Introduction

Small non-coding RNAs (sncRNAs) are essential regulatory elements in animal reproduction. In mammals, the onset of germline differentiation is established at early stages of embryonic development. In mice, germ cell differentiation initiates around 6 days post-coitum (dpc) (also referred to as Embryonic Day 6 or E6) in a very limited number of cells and is induced by morphogenetic proteins secreted by nearby extraembryonic tissue1. These cells, located in the epiblast and expressing alkaline phosphatase, are considered the precursors of the germ line (primordial germ cells or PGCs)2. PGCs’ differentiation and proliferation are highly regulated by the transcription factors Blimp1, Tfap2c, and Prdm143,4. Afterwards, PGCs migrate, colonize, and proliferate in the so-called genital ridges around Day 10.5 (E10.5). During the proliferation phase, epigenetic reprogramming of the genome takes place, erasing the previous epigenetic imprints, losing DNA methylation patterns and other molecular pathways, such as those mediated by non-coding RNAs (ncRNAs)5,6,7,8. In the female lineage, the formation of functional ovaries also entails the development and differentiation of multiple associated somatic cell constituents of ovarian follicles, such as theca and granulosa cells, in addition to oogonia. Meiotic activation starts around E12.5 with cytologic meiotic prophase initiation at E13.5. The maximum number of germ cells is reached at E14.5, with about 18,000 oogonia in the mouse ovary. Later, oogonia numbers will drastically drop by apoptosis, induced by meiotic checkpoints and follicle formation, leading to around 8,000 on the second postnatal day9,10. This apoptosis-mediated oocyte selection during fetal life is a key element in the establishment of the ovarian reserve and can be altered during the embryonic stage by exposure to reproductively toxic compounds, such as endocrine disruptors (ED)11.

Experimental evidence indicates that the correct differentiation of germ cells is strongly influenced by the environment in which these cells differentiate during early embryonic development. Exposure to environmental toxicants induces changes in the patterns of DNA methylation, histone modification, and disruption of regular expression of sncRNAs12,13,14,15. In fact, these alterations can also be transmitted to next generations. Transmission of epigenetic modifications can produce deleterious phenotypic effects; these epigenetic modifications are mostly detected by paternal inheritance16,17,18. Although maternal transmission of epigenetic modifications has been described, the mechanisms involved have been less explored19.

Endocrine disruptors such as vinclozolin (VZN) and phthalate monoester (MEHP) have been shown to alter the expression profiles of sncRNA12,16. Vinclozolin, commercialized as Ronilan, Curalan, Vorlan, Touche, etc., is a widely used fungicide of the dicarboximide group, detected as a contaminant in many plant foodstuffs with, mainly, an observed anti-androgenic effect20,21, but also with agonist effects on estrogenic receptors22. MEHP (mono-2-ethylhexyl phthalate), which is the bioactive metabolite of the di-2-ethylhexyl phthalate (DEHP), is a widely used plasticizer that alters steroidogenesis23,24, which in recent in vitro assays, has been reported to have both anti-estrogenic and anti-androgenic effects25,26. In vitro studies performed on human cell cultures have demonstrated that DEHP could have a negative impact on human granulosa cells at nanomolar concentrations27. Other studied EDs, such as bisphenol A, have also been reported to alter sncRNAs15,28,29, and later studies confirmed these effects on other EDs16,30,31, contributing to our current understanding of the implication of ED exposure in gene expression alterations and reproductive health problems. The transgenerational effect of EDs has been studied in male lineage and sperm alteration patterns32,33, but effects on the ovary cells and subsequent effects remain unclear.

Non-coding RNAs (ncRNAs) play an essential role in both cell differentiation and development, as well as being involved in pathogenesis in eukaryotic organisms34. ncRNAs can be categorized on the basis of their length, structure and initial putative functions. They are divided into long ncRNAs and small ncRNAs. Long ncRNAs are longer than 200 nucleotides (nt); sncRNAs usually range from 18 to 200 nt. sncRNAs can be further classified into functional or housekeeping ncRNAs and regulatory ncRNAs. Functional ncRNAs, such as rRNAs, tRNAs, and snoRNAs, have well-defined roles and have been extensively studied35,36. Regulatory ncRNAs include miRNAs and piRNAs, which are involved in gene regulation and have diverse functionalities that are still being explored. Among the sncRNAs, miRNAs (~ 22 nt in length) were the first to be identified and have been most extensively studied via their initial characterization as negative post-transcriptional regulators of mRNAs, through RNA interference mechanisms coupled to the RISC (RNA-induced silencing complex)37,38,39. In addition to reproductive effects, the deregulation of miRNA pathway has been extensively investigated and has been related to several diseases such as cancer40, neurological disorders41, cardiovascular disorders42.

piRNAs (P-element-induced wimpy testis (PIWI)-interacting RNAs) were discovered in testis43 and are most abundant in, but not restricted to, the germ line44. Initially, they were associated with control of the functional genomic integrity of germ cells through suppression of the transposability of transposable elements45. These single-stranded RNAs are usually longer than miRNA (24–31 nt). It has been reported that they generally undergo modifications in their 3’ terminal region (specifically, a 2′-O-methylation) and associate with PIWI proteins (a subfamily of the AGO family proteins) to generate specific RISCs46. Moreover, to their initial main function of enabling genome defense of transposons47, piRNAs appear to be important regulatory elements of gene expression, with capabilities not yet understood in depth48,49. Dysregulation of piRNAs has been associated with various pathologies in both germ cells and somatic cells50,51. Reported effects of misregulation of piRNAs are male infertility52,53 and cancer, including ovarian cancer54,55.

Here, we report the impact of two well-known EDs, VZN and MEHP, on sncRNA expression during early ovary development and germ cell differentiation. Examining the expression patterns of miRNAs and piRNAs at a critical developmental stage, we aimed to elucidate the effects of early exposure to these disruptors on the exposed generation and the subsequent two generations. Although there were limited alterations overall, specific significant changes in the expression of certain miRNAs and piRNAs were observed, and a significantly increased number of apoptotic cells were detected. The alterations have the potential to activate apoptotic pathways, posing a significant risk to ovarian cell development, with possible deleterious consequences in the ovarian reserve parameters and consequently in female reproductive health.

