GCNA is a histone binding protein required for spermatogonial stem cell maintenance

Abstract Recycling and de-novo deposition of histones during DNA replication is a critical challenge faced by eukaryotic cells and is coordinated by histone chaperones. Spermatogenesis is highly regulated sophisticated process necessitating not only histone modification but loading of testis specific histone variants. Here, we show that Germ Cell Nuclear Acidic protein (GCNA), a germ cell specific protein in adult mice, can bind histones and purified GCNA exhibits histone chaperone activity. GCNA associates with the DNA replication machinery and supports progression through S-phase in murine undifferentiated spermatogonia (USGs). Whilst GCNA is dispensable for embryonic germ cell development, it is required for the maintenance of the USG pool and for long-term production of sperm. Our work describes the role of a germ cell specific histone chaperone in USGs maintenance in mice. These findings provide a mechanistic basis for the male infertility observed in patients carrying GCNA mutations.


INTRODUCTION
The germline is tasked with faithfully transmitting genetic information from one generation to the next. The earliest mammalian germ cells, Primordial Germ Cells (PGCs), emerge during early embryonic development and migrate to the genital ridges (1). In the embryonic gonad, these cells undergo rapid proliferation and extensive epigenetic changes. Subsequently, female embryonic germ cells initiate meiosis, a process that does not begin until postnatal life in males. The male embryonic germ cells, in contrast, give rise to a population of self-renewing undifferentiated spermatogonia (USG), which comprises spermatogonial stem cells (SSCs) and progenitor germ cells (2). At the onset of puberty, USGs resume proliferation, differentiate and initiate spermatogenesis. The capacity to self-renew underlies the ability of males to produce gametes throughout their entire lifetime. In order to both self-renew and also generate daughter cells that ultimately differentiate into sperm, USGs must undergo repeated rounds of DNA replication.
The process of DNA replication poses a challenge to the fidelity of the genome due to both errors introduced by the DNA polymerases as well as impediments encountered by the polymerases (3). Chromatin also poses additional challenges for eukaryotic cells as during DNA replication nucleosomes must be disassembled ahead of the replication fork and re-allocated in the newly synthesised daughter strands (4). To do this, eukaryotic cells evolved a set of replication-associated histones chaperones. The CAF-1 (chromatin assembly factor-1) complex and ASF1 (anti-silencing function 1) are among the best characterised replication-associated histone chaperones and are recruited through interactions with key replisome components (5)(6)(7). Demonstrating the importance of histone chaperones for DNA replication and cell cycle progression, loss of CAF-1 or ASF1 results in impaired S-phase progression and DNA synthesis (8)(9)(10)(11)(12).
Given the highly proliferative capacity of germ cells and dramatic changes to chromatin encountered during their development, it is likely that unique challenges must be overcome during gametogenesis. Recent reports support this idea showing that histone chaperones are critical at distinct steps of gametogenesis. For example, the homolog of the P150 large subunit of CAF1 is essential for the maintenance of gonadal stem cells in female Drosophila (12). In mice, a homolog of ASF1, ASF1B, supports initiation of meiosis (13). ASF1B is so far the only histone chaperone identified to be important for gametogenesis in mammals. Generally, the chromatin transactions during DNA replication in mammalian USGs are poorly characterised. Given the importance of histone chaperone activity for DNA replication and the proliferative nature of USGs, it is likely that male fertility will be particularly reliant on histone chaperones.
Germ cell nuclear acidic protein (GCNA) has been extensively used as a germ cell marker for almost 30 years (14,15). Recently, mutations in GCNA have been linked to azoospermia in humans, defining GCNA as a clinical determinant of human infertility (16,17). Despite this, little is known about the function in mammals of this evolutionarily conserved factor. GCNA, including the human GCNA protein, con-tains a SprT protease domain closely related to that found in SPRTN, a critical factor in DNA-protein crosslink (DPC) repair (18). Studies in invertebrates report a role for GCNA in DPC repair in the germline (19)(20)(21). However, mouse GCNA lacks the SprT protease domain suggesting a role independent of DPC repair. Consistent with this, we find that GCNA is dispensable for maintaining cellular resistance to DPC-causing agents in mice. Instead, we observe that both mouse GCNA and human GCNA interact with core histones and that mouse GCNA (mGCNA) has histone chaperone activity. Moreover, we find that mGCNA can associate with the DNA replication machinery and consistent with a role as a histone chaperone supporting DNA replication, GCNA-deficient undifferentiated spermatogonia (USG) accumulate in S-phase. Consequently, GCNA deficient mice fail to maintain the USG pool through their lifetime, leading to an age-dependent reduction in sperm production. Therefore, our results suggest that GCNA is a histone chaperone necessary for maintenance of USGs in mammals.

Mice
All animal experiments undertaken in this study were approved by the Medical Research Council's Laboratory of Molecular Biology animal welfare and ethical review body and the UK Home Office under the Animal (Scientific Procedures) Act 1986 (license no. PP6752216). All mice were maintained under specific pathogen-free conditions in individually ventilated cages (GM500; Techniplast) on Lignocel FS-14 spruce bedding (IPS) with environmental enrichment (fun tunnel, chew stick and Enviro-Dri nesting material (LBS)) at 19-23 • C with light from 7:00 a.m. to 7:00 p.m. and fed Dietex CRM pellets (Special Diet Services) ad libitum. No animals were wild and no field-collected samples were used. Unless otherwise stated, mice were maintained on a C57BL/6J background. GOF18-GFP (Tg(Pou5f1-EGFP)2Mnn) (MGI ID: 3057158) JAX (stock ID: 004654) mice were purchased from The Jackson Laboratory (22). Plzf-tdTomato mice were a kind gift from Andrew McKenzie (23). Gcna tm1.1Dcp (Gcna 2lox (2L)) MGI ID: 5910931, JAX stock ID: 031055 (18) mice were purchased from The Jackson Laboratory. To generate the Gcna exon 4 allele, or Gcna knock-out allele, Gcna tm1.1Dcp mice were bred with mice carrying the Stella-Cre allele, or Dppa3 tm1(cre)Peli (24) (MGI ID: 5004882). Genotyping of the Gcna knockout allele was performed by PCR using the following primers: Gcna Fw (5 GGATAGCAAAGGTTTATCAAC 3 ) and Gcna Rv (5 TGTGGTCCATAGCAAAATAAGG 3 ) (Supplementary Figure S1A and B). Gender of mice was determined by PCR as previously described (25).

Protein extracts preparation for western blotting
Mouse ESC were lysed in ice cold RIPA buffer (25 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100) complemented with protease inhibitors cocktail (catalog no. 11873580001, Roche) and let chill on ice for 10 min. Then, extracts were sonicated for 10 min at 4 • C with a non-contact sonicator (30 s ON/30 s OFF cycles, 40% amplitude, catalog no. Vibra-Cell VC 750, Sonics & Materials). Extracts were spun at 16 200g for 10 min at 4 • C, supernatants were collected and analysed by western blot. Testes were processed in the same way but were first homogenised in RIPA buffer with protease inhibitors using a Dounce Homogeniser (10 strokes with loose pestle then 10 strokes with the tight one).

