Early-life microbiota disruption-induced deficits in the social brain are sensitive to diet

Summary

Diet is one of the major modulators of the microbiota-gut-brain axis across the lifespan. Milk bioactive components, including human milk oligosaccharides such as fucosyllactose and sialyllactose, and prebiotics, including GOS and FOS, promote the viability of commensal bacteria, fortify the intestinal barrier, and improve cognitive development. Here, we investigate the ability of these dietary components alone or in combination to counter the behavioral and physiological effects of early-life microbiota depletion via broad-spectrum antibiotics in mice. Microbiota depletion impaired social recognition in juvenile mice, which was reversed by supplementation with human milk oligosaccharides, GOS/FOS, and their combination. Transcriptomic analysis in brain areas linked to social memory (amygdala and prefrontal cortex), revealed that pathways for central nervous system development, learning, learning and memory are sensitive only to the combined supplementation. Together, our data show that prebiotics and milk bioactive components exert beneficial effects on the host by reversing microbiota depletion-related deficits on the brain and behavior.

Graphical abstract

Highlights

• •HMOs and GOS/FOS rescue social behaviour deficits induced by early-life microbiota depletion
• •HMOs and GOS/FOS restored the abundance of microbiota enzymatic genes for carbohydrate utilization
• •HMOs and GOS/FOS restored myelination and learning and memory pathways in the juvenile mouse brain
• •Combining HMOs with GOS/FOS restores microbiota, brain and behaviour more effectively than alone

Neuroscience; Microbiology; Microbiome

Introduction

Starting at birth, colonization of the gastrointestinal tract with microbes begins and this microbial community accompanies us throughout our lifespan. There is increasing evidence from both preclinical and human studies supporting the connection between the microbes in the gut with brain development and behavior.^1,2,3,4 This communication pathway between gut microbes and the brain is reciprocal and is defined as the microbiota-gut-brain axis.⁵ Emerging data suggest that microbiota disruption during early life, for instance, via antibiotic (ABX) administration, has major and long-lasting implications on intestinal microbes, brain development, and behavior throughout life.^6,7,8,9,10 Social behavior and social circuits in the brain are sensitive to antibiotic-induced microbiota depletion.^9,11 The early postnatal period is of particular interest for brain, behavior, and gut microbiota development^12,13 as it is characterized by increased plasticity for the brain and gut microbes. During this period, there is a constant stream of nutrients and microbes from the mother via the breastmilk to the infant.¹⁴

Breastfeeding is the gold standard for infant nutrition as it is tailored to provide various micro- and macronutrients to the rapidly developing infant.^15,16 Among those nutrients are human milk oligosaccharides (HMOs), indigestible glycans that pass through the gastrointestinal tract. More than 200 separate types of HMOs have been identified in maternal milk with protective effects on the infant. Upon reaching the intestines, HMOs serve as prebiotics promoting microbial growth in a strain-specific manner as they are fermented by members of the gut microbiota.¹⁷ They are also known to induce the maturation of epithelial cells and improve gut barrier function, protecting the infant gut from infections and promoting the viability and diversity of commensals^18,19 while inhibiting the blooming of pathogenic bacteria.^20,21

In the brain, HMOs improve neurodevelopment and cognition in both human²² and rodent studies.^23,24,25 HMO supplementation with either fucosyllactose (FL) or sialyllactose (SL) increased sialic acid concentration, relevant for ganglioside formation and myelination,^26,27 improved learning and memory formation^25,28 and enhanced cognition all the way through adulthood in rodent models.²⁴ In humans, a link has been identified between 2-fucosyllactose consumption, breastfeeding frequency, and cognitive development in the first month of life of breastfed infants.²⁹ Therefore, HMOs benefit the brain and behavior by promoting the growth of beneficial bacteria and by serving as a source of sialic acid.

In addition, prebiotic approaches have received much attention as a means to target the microbiome.^30,31 Galacto-oligosaccharides (GOS) and fructo-oligosaccharides (FOS) are non-digestible oligosaccharides with high prebiotic potential and supplementation with GOS and FOS in early life led to increased abundance of commensals Bifidobacterium^32,33 and Lactobacillus.³² In contrast to HMOs, GOS and FOS are efficiently metabolized by commensals such as Bifidobacterium.³⁴ Among the benefits of a microbiota composition rich in Bifidobacteria is the enzymatic degradation of non-digestible carbohydrates, the subsequent production of metabolites short-chain fatty acids³⁵ and neurotransmitters^36,37 both of which are relevant for brain homeostasis and crucial for appropriate CNS function.^38,39,40

The early life is a pivotal time frame for microbiota and brain development and function. Critical windows of plasticity coincide for both microbiota maturation and brain development.⁴¹ More specifically, throughout early life, the microbiota slowly matures and adapts to nutrient changes, while neurodevelopmental disorders such as autism spectrum disorder (ASD) and attention deficit/hyperactivity disorder (ADHD) arise. It is now recognized that microbiota signals during early life are essential for the precise calibration along the gut-brain axis.^7,14,41 Potential disturbances of the microbiota during critical windows might interfere with normal brain development and subsequent function later in life. Microbiota disturbances, including ABX-induced microbiota depletion during critical windows, change the composition and function of the microbes, along with brain function.⁷ Targeting the gut microbiota through prebiotic supplementation might ameliorate the negative effects of microbiota depletion on brain and behavior via the microbiota-gut-brain axis during critical developmental windows.

While the relationship of gut microbiota and brain development is recognized as crucial, the ability of HMOs alone or in combination with prebiotics to act as mediators of brain development and behavior remains unclear, especially in the context of early life microbiota perturbations. To this end we studied the effects of a mixture of HMOs (2′- and 3′-fucosyllactose and 3′- and 6′-sialylactose) alone or in combination with a mixture of short chain GOS and long chain FOS (from P3 until P23) after early-life antibiotic-induced microbiota depletion (P3-P7) and investigated the gut microbiota composition and function along with social behavior and brain transcriptomics in juvenile mice at P23.

