKO testes at both P8 and P17 days, suggesting that Prmt1 depletion resulted in deficient dif ferentia tion of spermatogonial cells (Supplementary Figure S4G).The depleted germ cells likely underwent cell apoptosis, since the TUNEL assay re v ealed a vast number of positi v ely stained germ cells in the Prmt1-sKO testes at P21(Supplementary Figure S4H-I).This evidence altogether led us to conclude that Prmt1 is required for sperma togonial dif ferentia tion and spermatocyte de v elopment.
Gi v en that we have previously shown that Stra8-Cre is not proficient in excision of two loxp alleles in spermatogonial population ( 34 ), it is thus unclear whether Prmt1 is essential for the undif ferentia ted sperma togonia, i.e. sperma togonial stem and pro genitor cells, especiall y w hen considering the highest le v els of Prmt1 protein detectab le in the spermatogonia as described before (Figure 1 ).To explore this postulation, we ne xt e xploited both Tam-inducib le Prmt1-uK O and Prmt1-dK O lines and performed the tamoxifen induction during the first wave of spermatogenesis.Due to the lethality of the Prmt1-uKO mice following tamoxifen injection, the testes were allowed for 8 days in Prmt1-uKO mice and 24 days in Prmt1-dKO prior to collection for H&E analyses after three-day tamoxifen injection (Figure 5 C, D).This unveiled that both undif ferentia ted and dif ferentia ted  sperma togonial cell popula tions were substantially eliminated from the seminiferous tubules (Figure 5 E, F).We further executed co-staining by Sox9 (Sertoli-cell-specific marker) in conjugation with Plzf or Pcna (proliferating spermatogonia) on the testicular cryosections, and uncovered that nearly the whole spermatogonial cell populations were essentially eradicated from the tubules (Figure 5 G-J).This evidence shows that Prmt1 is pivotal to orchestrate sperma togonial prolifera tion and dif ferentia tion during the first wave of spermatogenesis.
The maintenance of male fertility in the adult mammalian testis depends on the orderly cycling of spermatogonial cells between self-proliferation and differentiation prior to meiotic division.To interro gate w hether Prmt1 is critical for the maintenance of spermatogonia in the adult testis, we took advantage of the Prmt1-dKO line by administering the tamoxifen into the 8-week-old adult mice (Supplementary Figure S4J), and implemented whole-mount immunofluorescence staining with Gfr ␣1 antibody.In the WT testes, Gfr ␣1 typically marks undif ferentia ted sperma togonial population, which is characterized by the As, Apr, and Aal4 spermatogonial cells seen in whole-mount sections.We observed the elongated Pseudopod which is usually indicati v e of the acti v e stem cell population in the WT testes.By comparison, the number of the Gfr ␣1-positi v e spermatogonia and the length of elongated Pseudopod were evidently reduced in the Prmt1-dKO testes at 7-day post-Tam treatment (Supplementary Figure S4K, L).An e xhausti v e timecourse tracking of Tam-treated testes by co-staining with c-Kit and Gfr ␣1 re v ealed that the whole spermatogonial pool was essentially eradicated (Supplementary Figure S4M-P), verifying the necessity of Prmt1 in maintaining the homeostasis of spermatogonia in adult testes.Altogether, this evidence unraveled that Prmt1 is a core player r equir ed to establish and maintain the developmental trajectory of spermatogonial cells in the testis.

