Polycystic Ovary - Cell Metabolism

Article
Polycystic ovary syndrome is transmitted via a

transgenerational epigenetic process
Graphical abstract Authors
Nour El Houda Mimouni, Isabel Paiva,
Anne-Laure Barbotin, ...,
Vincent Prevot,
Anne-Laurence Boutillier,
Paolo Giacobini
Correspondence
laurette@unistra.fr (A.-L.B.),
paolo.giacobini@inserm.fr (P.G.)
In Brief
Polycystic ovary syndrome (PCOS) is the
most common female reproductive and
metabolic disorder, but its pathogenesis
is still unknown. Here, Mimouni, Paiva
et al. report that in mice, transmission of
PCOS-like traits to multiple generations
occurs via an altered landscape of DNA
methylation and transcriptome
expression. These authors also identify
methylome markers in women with PCOS
as possible diagnostic landmarks and
Highlights candidates for epigenetic-based
d Transmission of PCOS traits in mice occurs via an altered therapies.
DNA methylation landscape
d Metabolic- and inflammatory-related pathways are

dysregulated in models of PCOS
d Common hypomethylation signatures occur in a mouse

model of PCOS and in humans
d Identification of a novel epigenetic-based therapeutic

strategy for PCOS
Mimouni et al., 2021, Cell Metabolism 33, 513–530

March 2, 2021 ª 2021 The Author(s). Published by Elsevier Inc.
https://doi.org/10.1016/j.cmet.2021.01.004 ll
ll
OPEN ACCESS
Article
Polycystic ovary syndrome is transmitted
via a transgenerational epigenetic process
Nour El Houda Mimouni,1,6 Isabel Paiva,2,6 Anne-Laure Barbotin,1 Fatima Ezzahra Timzoura,1 Damien Plassard,3
Stephanie Le Gras,3 Gaetan Ternier,1 Pascal Pigny,4 Sophie Catteau-Jonard,1,5 Virginie Simon,1,5 Vincent Prevot,1
Anne-Laurence Boutillier,2,6,* and Paolo Giacobini1,6,7,*
1Univ. Lille, Inserm, CHU Lille, Laboratory of Development and Plasticity of the Postnatal Brain, Lille Neuroscience & Cognition, UMR-S1172,
FHU 1000 days for health, 59000 Lille, France

2Université de Strasbourg, UMR 7364 CNRS, Laboratoire de Neurosciences Cognitives et Adaptatives (LNCA), 12 Rue Goethe,
Strasbourg 67000, France

3CNRS UMR 7104, INSERM U1258, GenomEast Platform, Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC),
Université de Strasbourg, Illkirch, France

4CHU Lille, Service de Biochimie et Hormonologie, Centre de Biologie Pathologie, Lille, France
5CHU Lille, Service de Gynécologie Médicale, Hôpital Jeanne de Flandre, Lille, France
6These authors contributed equally
7Lead contact
*Correspondence: laurette@unistra.fr (A.-L.B.), paolo.giacobini@inserm.fr (P.G.)

https://doi.org/10.1016/j.cmet.2021.01.004
SUMMARY
Polycystic ovary syndrome (PCOS) is the most common reproductive and metabolic disorder affecting women
of reproductive age. PCOS has a strong heritable component, but its pathogenesis has been unclear. Here, we
performed RNA sequencing and genome-wide DNA methylation profiling of ovarian tissue from control and
third-generation PCOS-like mice. We found that DNA hypomethylation regulates key genes associated with
PCOS and that several of the differentially methylated genes are also altered in blood samples from women
with PCOS compared with healthy controls. Based on this insight, we treated the PCOS mouse model with
the methyl group donor S-adenosylmethionine and found that it corrected their transcriptomic, neuroendo-
crine, and metabolic defects. These findings show that the transmission of PCOS traits to future generations
occurs via an altered landscape of DNA methylation and propose methylome markers as a possible diagnostic
landmark for the condition, while also identifying potential candidates for epigenetic-based therapy.
INTRODUCTION of anti-Mu€ llerian hormone (AMH) exposure (Tata et al., 2018), may
be in part responsible for the development of PCOS. Indeed,
Polycystic ovary syndrome (PCOS) is the main cause of female recent preclinical evidence demonstrated that PCOS may origi-
infertility, affecting 6%–20% of women of reproductive age world- nate in the womb due to the ‘‘programing’’ effect of excessive pre-
wide (Dumesic et al., 2015; March et al., 2010). It is characterized natal AMH exposure (Tata et al., 2018). This animal model, named
by a wide range of clinical symptoms including hyperandrogenism, PAMH, recapitulates all the diagnostic criteria for PCOS in women,
oligo-anovulation and, in many cases, metabolic disorders (type 2 including hyperandrogenism, oligo-anovulation, altered fertility,
diabetes [T2D], hypertension, and cardiovascular disease) (Boyle together with increased gonadotropin-releasing hormone (GnRH)
and Teede, 2016; Dokras et al., 2017). Despite the detrimental ef- and luteinizing hormone (LH) secretion, which exacerbate the hy-
fects on women’s health, progress toward a cure for PCOS has perandrogenism in mice (Tata et al., 2018) and humans (Stener-
been hindered by the absence of a clear mechanistic etiology, Victorin et al., 2020; Walters et al., 2018).
lack of prognostic markers and by the complexity of the disease. Preclinical PCOS models provide translatable organisms to
PCOS has a strong heritable component (Crisosto et al., 2007; investigate the mechanisms underlying the etiology of the disease
Gorsic et al., 2017, 2019), as witnessed by the fact that 60%– (Stener-Victorin et al., 2020). Consistently, prenatally androgenized
70% of daughters born to women with PCOS will eventually man- (PNA) mice derived from dams exposed to dihydrotestosterone
ifest the disease (Crisosto et al., 2019; Risal et al., 2019). In line with (DHT) during late pregnancy, display PCOS-like phenotypes
that, daughters of mothers with PCOS have a 5-fold-increased risk (Moore et al., 2015; Roland et al., 2010; Sullivan and Moenter,
of being diagnosed with PCOS later in life (Risal et al., 2019). An 2004), which are transmitted across three generations (Risal
altered in utero milieu, such as excessive androgen (Abbott et al., 2019).
et al., 2002; Franks and Berga, 2012; Padmanabhan and Veiga-Lo- Environmental factors exert their effects via the induction of
pez, 2013; Risal et al., 2019; Walters et al., 2018), or elevated levels epigenetic changes such as DNA methylation and these
Cell Metabolism 33, 513–530, March 2, 2021 ª 2021 The Author(s). Published by Elsevier Inc. 513
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
ll
OPEN ACCESS Article
Figure 1. Prenatal AMH exposure induces transgenerational transmission of PCOS neuroendocrine traits to multiple generations
(A) Schematic illustration of experimental design employed to generate F1, F2, and F3 offspring.
(B) Anogenital distance (AGD) measurement over postnatal days (P) 30, 40, 50, and 60 in adult control females (n = 14), PAMH F1 (n = 13–16), PAMH F2 (n = 14),
and PAMH F3 (n = 14).
(legend continued on next page)
514 Cell Metabolism 33, 513–530, March 2, 2021
ll
Article OPEN ACCESS
modifications can lead to increased disease susceptibility later in in humans, including hyperandrogenism, oligo-anovulation,
life. However, there are very few studies focusing on the epige- increased LH levels, and fertility impairments (Qi et al., 2019;
netic changes associated with PCOS development, with only a Tata et al., 2018). We then assessed whether these neuroendo-
handful of genome-wide studies conducted so far (Makrinou crine reproductive alterations were systematically present in
et al., 2020; Shen et al., 2013; Wang et al., 2014; Xu et al., PAMH F2 and F3 offspring. From postnatal day 30 (P30) to
2010, 2016; Yu et al., 2015). In this study, we provide compelling P60, F1, F2, and F3 female PAMH lineage exhibited longer ano-
evidence that PCOS neuroendocrine reproductive and meta- genital distance than control offspring (Figure 1B), indicating a
bolic dysfunctions are transmitted in PAMH mice for at least higher androgenic impregnation in the PAMH lineage. Moreover,
three generations. We employed genome-wide methylated PAMH F1-F3 female offspring exhibited delayed vaginal opening
DNA immunoprecipitation (MeDIP) analysis to characterize and delayed puberty onset (Figures S1A and S1B). At P60, the
methylated genes in ovaries from control and PAMH mice of PAMH lineage did not show any difference in body weight as
the third generation, the first unexposed transgenerational compared with control females (Figure S1C).
offspring, together with transcriptome analysis in these tissues. Subsequently, we uncovered a significant and persistent
We identified many genes with altered transcriptome expression elevation in both circulating levels of testosterone and LH in adult
in ovarian tissues of PCOS animals and we show that several key PAMH F1-F3 females in comparison with the control group (Fig-
molecules associated to the PCOS phenotype are epigenetically ures 1C and 1D). Ovarian histological analysis of PAMH animals
regulated through DNA hypomethylation. We report that several pointed to comparable abnormalities at F1 and F3 consistent
differentially methylated signatures found in the ovaries of with their oligo-ovulatory phenotype, characterized by the pres-
PCOS-like mice are also present in blood samples from women ence of fewer post-ovulation corpora lutea as compared with
with PCOS and from daughters born to women with PCOS. control animals (Figure 1E). We confirmed such ovulatory prob-
Finally, we provide evidence that treatment of PAMH F3 fe- lems by monitoring the estrous cycles of these animals over
male offspring with a methylating pharmacological agent res- 3 weeks and showing that F1, F2, and F3 offspring in the
cues the neuroendocrine and metabolic alterations of PCOS, PAMH lineage displayed disrupted estrous cycles with pro-
thus highlighting a roadmap to new avenues for epigenetic ther- longed time spent in metestrus and diestrus as compared with
apies of the disease. the control offspring (Figures 1F and 1G). PAMH lineage also
showed diminished fertility and fecundity from F1 to F3, as indi-
RESULTS cated by fewer pups per litter produced over a 3-month period
(Figure 1H), by a significant delay in their first litter (Figure 1I)
Prenatal AMH treatment drives transgenerational and by fewer litters produced during the 90 days mating protocol
transmission of reproductive and metabolic PCOS (Figure 1J). Similar ovulatory and fertility defects were detected
alterations across multiple generations when PAMH female offspring were mated with control naive
Given the strong heritability of PCOS and the well-documented males in a matriline breeding scheme (Figures S2A–S2E).
transmission of the cardinal neuroendocrine and metabolic fea- We then checked whether PAMH F1-F3 female offspring pre-
tures observed in first degree relatives of PCOS women (Sir-Peter- sented PCOS-like metabolic alterations. At 6 months of post-
mann et al., 2002, 2012), we sought to test whether female PCOS- natal life, PAMH F1-F3 animals had higher body weight and fat
like offspring (F1) of gestating mice prenatally exposed to high AMH mass, compared with controls (Figure 2A). The percentage of
(F0) (Tata et al., 2018) are susceptible to transfer PCOS-like traits to free body fluids was comparable between all groups (Figure 2A).
F2 (intergenerational) and to F3 (transgenerational) offspring. In addition, glucose tolerance and insulin sensitivity were lower in
We injected pregnant dams (F0) intraperitoneally with PBS 6-month-old PAMH F1 offspring compared with controls (Fig-
(CNTR) or with AMH (AMHC, 0.12 mg/Kg/day; prenatal AMH- ures 2B and 2C).
treated, PAMH) from embryonic day E16.5 to E18.5, to generate We then measured fasting glucose levels, upon 12-h overnight
CNTR F1 and PAMH F1, respectively. PAMH F1 females mated fasting conditions, in control and PAMH F1 and F3 female
with PAMH F1 unrelated males generated PAMH F2 offspring offspring. Fasting glucose levels were significantly higher in
and F2 female offspring mated with another group of unrelated PAMH F1 and PAMH F3 animals than in controls (Figure 2D),
males generated F3 offspring (Figure 1A). PAMH F1 female supporting a hyperglycemic phenotype of these animals. In
offspring manifest all the major criteria of PCOS diagnosis order to investigate whether PAMH mice displayed a T2D-
(C) Plasma testosterone concentration in adult females (P60–P90) in diestrus (CNTR F1, n = 12; PAMH F1, n =12; PAMH F2, n = 14; PAMH F3, n = 15).
(D) Plasma LH levels in adult (P60–P90) diestrus females (CNTR F1, n = 14; PAMH F1, n = 11; PAMH F2, n = 17; PAMH F3, n = 17).
(E) Quantification of the number of corpora lutea (CL) in the ovaries of adult diestrus female mice (CNTR F1, n = 8; PAMH F1, n = 3; PAMH F3, n = 3).
(F) Representative estrous cyclicity of 8 mice/treatment group during 16 consecutive days. M/D: metestrus/diestrus phase, P, proestrus; E, estrus.
(G) Quantitative analysis of estrous cyclicity in adult (P60–P90) mice from control and PAMH lineages. Scatterplot representing the percentage (%) of time spent in
each estrous cycle in CNTR F1 (n = 19), PAMH F1 females (n = 19), PAMH F2 females (n = 14), and PAMH F3 females (n = 12), respectively. The horizontal line in
each scatter plot corresponds to the median value. The vertical line represents the 25th–75th percentile range.
(H) Number of pups per litter.
(I) Time to first litter (number of days to first litter after pairing).
(J) Fertility index: number of litters per females over 3 months, quantified per generation and pairing.
Data in (B)–(E) and (H)–(J) are represented as mean ± SEM. For statistical analysis, p values were calculated by Kruskal-Wallis followed by Dunn’s multiple
comparisons post hoc test (B and G) or by one-way ANOVA followed Tukey’s multiple comparison post hoc test. *p < 0.05; **p < 0.005; ***p < 0.0005;
****p < 0.0001.
Cell Metabolism 33, 513–530, March 2, 2021 515

ll
OPEN ACCESS Article
Figure 2. Prenatal AMH exposure causes a transgenerational transmission of metabolic derangements in 6-month-old female offspring
(A) Body composition of CNTR (n = 16; 6 months old), PAMH F1 (n = 16; 6 months old), PAMH F2 (n = 11–12; 6 months old), and PAMH F3 (n = 16; 6 months old),
presented as body weight (g), percent fat mass normalized to body weight (g), and percent of lean mass.
(B and C) Oral glucose tolerance test (GTT) upon 14 h of fasting (B) and insulin tolerance test (ITT) upon 4-h fasting (C) in CNTR (n = 7; 6 months old) and PAMH F1
adult female offspring (n = 7; 6 months old).
(D) Glucose levels upon 12 h of fasting in CNTR (n = 10; 6 months old), PAMH F1 (n = 10; 6 months old), and PAMH F3 (n = 7; 6 months old) female
offspring.

