Maternal emulsifier consumption alters the offspring early-life microbiota and goblet cell function leading to long-lasting diseases susceptibility

Abstract

Early-life acquisition of microbiota and, consequently, immune system development, both lastingly impacts health. Accordingly, we hypothesized that disturbing the microbiota of lactating mothers via consumption of dietary emulsifiers might alter the microbiota, and perhaps the immune system, of their offspring, thereby increasing susceptibility to microbiota-mediated diseases, including colitis and metabolic syndrome. Here we report that, in mice, maternal consumption of carboxymethylcellulose and polysorbate-80 resulted in transient alterations in offspring microbiotas that were necessary and sufficient to increase proneness to colitis and metabolic syndrome in young adulthood. Offspring microbiome alterations induced by maternal emulsifier consumption resulted in elevated levels of pro-inflammatory flagellin, bacterial encroachment, and premature closure of goblet cell associated antigens passages (GAPs). The latter event was linked to phenotypic outcome in that pharmacologically preventing GAP closure eliminated the detrimental of maternal emulsifier consumption. Collectively, these results illustrate the potential of dietary emulsifiers to drive transgenerational microbiota alteration and, consequently, hastened immune development that increases susceptibility to inflammatory diseases.

Here, Delaroque et al. show that maternal intake of dietary emulsifiers alters offspring microbiota early in life, which, associated with perturbation of goblet cell function, promotes increased susceptibility to inflammatory diseases in adulthood.

Introduction

Birth is accompanied by the initiation of vertical transmission of microbiota from mother to offspring. Such microbiota acquisition continues during lactation, during which it is influenced by immune systems of mother and offspring. Cessation of lactation ablates the contribution of the maternal immune system to this process, leading to the “weaning reaction” that drives immune system development and maturation of the offspring^1–8 and establishes a symbiotic relationship between the host and its microbiota. Disruption of early life microbiota acquisition results in irreversible immunological and metabolic defects^1,4 that increase susceptibility to various diseases, including inflammatory diseases^1,9,10, asthma^11,12, cancer¹, and metabolic syndrome^2,13–15. One proposed explanation for the importance of this critical time window involves the early-life specific circulation of microbiota-derived antigens through colonic goblet-cell associated antigens passages (GAPs)^16–18, which was reported to be central in driving immune tolerance towards commensals^16,18,19.

Recognition of the pivotal role played by early-life microbiota in adulthood host health has driven efforts to identify environmental factors that influence the establishment of this microbial community. Environmental factors such as early-life antibiotic treatments^2,20, maternal influence through transmission of select bacterial strains^21,22, as well as indirect mechanisms including the involvement of milk components and genetic factors^23–25, have emerged as a contributors of early-life microbiota settlement²⁶. However, the exact human relevance of these findings, as well as the underlying long-lasting mechanisms at play, both remain unclear. Hence, we investigated whether maternal exposure to microbiota stressors could influence offspring’s early-life microbiota and, subsequently, proneness to inflammation. We focused on dietary emulsifiers, a ubiquitous class of food additive known to perturb microbiota in those who consume them. Dietary emulsifiers directly impact the intestinal microbiota in a manner that promotes chronic intestinal inflammation, and consequently metabolic syndrome^27–30. Moreover, recent epidemiological data from the NutriNet Santé cohort, including >175,000 participants, reported positive associations between intake of dietary emulsifiers and risk of cardiovascular diseases, various cancers as well as type 2 diabetes^31–33.

We observed that offspring from emulsifier-exposed dams exhibited early-life microbiota alterations, at both the compositional and functional levels, which associated with increased adulthood susceptibility to diet-induced obesity and colitis. Microbiota normalization through cross-fostering procedures revealed the causal relationship between early-life microbiota alterations and long-lasting disease susceptibility. Mechanistically, offspring from emulsifier-exposed dams displayed premature GAP closure that linked microbiota alterations to long-lasting susceptibility to intestinal inflammation and metabolic syndrome. Altogether, these data suggest that maternal exposure to common dietary factors is sufficient to perturb early-life host-microbiota interactions in a manner that lastingly promotes inflammation.

Results

Maternal emulsifier consumption induced early-life alterations in offspring’s microbiota composition

To investigate the trans-generational effect of dietary emulsifiers consumption on early life microbiota, C57Bl/6J dams were subjected to 1% of carboxymethylcellulose (CMC) or polysorbate-80 (P80) in drinking water for 10 weeks prior breeding (Fig. 1a). Such treatment was sufficient to reproduce previously observed microbiota alteration in dams^27,30 (Fig. S1). After 10 weeks of treatment, breeding pairs (N = 6 per experimental group) were formed with unexposed male, and pregnant females were individually housed from gestation until weaning. At birth, offspring were either kept with their biological mother (experimental groups Water; CMC; P80) or cross-fostered to a water-treated dam with age-matching and size-matching litters (experimental groups Water→Water; CMC→Water; P80→Water), with 3–7 litters used per experimental group. Male offspring were weaned at 3 weeks under emulsifiers-free conditions, as described Fig. 1a. While dams were kept under emulsifier treatment during the pre-weaning phase, we confirmed that offspring were not directly exposed to dietary emulsifiers through milk consumption, with the observation that CMC and P80 were both undetectable by NMR in milk²⁹.

Fig. 1. Maternal emulsifier consumption induced early-life functional and compositional microbiota alterations in offspring.

Dams were subjected to 1% CMC or P80 in their drinking water for 10 weeks prior to breeding. At birth, offspring were either kept with their biological mother or cross-fostered to a water-treated dam. Following weaning of male offspring under emulsifier-free conditions, longitudinal compositional and functional analysis of the offspring’s microbiota were performed. a Schematic representation of the experimental design used. b, c Principal coordinates analysis of the Bray–Curtis distance at week 4, representing offspring from Water, CMC-, and P80-treated dams (b), or Water→Water, CMC→Water, and P80→Water cross-fostered offspring (c). d, e Bray–Curtis distance separating the various experimental groups from the water or Water→Water offspring at week 4 (d) and at week 16 (e). f Relative abundance of MaAsLin2-identified microbiota members at the family level with significantly altered abundance in CMC and P80 offspring compared to control at week 4. Bold indicates families containing flagellated members. g, h Fecal levels of bioactive flagellin (g) and LPS (h), with data normalized to either Water or Water→Water offspring. i Distance of closest bacteria to intestinal epithelial cells (IECs) at week 3, over five high-powered fields per mouse. Data are presented as means ± s.e.m (a–h, N = 13–15; i, N = 19–20; a–i, mice from ≥ 3 independent litters). Significance was assessed by ANOSIM (b, c), one-way ANOVA (d, e, i) or two-way ANOVA (g, h), and is indicated as follow: n.s. indicates non-significant * p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

Offspring’s fecal microbiota composition was next investigated by 16S rRNA gene sequencing. At 4 weeks of age, the overall microbiota composition was significantly divergent between offspring from water-, CMC-, and P80-treated dams (Fig. 1b, d). Moreover, such alteration in composition was prevented when pups were cross-fostered by a water-treated dam (Fig. 1c, d). In adulthood, at 16 weeks of age, no significant alteration in microbiota composition was observed between CMC, P80, and water offspring (Fig. 1e), suggesting that the observed trans-generational alterations are restricted to the early-life period. Use of machine learning to predict dams’ treatment based on offspring microbiota composition revealed a loss of prediction accuracy throughout time (Fig. S2a–h), further highlighting that microbiota alterations in offspring from emulsifier-treated dams are occurring in early-life before normalizing at adulthood.

We next performed Microbiome Multivariable Associations with Linear Models (MaAsLin2)-based analysis to identify microbiota members significantly altered in their abundance at the family taxonomic level, in CMC and P80 offspring compared to water offspring, at week 4. Such an approach revealed 11 families that differed in abundance at week 4 in CMC or P80 offspring compared to control (Fig. 1f). These alterations in bacterial families abundance were not sustained in adulthood, suggesting that they are specific to the early-life period (Fig. S2i–r), which, together with the absence of alteration in caecal-derived metabolites at week 16 (Fig. S2s, t), further highlight that trans-generational alterations in microbiota composition vanished by adulthood.