Materials and methods

Biological samples and ethical procedures

We used the outbred CD1 mouse strain (Mus musculus, Crl:CD1) provided by Charles River laboratories (France) as our animal model. All procedures related to animal care and handling were conducted at the CIB-CSIC Animal Facility under specific pathogen-free conditions, with controlled temperature (22 ± 1 °C) and humidity (50–55%). Mice were maintained on a 12-h light–dark cycle with ad libitum access to food and water. Animal management complied with the European Commission directives (2010/63/EU and 86/609/EEC) and the Spanish Royal Decree RD 53/2013 on the protection of animals used for scientific purposes. Experimental protocols were approved by the Animal Experimentation Committee of the Comunidad de Madrid and the CSIC Bioethics Committee (Permit: CAM/PROEX 054/15), in alignment with the ARRIVE guidelines to promote transparency, reproducibility, and ethical integrity. All efforts were made to minimize animal suffering and reduce overall numbers. A total of 287 female mice were used across control and experimental groups (VNZ and MEHP exposure). Detailed group sizes and the number of pregnant females and embryos are reported in Table 1 and Supplementary Excel Table S1. In order to prevent potential alterations in molecular profiles caused by anesthetic agents, the animals were euthanized by cervical dislocation without prior anesthesia, in accordance with approved protocols. In all cases, the animals were euthanized on day 13.5 of pregnancy, embryos were collected euthanized by decapitation prior to ovary dissection.

Table 1 Collection records by exposure group and generation.
Full size table

Endocrine disruptor dosage and administration

We evaluated the effect on sncRNA expression and apoptosis in embryonic ovary cells of two different known EDs: vinclozolin (3-(3,5-Dichlorophenyl)-5-ethenyl-5-methyl-2,4-oxazolidinedione) (CAS: 50471-44-8) and MEHP, mono (2-ethylhexyl) phthalate, (CAS: 4376-20-9), both from Sigma-Aldrich. Three different groups were established according to their exposure to the endocrine disruptors: (i) an unexposed or control group of mice, which were administered only the vehicle, DMSO, at a final solution of 0.3% (v/v); (ii) a VZN group, corresponding to mice exposed to 50 mg/kg bw/day VZN (using the same amount of DMSO as in the control group); (iii) the MEHP group, to which 35 mg/kg bw/day MEHP (and same amount of DMSO) was administered. These doses were selected to reflect environmental relevance and typical daily human exposures to endocrine disruptors, while also considering established thresholds such as the NOAEL (No Observed Adverse Effect Level) and LOAEL (Lowest Observed Adverse Effect Level). In the case of vinclozolin, according to the U.S. Environmental Protection Agency (EPA), the NOAEL for humans is established at 1.2 mg/kg/day56. The LOAEL is set at 50 mg/kg/day56 and that was the dosage used in this study based on previous studies57 and others in literature58,59. In the case of phthalates, the Agency for Toxic Substances and Disease Registry estimated that the daily exposure to DEHP (which is metabolized into MEHP) ranges from 3 to 30 µg/kg/day60. The European Food Safety Authority (EFSA) reports the NOAEL for DEHP as 5.8 mg/kg/day61. The most commonly reported NOAEL for MEHP is 140 mg/kg/day, based on reproductive studies by Lamb et al.62. However, more recent studies have reported effects at lower doses; for example, previous experiments in our laboratory identified transcriptomic alterations at 100 mg/kg/day, which was the medium dose tested63. Because our study aimed to assess effects across multiple generations, we selected a lower dose of 35 mg/kg/day to minimize severe reproductive outcomes, mimicking human exposure thereby allowing intergenerational analyses without blocking reproductive capacity in the treated first generation. All treatments were administered via drinking water. Estimated intakes were calculated from mean water consumption and body weight data obtained from previous experiments in our laboratory, as well as published literature14,16,64.

Experimental design

We tested the direct and indirect (multigenerational and transgenerational) effects of maternal exposure to the selected EDs (Fig. 1a). First, we set the breeding duos, which are considered generation 0 (F0 in the scheme of Fig. 1). Then, we monitored daily to detect the presence of the vaginal plug. After confirming the vaginal plugs, females were removed from the cages and randomly assigned to Control, VZN, or MEHP groups using GraphPad QuickCalcs random assignment tool. Animals were exposed during the first 13.5 days of gestation to one of the two EDs or control; specifically, upon confirmation of the vaginal plug, the regular water supply was modified with an ED or control. Twenty females from among these pregnant mice were euthanized at 13.5 dpc to obtain the embryonic gonads, and the rest of them were withdrawn from exposure to the ED and used to generate next generations. Pups from these mothers were used to generate F2 and F3. This process was repeated to study both the F1 generation (directly exposed, during their embryonic period, to EDs), the F2 generation (indirectly exposed to EDs) and generation F3 (not exposed to EDs). The experimental design allowed us to study the maternally transmitted transgenerational effect65.

Fig. 1
figure 1

Experimental design of endocrine disruptor exposure and sncRNA identification. Experimental design of ED exposure to (a) obtain multi- and transgenerational effect upon maternal exposure and (b) evaluate the potential additive effect of maternal and paternal exposure in the second-generation (Sy group) identification of sncRNA sequences in E13.5 ovaries. The number in brackets represents the number of pregnant females harvested to collect E13.5 embryonic ovaries for each exposed group. (c) Number and proportion of each sncRNA found in total pool of identified sequences. (d) Proportion of miRNAs (green), piRNAs (pink), and mi-piRNAs (blue) in each sample, Sample names are structured as follows: C = control; M = exposure to MEHP; V = exposure to vinclozolin; P = PGCs; S = somatic cells; 1, 2, 3 = successive generations; Sy = Synergic group.

Full size image

Additionally, to test the potential synergies of maternal and paternal effects, we also had an extra group, called the “synergic group” (Sy), following a similar protocol. Male offspring of the F1 generation (F1 adult in Fig. 1a), that had been exposed from conception to Embryonic Day 13.5, were crossbred to obtain the embryos of group F2.Sy (Fig. 1b), in which both parents had been exposed during the embryonic period to the EDs or control. The Sy group was used to evaluate the additive effect of in utero exposure by both mother and father.

Birth rate data collection

We systematically recorded the total number of embryos, their respective sex distribution, and the frequency of observed embryo resorptions at the time of embryo collection, which occurred at 13.5 dpc. These data were collected to assess the overall viability and developmental progression of the embryos, as well as to explore potential sex-related differences in embryonic survival and reabsorption rates (Excel Table S1).

Gonad extraction and PGC enrichment

After euthanasia of the pregnant females, the embryos were extracted by ventral dissection of the mothers and placed in M2 medium (MR-015, Sigma-Aldrich). Embryonic gonads were obtained by performing a ventral incision; dissection of gonads included the mesonephros, which was subsequently removed using tungsten needles. Gonads were placed in M2 medium. Sexual differentiation of the gonads at 13.5 dpc was carried out by morphological characterization of the gonads. Embryonic gonads were washed in PBS prior to further processing. The female gonads of the offspring from at least 4 mothers, collected on the same days and from the same exposure group, were pooled together (depending on litter sizes, we pooled an average of 54 (± 23.2) embryonic ovaries per batch) and further processed as a one unique biological sample (Excel Table S1). We obtained a total of 23 pooled samples from 87 different pregnant mothers for a control group that included 507 female embryos or 1014 gonads. From the exposed groups, MEHP and vinclozolin, we obtained 22 pooled samples per group from 96 and 103 different mothers, respectively, obtaining a total 615 (MEHP) and 678 (vinclozolin) female embryos, taking care, always, to note the embryos’ generation (see details at Table 1).