Prmt1 establishes the high-fidelity RNA transcriptome in spermatogonia
To interrogate how Prmt1 loss impacts the spermatogonia at the transcriptomic le v el, we ne xt conducted RNA sequencing using either whole mouse testis or purified mouse spermatogonia.The samples were collected at P7 following three consecuti v e doses of Tam injection (Figure 6 A and Supplementary Figure S5A).The spermatogonial population was sorted through Fluorescence-activated Cell Sorting (FACS) by utilizing our in-house generated Ddx4-CreERT2 KI mouse model, which carries an EGFP tag dri v en by the Ddx4 promoter.Conventional bulk RNAseq using whole mouse testis at P7 identified a large number of d ys-regula ted genes (1365 up-regulated vs 293 downregulated) in the Prmt1 KO testes upon Tam treatment (cutoff: fold change ≥ 2, adj-P -value < 0.05) (Supplementary Figure S5B, C).As shown above, loss of Prmt1 led to the spermato gonial loss, likel y causing a disproportional population of spermatogonia between whole WT and KO testes.Hence, to capture the accurate transcriptomic change, we further performed the mini-Smart-seq2 analysis, w hich was specificall y optimized for low-input cell samples ( ∼ranging from 10 000-100 000), using FACS-sorted, EGFP-positi v e spermatogonial cells from P7 testes.Consistently, this re v ealed 1231 upregulated along with 660 downregulated genes (Figure 6 B, C) in Pr mt1-dKO sper matogonia (cutoff: fold change ≥ 2, adj-P -value < 0.05).GO enrichment analysis unveiled that those d ys-regula ted genes ar e pr edominantly involv ed in cell-cy cle and meiotic regulation (Figure 6 D).Notoriously, in support of previous phenotypic defects seen in Prmt1-null spermatogonia, a handful of genes involved in spermatogonial self-renewal (Id4, Plzf, Bmi1, etc.) and dif ferentia tion (Ccnd1, C-kit, Sohlh-1, etc.) wer e markedly r educed in Pr mt1-dKO sper matogonia as v alidated b y qPCR (Figure 6 E), suggesting Prmt1 is crucial to establish and maintain the transcriptomic integrity of spermatogonia.
PRMT enzymes are known to impact alternati v e splicing through methylating the splicing factors, e.g.SmB and hnRNP members (35)(36)(37)(38).We thereby analyzed alternati v e splicing (AS) e v ents, which were di vided into fiv e categories --skipped exon (SE), Alternati v e 5 splice site (A5SS), Alternati v e 3 splice site (A3SS), m utuall y exclusi v e e xon (MXE) and retained intron (RI).Among them, the SE dominates the alternati v e splicing e v ents as detected by both whole bulk RNA-seq and mini-Smart-seq analyses (Figure 6 F and Supplementary Figure S5D, E).GO examination unraveled the AS genes are enriched in cell-cycle regulation and mRNA processing (Figure 6 G).A visual inspection of the integrati v e genomic vie wer (IGV) for AS genes with SE, such as Kdm5a, Brca1, Clk4, Srsf9, Map7, Ccna1, Srsf7 and Lsm5, was further corroborated by the semi-quantitati v e PCR validation, showing that those exons were aberrantly skipped or retained in the Prmt1-null spermatogonia (Figure 6 H-J and Supplementary Figure S5F-N).Moreover, the CENP gene family, including CENP-C, CENP-H, CENP-M, CENP-N, CENP-U and CENP-T, assembles the CENP-A NAC complex, which safeguards chromosomal alignment and segregation ( 39 ).Those members were all d ys-regula ted upon Prmt1 loss as re v ealed by Smart-seq2 examination.Among them, Cenpt is prominently enriched in spermatogonial population, and both IGV browser snapshot and qPCR inspection verified that the intron 11 is significantly retained in the Prmt1-null spermatogonia (Figure 6 K-N).Together, this evidence indicates that Prmt1 maintains the transcriptomic high-fidelity essential for spermatogonial de v elopment.