Results

Human milk oligosaccharides and/or galacto-oligosaccharides/fructo-oligosaccharides ameliorated the social behavior deficits in juveniles induced by early-life microbiota depletion

Social behavior impairments are common features among a variety of neuropsychiatric conditions, including ASD. Microbiota-deficient mice develop social deficits.⁴² Here, we use the three-chamber social interaction test to assess the social behavior in juvenile mice. Microbiota depletion by ABX administration in early life results in significantly decreased social recognition during juvenile period (t₁₀ = −0.723, p = 0.486), that is restored by HMOs (t₇ = −2.422, p = 0.046), GOS/FOS (t₉ = −2.815, p = 0.02) and HMOs+GOS/FOS (t₈ = −4.276, p = 0.003) treatment (Figure 1B).

Figure 1.

HMOs and GOS/FOS alone or in combination improve social behavior and alter microbiota composition and function

(A) Study design and treatment groups.

(B) Interaction time in social novelty as part of the three-chambers social interaction test at P22 (F: familiar conspecific mouse, N: novel conspecific mouse. Data were analyzed with paired-sample t test, presented here with individual values as mean+−SEM).

(C) PCA plot of beta diversity of caecum microbiota at P23: p < 0.001 Control Vehicle vs. ABX Vehicle.

(D) Alpha-diversity in caecum microbiota at P23: Chao, Shannon, and Simpson indices: p < 0.001 Control Vehicle vs. ABX Vehicle.

(E) Stacked bar plot of the 10 most abundant families present in the caecum microbiota at P23.

(F) Significantly restored taxa in caecum microbiota at P23 in animals that received HMOs, GOS/FOS, or HMOs+GOS/FOS.

(G) Abundance of enzymes in the genome of microbes from caecum at P23: Z-scores of mean abundance ###p < 0.001 ABX Vehicle vs. Control Vehicle and ∗p < 0.2 ABX HMOs+GPS/FOS vs. ABX Vehicle. n = 8–10 per treatment, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Ultrasonic vocalizations produced by rodents are consistent and robust phenomena used as an index of social interest, motivation, and emotional development.⁴³ In mouse models for ASD, such as the BTBR mouse, USV patterns are impaired, emitting significantly higher number of calls than the C57/Bl6 mice.^44,45 In our data, ABX administration and treatment with HMOs and/or GOS/FOS did not impact the number of calls (Figure S1A).

The maternal attachment test is used to evaluate the recognition ability of the maternal bedding. Animals with perturbed gut microbiota display reduced time spent in the maternal bedding compared with a neutral bedding.⁴⁶ In our study, there was a slight decrease on the time spent in the maternal bedding in ABX-treated animals, but this effect was not significant (Figure S1B). Treatment with HMOs and/or GOS/FOS did not impact the time spent in the maternal bedding (Figure S1B).

Human milk oligosaccharides and galacto-oligosaccharides/fructo-oligosaccharides selectively altered gut microbiota abundance in antibiotic-treated animals

Broad-spectrum ABX administration in early life led to altered microbiota composition at P23 (Figures 1C and 1D). Beta diversity, which describes the variability in microbiota composition, was significantly altered among groups (pseudo-F_8,68 = 2.883, p = 0.0001). Alpha diversity indices, including Chao1, Shannon, and Simpson, were significantly altered among groups (p < 0.001). Chao1 index describes the adjusted richness of the microbiota composition, which was significantly altered among groups (F_4,40 = 27.145, p < 0.001) with ABX decreasing the overall richness of the microbiota community at P23 (Control Vehicle vs. ABX Vehicle: p_adj < 0.001). Additionally, microbiota depletion via ABX changed the Shannon and Simpson indices, describing the richness and evenness of the community (Shannon: F_4,40 = 9.486, p < 0.001 and Simpson: F_4,40 = 11.088, p < 0.001). However, the detrimental effect of early-life ABX on the composition was not altered by HMOs, GOS/FOS, or HMOs+GOS/FOS (Figures 1C and 1D, for statistics see Table S1), possibly due to the short time between the ABX treatment and sampling and the short-term intervention duration of HMOs, GOS/FOS, or HMOs+GOS/FOS.

However, the abundance of specific taxa was restored by HMOs, GOS/FOS, or the combination of them (Figure 1F). Treatment with HMOs alone restored the abundance of ASV738, Lachnospiraceae (Unclassified ASV72, ASV102, ASV253, and ASV374), Muribaculaceae (Unclassified ASV46, ASV258, and ASV275), and Anaerovoracaceae families. GOS/FOS alone restored the abundance of Lachnospiraceae (Unclassified ASV77 and ASV111), Anaerovoracaceae, Ruminococcaceae, and Muribaculaceae (Unclassified ASV258). The combination of HMOs+GOS/FOS had the biggest effect on the bacteriome in the juvenile caecum (Figures 1E and 1F). HMOs+GOS/FOS restored the abundance of families Bacteroides, Lachnospiraceae (Unclassified ASV69, ASV77, ASV253, and NK4A136 ASV111), Ruminococcaceae, Anaerovoracaceae, and Muribaculaceae (Unclassified ASV258 and ASV275) in the gut compared with ABX-treated animals at P23.

Human milk oligosaccharides and galacto-oligosaccharides/fructo-oligosaccharides restored the abundance of N4-(β-N-acetylglucosaminyl)-L-asparaginase in the microbiota

Microbes utilize oligosaccharides via enzymatic breakdown. We further interrogated the inferred genome of microbes present in the community for enzyme-encoding genes that are relevant for HMOs, GOS, and FOS utilization. The inferred abundance of the enzyme N4-(β-N-acetylglucosaminyl)-L-asparaginase (EC 3.5.1.26), encoded by the NagZ gene in bacteria, decreased significantly in response to early-life ABX (ABX Vehicle vs. Control Vehicle β estimate = 2.35, p < 0.001) (Figure 1G). The abundance of this enzyme increased significantly only by the combination of HMOs with GOS/FOS treatment (ABX HMOs+GOS/FOS vs. ABX β estimate = −1.20, p_adj = 0.110) while treatment with HMOs or GOS/FOS alone did not alter the abundance of this enzyme in the microbiota at P23 (Figure 1G).