ll
Article OPEN ACCESS
like phenotype, we first applied a tissue-clearing technique, tion, glucose, and energy homeostasis (Dupont and Scaramuzzi,
iDISCO+ (Renier et al., 2014), and whole-organ 3D imaging to 2016; Richards and Pangas, 2010) (Figure 3H).
determine the 3D distribution of insulin-producing b cells and Notably, among the upregulated genes, we found some signif-
glucagon-producing a cells, across the whole pancreas in 6- icant enrichment in PAMH F3 offspring of genes involved in the
month-old CNTR and PAMH F1 female mice (Figure 2E). negative regulation of insulin secretion and in the control of folli-
Whole-organ immunofluorescence revealed that pancreatic is- culogenesis and ovarian steroidogenesis (Findlay, 1993; Poulsen
lets of Langerhans were hypertrophic in PCOS-like animals as et al., 2020), such as inhibin b (Inhbb) and insulin degrading
compared with controls (Figure 2E). To determine whether the enzyme (Ide) (Figure 3I), as well as follistatin (Fst; Figure 3J).
islet hyperplasia, observed in PCOS-like animals, was associ- Moreover, we identified a significant enrichment in the ovaries
ated with altered insulin secretion, we then measured fasting in- of PAMH F3 mice of genes involved in lipid metabolism (Fig-
sulin levels in 6-month-old CNTR and PAMH F1 mice and ure 3K) and inflammatory response (Figure 3L).
demonstrated that they are significantly higher in PAMH F1 fe- The top 20 significant upregulated and downregulated genes
male offspring than in control animals (Figure 2F). by fold-change in PAMH F3 ovaries versus control ovaries are
presented in Figure S5. The top upregulated genes in third-gen-
Prenatal AMH exposure results in altered ovarian eration PCOS-like ovaries are mainly related to ovarian function,
transcriptomic profiles in the third-generation offspring insulin metabolic process, inflammation, angiogenesis, cell-cy-
To dissect the molecular mechanisms and the affected gene cle progression, and axon guidance (Figure S5A). The top 20
pathways underlying ‘‘fetal reprogramming’’ of PCOS, we per- downregulated genes are mainly linked to epigenetic modifica-
formed RNA sequencing (RNA-seq) analysis in ovaries dissected tions, such as histone acetylation or methylation, apoptotic pro-
from control diestrus offspring (CNTR) and from PAMH F3 dies- cess, cell proliferation, and regulation of cell-cycle progression
trus and performed differential gene expression analysis (Figures (Figure S5B). Notably, the expression of 7 genes among the
3A and S3; Data S1). We identified 102 differentially expressed top 20 upregulated ones (Figure S5A, asterisks) and 1 gene,
genes (DEGs; 54 downregulated and 48 upregulated; adjusted among the top 20 downregulated ones (Figure S5B, asterisks),
p % 0.05) in PAMH F3 ovaries compared with control ovaries were previously reported to be altered in women with PCOS
(Figures 3B and 3C). Next, we generated heatmaps showing (Table S1), strengthening the validity of our animal model.
the expression patterns of the 102 DEGs in control and PAMH We then validated the RNA-seq results performing qRT-PCR
F3 offspring (Figure S4). Several differentially downregulated analyses of 6 upregulated genes and 6 downregulated genes
genes are involved in regulating insulin-like growth factor (IGF) related to ovarian function, metabolism, inflammation, axon
transport and uptake by IGF-binding proteins (IGFBPs) as shown guidance, and cell migration (Figures S5C and S5D), and we
in the STRING protein interaction network (Figure 3D). found that the expression of those transcripts is in accordance
To further gain insight into gene function, we performed a gene with the RNA-seq data (Figure S4; Data S1).
enrichment analysis with the DEGs using the database for annota-
tion, visualization, and integrated discovery (DAVID) functional Alterations of DNA methylation patterns in ovaries of
annotation tool (p % 0.05; Figures 3E and 3G). The downregu- PAMH mice
lated-genes-associated biological processes in PAMH F3 Since ancestral prenatal AMH exposure leads to alterations
offspring are involved in DNA repair, cell-cycle arrest, negative in ovarian gene expression in the third generation, we next inves-
regulation of phosphorylation, and negative regulation of cell pro- tigated whether it could modulate the epigenome in PAMH
liferation (Figure 3E). We used pathway analyses to identify the F3 offspring. We used methylated DNA immunoprecipitation
significant pathways associated with the DEGs according to (anti-5’methyl-cytosine, 5mC) combined with deep sequencing
Kyoto Encyclopedia of Genes and Genomes (KEGG) (Figure 3F). (MeDIP-seq) to profile the methylomic landscape in control dies-
Our analysis revealed that the most affected pathway is the trus ovaries (CNTR, prenatally PBS-treated) versus PAMH F3
FoxO signaling pathway, which is related to the regulation of cell diestrus ovaries (Figure 4A).
cycle and control of quiescence of primordial follicles, steroido- We assessed MeDIP efficiency using spike-in controls for
genesis in ovarian granulosa cells, apoptosis, and insulin signaling unmethylated and methylated DNA regions from Arabidopsis
(Dupont and Scaramuzzi, 2016; Richards and Pangas, 2010). The thaliana (Figure S6A). Principal component analysis, particularly
upregulated genes in the PAMH lineage are involved in axon guid- the PC2, indicates an evident separation of CNTR and PAMH F3
ance, fatty acid biosynthetic process, transforming growth factor groups (Figure S6B).
beta (TGF-b) production, and metabolic processes (Figure 3G). We then calculated the differentially methylated windows be-
KEGG pathway analysis showed that the most affected pathway tween the two groups. Applying an adjusted p value % 0.05 re-
among the upregulated genes is the TGF-b signaling pathway, turned 185 significant hypermethylated regions and 887 hypo-
which is involved in folliculogenesis, ovarian function, inflamma- methylated regions in the ovaries of PAMH F3 offspring as
(E) LSFM images of solvent-cleared pancreata dissected from 6-month-old CNTR and PAMH F1 female mice. Left: 3D projection of immunostaining for insulin
(red) and glucagon (white); scale bars, 150 mm. Middle: 3D analysis of rendered pancreatic islets expressing insulin; scale bars, 200 mm. Right: single plane optical
reslice of the pancreata; scale bars, 200 mm.
(F) Plasma insulin levels upon 12 h of fasting in CNTR (n = 10; 6 months old) and PAMH F1 (n = 10; 6 months old) female offspring.
Values are represented as the mean ± SEM. For statistical analysis, p values were calculated by one-way ANOVA followed by Tukey’s multiple comparison post
hoc test (A, body mass), by Kruskal-Wallis followed by Dunn’s multiple comparisons post hoc test (A, % fat mass, % lean mass; D) or by an unpaired two-tailed
Student’s t test (B and E). Statistical significance for all analyses were *p < 0.05; **p < 0.005; ***p < 0.0005; ****p < 0.0001.

ll
OPEN ACCESS Article
(legend on next page)

ll
Article OPEN ACCESS
compared with control offspring (Figures 4B and 4C; Data S2), Depicting the differentially methylated genes along the chro-
corresponding to 173 exclusively hypermethylated genes and mosomes in a Manhattan plot confirmed a preponderance of hy-
858 hypomethylated genes. pomethylation in the PAMH F3 samples and indicated that
We defined feature sets spanning sub-typed by location epigenetic changes occur quite homogenously across all chro-
(exon, intergenic, intron, promoter-transcription start site [TSS], mosomes (Figure 5A). Further, we identified significant changes
and transcription termination site [TTS]) of the hypomethylated in DNA methylation in the loci of key genes involved in demethy-
and hypermethylated regions (Figure 4D). We also observed lase activities such as ten-eleven translocation methylcytosine
that the hypermethylated regions are mostly localized in intronic dioxygenase 1 (Tet1) and factors responsible for DNA methyl-
and intergenic regions, whereas hypomethylated regions are ation maintenance, such as ubiquitin-like, containing PHD- and
mostly found into upstream-promoters and TSS, thereby most RING-finger domains, 1 (Uhrf1), in PAMH F3 ovaries compared
likely affecting gene expression. with control ovaries (Figure 5B). Both the Tet1 and Uhrf1 loci
To determine whether DNA methylation changes might asso- are significantly hypomethylated in the third-generation PCOS-
ciate with gene expression variations, we looked for overlap be- like ovaries compared with controls (Figure 5B). Consistent
tween differentially methylated genes with DEGs of PAMH F3 with these findings, we found a higher expression of Tet1 gene
ovaries (Figure 4E). We identified four common genes between in our RNA-seq analysis of PAMH F3 ovaries versus control
MeDIP-seq and RNA-seq: roundabout homolog 1 (Robo-1), sor- ovaries (p = 0.009; Data S1), even though the adjusted p value
bin and SH3 domain-containing protein 2 (Sorbs2), cyclin- did not reach statistical significance (padj = 0.36; Data S1). Over-
dependent kinase inhibitor 1A (Cdkn1a), and histidine decarbox- all, these experiments identified many genes and pathways
ylase (Hdc) (Figures 4E and 4F). They are respectively implicated associated with the PCOS reproductive and metabolic pheno-
in Slit/Robo pathway, Notch signaling, inhibition of cell prolifera- types, with altered DNA methylation profile in ovaries of the third
tion, and inflammation (Figure 4F). generation of PAMH offspring.
We next performed GO-term enrichment analysis and revealed To assess whether changes in DNA methylation detected in
distinct functional categories for the PAMH-associated gene lists ovaries of PAMH F3 animals may also occur in other tissues,
(p % 0.05; Figures 4G–4K). Within the category of biological pro- we performed MeDIP-PCR experiments in the hypothalamus,
cesses for the hypomethylated genes, chromatin remodeling and as it is the central regulator of both reproductive and metabolic
chromatin modification, cell cycle, cell differentiation, lipid meta- functions, and in the liver, which is responsible for controlling
bolism, and insulin response are listed in the top related functions the internal supply of glucose, in CNTR and PAMH F3 animals
(top 20 most significant processes; Figure 4G). Regarding the (Figure 5C). We assessed MeDIP efficiency using primers
KEGG pathways, the hypomethylated genes are enriched with directed against glyceraldehyde 3-phosphate dehydrogenase
metabolic pathways and T2D (Figures 4H and 4I). Genes associ- (GAPDH), as a negative control, and the testicular gene testis-
ated with insulin regulation (glycolysis/gluconeogenesis and specific histone H2B (TSH2B), as a positive control. We found
T2D) are represented in a STRING protein network predicting in- a strong hypomethylation of GAPDH and a hypermethylation of
teractions of those proteins associated with glucose metabolism, TSH2B, confirming the efficiency of the immunoprecipitation
insulin signaling, insulin response, and insulin receptor binding (Figures S6C and S6D). We screened by MeDIP-PCR the methyl-
(Figure 4I). Within the category of GO biological processes for ation levels of Tet1, Uhrf1, and of the four common genes be-
the hypermethylated genes, nervous system development, axons tween MeDIP-seq and RNA-seq of ovarian tissue: Hdc, Cdkn1a,
guidance, heart development, transcription, and methylation Sorbs2, and Robo-1 (Figure 5C). Similar to the methylation
pathways are listed in the top related functions (Figure 4J). changes observed in the ovaries of PCOS-like animals, we found
KEGG analysis revealed that GABAergic synapse pathway is a trend for a hypomethylation in the locus of Tet1 (p = 0.07;
significantly enriched in the hypermethylated genes (Figure 4K). Mann-Whitney test) and a significant hypomethylation in the lo-
In accordance with these changes, preclinical investigations in cus of Hdc (p = 0.02; Mann-Whitney test) in the hypothalami of
PCOS animal models have reported an ovarian hyperinnervation PAMH F3 animals as compared with controls (Figure 5C),
and a potential contribution of the peripheral sympathetic system whereas we did not detect any methylation changes in the other
in the initiation and/or perpetuation of PCOS has been proposed three genes of CNTR and PCOS animals (Figure 5C). qRT-PCR
(Stener-Victorin et al., 2005). validation experiments revealed that among those six genes,
Figure 3. RNA-seq analysis of ovarian tissue in control and F3 PCOS animals points to altered gene expression linked to ovarian and meta-
bolic functions and to inflammatory response
(A) Schematic illustration of the experimental design.
(B) MA plot of gene expression changes in the PAMH F3 ovaries (n = 4) versus control (prenatally PBS-treated; CNTR, n = 3) for all experimental conditions.
Ovaries were dissected from CNTR or PAMH F3 adult females (P60) at diestrus. The MA plot represents the estimated log2 fold change as a function of the mean
of normalized counts. Significant genes were selected when adjusted p value lower than 0.05. For significant genes, a selection of first gene names according to
the adjusted p value is displayed in red for the upregulated genes and blue for the downregulated genes.
(C) Pie chart refers to the number of genes upregulated and downregulated when comparing PAMH F3 with CNTR ovaries (padj % 0.05).
(D) STRING protein network analysis of the downregulated genes revealed strong interaction between genes involved in regulating IGF transport and uptake by
IGFBPs.
(E–H) Functional annotation charts using DAVID performed on the differentially regulated genes either decreased in PAMH F3 versus CNTR (blue, E and F) or
increased in PAMH F3 versus CNTR (red, G and H). Significance is indicated as log10 p value.
(I–L) Histograms significantly show enrichment in the PAMH F3 ovaries versus CNTR of genes involved in the negative regulation of insulin secretion (I), follistatin
(Fst; J), lipid metabolism (K), and inflammatory response (L). *padj < 0.05; **padj < 0.005; ***padj < 0.0005.