Immunofluorescence on testis sections
When the experiment involved the incorporation of 5ethynyl-2 -deoxyuridine (EdU) into germ cells, mice were given a single intraperitonal injection of EdU (catalog no. A10044, ThermoFisher Scientific), 50 mg/kg at 10 ml/kg. Four hours after injections, testes were biopsied, fixed in formalin (catalog no. HT5011-15 ml, Sigma-Aldrich) and processed for histological analysis as described earlier. Sections of formalin-fixed, paraffin-embedded samples were deparaffinised and rehydrated following standard procedure. Slides were boiled for 20 min into antigen retrieval buffer (10 mM sodium citrate, pH 6.0). Slides were allowed to chill at room temperature then slides were washed three times in water for 5 min and then once for 5 min in Trisbuffered saline (TBS) containing 0.1% Tween 20. If needed, labelling of EdU was then performed using the Click-iT Plus EdU Cell Proliferation Kit, Alexa Fluor 594 dye (catalog no. C10639, ThermoFisher Scientific). Sections were then incubated in blocking buffer (TBS, 0.1% Tween 20, 5% fetal bovine serum) for 1h at room temperature. Sections were incubated ON at 4 • C with the following primary antibodies diluted in blocking buffer: anti-GCNA clone GCNA- A-11029, ThermoFisher Scientific). After incubation with secondary antibodies, slides were washed three times in TBS + 0.1% Tween 20 for 5 min. After washes, slides were incubated for 10 min with 2 g/ml DAPI in TBS then washed three times in TBS for 5 min. Slides were then briefly dipped into water and mounted with Prolong Gold Antifade Mountant (catalog no. P36934, ThermoFisher Scientific). Images were captured with a LSM 780 confocal microscope (ZEISS). Human testis sections (catalog no. 11-701 YC1, ProSci Inc.) were processed on the same way than mice testis sections.

Epididymal sperm counts
One epididymis (caput, corpus and cauda) per mouse is placed in a 35 mm dish with 1 ml of PBS, pre-warmed at 37 • C. Then, epididymes were punctuated with 19G needles. Dishes were then incubated at 37 • C to let the spermatozoa swim-out. After 30 min, epididymes were flushed using the supernatant. Supernatants were collected and placed in clean, new, 2 ml tubes. Then, epididymes were flushed again with 400 l of fresh PBS and supernatants were collected and added to the previous 2 ml tubes with supernatants. Collected spermatozoa, in approximatively 1.4 ml of PBS, were then counted using a Haemocytometer.

Assessment of the fertility of mice
To determine if GCNA-deficient mice were capable of giving rise to offspring, test mice were paired with wild-type C57BL/6J mates of the opposite sex. Females were examined daily for the presence of copulation plugs. Test mice were mated at 7 weeks old and remained mating until they reached 23 weeks old. The numbers of pups born were recorded. Individuals performing the copulation plug checks were blinded to the genotype of mice.

Quantification of primordial germ cells
Urogenital ridges of E12.5 embryos carrying the GOF18-GFP reporter were isolated and placed into 150 l of trypsin solution (2.5 g/ml trypsin (Gibco), 25 mM Tris, 120 mM NaCl, 25 mM KCl, 25 mM KH 2 PO 4 , 25 mM glucose, 25 mM EDTA, pH 7.6) pre-warmed to 37 • C and incubated for 10 min at 37 • C. Subsequently, 1 l of Benzonase endonuclease (catalog no. 70664, Merk Millipore) was added per sample and samples were disaggregated by gentle pipetting and incubated for a further 5 min at 37 • C. The trypsin was inactivated by adding 1 ml of PBS + 5% fetal bovine serum (FBS). Following 10 min of centrifugation at 3300 r.p.m., the cell pellet was resuspended in 100 l of Alexa Fluor 647-conjugated anti-human/mouse SSEA1 antibody (catalog no. MC-480; BioLegend) diluted 1:100 in PBS + 2.5% FBS and incubated at room temperature for 10 min. Then, 300 l of PBS + 2.5% FBS were added to the cell suspension and samples were immediately run on an ECLIPSE analyzer (Sony Biotechnology) and the data analysed using FlowJo v.10.1r5 (FlowJo LLC).

Expression analysis of GCNA
PGCs were isolated from male GOF-18 embryos by fluorescence activated cell sorting (FACS). Individual gonads of developing embryos carrying the GOF18-GFP reporter were isolated and placed into pre-warmed trypsin solution (2.5 g ml −1 trypsin (Gibco), 25 mM Tris, 120 mM NaCl, 25 mM KCl, 25 mM KH 2 PO 4 , 25 mM Glucose, 25 mM EDTA, pH 7.6) and incubated for 10 min at 37 • C. Subsequently 1 l of Benzonase (EMD Millipore) was added and the sample disaggregated by pipetting and incubated for a further 5 min at 37 • C. The trypsin was inactivated by the addition of 1 ml of PBS, 5% v/v fetal calf serum (FCS). Following 10 min of centrifugation the cell pellet was resuspended in 100 l of Alexa Fluor 647-conjugated anti-human/mouse CD15 (SSEA1) (BioLegend, MC-480) diluted 1:100 in PBS, 2.5% v/v FCS and incubated at room temperature for 10 min. 300 l of PBS, 2.5% v/v FCS was added to the cell suspension and the samples passed through a 70 m cell strainer. Samples were immediately run on a Synergy sorter (Sony Biotechnology Inc.) and the cells sorted into 10 l PBS. Sorted PGCs were centrifuged at 13 000 r.p.m. in a centrifuge chilled to 4 • C and the cell pellets stored at -80 • C until further analysis. Testis were dissected from postnatal mice at the days indicated lysis was performed using a TissueLyser II (catalog no. 85300, Qiagen). RNA was isolated PicoPure kit following manufactuerer's instructions (catalog no KIT0204, Thermo Fisher Scientific). cDNA was prepared using QuantiTect Revere transcription kit (catalog 205313, Qiagen). Taqman probes were used for qRT-PCR expression analysis of Gcna (PN4441114). PCR amplification was performed using the TaqMan Fast Advanced Master Mix (ThermoFisher Scientific). PCR amplification was performed on a Viaa7 cycler for 95 • C for 15 s and 60 • C for 1 min. Mean threshold cycles were determined from three technical repeats using the comparative CT methodology. All expression levels were normalised to Gapdh (mm99999915 g1).

Generation of 3T3 clones stably expressing FLAG-2xNLS-EGFP or FLAG-mGCNA
Mouse NIH3T3 cells were transfected with pExpress-LoxBsr-FLAG-2xNLS-EGFP or pExpress-LoxBsr-FLAG-mGCNA then later selected for resistance to Blasticidin (5 g/ml, catalog no. A11139-03, Gibco). Resistant clones were then screened for the expression of FLAG-2xNLS-EGFP or FLAG-mGCNA by immunoblotting of FLAG (see Western Blotting). For each constructs, two independent and successfully transfected clones were then expanded and maintained in culture with Blasticidin. To determine the growth rate, cells were counted at the indicated times using a cell counter (Vi-CELL XR Cell Viability Analyser, Beckman Coulter).

Cellular fractionation for chromatin isolation
Cells were lysed in one volume of CSK buffer (10 mM PIPES pH 7.5, 100 mM NaCl, 300 mM Sucrose, 3 mM MgCl 2 ) with 0.5 % Triton X-100 and protease inhibitors cocktail (catalog no. 11873580001, Roche) and let chill on ice for 10 min. Cells were then spun at 845g for 5 min at 4 • C Nucleic Acids Research, 2023, Vol. 51, No. 10 4795 and the supernatant containing the cellular soluble fraction was collected and stored. The pellet containing the chromatin fraction was washed two times with three volumes of CSK buffer containing 0.5 % Triton X-100 and protease inhibitors cocktail. For DNAse I digestion of the chromatin fraction, chromatin was incubated in DNAse I buffer (10 mM Tris-HCl pH 7.6, 2.5 mM MgCl 2 , 0.5 mM CaCl 2 ) with 20 U of DNAse I for 45 min at 37 • C. After digestion, the chromatin fraction was then washed two times with five volumes of CSK buffer containing 0.5 % Triton X-100 and protease inhibitors cocktail. The chromatin fraction was finally resuspended in 1.5 X LDS buffer (catalog no. NP0007, The-moFisher Scientific) containing 5% ␤-mercaptoethanol and sonicated for 10 min at 4 • C with a non-contact sonicator (30 s ON/30 s OFF cycles, 40% amplitude, catalog no. Vibra-Cell VC 750, Sonics & Materials). Samples were then analysed by Western Blot.