Global profiling identified the genomic distribution of histone arginine methyl marks and a direct competition between H4R3me2s and H4R3me2a at promoters in regulating gene expression
Prmt1 deposits the asymmetrical H4R3 methylation (H4R3me2a) known as an acti v e histone mark for gene e xpression.To e xplore the mechanism by which Prmt1 loss caused the transcriptomic d ys-regula tion, we a ttempted to execute the highly sensitive and efficient CUT&Tag profiling recently de v eloped in Henikoff's Lab ( 27 ).As a fact, the conventional ChIP-seq assay rarely succeeded in identifying the genome-wide distribution of histone arginine methyl marks in the past decades (as re vie wed ( 40 )).To enrich the spermatogonia in P7 testes, we removed the Sertoli cells by filtering the single-cell suspension through 40 m strainer, and determined that the ratio of germ cells to somatic cells in the filtrate exceeded 95% (Supplementary Figure S6).Next, we selected ChIP-qPCR validated antibodies against the histone methylarginine marks with compensatory increases in the Prmt1-dKO testes, including H4R3me2a for Prmt1, H3R2me2a for Pr mt6, H4R3me2s for Pr mt5, H3R8me2a for Pr mt2 and H4R3me1 for Prmt7.Meanwhile, both H3K4me3 and H3K27ac were included for comparison, which specifically delineates promoter and enhancer elements respecti v ely.This high-throughput profiling re v ealed > 14000 peaks on average for each methyl mark with high confidence (Cutoff: P < 0.005) (Supplementary Figure S7).Strikingly, all fiv e histone arginine methyl marks (H4R3me2a, H3R2me2a, H4R3me2s, H3R8me2a, H4R3me1) are highly enriched in the TSS region, albeit with much lower peak densities, compared with the H3K4me3 peak at the TSS region (Figure 7 A-C, Supplementary Figures S7A-N and S8A-I).Mor eover, ther e is a high overlapping of the TSS peaks among the fiv e methylarginine mar ks, H3K4me3 and H3K27ac (Supplementary Figure S8C).An overlapping between the H4R3me2a-enriched peaks and DEGs in Prmt1-null spermatogonia unveiled that a large number of genes, in particular, among which a total of 277 genes, ar e r egulated by the Prmt1-deposited H4R3me2a (Figure 7 F).This signified that PRMT members are capable of directly tuning gene expression by depositing histone methylarginine marks at gene promoters.
Prior studies have documented that Prmt1-deposited H4R3me2a most often functions as an activator, while Prmt5-catalyzed H4R3me2s predominantly functions as a r epr essor ( 41 , 42 ).Since both marks occur at the same arginine residue in histone 4, it is thus tempting to postulate that the intricate competition between Prmt1 and Prmt5 likely determines the ultimate outcome of gene expression at specific genomic loci.We thus next assessed how H4R3 methylation is impacted in the absence of Prmt1.Surprisingly, we found that the H4R3me2s intensity was reduced to one-half in the Prmt1-dKO testes as compared to that in WT testis at P7 (Supplementary Figure S8I-L).Among a total of 3165 promoter peaks for H4R3me2s identified in the Prmt1-dKO testes, a pproximatel y a half (1613) were newly synthesized, and this might directly finetune the incr eased expr ession levels for Prmt2 / 5 / 6 in Prmt1-deficient germ cells (Supplementary Figure S8J-L).Further overlapping with H4R3me2a-enriched peaks at TSS region showed that a total of 330 H4R3me2s peaks competed for the endogenous H4R3 loci (Supplementary Figure S8J).In addition, the overlapping between H4R3me2s-enriched TSS peaks and d ys-regula ted genes uncovered that a total of 296 affected genes lost the H4R3me2s mark while 152 genes gained new H4R3me2s decoration (Figure 7 G).This evidence suggests that the endogenous competition between H4R3me2s and H4R3me2a likely accounted for the aberr ant tr anscriptomic regula tion in the Prmt1-null sperma togonia.We ther efor e next visually inspected the IGV browser for a panel of transcription factors essential for spermatogonial de v elopment, which e xhibited down-regulated e xpression upon loss of Prmt1.They can be divided into three categories -SSC self-r enewal, SSC self-r enewal promoting, and spermatogonial differentiation promoting transcrip-tion factors (Figure 7 H-J).The IGV tracks comprise a total of fiv e methylarginine mar ks as well as two well-known acti v e mar ks (H3K4me3 and H3K27ac) as positi v e controls and for promoter identification.This e xhausti v e e xamination re v ealed that, in support of the down-regulated mRNA le v els, the r epr essi v e H4R3me2s peak intensities for three SSC self-renewal transcription factors (Bcl6b, Id4, Lhx1 and Pou3f1) were significantly elevated in their TSS regions in the Prmt1-dKO testes at P7 (Figure 7 H).By comparison, there is no difference for H4R3me2s peak enrichment in the TSS region for Etv5, which showed comparable mRNA e xpression le v els between WT and Prmt1-dKO testes (Figure 7 H).Further visual inspections for three SSC selfrenewal promoting transcription factors (Pou5f1, Zbtb16 and Taf4b) (Figure 7 I) and spermatogonial differentiation promoting transcription factor (Sohlh2) (Figure 7 J) are also supporti v e of the notion that re-occupancy of H4R3me2s upon Prmt1 loss at the promoters, at least in part, accounted for the down-regulated mRNA le v els for genes that are critical to maintain spermatogonial homeostasis (Figure 7 H-J).Together, this evidence suggests that Prmt1-deposited H4R3me2a intrinsically competes with H4R3me2s to establish transcriptomic identity that coordinates spermatogonial self-renewal and dif ferentia tion.