Various mechanisms are at play in shaping early-life microbiota, including direct mom-to-infant microbiota members transmission. In addition, milk components are known to influence the developing microbiota in offspring^23,25, such as milk-derived IgA, since mice pups start to produce their own IgA only after weaning, making maternal milk as the only source of IgA in early life²⁴. CMC offspring displayed a significant reduction in the proportion of IgA-bound bacteria, which was fully prevented by cross-fostering (Fig. S3a). While we observed modest modifications in select antigen-specific milk-derived IgA, including anti-LPS IgA (Fig. S3b–d), they were not sufficient to significantly impact the overall bacterial community targeted by these antibodies (Fig. S3e). We next investigated if milk-derived bacteria were impacted by emulsifier consumption in a way that drove the observed early life microbiota alterations in the descendants. 16S rRNA sequencing-based approach revealed similar composition in bacterial community in milk from CMC- or P80-treated compared to unexposed dams (Fig. S3f, g). We next assessed maternal milk metabolic profile, known to be a player in offspring’ microbiota establishment^34,35. NMR-based quantification identified 17 metabolites with a similar overall profile (Fig. S3h, i), and only modest emulsifier-induced alterations in alanine (Fig. S3j), creatine (Fig. S3k), and betaine (Fig. S3l) being observed. Altogether, these data indicate that maternal emulsifiers intake induced transient transgenerational microbiota alteration in composition, which may have resulted from a combination of direct mom-to-pups microbiota transmission, together with modest alterations in milk-derived IgA and metabolites.

Maternal emulsifier consumption induced early-life alterations in offspring’s intestine-microbiota interactions

We next investigated the potential trans-generational impact of emulsifier consumption on the offspring’s microbiota at the functional level. For this purpose, we first quantified microbiota-derived pro-inflammatory molecules lipopolysaccharide (LPS) and flagellin, proxi markers of a given microbiota’s ability to stimulate host innate receptors (pro-inflammatory potential)^36,37, and previously reported to be significantly increased upon emulsifier exposure^27,30,38. Longitudinal fecal levels of bioactive flagellin, but not LPS, displayed significant increase at week 4 in CMC and P80 offspring compared to water offspring (Fig. 1g, h). Flagellin increase was paralleled by, also not directly correlated with, changes in abundance of bacterial families comprising flagellated members (Fig. 1f, with families containing flagellated members highlighted in bold), suggesting that these changes arise from select bacterial species and/or transcriptional activity. Such observation indicates that offspring from emulsifier-treated dams display early-life restricted microbiota with increased pro-inflammatory potential, which were fully prevented by cross-fostering (Fig. 1g, h).

Previous observations reporting that microbiota pro-inflammatory potential positively correlates with microbiota encroachment (i.e., the microbiota-epithelium distance)^27,39–41 led us to quantify such proximity at weaning (Figs. 1i and S2u). This distance was significantly reduced in CMC and P80 offspring compared to control offspring, which was potentially explained, at least in part, by the observed increased level of bacterial motility factor flagellin (Fig. 1g) and impaired IgA-mediated mucosa extrusion (Fig. S3b–d). These data further indicate that maternal emulsifier consumption had detrimentally impacted offspring’s intestinal microbiota, at both the compositional and functional levels.

Maternal emulsifier consumption induced long-lasting susceptibility to diet-induced obesity

We next investigated impacts of maternal emulsifier consumption on offspring metabolism. Males weaned under emulsifiers-free conditions were subjected to a western diet (WD, high in fat and low in soluble fibre) regimen, for 13 weeks, as presented Fig. 2a. Offspring of CMC- and P80-fed dams gained significantly more weight upon WD regimen compared to water controls (Fig. 2b), despite normalization of microbiota composition in adulthood (Fig. S4a). Moreover, early-life microbiota normalization through cross-fostering procedure was sufficient to prevent such increased body weight gain (Fig. 2b). Furthermore, offspring from P80-treated dams additionally presented increased peri-epididymal fat pad weight (Fig. 2c) as well as fasting glycemia (Fig. 2d), which were both prevented by cross-fostering. Based on the major impact of WD regimen on hepatic health, we next investigated liver homeostasis by histology and RT-q-PCR. While no differences were observed in liver weight between groups (Fig. S4b), major WD-induced liver alterations were observed in CMC and P80 offspring compared to water offspring, as evidenced by significantly increased steatosis (Figs. 2e and S4c), early-stage fibrosis evidenced by Sirius red staining (Fig. 2f) and altered expression of steatosis-associated genes Acta2, Pparg, and Il1b (Figs. 2g and S4d, e). In addition to liver damages, WD regimen is also known to induce chronic low-grade intestinal inflammation⁴². Histological scoring of colonic sections revealed that CMC and P80 offspring displayed exacerbated WD-induced low-grade intestinal inflammation (Fig. 2f), further evidenced by colon shortening and increased weight, both markers of colonic inflammation and immune cells infiltration (Fig. S4f, g). Importantly, all these WD-induced perturbations observed in CMC and P80 offspring were fully abrogated in water cross-fostered offspring, thus revealing that transgenerational impact of maternal emulsifier consumption have potent and long-lasting deleterious impact on metabolic health, despite only transient detrimental effect on offspring’s intestinal microbiota.

Fig. 2. Maternal emulsifiers consumption induced long-lasting susceptibility to diet-induced obesity and colitis in offspring.

Dams were subjected to 1% CMC or P80 in their drinking water for 10 weeks prior to breeding. At birth, offspring were either kept with their biological mother or cross-fostered to a water-treated dam. Following weaning of male offspring under emulsifier-free conditions, susceptibility to metabolic deregulations and intestinal inflammation was assessed by subjecting mice to a WD regimen or DSS treatment, respectively. a Schematic representation of the experimental design used to investigate susceptibility to WD-induced metabolic deregulations. b Body weight over time, with data being expressed as percentage compared to the body weight at weaning (week 3). c Epididymal fat deposition measured at sacking, d 15 h fasting glycemia at week 14. After sacking, liver histological samples were stained with HE or Sirius red, allowing e steatosis histological scoring and f fibrosis quantification as the percentage of Sirius red positive area. g Liver expression fold change of Acta2. h Histopathological scoring of HE stained colon samples following WD challenge. i Schematic representation of the experimental design used to investigate susceptibility to DSS-induced colitis. j Histopathological scoring of HE stained colon samples following DSS challenge and k colon length. Data are the means ± s.e.m (b–h, N = 13–15; j, k, N = 10–28; b–h, j, k, mice from ≥ 3 independent litters). Significance was assessed by two-way ANOVA (b) or one-way ANOVA (c–h, j, k) and is indicated as follow: n.s. non-significant; * ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

Maternal emulsifier consumption induced long-lasting susceptibility to colitis

In light of the central role played by the early-life intestinal microbiota on long-lasting susceptibility to intestinal inflammation^1,2,13, we next investigated colitis susceptibility in offspring from CMC- and P80-treated dams. For this purpose, males weaned under emulsifier-free conditions were subjected, 4 weeks after weaning (at 7 weeks of age), to 1% dextran sodium sulfate (DSS) diluted in the drinking water for 7 days, as presented Fig. 2i. Despite no significant differences in microbiota profiles before or during DSS treatment (Fig. S5a, b), offspring from CMC- and P80-treated dams displayed increased DSS-induced intestinal inflammation, as evidenced by an increased colonic histopathological score and colon shortening (Fig. 2j, k). CMC and P80 offspring also harbored an increased colon weight (Fig. S5c), suggesting immune cells infiltration, further evidenced by an increased infiltration of CD68-positive immune cells (Fig. S5d). Such increased colitis susceptibility in offspring of CMC- and P80-fed dams was associated with alteration in mucus layer homeostasis, as revealed by decreased Muc2 gene expression (Fig. S5e) and alterations in colonic and ileal lamina propria immune cell profiles (Fig. S6). These alterations were compound-specific, aligning with previous observations on CMC and P80^27,28. Moreover, early life microbiota normalization through cross-fostering procedure in CMC→water and P80→water groups (Fig. 2i) fully prevented the increased colitis susceptibility normally observed in CMC and P80 offspring, suggesting that early-life microbiota alterations had driven the increased colitis susceptibility (Figs. 2j, k and S5c, d). Next, we examined metabolic deregulations potentially associated with low-grade inflammation induced by this relatively low dose of DSS (1%) in chow-fed mice. Despite no changes in body weight (Fig. S5f), we observed increased adiposity, namely peri-epididymal fat deposition, in CMC and P80 offspring compared to water control offspring. This difference was fully prevented through cross-fostering by untreated dams (Fig. S5g).