PGC enrichment was performed with paramagnetic sorting technologies, using the PGC-specific antibody SSEA1 (Miltenyi Biotec) and validating it by alkaline phosphatase reaction66. Briefly, we performed the desegregation of the sample of pooled embryonic ovaries, using Trypsin–EDTA (0.25% Gibco). Gonads were incubated in trypsin for 15 min at 36 °C, followed by PBS washes to obtain single-cell suspension. After that, the single-cell suspension was incubated for 45 min with Anti-SSEA1 antibody bound to paramagnetic microbeads (Cat. 130-094-530 from Miltenyi Biotec). Following the incubation, cells were sorted using a magnetic column (miniMAcs, Miltenyi Biotec). Two cellular populations were collected: a PGC fraction and a somatic cell fraction. To determine the level of PGC enrichment, samples of suspension cells from both fractions were spread on slides, fixed with paraformaldehyde (4% in PBS) for 5 min, processed with naphthol AS-MX/ FAST-RED (Sigma-Aldrich) to detect alkaline phosphatase PGC-specific reaction, and evaluated at the microscope. Using this technique, we discarded samples with lower than 90% of enrichment keeping samples with an PGC-enriched fraction with an average of 7254 ± 2042 PGCs per embryo, with 92–96% purity (representative image of fraction enrichment at Fig. S1), and 30,581 ± 9269, with 94–96% purity in the somatic cell fraction. Both cell fractions were stored in NYZol (NZYtech) at − 80 °C for under 6 months before RNA isolation.

RNA isolation and small RNA sequencing

Total RNA was extracted using Direct-zol DNA/RNA Miniprep (Zymo Research), following the manufacturer’s instructions. The quality and quantity of RNA was evaluated in the Genomic Unit at Universidad Complutense de Madrid, using a Bioanalyzer 2100 (Agilent Technologies). We sequenced 48 samples, all of them with RIN (RNA integrity number) above 8, as a unique batch to avoid batch effects. The RNA sequencing was carried out in Beijing Genomics Institute (Beijing, China), using a DNBSeq technology specifically designed for small RNA analysis. We sent 1.5 µg of total RNA to the facility, where RNA size fragmentation was performed by electrophoresis on acrylamide gels, retaining only fragments of approximately 100 nt for further library preparation. All samples were sequenced using a DNBSEQ-G50 machine, configured to get single-end 50-nt-long sequences (SE50) with around 10 M sequencing depth. Two replicates of each sample were sequenced.

Quality control and sequence filtering

The step directly following the sequencing of the samples was quality control, using the FastQC program (www.bioinformatics.babraham.ac.uk/projects/fastqc). From the raw data, adaptor sequences were removed, poor quality reads were filtered out (sequences below a Phred score of 32, which would be equivalent to a probability of error of 5.01 × 10–4), and poorly expressed sequences (under 10 CPM, counts per million) were removed, which in our samples, is equivalent to about 100 reads per defined sequence. To assess potential batch effects, a PCA including samples from each condition was performed, as shown in Figure S2.

To speed up the analysis and sequencing process, we used a software called FASTX-Toolkit (available at https://github.com/agordon/fastx_toolkit) after performing quality control. This software helped us combine identical sequences, removing duplicates while keeping track of the total number of copies found. This change does not decrease the information but significantly reduces the number of unique sequences that need to be aligned and identified. We use the term “sequences” when talking about different detected sequences (regardless of the number of copies), and “reads” when referring to the total number of copies for each unique sequence.

Identification of sncRNAs

We compared our small RNA sequences to publicly available databases of miRNA and piRNAs. Bowtie v 1.3.1 was used as a sequence aligner; we set the mismatch allowance at one mismatch between the query sequence and the database sequence. To identify miRNAs, we used the comprehensive database of miRNAs miRBase v22.167, and to identify piRNA sequences, we utilized piRNA-IPdb68, an improved version of the piRBase v2 database69, which only considers as piRNAs the sequences detected after immunoprecipitation with PIWI proteins. Since we conducted sncRNA identification in parallel, some sequences were identified as both miRNAs and piRNAs and are therefore referred to as mi-piRNAs. For analysis of mitochondrial sequences, those identified as miRNAs or piRNAs were compared and mapped against the mitochondrial DNA (accession number AY172335), using bowtie v1, with up to three mismatches (-p 7 -v 3 -k 3 – best –strata -y –chunkmbs 256 –sam –no-unal), considering the known high variability generated by the evolution of mitochondrial DNA sequences70.

Data analysis and visualization tools

Differential expression analysis was performed using the DESeq2 package in R software version 4.3.1. DESeq2 was used to normalize raw sequence counts and identify statistically significant differences in expression between experimental groups. Small RNAs with an FDR-adjusted p-value (padj) < 0.05 were considered significantly differentially expressed, in the case of piRNAs we have increased the astringency of the result, and we include a cutoff of Foldchange of 2 to be considered as a significant result. Given the large number of piRNA sequences and to increase the biological relevance of our results, we further filtered the dataset to include only those piRNAs with an absolute log₂ fold change greater than 1, corresponding to a minimum two-fold change between control and exposure groups. The different plots used to visualize the obtained data were created mainly by the ggplot2 package of R. Heatmaps were made with the ComplexHeatmap package of R. Control groups across the generations (F0-F3 and Sy) were evaluated; no statical differences were observed. To simplify the heatmap, all control groups are represented as one.

Validation of miRNAs by RT-qPCR

The same RNA samples were used for sequencing to perform miRNA-specific RT-qPCRs. Three different biological replicates of the first generation were used for miRNA expression validation, using more replicates when RNA was available, up to a maximum of nine. The complementary DNA was synthesized using the TaqMan MicroRNA Reverse Transcription kit (Life Technologies) with specific primers for each miRNA to be validated. CFX96 Real-Time PCR Detection System (Bio-Rad) was used to run the miRNA-specific RT-qPCRs, and the expression of each miRNA was evaluated using the 2-∆∆CT method71. Since there was not a gold-standard endogenous control to use, we picked the miRNA mmu-miR-425-5p because it had one of the lowest variabilities, across all the samples, of the RNA sequencing data collected and analyzed in this study. The TaqMan probes were ordered from Thermo Fisher Scientific and used to target each of the miRNAs: mmu-miR-451, hsa-miR-484, mmu-miR-871-3p, hsa-miR-107, hsa-let-7c, hsa-miR-425-5p; all of them were acquired from Thermo Fisher and were suitable to be used in mouse samples (detailed descriptions in Table S1). In all cases, we followed the manufacturer’s instructions, using as a starting point, an RNA amount of 6.3 ng/µL. The RT-qPCR data were analyzed using CFZ Maestro software, version 2.3 (Bio-Rad).

sncRNA target prediction

The miRanda software analyzes the base complementarity and thermodynamic stability of potential sequence pairs to estimate the likelihood of interaction between a piRNA sequence and its corresponding mRNA region. To focus on the most relevant targets, we considered the three most likely targets that surpassed a complementarity score of 170 and an energy threshold of − 20 kcal/mol. This approach helps reduce the number of potential false positives and enriches the selection of significant targets. Although originally intended for detecting miRNA-mRNA interactions, the miRanda software is widely utilized for analyzing interactions between piRNAs and mRNAs as well. This is because miRNAs and piRNAs could share functional similarities, such as their ability to bind to AGO or PIWI proteins72.

