DNA oxidation induced by fetal exposure to BPA agonists impairs female meiosis

Many endocrine disruptors have been proven to impair the meiotic process that is mandatory to produce healthy gametes. Bisphenol A is emblematic as it impairs meiotic prophase I and causes oocyte aneuploidy following in utero exposure. However, the mechanisms underlying these deleterious effects remain poorly understood. Furthermore, the increasing uses of BPA analogs raise concerns for public health. Here, we investigated the effect on oogenesis in mouse of fetal exposure to two BPA analogs, Bisphenol A Diglycidyl Ether (BADGE) or Bisphenol AF (BPAF). These analogs delay meiosis initiation, increase MLH1 foci per cell and induce oocyte aneuploidy. We further demonstrate that these defects are accompanied by a deregulation of gene expression and aberrant mRNA splicing in fetal premeiotic germ cells. Interestingly, we observed an increase in DNA oxidation after exposure to BPA analogs. Specific induction of oxidative DNA damages during fetal germ cell differentiation causes similar defects during oogenesis, as observed in 8-Oxoguanine DNA Glycosylase (OGG1) deficient mice or after in utero exposure to potassium bromate (KBrO3), an inducer of oxidative DNA damages. Moreover, the supplementation of N-acetylcysteine (NAC) with BPA analogs counteracts the bisphenol-induced meiotic effect. Together our results position oxidative stress as a central event that negatively impacts the female meiosis with major consequences on oocyte quality. This could be a common mechanism of action for so called endocrine disruptors pollutants and it could lead to novel strategies for reprotoxic compounds.


Introduction
In females, aneuploidy (aberrant number of chromosomes) is an important cause of adverse reproductive outcomes such as miscarriages and congenital abnormalities. Aneuploid eggs can be induced by numerous lifestyle factors (age, obesity, environmental pollutants…) (Nagaoka et al., 2012) and can derive from alterations occurring during meiotic prophase I in fetal life. Indeed, proper chromosome segregation at adulthood requires organized reciprocal DNA exchanges between homologous chromosomes (crossover) occurring during prophase I as a result of the repair of meiotic double strand break (DSB) by homologous recombination (Patricia A. Hunt & Hassold, 2008;Ottolini et al., 2015;. Crossover regulation depends on a correct implementation of the meiotic program in primordial germ cells (PGCs) initiated after their migration into the gonad. At this stage, pluripotent and proliferative PGCs acquire the competence to initiate meiosis through the expression of Deleted In Azoospermia-like (Dazl) (Nicholls et al., 2019). Depending on the somatic environment and under the control of meiotic orchestrators such as Stra8 that direct the switch from mitosis to meiosis, female PGCs initiate prophase I at 13.5 day post-conception (dpc) (Bailey et al., 2017;Hargan-Calvopina et al., 2016a;Ishiguro et al., 2020;Le Bouffant et al., 2010;Spiller & Bowles, 2019;Trautmann et al., 2008). Alterations that occur during the establishment of the meiotic program lead to meiotic defects in prophase I that can hamper future fertility (Bailey et al., 2017;Hargan-Calvopina et al., 2016a;Ishiguro et al., 2020;Nicholls et al., 2019). It is well known that implementation and progression of prophase I are extremely sensitive to environmental factors such as toxicants and endocrine disrupting chemicals. Among those, bisphenol A (BPA) is the first and the most studied environmental compound known to alter meiosis and folliculogenesis in females in numerous mammalian and non-mammalian organisms (Brieno-Enriquez et al., 2012;Brieño-Enríquez et al., 2011;Lawson et al., 2011;Susiarjo et al., 2007;T. Zhang et al., 2014).
In female primates and rodents, fetal exposure to BPA induces alteration of the expression of meiotic genes at the time of meiosis onset and during prophase I (Brieno-Enriquez et al., 2012;Lawson et al., 2011;. Moreover, fetal exposure to BPA alters the distribution of recombination events signaled by MLH1, a DNA mismatch repair protein required for

