In recent years, over 350,000 chemicals and chemical mixtures have been registered for production, resulting in the continuous exposure of the general population to multiple contaminants, including airborne particles, industrial chemicals, heavy metals, and persistent organic pollutants (POPs) ¹. These toxicants can enter the body via inhalation, ingestion, or dermal absorption and, once present in the circulatory system, they may interact with cells and tissues, interfering with essential biological processes. Such effects can be of particular concern during pregnancy, when maternal exposures can impact foetal development.

In this context, the Developmental Origins of Health and Disease (DOHaD) framework, also known as the Barker hypothesis, postulates that insults that occur during sensitive developmental windows can alter disease susceptibility across the lifespan ². According to the “two-hit hypothesis”, early-life exposures may induce latent biological changes that manifest later in life when a second environmental or physiological stressor occurs. Indeed, there are epidemiological studies which support this model, showing associations between prenatal exposure to environmental toxicants and adverse birth and childhood outcomes, including impaired growth, neurodevelopmental deficits, and immune dysfunction ^{3, 4}. These findings highlight that foetal development is shaped not only by genetic determinants but also by modifiable environmental factors ⁵.

A growing body of research indicates that epigenetic mechanisms mediate the long-term health effects of prenatal exposures. DNA methylation, histone modification, and gene silencing by microRNAs (miRNAs) are key processes linking environmental insults to altered gene regulation ^{6, 7, 8, 9}. miRNAs, small non-coding RNAs that regulate gene expression and cellular homeostasis, are highly responsive to environmental stimuli ¹⁰. Their stability in biofluids and detectability in blood make them promising, non-invasive early biomarkers of exposure and biological impact ¹¹. Several studies have shown that maternal exposure to chemical mixtures is associated with changes in circulating miRNA expression, with potential consequences for offspring development and later disease risk ^{12, 13, 14}. Since many of these molecular alterations likely originate at the maternal–foetal interface, the placenta emerges as a key organ of interest for the way it mediates this interface by ensuring nutrient and gas exchange and acting as a barrier to xenobiotics. However, evidence indicates that this barrier is not impermeable: POPs, polycyclic aromatic hydrocarbons (PAHs), and even particulate matter can cross into the foetal compartment ^{15, 16}. Such transplacental transfer raises concerns about direct foetal exposure to complex contaminant mixtures during critical windows of intrauterine development.

Despite increasing evidence linking pollutant exposure, miRNA dysregulation, and health outcomes, very few studies have simultaneously investigated both maternal and cord blood to assess maternal-foetal transfer and intergenerational effects. To address this knowledge gap, birth cohorts enrolled in highly contaminated areas are a powerful tool for investigation. The Neonatal Environment and Health Outcomes (NEHO) cohort, established in southern Italy, provides such an opportunity. Conducted in the industrialized Augusta-Priolo area, recognized as a National Priority Contaminated Site (NPCS), NEHO was specifically designed to evaluate how environmental pollutants can affect the health of mothers and their children ^{17, 18, 19}.

For this study, we investigated the relationships between maternal exposure to environmental pollutants and circulating miRNA expression in maternal serum during the third trimester of pregnancy and in cord serum at birth. We i) assessed the concentration of selected inorganic and organic contaminants and their placental transfer, ii) examined their associations with miRNA expression in both maternal and foetal sides, and iii) explored the biological pathways potentially linking these molecular alterations to health outcomes. By integrating biomonitoring data, the evaluation of biomarkers, and questionnaire data, this work aims to advance our understanding of maternal-foetal health dynamics in polluted environments, as well as to provide insights which are relevant to public health interventions.

NEHO participants exhibited widespread exposure to a variety of environmental pollutants. Seven POPs — including HCB, p,p′-DDE, PCB-118, PCB-138, PCB-153, and PCB-180 — and five trace elements, comprising three EEs (Cu, Zn, Se) and two non-essential elements known for their toxicity (Hg, As), were quantified in maternal and cord serum samples as described above. The concentrations of analytes in both maternal and cord serum are presented in Table S1, along with the percentage of values below the LOQ in the 95 samples analysed. Compounds with LOQ frequencies exceeding 15% in either matrix were excluded from further analysis.