Human milk oligosaccharides and/or galacto-oligosaccharides/fructo-oligosaccharides restored pathways involved in myelination, CNS development, and learning and memory in the juvenile amygdala

Amygdala is a brain area relevant for social behavior, as it is the center for emotional processing.^39,47 The overall composition of genes in P23 amygdala was altered by early-life ABX administration (Figure 2A) (Control Vehicle vs. ABX vs. ABX+HMOs: F_2,23 = 1.4983, p = 0.005, Control Vehicle vs. ABX vs. ABX+GOS/FOS: F_2,25 = 1.4693, p = 0.0016, Control Vehicle vs. ABX vs. ABX+HMOs+GOS/FOS: F_2,24 = 1.6693, p = 0.0001). There is a trend for GOS/FOS and HMOs+GOS/FOS to altered transcriptomic profile in the P23 amygdala (p = 0.053 and p = 0.079, respectively) compared with ABX Vehicle, but not HMOs (p = 0.339).

Figure 2.

HMOs, GOS/FOS, and HMOs+GOS/FOS restoration of gene expression in pathways relevant to learning and memory and CNS development in P23 Amygdala in response to ABX-induced microbiota depletion

(A) PCA plots of overall gene expression in P23 amygdala in response to HMOs, GOS/FOS, and HMOs+GOS/FOS compared to ABX Vehicle: in each of the three PCA plots, the comparisons are ABX Vehicle vs. Control Vehicle and ABX Vehicle vs. ABX Treatment (HMOs, GOS/FOS or HMOs+GOS/FOS).

(B) Pathway enrichment analysis of genes whose expression is restored in response to ABX HMOs, ABX GOS/FOS, or ABX HMOs+GOS/FOS compared with ABX Vehicle in amygdala at P23: Significance (p < 0.05) is indicated with a gray circle, and the magnitude of the enrichment of each pathway is indicated by the enrichment ratio.

(C) Heatmaps of the significantly altered genes (p < 0.05 Control Vehicle vs. ABX Vehicle and p < 0.05 ABX Vehicle vs. ABX+<Treatment>) in P23 amygdala. n = 8–10 per treatment.

When investigating the transcriptomic profile in the amygdala, we found that 2548 genes were restored by HMOs, 2430 were restored by GOS/FOS, and 1640 were restored by HMOs+GOS/FOS treatment. Pathway enrichment analysis revealed that pathways involved in learning and memory, CNS development, glutamate reuptake, and establishment of BBB were significantly enriched in animals that received any of the three treatments compared with ABX-treated animals (Figure 2B). However, the combination of HMOs with GOS/FOS displayed a slightly higher ratio in those pathways compared with the HMOs and GOS/FOS alone within the ABX-treated animals (Figure 2B). Additionally, pathways relevant for myelination, such as oligodendrocyte differentiation, ensheathment of neurons, and structural constituents of myelin sheath were significantly altered in response to HMOs, GOS/FOS, and HMOs+GOS/FOS compared with ABX Vehicle animals (Figure 2B). However, ganglioside metabolism, which is also relevant for myelination, was significantly enriched by GOS/FOS and HMOs+GOS/FOS but not HMOs alone. Other pathways of interest that were significantly enriched for any of the prebiotic treatments were neurogenesis, neuron projection morphogenesis, and response to vasopressin (Figure 2B).

Human milk oligosaccharides and galacto-oligosaccharides/fructo-oligosaccharides restored pathways involved in CNS development and learning and memory in the juvenile prefrontal cortex

The PFC is an area implicated in social recognition memory, social motivation, decision making, and is a crucial brain area for social behavior. The overall composition of genes in P23 PFC was not altered by ABX or any of the three prebiotic treatments (Figure 3A) (Control Vehicle vs. ABX vs. ABX+HMOs: F_2,24 = 1.065, p = 0.1466, Control Vehicle vs. ABX vs. ABX+GOS/FOS: F_2,25 = 0.993, p = 0.4448, Control Vehicle vs. ABX vs. ABX+HMOs+GOS/FOS: F_2,24 = 0.9942, p = 0.3902).

Figure 3.

HMOs, GOS/FOS, and HMOs+GOS/FOS restoration of gene expression in pathways relevant to learning and memory and CNS development in P23 PFC in response to ABX-induced microbiota depletion

(A) PCA plots of overall gene expression in P23 PFC in response to HMOs, GOS/FOS, and HMOs+GOS/FOS compared to ABX Vehicle: in each of the three PCA plots, the comparisons are ABX Vehicle vs. Control Vehicle and ABX Vehicle vs. ABX Treatment (HMOs, GOS/FOS or HMOs+GOS/FOS).

(B) Pathway enrichment analysis of genes whose expression is restored in response to ABX HMOs, ABX GOS/FOS, or ABX HMOs+GOS/FOS compared with ABX Vehicle in the PFC at P23: Significance (p < 0.05) is indicated with a gray circle, and the magnitude of the enrichment of each pathway is indicated by the enrichment ratio. C. Heatmaps of the significantly altered genes (p < 0.05 Control Vehicle vs. ABX Vehicle and p < 0.05 ABX Vehicle vs. ABX+<Treatment>) in P23 PFC. n = 8–10 per treatment.

However, closer investigation revealed that microbiota depletion by ABX changed the expression of 2344 genes in the PFC, which was restored by each of the treatments; 982 were restored by HMOs, 1657 by GOS/FOS, and 1811 by HMOs+GOS/FOS in the P23 PFC. Pathway enrichment analysis revealed that learning and memory, CNS development, and cytokine production were significantly upregulated only in the PFC of animals that received HMOs+GOS/FOS (Figure 3B). Other pathways of interest that are enriched in PFC in response to any of the prebiotic supplementation is fatty acid catabolism (Figure 3B). Interestingly, SCFAs acid metabolism was only significantly enriched by GOS/FOS treatment in the PFC (Figure 3B).