BADGE or BPAF fetal exposure reproduces BPA defects on folliculogenesis and recombination events.
In mammals, fetal exposure to BPA disrupts follicle formation causing multioocyte follicles and increases the incidence of missegregation after meiosis resumption in post-natal ovaries. To study the effects of analogs of BPA on oocyte and folliculogenesis at adulthood, we exposed pregnant mice to 10 µM of BADGE or BPAF in drinking water from 10.5 to 18.5 days post-conception. Haematoxylin and eosin staining of ovarian section from 3 months old mice exposed during fetal life to vehicle (ETOH), BADGE (BADGE) or BPAF (BPAF). Scale bar, 200µm or 10µm for the higher magnification (multi-oocyte follicle). (B) Quantification of total follicle count in bisphenol-treated or control adult ovaries. Each bar represents the percentage of primordial, primary, secondary and antral follicles. (C) Percentage of multioocytes follicle in bisphenols-treated or control post-natal (8 dpp) and adult (3 months) ovaries. Error bars indicate mean ± s.e.m. n=3-5 mice from 3 independent exposures, * p <0.05 (Mann-Whitney's test).
This defect in meiosis progression could be the consequence of a delay in meiosis initiation.
Immunostainings for BrdU, SYCP3 and a germ cells marker (TRA98), were used to identify three germinal populations 14.5 dpc ovaries: oogonia, the female PGCs (SYCP3-negative cells), premeiotic cells in S-phase also called pre-leptonema (BrdU-positive/SYCP3-positive) and oocytes (BrdU-negative/SYCP3-positive; Figure 3B-C). In untreated 14.5 dpc ovaries almost all germ cells had initiated meiosis (over 96% are SYCP3+) and very few oogonia and preleptotene cells were still present. In bisphenols-treated mice, we observed a significant increase in mitotic PGCs (Supplementary Figure 1A) and pre-leptonema while oocyte number was reduced ( Figure 3C).  178  179  180  181  182  183  184  185  186  187  188  189 190 191 192 193 194 195 196 197 198 199   In addition, we observed an increasing trend for STRA8 positive germ cells in 14.5 dpc female gonads (20% ± 4 in ETOH group VS 36% ± 1 p=0.08 and 39% ± 6, p=0.07 in BADGE and BPAF groups respectively, Supplementary Figure 1B). The presence of STRA8 correlated with the initiation of the meiotic program and declines rapidly just after the initiation in prophase I. The observed increase confirmed the bisphenols-induced delay of meiotic initiation. A defect of crossover distribution could be the consequence of a delay and/or an alteration of meiosis initiation. To confirm this hypothesis, we performed exposure to bisphenols until meiosis initiation (short term exposure).

Supplementary Figure 2:
Ovaries from fetuses exposed to vehicle (ETOH) or bisphenols (BADGE and BPAF) from 10.5 dpc to 14.5 dpc were used for the MLH1 quantification (short exposure).