Table 2 lists the analytes included in the subsequent analyses. All selected chemicals showed statistically significant differences between the two matrix concentrations (p < 0.001). The placental transfer ratio provides insight into placental permeability and potential foetal exposure to environmental contaminants. A higher ratio indicates greater transfer efficiency from mother to foetus. Among EEs, the placental transfer ranged from 0.13 for Cu to 0.76 for Se, with Zn being the only element exhibiting a median ratio above 1 (1.41 [1.17;1.59]). In contrast, OC compounds displayed consistently low placental transfer ratios, with values ranging from 0.21 to 0.52, indicating limited permeability. Similarly, PCBs showed low transplacental transfer, with values ranging from 0.12 to 0.24 across individual congeners, reflecting their restricted transplacental movement.

To further investigate the distribution of analytes at the maternal–foetal interface, we examined the statistically significant correlations among analytes in maternal serum (Fig. 1A), in cord serum (Fig. 1B), and between the two matrices for each analyte (Fig. 1C). Strong intra-group correlations were observed among substances with similar chemical characteristics in maternal serum (Fig. 1A). EEs were weakly correlated with each other. Zn and Se did not show any significant inter-element correlations. Among OCs, HCB and p,p’-DDE were highly correlated (r = 0.79). HCB also showed significant correlations with all PCB congeners; however, the strength of association showed an evident reduction in the Pearson correlation index with increasing chlorine substitution. Indeed, the correlation was strongest with PCB-118 (r = 0.68, five chlorine atoms) and lowest with PCB-180 (r = 0.33, seven chlorine atoms). A similar trend was observed for p,p’-DDE, but only for the less chlorinated PCBs (PCB-74, PCB-118, PCB-138, and PCB-153). Finally, high inter-correlations were also found among PCB congeners, especially those with similar chlorination patterns.

In cord serum (Fig. 1B), a strong correlation between Se and Cu (r = 0.66) was observed. Both elements also exhibited weaker correlations with compounds having different chemical characteristics. Specifically, Se showed a slight positive correlation with PCB-138 (r = 0.22) and a negative correlation with HCB (r = -0.20), whereas Cu was positively correlated with PCB-138 (r = 0.36) and PCB-153 (r = 0.27). OCs were not correlated with each other in cord serum but showed weak correlations with PCB congeners. As shown for the maternal side, the PCB congeners in cord serum remained strongly inter-correlated. Finally, several analytes showed moderate correlations between both the maternal and cord compartments (Fig. 1C), reinforcing the idea that, although placental transfer is generally limited, foetal exposure is more likely to be influenced by maternal levels. Although the placental transfer values of OCs and PCBs were very low, p,p’-DDE exhibited the highest correlated index between maternal and cord serum (r = 0.96), followed by various PCBs. Among the EEs, although Zn was the only analyte showing placental transfer above 1, Se was the only one to show a slight maternal-cord correlation (r = 0.22).

A total of 84 miRNAs were analysed across maternal and cord serum samples. Of these, 75 from maternal serum and 79 from cord serum met quality thresholds (Ct < 40) in more than 70% of samples. The full list of miRNAs, along with their ΔCt values and the percentage of imputed values in each compartment, is reported in Table S2. Five miRNAs were excluded from further analysis in both maternal and cord serum due to having more than 30% imputed values. Additionally, four miRNAs were removed from maternal serum because their threshold Ct values were below 40 in fewer than 70% of subjects. We compared the median ΔCt values between maternal and cord serum to identify statistically significant differences in their expression levels. 31 miRNAs showed significantly lower ΔCt in maternal serum, indicating higher relative expression compared to the cord group. On the other hand, 35 miRNAs showed a higher median ΔCt value in maternal serum, reflecting greater expression in cord serum. The remaining 18 did not show significant differences between compartments.

These results suggest that specific circulating miRNAs are differentially expressed in maternal and cord blood, potentially reflecting foetal-specific gene regulation, maternal-foetal exchange, and exposure-related molecular signaling.