Finally, to assess whether emulsifier consumption induced transgenerational early life microbiota alterations were sufficient to drive colitis susceptibility, another cross-fostering approach was applied in which pups from water-treated dams were either kept with their own mother or cross foster to a water (water→water group), CMC- (water→CMC group) or P80- (water→P80 group) treated dams (Fig. 2i). Such cross-fostering resulted in early life microbiota alteration (Fig. S5h) and a stark increased in colitis susceptibility, compared to water→water control group, in adulthood (Figs. 2j, k and S5c). Altogether, these data highlight that early life microbiota alterations induced by maternal emulsifier intake were both necessary and sufficient for long-lasting increased susceptibility to colitis.

Early-life microbiota alterations resulting from maternal exposure to dietary emulsifiers induce impaired goblet cells-associated passage at weaning

The observation of transgenerational susceptibility to metabolic deregulations and colitis despite microbiota normalization in adulthood prompted us to focus on the early-life time window, previously described as having long-lasting impact on disease susceptibility^1,2,43. Hence, we investigated the effect of maternal emulsifier intake on the offspring’s intestinal compartment at weaning, as described in Fig. 3a. We found that, at weaning, and even in the absence of WD or DSS insult, offspring from emulsifier-exposed dams displayed increased body weight, adiposity (Fig. 3b, c) and low-grade colonic inflammation (Fig. 3d). Next, to probe mechanisms that linked maternal emulsifier consumption, early-life microbiota alterations and increased disease susceptibility, we performed bulked RNA-seq on colonic samples from weanling pups (Fig. 3e–h). This approach revealed differentially expressed genes in CMC (23 genes, Fig. 3e) and P80 offspring (125 genes, Fig. 3f) compared to water control offspring. Gene enrichment pathways analysis revealed that pathways involved in response to microorganisms are highly impacted by maternal emulsifier consumption (Fig. 3g, h), aligning with the observed increased microbiota pro-inflammatory potential in CMC and P80 offspring (Fig. 1g, i). For example, CMC and P80 offspring were characterized by a decreased colonic expression of Ifng, encoding the IFN-γ cytokine (Fig. 3i), for which microbiota-driven increased expression at weaning was previously reported to be central in driving long-term protective imprinting, a phenomenon termed the weaning reaction¹. With the previous observation that intestinal immune cells are involved in such imprinting, we next assessed the intestinal immune compartment by investigating immune populations and cytokine ex vivo production in weanling mice, using mesenteric lymph node isolated cells. This revealed increased expression of Il1b in mesenteric lymph nodes-derived cells after 24 h ex vivo culture (Fig. 3j), aligning with the observed increased inflammatory tone in CMC and P80 offspring and highlighting the inflammatory skewing of intestinal immune cell activity. Quantification of immune cell populations using flow cytometry revealed that pups born from CMC-exposed dams displayed significantly decreased B cells and increased in T cells (Fig. 3k, l). Moreover, despite no difference in overall T cell population in P80 offspring (Fig. 3l), we observed a reduction in total FoxP3⁺ T_reg (Fig. 3m), mostly driven by RorɣT⁺ tissue induced T_reg⁴⁴ rather than natural T_reg (Fig. 3n, o), suggesting that early life microbiota of P80 offspring may have reduced capacity to promote tolerogenic immune cells.

Fig. 3. Maternal emulsifiers consumption induced intestinal inflammation and perturbed the weaning reaction.

Dams were subjected to 1% CMC or P80 in their drinking water for 10 weeks prior to breeding, and males offsprings were used just prior weaning for the analysis of intestinal inflammation and goblet-cell mediated molecular communication. a Schematic representation of the experimental design used. b Body weight, c peri-epididymal fat deposition, and d histopathological scoring of HE stained colon samples at weaning. RNA seq was performed on colon samples collected at weaning; e, f volcano plots of genes expression in Water vs CMC offspring (e) or Water vs P80 offspring (f). g, h Significantly altered pathways between Water and CMC offspring (g) or between Water and P80 offspring (h). i Colonic expression fold change of Ifng. j Il1b fold change expression by mLN cells after 24 h ex vivo culture. k–o Mesenteric lymph nodes (mLN) cells were analyzed by flow cytometry; k B220⁺ B cells, l CD3⁺ T cells, m Foxp3⁺ CD25⁺ CD127⁻ T_reg, n Foxp3⁺ RORɣt⁺ T_reg, and o Foxp3⁺ RORɣt⁻ T_reg. p, q Quantification of colonic goblet-cell associated antigens passages at 18 days, expressed as the mean of GAP/crypt with the measurement of ovalbumen-positive goblet cells performed in 50 colonic crypt per samples (p). Representative images of ovalbumin-labelled GAPs in transversal colonic sections (Ovalbumin labeled in pink, actin labeled in light blue, scale = 50 μm) (q). r Goblet cells per colonic crypt, quantified on alcian blue strained colonic section. Data are the means ± s.e.m (b–i, N = 19–20; j, N = 11–19; k–o, N = 13–24; p–r, N = 12–20; b–r, mice from ≥ 3 independent litters). Significance was assessed by two-way ANOVA (b–d, i–p, r) or using the two-sided quasi-likelihood F-test (e–h) and is indicated as follow: n.s. non-significant; * ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001.

We next investigated mechanisms that might link reduced Ifng expression to alterations in tolerogenic immune cells. We envisaged that one possible mediator might be goblet cells associated passage (GAPs), which are only observed during the early life window of opportunity and are suspected to be important for proper immune development and intestinal health¹⁶. GAP closure, as well as increase in TLR5 signaling that results in CMC and P80 offspring from their elevated levels of microbiota flagellin, both rely on MyD88-dependent signaling^16,45, which led us to quantify colonic GAPs in 18 day-old offspring from water-, CMC- or P80-treated dams. Using colonic instillation with fluorescently labelled ovalbumin, followed by confocal-based quantification of ovalbumen-positive goblet cells, we observed a significant decrease in the number of colonic GAPs in CMC and P80 offspring (Fig. 3p, q), despite no reduction in the total number of goblet cells (Fig. 3r).

To rule out the possibility that the observed phenotypes in the offspring result from direct exposure to emulsifiers (i.e., through emulsifiers present in the dams’ drinking water), we next conducted a complementary experiment in which P80 treatment was discontinued at birth instead of weaning (Fig. S7a), ensuring that pups had never direct access to the treatment. At weaning, offspring from such P80-treated dams still exhibited a comparable reduction in colonic GAPs (Fig. S7b), decreased colonic expression of the weaning reaction–associated gene Ifng (Fig. S7c), as well as increased body weight, adiposity, and intestinal inflammation (Fig. S7d–f). These findings strongly suggest that the observed phenotypes are not attributable to direct exposure of the offspring to maternal treatment.

Altogether, these results suggest that maternal emulsifier consumption altered offspring microbiota to elevate MyD88-mediated pro-inflammatory gene expression that led to premature GAPs closure in a way that aborted establishment of proper host-microbiota interaction.