Discussion

Investigating the effects of bioactive components of breast milk, such as HMOs, alone or in combination with prebiotics, is of great importance in order to understand the mode of action of those molecules for brain development and to facilitate more insights to guide improvements in infant nutrition. Recent studies show growing interest in the prebiotic effect of HMOs and GOS/FOS, via which these compounds, at least partly, exert their benefits on brain development throughout the lifespan.^23,26,48,49 In the current study, we report the positive effects of HMOs, GOS/FOS, and their combination on social behavior and brain transcriptomics via the host gut microbiota function following early-life microbiota depletion.

As expected, we observed widespread changes induced by antibiotics on microbiota composition and function. Exposure to a broad-spectrum antibiotic cocktail altered the overall abundance of microbes with implications for community dynamics within the gut microbiota during a critical developmental window. While our antibiotic model was not intended to replicate a clinical scenario, the profound microbiota depletion it induces, serves as a tool to explore the capacity of prebiotics to influence host outcomes under conditions of extreme microbial disruption. Treatment with HMOs and GOS/FOS selectively restored the abundance of the Muribaculaceae family, which has been previously found to protect the gut epithelium in the context of colitis.⁵⁰ Moreover, our result regarding the decreased relative abundance of Bacteroides in response to ABX is also supported by literature.^51,52 The combination of HMOs with GOS/FOS was able to restore the abundance of Bacteroides toward control levels. However, neither GOS/FOS nor HMOs alone restored Bacteroides abundance per se, which is perhaps surprising considering that this microbe utilizes, among other substrates, HMOs isolated from human milk in vitro and, in vivo, in adulthood.⁵³ Interestingly, an infant cohort revealed that a Bacteroidetes-dominant microbiota was associated with enhanced cognition and language performance for up to 2 years in male infants.⁵⁴ Therefore, the restored abundance of this microbe by the combination of prebiotics might be beneficial for signaling across the gut-brain axis.

The hormonal and biological alterations during puberty are accompanied by psychological and social transformation during this period. This period of life is characterized by increasing need to spend time with peers, with heightened attention to acceptance, rejection, and approval from peers.^55,56,57 This transformative time window is characterized by the establishment of social self-identity and the development of higher-level cognitive processes, such as understanding the reciprocity of social interactions in humans.⁵⁸ During the juvenile period, rodents display a peak in motivation for social interaction and play.^59,60 In the present study, ABX-induced microbiota depletion in early life results in impaired social behavior, and this effect was ameliorated by HMOs, GOS/FOS, and HMOs+GOS/FOS supplementation in male mice. In agreement with our study, it has previously been shown that supplementation with a prebiotic mixture containing GOS during early post-natal development enhanced positive social interactions during late adolescence compared with controls.⁶¹ In a more recent animal study, male mice that received GOS/FOS supplementation since birth displayed improved social behavior compared with controls in adulthood.⁶² Regarding HMOs, higher levels of HMO 6′-SL in breast milk were associated with better social outcomes and increased myelination in areas involved in social behavior in infants at 12 months of age, while levels of 3-FL were associated with improved language skills.⁶³ Notably, in our study the combination of HMOs+GOS/FOS has reversed the effect of antibiotics on social behavior more significantly compared to when HMOs or GOS/FOS are administered alone. This interesting effect might shed light on the additive effect of bioactive components of breast milk with non-digestible prebiotics on the brain and behavior.

Various studies have unmasked links between multiple brain areas implicated in social behavior.^39,47 Among these areas, connections between amygdala and PFC are crucial for social recognition memory.⁶⁴ In our study, pathway enrichment analysis revealed a common pathway that was significantly upregulated in both amygdala and PFC, termed “Learning and Memory” in response to HMOs+GOS/FOS treatment compared with ABX in juvenile male mice. Moreover, the pathway relevant for response to vasopressin was upregulated in the amygdala receiving each of the treatments compared with ABX Vehicle. Vasopressin is a neuropeptide involved in social behavior that is expressed, among other regions, by neurons in amygdala.⁶⁵ Each of the prebiotic treatments ameliorated the social deficits induced by early-life ABX in the social novelty, revealing that social recognition memory was enhanced in those animals that received prebiotics. This conservation of pathways relevant for brain development and learning and memory in both PFC and amygdala might be linked with the enhanced social recognition memory in the animals that received the combination of HMOs+GOS/FOS. Even though this pathway was not significantly altered in the PFC following supplementation with HMOs or GOS/FOS alone, the significance persisted in the amygdala. Moreover, the behavior of those animals supports the interpretation that recognition memory was enhanced in response to HMOs or GOS/FOS compared with control male mice at P22.

In terms of brain anatomy, the social brain, which includes the PFC and amygdala,⁶⁶ is going through continuous transformations/rewiring from childhood all the way to adolescence. The macrostructural changes in the PFC correspond to neurodevelopmental mechanisms at the microstructural level, with myelination, axonal growth, and synaptic refinement taking a central role during this period.⁶⁷ Pathways relevant to neuron projection development and myelination were significantly enriched in the amygdala of the juvenile offspring in response to HMOs, GOS/FOS, or HMOs+GOS/FOS. We therefore suggest that changes in gene expression patterns in the amygdala and PFC underlie the improvement of social behavior in prebiotic-treated juvenile mice compared with ABX-treated controls.

Prebiotics GOS and FOS and HMOs support gut health and modulate intestinal microbiota composition,^68,69 with increased levels of Bifidobacterium and decreased levels of Clostridium in the gut.⁷⁰ However, in our study, we did not find significant differences in Bifidobacterium and Clostridium abundance among groups. This result might be due to the widespread effect of antibiotics on the gut, and the intervention timing with the prebiotics might have been too narrow to observe such differences. Additionally, microbiota dynamics regarding metabolic activity, utilization of prebiotics, and cross-species interactions might explain the differences seen in our study and others. At this point, it is important to acknowledge that the timepoint of assessing this effect (P23) coincides with the solid food introduction phase and the weaning of animals to independent cages from the mother, and, from the litter. This alone has an effect on microbiota composition, adjusting to a new norm of solid food.⁷¹