Associations between individual maternal miRNAs and analytes, correcting for covariates indicated in section 2.6 of “Materials and Methods”, are reported in Fig. 2. Only statistically significant associations after FDR correction (FDR-adjusted p < 0.05) are listed.

Maternal concentrations of Cu and Se were positively associated with a higher ΔCt of miR-7-5p and miR-214-3p, respectively. Both miRNAs also showed statistically significant associations with maternal Zn levels. In addition, increasing Zn concentrations were positively associated with the higher ΔCt values of five additional miRNAs. In contrast, miR-30d-5p and miR-374a-5p were inversely associated with Zn concentrations in maternal serum. Regarding POPs, maternal p,p’-DDE and HCB concentrations were associated with increased ΔCt values of miR-1-3p and miR-133a-3p. Although PCB-74 and PCB-118 are grouped in a different chemical class, we identified a similar trend with respect to the OCs. This included positive associations with miR-1-3p and miR-133p. Furthermore, PCB-118 was also associated with miR-195-5p. Higher concentrations of PCB-138 in maternal serum were positively associated with miR-1-3p, miR-195-5p, and miR-30e-5p. The most chlorinated PCBs (PCB-153 and PCB-180) were primarily linked to the latter two of these miRNAs. These patterns were consistent with results obtained using the sum of maternal PCB concentrations: high concentrations of PCBs were positively associated with miR-195-5p and miR-30e-5p. Detailed information about beta estimates are reported in Table S3.

We investigated associations between miRNA ΔCt values and analyte concentrations in cord serum, correcting for covariates indicated in section 2.6 of “Materials and Methods”. Figure 3 reports the results, focusing on statistically significant associations after FDR correction (FDR-adjusted p < 0.05).

Increasing concentrations of Cu and Se in cord serum were positively associated with increasing ΔCt values of miR-30d-5p and miR-7-5p. Additionally, cord Cu concentrations were positively associated with seven miRNAs: miR-195-5p, miR-122-5p, miR-134-5p, miR-200c-3p, miR-574-3p, miR-10a-5p, miR-376-3p. Higher Se concentrations in cord serum were associated with lower ΔCt values of miR-29a-3p and miR-22-3p. Moreover, miR-25-3p and miR-16-5p were inversely associated with Zn concentrations.

Across the POPs, HCB and PCB congeners were associated with distinct miRNAs. Specifically, cord HCB concentrations were associated with increased ΔCt values of let-7c-5p, miR-374a-5p, and miR-191a-5p. Two miRNAs showed consistent associations with all measured PCB congeners in cord serum (PCB-138, PCB-153, and PCB-180) and confirmed their behaviour when analysing the sum of PCB congeners: miR-25-3p was positively associated with each congener, whereas miR-29a-3p was negatively associated with the same congeners. In addition, higher levels of cord PCB-138 and PCB-153 were positively associated with increased ΔCt values of miR-30d-5p and miR-7-5p, they were also negatively associated with miR-133a-3p and miR-133b. These results were confirmed by analyses using the sum of PCBs, while showing no statistically significant results with PCB-180. The β coefficients and confidence intervals for each significant association are reported in Table S4.

Following the same approach, we applied bWQS regression to examine the impact of three distinct exposure groups: EEs, POPs, and a combined total mixture of both EEs and POPs. Statistically significant associations for each mixture are presented in Fig. 5.

Eight out of eleven associations were confirmed through the bWQS analysis performed on the EE mixture in cord serum (Fig. 5 panel A and B). Specifically, the four positively associated miRNAs previously reported for Cu and the two negative associations previously reported for Se were confirmed in the mixture analysis. No associations in common with Zn were observed. In addition, two associations, previously observed in the single-chemical models for both Se and Cu, were confirmed by bWQS analysis (Fig. 5 panel A and B).

Eight cord serum miRNAs were newly identified as being associated with the EE mixture, revealing five positive and three negative associations.

Analysing the POP mixture in cord serum, four associations previously identified in the single-chemical regression were confirmed in the mixture model (Fig. 5 panels and C). Of these, three miRNAs were also negatively associated with at least one PCBs, while the remaining miRNA was positively associated with HCB. In addition, the bWQS analysis of the POP mixture revealed nine novel associations which had not previously been identified in single-chemical regression, with a general negative trend.