BADGE or BPAF fetal exposures alter gene expression and mRNA splicing in germ cells
In order to understand the bases of bisphenols-induced alterations in PGCs during acquisition of meiotic competence and initiation of the meiotic program, we performed transcripomic analyses on 11.5 dpc (during acquisition of the meiotic competence) and 13.5 dpc (during meiosis initiation) germ cells. We sorted PGCs by Magnetic Activated Cell Sorting (MACS) using the cell surface protein stage-specific embryonic antigen 1 (SSEA-1). Gene expression analysis was conducted using murine Affymetrix GeneChip Gene 2.0 TS (11.5 dpc) and Clariom TM D (13.5 dpc) microarrays. At 11.5 dpc, we identified 1817 and 733 differentially expressed genes (DEGs) in BADGE and BPAF treated germ cells respectively (compared to vehicle treated PGCs with a |Log2 Fold-Change| ≥ 0.5 and p < 0.05) (Figure 4). DEGs were mostly downregulated (56 to 65 % of DEGs; Figure 4A). At 13.5 dpc when the program of meiosis is initiated, we identified 886 and 1376 DEGs in BADGE and BPAF treated germ cells respectively ( Figure 4A). Contrary to what it is observed at 11.5 dpc, two third of the DEGs were upregulated in 13.5 dpc bisphenols-treated germ cells ( Figure 4A). Using EnrichGO function from Clusterprofiler package to identify enrichment of gene ontologies, we observed that DEGs at 11.5 dpc as well as 13.5 dpc were mostly related to meiosis, stem cell differentiation and regulation of stem cell signaling pathway such as Wnt pathway ( Figure 4B and Supplementary Figure 3). Meiosis-associated DEGs such as Stag3, Sycp1&3,Hormad1&2,Spo11,Spata22,Meiob,Dmc1 or Brca2 were strongly enriched in genes associated to synapsis and DNA recombination processes ( Figure 4A-C, Supplementary Table 1).
Surprisingly, genes linked to meiosis and stem cell differentiation showed opposite transcriptional response at 11.5 dpc and 13.5 dpc (Supplementary Figure 3 and Figure 4C). Stemness genes tend to be up-regulated at 11.5 dpc and mostly down-regulated at 13.5 dpc and meiotic genes were preferentially down-regulated at 11.5 dpc and up-regulated at 13.5 dpc ( Figure 4C). : BADGE or BPAF fetal exposures induce transcriptional alteration before meiosis initiation. SSEA1 positive germ cells from 11.5 or 13.5 dpc ovaries treated by vehicle (ETOH) or bisphenols (BADGE and BPAF) were used for transcriptomic analyses. Differentially expressed genes (DEGs) between bisphenols and control condition (BPAF vs ETOH or BADGE vs ETOH) were filtered according to the log fold change (abs[Log2FC] ≥ 0.5) and the significativity (p < 0.05). Three pools of cells each of them from 10-15 fetuses were analyzed. (A) The table represents the number (and percentage) of DEGs after BADGE or BPAF exposure in 11.5 or 13.5 dpc germ cells. (B) Gene Ontology enrichment associated to meiosis and stem cell differentiation among DEGs. (C) Correlation analysis between BADGE and BPAF condition of the differential expression (Log2FC) of genes associated to meiosis or stem cell differentiation. r= Pearson's correlation coefficient (Pearson's test).  canonical Wnt signaling pathway cell junction assembly cell junction organization cell−cell junction assembly cell−cell junction organization cell−cell signaling by wnt cell−matrix adhesion chromosome organization involved in meiotic cell cycle chromosome segregation cofactor metabolic process collagen−containing extracellular matrix condensed chromosome condensed nuclear chromosome developmental cell growth developmental growth involved in morphogenesis DNA recombination drug catabolic process embryonic appendage morphogenesis embryonic organ development embryonic organ morphogenesis endocrine system development extracellular matrix female gamete generation gland development gland morphogenesis homologous chromosome segregation hormone metabolic process hormone secretion male meiotic nuclear division meiosis I meiosis I cell cycle process meiotic cell cycle meiotic cell cycle process meiotic chromosome segregation meiotic nuclear division mesenchymal cell development mesenchymal cell differentiation mesenchyme development mesonephric epithelium development mesonephric tubule development mesonephros development negative regulation of canonical Wnt signaling pathway negative regulation of cell development negative regulation of response to wounding negative regulation of Wnt signaling pathway negative regulation of wound healing Notch signaling pathway nuclear chromosome segregation nuclear division oogenesis organelle fission organic acid binding organic hydroxy compound metabolic process oxidoreductase activity, acting on CH−OH group of donors oxidoreductase activity, acting on peroxide as acceptor oxidoreductase activity, acting on the CH−OH group of donors, NAD or NADP as acceptor oxygen binding oxygen carrier activity positive regulation of epithelial cell proliferation positive regulation of ERK1 and ERK2 cascade positive regulation of reproductive process proximal promoter DNA−binding transcription activator activity, RNA polymerase II−specific regulation of developmental growth regulation of epithelial to mesenchymal transition regulation of stem cell differentiation regulation of transforming growth factor beta receptor signaling pathway regulation of transmembrane receptor protein serine/threonine kinase signaling pathway regulation of Wnt signaling pathway regulation of wound healing renal system process renal system vasculature development retinoic acid receptor signaling pathway sex differentiation SMAD protein signal transduction small molecule catabolic process smooth muscle cell proliferation stem cell development stem cell differentiation steroid biosynthetic process steroid metabolic process synapsis synaptonemal complex synaptonemal complex assembly synaptonemal complex organization synaptonemal structure transforming growth factor beta receptor signaling pathway transmembrane receptor protein kinase activity transmembrane receptor protein serine/threonine kinase signaling pathway Wnt signaling pathway wound healing logpvalue BADGE BPAF