Evidence that H4R3 methylation governs alternative splicing through dir ectly r egulating expr ession of splicing-r elated factors as well as shaping of local chromatin signature
Since we have attained a global landscape of histone methylarginine mark distribution, we next evaluated how the histone marks are linked to the aberrant alternative splicing as seen in Prmt1-null spermatogonia.An examination of Smart-seq2 data identified that > 5000 transcripts for a total of 3938 genes were aberrantly spliced in mouse spermatogonia (Figure 6 F).Among them, GO enrichment showed that a group of defecti v ely spliced transcripts are closely linked to splicing machinery assembly and function (Figure 8 A), and displayed skipped exon (SE), retained intron (RI) and alternati v e 3 splice site (A3SS) (Figure 8 B and Supplementary Figure S9A-C).Some of them, such as Srsf4, Hn-RNPA1 and Ptbp1, were significantly d ys-regula ted in the Prmt1null spermatogonia (Figure 8 C).Likewise, Bud31and Ddx5-mediated alternati v e splicing in SSCs governs the dif ferentia tion and self-renewal of mammalian male germ cells ( 43 , 44 ).This evidence suggests that Prmt1 likely coordina tes alterna tive splicing indirectly by regulating expression of splicing factors.Since H4R3 methyl marks are highly enriched in the promoter region of protein-coding genes as discovered by CUT&Tag profiling, we next asked whether there is a direct regulation of the AS-regulating transcripts by H4R3 methylation.The overlapping with the H4R3me2a-enriched peaks at promoters identified a total of 893 defecti v ely splicing genes in the Prmt1-dKO spermato gonia, w hich are presumably under the direct transcriptional control of Prmt1-catalyzed H4R3me2a modification.H4R3me2a is enriched on promoter and gene bod y tha t help in normal transcription initiation.We argued tha t these modifica tions influence alterna ti v e splicing because transcription is dependent on arginine histone PTM and those transcribed exons are spliced further to mature  transcript as we identified 1949 genes that were overlapped with defecti v ely spliced genes (Figure 8 D).This evidence suggests that Prmt1 is capable of directly modulating the alternati v e splicing, at least in part, by H4R3me2a deposition at promoters for splicing-related factors.
Surprisingly, distinct from the known peak enrichment at promoters for the acti v e H3K4me3 ( > 80%) and H3K27ac ( > 40%) marks (Supplementary Figure S8G-H), the global CUT&Tag profiling uncovered prominent proportions of peaks enriched at the exons ( ∼4%) and introns ( ∼30%) for the fiv e methylarginine mar ks (Figure 7 A-E and Supplementary Figure S8D-I).This led us to speculate that these methyl marks likely execute other regulatory functions, in addition to the fraction of peaks distributed at promoters in driving gene expression.Indeed, emerging evidences have shown that the non-random distribution of H3K36me3 occupies mor e fr equently in exonic than in intronic regions in the protein-coding gene bodies, and this is intimately linked to the characteristic alternati v e splicing pattern in a context-specific manner ( 45 ).To test this hypothesis, we first assessed how the various types of histone marks correlate with each other for the SE and RI splicing e v ents in mouse sperma togonia.Pearson correla tion analyses unveiled that there is closer relationship of the peaks among H4R3me2a, H3R2me2a, H4R3me1 and H3R8me2a (Pearson correlation coefficient > 0.69) (Supplementary Figure S9D, E).By comparison, there is a less correlation for H4R3me2s (Pearson correlation coefficient ≈ 0.4) as compared to other four histone arginine methyl marks (Supplementary Figure S9D, E), suggesting there might exist a counteractive effect between H4R3me2s and H4R3me2a (possibly other methylarginine marks) during RNA transcription.As a control, the H3K4me3 and H3K27ac marks are known as acti v e mar ks without known functional roles in regulating exon skipping or intron retention, and are thereby not surprising to cluster separately.Next, we extracted the intronic and exonic regions with 500-bp flanking sequences across the exon-intron junctions separately for the aberrantly spliced genes, and calculated the averaged read tag densities for different histone modification marks.This uncovered that, consistent with the reported selecti v e distribution of H3K36me3 mark involved in splicing ( 45 , 46 ), the peaks for H4R3me2a, H4R3me2s and H3R2me2a are highly enriched in the exonic regions, but not in the intronic r egions (Figur e 8 E-F and Supplementary Figur e S9F-H).On the other hand, for the alternati v ely spliced genes, we isolated both the constituti v e and alternati v e e xon regions with flanking 100-bp intronic sequence, and calculated the read density enrichment across the exon-intron junctions.In support of previous findings, there exists an apparent enrichment of H4R3me2s and H4R3me2a peaks for both the 5 -and 3 -ends acr oss exon-intr on junctions, but not for the H3K4me3 and H3K27ac peaks (Figure 8 G, H and Supplementary Figure S9IK).In the Prmt1-null spermatogonia, the enrichment trend for H4R3me2s peaks completel y disa ppear ed (Figur e 8 H), indicti v e of a reciprocal interplay between H4R3me2s and H4R3me2a in regulating alternati v e splicing of the e xons / introns in spermatogonia.In summary, these data provide evidence that histone methylarginine-directed chroma tin signa ture, consisting of H4R3me2s and H4R3me2a, harbors profound impact on alternati v e splicing in spermatogonial de v elopment.

DISCUSSION
Maintenance of male fertility hinges on continuous spermato genesis, w hich is established and sustained through the timely self-renewal and co-ordinated differentiation programming in the highly heterogeneous population of spermatogonial cells in the mammalian testis.Herein, we identified a core epigenetic factor, namely Prmt1, which is required to coordinate spermatogonial de v elopment through (i) governing gene expression for germline transcription factors associated with spermatogonial self-renewal and differentia tion; (ii) modula ting the expression of splicing-related factors by depositing H4R3me2a at gene promoters; (iii) fine-tuning the alternati v e splicing through delineation of H4R3me2a mark at exonic regions.