The bWQS analysis of the total mixture in cord serum identified fifteen associations (Fig. 5 panel A and D). Among these, one negative association was commonly identified analysing both the EE and POP mixtures. Eight associations overlapped with the EE mixture results, of which six were positive. In addition, two new associations emerged: a positive association with miR-200c-3p and negative association for miR-19b-3p were observed.

The β coefficients and confidence intervals for each significant association are reported in Table S6.

To further explore the drivers of the observed miRNA alterations, we examined the contribution of the individual components within pollutant mixtures to overall miRNA expression patterns on both the maternal and foetal side. The bWQS chemical weight thresholds are listed in Table S7.

As shown in Fig. 6 (left side, EE section), Zn exceeded the contribution threshold (t_EEm=0.33) in 94% of the significant associations, followed by Cu (22%), and Se (17%) in maternal serum.

In the POP section, PCB-118 emerged as the most frequent contributor (61% of the significant associations), followed by HCB and PCB-74 (both 54%). PCB-138 and PCB-180 exceeded the weight threshold in 31% of the associations, while p,p’-DDE and PCB-153 exceeded the 0.14 threshold (t_POPm=0.14) in 23% of the associations.

Overall, in maternal serum, as highlighted in Fig. 6 (left side, TOTAL section), Se and Zn emerged as the predominant elements across the total mixture, exceeding the weight threshold in 81% and 72% of the significant associations, respectively. In contrast, Cu overcame the threshold in 36% of associations. Among the POPs, PCB-118 and HCB were consistently above the threshold (t_TOTm=0.10) in 54% and 45% of the associations, respectively. PCB-74 and PCB-138 showed weights above the threshold in 27% of the cases, while p,p’-DDE and PCB-153 exceeded the 0.10 threshold in 18% of the associations. PCB-180 was the least influential as 10% of associations reported PCB-180 as having weight above the threshold.

Regarding the cord serum, among the elements included in the EE mixture, Cu most frequently emerged as a key driver of the observed associations (62%), while Se and Zn showed a lower contribution frequency (both 37%) (Fig. 6 right side, EE section).

For the POP section, as illustrated in Fig. 6 (right side), PCB-138 emerged as the most recurrent contributor across the significant associations, exceeding the relevant threshold (t_POPc=0.20) in 85% of the cases. PCB-153 and HCB also demonstrated substantial contributions, with weights over the threshold in 61% and 54% of the significant associations, respectively. In contrast, PCB-180 exceeded the threshold in only 23% of the associations, while p,p’-DDE was the least frequent (15% of the associations).

Overall, in cord serum, Se emerged as the most frequent component above the threshold, in 53% of significant associations (Fig. 6 right side, TOTAL section), followed by Cu and PCB-153, each contributing in 47% of the cases. HCB was involved in 40% of associations, while Zn and p,p’-DDE exceeded the threshold (t_TOTc=0.125) in 33% of associations. Finally, PCB-138 and PCB-180 were the least represented, contributing above the threshold in 27% and only 0.7% of associations, respectively.

This study provides a comprehensive understanding of the epigenetic consequences of prenatal exposure to essential elements and toxic environmental contaminants in a birth cohort enrolled in a highly industrialized area, integrating pollutant and miRNA expression profiling. Our work aimed to investigate i) placental function in pollutant transfer, ii) differential circulating miRNA expression between maternal and foetal compartments, and iii) the influence of chemical mixtures on miRNA expression on both sides.

By simultaneously analysing EEs and POPs in matched maternal and cord serum samples of the NEHO birth cohort, we provide evidence that the placental maternal–foetal interface is not only a site of selective chemical permeability but also a determinant of compartment-specific epigenetic regulation. Placental transfer behaviour varied considerably across compounds. Among the EEs analysed, Zn showed a median transfer ratio greater than 1, consistent with active and regulated transport mechanisms supporting foetal development ^{20, 21}. In contrast, Se and Cu displayed moderate to low transfer efficiencies, while lipophilic POPs such as OCs and PCBs exhibited limited placental passage, with transfer ratios well below 0.5.