Pharmacologic restoration of GAPs prevented transgenerational consequences of maternal emulsifier consumption

We next investigated whether premature GAPs closure contributed to deregulated inflammatory tone at weaning and the associated long-lasting consequences. For this purpose, pups from water-, CMC- or P80-treated dams were intraperitoneally injected daily from day 10 to day 21 with Tyrphostin, an inhibitor of the Epidermal Growth Factor Receptor (EGFR), known to prevent GAPs closure in the colon¹⁶ (Fig. 4a). As previously reported, this regimen fully prevented the reduction in GAPs that was otherwise observed in CMC and P80 offspring (Fig. 4b, c). Such maintenance of GAPs not only eliminated the decreased in colonic expression of Ifng that had resulted from maternal emulsifier consumption, but in fact led to an increase in this parameter (Fig. 4d). These results suggest that the reduction in IFNγ induced by maternal emulsifier consumption reflected premature GAP closure, potentially to avoid the increased microbiota-induced immune response that occurred when GAPs were pharmacologically forced to remain open (Fig. S8a–f). Phenotypically, maintenance of GAPs opening in CMC and P80 offspring was associated with restoration of normal body weight at weaning (Fig. 4e), peri-epididymal fat deposition (Fig. 4f), as well as colonic inflammatory level (Fig. 4g).

Fig. 4. Chemically-induced GAP restoration prevented intestinal inflammation and susceptibility to diet-induced obesity that otherwise resulted from maternal emulsifier consumption.

Dams were subjected to 1% CMC or P80 in their drinking water for 10 weeks prior to breeding, and obtained pups were daily injected, from day 10 to day 21, with tyrphostin in order to maintain GAPs opening. Female offspring were sacked at weaning to quantify GAPs, intestinal inflammation, and metabolism, while male offspring were weaned under emulsifiers-free conditions and challenged with a WD for 13 weeks. a Schematic representation of the experimental design used. b Quantification of colonic goblet-cell associated antigens passages and c representative images of ovalbumin-labelled GAPs in transversal colonic sections (Ovalbumin labeled in pink, actin labeled in light blue, scale = 50 μm). d Colonic expression fold change of Ifng. e Body weight at weaning, f peri-epididymal fat deposition, and g histopathological scoring of HE stained colon samples. h Schematic representation of the experimental design used to investigate susceptibility to WD-induced metabolic deregulations of tyrphostin-treated mice. i Body weight over time, with data being expressed as percentage compared to the body weight at weaning (week 3). j Peri-pididymal fat deposition measured upon sacking and k 15 h fasting glycemia measured at week 14. l, m Liver histological samples were stained with HE or Sirius red for steatosis scoring (l) and fibrosis quantification as the percentage of Sirius red positive area (m). n Histopathological scoring of HE stained colon samples following WD challenge. Data are the means ± s.e.m (b, c, N = 7; d–g, N = 7–12; i–n; N = 6–8; b–n, mice from ≥ 3 independent litters). Significance was assessed by one-way ANOVA (b, d–g, j–n) or two-way ANOVA (i) and is indicated as follow: n.s. non-significant; * ≤ 0.05; **p ≤ 0.01.

Finally, we investigated whether early life anticipated GAPs closure contributed to the long-lasting susceptibility to diet-induced obesity. Mice subjected to tyrphostin treatment between day 10 to day 21 were subsequently challenged with a WD for 13 weeks, starting at weaning (Fig. 4h). Maintenance of GAPs opening in CMC-Tyrphostin and P80- Tyrphostin mice was associated with a full prevention of the increased susceptibility to diet-induced metabolic deregulations normally observed in CMC and P80 offspring, as evidenced by normalization of body weight gain (Fig. 4i), peri-epididymal fat deposition (Fig. 4j), glycemia (Fig. 4k), liver steatosis and fibrosis (Fig. 4l, m), as well as intestinal inflammation (Fig. 4n). Thus, premature GAP closure was driving the adverse transgenerational metabolic consequences of effects of maternal emulsifier consumption.

Discussion

Data accumulating over the last 15 years, from ourselves and others, have revealed the deleterious impact of dietary emulsifier on the intestinal microbiota, with subsequent detrimental consequences for host health. This includes the observation that dietary emulsifiers consumption, in both mice and humans, alter microbiota composition and increase its pro-inflammatory potential in a way that could drive intestinal inflammation and metabolic dysregulation^{27,29,30,38,39,46}. More recently, large epidemiological studies have reported positive associations between emulsifiers intake and risk of various chronic diseases^31–33. That microbiota is vertically transmitted from mother to offspring suggests the potential existence of trans-generational consequences of dietary emulsifier consumption, but this possibility had not been previously investigated. While the role played by the early life microbiota in shaping long-term host health has been recently documented, the exact actors at play, as well as the associated molecular mechanisms, remain largely unknown. Hence, we investigated here the potential consequence of maternal emulsifier consumption and its associated microbiota alterations on the next generation’s microbiota and health.

We observed that maternal emulsifier consumption induced offspring’s microbiota alteration in early life, at both the compositional and functional levels. While we previously reported that emulsifier treatment removal in adult mice reversed most associated health consequences²⁷, we report here that maternally-induced microbiota alterations associate with long-lasting susceptibility to diet-induced obesity and colitis, and cross-fostering approaches revealed a central role played by the intestinal microbiota in dictating such long-lasting consequences. We hypothesize that such long-lasting deleterious imprint of maternal-driven early-life microbiota alteration is mediated by impaired immune tolerance settlement in early-life, resulting in life-lasting hyper-inflammatory immune system, known to have major impact on host metabolism^47–49. Indeed, in early life, colonic goblet cells-associated antigen passages (GAPs) quantification led to the observation that offspring from emulsifier-exposed dams displayed a stark reduction in GAPs levels. Based on the previous observation that MyD88 signaling is central in modulating GAPs opening, with both caecal content from SPF mice and luminal LPS installation being sufficient to induce GAPs premature closure in the colon of WT but not in MyD88^−/− host¹⁶, we hypothesize that microbiota encroachment within the normally sterile inner mucus layer contributes to increased MyD88 signaling prior to weaning that drives premature GAP closure. This hypothesis is supported by our observation that maternal emulsifier consumption led to increased levels of bioactive flagellin in offspring, which could activate MyD88-dependant signaling, especially during this early-life time period characterized by high Tlr5 colonic expression^3,50. Importantly, chemical maintenance of GAPs opening was associated with the prevention of long-lasting transgenerational consequences of maternal emulsifier consumption, highlighting the crucial role played by GAPs-mediated early-life host-microbiota communication in host long-lasting health, as illustrated Fig. S9. Furthermore, increased GAP levels was strongly associated with IFNγ levels, suggesting that GAP modulate the previously described IFNγ pick observed during the weaning reaction (Figs. S8 and S9). The association between early-life GAP levels and adult health status may stem from early-life imprinting of long-lasting tolerogenic cell populations. This is particularly relevant given recent literature linking both Ifng expression and GAP levels at weaning with the induction of T_reg—both of which have been shown to be critical determinants of long-term health^1,16,26. Increased IFNγ following Tyrphostin-induced GAPs opening boost was even further enhanced in CMC and P80 offspring compared to water offspring, which we suspect to be driven by the increased immunostimulatory potent of the luminal microbiota-derived antigens trafficking through GAPs in these emulsifier-treated mice. In addition, we also observed decreased levels of S100-alarmin in offspring from P80-treated dams, which were reported to prevented hyperinflammatory responses in early life through altering MyD88-dependent pro-inflammatory gene programs⁵¹. Impaired S100-alarmine resulting from maternal emulsifiers intake may hence contribute to the observed intestinal inflammation and anticipated GAP closure.