The functions of the metagenomes identified that the abundance of N4-(β-N-acetylglucosaminyl)-L-asparaginase was significantly decreased in response to antibiotics, while only treatment with HMOs+GOS/FOS reversed this effect. However, neither the HMOs nor the GOS/FOS administration alone was able to increase the abundance of this enzyme in microbial genomes, revealing a possible interaction effect of the HMOs+GOS/FOS with the gut microbiota. N4-(β-N-acetylglucosaminyl)-L-asparaginase is catalyzing the hydrolysis of N-acetylglucosamine (GlcNAc) and asparagine residues within glycoproteins and is active in lysosomes. Considering that the combination of HMOs with GOS/FOS had a stronger effect on modulating the gut microbiota composition compared with HMOs or GOS/FOS alone, it is reassuring that microbiota enzymatic function is also highly restored only by the combination of HMOs with GOS/FOS. It is therefore evident that the combination of HMOs+GOS/FOS in early life has stronger benefits for the microbiota community and function in juveniles compared to when those prebiotics are administered alone. The increased inferred enzymatic gene expression of N4-(β-N-acetylglucosaminyl)-L-asparaginase could increase the levels of peptidoglycan fragments in the gut of mice that are treated with HMOs+GOS/FOS. Peptidoglycans originating from gut microbes can be transported to the brain, sensed by pattern-recognition receptors (PRRs),^72,73 and have been suggested to alter long-term social behavior and brain gene expression patterns.⁷⁴ We hypothesize that the combination of HMOs with GOS/FOS may enhance the microbial expression of N4-(β-N-acetylglucosaminyl)-L-asparaginase, potentially increasing the production of peptidoglycan fragments that could influence long-term social behavior through PRR-mediated neuroimmune mechanisms.

There is multiple evidence from in vitro and in vivo studies illuminating the mechanisms via which these prebiotic oligosaccharides exert the benefits in the central nervous system through the microbiota-gut-brain axis. For example, the by-products of microbiota fermentation of GOS and FOS have immune,⁷⁵ neuronal and endocrine potential,⁷⁶ signaling from the gut to the periphery and the brain. GOS supplementation in early life resulted in altered BDNF levels and synaptic protein expression in adult animals, suggesting that neonatal GOS supplementation is capable of manipulating gut microbiota in early life and its positive effects on the brain persist at least up to young adulthood in rodents.⁴⁹

In conclusion, our study highlights the power of dietary manipulations in early life to counter the effects of microbiota disturbance-induced changes in brain and behavior. Specifically, we show that HMOs, GOS/FOS, and HMOs+GOS/FOS supplementation may be a useful strategy to bolster microbiota-gut brain axis function in early life with positive knock-on effects on the social brain. Together, our data show that studies with prebiotics and milk bioactive components in humans are now warranted and highlight the importance of critical windows in driving behavioral changes across the microbiota-gut-brain axis.

Limitations of the study

The current study investigated the effects of HMOs and GOS/FOS in male gut microbiota, brain and behavior. Males were chosen due to the fact that the majority of effects of microbiota manipulations on the social brain are more robust in those animals.⁷⁷ However, evidence shows that there are sex-dependent effects of diet on the gut microbiota and brain development along with behavioral outputs, due to hormonal and physiological changes in juveniles and future studies in females should also be carried out.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Professor John F. Cryan at j.cryan@ucc.ie.

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Metagenomics (https://www.ebi.ac.uk/ena/browser/view/PRJEB94443) and transcriptomic data (https://zenodo.org/records/17396065) have been deposited at the European Nucleotide Archive and Zenodo and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.
• •All original code has been deposited at Zenodo and is publicly available. The URL (https://zenodo.org/records/17396065) is listed in the key resources table.
• •Any additional information required to reanalyze the data reported in this article is available from j.cryan@ucc.ie upon request.

Acknowledgments

We wish to thank G. Moloney, P. Fitzgerald, C. Manley, J. O’Reilly for the technical assistance throughout the study. This work was supported by a research collaboration with Danone Global Research & Innovation Centre. Prof. Cryan is funded by the Science Foundation Ireland (SFI/12/RC/2273_P2), Saks Kavanaugh Foundation and Swiss National Science Foundation project CRSII5_186346/NMS2068.

Author contributions

Conceptualization, A.R., C.A.B., I.B.R., J.K., E.G., G.K., and J.F.C.; data curation: A.R., C.A.B., B.V., and T.F.S.B.; formal analysis, A.R., C.A.B., B.V., T.F.S.B., and B.L.S.; funding acquisition: I.B.R., J.K., and J.F.C.; investigation: A.R., C.A.B., I.B.R., J.K., E.G., G.K., and J.F.C.; methodology: A.R., C.A.B., B.V., T.F.S.B., B.L.S., I.B.R., J.K., E.G., G.K., and J.F.C.; project administration, A.R., C.A.B., E.G., and J.F.C.; resources: A.R., C.A.B., B.V., T.F.S.B., I.B.R., J.K., and E.G.; software, A.R., C.A.B., B.V., T.F.S.B., and B.L.S.; supervision, A.R., C.A.B., I.B.R., J.K., E.G., G.K., and J.F.C.; validation, A.R., C.A.B., B.V., T.F.S.B., B.L.S. I.B.R., J.K., E.G., G.K., and J.F.C.; visualization, A.R., B.V., and T.F.S.B.; writing – original draft, A.R. and J.F.C.; writing – review and editing, A.R., B.V., T.F.S.B., I.B.R., J.K., E.G., G.K., and J.F.C.