These findings align with prior reports which observed that the physicochemical properties of POPs, particularly their hydrophobicity and protein-binding affinity, limit their mobility across the placenta ²². Strong intra-group correlations were found among chemicals with similar structures, while cross-group correlations were generally weak; notable exceptions included Se with Cu and PCB-156 in maternal serum, and Se with Cu in cord serum. Indeed, HCB and p,p’-DDE were highly correlated, as were PCB congeners, with correlation strength decreasing as chlorine substitution increased. These correlations highlight the placenta’s role as a selective barrier that modulates, rather than simplistically prevents, pollutant transfer to the foetus.

Furthermore, the literature data indicates that stressors such as environmental pollutants can induce compartment-specific molecular alterations, including differential miRNA expression, reflecting maternal adaptation and foetal vulnerability ²³.

These data support our initial approach in which we analysed the chemical-specific transfer at the placenta maternal–foetal interface shaping differences in miRNA regulation, with profiles serving as a readout of compartment-specific biology during critical developmental windows. For these reasons, we evaluated the association between environmental pollutants and circulating miRNAs in the maternal and cord blood sides of the placenta in the NEHO birth cohort samples. In maternal serum, we observed that higher concentrations of Cu, Se, and Zn were significantly associated with changes in several miRNAs ²⁴. Specifically, Cu and Se concentrations were positively associated with miR-7-5p and miR-214-3p, respectively. In this context, miR-214-3p is known to be as a putative regulator of GPx4, a Se-dependent selenoprotein, whereas miR-7 is recognized as a key player in cancer biology and in the modulation of drug resistance. By contrast, Zn exhibited both positive and negative associations across different miRNAs. Notably, miR-30d-5p and miR-374a-5p were inversely associated with maternal Zn levels, suggesting a possible specific regulatory mechanism of Zn on miRNA expression participating in maternal–foetal communication and pregnancy complications ¹³.

In cord serum, Cu levels were positively associated with a broader set of miRNAs, including miR-7-5p, miR-10a-5p, miR-30d-5p, miR-134-5p, miR-122-5p, miR-195-5p, miR-200c-3p, miR-376c-3p, and miR-574-3p. Conversely, Se showed opposite trends depending on the selected miRNAs, with inverse associations observed for miR-22-3p and miR-29a-3p showing that many cord plasma miRNAs respond to maternal and neonatal metabolic traits, reinforcing the idea that cord miRNA profiles are broadly exposure-sensitive in pregnancy ²⁵.

Exposure to OCs, including p,p’-DDE, HCB, and various PCB congeners, was also associated with distinct miRNA expression patterns. In maternal serum, p,p’-DDE, HCB, and lower-chlorinated PCBs (e.g., PCB-74, PCB-118) were positively associated with miR-1-3p and miR-133a-3p, which are known to be involved in muscle development and cardiac function (myomiRs) ²⁶. Interestingly, PCB-118 and PCB-138 showed broader effects, being associated with multiple miRNAs. The most chlorinated PCBs (PCB-153 and PCB-180) were primarily linked to miR-195-5p and miR-30e-5p, associations that were also confirmed using the sum of PCB concentrations. Increasing data demonstrates that both miR-195-5p and miR-30e-5p are key regulatory miRNAs in pregnancy. MiR-195-5p supports trophoblast proliferation, migration, and invasion and interfaces with angiogenic and activin/Nodal pathways. Its circulating levels also track the anti-angiogenic imbalance characteristic of preeclampsia ²⁷. MiR-30e-5p has been linked to vascular and immune/inflammatory adaptations since it associates with preeclampsia trajectories and microvascular dysfunction and varies with late-pregnancy blood-pressure phenotypes ²⁸.