Chromatin isolation on synchronised cells
NIH3T3 cells were synchronised in S-phase as follow: cells were plated at a density of 5 × 10 5 cells per 10 cm dish. Twenty-four hours after plating, cells were synchronised by being washed two times with DMEM (catalog no. 31966-021, ThermoFisher Scientific) and then placed in DMEM supplemented with 0.5% fetal bovine serum (catalog no. 10270-106, ThermoFisher Scientific). Cells were incubated for 48 h and then released from serum starvation by incubating them for 15 h in DMEM supplemented with 10% fetal bovine serum. After 15 h of culture, cells in S-phase were collected and processed for chromatin isolation as described earlier.

Immunofluorescence on cultured cells
NIH3T3 cells were seeded on no. 1.5 coverslips (catalog no. 631-0150, VWR) 24 h before experiments. If needed, cells were then exposed for 30 min with 10 M EdU (catalog no. A10044, ThermoFisher Scientific). For the classic fixation, cells were washed twice for 5 min with PBS supplemented with 500 M MgCl 2 and 0.5 M CaCl 2 (PBS-S) then fixed with 4% paraformaldehyde (catalog no. 43368, Alfa Aesar) for 20 min and washed with PBS-S twice for 5 min. If cells were previously exposed to EdU, paraformaldehyde was replaced with formaldehyde (catalog no. 28906, Thermo Scientific). Cells were then permeabilised for 10 min with PBS-S containing 0.1% Triton X-100 and washed PBS-S twice for 5 min.
For the pre-extraction before fixation, cells were washed once with PBS-S for 3 min, then with CSK buffer (10 mM Pipes, pH 7.0, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 ) for 3 min and incubated twice for 3 min with CSK buffer containing 0.5% Triton X-100 and complete protease inhibitors cocktail (catalog no. 11873580001, Roche). Cells were after washed with PBS-S, fixed with 4% paraformaldehyde (catalog no. 43368, Alfa Aesar) for 20 min and washed with PBS-S. If cells were previously exposed to EdU, paraformaldehyde was replaced with formaldehyde (catalog no. 28906, Thermo Scientific).

Derivation of mESCs
Gcna +/− females were mated with wild-type males and the next day, embryos were collected and incubated in Embryomax KSOM Mouse embryo Media (catalog no. MR-020P-5F, Sigma-Aldrich) in a 5% CO 2 , 1% O 2 , humidified incubator at 37 • C. After several days of culture, embryos hatched and were then placed on irradiated feeder mouse embryonic fibroblasts. Growing colonies of mESCs were then picked and mESC were expanded in 2i + LIF media (see previously, in 'Cell culture' part), without feeders, on fibronectin-coated plates and in a 5% CO 2 humidified incubator at 37 • C. Clones were screened for their gender and genotype, as described earlier (see 'Mice' part).

2D cell cycle analysis
Cell cycle analysis of mESCs was performed using the Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kit (catalog C10634, Thermo Fisher Scientific) following the manufacturer's instruction. mESC were split 1:5 the day before the assay was conducted. On the day of the assay cells were grown in media supplemented with 10 M EdU for 1h. Following this cells were washed twice in PBS and fixed, permeabilised prior to the click reaction. The cells were resuspended in PBS supplemented with 1ug/ml DAPI and immediately analysed by Flow cytometry on a Fortessa (BD Bioscience).

Phylogenetic analysis of IDRs of metazoan GCNA proteins
IDRs of metazoans GCNA proteins were determined using GlobPlot 2 (29). Electrostatic charges of residues of these IDR were then determined using the EM-BOSS charge software (https://www.bioinformatics.nl/cgibin/emboss/charge). Finally, representations of charges distributions in IDRs were obtained using the Heatmapper software (30).

Chromatin isolation and NaCl gradient washes
Mouse embryonic stem cells were lysed in one volume of CSK buffer (10 mM PIPES pH 7.5, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 ) containing 0.5 % Triton X-100 and protease inhibitors cocktail (catalog no. 11873580001, Roche) and let chill on ice for 10 min. Cells were then spun at 845g for 5 min at 4 • C and the supernatant containing the cellular soluble fraction was collected and stored. The pellet containing the chromatin fraction was washed two times with three volumes of CSK buffer (with modified NaCl concentration ranging from 100 to 700 mM) + 0.5 % Triton X-100 and protease inhibitors cocktail. Then, chromatin fraction was washed two times with the regular CSK buffer containing 0.5 % Triton X-100 and protease inhibitors cocktail. The chromatin fraction was finally resuspended in 1.5

Immunoprecipitation
Cells were lysed in CSK buffer (10 mM PIPES pH 7.5, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 ) with 0.5 % Triton X-100 and protease inhibitors cocktail (catalog no. 11873580001, Roche) and let chill on ice for 10 min. Cells were then spun at 845g for 5 min at 4 • C and the supernatant containing the cellular soluble fraction was collected. One volume of soluble fraction was then diluted with 5 volumes of Binding buffer (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 10% glycerol, 1 mM CaCl 2 , 1 mM MgCl 2 , 0.1% Bovine serum albumin, 1 mM ␤-mercaptoethanol, protease inhibitors cocktail (catalog no. 11873580001, Roche)) then 90 U of Benzonase (catalog no. 70664, Merk Millipore) and 3 U of DNAse I (catalog no. M0303, New England Bio-Labs) were added per ml of mix. For each immunoprecipitation, 25 l of anti-FLAG (catalog no. M8823, Sigma-Aldrich) or anti-GFP (catalog no. gtma-20, Chromotek) magnetic beads were then washed with PBS + 0.05% Tween 20 and then with Binding buffer. After washes, magnetic beads were added to the cellular soluble fraction diluted in Binding buffer. The immunoprecipitation mix was then incubated ON at 4 • C with gentle rotation. The next day, magnetic beads were washed three times with CSK buffer + 0.5 % Triton X-100 and protease inhibitors cocktail, 5 min with gentle rotation at 4 • C for each wash. Elution was then done by adding 1.5× LDS buffer (catalog no. NP0007, The-moFisher Scientific) containing 5% ␤-mercaptoethanol to the beads. Samples were then analysed by western blot.