Declaration of interests

The work in the article “Early-life Microbiota Disruption-Induced Deficits in the Social Brain are Sensitive to Diet” was funded by a research grant from Danone to J.F.C. J.F.C. has received research funding from IFF, Reckitt, and Danone Nutricia. J.K. and I.B.R. are employed by Danone Global Research & Innovation Center. G.C. has received honoraria from Janssen, Probi, Apsen, and Ingelheim Boehringer as an invited speaker; is in receipt of research funding from Pharmavite, Fonterra, Reckitt, Nestle, and Tate and Lyle; and has been paid for consultancy work by Yakult, Zentiva, Bayer Healthcare, and Heel Pharmaceuticals. This support neither influenced nor constrained the contents of this article. A.R., C.A.B., B.V., T.F.B., B.L.S., and E.G. declare no conflict of interest.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Caecal content	This study	–
Amygdala	This study	–
PFC	This study	–
Chemicals, peptides, and recombinant proteins
Ampicillin Sodium Salt 69-52-3	Discovery Fine Chemicals	69-52-3
Gentamicin Sulfate	Discovery Fine Chemicals	1405-41-0
Vancomycin Hydrochloride	Discovery Fine Chemicals	1404-93-9
Imipenem Monohydrate	Discovery Fine Chemicals	74431-23-5
4HMO mix	Jennewein	4101
GOS	Friesland Campina	Vivinal® GOS Syrup
FOS	Beneo	Orafti®HP
Critical commercial assays
Qiagen QiaAMgenP DNA Fast Stool Kit	Qiagen	51504
mirVana RNA isolation kit	Invitrogen	AM1561
Nextera XT Index Kit	Illumina	15032354 and 15032355
Qubit dsDNA High Sensitivity Assay Kit	Thermo Fisher Scientific	Q33230
AMPure XP Beads for DNA Cleanup	Beckman Coulter	A63881
Deposited data
Raw gut microbiota data	European Nucleotide Archive	https://www.ebi.ac.uk/ena/browser/view/PRJEB94443
Gene expression count tables of amygdala and prefrontal cortex	Zenodo	https://zenodo.org/records/17396065
Experimental models: Organisms/strains
Mouse NIH Swiss	Envigo	035
Software and algorithms
Avisoft	Avisoft Bioacoustics, Berlin, Germany	https://avisoft.com
BORIS	Friard and Gamba78	https://www.boris.unito.it
R and RStudio	R Core Team, 2022; RStudio Team, 2022	https://www.r-project.org/and https://posit.co/
IBM, SPSS Statistics V29	Version 29.0; IBM Corp., 2022	https://www.ibm.com/spss
FastQC	Babraham Bioinformatic	https://www.bioinformatics.babraham.ac.uk/projects/fastqc/
DADA2	Callahan et al.79	https://benjjneb.github.io/dada2/
SILVA	Quast et al.80	https://www.arb-silva.de/documentation/release-128/
PICRUSt2	Douglas et al.81	https://github.com/picrust/picrust2
Tjazi	Bastiaanssen et al.82	https://github.com/thomazbastiaanssen/Tjazi
Kallisto	Bray et al.83	https://pachterlab.github.io/kallisto/
Vegan	Dixon84	https://cran.r-project.org/web/packages/vegan/index.html
Original code of metagenomics and transcriptomics analyses	Zenodo	https://zenodo.org/records/17396065
Other
Nanodrop Spectrophotometer	Nanodrop Technologies	RRID:SCR_016517
Purified rodent diet Control	Ssniff Spezialdiaten	S9262-E360 (custom-made)
Purified rodent diet 1.29% HMO mix	Ssniff Spezialdiaten	S9262-E367 (custom-made)
Purified rodent diet 6% GOS, 0.3% FOS	Ssniff Spezialdiaten	S9262-E364 (custom-made)
Purified rodent diet 3.755% GOS, 0.33%FOS, 1.279% HMO mix	Ssniff Spezialdiaten	S9262-E366 (custom-made)

Experimental model and study participant details

Studies in animals

NIH Swiss (HSD:NIHS) female and male breeders were purchased at 8 weeks old from Envigo, UK. After they were acclimated in the animal unit for 1 week in 33x15x13cm white cages with environmental enrichment (bedding, nesting material, cylindrical tube) and access to tap water and chow food ad libitum at 26 degrees Celsius in a 12-hour light, 12-hour dark cycle for all animals, independently of experimental treatment. Animals were bred for the first time (primiparous), with plug defining the beginning of gestation and the moment of separation of females from the male breeders into their own cage. After separation, the male breeders were sacrificed. Approximately 20 days after plug detection offspring was produced and each litter was randomly assigned to a treatment group. The offspring in each cage was administered either a broad-spectrum antibiotics (ABX) mixture (amcillin: 0.2mg/g, gentamicin: 0.2mg/g, vancomycin: 0.1mg/g and imipenem: 0.05mg/g) or tap water (Vehicle) via gavage once a day between 9-10am from postnatal day 3 (P3) to P7. A mixture of HMOs (2’-FL, 3’-FL, 3’-SL and 6’-SL) or GOS/FOS or a combination of HMOs+GOS/FOS was provided to the offspring via gavage once a day between 2-3pm from P3 until P20. The prebiotic mixture of short-chain galacto-oligosaccharides (scGOS) and long-chain fructo-oligossaccharides (lcFOS) (3% final concentration), scGOS/lcFOS were given in a 9 (sc-oligosaccharide): 1 (lc-oligosaccharide) ratio. Pups were gavaged either GOS/FOS (9:1) mixture alone, HMOs (0.94% final concentration) or a combination of HMOs (at 3.3 % final concentration) with GOS/FOS (9:1). From birth to PND20 prebiotic interventions were given in PBS. As of P21, animals were weaned and caged in groups of 2-3 animals per cage and prebiotics were given in the custom-made rodent diet (Ssniff Spezialdiaten GmbH, D-59494 Soest, Germany) ad libitum. Total fiber concentration of all diets was 5%. For specification on diets see key resources table and Table S2. We have therefore 5 experimental groups included in our study (Figure 1A): Control Vehicle (4 litters), ABX Vehicle (6 litters), ABX HMOs (3 litters), ABX GOS/FOS (6 litters) and ABX HMOs+GOS/FOS (4 litters). At P23, animals were weighed and euthanized by decapitation between 9am and 12pm. Caeca with content were weighed and collected, the right side from the prefrontal cortex (PFC) and amygdala were isolated and frozen on dry ice until further processing. Overall, 2-3 male animals per litter originating from 3-6 litters per treatment were used for this study, with a final n of 8-10 male animals per treatment for each experiment, unless otherwise specified (see behavioural testing in method details).