In cord serum, HCB modulated a specific subset of miRNAs with increased ΔCt values, let-7c-5p, miR-191a-5, and miR-374a-5p. Importantly, miR-25-3p and miR-29a-3p ²⁹ showed consistent associations with all measured PCB congeners, though miR-25-3p was positively associated whereas miR-29a-3p was inversely associated, suggesting divergent regulatory responses to these chemicals. Data from the literature shows that miR-25-3p regulates myometrial contractility by being exported in trophoblast extracellular vesicles, and its downregulation enhances Ca²⁺ signaling, suggesting a role in both uterine activity and foetal development ³⁰. Furthermore miR-29a-3p, enriched in villus-derived exosomes, promotes immune tolerance by suppressing decidual NK-cell IFN-γ production; its reduction has been reported in recurrent pregnancy loss ³¹. Finally, miR-7-5p ³²and miR-30d-5p ³³ were positively associated with PCB-138 and PCB-153, whereas miR-133a-3p and miR-133b-3p were negatively associated with the same congeners. These trends suggest a potential disruption of miRNAs involved in muscle and cardiovascular development and reinforce the sensitivity of the foetal miRNome to environmental exposures.

As a first approach, these single-pollutant models were very informative for identifying specific associations between exposures (chemicals) and biological responses (miRNAs), also showing qualitative and quantitative responses in both the maternal and foetal compartments. However, their use would oversimplify the real-world scenario in which individuals are simultaneously exposed to a complex mixture of contaminants.

Therefore, moving toward multipollutant approaches provides a more realistic framework for analysing the effects of mixtures, highlighting potential key drivers, and can better capture the overall impact of environmental exposures on maternal and foetal health, particularly in highly industrialized areas characterized by complex pollutant profiles. By extending the analysis to include mixture models through bWQS regression, we identified several additional miRNAs associated with combined exposures to EEs and POPs, many of which were not detectable in single-chemical models.

Our multipollutant analysis revealed both consistent and divergent associations between environmental mixtures and circulating miRNAs in maternal and cord serum, suggesting shared as well as compartment-specific regulatory responses. For instance, miR-7-5p and miR-122-5p were positively associated with essential element mixtures in both maternal and cord samples, indicating stable modulation by micronutrient-related exposures and supporting their role as common biomarkers of exposure across compartments. In contrast, several miRNAs demonstrated opposite directions of association between mother and child, including miR-133a-3p, miR-133b, miR-200a-3p, miR-205-5p, and miR-214-3p, which may reflect differential maternal–foetal adaptive responses or compartment-specific regulation of gene expression in response to environmental stressors (consistent with the known roles of these miRNAs in trophoblast invasion, Epithelial–mesenchymal transition, endothelial signaling, and placental dysfunction) ³⁴.

Such divergent associations suggest that maternal and foetal systems develop distinct molecular programs to cope with environmental stress, reflecting compartment-specific priorities in maintaining pregnancy homeostasis. Notably, miR-195-5p showed consistent positive associations with both POP and EE mixtures in maternal and cord serum, reinforcing its central role in placental and vascular biology ³⁵. By contrast, miR-30d-5p, miR-134-5p, and miR-376c-3p shifted direction depending on the exposure mixture considered, underlining the complexity of multipollutant effects and the necessity of integrated analytical approaches to capture them. Together, these findings suggest that, while some miRNAs act as robust biomarkers of shared maternal–foetal exposure, others capture regulatory programmes, underscoring the value of mixture-aware models in elucidating the molecular impact of environmental exposures during pregnancy.

Considering the placental transfer dynamics, our results illustrate how complex chemical exposures shape compartment-specific miRNA responses. Overall, the bWQS regression demonstrated that, in the maternal serum, Zn emerged as the most frequent contributor to mixture-associated miRNA expression, whereas Cu dominated on the foetal side, emphasizing not only the chemical-specific but also the compartment-specific drivers of epigenetic regulation. Among POPs, PCB-118 and HCB were the most influential components in maternal serum, whereas PCB-138 and HCB were predominant in the cord compartment. These patterns suggest that the most biologically active or bioavailable pollutant components may differ across maternal and foetal systems, probably due to differences in toxicokinetics, metabolism, or placental transfer efficiency. Indeed, few miRNA–pollutant associations overlapped across compartments, despite strong correlations in pollutant levels. This divergence points to independent molecular responses shaped by tissue-specific regulation, chemical concentration, developmental context, and possibly differences in cell-type composition of circulating shuttles such as extracellular vesicles.