Immunofluorescence and apoptotic cell count

A fraction of 13.5-dpc ovaries were formalin fixed, and paraffin embedded (FFPE) for immunofluorescence staining. The immunofluorescence process was performed on five sections of three different FFPE ovaries of each group, following our previously described technique73. Briefly, slides were deparaffinized and rehydrated with three washes of SafeClear (Fisher Scientific) for 10 min each, followed by three washes of each concentration in a graded series of ethanol (100%, 95%, 80%, 70%) for 5 min each. After rinsing the slides twice for 5 min in distilled water, they were incubated in sodium citrate, pH 6.0, for 40 min at 95 °C. Permeabilization was performed in 0.2% Triton-X 100 in PBS for 1 h. Sections were blocked for 4 h in blocking solution (2.52 mg/ml glycine, 10% goat serum, 3% BSA, 0.2% Tween20 in PBS) and then incubated with primary antibody for SSEA1 (ab16285, Abcam) and cleaved PARP1 (ab32064, Abcam), diluted in blocking solution at 1:100 concentration overnight at room temperature (RT). After 2 washes of 10 min with 0.1% Tween20 in PBS (PBS-Tw20), the slides were incubated with secondary antibodies for 2 h at RT and for 15 min in DAPI (1.5 mg/mL). The slides were rinsed twice with PBS-Tw20 for 5 min and mounted in Vectashield (Vector Laboratories). Image acquisition was performed using an Axio Imager.M2 microscope (Carl Zeiss). Images were processed using Zen 2 (Carl Zeiss). The cells positive for each antibody were counted in each section, and the total surface area was noted, measured in mm2, to normalize.

SPORT software analysis

To compare the effect of sncRNA identification depending on the pipeline used, we also checked the number of piRNA and miRNA identified. We ran the software as similar as the original pipeline, allowing on mismatch and using the same miRbase and piRNA databases alongside the other databases proposed in the pipeline such as rRNA database, GtRNAdb database, Ensembl ncRNA database and Rfam database74.

Histological count statistical analysis

Data in the bar plots are shown as the mean ± SEM All experimental data were analyzed using unpaired t-tests when comparing two groups or one-way analysis of variance (ANOVA) to compare differences between more than two groups (GraphPad v.10.3.0 Software). FDR-adjusted p-values < 0.05 were statistically significant and marked with an asterisk (*).

Results

Effects of ED exposure on mouse reproduction

The intrinsic biology of the generation of PGCs makes it exceptionally susceptible to environmental exposures during embryo development since it will also imply exposure of the in utero germline and the PGCs’ development and differentiation. So, as described in the experimental design (Fig. 1a), to study the effect of ED exposure during embryo development, we administered VZN or MEHP to pregnant mice from pregnancy initiation. We analyzed the litter sizes, the proportion of female offspring, and the number of reabsorbed embryos at the moment of collection. From the total of 287 harvested mothers, 87 made up the control group, 96 MEHP-exposed, and 104 VZN-exposed (Table 1). We did not detect any reproductive alterations; the assessed reabsorption rates were not increased by ED exposition in any group (p = 0.25), with an overall average of 0.63 ± 0.8 per litter; and the litter sizes were similar in all exposure groups, with a non-significant higher rate in both exposed groups (p = 0.22), for an average of 13.6 ± 3.1 embryos per litter. We checked the sex ratio embryos and the detected reabsorbed embryos without finding any differences between exposed groups and controls (p = 0.39 for MEHP group and p = 0.69 for VZN). Detailed data can be found on Excel Table S1.

Identification and characterization of sncRNA sequences

The identification of piRNA and miRNA sequences from PGCs and somatic cells from fetal ovaries was performed using our previously published pipeline75, which consists of a quality control, removing low-quality reads and low expressed reads, and a second step consisting in aligning each sequenced read to two different databases, piRNA-IPdb for piRNAs and miRbase for miRNAs (as indicated in the Materials and Methods section). Each sequence matched with the database was labeled as miRNA or piRNA (or both). We detected a total of 12,078 piRNA and 3498 miRNA sequences. In agreement with previous publications, piRNAs and miRNAs were detected in both germ cells and in gonadal somatic cells75. However, miRNAs exhibited a clear preferential expression in somatic cells (p = 5.8 × 10–14) (Table 2). piRNA sequence analysis showed an average of 4007 different sequences per sample in PGCs and 3385 in somatic cells (p = 1.49 × 10–08). Analysis of miRNAs in female embryonic gonads showed that the number of miRNA reads also depends on cell type: around 60% of total detected reads in PGCs and 84% in ovarian somatic cells (SCs) (percentage and number of sequences for each sample in Table S2). We detected an average of 1542 miRNA sequences in PGCs and about 2074 in SCs, including all three groups. Exposure to EDs did not show a direct effect over the total number of sequences observed in any of the cell types for either miRNAs (p = 0.94) or piRNAs (p = 0.99).

Table 2 Summary of sequences and reads in each of the samples analyzed.
Full size table

Certain miRNA sequences have been described that also correspond to detectable piRNAs in their databases76. Consequently, those molecules identifiable as both piRNAs and miRNAs are referred to here as mi-piRNAs. To identify these sequences, we used the piRNA-IPdb database, built only with sequences detected after PIWI protein immunoprecipitation, and the miRBase database. In previous publications, we observed that 607 sequences were identical between miRNAs and piRNAs68, and this number increased to 40,000 sequences when we included the possibility of a single mismatch. Following the same approach, here we classified these sequences as miRNAs when they matched miRBase sequences and as piRNAs when they matched piRNA-IPdb sequences. We found a total of 12,189 different sequences (12,078 identified as piRNA and 3498 miRNAs) from which 27.8% (3,387) were considered mi-piRNA. A total of 8691 sequences were identified only as piRNAs (Fig. 1c), and just 111 sequences were identified only as miRNAs. In fact, when we analyzed the miRNA sequences between samples, we observed that the previously described higher number of miRNAs in somatic cells were in fact mi-piRNA sequences (Fig. 1d).