We observed that microbiota composition differed between CMC- and P80-exposed dams, as previously described^27,38, suggesting compound-specific effects. Additionally, P80- and CMC-specific metabolic deregulations were noted in the offspring following a WD regimen, suggesting that there may not be a single, uniform imprinting mechanism, but rather a combination of long-lasting effects driven by emulsifier-specific microbiota alterations, at the compositional and/or functional level. In addition, in CMC offspring, some phenotypes were not fully restored upon microbiota normalization at birth, suggesting that this compound and/or the associated altered maternal microbiota may have an in utero impact on the offspring. During fetal life, mother’s microbiota could indeed produce compounds that are transferred to the fetus in a way that can impact innate immune cell populations⁵². The deleterious transgenerational effects of maternal emulsifier intake on offspring health highlighted in this study underscore the need for further research into the impact of direct emulsifier exposure during the early life period, especially in Humans given the presence of these additives in various infant formula products⁵³. Another point in dire need of further investigations relates to the role played by the recently observed heterogeneous response to emulsifiers in humans on the transgenerational effects described here. We indeed recently reported that microbiota inter-individual variation are key in driving host responses to emulsifier consumption²⁹, leading to some individuals being highly sensitive to emulsifier-induced perturbations. Investigating the trans-generational consequences of such personalized responsiveness toward emulsifier appears essential to fully appreciate the impact of maternal emulsifier consumption across generations.

To conclude, our observation that maternal emulsifiers intake induces offsprings microbiota alteration in a way that disrupt host-microbiota interactions, leading to long-lasting susceptibility to metabolic and inflammatory diseases, underscores the need to carefully consider maternal exposure and dietary recommendations to ensure the long-term health of future generations.

Limitations of the study

While this study offers novel insights into how maternal emulsifier intake shape offspring health through early-life microbiota alterations, several limitations must be acknowledged.

• First, the mechanistic underpinnings of the observed increase in flagellin remain incompletely resolved. Due to sequencing methods resolution, we were unable to identify which specific bacterial taxa or transcriptional changes account for increased flagellin levels observed during the peri-weaning phase. This is further complicated by the fact that both LPS and flagellin exist in multiple forms, each with distinct capacities to engage immune signaling pathways^54,55. As a result, accurately quantifying their functional impact remains technically challenging.

• Next, although we observed associations between reduced GAP formation and early-life microbiota compositional and functional alterations, the precise drivers of GAP modulation remain unclear. It is currently unknown whether GAP closure is induced by microbial proximity, flagellin signaling, milk components, or a combination thereof. Similarly, while pharmacological restoration of GAPs in early life appeared to rescue inflammatory and metabolic outcomes, we cannot exclude the possibility that this reflects a correlative rather than causal relationship.

• Moreover, although we observed alterations in mucosal immune populations and transcriptional signatures, we could not determine which specific immune cell types are directly responsible for the lasting phenotypes. T_reg have been identified as central in driving the lasting protection associated with the weaning reaction and early-life colonic GAPs^1,16,26 and form the basis of our working hypothesis, but this remains to be formally demonstrated in our model.

• Finally, we focused here primarily on structural features such as GAPs, without broad profiling of intestinal metabolites in offspring. Since GAPs do not themselves modulate immunity but rather serve as conduits for luminal antigens and microbial signals, identifying the bioactive metabolites they transport remains an important next step.

Methods

Mice

Four-week-old male and female C57Bl/6J mice were housed in cages of five individuals and maintained under a 12-h light/dark cycle. The male mice were housed with ad libitum access to standard chow diet (SAFEA03, Scientific Diets) and water, while the female mice were divided into three groups: a water-treated group, a group treated with 1% carboxymethyl cellulose (CMC), and a group treated with 1% polysorbate 80 (P80) in the drinking water. Cages for all groups were changed every other week. After 10 weeks of treatment, breeding pairs were formed by pairing two females with one male. Pregnant females were subsequently individually housed from gestation until weaning. On the day of birth, pups were either left with their biological mother or cross-fostered to dams with age-matching and size-matching litters. The resulting pups were categorized as follows:

Experiment 1—Figs. 1, 2, S2, and S4

Pups born to water-, CMC-, or P80-treated dams were either kept by their biological mother or cross-fostered with water-treated dams at birth. Male pups were weaned at 21 days-old under emulsifier-free conditions, with ad libitum access to water, and then subjected to a western diet (WD, D12492, Research Diets) for 13 weeks.

Experiment 2—Figs. 2, S5, S6

Pups born to water-, CMC-, or P80-treated dams were either kept by their biological mother or cross-fostered with water-, CMC-, or P80-treated dams at birth. Male pups were weaned at 21 days old under emulsifier-free conditions, with ad libitum access to standard chow diet and water. After 4 weeks, these mice were subjected to 1% dextran sodium sulfate (DSS) in the drinking water for 7 days.

Experiment 3—Fig. 3

Half of the male pups born to water-, CMC-, or P80-treated dams were used at day 18 to quantify goblet cells-associated antigen passage (GAP), while the other half were euthanized on the day of weaning and used for either RNA-seq or intestinal inflammation quantification.

Experiment 4—Fig. 4

Pups born to water-, CMC-, or P80-treated dams were injected daily intraperitoneally (IP) with 0.5 μg of Tyrphostin per gram of body weight from day 10 to day 21¹⁶. Female pups were used to quantify GAP at weaning. Male pups were weaned at 21 days old under emulsifier-free conditions, with ad libitum access to water, and then subjected to a WD for 13 weeks. Fasting glycemia was measured after 10 weeks of WD.

Experiment 5—Fig. S7

Dams’ P80 treatment was discontinued on the day of birth. Resulting pups, as well as control pups from water-treated dams, were used at day 18 to quantify colonic GAP, fat deposition, and intestinal inflammation.

Body weight was measured and feces were collected longitudinally throughout the experiments. Upon euthanasia, mice were anesthetized with isoflurane, blood samples collected, and mice were euthanized by cervical dislocation. Measurements of colon length, colon weight, spleen weight, and adipose weight were recorded, and tissue samples were collected for downstream analysis. Animal welfare and experimental protocols followed the ARRIVE guidelines (Animal Research: Reporting of in vivo Experiments). All procedures involving animals were approved by the French Ministère de la l’enseignement supérieur, de la recherche et de l’innovation, APAFIS#24788-2019102806256593 v8.

Milk collection

Following 6 h of pups isolation from their mother, dams are injected IP with 1.5 UI of Ocytocine-S, (Sigma, O3251) prior anesthesia with isoflurane for milk collection. Milk samples were snap frozen for downstream analysis.

Colonic GAP staining and quantification

Pups were subjected to intro-colonic injection with 0.25 mg of Ovalbumin Alexa Fluor 647 conjugated (Invitrogen 034784). After 1 h, mice were euthanized and colons were opened longitudinally and washed twice in PBS prior fixation in 4% paraformaldehyde. Following embedding in paraffin with a vertical orientation, five-μm sections were cut and dewaxed by bathing in xylene at 60 °C for 10 min, xylene at room temperature for 10 min, and 99.5% ethanol for 10 min. Sections were marked using a PAP pen (Sigma, St. Louis, MO, USA) and block solution (5% FBS in PBS) was added for 30 min at 4 °C. Mucin-2 primary antibody (rabbit H-300, [C3], C-term, Genetex, GTX100664) was diluted to 1:100 in block solution and applied overnight at 4 °C. After washing 3 × 10 min in PBS, block solution containing anti-rabbit Alexa 488 secondary antibody diluted to 1:300, PhalloidinTetramethylrhodamine B isothiocyanate (Sigma-Aldrich) at 1 mg ml⁻¹ and Hoechst 33258 (Sigma-Aldrich) at 10 mg ml⁻¹ was applied to the section for 2 h. After washing 3 × 10 min in PBS slides were mounted using Prolong anti-fade mounting media (Life Technologies) and kept in the dark at 4 °C. Observations and measurement of the distance between bacteria and epithelial cell monolayer were performed with a Spinning Disk IXplore using the Olympus cellSens imaging software 421 (V2.3) at a frame size of 2048 × 2048 with 16-bit depth. A 405 nm laser was used to excite the 422 Hoechst stain (epithelial DNA), 488 nm for Alexa Fluor 488 (mucus), 488 nm for Phalloidin (actin), 423 and 640 nm for Alexa Fluor 647 (ovalbumin). Samples were imaged with a ×10 objective, and GAP identified as ovalbumin-labeled goblet cells were quantified over 50 villus using QuPath 0.5.0 software.