To further derive the biological significance of these epigenetic signals, we performed pathway enrichment analysis of the predicted miRNA target genes. In order to analyse the structures from these multi-dimensional signals in detail, miRNAs were grouped by exposure class [EEs, POPs, or total mixture (TOT)], direction of association in bWQS (positive vs negative), and biological origin (maternal vs cord). The general structure of the enriched pathways was organized into general hubs. Only pathways supported by at least four miRNAs are displayed, increasing confidence that the observed communities are not driven by single markers.

The analysis indicates that miRNAs associated with pollutant mixtures converge on biological processes of central relevance to pregnancy. Among the most prominent are apoptosis, cell cycle regulation, and cellular stress responses, all of which are essential for placental development and trophoblast turnover ³⁶. Enrichment of immune-related pathways, including cytokine signaling, innate immune responses, and the TLR4 cascade, is particularly noteworthy as these processes underpin the delicate balance between maternal immune tolerance and controlled inflammation at the maternal–foetal interface ³⁷. In addition, pathways such as NOTCH signaling, Rho GTPase signaling, and adherens junction interactions suggest potential effects on trophoblast migration, vascular remodelling, and intercellular communication—mechanisms frequently disrupted in pregnancy complications ¹³. The enrichment of metabolic pathways (e.g., fatty acid metabolism, glycosylation, nuclear receptor signaling) further points to a possible impact of pollutant exposures on maternal–foetal metabolic homeostasis ³⁸. Although maternal and cord compartments showed overlapping pathway signatures, notable differences were observed: maternal profiles were more strongly enriched in stress- and apoptosis-related pathways, whereas cord profiles were characterized by immune and developmental signaling, suggesting compartment-specific adaptive responses. Collectively, these findings support the view that environmental mixtures influence pregnancy through interconnected networks regulating placental growth, immune balance, and metabolic function, with potential implications for adverse outcomes. These pathway structures demonstrate that mixture composition and biological compartment jointly determine miRNA-mediated regulation.

The maternal programme emphasizes growth factor signaling, nuclear receptor integration, and immuno-metabolic coupling, whereas the foetal programme highlights neurotrophic, innate immune, and cytoskeletal/adhesion axes.

Because miRNAs typically repress their targets, apparent increases or decreases in pathway enrichment must be interpreted in light of target directionality and network feedback, although validation at the mRNA/protein level is warranted.

The strength of this study lies in its integrative design, combining high-resolution chemical analysis with miRNA profiling in matched maternal and foetal serum, and applying both single-pollutant and advanced mixture modelling approaches. To the best of our knowledge, this combined approach represents a significant and original improvement in this field. However, some limitations must be acknowledged. The sample size, while sufficient for detecting robust associations, may have limited our ability to explore more specific interactions, and the cross-sectional design does not permit inference of causality. In addition, while circulating miRNAs are promising biomarkers, they may not fully capture the tissue-specific regulatory changes occurring at the maternal–foetal interface or within target organs. Nonetheless, the consistency of key findings across models and the alignment with established biological pathways support their interpretive validity.

In summary, our data demonstrate that environmental exposures during pregnancy induce distinct and compartment-specific miRNA responses in maternal and foetal circulation, shaped by individual chemical properties, complex mixture effects, and the placenta’s filtering role. The identification of exposure-sensitive miRNAs involved in developmental, oxidative stress, and immune pathways provides mechanistic insight into how early-life exposures may contribute to long-term health trajectories. These results emphasize the importance of considering both the maternal and foetal compartments, accounting for chemical mixtures and integrating molecular biomarkers to better understand the developmental origins of health and disease. Furthermore, our findings also highlight the pivotal role of the placenta as a selective barrier and an active regulator in mediating foetal exposure and shaping molecular responses to environmental stressors. However, whether the placenta actively sends distinct regulatory signals to the maternal and foetal sides or rather acts as a selective barrier that separates or integrates molecular cues, remains to be clarified and warrants further investigation.

Altogether, these results advance our understanding of how the interplay between environmental factors and placental function contributes to the regulation of gene expression in early development, setting the stage for a detailed discussion of the observed pollutant-specific transfer dynamics and miRNA expression profiles.