Endocrine disruptor exposure alters expression patterns of piRNAs

To determine whether there were significant differences in piRNA expression among the analyzed samples, we input the sncRNA expression levels of all 48 samples—including all exposed groups, three generations, in both cell types and in Sy group samples—into an unsupervised hierarchical clustering and plotted the results on a heatmap specific to piRNAs (Fig. 2a). As expected, PGCs and SCs showed different patterns of expression. This differential expression was further confirmed by DESeq2 software, which identified 3,677 piRNAs (30.44%) as differentially expressed (padj < 0.05) between PGCs and SCs, as shown in a volcano plot (Fig. 2b).

Fig. 2
figure 2

Analysis of global piRNA expression represented as: (a) heatmap of expression levels generated by unsupervised hierarchical clustering and (b) volcano plot showing differences between PGC and SC samples. The pie chart inside the volcano plot shows the percentage of upregulated and downregulated piRNAs.

Full size image

Using this approach, we observed that exposure to EDs at the analyzed doses did not quantitatively affect global piRNA expression levels but did alter the expression of specific sequences. An assessment of these changes was conducted using the DESeq2 R package, analyzing each ED and cell type individually to avoid confounding factors. The effects vary, depending on the endocrine disruptor and the cell type analyzed. The most pronounced deregulation occurred after VZN exposure in SCs, where 407 piRNA sequences were identified as deregulated; but in other cases, for example, SCs under MEHP exposure, just 34 sequences were deregulated (Table 3). The full list of differentially expressed piRNA sequences is provided in Supplementary Data (Excel Table S2).

Table 3 Summary of dysregulated sncRNAs in PGCs and somatic cells (SCs) for each ED and cell type, considering only the piRNA sequences with adjusted p-value < 0.05 and a fold change > 2.
Full size table

Considering the potential transgenerational effects reported in other studies16,77, we also evaluated the maintenance of piRNA expression patterns across successive generations. About 3% of piRNAs deregulated in the first generation kept an ED effect in the subsequent generation after ED exposure in the first generation, but only eight piRNA sequences carried the deregulated pattern through to F3 (Table 4). A comprehensive list of piRNAs with sustained differential expression in the second and third generations is provided (Excel Table S3). Notably, some sequences (piR-mmu-72438, piR-mmu-72623, piR-mmu-72444, and piR-mmu-5901931, all overlapping with miRNA miR-107-3p) displayed similar expression trends. In the case of VZN exposure in SCs, all piRNAs like miR-107-3p were consistently downregulated across all generations, indicating a transgenerational effect. Given the reported interaction between miR-107 and let-778, where miR-107 represses let-7, and the role of let-7 as a regulator of BLIMP116, the persistent deregulation of these piRNAs may represent a biomarker for the multigenerational impact of ED exposure.

Table 4 Differentially expressed piRNA sequences maintained in successive generations in each case.
Full size table

We also assessed the effect of ED exposure on animals with both parents exposed, not only the mother, on potential inheritance. For that purpose, as indicated in Fig. 1b, we crossed mice from F1 (males and females exposed during the embryo stage) to get group Sy. Comparing F2 generations with and without paternal exposure, we observed a 4.06-fold increase in deregulated piRNAs (from 668 to 2715 sequences) when both progenitors were exposed (Table 5). These findings highlight the aggravating effect of combined maternal and paternal exposure to EDs on piRNA deregulation in embryonic ovary cells.

Table 5 Comparison of observed alterations in piRNA and miRNA expression in F1 after direct exposure to EDs and paternal addition (Sy group) effect.
Full size table

Endocrine disruptor exposure alters expression patterns of miRNAs

The effect of ED exposure on miRNA expression was also tested, using the same RNA sequencing data. Samples from across three exposed groups, three generations, both cell types, and the Sy group samples were used to generate an unsupervised hierarchical clustering and plot the results on a heatmap specific to miRNAs (Fig. 3a). PGCs and SCs exhibited markedly distinct miRNA expression patterns (as had been the case with piRNAs). A total of 397 miRNAs (10.97% of the detected miRNAs) were identified as differentially expressed between PGCs and SCs, as shown in the volcano plot (Fig. 3b).

Fig. 3
figure 3

Analysis of global miRNA expression patterns represented as: (a) heatmap of expression levels generated by unsupervised hierarchical clustering and (b) volcano plot showing differences between PGC and SC samples. The pie chart inside volcano plot shows the percentage of upregulated and downregulated miRNAs. The 5 miRNAs (miR-451, miR-484, miR-871-3p, miR-107-3p, and let-7c-3p) validated by RT-qPCR are highlighted.

Full size image

Although the number of miRNA expression levels remained quantitatively unchanged under the analyzed ED doses (as previously mentioned), specific miRNA sequences exhibited detectable alterations. The DESeq2 R package was used to assess these changes, ensuring that each ED and cell type was analyzed individually to avoid confounding factors. The most deregulated scenario for miRNAs was observed in somatic cells exposed to MEHP or VZN, with up to 28 and 6 miRNA sequences altered, respectively (Table 3). The full list of differentially expressed miRNAs is provided in Supplementary Data (Excel Table S4).

We also assessed the transgenerational maintenance of miRNA expression patterns. As previously mentioned, miRNA sequences such as miR-107-3p were consistently downregulated across all generations in VZN-exposed SCs (Table 4). The interaction between miR-107 and let-7, combined with let-7's role as a BLIMP1 regulator, underscores the potential significance of these findings for PGC specification and broader developmental processes16.

Comparing F2 generations with and without paternal exposure, we observed a 5.5-fold increase in deregulated miRNAs (from 34 to 187 sequences) when both parents were exposed to EDs (Table 5). Concerning specific miRNAs altered in both comparisons (the 57 altered sequences with maternal-only inheritance and the 187 altered sequences of the Sy group), they only share nine of them (Excel Table S5), which suggests that the mechanism involved in the paternal transmission of dysregulation might be different from the maternal one. These results enhanced the hypothesis that the exposure of both progenitors to analyzed endocrine disruptors vinclozolin and MEHP aggravates the deregulation of expressed piRNAs and miRNAs in embryonic ovary cells.

Comparison between the effects of each endocrine disruptor

VZN and MEHP are both endocrine disruptors with different molecular pathways triggering differential responses in cells, tissues, and organisms. VZN primarily acts via its anti-androgenic properties20, interfering with androgen receptor signaling, while MEHP has an anti-estrogenic effect through disruption of estradiol biosynthesis23. Despite these mechanistic differences, both exposures may induce alterations in small non-coding RNAs, and transcriptomic deregulation associated with germ cell quality has already been described28. Building upon the characterization of the effects of VZN and MEHP on piRNAs and miRNAs, we compared the possible overlapping sequences present in several of those results. Our results define a strong tissue specific effect for both endocrine disruptors tested. Sequence overlap between conditions was minimal, with 616 piRNA sequences (95.5% of the total) exhibiting differential expression exclusively to a single experimental group. The miRNAs show a similar pattern, with 30 (93.7%) unique miRNAs per exposure group (Figure S3).