Genome-wide profiling of histone methylarginine marks by CUT&Tag identified the permissive but low-abundance distribution at promoters
To gain a deep understanding of epigenetic regulation at the molecular le v el, it is of paramount importance to attain a thorough landscape of genome-wide distribution loci for a specific histone PTM.As described above, arginine methylation is one of the widespread histone PTMs occurring in mammalian cells, yet good datasets generated by ChIP-seq approach are by far rarely availab le.An e xhausti v e scrutiny of the published papers discovered that the global distribution of most PRMTs and their deposited histone arginine methyl marks is verified through conventional ChIP-PCR or ChIP-qPCR ( 40 ).Only until recently, the genomic distribution loci for H3R17me2a ( 47 ), H4R3me2s ( 48 ) and H3R2me2a ( 49 , 50 ), which are catalyzed by Prmt4, 5 and 6, respecti v ely, hav e been pub lished using ChIP-seq.In this study, after verifying the validity of the antibodies for immunoprecipitation, we chose to utilize the recently de v eloped CUT&Tag approach to profile the genomic loci distribution in the de v eloping mouse testis.We found that all fiv e histone arginine methyl marks (H4R3me2a, H3R2me2a, H4R3me2s, H3R8me2a, H4R3me1) are in general ubiquitously enriched at promoter regions.Howe v er, their enrichment intensities are much lower than that of H3K4me3 in acti v ely transcribed genes (Figure 7 and Supplementary Figur e S8).Mor eover, their enrichment le v els increase substantially from P7 (spermatogonia) to P14 (meiotic sperma tocytes) a t promoters during postna tal testicular de v elopment (Supplementary Figure S8), suggesting that the germline de v elopment is subject to dynamic expr ession r egulation during mitosis-to-meiosis transition.
In addition, prior studies have reported that CpG islands (CGIs) are frequently present in the promoter regions, and ar e often differ entially methylated in a cell context-specific pattern (51)(52)(53)(54)(55).We ther efor e extracted and divided the promoter peaks into two proportions: CGI peaks and non-CGI peaks (Supplementary Figure S7).Overall, we discovered that CGI promoters are more likely subject to modulation by histone arginine methylation as compared to the non-CGI promoters in the de v eloping testis from P7 to P14, presumably indicating a synergistic role of arginine methylation along with DNA methylation directly in controlling germline gene expression.
We reasoned that the success of CUT&Tag approach that outcompetes conventional ChIP-seq in profiling the histone arginine methyl marks could be attributed to se v eral aspects.First, the whole steps of target chromatin capture and DNA fragmentation for CUT&Tag were executed on li v e cells without any fixation step.It is generally accepted that the typical fixati v e used in conventional ChIP-seq, e.g.formaldehyde, can mask the target antigen causing failure of antibody recognition ( 56 ).This assumption is manifested by a more recent study, which unambiguously demonstrated that H3K36me3 is also highly enriched in gene promoters, in addition to the gene body enrichment, by utilizing both CUT&Tag and nati v e ChIP-seq (N-ChIP-seq) methods ( 57 ).In contrast, the crosslinking ChIP-seq (X-ChIPseq) seldomly identified the H3K36me3 in the promoters but e xclusi v ely detected the H3K36me3 in the gene body r egions, as often r eported by pr evious studies ( 58 , 59 ).Second, although histone arginine methylation is permissi v ely distributed across the genome, we and others found that the methylation le v els in histones are generally v ery low as compared with those of histone lysine methylation (Supplementary Figures S7 and S8).In a typical ChIP-seq libr ary prepar ation pipeline, the imm unoprecipitated DN A fragment is directly subject to end repair and adaptor ligation for library pr eparation ( 60 ).Ther e is a lack of further target-specific DNA enrichment step that can distinguish them from the background DNA noise as carried over by the non-specific binding of both beads and antibodies.In comparison, CUT&Tag executes the target DNA fragmentation and adapter ligation in situ , without further introduction of non-target genomic DNA fragments into the library, thereby yielding extremely low background signals.As such, it is advisab le to e xploit CUT&Tag, rather than conv entional fixati v e / sonication-based ChIP-seq appr oach, to pr ofile low-abundance histone PTMs, such as arginine methyl marks.