ll
OPEN ACCESS Article
Figure 4. Preponderance of hypomethylations in the ovarian tissue of F3 PCOS animals compared with controls and biological process of
hypomethylated and hypermethylated genes
(A) Schematic illustration of the experimental design. Methylated DNA immunoprecipitation and deep sequencing (MeDIP-seq) was performed in control ovaries
(CNTR n = 3, independent samples at diestrus, P60) and PAMH F3 ovaries (n = 3, independent samples at diestrus, P60).
ll
Article OPEN ACCESS
three (Hdc, Cdkn1a, and Robo1) were significantly upregulated hyperplasia detected in PAMH F1 animals (Figure 2E) was still pre-
in the hypothalamus of PAMH F3 mice compared with CNTR sent in the third generation (Figure 6H) and that it was normalized
(Figure 5D). We did not detect any differences in methylation upon SAM administration (Figures 6H and 6I).
levels in the loci of all the above-mentioned genes in the liver To further explore the effect of SAM on gene expression levels,
of either animal group (Figure S6F). we harvested the ovaries from CNTR, PAMH F3, and PAMH F3-
SAM animals at the end of the treatment period and performed
Methyl donor S-adenosylmethionine treatment of PAMH qRT-PCR experiments (Figure S7A). While Tet1 transcript levels
F3 mice normalizes their neuroendocrine reproductive were comparable in the three animal groups, Uhrf1 was signifi-
and metabolic phenotypes cantly higher in PAMH F3 ovaries (Figure S7A). Notably, SAM
Since our MeDIP-seq analyses pointed to a preponderance of hy- treatment restored Uhrf1 expression in PAMH F3 mice to normal
pomethylation in ovarian tissues of PCOS-like animals compared conditions (Figure S7A). We then selected three genes involved
with control animals, we then examined the therapeutic potential in ovarian function and insulin transport that we found both
of using the universal methyl group donor S-adenosylmethionine differentially expressed and/or methylated in PAMH F3 offspring
(SAM) in an epigenetic preclinical investigation (Figure 6A). SAM versus controls; namely Sorbs2, Grem1, and Igfbp6. Our qRT-
is an important and naturally occurring biomolecule found ubiqui- PCR experiments confirmed the RNA-seq data and showed a
tously in all living cells and functions as the primary methyl donor significant downregulation of Sorbs2 in the ovaries of PCOS an-
for all transmethylation reactions (Bottiglieri, 2002) and can imals, while its expression remained unaltered after the SAM
thereby be used to promote methylation of otherwise hypomethy- treatment (Figure S7A). In agreement with our RNA-seq data,
lated tissues (Figure 6A). We first analyzed estrous cyclicity of Grem1 and Igfbp6 transcripts were higher and lower, respec-
adult control (CNTR; 6-month-old group 1) and PAMH F3 offspring tively, in PAMH F3 mice as compared with controls (Figure S7A).
(6 months old) for 25 and 10 days, respectively, to confirm the The methylating agent did not affect expression levels of those
oligo-anovulatory phenotype of PCOS-like animals (Figure 6B). genes neither (Figure S7A).
Thereafter, we monitored for additional 15 days vaginal cytology As evidence suggests that several metabolic disturbances,
of PAMH F3 animals treated either with intraperitoneal (i.p.) injec- including obesity and T2D, are related to the generation of low-
tions of PBS (group 2) or with 50 mg/kg daily injections of SAM grade, chronic inflammation (Reilly and Saltiel, 2017), we next
(group 3). We collected tail-blood samples for LH and T measure- wondered whether a pro-inflammatory state could underlie the
ments at diestrus (day 10), before the beginning of the treatment, metabolic derangements of PAMH F3 mice. We investigated
and we harvested trunk blood and ovaries at day 25 (diestrus) at the expression changes in the ovaries of these animals of three
the moment of the sacrifice, corresponding to the end of the treat- genes involved in inflammation and immune response (Hdc,
ment period (Figure 6B). PAMH F3 animals of group 2 displayed Ptgs2, and Rela). Consistent with our RNA-seq analysis, we
prolonged time in metestrus and diestrus, as compared with con- identified a 2-fold increase in Hdc and Ptgs2 transcript levels in
trol offspring, whereas SAM treatment restored normal ovulation PAMH F3 ovaries as compared with control animals, which
of PCOS animals of group 3 (Figure 6C). In addition, SAM treat- was normalized by the methylating agent (Figure S7A). Similarly,
ment restored the aberrant LH and T concentrations of PAMH Rela (alias NF-kB) was upregulated in PCOS-like animals and
F3 offspring (Figures 6D and 6E), as well as their body weight restored to normal levels upon SAM treatment (Figure S7A).
and their body mass composition to control conditions (Figure 6F). A positive energy balance is known to initiate an inflammatory
Similar to the PAMH F1 animals, the third generation of PCOS-like response in adipocytes and in hypothalamic centers regulating
females were also hyperglycemic (Figure 6G), suggesting that hy- energy homeostasis (Jais and Bru €ning, 2017; Reilly and Saltiel,
perglycemia was transgenerationally passed from F1 to F3 PAMH 2017). Thus, we next investigated gene expression changes in
offspring. The methylating agent lowered total glucose levels of visceral fat and hypothalamus of CNTR, PAMH F3, and PAMH
PAMH F3 animals close to control conditions. However, no statis- F3-SAM animals. We found that Uhrf1 was significantly overex-
tical significance was reached upon 15 days of SAM treatment pressed in PAMH F3 visceral fat compared with CNTR (Fig-
(Figure 6G). Furthermore, we observed that the pancreatic islet ure S7B) and completely restored by the SAM treatment
(B) MA plot of MeDIP-seq reads. The MA plot represents the log2 fold change as a function of the mean of normalized counts. Log2 fold change corresponds to
shrunk fold change as calculated with the method proposed in Love et al. (2014). Red dots correspond to significantly different hypermethylated and hypo-
methylated regions in the ovaries of PAMH F3 animals as compared with controls, adjusted p % 0.05.
(C) Total number of methylated regions detected as hyper- or hypomethylated after comparing MeDIP-seq data of the two groups (PAMH3 F3 versus CNTR).
(D) Functional annotation plot showing the proportion of methylated regions falling into several genomic features. Upstream refers to upstream regulatory regions.
These are regions located 20/1 kb away from the TSS. Promoter-TSS refers to regions located 1 kb/+100 bp around the TSS and TTS refers to regions
located 100 bp/+1 kb around the TTS. Plots are shown for hypomethylated and hypermethylated regions.
(E) Venn diagram shows the overlap between genes associated with hypermethylated and hypomethylated regions in ovarian tissues of PAMH F3 mice (MeDIP-
seq data) with the 102 DEGs obtained from the RNA-seq analysis.
(F) The table shows the list of the 4 common genes found between the MeDIP-seq and the RNA-seq. It details the methylation status, the gene expression
changes, and gene ontology related either to ovarian function or PCOS and references when it applies.
(G and H) Functional enrichment analysis performed by DAVID on the genes associated with hypomethylated regions when comparing PAMH F3 versus CNTR.
GO biological processes (top 20 most significant processes) and KEGG pathway are shown. Significance is indicated as log10 p value.
(I) STRING protein network prediction interaction of proteins associated with glucose metabolism, insulin signaling, insulin response, and insulin receptor binding.
(J and K) Functional enrichment analysis performed by DAVID on the genes associated with hypermethylated regions when comparing PAMH F3 versus CNTR.
GO biological processes and KEGG pathway are shown. Significance is indicated as log10 p value.

ll
OPEN ACCESS Article
Figure 5. Chromosomal distribution of DNA methylation reads and methylation signatures in the ovary and hypothalamus of PAMH F3 mice
(A) Manhattan plot showing the association of methylated positions along the chromosomal positions. x axis represents methylated regions along the chro-
mosomes. y axis is the log10 (adjusted p value), which is the significance of differentially methylated regions when comparing PAMH F3 versus CNTR. Sign of

ll
Article OPEN ACCESS
(Figure S7B). Similar to the ovaries of PAMH animals, we found in the six genes selected from the genome-wide methylation profile
the visceral fat of PAMH F3 mice a significantly higher transcript of the PAMH lineage as being also differentially methylated in
expression of inflammatory genes (Hdc and Rela) and pro-in- blood samples of women with PCOS as compared with healthy
flammatory cytokines (Il1b and Il6) as compared with controls women (Figure 7B). Notably, all changes identified were located
(Figure S7B). Among those genes, the epigenetic treatment in the promoter regions of these genes.
restored only Rela expression levels (Figure S7B). Finally, we ROBO-1, HDC, and IGFBPL1 were also hypomethylated in
found a significant overexpression of Hdc, Ptgs2, Il1b, Il6, and blood samples of post-pubertal daughters, diagnosed with
Tnfa in the hypothalamus of PAMH F3 animals (Figure S7C). PCOS, and born to mothers with PCOS as compared with con-
The methylating agent normalized the expression of all the in- trol daughters (Figure 7C). Daughters with PCOS also showed a
flammatory pathways, except for Il1b (Figure S7C). trend to lower methylation levels of TET1 as compared with con-
To determine whether the rescue of LH and T levels observed trols (p = 0.07; Figure 7C).
in PAMH F3 females upon SAM treatment (Figures 6D and 6E)
was associated with a central neuroendocrine effect of the DISCUSSION
methylation agent, we next investigated the expression of two
genes regulating proper hypothalamic-pituitary-gonadal (HPG) Familial clustering and twin studies point to PCOS as a pathology
functions, Gnrh1 and Kiss1, in the hypothalamic preoptic area with a strong heritable component (McAllister et al., 2015). How-
(POA) dissected from the three groups of animals (Figure S7D). ever, the human PCOS loci identified by genome-wide associa-
We found a significant upregulation of both transcripts in the tion studies account for less than 10% of heritability (Azziz,
POA of PAMH F3 mice compared with CNTR (Figure S7D). How- 2016), suggesting that environmental and epigenetic mecha-
ever, the methylating agent was not able to restore Gnrh1 or nisms may play an important role in the etiology of this disease.
Kiss1 levels of PCOS animals to control conditions (Figure S7D). Clinical and preclinical studies suggest that altered levels of an-
drogens or AMH during pregnancy may be responsible for the
Common epigenetic signatures in ovarian tissue of fetal programing of PCOS (Stener-Victorin et al., 2020). Hence,
PAMH lineage and blood of women with PCOS PNA and PAMH animals are excellent preclinical models to mimic
To investigate how our findings in mice might relate to human a key maternal PCOS condition in which to investigate whether
PCOS, we searched by MeDIP-PCR for common epigenetic sig- exposed lineages have increased susceptibility to develop
natures in blood samples of women with PCOS and control PCOS-like traits in F1–F3 offspring (Stener-Victorin et al., 2020).
women (CNTR), as well as in post-pubertal daughters born to A recent study highlights transgenerational transmission of
mothers with or without PCOS (PCOS-D and CNTR-D; Figure 7A; PCOS-like phenotypes in PNA mice exposed or not to maternal
Tables S2 and S3). We assessed MeDIP efficiency using primers high-fat diet (Risal et al., 2019). That investigation also identifies
directed against GAPDH and TSH2B (Figure S6E). Taking into ac- several genes with altered expression in mouse oocytes from
count the hypomethylation of two key DNA methylation related F1–F3 offspring and in the serum of daughters of women with
genes in PAMH F3 ovaries, Tet1 and Uhrf1, that could contribute PCOS (Risal et al., 2019). However, the mechanisms underlying
to the preponderance of global DNA hypomethylation identified in the inheritance and transmission of PCOS traits to subsequent
PCOS-like mice, we first assessed the methylation levels of TET1 generations have not been elucidated. To dissect the cascade
and UHRF1 in blood samples of women with PCOS and in healthy of molecular events leading to increased disease susceptibility,
women. Interestingly, TET1 was significantly hypomethylated in we used the PAMH mouse model, which recapitulates the major
PCOS women as compared with controls, whereas methylation neuroendocrine reproductive traits of PCOS (Tata et al., 2018).
levels of UHRF1 were comparable in the two groups (Figure 7B). Compared with the PNA mouse model, in which litter size,
We next selected the four ovarian genes that were differentially estrous cyclicity, and androgen levels of the offspring are unaf-
expressed and methylated in the PAMH F3 offspring versus con- fected in the second and third generation (Risal et al., 2019),
trols and investigated the expression of the human equivalent of the transmission of PCOS-typical reproductive defects is more
those genes, ROBO-1, SORBS2, CDKN1A, and HDC as well as severe in the PAMH lineage, which pass on to subsequent gen-
IGFBPL1 (encoding for insulin-like growth factor binding protein erations all the major diagnostic features of PCOS: hyperandro-
like 1) and IRS4 (encoding for insulin receptor substrate 4), which genism, ovulatory dysfunctions, and altered fertility. This is most
may be related to defects in insulin signaling in PCOS. We iden- likely the result of the different prenatal hormonal exposure pro-
tified five (ROBO-1, CDKN1A, HDC, IGFBPL1, and IRS4) out of tocols used to generate the two mouse models. The PNA model
log10 (adjusted p value) corresponds to the direction of methylation change (hyper- or hypomethylated). Red dots show the peaks related to genes associated to
hypomethylated regions and whose functional annotations are associated with insulin stimulus, glycolysis/gluconeogenesis and T2D, as depicted in the STRING
analysis in Figure 4I. Numbers depicted below the Manhattan plot refer to the total number of significant deregulated peaks with padj % 0.05 per chromosome.
(B) Representative UCSC Genome Browser views of Tet1 and Uhrf1 locuses with DNA methylation peaks in ovarian tissues of CNTR versus PAMH F3 mice.
Differential methylation analyses revealed that the 5-mC is decreased at the highlighted regions in PAMH F3 mice compared with the CNTR. Tet1, padj = 0.018;
Uhrf1, padj = 0.01 (peak 1)/0.02 (peak 2).
(C) Genomic DNA was isolated from hypothalami dissected from CNTR (n = 6–7) and PAMH F3 offspring (n = 4–5) and MeDIP-PCR experiments performed in the
two groups of animals. Unpaired two-tailed Mann-Whitney U test, *p < 0.05, n.s. not significant.
(D) qRT-PCR analyses using primers against the genes listed were performed in hypothalamic tissues of CNTR (n = 6–8) and PAMH F3 offspring (n = 4–5).
Unpaired one-tailed Mann-Whitney U test, *p < 0.05, **p < 0.005; n.s. not significant.
Data in (C) and (D) are presented as mean ± SEM.

ll
OPEN ACCESS Article
Figure 6. Epigenetic therapy restores PCOS neuroendocrine, reproductive, and metabolic traits in PAMH F3 adult females
(A) Schematic of experimental design whereby adult (6 months old) PAMH F3 females have been treated or not with i.p. injections of SAM. SAM functions as the
primary methyl donor for transmethylation reactions and acts by adding 50 methylcytosine groups to the otherwise hypomethylated DNA.