284 285
Interestingly, we observed a significant correlation between BPAF and BADGE of the differential expression of genes associated to meiosis after bisphenols exposure (r=0.37, pvalue < 0.0001, 11.5dpc and r=0.87, pvalue < 0.0001, 13.5 dpc) and stem cell (r=0.32, p<0.0001, 11.5 dpc and r=0.45, p < 0.0001, 13.5 dpc) at 11.5 dpc and 13.5 dpc ( Figure 4C). These positive correlations demonstrate a common transcriptional signature between BPAF and BADGE exposure and suggest similar mechanisms of action for both bisphenols. Interestingly, as observed in the chromosomes 3, 4, 7, 10, 14, and 17, genome mapping of DEGs (up-and down-regulated) showed a concentration of genes on specific genome locations that were differentially expressed after BADGE and BPAF exposure at 11.5 dpc as well as 13.5 dpc (Supplementary Figure 5).
Genes differentially spliced in BPAF or BADGE conditions were not preferentially differentially expressed compared to others genes (Supplementary Figure 4). As observed with non-spliced genes, less than 8 % of spliced genes were also differentially expressed in BPAF or BADGE conditions. Interestingly, we observed closer transcription behaviors after BPAF or BADGE exposure of differentially spliced genes in comparison to non-differentially spliced genes randomly chosen (Supplementary Figure 4). This observation suggests a common alteration impacting the transcriptional response and RNA splicing after BADGE and BPAF exposure.
Supplementary Figure 4: 2d density plot between BADGE and BPAF condition of the differential expression (Log2FC) of differentially spliced genes in BADGE or BPAF condition. The plot area is divided in 200 hexagons and the color of the hexagon correspond to number of genes inside the hexagon. A simple random sampling of 3543 non differentially spliced genes (maximal number of differentially spliced genes) was used to represent transcriptionnal behavior after BADGE or BPAF exposure of non differentially spliced genes.  Figure 5). Therefore, we analyzed the induction of oxidative DNA damages in response to bisphenols in 12.5 dpc proliferative oogonia by detection of 8-OdG ( Figure 6A-B). As a positive control, we added the pro-oxidant potassium bromate (KBrO3) in drinking water from 10.5 dpc. As observed after KBrO3 exposure, bisphenols-exposed PGCs showed a significant increase in 8-OdG when compared to the vehicle-treated germ cells ( Figure 6B).  384  385  386  387  388  389  390  391  392  393  394  395  396  397  398  399  400  401  402  403  404  405  406  407  408  409  410  411  412  413  414  415  416  417  418  419  420  421 To explore the relationship between meiosis initiation delay and oxidative stress, we quantified meiosis initiation in KBrO3 treated ovaries at 14.5 dpc. KBrO3 exposure significantly decreased the proportion of meiotic oocytes (SYCP3 positive cells; Figure 6C).
On the contrary, when we treated pregnant mice with bisphenols supplemented with the antioxidant N-acetylcysteine (NAC) to pregnant mice, we restored the percentage of oocytes to that of vehicle treated (ETOH) 14.5 dpc ovaries. To further asses the impact of oxidative lesions, we used mice deficient for OGG1, a DNA glycosylase involved in the removal of oxidative DNA damage through the Base Excision Repair (BER). Ogg1 -/mice showed increased levels of 8-OdG in proliferative 12.5 dpc oogonia in comparison to Ogg1 +/+ mice (Supplementary Figure 6). The lack of OGG1 also impaired meiosis initiation as we observed a decreasing trend of oocyte number at 14.5 dpc (48.0% ± 3.5, Ogg1 +/+ vs 35.7% ± 1.8, Ogg1 -/-, p=0.0832; Figure 6D). To link oxidative DNA damage and recombination events, we quantified the number of MLH1 foci at 18.5 dpc in OGG1 deficient mice. Interestingly, as observed for bisphenols treated mice, the number of MLH1 foci was significantly increased in the mutant when compared to wild type mice ( Figure 6E). Taken together these data reveal that bisphenols induce oxidative DNA damages which in turn could delay fetal meiosis and induce recombination defects. Figure 6: Quantification of the total area of 8OdG normalized by the nucleus area in 12.5 dpc OGG +/+ and -/oogonia .n= 51 (OGG +/+ ), n=58 (OGG -/-). *** p<0.001 (Mann-Whitney's test).