Protein purification
Purification of MBP-mGCNA. Mouse GCNA Nterminally fused to MBP was purified from Expi293 cells as follow: 3 × 10 9 Expi293 cells were transfected with pcDNA3.1(+)-MBP-mGCNA by PEI transfection (31). Cells were then incubated in presence of 3.5 mM valproic acid (catalog no. P4543, Sigma-Aldrich) in culture media for 4 days at 37 • C. Cells were then pelleted at 300g for 5 min and lysed in lysis buffer (20 mM Tris ph 7.4, 200 mM NaCl, 1 mM EDTA, 1X Protease inhibitors (catalog no. 11873580001, Roche), 0.5% Triton X-100, 5 mM CaCl 2 , 20 g/ml RNAse A (catalog no. R6148, Sigma-Aldrich), 1000 U/ml MNase (catalog no. M0247S, New England Biolabs)) at a ratio of 1 ml of lysis buffer per 100 mg of cell pellet. Cells were sheared with a 21G needle then sonicated for 10 min at 4 • C with a non-contact sonicator (30 s ON/30 s OFF cycles, 40% amplitude, catalog no. Vibra-Cell VC 750, Sonics & Materials). Extracts were then spun at 16 200g for 20 min at 4 • C and supernatants were collected. Cleared extracts were diluted ten times in column buffer (20 mM Tris ph 7.4, 200 mM NaCl, 1 mM EDTA, 1× protease inhibitors (catalog no. 11873580001, Roche)). One litre of diluted extracts were incubated with 9 ml of packed amylose resin (catalog no. E8021S, New England Biolabs) and incubated at 4 • C for 2 h with gentle rotation. Extracts with resin were then loaded onto a column and washed with 3 volumes of column buffer, then 6 volumes of column buffer with 1M NaCl and with 3 volumes of elution buffer without Maltose (25 mM Tris ph 7.4, 10% glycerol, 75 mM NaCl, 2 mM DTT, 1 mM EDTA). Proteins were then eluted with 25 ml of elution buffer supplemented with 10 mM maltose and several fractions were made. Fractions were then analysed by SDS-PAGE and fractions containing MBP-mGCNA were pooled. Pooled fractions were loaded onto an anion exchange Maxi column (catalog no. 78243, ThermoScientific). The column was washed three times with buffer A (25 mM Tris pH 7.4, 75 mM NaCl, 10% glycerol, 1 mM EDTA, 2 mM DTT), 10 ml per wash. Then, the column was serially washed with buffer A containing increasing concentrations of NaCl (100, 150, 200, 300 mM NaCl), 10 ml per wash. Then, MBP-mGCNA was eluted three times with 2 ml buffer A containing 500 mM NaCl. Elutions with pure MBP-mGCNA were aliquoted and stored at -80 • C.
Purification of MBP. MBP was purified from E. coli as follow: Rosetta (DE3) cells (catalog no. 70954, Sigma-Aldrich) transformed with pOPTM were incubated ON at 37 • C in SOB media (2% Tryptone, 0.5% Yeast extract, 10 mM NaCl, 2.5 mM KCl, 10 mM MgCl 2 ) containing chloramphenicol (25 g/ml) and ampicillin (100 g/ml). The next day, 5 ml of ON culture was diluted in 500 ml 2× TY media (16 g/l tryptone, 10 g/l yeast extract, 5 g/l NaCl) containing chloramphenicol and ampicillin and the culture was incubated for 4 h at 37 • C. Then, 500 ml 2X TY media with chloramphenicol and ampicillin and 1 mM IPTG were added to the culture. The culture was then incubated for 15 h at 20 • C. Culture was then pelleted and stored at -80 • C. The cell pellet was then lysed with B-PER buffer (catalog no. 78243, ThermoFisher Scientific) according to the manufacturer's instructions. Lysate was then spun at 16 200g for 15 min at 4 • C and supernatant was collected. The supernatant was then diluted ten times in column buffer (20 mM Tris ph 7.4, 200 mM NaCl, 1 mM EDTA, 1X× protease inhibitors (catalog no. 11873580001, Roche)). Diluted lysate was then incubated with 9 ml of packed amylose resin (catalog no. E8021S, New England Biolabs) and incubated at 4 • C for 2 h with gentle rotation. Extracts with resin were then loaded onto a column and washed with 3 volumes of column buffer, then 6 volumes of column buffer with 1M NaCl and with 3 volumes of elution buffer without Maltose (25 mM Tris ph 7.4, 10% glycerol, 75 mM NaCl, 2 mM DTT, 1 mM EDTA). Proteins were then eluted with 25 ml of elution buffer supplemented with 10 mM Maltose and several fractions were made. Fractions were then analysed by SDS-PAGE and fractions with MBP were pooled. The buffer containing MBP was then exchanged with the storage buffer of MBP-mGCNA (25 mM Tris pH 7.4, 500 mM NaCl, 10% glycerol, 1 mM EDTA, 2 mM DTT) using a Vivaspin-20 concentrator (catalog no. VS2021, Sartorius). Five buffer exchanges with the storage buffer were performed and purified MBP was then aliquoted and stored at -80 • C.

Limited proteolysis assay
In a 100 l reaction, 1.4 g of H3-H4 tetramers (catalog no. M2509S, New England Biolabs) or H2A-H2B dimers (catalog no. M2508S, New England Biolabs) were incubated with 2.5 g of MBP or MBP-mGCNA in Binding buffer (25 mM Tris pH 7.5, 150 mM NaCl, 5% glycerol). Then, 0.5 ng of trypsin gold (for the proteolysis with H3-H4, catalog no. V528A, Promega) or 1 ng of trypsin gold (for the proteolysis with H2A-H2B) were added to the 100 l reactions. Aliquots of 20 l were then collected at the indicated time points and later analysed by SDS-PAGE. Polyacrylamide gels were then stained by using the silver stain method (catalog no. 24612, ThermoFisher Scientific) and stained gel were imaged by using a ChemiDoc MP Imaging System, Biorad.

Plasmid supercoiling assay
The circular plasmid, phiX174 RF1 (catalog no. SD0031, ThermoFisher Scientific) was pretreated for 24 h at 37  , 60 mM Tris pH 8, 30 mM EDTA, 2% SDS, 2 mg/ml proteinase K) and incubated for 1 h at 37 • C. Products were then analysed by 1% agarose gel electrophoresis in TBE buffer followed by SYBR Gold staining (catalog no. S11494, ThermoFisher Scientific). Gels were imaged with a ChemiDoc MP Imaging System, Biorad. Supercoilling was quantified using Fiji software (32). Each lane were manually selected with the software (Gels > Select First Lane, then Gels > Select Next Lane) and signal intensity for each lane was quantified by Fiji (Gels > Plot Lanes, then use the Wand tool to obtain the quantifications).

Image analysis
Image were analysed using Fiji software (32). To quantify the amount of fluorescence per cell, cell nuclei were manually selected with the software and the fluorescence (IntDen) was quantified by Fiji (Analyse > Measure). Profiles of fluorescence intensity were obtained by first drawing a line on the region of interest then by quantifying the fluorescence intensity with Fiji (Analyse > Plot Profile).

Statistical analysis
The number of independent biological samples and technical repeats are indicated in the figure legends. Unless otherwise stated, data are shown as the mean ± S.D. Analysis was performed in Prism 9 (GraphPad Software).

GCNA supports long term gametogenesis in male mice
Although mutations in the GCNA locus have been associated with azoospermia in humans, the mechanism behind this is not understood (16,17). Therefore, we sought to define the aetiology of this defect using a previously generated Gcna knock-out mouse allele (Supplementary Figure  S1, Exon 4 allele, MGI ID: 5910931, JAX stock ID: 031055) (18). The allele was first validated by western blot and immunofluorescence using two well characterised antibodies (Tra98 and GCNA-1) that recognise the C-terminus of mouse GCNA, a region outside the deletion (14-15,18. We were unable to detect a GCNA signal in the knock-out mice, confirming loss of the protein ( Figure 1A-C, Supplementary Figure S1).
We first set out to ask if GCNA was required for mouse development. GCNA-deficient male mice were born at Mendelian ratios (Supplementary Figure S2A and B). Furthermore, the body weights and one year survival of male mice lacking GCNA were indistinguishable from wild-type littermates ( Figure 1C and Supplementary Figure S2C and D). We then confirmed that the expression of GCNA is restricted to the testis, in adult mice ( Supplementary Figure S2E) (14,15). As these results suggest a specific role of GCNA in testis physiology, we then focused our attention on the testis development of Gcna knock-out mice. Testis mass at 6 weeks old is indistinguishable from wildtype males ( Figure 1D). However, it decreases over time in the absence of GCNA, but not in wild-type controls. By one year of age, GCNA deficient males show approximately a 40% reduction in testis mass compared to littermate controls ( Figure 1D and E). Consistently, the testis weight reduction is accompanied by a time-dependent decrease of sperm number in the epididymis ( Figure 1F). We did not observe any difference in the mass of the seminal vesicle nor the serum testosterone concentration suggesting that hormonal signalling is intact in the absence of GCNA and it is not causing the reduction of the sperm production (Supplementary Figure S2F and G). Although Gcna knock-out male mice produce less sperm upon ageing, these remaining sperm are functional (Supplementary Figure S2H and I). These phenotypes are consistent with the previous description of GCNA deficient mice (18). Together, our observations reveal that whilst spermatogenesis is not blocked in the absence of GCNA, males lacking GCNA are unable to sustain sperm production over time.
To determine the basis of this defect, we histologically assessed the testes at different ages ( Figure 1G). We observed a time-dependent increase in seminiferous tubes lacking germ cells, or Sertoli cell only tubes (SCOs), in Gcna knock-out male mice ( Figure 1H). These results provide an explanation for the decrease of the testis mass and sperm produc- tion over time, showing that GCNA supports the maintenance of the male germ line with time.
As GCNA is important for homeostasis of male germ cells, we sought to assess if it also has a role in females. Gcna knock-out female mice were born at the Mendelian ratio, suggesting that GCNA is not necessary for female development (Supplementary Figure S3A and B). We then counted follicles in ovaries of young and aged mice and did not observe any difference between wild-type and Gcna knockout females (Supplementary Figure S3C and D). Consistent with this result, no reduction in fertility was observed in Gcna knock-out female mice (Supplementary Figure S3E-G). These results indicate that GCNA is dispensable for the female reproductive function and that maternal GCNA is dispensable for embryonic development.