ll
Article OPEN ACCESS
relies on the prenatal non-aromatizable DHT exposure during metabolic manifestations in these animals, occurring only few
late gestation while the latter model is derived from exposing months later, at 6 months. Our preclinical data are also in agree-
the dams to AMH during the same gestational window as the ment with several clinical studies reporting DNA methylation
PNA mice (Stener-Victorin et al., 2020). Such treatment drives changes related to inflammation, hormone-related processes,
a 3-fold rise in aromatizable T levels in the dams with subsequent and glucose and lipid metabolism in various tissues of women
changes in the HPG axis and hormone levels of both the dams with PCOS (Makrinou et al., 2020; Shen et al., 2013; Vázquez-
and the progeny (Tata et al., 2018). In this study, we also report Martı́nez et al., 2019).
the appearance of the major metabolic derangements of the hu- Genetic and epigenetic modifications are implicated in the
man PCOS condition in PAMH mice at six months of postnatal transgenerational inheritance of prenatally programed diseases
life, including typical traits of T2D and a b cell hyperplasia in (Cavalli and Heard, 2019; Gapp et al., 2014) and in recent years,
the pancreas of PAMH animals of the first and third generation. epigenetic factors gained considerable attention in the study of
Islet cell hyperplasia is associated with T2D in the leptin-deficient PCOS (Escobar-Morreale, 2018; Makrinou et al., 2020; Patel,
ob/ob mouse, which has been extensively studied as a model for 2018; Vázquez-Martı́nez et al., 2019). Here, we identify many
this disease for decades (Bock et al., 2003). Our preclinical data differentially methylated genes in PAMH F3 ovaries associated
are in line with the observation that up to 30%–40% of women with the PCOS phenotype, with a preponderance of hypomethy-
with PCOS have impaired glucose tolerance, and as many as lation detected in these animals. Interestingly, we find that some
10% of those women develop T2D by the age of 40 (Diamanti- of the epigenetic signatures and associated gene expression
Kandarakis and Dunaif, 2012; Wild et al., 2010). Consistently, changes detected in the ovaries of PAMH F3 animals are also pre-
our RNA-seq analysis reveals a significant enrichment in the sent in other metabolic tissues, such as the hypothalamus and the
ovaries of PAMH F3 mice of genes involved in the negative regu- visceral fat. These results suggest that altered methylation land-
lation of insulin secretion, including Inhbb, Ide, Fst, and TGF-b scape and transcriptional changes, transgenerationally inherited
signaling pathway, which also regulate folliculogenesis and in PCOS animals, span across multiple tissues involved in the con-
ovarian steroidogenesis (Findlay, 1993; Liu et al., 2016) and trol of ovarian function and metabolism. Such DNA methylation
which are affected in human PCOS ovarian tissues (Liu et al., differences are not just general hypomethylation events, but
2016; Pan et al., 2018). Dysfunction of TGF-b signaling in they occur in specific genes and pathways associated with repro-
PCOS may be generalized in different organs. Indeed, Dumesic ductive and metabolic traits associated with PCOS. As we find
and colleagues (Dumesic et al., 2019) report TGF-b signaling as that all hormonal, reproductive, and metabolic alterations of
the ‘‘master upstream regulatory gene’’ in subcutaneous adi- PAMH F1 offspring are maintained in the third generation, we
pose stem cells from women with PCOS in contrast to non- cannot exclude that AMH fetal-determined changes in epigenetic
PCOS women. In addition, in PCOS-like, prenatally androgen- marks, in combination with the recurring high testosterone levels
ized female rhesus monkeys, similar disruption of TGF-b and perhaps with the metabolic disturbances of the PAMH lineage
signaling accompanies altered DNA methylation patterns of may be the programing agents responsible for the acquisition and
visceral fat, implying an androgen-related developmental origin transmission of PCOS-like traits through modifications of DNA
in PCOS for abdominal fat accumulation (Xu et al., 2011). methylation landscapes. Also, we cannot exclude a mechanism
Our MeDIP-seq and RNA-seq data show that many affected involving de-methylation and re-methylation that occurred during
genes in the ovarian tissue of PCOS animals are correlated development in primordial germ cells. Indeed, the maternal
with inflammatory and metabolic pathways, typical of the human genome undergoes passive demethylation before subsequent
PCOS condition (Boyle and Teede, 2016; Dumesic et al., 2015; re-methylation at specific loci to ensure epigenetic inheritance,
Vázquez-Martı́nez et al., 2019). Interestingly, alterations in in which Uhrf1 is a key component (Zeng and Chen, 2019). Our
ovarian gene expression patterns, detected in 2-month-old data showing overexpression of Uhrf1 in PAMH F3 ovary and
PCOS-like animals, precede the appearance of phenotypic visceral fat, that is rescued by SAM treatment, support the notion
(B) Representative estrous cyclicity and experimental design. Prenatally PBS-treated, group 1 (n = 5, 6 months old); PAMH F3 animals, group 2 (n = 5, 6 months
old); and SAM-treated, group 3 (n = 5, 6 months old). The y axis refers to the different stages of the estrous cycle: metestrus/diestrus (M/D), estrus (E), and
proestrus (P). The x axis represents the time course of the experiments (days). Tail-blood samples were collected for LH and T measurements at day 10 (diestrus),
before the beginning of the treatment, and trunk blood was collected at day 25 (diestrus) at the moment of the sacrifice, corresponding to the end of the treatment
period.
(C) Scatterplot representing the percentage (%) of time spent in each estrous cycle in the three groups of animals, respectively. The horizontal line in each scatter
plot corresponds to the median value. The vertical line represents the 25th–75th percentile range.
(D) Mean LH levels were measured in diestrus CNTR mice (n = 10) and in group 2 (n = 5) and group 3 (n = 5) before the treatment (day 10) and after the treatment
(day 25).
(E) Mean T levels measured in diestrus in CNTR mice (n = 10) and in group 2 (n = 5) and group 3 (n = 5) before the treatment (day 10) and after the treatment (day 25).
(F) Body composition in the three experimental groups (n = 5 for each group, 6 months old) presented as body weight (grams), percent fat mass normalized to
body weight (grams), and percent lean mass normalized to body weight (grams).
(G) Mean total glucose levels (mg/dL) in diestrus CNTR mice (n = 11), in PAMH F3 mice (group 2; n = 10), and in PAMH F3 mice after SAM treatment (group 3; n = 5).
(H) Representative photomicrographs showing insulin-expressing b cells (red) and glucagon-expressing a cells (white) in the pancreata of CNTR, PAMH F3, and
PAMH F3 + SAM mice (females, 6 months old). Scale bar, 50 mm.
(I) Quantitative analysis of the mean area of the islets in CNTR (n = 4), PAMH F3 (n = 5), and PAMH F3 after SAM treatment (n = 5).
Values in (D)–(G) and (I) are represented as the mean ± SEM. For statistical analysis, p values were calculated by Kruskal-Wallis test followed by Dunn’s multiple
comparison post hoc test (C), by one-way ANOVA followed by Tukey’s multiple comparison post hoc test (D, E, G, and I), or by Kruskal-Wallis followed by Dunn’s
multiple comparisons post hoc test (F). Statistical significance for all analyses was *p < 0.05; **p < 0.005; ***p < 0.0005; ****p < 0.0001; n.s. not significant.

ll
OPEN ACCESS Article
Figure 7. Common epigenetic signatures in human blood samples from women with PCOS
(A) Schematic illustration of the experimental design. Genomic DNA was isolated from blood samples of a case-control study comprising two cohorts of women.
Group 1: women with and without PCOS (CNTR). Group 2: post-pubertal control daughters born to mothers without PCOS (CNTR-D) and daughters with PCOS
born to mothers with PCOS (PCOS-D). Methylated DNA immunoprecipitation using antibody against anti-5mC, followed by PCR (MeDIP-PCR) using specific
primers against the genes listed in (B) and (C) was performed in the two groups.
(B) MeDIP-PCR analyses in CNTR women (n = 15) and women with PCOS (n = 32).
(C) MeDIP-PCR analyses in daughters from the control group (CNTR-D, n = 3) and daughters with PCOS of women with PCOS (PCOS-D, n = 5).
Data are presented as mean ± SEM. Unpaired two-tailed Mann-Whitney U test. *p < 0.05; **p < 0.005; n.s. not significant.
that maintenance of the methylation machinery is a crucial event in responsible for genomic instability in the disease condition.
the inheritance of PCOS. Consistently, a genome-wide DNA methylation study on umbili-
We speculate that a global loss of DNA methylation, particu- cal cord blood reports a prevalence of hypomethylation in
larly in promoter-TSS and upstream-promoters, could be women with PCOS compared with unaffected women

ll
Article OPEN ACCESS
(Lambertini et al., 2017). As genomic instability is highly corre- lin, fasting glycemia, and triglycerides) are within the normal range
lated with DNA damage, excessive DNA demethylation could in the PCOS groups, we can exclude a priori the contribution of
be thus associated with impaired DNA damage repair. This is metabolic alterations on the differential methylation landscape
in line with many reports describing a strong association be- of women with PCOS versus control women. We should also
tween PCOS and malignancies, such as ovarian and endometrial note that control women recruited in this study are significantly
cancer (Escobar-Morreale, 2018), and suggest that a higher pre- older than women with PCOS (Table S2), raising the possibility
disposition to cancer detected in women with PCOS could be of the existence of an age-dependent epigenetic effect. However,
due to altered DNA methylation landscapes. the observation that hypomethylation occurs in daughters with
Mechanistically, we identify a significant hypomethylation in PCOS age-matched with control daughters (Table S3; age
the locus of two key genes involved in DNA demethylation CNTR-D: 24 ± 4.4, PCOS-D: 26 ± 2.0), in the same genes as those
(Tet1) and DNA methylation maintenance (Uhrf1) in PAMH F3 detected in blood samples of women with PCOS suggests that
ovaries compared with control ovaries of which our RNA-seq DNA methylation changes are mostly associated with the PCOS
data show increased Tet1 expression with a significant p value. status rather than being influenced by age.
Consistently, we find TET1 significantly hypomethylated in blood As DNA methylation epigenetic changes can precede pheno-
samples of women with PCOS as compared with control women typic manifestation and display more stability than gene expres-
and a trend to a hypomethylation of this gene is also present in sion alterations (Kelly et al., 2010), the differentially methylated
daughters with PCOS. As TET1 is one of the family members genes offer the opportunity to develop valuable diagnostic indica-
of 5mC dioxygenases, which initiate demethylation, it is possible tors for PCOS risk or prognostic indicators for the disease
that the decreased levels of TET1 methylation observed in progression. Even more importantly, the reversible nature of
women with PCOS could contribute to the preponderance of epigenetic modifications makes them more ‘‘druggable’’ than at-
global DNA hypomethylation characterizing the disease. tempts to target or correct defects in gene expression itself. Both
In this study, we report that only 4 genes showed concordance hypermethylation and hypomethylation are involved in several dis-
between RNA-seq and methylation analyses in ovarian tissues of ease conditions (Kelly et al., 2010). Several inhibitors of DNA
PCOS animals (Robo-1, Sorbs2, Cdkn1a, and Hdc). However, a methylation are currently approved for many pathologies by the
far more complex relationship exists between DNA methylation US Food and Drug Administration (FDA) and have been in clinical
and transcriptional processes, which can be regulated by other use for many years (Kelly et al., 2010). However, at present, there
epigenetic mechanisms. Indeed, the methylation of the first are no FDA-approved therapeutic modalities that target hypome-
exon is negatively correlated to gene expression in a more pro- thylation. Here, we examined the therapeutic potential of SAM, a
nounced way than methylations of the promoter regions (Brenet known natural agent causing methylation of several genes (Chik
et al., 2011). Moreover, previous studies suggest a permissive et al., 2014). To our knowledge, this is the first direct evidence
state of gene expression linked with low methylation, but not a for the potential therapeutic effect of SAM in a preclinical model
linear inhibitory link with high methylation (Anastasiadi et al., of PCOS. Our investigation showed that SAM treatment can
2018). Based on the canonical view of 5mC being a repressor rescue the major PCOS reproductive neuroendocrine and meta-
of transcription (Deaton and Bird, 2011), our results indicate a bolic alterations of PAMH F3 mice, thus highlighting the therapeu-
mismatch between the methylation state and the level of gene tic potential of methylating agents as promising epigenetic thera-
expression for Robo-1 and Cdkn1a. Considering that the most pies aimed at treating women with PCOS. We provide evidence
hypomethylated genes emerging from our analysis are related that the methylating agent restores the aberrant expression of
to chromatin remodeling and chromatin modification, it is likely most inflammatory genes investigated in the ovaries as well as
that, besides DNA methylation, other epigenetic events, such in metabolic tissues of PAMH F3 adult mice. Numerous studies
as histone acetylation/methylation are modulated altering gene show a causal link between low-grade inflammation and meta-
expression, which could in part explain the weak correlation bolic diseases, including T2D (Reilly and Saltiel, 2017). Moreover,
that we observe between MeDIP-seq and RNA-seq. In line the degree of inflammation correlates well with the severity of in-
with these findings, histone acetylation alterations are present sulin resistance, T2D, and hyperandrogenism related to PCOS
in various tissues of women with the disease (Qu et al., 2012; (González et al., 2006; Zhao et al., 2015). However, the precise
Vázquez-Martı́nez et al., 2019). triggers of PCOS-associated inflammation are poorly understood.
Remarkably, we report that several of the differentially methyl- Based on our findings we can speculate that the trigger for tissue
ated genes identified in ovarian tissues of PCOS mice of the third inflammations could emanate from altered DNA methylation land-
generation are also altered in blood samples from women with scapes, which can be corrected by the SAM.
PCOS and from daughters of women with PCOS compared with Taken together, this study points to AMH excess during gesta-
healthy women. Six genes associated with DNA demethylation tion as a detrimental factor leading to the transgenerational
(TET1), axon guidance (ROBO-1), inhibition of cell proliferation transmission of PCOS cardinal neuroendocrine, reproductive,
(CDKN1A), inflammation (HDC), and insulin signaling and metabolic alterations and shed lights into the epigenetic
(IGFBPL1, IRS4) are hypomethylated in women with PCOS as modifications underlying the susceptibility of the disease while
compared with controls, and three genes (ROBO-1, HDC, and pointing to novel diagnostic tools and epigenetic-based thera-
IGFBPL1) are also hypomethylated in daughters diagnosed with peutic avenues to treat the disease.
PCOS. As the BMI of women with PCOS included in this investi-
gation is not significantly different compared with controls, both Limitations of study
in unrelated women and in CNTR-D and PCOS-D (Tables S2 In this study, we demonstrate the involvement of epigenetic
and S3), and because several metabolic parameters (fasting insu- changes at the ovarian level as drivers of the susceptibility of

ll
OPEN ACCESS Article
PCOS across generations. However, we have not performed ACKNOWLEDGMENTS

genome-wide DNA methylation and RNA-seq profiling of the
We thank M.-H. Gevaert (histology core facility of Lille), J. Devassine and Yann
hypothalamic tissue of control and PCOS animals. Hence,
Lepage (animal core facility, BICeL core facility of Lille, University of Lille,
we cannot rule out that gene expression modifications CNRS, INSERM, CHU Lille, Institut Pasteur de Lille, US 41-UMS 2014-PLBS,
and DNA methylation changes possibly affecting the hypothal- F-59000 Lille, France), and the IGBMC core facility of Strasbourg for expert
amus of PCOS-like animals may also play a role in this phe- technical assistance. This work was supported by the European Research
nomenon. Moreover, we have not fully assessed if SAM treat- Council (ERC) under the European Union’s Horizon 2020 research and innova-
ment can also normalize the reproductive deficits of PAMH tion program (ERC-2016-CoG to P.G., grant agreement no. 725149/REPRO-
DAMH); Institut National de la Santé et de la Recherche Médicale (INSERM),
animals.
France (grant number U1172 to P.G. and V.P.); Centre Hospitalier Régional
Future studies, using genome-wide DNA methylation ap-
Universitaire, CHU de Lille, France (Bonus H to P.G. and Ph.D. fellowship to
proaches, in combination with preclinical phenotypic investiga- N.E.H.M.); Fondation pour la Recherche Médicale, France (FRM, fellowship
tions, are also necessary to determine whether re-methylation to I.P.); Centre National de la Recherche Scientifique (CNRS), France; and Uni-
after the epigenetic treatment could potentially drive adverse versité de Strasbourg, France (to A.-L.B.).
secondary phenotypic effects at long term. Finally, our clinical
investigation identifies methylome markers in the blood of AUTHOR CONTRIBUTIONS
women with PCOS as possible diagnostic landmarks and candi-
dates for epigenetic-based therapies. However, these markers N.E.H.M. designed and performed the experiments involving animal and hu-
man investigation, analyzed the data, and was involved in all aspects of manu-
need to be confirmed in larger cohorts of patients, segregating
script preparation; I.P. designed and performed the epigenetic and transcrip-
endocrine and metabolic characteristics related to risk and tomic experiments, analyzed the data, and was involved in all aspects of
severity of PCOS, to suggest their use as an effective personal- manuscript preparation; A.-L.B. performed animal investigation and analyzed
ized screening and treatment intervention. the data; F.E.T. assisted in the preparation and realization of experiments; D.P.
and S.L.G. performed RNA-seq and MeDIP-seq data analyses and reviewed
and edited the manuscript; G.T. performed tissue-clearing experiments,
STAR+METHODS light-sheet imaging, and analysis of pancreatic tissues in 2D and 3D; S.J.
and P.P. recruited the patients and provided clinical information for the human
Detailed methods are provided in the online version of this paper study and biological samples, respectively; V.S. helped with the analysis of the
and include the following: clinical data; V.P. was involved in the interpretation of the results and prepara-
tion of the manuscript; A.-L.B. was involved in the design and analyses of the
d KEY RESOURCES TABLE epigenetic investigation and reviewed and edited the manuscript; and P.G. de-
d RESOURCE AVAILABILITY signed the study, analyzed data, prepared the figures, and wrote the manu-
script. All authors read and approved the final manuscript.
B Lead contact
B Materials availability
B Data and code availability DECLARATION OF INTERESTS
d EXPERIMENTAL MODEL AND SUBJECT DETAILS P.G., N.E.H.M., I.P., A.-L.B., and V.P. disclose that they are inventors of a sub-
B Human studies mitted patent application by the INSERM (Institut National de la Santé et de la
B Mouse models Recherche Médicale) covering methods and kits for diagnostic and treatment
d METHOD DETAILS of PCOS. All other authors do not have competing interests.
B Prenatal anti-Mu€llerian hormone (PAMH) treatment
B Mouse breeding scheme and feeding paradigm to Received: August 6, 2020
Revised: November 20, 2020
generate F1–F3 offspring
Accepted: January 5, 2021
B Assessment of phenotype, estrous cycle, and fertility
Published: February 3, 2021
B Ovarian histology
B LH and T ELISA assays
SUPPORTING CITATIONS
B Body weight and composition
B Measurement of fasting blood glucose and insulin level The following reference appears in the Supplemental Information: Schulze
B Glucose and insulin tolerance tests Schulze et al. (2012).
B Optical clearing of mouse pancreas
B Light-sheet imaging
REFERENCES
B Immunohistochemistry on pancreatic sections
B Image acquisition and analysis Abbott, D.H., Dumesic, D.A., and Franks, S. (2002). Developmental origin of
B RNA extraction and RT-qPCR polycystic ovary syndrome - a hypothesis. J. Endocrinol. 174, 1–5.
B RNA libraries and sequencing Anastasiadi, D., Esteve-Codina, A., and Piferrer, F. (2018). Consistent inverse
B MeDIP correlation between DNA methylation of the first intron and gene expression
B MeDIP-seq - libraries construction and sequencing across tissues and species. Epigenet. Chromatin 11, 37.
B Methyl donor S-adenosylmethionine (SAM) treatment Anders, S., and Huber, W. (2010). Differential expression analysis for sequence
d QUANTIFICATION AND STATISTICAL ANALYSIS count data. Genome Biol 11, R106.
Anders, S., Pyl, P.T., and Huber, W. (2015). HTSeq–a Python framework to
SUPPLEMENTAL INFORMATION work with high-throughput sequencing data. Bioinformatics 31, 166–169.
Supplemental information can be found online at https://doi.org/10.1016/j. Azziz, R. (2016). PCOS in 2015: new insights into the genetics of polycystic
cmet.2021.01.004. ovary syndrome. Nat. Rev. Endocrinol. 12, 183.