BADGE or BPAF fetal exposures alter DNA demethylation in PGCs.
BPA has been reported to alter DNA methylation in various cell types and removal of DNA PGC compared to unmethylated/differentiating PGC by immunodetection in 12.5 dpc bisphenols exposed and unexposed ovaries of the 5-hydroxymethylCytosine (5hmC), the first intermediate form during active demethylation ( Figure 7A). After bisphenols exposure, we observed an enrichment of methylated germ cells with a global staining (homogeneous staining) of 5hmC in the nucleus in TRA98 cells ( Figure 7B). This result suggests that bisphenols exposure alters or delays the DNA demethylation in PGC increasing the proportion of methylated germ cells compared to unmethylated ones.  City individuals (7.15 ng/ml for BADGE·2H20 and 2.26 ng/ml for BADGE·H20 versus 1.77 ng/ml for BPA) likely due to similar exposure source to BPA (ie. from epoxy coatings into canned food) (L. Wang et al., 2015). In China, BPAF mean concentration in human plasma is 0.073 ng/mL versus 0.4 ng/ml for BPA in the same samples from 81 individuals in general population (Jin et al., 2018).
For this reason, we investigated the consequences of BPAF and BADGE exposure on female germ cell differentiation. Although BPAF concentration detected in human samples is lower than the one of BADGE or BPA, we chose to expose mice by drinking water to BADGE or BPAF at the same concentrations (ie 10µM). This allows to compare BPAF and BADGE and to compare our results to previous studies, including ours, that used BPA at environmentally doses (Eladak et al., 2018). In the present study, we exposed pregnant mice to bisphenols from 10.5 to 18.5 dpc. The concentration of total BPAF detected in the plasma of the pregnant mice was of the a same order of magnitude (5.96 ± 0.78 ng/ml) than the one we previously observed for BPA (Eladak et al., 2018 Proper establishment and orchestration of the meiotic program are essential to ensure a correct regulation of crossover distribution [57]. The establishment of prophase I requires two events: acquisition of meiotic competence and initiation of the meiotic program . Meiotic competence is acquired after PGCs colonization. Migratory PGCs have a genomic program associated with stemness and express pluripotent genes. In the gonad, at 10.5 dpc, PGCs switch off the pluripotency program and activate genes involved in gametogenesis. DAZL, a germ cellspecific RNA binding protein, binds germ cell specific mRNA and promotes the stabilization of meiotic mRNA allowing the initiation of the meiotic program (Atala, 2012;Hu et al., 2015;Kato et al., 2016;Nicholls et al., 2019). Our transcriptomic analyses revealed that bisphenol exposure alters the acquisition of the gametogenic competence and the initiation of the meiotic program. gametogenesis/meiosis competence. This is illustrated by a global down-regulation of meiosis associated genes such as Sycp1&3 and Hormad2 and an up-regulation of pluripotency associated genes such as Msx1 and Gata4. Second, bisphenols exposure also impacts the initiation of the meiotic program at 13.5 dpc as illustrated by the up-regulation of meiotic genes. These observations are consistent with previous studies on the effect of BPA on human and murine meiosis initiation (Houmard et al., 2009;Lawson et al., 2011). However, it is unclear whether the up-regulation of meiotic genes induced by bisphenols is the consequence of the PGC differentiation delay observed earlier, or a real induction of gene expression and/or RNA stabilization. Indeed, numerous meiotic genes such as Stra8 or Rec8 are drastically downregulated just after meiosis onset. Therefore, a delay of meiosis initiation would lead us to observe a higher level of expression of some meiotic transcripts . Moreover, bisphenol exposure also induced a global alteration of splicing events during this step. Recent studies have highlighted the role of mRNA splicing on the mitosis to meiosis transition in male and female germ cells and lack of splicing regulators in male germ cells leads to meiotic defects Naro et al., 2017;Schmid et al., 2013;. In consequences, minor modifications of the splicing of meiotic genes after bisphenols exposure could impact meiosis. As observed for BPA exposure, changes in level of gene expression were subtle but, interestingly, we observed common transcriptional signature between BPAF and BADGE suggesting common bisphenol molecular targets.
Transcriptional modifications observed after bisphenol exposure impacted the transition from mitosis to meiosis and were certainly the origin of the observed delay of meiosis initiation and progression, the abnormal distribution of MLH1 foci but also the cause of folliculogenesis alterations (ie.MOF, follicle number). Indeed, the initiation of follicle assembly requires completion of meiotic prophase and any delay in meiotic progression could interfere with this process A. Paredes et al., 2005;. Thus all these events may be related Interestingly, common cellular (ie MLH1 and chiasmata distribution and MOF) and transcriptional meiotic signatures observed after bisphenols exposure and other pollutants such as phthalates in multiple organisms suggest a common mechanism of action of these molecules (Allard & Colaiácovo, 2010b;Cuenca et al., 2020;Gely-Pernot et al., 2017;Parodi et al., 2015;Shin et al., 2019;Susiarjo et al., 2007;Tu et al., 2019). In this study, we propose oxidative stress as an new player involved in bisphenol response that could impair the PGC differentiation. One of the major targets of reactive oxygen species is the DNA, and 8OdG is the most prominent lesion in the genome (Ba & Boldogh, 2018). We showed that bisphenol exposure quickly induces the formation of this oxidative DNA lesion in proliferative germ cells. Interestingly, analyzing the genomic landscape of oxidative DNA damage revealed a specific susceptibility to oxidation in genomic regions harboring genes differentially expressed after bisphenol exposure. It is well known that oxidative DNA damages can modulate gene expression directly or undirectly. First, oxidative DNA damage induce antioxidant and inflammatory transcriptional responses and recruiting DNA repair proteins (Hörandl & Hadacek, 2013;Pan et al., 2016). Second, oxidized guanine biases the recognition of methylated CpG dinucleotides and alters the dynamic of DNA demethylation (Gruber et al., 2018;Pan et al., 2016). Interestingly, acquisition of the meiotic competence is completely dependent on an intensive germinal DNA demethylation (Hargan-Calvopina et al., 2016b;Yamaguchi et al., 2012). In this study, we observed that bisphenols exposure interferes with DNA demethylation and could be the consequence of the presence of oxidative DNA damages. The presence of 8OdG has also post-transcriptional consequences and could alter splicing events. Pre-mRNA splicing requires the identification of specific 5' donor and 3' acceptor sites. The 5' and 3' splicing sites mostly begin with dinucleotide GT(U) and end with dinucleotide AG that allows the major 5'_3' combination GT-AG (Calvello et al., 2013). Incorporation of the wrong base within these splicing signals would lead to an alteration of the fidelity of mRNA splicing. In mammalian cells, to the effect of molecules with a xenoestrogenic potential (Cuenca et al., 2020;Susiarjo et al., 2007;Tu et al., 2019). Our hypothesis is in agreement with an involvement of estrogen signaling during this process as DNA binding by the estrogen receptor drives the local production of oxidative DNA damages via LSD1, a Lysine-specific histone demethylase 1, activity in promoter and enhancer regions (Perillo et al., 2008).
Current knowledge on the potential toxicological effects of BPA analogs is limited. We provide here proofs that endocrine disruptors such as bisphenols negatively impact the female germline causing oocyte defects with dramatic consequences such as aneuploïdy. We also reveal that bisphenols effects are mediated by DNA oxidation. Numerous toxicological studies have linked prophase I alterations induced by pollutants, aneuploidy and folliculogenesis defect and, here, we demonstrated the central role of oxidative DNA damages in these ovarian reproductive failures (Cuenca et al., 2020;Gely-Pernot et al., 2017;Susiarjo et al., 2013) . This opens new research avenues considering DNA oxidation in the developing germline as the cause of adult reproductive defects. Such mechanism remains to be investigated in the human germline and could also be invoked for numerous pollutants, either considered or not as endocrine disruptors, whose reprotoxic potential has poorly been studied despite a strong oxidative potential. was examined the following morning. The day following overnight mating is counted as 0.5 day post conception (dpc). All mice were killed by cervical dislocation at 11.5 dpc, 12.5 dpc, 14.5 dpc, 18.5 dpc and 8 dpp or at 3 months after birth. For pregnant mice, fetuses were removed after euthanasia from uterine horns before gonad isolation under a binocular microscope . The mice used in this study were NMRI mice (Naval Maritime Research Institute) and OGG1 deficient C57/Bl6 mice obtained from the production colony at our laboratory (Klungland et al., 2002).