All procedures were conducted with approval from the Animal Experimentation Ethics Committee (AEEC) at University College Cork and the Health Products Regulatory Authority (HPRA) in accordance with the recommendations of the European Directive 2010/63/EU and authorization number AE19130/P087, prepared according to the ARRIVE 2.0 guidelines.

Method details

Behavioral testing

The short and long-term effects of microbiota depletion and treatment with either HMOs, GOS/FOS or HMOs+GOS/FOS intervention on behavior were evaluated in male offspring in early life (P9 and P10) (3-6 male animals per litter originating from 3-6 litters per treatment) and in juvenile period (P22) (2-3 males per litter originating from 3-6 litters per treatment). Behavioural tests were analysed by an independent experimenter blinded to experimental groups. All tests were performed during the lights on phase and between the hours of 9am and 2pm.

Isolation-induced ultrasonic vocalizations test

Isolation-induced ultrasonic vocalizations (USV) are produced by mouse pups during the first two weeks of life when separated from their mother and littermates.⁸⁵ USV phenotyping is a tool to identify behavioural patterns that could suggest early signs of neurodevelopmental disorder.⁴⁵ USVs was performed as described previously in.⁸⁶ Briefly, pups were isolated and placed into a clean plastic container (11x8.5x5.5cm) enclosed in a sound-attenuating chamber (44x34x24cm). Emission of USV calls were monitored by an ultrasound sensitive microphone suspended above the isolated pup for 3 min (for apparatus see Figure S1A). The number of calls was recorded and analysed by Avisoft software (Avisoft Bioacoustics, Berlin, Germany). The data presented included counts of calls from all of the male animals of each litter (n:15-18 per treatment originating from 3-6 litters per treatment). Some animals emitted large numbers of calls, failed the outlier test and removed from downstream data analysis.

Maternal attachment test

Maternal attachment test evaluates the ability of pups to differentiate their mother’s and littermates’ nest.⁸⁷ Maternal attachment test was performed as previously described in.⁸⁸ Briefly, at P10, the floor of a clean mouse cage (33x15x13cm) was subdivided into three equally sized areas by wire-mesh dividers. One area was uniformly covered with home cage bedding, thus containing familiar odour stimuli. The opposite area was covered with bedding from the cage of another litter (born at approximately the same time). The middle section was covered with clean bedding material. Pups were placed individually in the middle section for 1 min; the dividers were then removed, and the pups were allowed to freely explore the arena for 2 min. Total time spent in each area was recorded by a video camera mounted above the apparatus (for apparatus see Figure S1B). The videos were analyzed by semi-automated analysis using BORIS software.⁷⁸ The data presented included all of the male animals of each litter (n:14-26 per treatment originating from 3-6 litters per treatment).

Three-chamber social interaction test

The three-chamber social interaction test is widely used to evaluate social behaviour as previously described in.⁴² In brief, animals were placed in a 36x19x30cm rectangular light grey apparatus divided into three chambers (left and right chambers: 13x19x30cm each], and a smaller centre chamber: 9x19x30cm), small circular openings allowed easy access to all compartments. In total, there are three phases that last 10minutes each for a total of 30 minutes for the whole test:

• a.Habituation: animals are left to explore the apparatus and the empty wire meshes for 10 minutes.
• b.Sociability: animals are free to explore another sex-matched mouse conspecific or an inanimate rubber duckie under a wired mesh (circular, 9cm diameter) for 10 minutes.
• c.Social novelty: animals are free to explore a familiar or a novel sex-matched mouse conspecific under a wired mesh (circular, 9cm diameter) for 10 minutes. Social novelty recognition was assessed by interaction time recorded via a video camera mounted above the apparatus (for apparatus see Figure 1B). The social novelty index was calculated as described in,⁸⁹ where SNI= (Time with Novel – Time with Familiar) / (Time with Novel + Time with Familiar). Then 95% confidence intervals were calculated for each of the conditions of interest to assess the restoration potential of the prebiotic interventions on ABX-treated mice.

Total time spent in each area was recorded by a video camera mounted above the apparatus (for apparatus see Figure 1B). The videos were analyzed by semi-automated analysis using BORIS software.⁷⁸ The data presented included all of the male animals of each litter (n:8-11 per treatment originating from 3-6 litters per treatment).

Microbiota composition and enzyme abundance analysis

Microbial DNA was isolated from caecal content of P23 animals (n:8–10 per treatment originating from 3-6 litters per treatment) with Qiagen QiaAMgenP DNA Fast Stool Kit (51504, Qiagen, The Netherlands). Prior to library construction, the stool DNA samples are quantified using a Qubit Fluorometer (Thermo Fisher Scientific, USA) to ensure the template microbial DNA quality and quantity. At this stage, one sample from the ABX HMOs group was removed due to poor quality. A two-step PCR workflow is performed for library preparation according to the Illumina 16S sample preparation guide using the Ilumina MiSeq system (Illumina, USA). First stage PCR is performed to amplify the template from DNA sample using region of interest specific primers with overhang adapters attached. The gene-specific sequences used in this protocol target the 16S V3 and V4 region. Second stage PCR is performed to attach the Illumina sequencing adapters through PCR using Nextera XT Index Kit (Illumina, USA) according to the manufacturer’s protocol. PCR cleanup with AMPure XP beads (Beckman Coulter, USA) is then performed to purify the 16S V3 and V4 amplicon away from free primers and primer dimer species for each PCR process. FastQ files with raw sequences were loaded into R (version 4.1.1), where the Divisive Amplicon Denoising Algorithm (DADA2) version 1.22.0⁷⁹ was used to construct a count table of amplicon sequence variants (ASV) up to the genus level from n:8-10 animals per treatment originating from 3-6 litters per treatment. For each step, default parameters were used with the exception of the function filterAndTrim, where the argument maxEE was set as c(2,4), truncQ as 2 and trimLeft as 20. Taxonomic assignment was performed using the data base SILVA v138.⁸⁰ PICRUSt2 v2.4.1⁸¹ was used with its default parameters to infer the genomic content from 16S data in terms of Enzyme Commission (EC) numbers in stratified and unstratified mode.