GCNA preserves the undifferentiated spermatogonia pool in mice
We next sought to identify the cause of the age-dependent decrease in spermatogenesis in mice lacking GCNA. We first tested if the premature loss of germ cells could be due to a seeding defect of the embryonic gonads. We first assessed the expression of GCNA and found that expression began in PGCs at E11.5 and continued to increase in expression in the testis of postnatal mice (Supplementary Figure  S4A). We therefore crossed the Gcna knock-out allele with a primordial germ cell reporter which expresses GFP under the control of a fragment of the Oct4 promoter (also known as GOF18-GFP) (22) (Supplementary Figure S4B). We found that at E12.5 Gcna knock-out males have indistinguishable numbers of PGCs compared to wild-type littermates (SSEA1 + GOF18-GFP + PGGs, Supplementary Figure S4C-E). These results suggest that the age-dependent loss of germ cells in males lacking GCNA may not have an embryonic origin.
Our results indicate that GCNA-deficient male mice have normal numbers of germ cells and normal testis mass and sperm production in their first weeks of life. We hypothesised therefore that the undifferentiated spermatogonia (USG) pool, which comprises spermatogonial stem cells (SSCs) and progenitor germ cells, prematurely contracts during the post-natal life, in the absence of GCNA. To test this, we first verified that GCNA is expressed in USGs by immunofluorescence. USGs were identified by immunostaining of the USGs marker promyelocytic leukaemia zinc finger (PLZF, 33,34) and we observed that a high proportion of PLZF positive cells (approximatively 70%) also expressed GCNA (Figure 2A). Consistent with a role for GCNA in human spermatogenesis too, we observed the same expression pattern in human USGs ( Figure 2B). These results are intriguing as GCNA is thought to be a pan germ cell marker (14,15). However, it is worth noting that the expression of GCNA is cell-cycle regulated, with a peak of expression in G2 (20). Approximatively 40% of USGs are not committed into the cell-cyle (G0) (35). Part of these USGs in G0 may not express GCNA or express GCNA under the level of detection by immunofluorescence. This hypothesis can explain why only 70% of USGs are positive for the GCNA marker ( Figure 2C).
We then tested if GCNA is important for the maintenance of the USG population in mice. The USG population in adult animals (10 weeks old) was first assessed by quantification of PLZF by western blot. Interestingly, we observed reduced PLZF expression in the GCNA-deficient testes (Supplementary Figure S5A and B). In contrast, we did not observe any reduction of the expression of the Sertoli cell marker Wilms tumor 1 (WT1) suggesting that the Sertoli cell population is not altered in Gcna knock-out male mice (36) (Supplementary Figure S5A and C). These results suggest a reduction in the number of USGs or reduced expression of PLZF.
We therefore assessed the number of USGs in GCNAdeficient male mice and how this number changes over mice lifetime. We counted PLZF positive USGs in tubes exhibiting spermatogenesis, excluding SCO tubes, at different ages. As observed with the testis mass and the sperm concentration, a time-dependent decrease in the USG number was found, first apparent at 3 months of age ( Figure 2D and E). Consistent with this result, there was an age-dependent increase in the proportion of seminiferous tubes lacking any PLZF positive USG ( Figure 2F). In order to confirm the specificity of the immunostaining used in this study we employed a PLZF reporter in which the expression of td-Tomato is driven by the endogenous PLZF-promoter (23). Quantification of tdTomatdo positive cells in 6-month-old GCNA-deficient mice revealed a significant reduction in the number of PLZF positive cells per tubule and an increase in the frequency of tubules with no PLZF positive cells (Supplementary Figure 5D and E). The magnitude of defect observed using the Plzf-tdTomato reporter was comparable to that obtained by PLZF immunostaining. These data suggest that GCNA is required to maintain the USG population over time in mice, which permits the continuous production of sperm.

Neither apoptosis nor premature differentiation drive agedependent loss of USGs in Gcna knock-out males
As USGs in GCNA-deficient male mice are lost in an agedependent manner, we sought to elucidate how these cells are lost. As USGs can either be lost through apoptosis or differentiation, we first tested these two fates (37,38). We initially assessed apoptosis in testis of GCNA deficient males by measuring the frequency of cleaved-caspase 3 (CC3) positive cells (Supplementary Figure S6A-C). We could not detect any increase in CC3 cells in either wild-type or GCNAdeficient testes, at 1.5 or 6 months old. As USGs make up only a small proportion of cells in the testis we next assessed if there were an accumulation of apopotic USGs, but we could not detect any difference between wild-type and GCNA-deficient males (Supplementary Figure S6D and E). We then tested a potential clearance of defective USGs through increased differentiation rate by assessing the frequency of USGs expressing DNMT3B, a factor that drives differentiation of USGs (39). Frequencies of differentiating USGs were similar between wild-type and GCNAdeficient animals (Supplementary Figure S6F and G). Together, these results suggest that neither increased apoptosis nor differentiation of USGs is the origin of the agedependent loss of USGs in Gcna knock-out males.

GCNA is necessary for S-phase normal progression in murine USGs
As apoptosis or differentiation do not explain the USG loss in absence of GCNA, we tested an alternative hypothesis. It has been shown that reduced quiescence triggers premature exhaustion of the USG pool (35,40). To test if this mechanism could explain the age-dependent loss of USGs in Gcna knock-out males, we assessed the frequency of USGs negative for the proliferation marker Ki67 ( Figure 3A). Interestingly, whilst the proportion of quiescent USGs increases with age in wild-type mice, no such increase is observed in the absence of GCNA ( Figure 3B). These results suggest that GCNA-deficient USGs remain more frequently in the cell cycle, unlike wild-type USGs.
As GCNA-deficient USGs appear committed to the cell cycle, we tested if these cells are also frequently dividing. We therefore assessed the frequency of USGs positive for the M phase marker H3 pSer10. Surprisingly, old Gcna knockout male mice (>6 months old) do not exhibit an increased frequency of H3 pSer10 positive USGs compared to wildtype, indicating that GCNA-deficient USGs are not dividing more frequently ( Figure 3C and D). The observation that more USGs are engaged in the cell cycle but are not dividing more suggests that GCNA-deficient USGs may be accumulating in a particular phase of the cell cycle.
We consequently tested if USGs accumulate in S/G2, by using the cyclin A2 (CCNA2) marker. Interestingly, we found a significant increase in CCNA2 positive USGs, in old GCNA-deficient male mice suggesting an accumulation in S/G2 ( Figure 3E and F). We then decided to narrow down the USG defect by looking more closely at Sphase using the incorporation of the nucleotide analogue named 5-Ethynyl-2 -deoxyuridine (EdU). Strikingly, we observed a significant increase in the frequency of EdU positive USGs in old GCNA-deficient male mice suggesting that these USGs are more frequently in S-phase than wildtype USGs ( Figure 3G and H). Together with the previous data on the cell cycle properties of GCNA-deficient USGs, this result suggests that there is likely an elongation of Sphase in USGs of old mice lacking GCNA.