Identification of mitochondrial sncRNAs

In recent years, a key role in mitochondrial regulation via sncRNAs (both, piRNAs and miRNAs) has been reported79. Furthermore, a recent discovery has shed light on the interplay between the nuclear and mitochondrial genomes, which is facilitated by piRNAs80,81. In addition, previous experiments performed in our laboratory showed, in vitro and ex vivo, that in the first stages of ovarian development, VZN exposure induced a delayed meiotic differentiation and an altered mitochondrial metabolism57. In this study, we detected three sequences as potential mitochondrial miRNAs (miR-451a, miR-872-5p, and miR-484) and a total of 203 sequences of piRNAs with partial homology (up to three mismatches) with mitochondrial DNA, in which 59 of them have the exact mitochondrial DNA sequence (Fig. 4a). Most of the sequences showed similarity to mitochondrial DNA, specifically to the locus of mitochondrial tRNA (78 seq. 38.4%) or mitochondrial ribosomal transcripts (78 seq. 38.4%). The rest of the piRNAs matched with encoding mitochondrial genes (14 seq. 6.9%) or with the D-loop regulatory region (33 seq. 16.3%) as summarized in Fig. 4b. A detailed table of sequences detected in mitochondria is included in Excel Table S6. In 112 cases, the mitochondrial sncRNAs also mapped to the nuclear genome (which we called NUMT—nuclear mitochondrial DNA), and since the RNA was sourced from total RNA extraction, these sequences could originate also from nuclear DNA expression. Differential expression analysis showed that exposure to VZN or MEHP disrupts mitochondrial sncRNAs. The miRNAs miR-484 was found to have been altered by VZN in somatic cells.

Fig. 4
figure 4

Mitochondrial sncRNA detection. Overview of the mouse mitochondrial genome highlighting known features, with matched piRNA (green) and miRNA (orange) sequences detected (a), and the proportion of sncRNAs corresponding to each mitochondrial feature type (b).

Full size image

Validation of miRNA expression levels

In order to confirm the observations detected with small RNA-seq and differential expression, we assessed the expression levels by an alternative approach. We validated the results of the small RNA sequencing analysis by RT-qPCR, using TaqMan probes for six different miRNAs: mmu-miR-451, hsa-miR-484, mmu-miR-871-3p, hsa-miR-107, hsa-let-7c, hsa-miR-425-5p (all of them suitable to be used in mouse samples). As a housekeeping gene, we selected miR-425-5p as reference, since this miRNA showed a stable expression across all samples analyzed. We measured, by RT-qPCR, the relative expression levels in PGCs and SCs of samples directly exposed to EDs and controls. The effect of ED exposure on prenatal ovaries shows decreased levels, in PGCs, of miR-107, miR-871-3p, let-7c, and miR-484 (only in VZN-exposed cells), according to RT-qPCR results (Fig. 5a). The effect on somatic cells (Fig. 5b) also resulted in a decreasing level of miR-451 and miR-871-3p but increased in let-7c and miR-484. Consistent with the RNA-seq results, RT-qPCR detected similar changes in miRNA levels after ED exposure, with many cases showing even more pronounced distinctions. It is also noteworthy that, overall, PGCs exhibit fewer changes in miRNA expression compared to somatic cells. In certain cases, such as let-7c and miR-484, expression levels increased by approximately tenfold in exposed groups compared to controls, further emphasizing a cell type-specific effect in sncRNA regulation. However, miR-451a showed opposite results in both PGCs and SCs, suggesting that the TaqMan probe may be detecting a different isoform and not the one observed in the sequencing approach.

Fig. 5
figure 5

Expression levels of miRNAs confirmed by RT-qPCR using TaqMan probes: mmu-miR-451, hsa-miR-484, mmu-miR-871-3p, hsa-miR-107, hsa-let-7c, and hsa-miR-425-5p. All of them are suitable for use in mice, according to the manufacturer (Table S1). The relative expression of miRNAs was assessed using miR-425-5p (housekeeping gene) as basal expression. Relative expression of miRNAs in (a) primordial germ cells and (b) somatic cells. One asterisk (*) means p-value < 0.05, and two asterisks (**) means p-value < 0.01; three asterisks (***) means p-value < 0.001, and four asterisks (****) means p-value < 0.0001. All comparisons were made by one-way ANOVA with multiple comparisons against controls.

Full size image

Endocrine disruptors cause an increase in apoptotic cells

Given that dysregulated sncRNA can be involved in PGC development and apoptosis, we used immunofluorescence staining on ovary sections to evaluate the levels of apoptosis in E13.5 embryonic ovaries. Specifically, for PGC identification, we used an anti-SSEA1 antibody, considering that the experimental enrichment of PGCs for RNA-seq was performed using SSEA1 paramagnetic beads (see Materials and methods). To identify early apoptotic pathway activation, we used cleaved poly-ADP-ribose polymerase-1 (cPARP1) antibody. We analyzed 12 embryonic ovary sections from 3 different mice per exposure group, counting PGC number, all apoptotic cells, apoptotic PGCs, and the size of the ovary section to normalize. No significant changes in the global number of PGCs were detected (Fig. 6d). However, the apoptotic cells, including SCs and PGCs, showed a significant increase (double) in ED-exposed groups (p = 0.028 for VZN and p = 0.001 for MEHP). From an average of 35.4 ± 6 apoptotic cells per mm2 of ovary in the control group to averages of 70.8 ± 11 and 70.2 ± 5 in VZN and MEHP, respectively (Fig. 6e). Moreover, these apoptotic events were specifically observed in PGCs: from 18.3 ± 4 apoptotic cells per unit of ovary surface in controls to 35.8 ± 8 in VZN (p = 0.008) and 32.3 ± 3 in MEHP (p = 0.02), representing a fold increase of 2.44 and 1.75, respectively (Fig. 6f). Representative microscopy images are displayed in Fig. 6a–c. Consequently, our results show an increase in apoptotic levels in somatic and germ cells as a consequence of independent exposure to both of the EDs used (VZN or MEHP).

Fig. 6
figure 6

Exposure to vinclozolin and MEHP induce apoptosis in E13.5 ovaries. (a–c) Representative ovarian sections from E13.5 ovaries fluorescently labeled with SSEA1 (PGC surface marker) in green, cPARP1 (apoptosis marker) in red, and DAPI as counterstaining in blue. Apoptotic cells are indicated with arrows: orange for PGCs and white for non-PGCs. Bars = 20 μm. (d–f) Count of PGC and apoptotic cells normalized by ovary section surface in mm2. Quantification of PGCs (d); apoptotic cells, including PGCs and somatic cells (e); quantification of apoptotic PGCs (f). One asterisk (*) means p-value < 0.05, and two asterisks (**) means p-value < 0.01 in one-way ANOVA test.