Fasting blood glucose measurement

After 10 weeks of WD, mice were placed in a clean cage and fasted for 15 h. Blood glucose concentration was then determined using a Nova Max Plus Glucose Metre and expressed in mg/dL.

Steatosis and fibrosis scoring

Following euthanasia, livers were harvested and fixed in 4% paraformaldehyde, embedded in paraffin, five-μm sectioned, and stained with hematoxylin and eosin (steatosis) or Sirius red (fibrosis). Steatosis was evaluated blindly and as previously described⁵⁶. Briefly, on 5 equal size section of each slide, a score between 0 and 3 was given to macrovesicular steatosis, microvesicular steatosis, hypertrophy, and number of inflammation foci⁵⁶. Fibrosis was determined by measuring the percentage of Sirius-positive area on the whole section.

Staining of colonic tissue and histopathologic analysis

Following euthanasia, colons (proximal colon, 2 first cm from the cecum) were placed in Carnoy’s fixative solution (60% methanol, 30% chloroform, 10% glacial acetic acid). Tissues were then washed in methanol 2  × 30 min, ethanol 2 × 15 min, ethanol/xylene (1:1) 15 min, and xylene 2 × 15 min, followed by embedding in paraffin with a vertical orientation. Tissues were sectioned at 5-μm thickness and stained with hematoxylin & eosin (H&E) using standard protocols. H&E-stained slides were assigned four scores based on the degree of epithelial damage and inflammatory infiltrate in the mucosa, submucosa, and muscularis/serosa. Each of the four scores was multiplied by 1 if the change was focal, 2 if it was patchy, and 3 if it was diffuse, as previously described⁵⁷. The four individual scores per colon were added, resulting in a total scoring range of 0–36 per mouse.

Colonic sections (4 μm) were also stained with Alcian Blue, preferentially staining mucopolysaccharides, and 17–23 crypts of 3 regions per colonic sections were randomly selected per animal to determine goblet-cell number per crypt.

Liver and colon mRNAs extraction and q-RT-PCR analysis

Distal colons were collected during euthanasia and placed in RNAlater. Total mRNAs were isolated from colonic tissues using TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions and as previously described⁵⁷. Quantitative RT-PCR was performed using the Qiagen kit QuantiFast® SYBR® Green RTPCR in a LigthCycler® 480 instrument (Roche Molecular Systems, Inc.) with mouse gene-specific oligonucleotides (Supplementary Table 1). Gene expressions are presented as relative values using the Ct approach with the 36B4 housekeeping gene.

Ex-vivo cell culture and cytokine quantification

Cells from mesenteric lymph nodes (mLN) were isolated as follow: mLN were collected in HBSS prior cell dissociation by pushing cells through a 100 μm filter with a syringe plug. Cells were then washed once in HBSS before resuspending 500,000 cells in RPMI for culture. After 24 h, cells are collected and mRNAs were extracted as described above.

Small intestine, colon, and mLN cell preparation, and flow cytometry analysis

mLN cells were isolated as previously described in the method section. For small intestinal and colonic lamina propria cells isolation, Peyer’s Patches were removed, and whole small intestine and colon were opened longitudinally, cut into pieces, and incubated for 30 min in 30 mM EDTA, washed extensively, and incubated for 15 min at least twice in 1 mg/mL collagenase D and 1 U/mL DNAse I in HBSS. These preparations were then pushed through a 100 μm filter to generate single-cell suspensions. Cells were separated by a 40/80% (w/v) Percoll (GE Healthcare) density gradient and washed prior to staining for flow cytometry analysis. Cells were then pre-incubated with Zombie UV™ Fixable Viability Kit for 30 min, Fc-Block for 15 min, and stained for 20 min with the following antibodies: anti-CD45 PERCP (530-F11; Biolegend), anti-B220 Pacific Blue (RA3-6B2; BD Biosciences), anti-CD3 BV711 (17A2; Biolegend). Cells isolated from ileum and mLN were additionally stained with anti-CD138 BB515 (281-2; BD Biosciences), anti-IgA BV786 (C10-1; BD Biosciences), anti-CD4 (GK1.5; BD Biosciences), anti-CD8 V500 (53-6.7; BD Biosciences), anti-CD127 BV605 (A7R34; Biolegend), anti-CD25 BV650 (PC61; BD Biosciences) and intracellular staining was performed following fixation and permeabilization with anti-Tbet PE-CF594 (O4-46; BD Biosciences), anti-Gata3 AF700 (TWAJ; eBioscience), anti-RorɣT PE (Q31-378; BD Biosciences), anti-FoxP3 APC (FJK16s; eBioscience). Cells isolated from colon were additionally stained with anti-MHCII APC-Cy7 (M5/114.15.2; Biolegend), anti-CD115 APC (AFS98; eBioscience), anti-Ly6G/C BV650 (RB6-8C5; Biolegend), anti-CD103 PE-CF594 (M290; BD Biosciences), anti-CD11c BB515 (NA18; BD Biosciences), anti-CX3CR1 AF700 (SA011F11; Biolegend), anti-CD206 PE (C068C2; Biolegend), anti-CD27 BV786 (LG3A10; BD Biosciences) and anti-CD11b BV421 (M1/70; BD Biosciences). Samples were acquired on a BD LSRFortessa™ Cell Analyzer, and data were analyzed using a non-supervised approach using the Omiq software.

Colonic mRNAs sequencing

Distal colon was collected during euthanasia and placed in RNAlater. Total mRNAs were isolated from colonic tissues using TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions and as previously described⁵⁷.

Library preparation and sequencing

cDNA library was prepared using the Invitrogen^TM Collibri^TM Stranded RNA library Prep Kit for Illumina^TM with Collibri^TM H/M/R rRNA Depletion Kit according to the manufacturer’s instructions and starting with 500 ng of purified RNAs. Briefly, rRNA were first depleted, and enriched mRNAs subsequently used for fragmentation, adaptors ligation, and reverse transcription. After purification, libraries were PCR-enriched, further purified, and quantified and quality-assessed on an Agilent^TM 2100 Bioanalyzer^TM instrument. A master library was generated from the purified products in equimolar ratios. The pooled product was quantified using Qubit and sequenced using an Illumina Next-Seq sequencer (paired-end reads, 2 × 750 bp).

Data analysis

Cutadapt online tool⁵⁸ was used to remove adapter sequences as well as trim sequences from the first low-quality (<28) base. High-quality reads longer than 20 nucleotides were then aligned to the mm10 mus musculus reference genome using Bowtie2⁵⁹. Gene expression levels were next measured using Cufflinks⁶⁰ and differentially expressed genes between conditions were identified using Cuffdiff ⁶⁰. Fragments Per Kilobase of transcript per Million mapped reads (FPKM) unit was used and Log2 fold changes and p-values were computed for each comparison of interest. Gene level volcano plots were generated through R (R version 4.2.3) and differentially expressed genes enrichment analysis was performed using the edgeR R package⁶¹ (package version 3.40.2) to identify genes with Log2 fold changes ≥ log(2) or ≤ −log(2) and p-value ≤ 0.05. Differentially expressed genes pathways enrichment analysis was performed using the PANTHER classification sytem⁶² and significant pathways with FDR ≤ 0.05 and more than 6 DEG involved were represented as enrichment plot (enrichment_chart, R package v2.3.0).