ll
Article OPEN ACCESS
Belle, M., Godefroy, D., Dominici, C., Heitz-Marchaland, C., Zelina, P., Hellal, F., Franks, S., and Berga, S.L. (2012). Does PCOS have developmental origins?
Bradke, F., and Chédotal, A. (2014). A simple method for 3D analysis of immu- Fertil. Steril. 97, 2–6.
nolabeled axonal tracts in a transparent nervous system. Cell Rep 9, 1191–1201. Gapp, K., von Ziegler, L., Tweedie-Cullen, R.Y., and Mansuy, I.M. (2014). Early
Benjamini, Y.H., and Hochberg, Y. (1995). Controlling the false discovery rate: a life epigenetic programming and transmission of stress-induced traits in mam-
practical and powerful approach to multiple testing. J. R. Stat. Soc. B 57, 289–300. mals: how and when can environmental factors influence traits and their trans-
Bock, T., Pakkenberg, B., and Buschard, K. (2003). Increased islet volume but generational inheritance? BioEssays 36, 491–502.
unchanged islet number in ob/ob mice. Diabetes 52, 1716–1722. González, F., Rote, N.S., Minium, J., and Kirwan, J.P. (2006). Increased activa-
Bottiglieri, T. (2002). S-adenosyl-L-methionine (SAMe): from the bench to the tion of nuclear factor kappaB triggers inflammation and insulin resistance in
bedside–molecular basis of a pleiotrophic molecule. Am. J. Clin. Nutr. 76, polycystic ovary syndrome. J. Clin. Endocrinol. Metab. 91, 1508–1512.
1151S–1157S. Gorsic, L.K., Kosova, G., Werstein, B., Sisk, R., Legro, R.S., Hayes, M.G.,
Teixeira, J.M., Dunaif, A., and Urbanek, M. (2017). Pathogenic anti-Mullerian
Boyle, J.A., and Teede, H.J. (2016). PCOS: refining diagnostic features in
hormone variants in polycystic ovary syndrome. J. Clin. Endocrinol. Metab.
PCOS to optimize health outcomes. Nat. Rev. Endocrinol. 12, 630–631.
102, 2862–2872.
Brenet, F., Moh, M., Funk, P., Feierstein, E., Viale, A.J., Socci, N.D., and
Gorsic, L.K., Dapas, M., Legro, R.S., Hayes, M.G., and Urbanek, M. (2019).
Scandura, J.M. (2011). DNA methylation of the first exon is tightly linked to
Functional genetic variation in the anti-Mullerian hormone pathway in women
transcriptional silencing. PLoS One 6, e14524.
with polycystic ovary syndrome. J. Clin. Endocrinol. Metab. 104, 2855–2874.
Caldwell, A.S.L., Edwards, M.C., Desai, R., Jimenez, M., Gilchrist, R.B.,
Heinz, S., Benner, C., Spann, N., Bertolino, E., Lin, Y.C., Laslo, P., Cheng, J.X.,
Handelsman, D.J., and Walters, K.A. (2017). Neuroendocrine androgen action
Murre, C., Singh, H., and Glass, C.K. (2010). Simple combinations of lineage-
is a key extraovarian mediator in the development of polycystic ovary syn-
determining transcription factors prime cis-regulatory elements required for
drome. Proc. Natl. Acad. Sci. USA 114, E3334–E3343.
macrophage and B cell identities. Mol. Cell 38, 576–589.
Cavalli, G., and Heard, E. (2019). Advances in epigenetics link genetics to the
Ignatiadis, N., Klaus, B., Zaugg, J.B., and Huber, W. (2016). Data-driven hy-
environment and disease. Nature 571, 489–499.
pothesis weighting increases detection power in genome-scale multiple
Chik, F., Machnes, Z., and Szyf, M. (2014). Synergistic anti-breast cancer effect of testing. Nat. Methods 13, 577–580.
a combined treatment with the methyl donor S-adenosyl methionine and the DNA €ning, J.C. (2017). Hypothalamic inflammation in obesity and
Jais, A., and Bru
methylation inhibitor 5-aza-20 -deoxycytidine. Carcinogenesis 35, 138–144.
metabolic disease. J. Clin. Invest. 127, 24–32.
Crisosto, N., Codner, E., Maliqueo, M., Echiburú, B., Sánchez, F., Cassorla, F., Kelly, T.K., De Carvalho, D.D., and Jones, P.A. (2010). Epigenetic modifica-
and Sir-Petermann, T. (2007). Anti-Mullerian hormone levels in peripubertal tions as therapeutic targets. Nat. Biotechnol. 28, 1069–1078.
daughters of women with polycystic ovary syndrome. J. Clin. Endocrinol.
Lambertini, L., Saul, S.R., Copperman, A.B., Hammerstad, S.S., Yi, Z., Zhang,
Metab. 92, 2739–2743.
W., Tomer, Y., and Kase, N. (2017). Intrauterine reprogramming of the polycy-
Crisosto, N., Ladrón de Guevara, A., Echiburú, B., Maliqueo, M., Cavada, G., stic ovary syndrome: evidence from a pilot study of cord blood global methyl-
Codner, E., Paez, F., and Sir-Petermann, T. (2019). Higher luteinizing hormone ation analysis. Front. Endocrinol. 8, 352.
levels associated with antiMullerian hormone in postmenarchal daughters of
Langmead, B., Trapnell, C., Pop, M., and Salzberg, S.L. (2009). Ultrafast and
women with polycystic ovary syndrome. Fertil. Steril. 111, 381–388.
memory-efficient alignment of short DNA sequences to the human genome.
Deaton, A.M., and Bird, A. (2011). CpG islands and the regulation of transcrip- Genome Biol 10, R25.
tion. Genes Dev 25, 1010–1022.
Leek, J.T. (2014). svaseq: removing batch effects and other unwanted noise
Dewailly, D., Gronier, H., Poncelet, E., Robin, G., Leroy, M., Pigny, P., from sequencing data. Nucleic Acids Res 42.
Duhamel, A., and Catteau-Jonard, S. (2011). Diagnosis of polycystic ovary
Li, T.W., Yang, H., Peng, H., Xia, M., Mato, J.M., and Lu, S.C. (2012). Effects of
syndrome (PCOS): revisiting the threshold values of follicle count on ultra-
S-adenosylmethionine and methylthioadenosine on inflammation-induced co-
sound and of the serum AMH level for the definition of polycystic ovaries.
lon cancer in mice. Carcinogenesis 33, 427–435.
Hum. Reprod. 26, 3123–3129.
Liu, Q., Li, Y., Feng, Y., Liu, C., Ma, J., Li, Y., Xiang, H., Ji, Y., Cao, Y., Tong, X.,
Diamanti-Kandarakis, E., and Dunaif, A. (2012). Insulin resistance and the poly- and Xue, Z. (2016). Single-cell analysis of differences in transcriptomic profiles
cystic ovary syndrome revisited: an update on mechanisms and implications. of oocytes and cumulus cells at GV, MI, MII stages from PCOS patients. Sci.
Endocr. Rev. 33, 981–1030. Rep. 6, 39638.
Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, Livak, K.J., and Schmittgen, T.D. (2001). Analysis of relative gene expression
P., Chaisson, M., and Gingeras, T.R. (2013). STAR: ultrafast universal RNA-seq data using real-time quantitative PCR and the 2(-delta delta C(T)) method.
aligner. Bioinformatics 29, 15–21. Methods 25, 402–408.
Dokras, A., Saini, S., Gibson-Helm, M., Schulkin, J., Cooney, L., and Teede, H. Love, M.I., Huber, W., and Anders, S. (2014). Moderated estimation of fold
(2017). Gaps in knowledge among physicians regarding diagnostic criteria and change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550.
management of polycystic ovary syndrome. Fertil. Steril. 107, 1380–1386.e1.
Makrinou, E., Drong, A.W., Christopoulos, G., Lerner, A., Chapa-Chorda, I.,
Dumesic, D.A., Oberfield, S.E., Stener-Victorin, E., Marshall, J.C., Laven, J.S., Karaderi, T., Lavery, S., Hardy, K., Lindgren, C.M., and Franks, S. (2020).
and Legro, R.S. (2015). Scientific statement on the diagnostic criteria, epide- Genome-wide methylation profiling in granulosa lutein cells of women with
miology, pathophysiology, and molecular genetics of polycystic ovary syn- polycystic ovary syndrome (PCOS). Mol. Cell. Endocrinol. 500, 110611.
drome. Endocr. Rev. 36, 487–525.
March, W.A., Moore, V.M., Willson, K.J., Phillips, D.I., Norman, R.J., and Davies,
Dumesic, D.A., Phan, J.D., Leung, K.L., Grogan, T.R., Ding, X., Li, X., Hoyos, M.J. (2010). The prevalence of polycystic ovary syndrome in a community sam-
L.R., Abbott, D.H., and Chazenbalk, G.D. (2019). Adipose insulin resistance ple assessed under contrasting diagnostic criteria. Hum. Reprod 25, 544–551.
in normal-weight women with polycystic ovary syndrome. J. Clin. Martin, M. (2011). Cutadapt removes adapter sequences from high-
Endocrinol. Metab. 104, 2171–2183. throughput sequencing reads. EMBnet J 17, 10–12.
Dupont, J., and Scaramuzzi, R.J. (2016). Insulin signalling and glucose trans- McAllister, J.M., Legro, R.S., Modi, B.P., and Strauss, J.F., 3rd. (2015).
port in the ovary and ovarian function during the ovarian cycle. Biochem. J. Functional genomics of PCOS: from GWAS to molecular mechanisms.
473, 1483–1501. Trends Endocrinol. Metab. 26, 118–124.
Escobar-Morreale, H.F. (2018). Polycystic ovary syndrome: definition, aeti- Moore, A.M., Prescott, M., Marshall, C.J., Yip, S.H., and Campbell, R.E. (2015).
ology, diagnosis and treatment. Nat. Rev. Endocrinol. 14, 270–284. Enhancement of a robust arcuate GABAergic input to gonadotropin-releasing
Findlay, J.K. (1993). An update on the roles of inhibin, activin, and follistatin as hormone neurons in a model of polycystic ovarian syndrome. Proc. Natl. Acad.
local regulators of folliculogenesis. Biol. Reprod. 48, 15–23. Sci. USA 112, 596–601.