Full size image

Discussion

The recognized regulatory roles of sncRNAs, mainly miRNAs and piRNAs, are in remarkable expansion82. In this study, we identified a total of 3498 miRNA and 12,078 piRNA (Table 2) sequences expressed in embryonic ovary primordial germ and somatic cells. Maternal exposure to endocrine disruptors during early pregnancy deregulated sncRNA expression of embryonic ovary cells. VZN altered the expression of 486 piRNAs and 15 miRNAs, and MEHP deregulated 182 piRNAs and 42 miRNAs, adding up to a total of 668 piRNAs and 57 miRNAs (Table 5). Co-exposure of both parents has an additive deregulatory effect on the female gonad of the offspring. Analyzing sncRNA targets, we were able to find possible routes for the observed sncRNA deregulation impact on apoptotic pathways. We also evaluated the apoptosis in both PGCs and somatic cells in 13.5-dpc ED-exposed and control ovaries. Our results show that the evaluated ED exposure increased apoptosis in these ovaries, particularly affecting PGCs.

Methodological considerations

The knowledge of the role of sncRNAs in cell differentiation is still in its relatively early stages, with a primary focus on piRNAs and miRNAs, as in this study. However, other sncRNAs are now being recognized as potential mediators of transgenerational effects. Notably, tRNA-derived small RNAs (tsRNAs) and rRNA-derived small RNAs (rsRNAs) found in sperm have been reported as able to influence the phenotypes of descendants by shaping early embryonic development17,83. These sncRNAs were largely elusive to sequencing technologies until the advent of advanced techniques, such as PANDORA-seq84 or TANT-seq85, which utilize enzymatic treatments to reveal previously hidden sequences. In our study, we focus on miRNAs and piRNAs low-modification RNAs that are reliably detectable using standard RNA-seq approaches comparable in sensitivity to these advanced techniques. While we cannot ignore the potential role of these newly identified sncRNAs in mediating epigenetic interactions, this was beyond the scope of the present work.

We noted that some sequences could be dually annotated as piRNAs and miRNAs. This duality could be considered to be the result of annotation errors; in fact some popular pipelines86,87,88 exclude miRNA-identified sequences before proceeding to the piRNA identification phase in order to avoid this problem, arguing that these sequences could be database artifacts76. The popular pipeline SPORTS (Shi et al., 2018) uses this same approach. The identification of sncRNAs could dramatically be distorted in this way. To example that situation, we run the same sequences with SPORTS1.1 software, the results (Excel Table S10) showed that the largest groups of identified sncRNAs were piRNAs (30.18%) and miRNAs (22.02%) others sncRNAs including rRNA, tsRNAs or unknown sequences accomplish around 15% each.

In our study, recognizing the well-documented ambiguity between piRNAs and miRNAs exist68, we conducted parallel identification and comparison of these small RNA species. To enhance the reliability of our piRNA annotation and avoid inclusion of poorly curated or misclassified sequences, we limited our analysis to piRNAs identified through co-immunoprecipitation with PIWI proteins. This was the rationale behind using the piRNA-IPdb database, which exclusively includes sequences obtained from RNA immunoprecipitation followed by high-throughput sequencing targeting PIWI proteins68. This approach reduces the likelihood of misclassifying other small RNAs as piRNAs. Notably, similar overlaps between miRNAs and piRNA-like sequences have been described. For example, Yang and colleagues demonstrated regulation by a piRNA resembling miR-17-3p89, highlighting the complexity of distinguishing functionally and structurally similar small RNAs. As the piRNA field continues to evolve, the development of standardized protocols and more rigorously curated databases will be essential for improving classification accuracy and functional interpretation.

Functional implications of sncRNA dysregulation

Recent studies have highlighted the effects of endocrine disruptors such as vinclozolin and phthalates on reproductive health and the popularization of omics techniques continues to increase the understanding of ED exposure90. Our lab has previously demonstrated that VZN exposure disrupts oocyte development by altering mitochondrial energetic metabolism and causing accumulation of synaptonemal complex components during meiosis, indicative of apoptosis and impaired germ cell maturation57. MEHP reduces estradiol synthesis and triggers apoptosis and ferroptosis in mammary tissues by inhibiting ERK1/2 phosphorylation, impairing granulosa cell function, and promoting oxidative stress91. Other multiomics approaches also reveal female-specific effects: MEHP exposure alters gene expression in placental cells92, microbiota diversity and metabolite profiles93, exacerbates allergic inflammation94, among other effects illustrating complex systemic disruptions.

The lack of a mechanistic understanding of piRNA regulatory functions makes it difficult to correlate most piRNAs with cellular functions. However, considering their role as RNA guides in combination with the RISC, we can approximate their function by in silico approaches like TARGETSCAN95 and miRanda software, along with miRDB (the microRNA target prediction database)96. With this approach, we evaluated the potential targets of miRNAs or piRNAs, especially the sequences detected as differentially expressed as a consequence of exposure to the EDs assessed. For example, we found the sequence “TGACTGGCACCATTCTGGATAAT” to be differentially expressed in the four comparatives (exposure to two EDs in both cell types, PGCs and SCs) when the log2Fold change criteria is not apply and, in all four comparisons, to be downregulated in the exposed groups Table 6. This piRNA was annotated as piR-mmu-24728 in the corresponding database, but the same sequence was also annotated as microRNA miR-871-3p. This sequence is part of the X-linked germ cell-specific miRNAs (XmiRs) that have direct interaction with a Wnt receptor (part of the WNT/β-catenin pathway) protein, Fzd497. This pathway has a critical role in female PGC development98, so its alteration could cause dysfunction in this process. Moreover, Fzd4 is not the only protein of the WNT/β-catenin pathway targeted by the deregulated piRNAs; we also encountered Wnt11, Wnt7a, and Wnt5b between detected targets, indicating a relevant impact of both EDs in this pathway. Based on these results, the sequence could be considered a potential biomarker of endocrine disruptor exposure after further appropriate validation.