Arginine methyl marks cross-talk and are required to coordinate spermatogonial development
Prior studies showed that Prmt5 exhibits a dynamic expression pattern in the germline.It is abundant in the cytoplasm of PGC at embry onic da y 7.5 (E7.5), f ollowed by shuttling to the PGC nuclei prior to the sex dif ferentia tion at E12.5 (14).In the de v eloping testis, it is highly detectable in the nuclei of pro-spermato gonia, and graduall y translocates to the cytoplasm in the differentiating spermatogonia and early stage of spermatocytes ( 61 ).The nuclear Prmt5 is important to restrict the retrotransposon activity by depositing the r epr essi v e H4R3me2s mark in PGCs and possibly in postnatal mouse germline cells.Compared with somatic organs, we found that the primary Type I PRMT enzyme -Prmt1-displays higher protein e xpression le v els in the nuclei of germline in de v eloping testis (Figure 1 ), in particular in spermatogonia and early stages of meiotic spermatocytes.In line with its expression pattern, we generated three germline-specific and Tam-inducible KO mouse mod-els, and provided genetic evidence showing that inactivation of Prmt1 remar kab ly impeded the timely dif ferentia tion and self-renewal in the spermatogonial cells, implicating a central role of nuclear Prmt1 in establishment and maintenance of spermatogonial identity in testis (Figures 2 , 5 and Supplementary Figure S4).
With the pan-methylarginine antibodies against MMA, ADMA and SDMA mar ks, compelling e vidence showed that loss of Prmt1 activity induced a remar kab le increase in MMA and SDMA le v els of substrate methylation, but not in ADMA le v els, in somatic MEF cells ( 32 ).This evidence suggests a direct cross-talk between Prmt1-directed asymmetric arginine methylation and Prmt5 / 7-mediated SDMA / MMA deposition.In other words, there exists a substrate competition among three types of arginine methylation.Howe v er, in the spermatogonia of the de v eloping testis, we discovered that Prmt1 loss resulted in an eminent increase in both the ADMA and MMA le v els, along with a slight increase in SDMA le v els, which resulted from the compensatory expression of Prmt2 / 5 / 6 / 7 (Figure 4 ).By comparison, only elevated expression of Prmt5 was observed in spleen tissue.This signifies that those different members from the same Type I PRMT family, e.g.Prmt1, Pr mt2 and Pr mt6, as evidenced by both mRNA and protein verification, compete for each other in the germline, but not in the somatic tissues, such as spleen.On the other hand, our genome-wide CUT&Tag data showed that, in the Prmt1-null spermatogonia, the peak intensities of H4R3me2s were generally decreased at the promotor loci of Prmt2 / 5 / 6 genes.Since H4R3me2s is known as a r epr essi v e mar k, we specula te tha t the decreased H4R3me2s occupancy le v els might contribute to the ele vated e xpression le v els for Prmt2 / 5 / 6 in germ cells.
Interestingly, Prmt2 has been traditionally considered as an enzyme-dead arginine methyltr ansfer ase owing to its extremely weak enzymatic activity seen in vitro ( 62 ).Nonetheless, recent studies have unraveled its catalytic activity on asymmetric dimethylation of histone H3R8 (H3R8me2a), which is associated with the oncogenic activation of glioblastoma ( 63 ).Our genome-wide profiling by CUT&Tag also re v ealed that Prmt2-deposited H3R8me2a is capable of modulating gene expression synergistically with other arginine marks, since all five histone arginine methyl marks were abundantly enriched at gene promoters.

Arginine methyl signaling is important but play distinct functions during spermatogonial development and meiotic divisions
We have previously reported that genetic ablation of Pr mt4 / Car m1, which is present in the cytoplasm of both sper matogonia and sper matocytes but shuttles to the nuclei of haploid spermatids, se v er ely impair ed the elongation of haploid spermatids without apparent defects seen in meiotic spermatocytes ( 13 ).In this study, we found Prmt1 is localized abundantly in the nuclei of spermatogonia, but weakly in the nuclei of spermatocytes (Figure 1 G and Supplementary Figure S1C).In both the Prmt1-sKO and Prmt1-dKO males treated by Tam, we observed aberrant meiotic spermatocytes and loss of early stages of spermatocytes (Figure 5 and Supplementary Figure S4), implying that Prmt1 is necessary for the de v elopment of both spermatogonia and meiotic cell-cycle progression in the testis.
By the CUT&Tag profiling, we consistently observed the increased enrichment of all fiv e methyl arginine marks at gene promoters in spermatocytes (P14) as compared to spermato gonia (P7), w hich is concomitant with the sim ultaneous increase in H3K27ac enrichment, whereas there is no difference in H3K4me3 enrichment at promoters (Figure 7 and Supplementary Figure S8).This presumably suggests that Prmt1 might play a significant function in meiotic division, which is distinguished from that in spermatogonia.

Evidence supporting that histone arginine methyl marks constitute a chromatin signature that is linked to transcriptional alternative splicing
It is known that the pre-mRNA processing is cotranscriptionally coupled to the gene transcription, and emerging evidence supports that there exists cotranscriptional splicing of alternati v e e xons and introns concomitant with RNA transcription ( 45 , 64 ).Ther efor e, it is concei vab le that a combinatorial chromatin signature will impact the transcriptional outcome, including alternati v e splicing, in a context-dependent manner.However, often these biological e v ents were studied independently owing to the high complexity of factors involved in the transcriptional processing.Recent evidence has provided clues tha t rela te histone modifica tion marks functionally to alternati v e splicing ( 65 ).For instance, integrati v e analyses of large-scale RNA-seq and ChIP-seq datasets, in conjugation with machine learning, re v ealed that the histone marks, such as H3K36me3, H3K79me2 and H3K4me1, ar e functionally r ele vant to e xclusion / inclusion of e xons and introns in a variety of cellular processes ( 66 ).Among the known AS-related histone marks, H3K36me3 is the onl y fairl y-studied e xample involv ed in numerous cellular e v ents, and its mode of action is relati v ely clear.For instance, H3K36me3 is recognized by the chromodomain of MORF-related gene on chromosome 15 (Mrg15), which recruits the splicing regulator PTB1 to modulate alternati v e splicing ( 67 ).Among di v erse cellular conte xts, H3K36me3 is often highly enriched in the exonic regions of the proteincoding genes, along with the promoter regions as recently discovered by the optimized CUT&Tag profiling ( 57 ).Strikingly, we also unveiled that all fiv e histone arginine methyl marks are distributed in the exon / intron regions in addition to the promoter enrichment.Exhausti v e analyses suggest that the intricate interplay among these histone mar ks is responsib le for the characteristic splicing pattern in the germline essential for male fertility.