ll
OPEN ACCESS Article
Padmanabhan, V., and Veiga-Lopez, A. (2013). Sheep models of polycystic Stener-Victorin, E., Padmanabhan, V., Walters, K.A., Campbell, R.E., Benrick,
ovary syndrome phenotype. Mol. Cell. Endocrinol. 373, 8–20. A., Giacobini, P., Dumesic, D.A., and Abbott, D.H. (2020). Animal models to un-
Pan, J.-X., Tan, Y.-J., Wang, F.-F., Hou, N.-N., Xiang, Y.-Q., Zhang, J.-Y., Liu, derstand the etiology and pathophysiology of polycystic ovary syndrome.
Y., Qu, F., Meng, Q., Xu, J., et al. (2018). Aberrant expression and DNA methyl- Endocr. Rev. 41, 538–576.
ation of lipid metabolism genes in PCOS: a new insight into its pathogenesis. Steyn, F.J., Wan, Y., Clarkson, J., Veldhuis, J.D., Herbison, A.E., and Chen, C.
Clin. Epigenet. 10, 6. (2013). Development of a methodology for and assessment of pulsatile lutei-
Patel, S. (2018). Polycystic ovary syndrome (PCOS), an inflammatory, sys- nizing hormone secretion in juvenile and adult male mice. Endocrinology
temic, lifestyle endocrinopathy. J. Steroid Biochem. Mol. Biol. 182, 27–36. 154, 4939–4945.
Pigny, P., Desailloud, R., Cortet-Rudelli, C., Duhamel, A., Deroubaix-Allard, D., Sullivan, S.D., and Moenter, S.M. (2004). Prenatal androgens alter GABAergic
Racadot, A., and Dewailly, D. (1997). Serum alpha-inhibin levels in polycystic drive to gonadotropin-releasing hormone neurons: implications for a common
ovary syndrome: relationship to the serum androstenedione level. J. Clin. fertility disorder. Proc. Natl. Acad. Sci. USA 101, 7129–7134.
Endocrinol. Metab. 82, 1939–1943.
Tata, B., Mimouni, N.E.H., Barbotin, A.L., Malone, S.A., Loyens, A., Pigny, P.,
Poulsen, L.C., Englund, A.L.M., Andersen, A.S., Bøtkjær, J.A., Mamsen, L.S.,
Dewailly, D., Catteau-Jonard, S., Sundström-Poromaa, I., Piltonen, T.T.,
Damdimopoulou, P., Østrup, O., Grøndahl, M.L., and Yding Andersen, C.
et al. (2018). Elevated prenatal anti-Mullerian hormone reprograms the fetus
(2020). Follicular hormone dynamics during the midcycle surge of gonadotro-
and induces polycystic ovary syndrome in adulthood. Nat. Med. 24,
pins in women undergoing fertility treatment. Mol. Hum. Reprod. 26, 256–268.
834–846.
Qi, X., Yun, C., Sun, L., Xia, J., Wu, Q., Wang, Y., Wang, L., Zhang, Y., Liang, X.,
Wang, L., et al. (2019). Gut microbiota-bile acid-interleukin-22 axis orches- Vázquez-Martı́nez, E.R., Gómez-Viais, Y.I., Garcı́a-Gómez, E., Reyes-
trates polycystic ovary syndrome. Nat. Med. 25, 1225–1233. Mayoral, C., Reyes-Muñoz, E., Camacho-Arroyo, I., and Cerbón, M. (2019).
DNA methylation in the pathogenesis of polycystic ovary syndrome.
Qu, F., Wang, F.F., Yin, R., Ding, G.L., El-Prince, M., Gao, Q., Shi, B.W., Pan,
Reproduction 158, R27–R40.
H.H., Huang, Y.T., Jin, M., et al. (2012). A molecular mechanism underlying
ovarian dysfunction of polycystic ovary syndrome: hyperandrogenism induces Walters, K.A., Gilchrist, R.B., Ledger, W.L., Teede, H.J., Handelsman, D.J.,
epigenetic alterations in the granulosa cells. J. Mol. Med. (Berl.) 90, 911–923. and Campbell, R.E. (2018). New perspectives on the pathogenesis of PCOS:
Quinlan, A.R., and Hall, I.M. (2010). BEDTools: a flexible suite of utilities for neuroendocrine origins. Trends Endocrinol. Metab.: TEM 29, 841–852.
comparing genomic features. Bioinformatics 26, 841–842. Wang, L., Wang, S., and Li, W. (2012). RSeQC: quality control of RNA-seq ex-
Reilly, S.M., and Saltiel, A.R. (2017). Adapting to obesity with adipose tissue periments. Bioinformatics 28, 2184–2185.
inflammation. Nat. Rev. Endocrinol. 13, 633–643.
Wang, X.X., Wei, J.Z., Jiao, J., Jiang, S.Y., Yu, D.H., and Li, D. (2014). Genome-
Renier, N., Wu, Z., Simon, D.J., Yang, J., Ariel, P., and Tessier-Lavigne, M. wide DNA methylation and gene expression patterns provide insight into poly-
(2014). iDISCO: a simple, rapid method to immunolabel large tissue samples cystic ovary syndrome development. Oncotarget 5, 6603–6610.
for volume imaging. Cell 159, 896–910.
Wild, R.A., Carmina, E., Diamanti-Kandarakis, E., Dokras, A., Escobar-
Richards, J.S., and Pangas, S.A. (2010). The ovary: basic biology and clinical
Morreale, H.F., Futterweit, W., Lobo, R., Norman, R.J., Talbott, E., and
implications. J. Clin. Invest. 120, 963–972.
Dumesic, D.A. (2010). Assessment of cardiovascular risk and prevention of
Risal, S., Pei, Y., Lu, H., Manti, M., Fornes, R., Pui, H.P., Zhao, Z., Massart, J., cardiovascular disease in women with the polycystic ovary syndrome: a
Ohlsson, C., Lindgren, E., et al. (2019). Prenatal androgen exposure and transge- consensus statement by the androgen excess and polycystic ovary syndrome
nerational susceptibility to polycystic ovary syndrome. Nat. Med. 25, 1894–1904. (AE-PCOS) society. J. Clin. Endocrinol. Metab. 95, 2038–2049.
Roland, A.V., Nunemaker, C.S., Keller, S.R., and Moenter, S.M. (2010).
Xu, N., Azziz, R., and Goodarzi, M.O. (2010). Epigenetics in polycystic ovary
Prenatal androgen exposure programs metabolic dysfunction in female
syndrome: a pilot study of global DNA methylation. Fertil. Steril. 94,
mice. J. Endocrinol. 207, 213–223.
781–783.e1.
Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshop group
(2004). Revised 2003 consensus on diagnostic criteria and long-term health Xu, N., Kwon, S., Abbott, D.H., Geller, D.H., Dumesic, D.A., Azziz, R., Guo, X.,
risks related to polycystic ovary syndrome (PCOS). Hum. Reprod. 19, 41–47. and Goodarzi, M.O. (2011). Epigenetic mechanism underlying the develop-
ment of polycystic ovary syndrome (PCOS)-like phenotypes in prenatally an-
Schulze, S.K., Kanwar, R., Gölzenleuchter, M., Therneau, T.M., and Beutler,
drogenized rhesus monkeys. PLoS One 6, e27286.
A.S. (2012). SERE: single-parameter quality control and sample comparison
for RNA-Seq. BMC Genomics 13, 524. Xu, J., Bao, X., Peng, Z., Wang, L., Du, L., Niu, W., and Sun, Y. (2016).
Shen, H.R., Qiu, L.H., Zhang, Z.Q., Qin, Y.Y., Cao, C., and Di, W. (2013). Comprehensive analysis of genome-wide DNA methylation across human
Genome-wide methylated DNA immunoprecipitation analysis of patients polycystic ovary syndrome ovary granulosa cell. Oncotarget 7, 27899–27909.
with polycystic ovary syndrome. PLoS One 8, e64801. Yu, Y.Y., Sun, C.X., Liu, Y.K., Li, Y., Wang, L., and Zhang, W. (2015). Genome-
Sir-Petermann, T., Ladrón de Guevara, A., Codner, E., Preisler, J., Crisosto, N., wide screen of ovary-specific DNA methylation in polycystic ovary syndrome.
Echiburú, B., Maliqueo, M., Sánchez, F., Perez-Bravo, F., and Cassorla, F. Fertil. Steril. 104, 145–153.e6.
(2012). Relationship between anti-Mullerian hormone (AMH) and insulin levels
Zeng, Y., and Chen, T. (2019). DNA methylation reprogramming during
during different tanner stages in daughters of women with polycystic ovary
mammalian development. Genes (Basel) 10, 257.
syndrome. Reprod. Sci. 19, 383–390.
Sir-Petermann, T., Maliqueo, M., Angel, B., Lara, H.E., Pérez-Bravo, F., and Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E.,
Recabarren, S.E. (2002). Maternal serum androgens in pregnant women with Nusbaum, C., Myers, R.M., Brown, M., Li, W., and Liu, X.S. (2008). Model-
polycystic ovarian syndrome: possible implications in prenatal androgeniza- based analysis of ChIP-seq (MACS). Genome Biol 9, R137.
tion. Hum. Reprod. 17, 2573–2579. Zhao, Y., Zhang, C., Huang, Y., Yu, Y., Li, R., Li, M., Liu, N., Liu, P., and Qiao, J.
€ng, A. (2005). Rats with
Stener-Victorin, E., Ploj, K., Larsson, B.M., and Holma (2015). Up-regulated expression of WNT5a increases inflammation and oxida-
steroid-induced polycystic ovaries develop hypertension and increased sym- tive stress via PI3K/AKT/NF-kB signaling in the granulosa cells of PCOS pa-
pathetic nervous system activity. Reprod. Biol. Endocrinol. 3, 44. tients. J. Clin. Endocrinol. Metab. 100, 201–211.

ll
Article OPEN ACCESS
STAR+METHODS
KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER

Antibodies
Mouse monoclonal anti-50 -methylcytosine Diagenode Cat#C15200081; RRID: AB_2572207
Mouse IgG Diagenode Cat#C15400001; RRID: AB_2722553
Guinea pig polyclonal anti-insulin DAKO Cat# IR00261-2; RRID: AB_2800361
Rabbit monoclonal anti-glucagon antibody Abcam Cat# ab92517; RRID: AB_10561971
Biological samples
Patient-derived blood Jeanne de Flandre http://maternite.chru-lille.fr
Hospital, France
Chemicals, peptides, and recombinant proteins
€llerian hormone (AMH)
Recombinant human anti-Mu R&D Systems Cat#1737-MS-10
Hematoxylin-eosin Sigma Aldrich Cat#GHS132
Trizol ThermoFisher Scientific Cat #15596026
S-adenosylmethionine (SAM) England Biolegends Cat #B9003S
Dichloromethane Sigma-Aldrich Cat# 270997
Benzyl ether Sigma-Aldrich Cat# 108014
Methanol VWR Chemicals Cat# 20847.360
Hydrogen peroxide solution Sigma-Aldrich Cat# 216763
Thimerosal Sigma-Aldrich Cat# T8784-5g
Triton X100 Sigma-Aldrich Cat# X100-500ml
Critical commercial assays
Mouse T ELISA Kit Demeditec Diagnostics Cat#DEV9911
RNeasy Lipid Tissue Mini Kit Qiagen Cat # 74804
RNA-to-cDNA kit Applied Biosystems Cat #4387406
TruSeq Stranded mRNA Library Prep Kit and TruSeq Illumina Cat # IP-202-1012; IP-202-1024
RNA Single Indexes kits A and B
MagMeDIP kit Diagenode Cat # C02010021
Qubit dsDNA HS Assay Kit Thermo Fisher Cat # Q32851
QIamp DNA blood Mini kit Qiagen Cat # 51104
DNeasy Blood & Tissue Kit Qiagen Cat #69504
Mouse Insulin ELISA Kit Mercodia Cat #10-1247-01
Deposited data
Raw and analyzed data This paper GEO: GSE148839
Experimental models: organisms/strains
C57BL/6J mice Charles River N/A
Oligonucleotides
Sorbs2-TTS_forward primer (MeDIP-PCR human samples): This paper N/A
CAGCCTCCTGGAGACACTTT
Sorbs2-TTS_reverse primer (MeDIP-PCR human samples): This paper N/A
CACGTCAAAATGTGGGATCA
Robo1_forward primer (MeDIP-PCR human samples): This paper N/A
AGGCAAGTTCTGCTCCTCAA
Robo1_reverse primer (MeDIP-PCR human samples): This paper N/A
TTCTCATTCGCCTGGATTTC
Cdkn1a-prom_forward primer (MeDIP-PCR human samples): This paper N/A
GCAGAGAGGTGCATCGTTTT
(Continued on next page)
Cell Metabolism 33, 513–530.e1–e8, March 2, 2021 e1

ll
OPEN ACCESS Article
Continued
Cdkn1a-prom_reverse primer (MeDIP-PCR human samples): This paper N/A
TCTGGCAGGCAAGGATTTAC
Hdc-inter_forward primer (MeDIP-PCR human samples): This paper N/A
CCTGCCCAATTCAATCTGTT
Hdc-inter_reverse primer (MeDIP-PCR human samples): This paper N/A
CCGGAATAGGGTGGTAGTCA
Igfbpl1-prom_forward primer (MeDIP-PCR human samples): This paper N/A
CACCCCTCATGTGTCTTGTG
Igfbpl1-prom_reverse primer (MeDIP-PCR human samples): This paper N/A
CCAAGAACGGTGTAGCTGGT
Irs4-prom_forward primer (MeDIP-PCR human samples): This paper N/A
TAACCAGTGTTGGGCTGTGA
Irs4-prom_reverse primer (MeDIP-PCR human samples): This paper N/A
TCTCGCTCAAGGAAAGGAAA
Uhrf1_forward primer (MeDIP-PCR human samples): This paper N/A
TCTTCCGGGTTTGTCATCTC
Uhrf1_reverse primer (MeDIP-PCR human samples): This paper N/A
CGAACTTGGCAGGTAGGAAG
Tet1-prom_forward primer (MeDIP-PCR human samples): This paper N/A
GAGGGCTCTCACACTTCCTG
Tet1-prom_reverse primer (MeDIP-PCR human samples): This paper N/A
GCTCAGTTTCCTCCAGCAAC
Sorbs2 _forward primer (MeDIP-PCR mouse samples): This paper N/A
CTATCATCATTGCGGGTTCA
Sorbs2_reverse primer (MeDIP-PCR mouse samples): This paper N/A
ATGGGACAACCTGACACACA
Robo1_forward primer (MeDIP-PCR mouse samples): This paper N/A
GCAAAGCTCTCTGCTTTTGAA
Robo1_reverse primer (MeDIP-PCR mouse samples): This paper N/A
AACGTGTCACTCCCTTCTCC
Cdkn1a _forward primer (MeDIP-PCR mouse samples): This paper N/A
GTATGCTGCCACAACCACAC
Cdkn1a_reverse primer (MeDIP-PCR mouse samples): This paper N/A
GCTGCTACTTGGCACCAGTT
Hdc _forward primer (MeDIP-PCR mouse samples): This paper N/A
GAACGTCATTCTCACAGAGCA
Hdc_reverse primer (MeDIP-PCR mouse samples): This paper N/A
CCCATTCTCACCACCGATTA
Uhrf1-peak1_forward primer (MeDIP-PCR mouse samples): This paper N/A
TCTTCCCTGGGCTTCCTAGT
Uhrf1-peak1_reverse primer (MeDIP-PCR mouse samples): This paper N/A
AGCGAGCTTCACACACACAG
Uhrf1-peak2_forward primer (MeDIP-PCR mouse samples): This paper N/A
TTTGTGGGTCTGACAACTGG
Uhrf1-peak2_reverse primer (MeDIP-PCR mouse samples): This paper N/A
TGGAAGGCATATGCTCAGTG
Tet1_forward primer (MeDIP-PCR mouse samples): This paper N/A
TGGGATCCTATGTCCTCTTCC
Tet1_reverse primer (MeDIP-PCR mouse samples): This paper N/A
TCCACCTGCCTCTACCTTTTT
Gremlin 1 TermoFisher Scientific Cat#Mm00488615_s1
Follistatin TermoFisher Scientific Cat#Mm00514982_m1
Roundabout homolog 1 (Drosophila) TermoFisher Scientific Cat#Mm00803879_m1
(Continued on next page)
e2 Cell Metabolism 33, 513–530.e1–e8, March 2, 2021

ll
Article OPEN ACCESS
Continued
insulin degrading enzyme TermoFisher Scientific Cat#Mm00473077_m1
prostaglandin-endoperoxide synthase 2 TermoFisher Scientific Cat#Mm00478374_m1
insulin-like growth factor binding protein 6 TermoFisher Scientific Cat#Mm00599696_m1
ephrin B1 TermoFisher Scientific Cat#Mm00438666_m1
tumor necrosis factor receptor superfamily, member 12a TermoFisher Scientific Cat#Mm01302476_g1
Jun D proto-oncogene TermoFisher Scientific Cat#Mm04208316_s1
cyclin-dependent kinase inhibitor 1A (P21) TermoFisher Scientific Cat#Mm04205640_g1
growth arrest specific 2 TermoFisher Scientific Cat#Mm00433519_m1
aquaporin 8 TermoFisher Scientific Cat#Mm01278161_m1
actin, beta TermoFisher Scientific Cat#Mm00607939_s1
ubiquitin-like, containing PHD and RING finger domains, 1 TermoFisher Scientific Cat#Mm00477872_m1
tet methylcytosine dioxygenase 1 TermoFisher Scientific Cat#Mm01169087_m1
sorbin and SH3 domain containing2 TermoFisher Scientific Cat#Mm01200787_m1
histidine decarboxylase TermoFisher Scientific Cat#Mm00456104_m1
nuclear factor of kappa light polypeptide gene enhancer in B cells TermoFisher Scientific Cat#Mm00476361_m1
mannan-binding lectin serine peptidase 1 TermoFisher Scientific Cat#Mm00434830_m1
allograft inflammatory factor 1 TermoFisher Scientific Cat#Mm00479862_g1
interleukin 6 TermoFisher Scientific Cat#Mm00446190_m1
tumor necrosis factor alpha TermoFisher Scientific Cat#Mm00443258_m1
interleukin 1 beta TermoFisher Scientific Cat#Mm00434228_m1
nuclear factor of kappa light polypeptide gene enhancer in B cells 1 TermoFisher Scientific Cat#Mm00476361_m1
Kisspeptin 1 N/A Cat#Mm03058560_m1
gonadotropin releasing hormone 1 N/A Cat#Mm01315605_m1
glyceraldehyde-3-phosphate dehydrogenase N/A Cat#Mm99999915_g1
Software and algorithms
Prism 8 Version 8.3.1 GraphPad Software, LLC https://www.graphpad.com/
scientific-software/prism/
Inspector software LaVision Biotec http://www.lavisionbiotec.com/
Imaris x64 software (version 8.0.1) Bitplane http://www.bitplane.com/imaris/imaris
iMovie (version 10.1.1) Apple http://www.apple.com/fr/imovie/
Other
MiniSpec mq 7.5, Nuclear Magnetic Resonance Bruker N/A
Glucometer Life scan OneTouch Verio
Bioruptor Plus sonicator Diagenode Cat # B01020001
RESOURCE AVAILABILITY
Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Paolo
Giacobini (paolo.giacobini@inserm.fr).
Materials availability
This study did not generate new unique reagents.
Data and code availability