Table 6 Summary of the statistical test of piRNA sequence “TGACTGGCACCATTCTGGATAAT” in all of the 4 comparisons.
Full size table

Other interesting possible targets of piRNA sequences are Sox genes. We detected up to eight different Sox genes: Sox5, Sox6, Sox8, Sox11, Sox13, Sox17, Sox18, and Sox30 as targets of deregulated piRNAs (such as: piR-mmu-50252965, piR-mmu-10844859, piR-mmu-5110545, piR-mmu-49610063). These genes are crucial in sex determination during embryo development. Knockout of Sox8 causes sexual reversion99. Related to the same pathway, we also detected altered expression of piR-mmu-72597 and piR-mmu-10835270 sequences targeting Xpo4 (exportin 4); an interactor of Sox proteins in mammalian embryonic cells100. Furthermore, although less pronounced, we also detected some piRNAs—piR-mmu-24728, piR-mmu-73689282, and piR-mmu-11457197—targeting genes involved in correct somatic cell function: Dnm3 (dynamin 3)101; ZP1 (zona pellucida glycoprotein 1)102; and Dicer1, a key protein in miRNA biosynthesis. The full descriptions of the deregulated piRNAs and their predictable targets are indicated in Excel Table S7.

The number of miRNAs with disrupted expression by the exposure to EDs was lower than we observed for piRNAs. However, the regulatory role of miRNAs on gene expression has been better studied. Due to their critical role during organogenesis, the disruption of a single miRNA may have a relevant negative impact on embryo development, specifically in the ovary, as in this study. Coinciding with targets detected for piRNAs, we detected the previously mentioned miR-871-3p as being capable of interfering in the WNT/β-catenin pathway. Several members of the let-7 miRNA family (let-7g, let-7f, let-7a) were found to be altered; this family has been shown to be a regulator of Blimp116.

The role of mitochondria in energy metabolism has been clearly established. In embryonic ovaries, this organelle plays a crucial role due to the energy consumption of proliferating PGCs and the growth of follicles during early embryo development103. Additionally, the role of piRNAs in mitochondrial linkage is not well known, but the mitochondrial membrane is not only the anchor point for piRNA biogenesis pathways, such as Zucchini/MitoPLD, but could also play a regulatory role104; piRNAs encoded by mitochondrial DNA have been described80. This organelle has also been found to be the target of endocrine disruptors105, including VZN57 and MEHP106. It is worth mentioning that, although our experimental design and sequencing approach were oriented toward piRNAs and miRNAs, we observed a notable correlation of sequence similarity with tsRNAs and rsRNAs. These RNA species have been increasingly recognized for their regulatory roles and have been implicated in disease-related molecular alterations, including in cancer and other pathologies107.

We also observed an increase in apoptotic cell numbers upon ED exposure on the first-generation embryonic ovaries, suggesting a connection between the alterations in sncRNA profiles and apoptosis induction. Similar effects have been observed in experiments with other endocrine disruptors, such as polychlorinated biphenyls108, bisphenol S109, and benzo(a)pyrenes110. Interestingly, among the reported mitochondrial sncRNAs detected as deregulated, we found miRNAs miR-484, along with several piRNAs with similar sequences that could interact in the same regulation pathways. As reported by Li et al.52, miR-484, in humans, has a direct impact on the Yes-associated protein isoform 1 (YAP1) gene, which is part of the Hippo pathway, involved in biological processes such as cell proliferation and apoptosis. Using overexpression/knockdown models and miR-484 inhibitors, they determined that miR-484 deregulation led to apoptosis, at least in granulosa cells. Considering the gentle impact detected in the transgenerational deregulation of these sncRNAs, potentially involved in apoptosis, the cytological assessment of apoptosis was only conducted in the generation directly exposed to the EDs. Nevertheless, assessing apoptosis in the Sy group to explore the effect of the exposure of both progenitors to ED would have been highly informative, as this group have a greater deregulation impact.

The alteration of sncRNA expression patterns observed following paternal exposure to endocrine disruptors has been widely reported16,17; however, reports analyzing the effects of maternal exposure are more sparse19. In our study, we show a mild transgenerational impact of EDs observed in a maternal exposure assay (Fig. 1a). This mild effect could be linked to the increase of apoptotic cells caused by these EDs. We hypothesize that the apoptotic process could work as a filter of “damaged” oocytes. These cells may undergo an apoptotic program similar to what happens during folliculogenesis in postnatal ovaries. This mechanism could mitigate deleterious transgenerational alterations in sncRNA profiles. The number of PGCs at E13.5 was maintained across the groups (Fig. 6d). However, the number of positive PGCs for apoptosis markers was increased in the exposed ovaries, suggesting that follicular ovarian reserves could be reduced in adulthood in the exposed groups. The remaining primordial germ cells continue to develop normally, preserving reproductive integrity but, importantly, with a potential reduction of the number of primordial follicles (ovarian reserve) during the fertile period of females exposed during the early embryonic period19,111,112,113,114. We could not conclusively attribute this effect to paternal exposure in our experimental design, leaving open the question of whether this alteration arises from a synergistic effect of both parental exposures or solely from maternal exposure.

We present the expression pattern of piRNAs and miRNAs in somatic cells and PGCs in mouse ovaries at 13.5 dpc and their alteration, produced by direct maternal exposure, during gestation, to VZN and MEHP. Globally, our findings suggest that direct exposure to some endocrine disruptors during pregnancy can lead to significant alterations in sncRNA expression in embryos. Moreover, we detected sncRNAs with conserved deregulation during successive unexposed generations. The levels of deregulation are further increased if both parents are exposed to such EDs. Our findings confirm the increase of apoptotic cells, specifically among PGC cells, occurring in early stages of ovarian development after exposure to endocrine disruptors vinclozolin and phthalate monoester and suggest the specific participation of deregulated sncRNAs, potentially associated with mechanisms of cell death by apoptosis, with the negative consequences that this could logically have on the oocyte reserve at reproductive age and, therefore, on female reproductive health.

Limitations of the study

The methodology used for RNA sequencing can significantly influence the detection of specific RNA species. In our study, we made two key decisions to enhance the detection of piRNAs and miRNAs. First, we applied a size-selection approach (as detailed in Materials and methods-RNA isolation and small RNA sequencing) combined with a sequencing strategy using 50-nucleotide single-end reads. As a result, larger RNA species in our samples were not captured, despite growing evidence that some of them—such as lncRNAs, snoRNAs, tsRNAs, and rsRNAs—are present in sperm and oocytes and may influence fetal development. Similarly, recently developed sequencing technologies, discussed in this work, have expanded the landscape of detectable sncRNAs. The traditional RNA-seq approach used in our study was intentionally designed to focus on miRNAs and piRNAs, thereby excluding other non-coding RNAs.

Another limitation of our study is the small amount of RNA that can be extracted from individual embryonic ovaries, which makes it difficult to perform direct small RNA sequencing on single ovaries or individual animals. The small RNA-seq protocol usually requires a minimum input of 1.5 µg of total RNA, a quantity that cannot be obtained from a single embryonic ovary at the developmental stages analyzed. To address this constraint, we pooled ovaries collected on the same day from animals within the same experimental group. This approach allowed us to meet the input requirements while minimizing biological variability and reducing potential individual-specific effects.