DA T A A V AILABILITY
All the sequencing data for RNA-seq, Smart-seq2 and CUT&Tag are publicly available in the Gene Expression Omnibus database under the Accession Code GSE227857.

SUPPLEMENT ARY DA T A
Supplementary Data are available at NAR Online.

Figure 1 .
Figure 1.Prmt1 is highly expressed and predominantly localized to the nucleus in spermatogonia in mouse testes.( A ) Relati v e quantification of Prmt1 mRNA e xpression le v els across different tissues in mice (n = 3 mice per genotype).( B ) Immunob lotting of Prmt1 in wild-type (WT) mouse tissues at postnatal day 7 (P7).Gapdh served as a loading control.(C ) Quantitati v e data show normalized protein intensity of Prmt1 at P7 in WT mice from 'B' ( n = 2).( D ) Relati v e mRNA le v els of Prmt1 in WT de v eloping testes ( n = 2).( E ) Immunoblotting of Prmt1 in WT developing testes ( n = 2).( F ) Quantitation of normalized protein intensity for Prmt1 in WT mouse testes from 'E' ( n = 2).( G ) Immunostaining of Prmt1 and Sox9 in testicular sections from WT mice at P7 and P38.Scale bars, 20 m. ( H ) A schema tic illustra tion showing the expression of Prmt1 in WT testis.Pink color r epr esents the nuclear localization of Prmt1 protein.( I ) mRNA expression of PRMT members at single-cell le v els in different de v elopmental stages of germ cells in P15 juvenile mice ( 68 ), TPM, transcript per million.

Figure 2 .
Figure 2. Prmt1 is indispensable for spermatogenesis and male fertility.( A ) Schematic diagram depicting the targeting strategy for Prmt1 conditional knockout construct.( B ) Schematic r epr esentation of br eeding scheme by crossing Prmt1 lo x / lo x with the thr ee Cr e-lines.( C ) Immunoblotting of Prmt1 in WT and Prmt1-uKO mice at P42. Gapdh served as a loading control.( D ) Immunoblotting of Prmt1 substrate (H4R3me2a) in WT and Prmt1-uKO testes.H3 served as a loading contr ol. ( E ) Gr oss morphology of adult mouse testis from Pr mt1-uKO, Pr mt1-dKO, and Prmt1-sKO mice as compared with WT mice.( F ) Averaged testicular weights derived from Prmt1-uKO, Prmt1-dKO, and Prmt1-sKO mice, as compared with those in WT.Error bars indicate SEM.**** P < 0.0001.(G) Fertility test for Prmt1-dKO, Prmt1-sKO and age-matched WT males for 6 months.The X-axis r epr esents the time periods during mating; Y-axis indicates the number of pups.Error bars indicate SEM.**** P < 0.0001.( H ) Immunostaining of Prmt1 and Sox9 in testicular sections from WT and Prmt1-dKO mice at P24. Scale bars, 20 m. ( I ) Immunostaining of H4R3me2a in testicular sections from WT and Prmt1-sKO mice at P21. Scale bars, 20 m.

Figure 3 .
Figure 3. Prmt1 modulates the substrate arginine methylome in the germline distinct from somatic cells.( A ) Immunoblotting of MMA, ADMA and SDMA in WT and Prmt1-sKO mice at P8 and P17.Short, short exposure; long, longer exposure.Gapdh serves as an internal control.The marker sizes are labelled (KDa).( B-D ) The line charts depict the relati v e e xpression intensities for the visible bands for MMA, ADMA and SDMA marks with normalization (log 2 ), respecti v ely.Each dot represents the relati v e intensity for each visible band from the top to the bottom in each sample, as calculated by densitometric analysis (using ImageJ).Student's t-test , mean ± SEM ( n = 2).( E ) Immunoblotting of MMA in WT and Prmt1-uKO mice at indicated time (Supplementary FigureS2A).( F ) Immunoblotting of SDMA in WT and Prmt1-uKO mice at indicated time (Supplementary FigureS2A).( G ) Immunoblotting of ADMA in WT and Prmt1-uKO mice at indicated time (Supplementary FigureS2A).( H-J ) Quantitation of normalized band densitometries for MMA, SDMA and ADMA immunoblots from 'E-G', respectively, as described in (B-D).Note the same blots for Prmt1 and Gapdh loading controls were used since they were deri v ed from the same samples.*P < 0.05, ** P < 0.01, *** P < 0.001 and **** P < 0.0001; mean ± SEM, and n = 3.