The accession number for the RNA-seq and MeDIPseq data from mouse ovaries from CNTR and PAMH F3 females reported in this
paper is GEO: GSE148839. All the other data supporting the findings of this study are available within the article and its supplemen-
tary information files and from the lead contact author upon request.

ll
OPEN ACCESS Article
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Human studies
Blood samples have been collected prospectively, from 2003 to 2008, to perform genetic studies, in the Reproductive Medicine
Department of Jeanne de Flandre in Lille University Hospital, France. Biological and clinical data about patients were collected at
the same time. This study was approved by the Ethics Committee of Lille University Hospital (DRC BT/JR/DS/N 0231 PROM
02-563 CP 03/11). Written informed consent was obtained for all patients. Patients were initially referred to our department for hyper-
androgenism (HA) and/or oligo-anovulation and/or infertility. The diagnosis of PCOS was based on the presence of at least 2 out of
the 3 following Rotterdam criteria (Rotterdam, 2004), i.e.,: 1) HA (clinical or biological). Clinical HA was defined by the presence of
hirsutism (modified Ferriman-Gallwey score over 7 and/or acne located in more than two areas). Hyperandrogenism was defined as a
serum TT level > 0,39 ng/ml and/or a serum androstenedione level (A) >2,2 ng/ml, as previously reported (Dewailly et al., 2011) 2)
oligo-anovulation, (i.e. oligomenorrhea or amenorrhea); 3) presence of Polycystic Ovarian Morphology (PCOM) at ultrasound (U/S),
with an ovarian area R 5.5 cm2 and/or a follicle number per ovary R 12, unilaterally or bilaterally. Women with congenital adrenal
hyperplasia, Cushing syndrome, androgen secreting tumor or hyperprolactinemia were excluded. Women with PCOS were asked
about familial history and the genetic study was also proposed to their mothers and sisters. The latter were asked about their personal
clinical history (age, body mass index, age of first menstruations, cycle length, presence of hirsutism or acne). For sisters who didn’t
have any contraceptive treatment, hormonal assays were also performed in the follicular phase. Based on this information, they were
classified as women with PCOS or control, if possible.
Biochemical and hormonal measurement tests were performed in the central biochemistry department of Lille and included:
estradiol, LH and FSH, total testosterone, delta4 androstenedione, 17-hydroxyprogesterone, SDHEA, SBP, prolactinemia, fasting
glucose, insulinemia and lipid profile. Estradiol, androstenedione, testosterone, LH and FSH were measured by immunoassays as
previously described (Dewailly et al., 2011; Pigny et al., 1997). Fasting serum insulin levels were measured in duplicate by an immu-
noradiometric assay (Bi-Insulin IRMA Pasteur, Bio-Rad, Marnes la Coquette, France) that uses two monoclonal anti-insulin anti-
bodies. Intra and interassay coefficient of variation were <3,8 and <7,5% respectively. Results are expressed as milli international
units per liter.
47 blood samples have been recently analyzed from 32 women with PCOS (18-65 years old) and 15 women without PCOS (22-66
years old). Among the 32 PCOS women, five were born from PCOS mothers (23-30 years old) and among the 15 control women,
3 were confirmed to be born from control mothers (22-36 years old).
All procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on
human experimentation and with the 1975 Declaration of Helsinki, as revised in 2008.
Mouse models
All C57BL/6J (B6) mice (Charles River, USA) were group-housed under specific pathogen-free conditions in a temperature-controlled
room (21-22 C) with a 12-h light/dark cycle and ad libitum access to food and water. Standard diet (9.5 mm Pelleted RM3, Special
Diets Services, France) was given to all mice during breeding, lactation and growth of young stock. Nutritional profile of the standard
diet RM3 is the following: Protein 22.45%, Fat 4.2%, Fiber 4.42%, Ash 8%, Moisture 10%, Nitrogen free extract 50.4%; Calories:
3.6 kcal/gr. F0 C57BL/6J mice were used directly after arrival into the animal facility and after an acclimatization period of at least
2 weeks.
Mice were randomly assigned to groups at the time of purchase or weaning to minimize any potential bias. No data sets were
excluded from analyses. The animals were daily for health issues by qualified personnel; health status was normal for all animals.
Animal studies were approved by the Institutional Ethics Committees of Care and Use of Experimental Animals of the University of
Lille (France; Ethical protocol number: APAFIS#2617-2015110517317420 v5 and APAFIS#13387-2017122712209790 v9). All exper-
iments were performed in accordance with the guidelines for animal use specified by the European Council Directive of 22 September
2010 (2010/63/EU). The sample size, sex and age of the animals used is specified in the text and/or figure legends.
METHOD DETAILS
Prenatal anti-Mu € llerian hormone (PAMH) treatment

PAMH animals have been generated as previously described (Tata et al., 2018). Timed-pregnant adult (3-4 months) C57BL6/J (B6)
dams were injected daily intraperitoneally (i.p.) from embryonic day (E) 16.5 to 18.5 with 200 mL of a solution containing respectively:
1) 0.01 M phosphate buffered saline (PBS, pH 7.4, prenatal control-treated, CNTR), 2) PBS with 0.12 mgKg-1/d human anti-Mu €llerian
hormone (AMH) (AMHC, R&D Systems, rhMIS 1737-MS-10, prenatal AMH (PAMH)-treated).
Mouse breeding scheme and feeding paradigm to generate F1–F3 offspring

PAMH female offspring (F1) were mated with F1 PAMH unrelated males to generate PAMH F2 offspring, and a subset of PAMH F2
female offspring were mated with PAMH F2 unrelated males to generate PAMH F3 offspring. The remaining F1, F2 and F3 female
offspring were subjected to phenotypic testing as described below. Control male or female offspring (CNTR) used in this study
were generated by prenatally treating gestating mice with PBS from E16.5 to E18.5 as described above.

ll
Article OPEN ACCESS
The exact number of mice used for each procedure and their sex and age are given in the figure legends and/or text. Details of the
number of mice used for (1) phenotypic testing and (2) breeding to generate F1, F2 and F3 offspring in each group are specified in in
the figure legends and/or text. To ensure variability within each group, offspring in each generation were randomly allocated for
phenotypic testing or breeding.
Assessment of phenotype, estrous cycle, and fertility

Control F1 and PAMH F1-F3 female offspring were weaned at post-natal day P21 and checked for vaginal opening (VO) and time of
first estrus. Anogenital distance (AGD) and body mass (grams, g) were measured at different ages during post-natal development
(P30, 35, 40, 50 and 60). At VO and in adulthood (P60), vaginal smears were performed daily for 16 consecutive days (4-cycles)
for analysis of age of first estrus and estrous cyclicity. Vaginal cytology was analyzed under an inverted microscope to identify
the specific stage of the estrous cycle. The reproductive competency of these animals was determined by pairing the following
mice: CNTR F1 females mated with CNTR F1 males, CNTR F1 males mated with PAMH F1-F3 females, PAMH F1-F3 females mated
with PAMH F1-F3 males, for a period of 3 months. Unexperienced males and primiparous females, selected from at least three
different litters, were used for the 90-days mating protocol test. Number of pups/litter (number of pups), fertility index (number of lit-
ters per females over 3 months), and time to first litter (number of days to first litter after pairing) were quantified per treatment and
pairing.
Ovarian histology
Ovaries were collected from 3-month-old diestrus mice, immersion-fixed in 4% PFA solution and stored at 4 C. Paraffin-embedded
ovaries were sectioned at a thickness of 5 mm (histology facility, University of Lille 2, France) and stained with hematoxylin-eosin
(Sigma Aldrich, Cat # GHS132, HT1103128). Sections were examined throughout the ovary. Total numbers of corpora lutea (CL)
were classified and quantified as previously reported (Caldwell et al., 2017). To avoid repetitive counting, each follicle was only
counted in the section where the oocyte’s nucleolus was visible. To avoid repetitive counting, CL were counted every 100 mm by
comparing the section with the preceding and following sections. CL were characterized by a still present central cavity, filled
with blood and follicular fluid remnants or by prominent polyhedral to round luteal cells.
LH and T ELISA assays

LH levels were determined by a sandwich ELISA, using a well-established ELISA method (Steyn et al., 2013; Tata et al., 2018). A
mouse LH-RP reference was provided by Albert F. Parlow (National Hormone and Pituitary Program, Torrance, California, USA).
The assay sensitivity of the LH ELISA was 0.04 ng/ml and intra-assay coefficient of variation was 4.3%. Plasma T levels were analyzed
using a commercial ELISA (Demeditec Diagnostics, GmnH, DEV9911) according to the manufacturers’ instructions as previously
validated in other studies (Moore et al., 2015; Tata et al., 2018). The assay sensitivity is 0.066 ng/ml at the 2 standard deviation con-
fidence limit. Intra-assay coefficient of variation for testosterone was 7 % and inter-assay coefficient of variation was 11.0%. 20 ml of
plasma was used for the testosterone assay, run in duplicate samples.
As T and LH levels can vary significantly due to handling, sampling techniques and depending on the time of the day, female adult
mice were habituated with daily handling for 3 weeks. Blood samples were taken from trunk blood, for T measurements, and from the
tail, for LH measurements, between 10h00 and 12h00 during diestrus.
Body weight and composition

Whole body fat, fluids, and lean tissue mass were determined by Nuclear Magnetic Resonance. (MiniSpec mq 7.5, RMN Analyser,
Bruker) according to the manufacturer’s recommendations.
Measurement of fasting blood glucose and insulin level

Fasting blood glucose levels were assessed after animals were fasted for 12 h (starting from 8:00 PM). Blood glucose levels were
determined in blood samples from the tail vein at 8:00 AM using an automatic glucometer (OneTouch Verio, Life scan). Fasting Insulin
levels were determined in plasma samples after sacrifice, by a sandwich Insulin ELISA (Mercodia, Cat #10-1247-01) according to the
manufacturers’ instructions.
Glucose and insulin tolerance tests

For intraperitoneal glucose tolerance test (ipGTT), animals were subjected to an overnight fasting (14 h food withdrawal). For intra-
peritoneal insulin tolerance test (ipITT), mice were fasted for 4 h. Either glucose (2 g/kg body weight) or human normal insulin (0.75 U/
kg body weight) were injected intraperitoneally at 0 (prior to glucose or insulin administration) and blood was collected from the tail
vein at different time points (0, 15, 30, 45, 60, 120, 150). Plasma glucose was measured using an automatic glucometer (OneTouch
Verio, Life scan).
Optical clearing of mouse pancreas

Whole-organ staining and clearing were performed using iDISCO+ (Renier et al., 2014). Pancreata were dissected from perfused an-
imals and processed as described below. Samples were dehydrated [20, 40, 60, 80, and 100% methanol at room temperature (RT)],
delipidated [100% dichloromethane (DCM; Sigma-Aldrich)], and bleached in 5% H2O2 (overnight, 4 C). Pancreata were rehydrated

ll
OPEN ACCESS Article
(80, 60, 40, and 20% methanol) and permeabilized [PBS/0.2% TritonX-100 (twice for 1h)] before proceeding to the staining proced-
ures. Samples were incubated at 37 C on an adjustable rotator in 10 ml of a blocking solution (PBSGNaT) of 1X PBS containing 0.2%
gelatin (Sigma-Aldrich), 0.5% Triton X-100 (Sigma-Aldrich) and 0.01% NaAzide for 3 nights. Samples were transferred to 10 ml of
PBSGNaT containing primary antibodies (FLEX polyclonal guinea pig anti-insulin ready-to-use, DAKO IR00261-2, 1:2; monoclonal
rabbit anti-glucagon antibody [EP3070], Abcam ab92517, 1:500) and placed at 37 C in rotation for 7 days. This was followed by
six washes of 30 min in PBSGT at RT and a final wash in PBSGT overnight at 4 C. Next, samples were incubated in secondary an-
tibodies: goat anti-rabbit Alexa Fluor 647 conjugated antibody (Life Technologies, 1:500), goat anti-Guinea Pig Alexa Fluor 568 con-
jugated antibody (Life Technologies, 1:500), diluted in 10 ml PBSGNaT for 5 days at 37 C in a rotating tube. After six 30-min washes in
PBS at room temperature, the samples were stored in PBS at 4 C in the dark until clearing. Samples were washed with PBSGNaT
(five times, RT) and PBS (five times, RT), dehydrated with a methanol gradient, then washed in 100% methanol (three times, 1 hr each)
and incubated overnight with 66% DCM/33% Methanol and then in 100% DCM until the sample sank (15 min to 45 min). The clearing
step was performed in 100% Benzyl ether (DBE; Sigma-Aldrich).
Light-sheet imaging
3D imaging was performed as previously described (Belle et al., 2014). An ultramicroscope (LaVision BioTec) using ImspectorPro
software (LaVision BioTec) was used to perform imaging. The light sheet was generated by a laser (wavelength 488 or 561 nm,
Coherent Sapphire Laser, LaVision BioTec) and two cylindrical lenses. A binocular stereomicroscope (MXV10, Olympus) with a
23 objective (MVPLAPO, Olympus) was used at different magnifications (1.63, 43, 53, and 6.33). Samples were placed in an im-
aging reservoir made of 100% quartz (LaVision BioTec) filled with DBE and illuminated from the side by the laser light. A PCO Edge
SCMOS CCD camera (2560 3 2160 pixel size, LaVision BioTec) was used to acquire images. The step size between each image was
fixed at 2 mm.
Images, 3D volume, and movies were generated using Imaris x64 software (version 7.6.1, Bitplane). Stack images were first con-
verted to imaris file (.ims) using ImarisFileConverter and 3D recontruction was performed using ‘‘volume rendering’’. Optical slices of
samples were obtained using the ‘‘orthoslicer’’ tools. The surface of the samples was created using the ‘‘surface’’ tool by creating a
mask around each volume. 3D pictures were generated using the ‘‘snapshot’’ tool. Adobe Photoshop CC 2019 (Adobe Systems, San
Jose, CA, USA) was used to process, adjust and merge the photomontages. Figures were prepared using Adobe Photoshop CC.
Immunohistochemistry on pancreatic sections