Exposition protocol
The exposition protocol of this study is shown in Figure 2C and  Figure 2) or to 18.5 dpc for long-term exposure ( Figure 2C).
The control group was given drinking water added with 0.1% ethanol. For transcriptomic analyses, pregnant mice were euthanasied at 11.5 dpc and 13.5 dpc to collect gonad for germ cell sorting. As one adult mouse drinks 150 ml per kg body weight and per day, the evaluated daily intake of BPAF and BADGE by treated mice is ~500 µg/kg/day. Internal BPAF concentration was evalueted. Total BPAF was measured by gas chromatography coupled to tandem mass spectrometry (GC-MS/MS) in the plasma of 18.5 dpc pregnant mice as previoulsy described (Eladak et al., 2018) The mean ± sem values in the plasma of BPAF group were 5.96 ± 0.78 ng/ml ( 'European Food Safety Authority (EFSA) stated that the NOAELs for BADGE and BPAF is 15 and 30 mg/kg/d respectively https://www.anses.fr/fr/system/files/CHIM2009sa0331Ra-1.pdf. Nacetylcystein (NAC; Sigma A9165, 4 mM) and potassium bromate (KBrO3; Merck 1.04212.0250, 0.15 mM) were also provided in water drink from day 10.5 dpc to the end of experiment.