Caecal metabolites quantification

Proton nuclear magnetic resonance (¹H NMR)—based metabolomics was performed on 100 mg of frozen caecal content as previously described⁶³. Data pretreatment and statistical analysis with principal component analysis was performed using full resolution spectra in MATLAB (R2021a; MathsWorks, Inc.).

Maternal milk metabolites qualification and detection of CMC and P80

Maternal milk metabolites were quantified using ¹H NMR-based metabolomics as described in our previous study with minor modifications⁶⁴. Approximately 200 μl of milk were mixed 1 mL of methanol:chloroform (2:1 v/v), vortexed for 30 s, and then 133 µL of nanopure distilled water and 333 µL of chloroform were sequentially added to the sample, vortexed for 30 s, and kept overnight at 4 °C. After centrifugation at 6000 × g at 4 °C for 20 min, the polar phase was collected in a 2 mL microcentrifuge tube and dried down using a SpeedVac vacuum concentrator. The dried extract was resuspended in 0.55 ml of 0.1 M PBS (50% D2O, 0.005% w/v TSP), and after centrifugation at 17000 × g at 4 °C for 10 min, the supernatant was transferred into 5 mm NMR tube for NMR analysis. The 1D ¹H spectra of milk extracts were acquired at 298 K using a Bruker Avance NEO 600 MHz spectrometer equipped with a SampleJet sample changer (Bruker Biospin, Germany). The standard pulse sequence (noesygppr1d) was used for recording ¹H NMR experiments with pre-saturation water suppression during relaxation and mixing time. All spectra were processed automatically with Chenomx NMR Suite 10 (Chenomx Inc., Edmonton, Alberta, Canada), and each spectrum was checked and adjusted manually for phase and baseline to satisfy quality requirements. Metabolites were identified and quantified using the built-in metabolite library and fitting algorithm in the Chenomx software, combined with the known concentrations internal standard (TSP, 0.29 mM).

Quantification of fecal IgA-coated bacteria

IgA-coated bacteria were quantified as previously described⁶⁵. In brief, frozen fecal samples were thoroughly homogenized in PBS to a final concentration of 20 mg/mL. Fecal suspensions were filtered through a 40-μm sterile nylon mesh, then centrifuged at 50 × g, for 15 min at 4 °C. 200 μL of supernatant was then washed with 1 mL PBS and centrifuged at 8000 × g, for 5 min at 4 °C. Resulting bacterial pellets were resuspended in 100 μl blocking buffer (staining buffer containing 20% Normal Rat Serum) and incubated for 20 min on ice before being stained with 100 μl of staining buffer containing PE-conjugated Anti-Mouse IgA (mA-6E1; eBioscience) for 30 min on ice, in the dark. Following two washes with staining buffer, pellets were resuspended in 200 μL of FACS buffer (PBS, 1% Normal Rat Serum). Data acquisition was performed on a Beckman Coulter Gallios flow cytometer. For each sample, 100,000 events were recorded and data was analyzed using FlowJo software v.10.8.2.

Milk total IgA and anti-flagellin/LPS IgA quantification by ELISA

Milk samples were resuspended 1 in 10 in collection media consisting of 0.05 mg soybean trypsin inhibitor per ml of a 3:1 mixture of 1× PBS and 0.1 M EDTA, pH 7.4. Following centrifugation at 400 g for 10 min, the supernatant was centrifuged again at 21,000 × g for 15 min at 4 °C, and final supernatant was collected and stored with 20% glycerol and 2 mM phenylmethylsulfonyl fluoride. Quantification of total IgA, anti-flagellin, and anti-LPS IgA was performed by coating 96-well microtiter plates (Costar, Corning, New York) with goat anti-mouse IgA (Southern Biotech), or 100 ng/well of laboratory-made Salmonella Typhimurium-derived flagellin, or 2 μg/well lipopolysaccharides (from E. coli 0128: B12, Sigma) in 9.6 pH bicarbonate buffer overnight at 4 °C. Serum or fecal samples from mice were then applied either pure at a final dilution of 1 in 10,000 for total IgA ELISA or 1 in 50 for anti-flagellin/LPS IgA ELISA for 1 h at 37 °C. After incubation and washing, the wells were incubated with horseradish peroxidase-linked anti-mouse IgA (Southern Biotech, 1040–05). Quantification of immunoglobulin was then developed by the addition of 3,3′,5,5′-Tetramethylbenzidine and the optical density was calculated by the difference between readings at 450 nm and 540 nm.

Microbiota analysis by 16S rRNA gene sequencing

16S rRNA gene amplification and sequencing were done using the Illumina MiSeq technology following the protocol of Earth Microbiome Project with their modifications to the MOBIO PowerSoil DNA Isolation Kit procedure for extracting DNA (www.earthmicrobiomeorg/emp-standard-protocols). Bulk DNA were extracted from frozen extruded feces using a PowerSoil-htp kit from MoBio Laboratories (Carlsbad, California, USA) with mechanical disruption (bead-beating). The 16S rRNA genes, region V4, were PCR amplified from each sample using a composite forward primer and a reverse primer containing a unique 12-base barcode, designed using the Golay error-correcting scheme, which was used to tag PCR products from respective samples⁶⁶. We used the forward primer 515 F 5′- AATGATACGGCGACCACCGAGATCTACACGCTXXXXXXXXXXXXTATGGTAATTGTGTGYCAGCMGCCGCGGTAA-3′: the italicized sequence is the 5′ Illumina adapter, the 12 × sequence is the golay barcode, the bold sequence is the primer pad, the italicized and bold sequence is the primer linker and the underlined sequence is the conserved bacterial primer 515 F. The reverse primer 806 R used was 5′-CAAGCAGAAGACGGCATACGAGATAGTCAGCCAGCC GGACTACNVGGGTWTCTAAT-3′: the italicized sequence is the 3′ reverse complement sequence of Illumina adapter, the bold sequence is the primer pad, the italicized and bold sequence is the primer linker and the underlined sequence is the conserved bacterial primer 806 R. PCR reactions consisted of Hot Master PCR mix (Quantabio, Beverly, MA, USA), 0.2 μM of each primer, 10–100 ng template, and reaction conditions were 3 min at 95 °C, followed by 30 cycles of 45 s at 95 °C, 60 s at 50 °C, and 90 s at 72 °C on a Biorad thermocycler. Products were then visualized by gel electrophoresis and quantified using Quant-iT PicoGreen dsDNA assay (Clariostar Fluorescence Spectrophotometer). A master DNA pool was generated in equimolar ratios, subsequently purified with Ampure magnetic purification beads (Agencourt, Brea, CA, USA) and sequenced using an Illumina MiSeq sequencer (paired-end reads, 2 × 250 bp) at the Genom’IC platform (INSERM U1016, Paris, France).

16S rRNA gene sequence analysis

16S rRNA sequences were analyzed using QIIME2—version 2022⁶⁷. Sequences were demultiplexed and quality filtered using the Dada2 method⁶⁸ with QIIME2 default parameters in order to detect and correct Illumina amplicon sequence data, and a table of QIIME2 artifact was generated using the folowing dada2 command: qiime dada2 denoise-paired –i-demultiplexed-seqs demux.qza –p-trim-left-f 0 –p-trim-left-r 0 –p-trunc-len-f 180 –p-trunc-len-r 180 –o-representative-sequences rep-seqs-dada2.qza –o-table table-dada2.qza –o-denoising-stats stats-dada2.qza –p-n-threads 6. A tree was next generated, using the align-to-tree- mafft-fasttree command, for phylogenetic diversity analyses, and alpha and beta diversity analyses were computed using the core-metrics-phylogenetic command. Principal coordinate analysis (PCoA) plots were used to assess the variation between the experimental group (beta diversity). For taxonomy analysis, features were assigned to operational taxonomic units (OTUs) with a 99% threshold of pairwise identity to the Greengenes reference database (version Greengenes2 2022.10).