Pancreata from control (n = 4), PAMH F3 (n = 5) and PAMHF3 + SAM (n = 5) animals were collected and immersed in 4% parafor-
maldehyde in PBS for 4 hours. Pancreata were then rinsed in PBS, dehydrated and embedded in paraffin, and the blocks were
sectioned on a microtome at 8 mm thickness.
Before immunofluorescence experiments, the sections underwent deparaffinization: incubation in xylene 2 times x 5 min, then
rinsed with 100% ethanol (2 times x 5 minutes), and rehydrated in a decreasing gradient of Ethanol/H20 (100%, 90%, 75%, 5 minutes
each) and finally transferred to PBS.
Antigen retrieval was performed as follows: the slides were incubated for 30 minutes in 10 mM citrate buffer heated to 95 C. They
were brought back to room temperature and rinsed with PBS. The sections were blocked for 30 minutes in blocking buffer (PBS + 7%
NGS + 0.3% triton X100), then incubated with primary antibodies in blocking buffer for 48 hours at 4 C. After three rinses in PBS, the
sections were incubated for 1h at RT with secondary antibodies diluted in blocking buffer. Finally, the sections were rinsed three
times in PBS before nuclear staining with DAPI (1:5000 in PBS) and mounting with Mowiol. The slides were allowed to dry before
imaging. Primary antibodies used were: FLEX polyclonal guinea pig anti-insulin ready-to-use (DAKO IR00261-2, 1:2), recombinant
monoclonal rabbit anti-glucagon antibody [EP3070] (Abcam ab92517, 1:500). Secondary antibodies: goat anti-rabbit Alexa Fluor
647 conjugated antibody (Life Technologies, 1:500), goat anti-Guinea Pig Alexa Fluor 568 conjugated antibody (Life Technolo-
gies, 1:500).
Image acquisition and analysis

Single-plane acquisitions of pancreatic sections from control (n = 4), PAMH F3 (n = 5) and PAMHF3 + SAM (n = 5) animals were per-
formed on a Zeiss LSM 710 confocal microscope equipped with a 20X/0.8 objective. Insulin staining was used to measure the area of
the pancreatic islets. For each animal, 3 to 4 random regions of the pancreas were imaged, and all pictures were processed in the
same manner using Fiji: pictures were first converted to 8bits, and the insulin staining was segmented by manual thresholding with
constant parameters (set to 20-255). The area of the segmented signal was then quantified using the measurement tool.
RNA extraction and RT-qPCR

Ovaries, perigonadal fat and hypothalami were harvest from control F1 and PAMH F3 (treated or not with SAM) female mice and were
snap-frozen in liquid nitrogen. Frozen tissues were homogenized using 1 ml of Trizol (ThermoFisher Scientific, Cat #15596026) with a
tissue homogenizer and total RNA was isolated using RNeasy Lipid Tissue Mini Kit (Qiagen; Cat # 74804) following the manufacturer’s
instructions. For gene expression analyses, cDNA was synthetized from 1000ng of total RNA using the High Capacity RNA-to-cDNA
kit (Applied Biosystems, Cat #4387406) using the manufacturer’s recommended cycling conditions. Real-time PCR was carried
out on Applied Biosystems 7900HT Fast Real-Time PCR system using exon-boundary-specific TaqMan Gene Expression Assays

ll
Article OPEN ACCESS
(TermoFisher Scientific). Data were analyzed by using the 2-DDCT method (Livak and Schmittgen, 2001) and normalized to
housekeeping genes Beta-actin (ActB) levels. Values are expressed relative to control values, as appropriate, set at 1.
RNA libraries and sequencing

RNA-Seq libraries were generated from 600 ng of total RNA using TruSeq Stranded mRNA Library Prep Kit and TruSeq RNA Single
Indexes kits A and B (Illumina, San Diego, CA), according to manufacturer’s instructions. Briefly, following purification with poly-T
oligo attached magnetic beads, the mRNA was fragmented using divalent cations at 94 C for 2 minutes. The cleaved RNA fragments
were copied into first strand cDNA using reverse transcriptase and random primers. Strand specificity was achieved by replacing
dTTP with dUTP during second strand cDNA synthesis using DNA Polymerase I and RNase H. Following addition of a single ’A’
base and subsequent ligation of the adapter on double stranded cDNA fragments, the products were purified and enriched with
PCR (30 sec at 98 C; [10 sec at 98 C, 30 sec at 60 C, 30 sec at 72 C] x 12 cycles; 5 min at 72 C) to create the cDNA library. Surplus
PCR primers were further removed by purification using AMPure XP beads (Beckman-Coulter, Villepinte, France) and the final cDNA
libraries were checked for quality and quantified using capillary electrophoresis. Libraries were then single-read sequenced with a
length of 50 pb, with 8 samples per lane on an Illumina Hiseq4000 sequencer. Image analysis and base calling were carried out using
RTA v.2.7.3 and bcl2fastq v.2.17.1.14. Reads were mapped onto the mm10 assembly of Mus musculus genome using STAR (Dobin
et al., 2013) v.2.5.3a. Gene expression was quantified from uniquely aligned reads using HTSeq-count (Anders et al., 2015) v.0.6.1p1
with annotations from Ensembl release and union mode. Data quality was evaluated with RSeQC (Wang et al., 2012). Comparisons of
read counts were performed using R 3.5.1 with DESeq2 (Love et al., 2014) v1.22.1 Bioconductor package. More precisely, counts
were normalized from the estimated size factors using the median ratio method and a Wald test was used for the statistical test. Un-
wanted variation was identified with sva (Leek, 2014) and considered in the statistical model. To reduce false positive, p-values were
adjusted by IHW method (Ignatiadis et al., 2016).
MeDIP
MeDIP was performed using MagMeDIP kit (Diagenode) according to the manufacturer’s instructions. Briefly, frozen mouse ovaries
(dissected at dioestrus) were chopped and lysed in 1mL GenDNA digestion buffer and DNA was extracted using phenol:chlorofor-
m:isoamyl alcohol (25:24:1). DNA extraction of the liver and hypothalamic tissues were performed using the DNeasy Blood & Tissue
Kit (Qiagen) according to the manufacturer’s instructions. DNA was quantified using the QubitTM DNA BR Assay kit. 1.1 mg of DNA
was sheared by sonication for six cycles with 30 s ON and 30 s OFF at 4 C using the Bioruptor Plus sonicator (Diagenode). Immu-
noprecipitation was performed using an anti-50 -methylcytosine mouse monoclonal antibody (Diagenode; Cat nr: C15200081; Lot nr:
RD004; 0.2 mg/immunoprecipitation) or a mouse IgG as a negative control (Diagenode; Cat nr: C15400001; Lot nr: MIG002S; 0.2 ug/
immunoprecipitation) and magnetic beads, following MagMeDIP kit settings. One-tenth of the DNA sample was set aside at 4 C for
input. To check the efficiency of the MeDIP experiment, spike-in controls including unmethylated (unDNA) and in vitro methylated
DNA (meDNA) from A. thaliana were used. After magnetic beads washes, methylated DNA was isolated using the DNA Isolation
Buffer protocol according to the MagMeDIP kit recommendations. DNA concentration was measured using Qubit dsDNA HS Assay
Kit (Thermo Fisher). Efficiency of the immunoprecipitation was assessed by performing qPCR using meDNA and unDNA primers.
MeDIP experiments from human blood were carried using the MagMeDIP protocol as described above with some modifications.
DNA was extracted from 200 mL of frozen blood using the QIamp DNA blood Mini kit (Qiagen) according to the manufacturer’s in-
structions. RNase A was added prior to cell lysis. DNA was eluted in 100 mL of water. Efficiency of the immunoprecipitation of the
liver, hypothalamus and human blood samples were assessed by performing qPCR for the mouse/human TSH2B (methylated region)
and GAPDH (unmethylated region) (primers provided in the MagMeDIP kit). Methylation quantification was calculated from qPCR
data and reported as the recovery of starting material: % (meDNA-IP/Total input) = 2^[(Ct(10%input)-3.32) Ct(meDNA-
IP)] 3 100%.
MeDIP-seq - libraries construction and sequencing

Libraries were prepared using the SMART cDNA Library Construction Kit and sequenced on Illumina Hiseq 4000 sequencer as sin-
gle-end 50 bp reads following Illumina’s instructions. Image analysis and base calling were performed using RTA 2.7.3 and bcl2fastq
2.17.1.14. Adapter dimer reads were removed using Dimer Remover. Data were preprocessed with Cutadapt v1.13 (Martin, 2011) to
remove the first 9 nucleotides and to remove sequences with a trailing polyT of at least 10 Ts. Cutadapt was used with the following
parameters ‘-u 9 -a T(10) –discard-trimmed’. Reads were mapped to the mouse genome (mm10) using Bowtie v1.0.0 (Langmead
et al., 2009) with default parameters except for ‘‘-p 3 –m 1 –strata –best’’. Methylated regions were detected using MACS v1.4.2
(Zhang et al., 2008) with default parameters except for ‘‘-g mm -p 1e-3’’. Regions were then annotated with the closest genes
with Homer v4.9.1 annotatePeaks.pl (Heinz et al., 2010) with Ensembl v90 annotations.
All regions found in at least 2 replicates of the same condition were retained for the detection of differentially methylated regions.
They were then combined to get the union of all peaks using the tool Bedtools merge v2.26.0 (Quinlan and Hall, 2010). Read counts
were normalized across libraries using the method proposed by (Anders and Huber, 2010). Statistical comparisons of interest were
performed using the method proposed by Love et al. (2014) implemented in the DESeq2 v1.22.2 Bioconductor library. p values were
adjusted for multiple testing using the Benjamini and Hochberg (1995) method. MA plot and Manhattan plots are been generated
using custom R scripts.

ll
OPEN ACCESS Article
Methyl donor S-adenosylmethionine (SAM) treatment

Vaginal cytology was analyzed under an inverted microscope to record the specific stage of the estrous cycle. PAMH F3 offspring
were injected intraperitoneally (i.p.) daily for 15 days with 200 mL of a solution containing 0.01M phosphate buffered saline (PBS, pH
7.4) or with SAM (50 mg/Kg/day; New England Biolegends, Cat. B9003S). This concentration was chosen based on previous in vivo
pharmacological studies using the same drug (Li et al., 2012). Estrous cyclicity was analyzed in adult CNTR offspring (prenatally PBS-
treated; Group 1, n = 5, 6 month-old) for 25 days and in PAMH F3 animals during 10 days before treatment. One group of PAMH F3
animals (Group 2; n = 5) was then injected daily with PBS and another group of animals (Group 3; n = 5) was injected with SAM for
15 days.
Tail-blood samples were collected for LH and T measurements at diestrus before the beginning of the treatments, at day 10, and at
the end of the treatment, at day 25.
QUANTIFICATION AND STATISTICAL ANALYSIS
All analyses were performed using Prism 8 (Graphpad Software, San Diego, CA) and assessed for normality (Shapiro–Wilk test and/or
D’Agostino & Pearson test) and variance before subsequent analyses by several statistical tests. Sample sizes were chosen accord-
ing to standard practice in the field. The investigators were not blinded to the group allocation during the experiments. However, an-
alyses were performed by two independent investigators in a blinded fashion. For each experiment, replicates are described in the
figure legends. All comparisons between groups, whose distribution was not normal, were performed using Mann-Whitney U test
(comparison between two experimental groups) or Kruskal-Wallis test (comparison between three or more experimental groups) fol-
lowed by a Dunn’s post hoc analysis. For analyses of populations normally distributed, data were compared using an unpaired two-
tailed Student’s t test or a one-way ANOVA for multiple comparisons followed by Tukey’s multiple comparison post hoc test. The
number of biologically independent experiments, sample size, p values, age and sex of the animals are all indicated in the main
text or figure legends. All experimental data are indicated as mean ± SEM or as the 25th–75th percentile, line at median. The signif-
icance level was set at p < 0.05.

Polycystic Ovary - Cell Metabolism

Uploaded by

Copyright:

Available Formats

Polycystic Ovary - Cell Metabolism

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Polycystic Ovary - Cell Metabolism

Uploaded by

Copyright:

Available Formats

Article

Polycystic ovary syndrome is transmitted via a

d Metabolic- and inflammatory-related pathways are

d Common hypomethylation signatures occur in a mouse

d Identification of a novel epigenetic-based therapeutic

Mimouni et al., 2021, Cell Metabolism 33, 513–530

FHU 1000 days for health, 59000 Lille, France

Strasbourg 67000, France

Université de Strasbourg, Illkirch, France

*Correspondence: laurette@unistra.fr (A.-L.B.), paolo.giacobini@inserm.fr (P.G.)

Cell Metabolism 33, 513–530, March 2, 2021 515

(legend continued on next page)

Cell Metabolism 33, 513–530, March 2, 2021 517

(legend on next page)

518 Cell Metabolism 33, 513–530, March 2, 2021

Cell Metabolism 33, 513–530, March 2, 2021 519

Cell Metabolism 33, 513–530, March 2, 2021 521

(legend continued on next page)

Cell Metabolism 33, 513–530, March 2, 2021 523

(legend continued on next page)

Cell Metabolism 33, 513–530, March 2, 2021 525

526 Cell Metabolism 33, 513–530, March 2, 2021

Cell Metabolism 33, 513–530, March 2, 2021 527

PCOS across generations. However, we have not performed ACKNOWLEDGMENTS

528 Cell Metabolism 33, 513–530, March 2, 2021

Cell Metabolism 33, 513–530, March 2, 2021 529

530 Cell Metabolism 33, 513–530, March 2, 2021

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER

Cell Metabolism 33, 513–530.e1–e8, March 2, 2021 e1

e2 Cell Metabolism 33, 513–530.e1–e8, March 2, 2021

Data and code availability

Cell Metabolism 33, 513–530.e1–e8, March 2, 2021 e3

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Prenatal anti-Mu € llerian hormone (PAMH) treatment

Mouse breeding scheme and feeding paradigm to generate F1–F3 offspring

e4 Cell Metabolism 33, 513–530.e1–e8, March 2, 2021

Assessment of phenotype, estrous cycle, and fertility

LH and T ELISA assays

Body weight and composition

Measurement of fasting blood glucose and insulin level

Glucose and insulin tolerance tests

Optical clearing of mouse pancreas

Cell Metabolism 33, 513–530.e1–e8, March 2, 2021 e5

Immunohistochemistry on pancreatic sections

Image acquisition and analysis

RNA extraction and RT-qPCR

e6 Cell Metabolism 33, 513–530.e1–e8, March 2, 2021

RNA libraries and sequencing

MeDIP-seq - libraries construction and sequencing

Cell Metabolism 33, 513–530.e1–e8, March 2, 2021 e7

Methyl donor S-adenosylmethionine (SAM) treatment

QUANTIFICATION AND STATISTICAL ANALYSIS

e8 Cell Metabolism 33, 513–530.e1–e8, March 2, 2021

You might also like