Mouse GCNA associates with the replication machinery
Given our observations, we sought to understand how GCNA can support the progression of S-phase. Therefore, we tested if GCNA associates with chromatin, specifically during S-phase. Subcellular fractionation experiments showed that endogenous mouse GCNA (mGCNA) was observed in the chromatin-associated fraction of asynchronous cultures of mouse embryonic stem cells (mESCs) ( Figure 4A). Additionally, we observed that this association with chromatin was sensitive to DNAse treatment. Then, we permeabilised 3T3 cells ectopically expressing mGCNA prior to fixation and immunofluorescence. This pre-extraction step allows the visualisation of proteins that remain tightly bound to chromatin. Using this protocol, we observed that mGCNA remains nuclear after preextraction and this is DNA-dependent ( Figure 4B). We then focused our attention on cells in S-phase. Cell fractionation of synchronised cells showed that mGCNA remains in the chromatin-associated fraction of cells in S-phase (Figure 4C). We also confirmed by immunofluorescence that mGCNA remains in the chromatin of replicating cells (EdU positive) ( Figure 4D and E). Taken together, these experiments strongly suggest that mGCNA is a chromatin associated protein and this association remains in S-phase.
Our experiments show that the association between mGCNA and chromatin is sensitive to DNAse treatment. We therefore tested if mGCNA could bind DNA by DNA pulldown. Mouse SPRTN was used as a positive control as it is a paralog of GCNA and it has been shown to bind single-and double-stranded DNA (ssDNA and dsDNA) (41). We used a catalytically inactive SPRTN (E113Q) as binding of SPRTN with DNA activates its protease activity and triggers autocleavage of the protein, making SPRTN detection difficult. We did not observe any interaction between mGCNA and DNA, suggesting that mGCNA could be recruited to chromatin through an interaction with a partner (Supplementary Figure S7).
As GCNA is associated with chromatin in S-phase but does not directly bind to DNA, we sought to investigate a functional interaction of mGCNA with the DNA replication. We first observed that expression of mGCNA reduces the proliferation of 3T3 cells ( Figure 4F). Then, we set out to test if this reduced proliferation could be due to mGCNA interfering with S-phase progression. We therefore investigated the DNA replication rate of 3T3 cells by measuring the incorporation of EdU, in replicating cells. Microscopic analysis showed that ectopic mGCNA expression reduces the incorporation of EdU ( Figure 4G and H). This result suggests that mGCNA interferes with DNA replication when it is expressed in cells.
Given the S-phase association of mGCNA with chromatin, its interference with replication but lack of direct DNA binding, we sought to directly test if there was an association of mGCNA with the DNA replication machinery. To answer this question, we performed isolation of proteins on nascent DNA in mESCs (iPOND) (28). In this assay, cells are first grown in media supplemented with EdU to label nascent DNA synthesis (the pulse). The EdU is then replaced with thymidine in order to perform a chase. After, cells are fixed to crosslink DNA with its associated proteins. Biotin-conjugation of the EdU is performed by a click chemistry reaction allowing the EdU-labelled DNA and its associated proteins to be purified by streptavidin pulldown. Finally, proteins associated to the EdU-labelled DNA are identified by Western-blot. True replication-associated proteins should be detected only in the pulse sample and not in the chase sample. Ubiquitously chromatin-bound proteins, such as H3, remain detectable both at the replication fork, i.e. after EdU pulse, and in thymidine-chased samples ( Figure 4I). However, replication-associated proteins like PCNA are enriched on the nascent DNA, after the EdU pulse. Interestingly, we observed that GCNA associates with nascent DNA, as does PCNA. Furthermore, the enrichment of both proteins is gradually lost from the nascent DNA during the thymidine chase. These results suggest that, as PCNA, GCNA can associate with the DNA replication fork.
We then sought to assess if GCNA associates with DNA synthesis in testes, where we have shown a physiological  role of GCNA. To test this, we analysed its co-localisation with PCNA and observed a high frequency of germ cells coexpressing both PCNA and GCNA in adult testes (Supplementary Figure S8A and B). Cells co-expressing both factors are predominantly in the pre-spermatid stages of the spermatogenesis. Therefore, we tested if the expression of GCNA correlates with PCNA expression in pre-spermatid cells. We observed a statistically significant and positive correlation between PCNA and GCNA expression levels (Figure 4J, Supplementary Figure S8C). This result suggests that the expression of GCNA is regulated similarly to the expression of PCNA in pre-spermatids, suggesting a potential role during DNA replication and DNA synthesis. Given these results, we sought to test whether GCNA and PCNA are in close proximity in pre-spermatid cells. Consistent with the co-expression pattern, we observed by proximity ligation assay (PLA) that pre-spermatid cells exhibited specific signals for the proximity of PCNA and GCNA ( Figure 4K and L). These results suggest that PCNA and GCNA associate in pre-spermatid cells, potentially during DNA replication and DNA synthesis.

GCNA is dispensable for DPC repair in mice
In invertebrates, GCNA is required for the repair of DNAprotein crosslinks (DPCs) during embryonic development and in the germ line (19)(20)(21). DPCs impede S-phase progression by inducing replication fork collapse, eventually leading to formation of DNA breaks (42). Here, we observe that mGCNA associates with the DNA replication, a localisation compatible with a role in DPC repair. Therefore, we tested if the absence of GCNA leads to DNA breaks formation in USGs, explaining their accumulation in S-phase. To do this, we assessed the presence of DNA breaks in USGs by using the ␥ H2AX marker. We did not observe any difference between wild-type and GCNA-deficient male mice in the frequency of USGs positive for ␥ H2AX nor in the ␥ H2AX signal intensity in USGs, at 1.5 or 6 months old ( Figure 5A-C and Supplementary Figure S9). These results suggest that GCNA-deficient USGs are not likely accumulating in S-phase because of DNA breaks.
A previous study reports that a fraction of GCNAdeficient spermatocytes in late meiotic prophase I retain DNA breaks, suggesting that GCNA may have a role in DNA repair in mice (20). Meiotic cells with unresolved breaks are typically cleared by apoptosis (reviewed in (43,44)). We therefore tested if GCNA-deficient germ cells exhibit high levels of apoptosis ( Supplementary Figure S6A-C). We could not detect any difference in the frequency of apoptotic germ cells between wild-type and GCNA-deficient testes. In agreement with this result, we observed normal sperm counts in 1.5 and 3 months old GCNA deficient mice (Figure 1). All these results suggest that (i) meiosis can proceed normally in GCNA-deficient male mice and (ii) GCNA-deficient spermatocytes are not cleared by apoptosis and therefore resolve meiotic DNA double-strand breaks.
Given our results, we sought to directly test if GCNA is involved in DPC repair in mice. To assess the putative role of GCNA in DPC repair in mouse cells, we took advantage of the fact that mESCs express GCNA. Loss of GCNA sen-sitises C. elegans to formaldehyde, which crosslinks a broad spectrum of proteins to DNA (21). Therefore, we derived mESCs from wild-type or Gcna − /Y embryos and exposed them to formaldehyde (Supplementary Figure S10A-C and Figure 5D). No significant difference in cellular formaldehyde resistance was observed between GCNA-deficient and wild-type mESCs.
We then asked if mGCNA could be involved in the repair of a specific and frequently formed class of DPC. It has been previously shown that mGCNA interacts with Topoisomerase 2 (TOP2) (20) suggesting a role in the repair of DNA crosslinked to TOP2. We first confirmed by immunoprecipitation that mGCNA interacts with TOP2 (Figure 5E). We then exposed mESCs to etoposide, which induces DNA-TOP2 crosslinks, and assessed their sensitivity. Again, no significant difference was observed between GCNA-deficient and wild-type mESCs ( Figure 5F). These results suggest that mGCNA is dispensable for maintaining cellular resistance to two archetypal DPC-inducing agents, in contrast to the reported role of GCNA in invertebrates. A common consequence of unrepaired DNA damage is perturbation of the cell cycle, we therefore assessed if loss of GCNA altered the cell cycle distribution. We performed 2D cell cycle analysis in wildtype and GCNA deficient mESCs and found no significant difference in the distribution of cells across the cell cycle (Supplementary Figure 10D and E).
Human GCNA can form nuclear foci after formaldehyde exposure in U2OS cells (21). We therefore exposed 3T3 cells stably expressing mGCNA to formaldehyde and assessed its chromatin localisation. We observe that formaldehyde exposure does not change the chromatin localisation of mGCNA and the protein does not form foci ( Figure 5G and H). Furthermore, formaldehyde exposure does not alter the association of mGCNA with chromatin in mouse cells ( Figure 5I). These data suggest that the chromatin localisation, and role, of mGCNA are DPC-independent.