Transcriptomics of amygdala and prefrontal cortex

Whole RNA from dissected P23 amygdala and PFC was isolated with the mirVana RNA isolation kit (Cat# AM1561, Invitrogen, Carlsbad, CA, USA) and maintained at -80°C until further processing (n:8-10 per treatment originating from 3-6 litters per treatment). Paired-end sequence reads were generated using the Illumina NovaSeq 6000. The sequences generated with the NovaSeq 6000 were performed under accreditation according to the scope of BaseClear B.V. (L457; NEN-EN-ISO/IEC 17025). FASTQ read sequence files were generated using bcl2fastq version 2.20 (Illumina). Initial quality assessment was based on data passing the Illumina Chastity filtering. Subsequently, reads containing PhiX control signal were removed using an in-house filtering protocol. In addition, reads containing (partial) adapters were clipped (up to a minimum read length of 50 bp). The second quality assessment was based on the remaining reads using the FASTQC quality control tool version 0.11.8. Transcripts were annotated using kallisto (v 0.48.0)⁸³ using the C57BL6 GRCm39 mouse reference genome using default parameters.

Quantification and statistical analysis

Microbiota biostatistical analysis

Statistical analysis was performed with R v4.2.0 and RStudio v2022.7.1.554. Alpha diversity metrices were calculated using the alpha function from the microbiome package v1.24.0. Statistical differences of alpha diversity metrics between experimental groups were calculated using ANOVA test from the R package rstatix v0.7.2. Principal component analysis (PCA) was performed using the prcomp function from the stats package in R, after performing a Centered Log Ratios (CLR) transformation over the count tables as implemented in the function decostand, from the vegan package v2.6.4⁸⁴ with a pseudocount of 2/3.⁹⁰ To test differences in beta diversity between groups, a PERMANOVA test was used as implemented in the adonis2 function from the vegan package, where 10.000 permutations were used. The distance metric applied was Aitchinson distance, as recommended for compositional data.⁹¹ Plots were generated using ggplot2 v3.4.4 and patchwork v1.1.3.

To assess the restorative effects of the prebiotic treatment (HMOs, FOS/GOS, HMOs+FOS/GOS) to animals exposed to ABX, ANOVA test was performed using the following formula: “ABX exposure ∗ Prebiotic treatment”, as implemented in the R package Tjazi v0.1.0.0.⁹² Subsequent pairwise comparisons were performed using Tukey test. In both tests, statistically significant results in the ANOVA were defined as Benjamini-Hochberg adjusted p values below 0.1. Bacterial taxa whose abundance was restored by prebiotics after antibiotic exposure were identified if the taxa met three criteria: (1) there is a difference between group means (i.e., statistically significant differences in the ANOVA), (2) the antibiotic exposure induced changes in the abundance of the taxa (i.e., statistically significant differences between the vehicle-fed antibiotic-exposed and antibiotics-free groups) and (3) antibiotic-exposed and antibiotic-free groups exposed to the same prebiotic show similar mean abundance (i.e., the difference between the group treated with antibiotics and any prebiotic and the group treated with the same prebiotic but not exposed to antibiotics is not statistically significant).

To assess the effects of prebiotic treatment on the abundance of enzymes after ABX exposure. GLMs were fitted, and the following formula applied: “Enzyme abundance ∼ Prebiotic treatment”. The control group was the ABX Vehicle. For visualization purposes the Control Vehicle group was added to visualize effects of the antibiotic interventions. Benjamini-Hochberg correction of p values was applied, and adjusted p values below 0.2 were considered significant for changes in the abundance of taxa and enzymes.

Brain transcriptomics biostatistical analysis

Further analysis of RNAseq data was handled in R (v 4.3.1). Gene transcription tables were filtered based on prevalence (50% threshold) and subsequently CLR-transformed, using the same 2/3 pseudocount imputation to deal with zeroes.⁹⁰

In order to determine whether a gene was restored to control levels, we used generalised linear models with a treatment-wise planned orthogonal contrast approach (Control Vehicle = 1/2, ABX Vehicle = -1, ABX + <Treatment> = 1/2). Briefly, genes were considered to be restored if 1) the pooled control vehicle and ABX + treatment groups together were different from the ABX Vehicle group and 2) additionally, the ABX Vehicle group was significantly different from Control Vehicle (post-hoc test). Principal component analysis was carried out in a treatment-wise fashion, using the R prcomp implementation in the “stats” library. Compositional differences were estimated using the adonis2 implementation of PERMANOVA.

Targeted enrichment analysis was performed by pre-defining GO terms of interest (astrocyte differentiation GO:0048708, CNS development GO:0007417, cytokine production GO:0001816, ensheathment of neurons GO:0007272, establishment of BBB GO:0060856, fatty acid catabolism GO:0009062, fucose metabolism GO:0006004, ganglioside metabolism GO:0001573, gliogenesis GO:0042063, glutamate metabolism GO:0006536, glutamate reuptake GO:0051935, learning and memory GO:0007611, myelin sheath GO:0043209, myelination GO:0042552, neurogenesis GO:0022008, neuron projection morphogenesis GO:0048812, oligodendrocyte differentiation GO:0048709, response to AVP GO:1904117, SCFAs biosynthesis GO:0051790, structural constituent of myelin sheath GO:0019911). Enrichment was assessed using hypergeometric tests (e.g., phyper() in the R stats package) on the subset of genes that were found to be restored on the p < 0.05 level. Figures involving were generated in R with ggplot2 (version 3.4.4).

Data analysis of non-omics data

All statistical analysis (apart from the caecal microbiota and RNA sequencing for amygdala and prefrontal cortex) was performed using IBM SPSS Statistics (IBM, SPSS Statistics V29). Before statistical analysis, data was analysed for normality using the Shapiro-Wilk test and homogeneity of variances using Levene’s test. For statistical analysis of parametric data, we used an independent sample Student t-test, while for non-parametric data we used the Mann-Witney U-test. For the behavioural results from three-chambers social interaction test we used repeated measures t-test. Data is shown as mean ±SEM. Statistical significance was set at p < 0.05. The number of animals in each experiment can be found in the figure legends and in the text throughout in the STAR Methods section.

Published: February 9, 2026