Figure 6 .
Figure 6.Prmt1 sculpts the transcriptomic identity in spermatogonia.( A ) The scheme for tamoxifen injection and tissue collection.( B ) Volcano plot illustra ting the d ys-regula ted genes in the Pr mt1-dKO sper matogonia purified by FACS from mice as indicated in 'A' using Smart-seq2 approach.( C ) A hea tmap illustra ting the d ys-regula ted genes in Pr mt1-dKO sper mato gonia compared with WT in biolo gical duplicates.Fold change > 2.0, and P < 0.05 (upr egulated in r ed; down-r egulated in blue).( D ) GO (gene ontology) terms of differentially expressed genes (DEGs) in FACS sorted, Prmt1-dKO sperma togonia.( E ) Rela ti v e mRNA le v els for genes involved in development of spermatogonia from Prmt1-dKO and WT mice ( n = 3).Two-tailed Student's t -test, the data are displayed as the mean ± SEM. ( F ) Statistical calculation of alternati v e splicing e v ents identified in sorted, Pr mt1-dKO sper matogonia via Smart-seq2.( G ) GO analysis of alternati v el y spliced genes that are significantl y affected in sorted, Pr mt1-dKO sper mato gonia.( H ) A sna pshot of IGV bro wser sho wing mRNA e xpression le v els for Kdm5a, Brca1, and Clk4 in sper matogonia from WT and Pr mt1-dKO mice; The red arrow indicates the splicing site.( I ) RT-PCR verification for Kdm5a, Brca1, and Clk4 genes sho wing AS pattern of ex on skipping (Ex on 18, 5 and 4), respecti v ely, in Prmt1-dKO testes; Gapdh was used as an internal control.( J ) Quantification of the expression for transcripts with skipped exon.FL, full length isoform, r epr esents skipped exon; ( n = 3).Two-tailed Student's t -test, the data are displayed as the mean ± SEM. ( K ) A snapshot of IGV bro wser sho wing Cenpt mRNA expr ession.The r ed arrow indicates the splicing site.( L ) mRNA e xpression le v els of Cenpt deduced from single-cell dataset in P15 mouse testis, transcript per million (TPM) ( 68 ).( M ) RT-PCR analysis of Cenpt mRNA expression showing AS pattern of intron 11 retention in Prmt1-dKO testis.( N ) Quantification of the expression for transcripts with retained intron in 'M' ( n = 3).Two-tailed Student's t -test, the data are displayed as the mean ± SEM.

Figure 7 .
Figure 7. Global profiling identified the genomic landscape of histone arginine methyl marks and a direct competition between H4R3me2s and H4R3me2a at promoters in regulating gene expression.( A , B ) Genomic distribution of H4R3me2a marks by CUT&Tag profiling in the WT mouse testes at P7 and P14. ( C ) Distribution of genomic loci for H4R3me2a in the WT mouse testis at P7 and P14. ( D ) Enrichment of H4R3me2a at TSS region in the WT mouse testis at P7 and P14. ( E ) Comparati v e analysis of genomic distribution for fiv e histone methylarginine mar ks, plus acti v e mar ks for promoter (H3K4me3), and enhancer (H3K27ac) in WT testis at P7. ( F ) Gene overlapping among H4R3me2a-P7 all genes, H4R3me2a enriched at promoters, and Prmt1-dKO DEGs.( G ) Overlapping genes for H4R3me2s enriched at promoters (WT) and H4R3me2s enriched at promoters (Prmt1-dKO) and Prmt1-dKO DEGs; GO terms for 296 genes that are only shared by WT H4R3me2s and Prmt1-dKO DEGs.( H-J ) IGV browser snapshot showing the peak distribution for histone methylarginine marks and mRNA gene expression for SSC self-renewal transcription factors, SSC self-renewal promoting transcription factors and spermatogonial differentiation promoting transcription factors.

Figure 8 .
Figure 8. H4R3 methylation governs alternati v e splicing through directly regulating expression of splicing-related factors as well as indirect shaping of local chromatin signature.( A ) GO enrichment for RNA splicing-related genes identified by Smart-seq2 in the Pr mt1-dKO sper ma togonia.( B ) Dif ferential expression and inclusion level difference for three types of splicing e v ents in RNA splicing-related genes.> 0 indicates a difference in the degree of inclusion in Prmt1-dKO and < 0 reflects inclusion in WT. ( C ) Volcano plot depicting the d ys-regula ted expression of RNA splicing-related genes.( D ) Gene overlapping among H4R3me2a-enriched genes, H4R3me2a enriched at promoters, and Prmt1-dKO AS genes.( E , F ) Comparison of the averaged peak intensities for H4R3me2s and H4R3me2a enrichment between introns and exons deduced from CUT&Ta g.The avera ge lengths for exons are ∼450 bp while intron lengths averaging ∼5 kb.( G , H ) C omparison of the averaged peak enrichment intensities for H4R3me2s and H4R3me2a across the 100 bp flanking regions for alternati v e and constituti v e e xons.