Mouse GCNA is closely related to the intrinsically disordered region (IDR) of human GCNA
As mGCNA appears dispensable for DPC repair, we sought to uncover the molecular functions that may be required for USG self-renewal. To gain insight on this, we first turned to the domain structure of GCNA. GCNA is conserved from Schizosaccharomyces pombe to humans and is characterised by the presence of an N-terminal intrinsically disordered region (IDR) and a C-terminal SprT protease domain (18). However, in agreement with the dispensable role in DPC repair, bioinformatic analysis reveals that GCNA in rodents does not possess the SprT protease domain which is essential for DPC repair in invertebrates (18,19) (Supplementary Figure S11A). In order to validate the in silico annotation of the mouse gene, we first tested if the endogenous protein in the testis and mESCs had the expected molecular mass of 53 kDa. Using the two characterised antibodies directed against GCNA, i.e. GCNA-1 and Tra98, we observed that the protein migrates between 75 and 100 kDa, consistent with previous reports, but larger than the predicted 53kDa mass (Supplementary Figure S11B and C) (14,15,18). Consequently, the cDNA corresponding to the annotated mouse Gcna locus was expressed in a heterologous expression system, E. coli (Supplementary Figure  S11B and C). We observed the same migration of the recombinant protein which strongly suggests that the annotation of mGCNA is correct and the protein migration is retarded, likely due to its highly acidic content.
As mGCNA is disordered with highly acidic content, we sought to study the phylogeny of GCNA IDRs among metazoans, particularly focusing our attention on the amount and distribution of charged residues in those IDRs. Interestingly, high content of negatively charged residues is a feature conserved in most mammalian GCNAs (Supplementary Figure S11D). Furthermore, GCNA in mammals is characterised by a large acidic region flanked by a basic one (Supplementary Figure S11E). This feature was not observed in other metazoan GCNAs examined. The conservation of the acidic region in the IDR of GCNA in mammals, plus the absence of the SprT domain in mouse GCNA suggest that i) the acidic nature of the IDR of GCNA is important for its function and ii) murine GCNA is a separationof-function model for the study of the IDR of GCNA in vivo.

Mouse and human GCNA bind histones
Our results suggest that mouse GCNA is highly acidic and is a chromatin-associated protein. We therefore hypothesized that the association of mGCNA with chromatin could rely on electrostatic interactions. To test this hypothesis, we washed the mESC chromatin fractions with increasing concentrations of salt and asked if mGCNA remained bound ( Figure 6A). Interestingly, mGCNA was almost completely lost at high salt concentrations, supporting an interaction of mGCNA with chromatin through electrostatic interactions.
Histones are a major component of chromatin and their interaction with DNA relies on charged-based interactions. Since histones are highly basic and mGCNA is highly acidic and associates with chromatin through electrostatic interactions, we hypothesised that mGCNA may interact with chromatin by binding to histones. We therefore tested if mGCNA physically interacts with histones by immunoprecipitation. As indirect chromatin interactions may confound our results, these experiments were conducted using the soluble, chromatin-free fraction obtained after cell fractionation. Additionally, samples were treated with nucleases to eliminate any potential contamination with DNA. We observed that mGCNA can bind to H3 and H2B ( Figure 6B-D). Histones were likely to be in the form of stable and soluble H3-H4 tetramers and H2A-H2B dimers as we used a nucleosome-free cell fraction during the immunoprecipitation experiments (45). Therefore, this result suggests that mGCNA can potentially bind H3-H4 tetramers and H2A-H2B dimers. Also note that mGCNA co-immunoprecipitated with two forms of both H3 and H2B, which appear as doublets by Western blot (Figure 6C and D). These doublets potentially represent post-translational modifications of N-terminal tails of histones or clipping of these tails (46). These results suggest that mGCNA also interacts with modified H3 and H2B. We could not observe any interaction between H1 and mGCNA, suggesting that the interaction between mGCNA and histones is specific to the core nucleosome components (Supplementary Figure S12). We then sought to test if human GCNA (hGCNA) also interacts with histones and found that it also interacts with H3 and H2B ( Figure 6B, E and F). It is worth noting that hGCNA also interacts with modified H2B ( Figure 6F). Moreover, we observed that the IDR of hGCNA, which is conserved with mGCNA, is sufficient for the interaction with H3 and H2B ( Figure 6B, E and F).
Next, we analysed if the interaction between mGCNA and core histones also occurred in vivo. We observed by PLA that mGCNA is in close proximity to H3, from spermatogonia to round spermatids ( Figure 6G). This result suggests that the chromatin association and histone binding of mGCNA occur in the germ line too.

Mouse GCNA is a histone chaperone
Histones-binding proteins such as histone chaperones are characterised by highly acidic and disordered regions akin to that observed in mGCNA (47). Given that mGCNA and hGCNA bind core histones, we sought to test if mGCNA exhibited histone chaperone activity. For this purpose, we purified N-terminally MBP-tagged mGCNA from human cells (Supplementary Figure S13). The MBP affinity tag was chosen over other tagged mGCNA forms used previously in this work to enhance the solubility of mGCNA during the purification protocol and to produce higher yields of the protein (48). A key property of histone chaperones is the ability to directly bind core histones. We therefore tested whether mGCNA could directly bind core histones. Limited proteolysis experiments showed increased protection of H3-H4 tetramers and H2A-H2B dimers in presence of MBP-mGCNA compared to the MBP control ( Figure  7A lanes 4 and 8, B and C lanes 4 and 8, D). These results suggest that mGCNA can directly interact with core histones and support the immunoprecipitation experiments presented earlier.
We then sought to assess the potential of mGCNA to act as a histone chaperone. Therefore, we tested the capacity of MBP-mGCNA, in the presence of H3-H4 tetramers and H2A-H2B dimers, to induce histone deposition using a plasmid supercoiling assay ((49); Figure 7E). In this assay, a plasmid is first relaxed by using topoisomerase 1. Then, the relaxed plasmid is incubated with histones and the putative histone chaperone. Deposition of histones into DNA induces supercoiling of DNA and DNA supercoils remain after deproteinisation of the plasmid. Finally, supercoiling of the plasmid is monitored by gel electrophoresis. MBP or MBP-mGCNA alone showed no supercoiling activity on the relaxed circular plasmid DNA ( Figure 7F, lanes 8 and  9). Histones alone or with MBP showed little supercoiling activity on the plasmid DNA ( Figure 7F, lanes 3 and 4). Interestingly, we observed that addition of increasing concentration of MBP-mGCNA to core histones progressively increased plasmid supercoiling ( Figure 7F, lanes 5-7 and Figure 7G). The plasmid supercoiling is enhanced in presence of mGCNA, compared to the tag-only control, suggesting a specific activity of mGCNA ( Figure 7F, lanes 4 and 7 and Figure 7G). These results indicate that mGCNA is able to  promote core histone deposition onto DNA as a bona fide histone chaperone.