Identification of microbiota members significantly altered in their relative abundance

Microbiota members presenting significant changes in their abundance at the family taxonomic level between Water and CMC/P80 offspring were identified using MaAsLin2 (Microbiome Multivariable Associations with Linear Models, version 2, R version 4.1.2, Maaslin2 version 1.12.0 package)⁶⁹. Microbiota members were reported as significantly altered in their relative abundance if corrected q-value < 0.05 (Fig. 1e).

Identification of bacteria targeted by maternal milk IgA

Milk samples were initially diluted at a ratio of 1:50 with a bacterial suspension derived from RagKO mice fecal material⁷⁰, which was previously resuspended in phosphate-buffered saline (PBS) to achieve a final concentration of 100 mg/mL after homogenization and filtration. Following a one-hour incubation period at 37 °C, IgA-coated bacteria were identified by flow cytometry following IgA staining, and a population enriched with IgA-positive bacteria, comprising 100,000 events, was isolated using a S3e Cell Sorter Biorad. Subsequently, DNA extraction was carried out from the sorted sample containing IgA-coated bacteria, as well as from the mixture resulting from the incubation of milk in the fecal suspension from Rag mice fecal material. The QIAamp Fast DNA Stool Mini Kit (Qiagen) was next used for DNA extraction. For taxonomic profiling, 16S sequencing was conducted, enabling the identification of bacteria bound by IgA. Significantly enriched bacteria were determined by comparing the IgA-coated bacteria enriched sorted sample with the product obtained from the incubation of milk in the fecal suspension from Rag mice fecal material, utilizing the MaAsLin2 approach.

Bioactive flagellin and LPS fecal load quantification

Levels of fecal bioactive flagellin and LPS were quantified, as previously described³⁷, using human embryonic kidney (HEK)-Blue-mTLR5 and HEK-Blue-mTLR4, respectively (Invivogen, San Diego, California). Fecal material were resuspended in PBS to a final concentration of 100 mg.mL⁻¹ and homogenized for 15 min using a vortex. We then centrifuged the samples at 8000 × g for 15 min and serially diluted the resulting supernatant and applied to mammalian cells. Purified Escherichia coli flagellin and LPS (Sigma, St Louis, Missouri) were used for standard curve determination using HEK-Blue-mTLR5 and HEK-Blue-mTLR4, respectively. After 24 h of stimulation, we applied cell culture supernatant to QUANTI-Blue medium (Invivogen, San Diego, California) and measured alkaline phosphatase activity at 620 nm after 30 min.

AhR ligands quantification in milk samples

Levels of AhR ligands were measured in milk samples using HT-29-Lucia™ AhR reporter cells (Invivogen, San Diego, California). Briefly, milk samples diluted 1:5 in PBS were centrifuged at 8000 × g, and the supernatant was applied to HT-29-Lucia™ AhR reporter cells. Luminescence was immediately measured, and sample concentrations were calculated based on the luminescence measurements of HT-29-Lucia™ AhR reporter cells stimulated with a known amount of the AhR ligand FICZ.

Immunostaining of mucins and localization of bacteria by fluorescent in situ hybridization

Mucus immunostaining was paired with fluorescent in situ hybridization (FISH), as previously described^27,71, in order to analyze bacteria localization at the surface of the intestinal mucosa. In brief, colonic tissues (proximal colon, second cm from the caecum) containing fecal material were placed in methanol-Carnoy’s fixative solution (60% methanol, 30% chloroform, 10% glacial acetic acid) for a minimum of 3 h at room temperature. Tissues were then washed in methanol 2 × 30 min, ethanol 2 × 15 min, ethanol/xylene (1:1) 15 min, and xylene 2 × 15 min, followed by embedding in paraffin with a vertical orientation. Five-μm sections were cut and dewaxed by preheating at 60 °C for 10 min, followed by bathing in xylene at 60 °C for 10 min, xylene at room temperature for 10 min and 99.5% ethanol for 10 min. The hybridization step was performed at 50 °C overnight with an EUB338 probe (59-GCTGCCTCCCGTAGGAGT-39, with a 59 Alexa 647 label) diluted to a final concentration of 10 mg.mL⁻¹ in hybridization buffer (20 mM TrisHCl, pH 7.4, 0.9 M NaCl, 0.1% SDS, 20% formamide). After washing for 10 min in wash buffer (20 mM Tris-HCl, pH 7.4, 0.9 M NaCl) and 3 × 10 min in PBS, a PAP pen (Sigma, St. Louis, Missouri) was used to mark around the section and block solution (5% FBS in PBS) was added for 30 min at 4 °C. Mucin-2 primary antibody (rabbit H-300, [C3], C-term, Genetex, GTX100664) was diluted to 1:100 in block solution and applied overnight at 4 °C. After washing 3 × 10 min in PBS, block solution containing anti-rabbit Alexa 488 secondary antibody diluted to 1:300, PhalloidinTetramethylrhodamine B isothiocyanate (Sigma-Aldrich) at 1 mg.ml⁻¹ and Hoechst 33258 (Sigma-Aldrich) at 10 mg.ml⁻¹ was applied to the section for 2 h. After washing 3 × 10 min in PBS slides were mounted using Prolong anti-fade mounting media (Life Technologies) and kept in the dark at 4 °C. Observations and measurement of the distance between bacteria and epithelial cell monolayer were performed with a Spinning Disk IXplore using the Olympus cellSens imaging software 421 (V2.3) at a frame size of 2048 × 2048 with 16-bit depth. A 405 nm laser was used to excite the 422 Hoechst stain (epithelial DNA), 488 nm for Alexa Fluor 488 (mucus), 488 nm for Phalloidin (actin), 423 and 640 nm for Alexa Fluor 647 (bacteria). Samples were imaged with a 20× objective. For quantification, three power fields per mouse were used to measure the distance between the 15 closest bacteria and the epithelial lining (for a total of 45 measurements per mouse).

Maternal treatment prediction based on offspring microbiota composition

The association between offspring microbiota composition and maternal treatment was assessed trough the prediction of maternal treatment based on offspring’s microbiota composition data. Prediction of CMC treatment (outcome CMC or Water, Fig. S2a–d), P80 treatment (outcome P80 or Water, Fig. S2e–h) was performed by computing receiver operating characteristic (ROC) curves (R version 4.1.2, randomForest 4.7-1.1 package, ROCR package) using training data set and validation data set containing randomly affected 80% and 20% of mice, respectively. Data set contained relative abundance data for microbiota members identified at the species level at weeks 4, 8, 12, and 16 of age. ROC calculation was repeated 50 times with random sampling of the training and validation data and area under curve (AUC) measurement for each iteration. Mean AUC and standard deviation are presented for each graph.

Statistical analysis

Significance was determined using, when normality and homoscedasticity postulates were valid, one-way group analysis of variance (ANOVA) with Sidak’s multiple comparisons test, or t-test when only two groups were involved. Significance of data that did not respect normality and homoscedasticity postulates was tested using Kruskal–Wallis corrected for multiple comparisons with a Dunn’s test or Brown–Forsythe and Welch ANOVA corrected for multiple comparisons with a Dunnett test, respectively. Significance of longitudinally-measured data was assessed using two-way ANOVA corrected for multiple comparisons with a Sidak’s test. Clustering significance in PCoA plot was determined using a ANOSIM analysis of similarities test. Differences were noted as significant *p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001; **** p ≤ 0.0001; n.s. indicates non-significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conception and design; analysis and interpretation of data; writing, review, and/or revision of the manuscript: C.D., B.C. Development of methodology: C.D., H.R., E.B., B.C. Acquisition of data: C.D., H.R., E.B., M.H., S.E.S., E.C., F.H., A.P., B.C.

Peer review

Peer review information

Nature Communications thanks Thomas Gensollen and Lindsay Hall for their contribution to the peer review of this work. A peer review file is available.

Data availability

Unprocessed sequencing data from both 16S rRNA sequencing and RNA-seq are deposited in the European Nucleotide Archive under accession number PRJEB93825 and PRJEB93824, respectively. All other data are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-62397-3.