BrdU incorporation and detection in fetal ovaries
For BrdU incorporation, 14.5 dpc ovaries were cultured in hanging drops with BrdU (1%). After three hours, ovaries were fixed in 4% paraformaldehyde and processed for histology.

Histology and immunofluorescence on ovarian sections
Protocols Proliferation kit (RPN20, GE Healthcare). Slides were mounted with Vectashield medium. Images acquisition was accomplished with a Leica DM5500 B epifluorescence microscope (Leica Microsystems) equipped with a CoolSNAP HQ2camera (Photometrics) and ImageJ Software.
Images were analyzed with the Image J software.

Immunofluorescence on chromosome spreads
Chromosome spreads were prepared using fetal gonads (12.5 and 18.5 dpc). Fetal ovaries were lacerated on precleaned/ready-to-use superfrost slides in 1X PBS, then they were supplemented with 0.2% sucrose before adding 1% paraformaldehyde/1% Triton. Slides were incubated for 1 h at room temperature in a humid chamber and then dried under a hood and then washed two times for Vectashield with or without DAPI medium. Images were processed and specific structures were quantified with the ImageJ software (Cell Counter plugin).

Germ cell isolation
11.5 dpc and 13.5 dpc germ cell isolation using SSEA-1 antigen was performed as previously

Affymetrix sample preparation
After isolation, cells were centrifuged and resuspended in RNeasy lysis buffer. RNA was extracted using Qiagen Rneasy miniKit as recommended by the manufacturer. Total RNA concentration and RNA integrity was monitored by electrophoresis (Agilent Bioanalyzer; RNA 6000 Pico Assay).
Three pools of cells from 20-30 fetus (5 independent exposures) were used for differential expression analyses. Gene expression analysis was conducted using Mouse Clariom S array (Thermo Fisher) at 11.5 dpc and a Mouse Clariom D array at 13.5 dpc (Thermo Fisher). 500 µg of total RNA were processed according to the manufacturer. Raw data were generated and controlled with Expression console (Affymetrix) at the Genomic's platform (Institut Cochin, Paris) for 11.5 dpc analyses and at the CEA for 13.5 dpc analyses.

Microarray analysis
For gene expression analyses (11.5 dpc or 13.5 dpc), R oligo package 1.42.0 (Carvalho & Irizarry, 2010) was used for probeset annotation with expression summarized at the transcript level and robust multi-array averaging-normalized. Differential expression testing was conducted by ANOVA after linear model fitting of expression intensities with the limma R package. Microarray expression values are represented as log(2) normalized intensities. Differentially expressed genes (DEG) was filtered with a logFC ≥ 0.5 and a pvalue < 0.05. For 13.5 dpc RNA splicing analyses, the expression was summarized at the exon level. Samples were analyzed using default parameters.
An alternative splicing event was defined as a differentially expressed exon with at least a 1 Log fold change in expression, at a pvalue of < 0.05.Downstream analyses were performed with R version 3.5.0 on a CentOS Linux 7 system (64-bit).Enrichment analyses were performed using ClusterProfiler, a R package for comparing biological themes among gene clusters (G. Yu et al., 2012).

Splicing variant detection by quantitative PCR
Total RNA from isolated germ cells was extracted using the RNeasy minikit (QIAGEN, Valencia, CA, USA). cDNA was obtained by reverse transcription using the high capacity kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's instructions. Set of primers specific to the truncated/spliced region and reference primers located to the core/unspliced of the transcript were used to quantify the region of splicing using 2 delta delta CT methods.

Statistical analysis
All data are presented as means ± s.e.m. Statistical analyses were performed using Graphpad software and the R version 3.5.0. All individual biological replicates were randomly sampled from 3 independent exposures (for bisphenols exposure). The statistical significance in the difference between control and bisphenols-treated data were evaluated using the non-parametric test Mann-Whitney. Statistical significance was set as p<0.05.

Author Contributions
Conceptualization