DISCUSSION
The expression of GCNA is restricted to germ line and embryonic stem cells in metazoans suggesting a role during germ cell and embryonic development. The presence of a SprT domain of GCNA shared with the SPRTN protease, implicated GCNA, including the human GCNA protein, in the repair of DNA-protein crosslinks (DPCs) to maintain genome stability. Indeed, studies in C. elegans, Drosophila and zebrafish provide evidence for a role of GCNA in the repair of DPCs in the germline and during early development (19,20). However, the absence of the SprT protease domain in mouse GCNA (mGCNA) suggests a function independent of DPC repair. Consistent with this, we have found that GCNA-deficient mESCs are not hypersensitive to DPC-inducing agents.
Although lacking the SprT domain, mGCNA shares a conserved intrinsically disordered region (IDR) with other GCNAs, including human GCNA. We therefore used the mouse protein as a model to explore these protease independent roles of GCNA. Our results show that mGCNA associates with chromatin, even in the absence of DNA damage. In addition, GCNA is highly acidic and predicted to be disordered akin to histone chaperones and we find that mGCNA directly binds core histones and also possesses histone chaperone activity in vitro. Moreover, we find that mGCNA is present at sites of nascent DNA synthesis, similar to other histone chaperones (50). Interestingly, GCNA interacts with components of the MCM2-7 complex in Drosophila suggesting that this association of GCNA with DNA replication is evolutionarily conserved (19). Consistent with DNA replication association and histone binding activity, we find that mGCNA ectopic expression causes alterations in the progression of DNA replication. Together, these data suggest that mGCNA is involved in chaperoning of histones during DNA replication.
Supporting this, we find that in vivo, the loss of GCNA leads to the accumulation of USGs in S-phase. In contrast, GCNA deficient mESCs do not show accumulation in Sphase. This may represent cell type specific functions of GCNA, differences in chromatin landscape between USGs and mESCs, or functional renduancy in mESCs. The accumulation of GCNA-deficient USGs in this stage of the cell cycle could be explained by the loss of histone chaperone activity. Firstly, mGCNA may be important in chaperoning parental histones evicted by the MCM 2-7 complex and/or chaperoning newly synthesised histones, promoting their loading after the replication machinery (Figure 8, see (1)). Whether mGCNA acts to recycle histones or in de novo histone deposition remains to be determined. In the absence of mGCNA, parental and/or new histones may be less efficiently chaperoned, slowing fork progression or leading to aberrant positioning of histones, e.g. leading to regions of single-stranded DNA (Figure 8, see (1 )). Interestingly, in a recent study in S.cerevisiae, authors proposed that the SPRTN/GCNA homolog Wss1 targets the degradation of histones bound non-covalently to single-stranded DNA, preventing impaired fork progression (51). There-fore, GCNA in rodents may have specialised in one ancestral function of the Wss1/SPRTN family which is the support of the management of histones during DNA replication.
Alternatively, mGCNA may promote DNA replication through the interaction with TOP2. We observed that whilst mGCNA is dispensable for the repair of TOP2 crosslinked to DNA, GCNA does interact with TOP2. During replication, positive supercoils are formed ahead of the replication fork as the MCM 2-7 complex progresses. To prevent arrest of the replication fork due to high topological constraints, TOP2 relaxes the DNA (52). However, in less accessible regions of chromatin, TOP2 may cooperate with a histone chaperone to have access to the DNA and we propose that GCNA could potentially help TOP2 at this step ( Figure 8, see (2)). In the absence of mGCNA, TOP2 may have less access to DNA in nucleosome-rich regions, causing the replication fork to progress more slowly and impairing S-phase progression ( Figure 8, see (2 )).
The loss of GCNA impacts male but not female germ cell production. GCNA-deficient male mice initially have intact spermatogenesis. However, this becomes compromised upon ageing. We hypothesised that this long-term failure to sustain spermatogenesis may be due to a defect in the USG pool. This is consistent with our observation that whilst GCNA is expressed during embryonic germ cell development no numerical defect is observed until males are aged. This differential requirement for GCNA in USGs and PGCs may be explained by the dramatic differences in chromatin landscape between these two stages of germ cell development (53)(54)(55). As adult females do not have an equivalent to USGs, this may explain why the phenotypes are restricted to males. GCNA-deficient males initially have normal number of USGs but this number decreases upon ageing, unlike in wild-type mice. This could be explained by either apoptosis, increased USG differentiation, failure of USG self-renewal, or a combination of these possibilities. We did not detect any increase in apoptosis or potentiated differentiation in the USG pool of GCNAdeficient mice. However, in the absence of GCNA, aged USGs show impaired S-phase progression. In mammals, the fine-tuned balance between self-renewal and differentiation of the USGs allows the continuous production of gametes throughout life ( Figure 8). As old GCNA-deficient USGs replicate less due to impaired S-phase progression, they are more likely to be lost through normal differentiation rate. Consequently, the USG pool is progressively depleted, and the production of gametes is reduced in a time-dependent manner.
Murine GCNA represents an attractive separation-offunction model to interrogate its human homolog. Indeed, we observe that the IDR of human GCNA (hGCNA) shows similarities with its murine homolog, with a highly acidic region. Akin to the murine protein, hGCNA interacts with core histones. Furthermore, we observe that the IDR of hGCNA is sufficient for this interaction. However, hGCNA retains a SprT domain, like the invertebrate homologs. So, one could ask what is the molecular role of hGCNA in the germ line. We propose that hGCNA targets histone proteolysis thanks to the coupling of the binding activity of its IDR and the protease activity of the SprT domain.  (1) and (2)). First, mGCNA is chaperoning histones that are then loaded after DNA synthesis (role (1)). Note that for simplification reasons, mGCNA is depicted as acting on the leading strand. Mouse GCNA could act as an histone chaperone on both strands, during DNA replication. Second, mGCNA interacts with TOP2 and supports relaxation of DNA supercoils ahead of the replication fork, in nucleosomes-rich regions (role (2)). Mouse GCNA supports the progression of S-phase in USGs and therefore, supports their self-renewal. Consequently, USGs are able to maintain homeostasis in the testis and mature sperm cells are produced through the mice life. However, in the absence of mGCNA, S-phase progression is slowed down because of a less efficient chaperoning of histones (see (1 )) or/and an accumulation of supercoils ahead of the replication fork (see (2 )). Thus, self-renewal of USGs is impaired, leading to their loss. Consequently, USG loss in the absence of GCNA triggers an age dependent reduction of the sperm production in males.
Functionally, GCNA appears to be of crucial importance for the maintenance of the human male germ line. Indeed, two recent reports describe the phenotype of several azoospermic men carrying mutations in the GCNA locus (16,17). These mutations may lead to loss of GCNA protein but it is worth noting that some mutations occur within the IDR. Strikingly, histological analysis of testis biopsies from two patients with GCNA mutations revealed a Sertoli-cell only phenotype in these two men (17). This re-sult strongly suggests that GCNA in humans may preserve the USG pool, like we characterise here in mice. Therefore, we show here for the first time that the mouse model is relevant for the study of GCNA as it phenocopies mutations of GCNA in humans, despite the absence of the SprT domain in the mouse protein. Moreover, this result strongly suggests that the IDR of GCNA plays a critical role in the function of GCNA during mammalian spermatogenesis. Finally, we propose that GCNA in mammals is a histone binding pro-tein required for USG maintenance, through its roles in the support of DNA replication.

DATA AVAILABILITY
All data generated or analysed during this study are included in this published article (and its supplementary data files).