The early life exposome and autism risk: a role for the maternal microbiome?

ABSTRACT

Autism spectrum disorders (ASD) are highly heritable, heterogeneous neurodevelopmental disorders characterized by clinical presentation of atypical social, communicative, and repetitive behaviors. Over the past 25 years, hundreds of ASD risk genes have been identified. Many converge on key molecular pathways, from translational control to those regulating synaptic structure and function. Despite these advances, therapeutic approaches remain elusive. Emerging data unearthing the relationship between genetics, microbes, and immunity in ASD suggest an integrative physiology approach could be paramount to delivering therapeutic breakthroughs. Indeed, the advent of large-scale multi-OMIC data acquisition, analysis, and interpretation is yielding an increasingly mechanistic understanding of ASD and underlying risk factors, revealing how genetic susceptibility interacts with microbial genetics, metabolism, epigenetic (re)programming, and immunity to influence neurodevelopment and behavioral outcomes. It is now possible to foresee exciting advancements in the treatment of some forms of ASD that could markedly improve quality of life and productivity for autistic individuals. Here, we highlight recent work revealing how gene X maternal exposome interactions influence risk for ASD, with emphasis on the intrauterine environment and fetal neurodevelopment, host–microbe interactions, and the evolving therapeutic landscape for ASD.

Introduction

Autism spectrum disorders (ASD) are characterized by heterogeneous presentation of social, communication, behavioral, and cognitive deficits. In 2023, the Autism and Developmental Disabilities Monitoring (ADDM) Network estimated that 1 in 25 boys and 1 in 100 girls eight years of age (or 1 in 36 children) in the United States is autistic,¹ a marked increase from its most recent estimate of 1 in 44 children.² Despite its prevalence and identification of molecular-to-neural circuit-level mechanisms contributing to syndromic and other genetically defined ASD populations,^3,4 risk profiles for ASD remain poorly defined, most cases are of unknown etiology, and effective preventative and therapeutic measures remain limited.

The genetic landscape of ASD is heterogenous and multifaceted, encompassing (1) syndromic forms, such as that associated with Fragile X Syndrome,⁵ (2) single nucleotide polymorphisms like mutations in the gene encoding voltage gated sodium channel 1.2 (Na_v1.2), SCN2A,⁶ and (3) copy number variants (CNVs), such as 15q11.2q13 deletions or duplications.⁷ This genetic complexity is reflected in the wide range of features, symptoms, and severity which characterize ASD (the ‘spectrum’). Despite these relatively well-characterized disorders, the underlying cause of most cases remains completely unknown. Given the high heritability of ASD, most studies have rationally focused on understanding how genetic variants contribute to disease risk and the underlying pathology. However, it is increasingly evident that environmental factors can modify ASD risk.^8,9 The concept that early-life environmental exposures that modify gene expression through epigenetic reprogramming could result in a phenotype indistinguishable from that of genomic variants is gaining traction, as the burgeoning field of exposomics attempts to capture how the full array of environmental exposures experienced from fetal development onward contribute to disease risk and outcomes, even across generations.

The “exposome,” a term coined by cancer biologist Christopher P. Wild in 2005,¹⁰ encompasses the myriad environmental factors to which an individual is exposed throughout life. Notably, a growing body of evidence highlights the role of the early life exposome in setting the stage for disease risk later in life. Beyond cancer, the interplay between genetics and the exposome appears to be particularly relevant to neurodevelopmental disorders, which typically emerge during early childhood.¹¹ Detrimental events, such as in utero exposure to valproic acid (VPA),¹² occurring during developmental critical periods can disrupt highly orchestrated changes in gene expression and thereby affect the formation of behaviorally relevant neural circuits. While investigation into the relationship between the exposome and neurodevelopmental health is in its infancy, exposure science is on the radar of major biomedical funding agencies, including the National Institute of Environmental Health Science (NIEHS).

Most environmental exposures that influence early life neurodevelopment likewise act on the maternal and offspring gut microbiome, as well as the host immune system.^13–15 Microbiota-immune interactions occur both in utero and during early-postnatal life, underscoring the importance of the maternal exposome in offspring risk for neurodevelopmental disorders. Maternal diet, infections, and medications can significantly alter the microbiome and the maternal inflammatory response. Each has been linked to risk for neurodevelopmental disorders in both preclinical and clinical studies.¹⁶

Multi-OMICs-based studies are beginning to unravel the complex relationship between the microbiome and its associated metabolome with the host immunome, epigenome, and, ultimately, the tissue-specific transcriptome. Intriguingly, some preclinical studies exploit a key feature of environmentally driven epigenetic alterations: their reversibility.^17,18 Alongside host genome-targeted approaches,¹⁹ modulation of the gut microbiome and the immune system during early development are being explored as innovative strategies for ASD. Here, we review recent evidence for complex genome–metagenome interactions in ASD and how early life environmental exposures, particularly those that affect the maternal and intrauterine environments, contribute to risk for ASD by influencing development of the brain and the immune system, and their interactions.

Autism spectrum disorder: a complex condition with increasing prevalence

Autism spectrum disorder (ASD) is characterized by presentation of persistent deficits in communication and social interaction as well as restricted and repetitive pattern of interests or activities which lead to significant impairment in social and occupational functioning.²⁰ The core features of ASD often emerge in early development, but the age of diagnosis varies. ‘Spectrum’ reflects the heterogeneous clinical manifestation of ASD, which can differ according to severity of the disorder as well as the developmental stage, biological sex, and age of the subject. While early detection and intervention, ideally in infancy, is associated with better outcomes,²¹ most cases are diagnosed after age three.

The prevalence and demographics of children diagnosed with ASD in the United States are changing.²² Biological sex is a well-established risk factor for ASD, with the DSM-V reporting that “[ASD] is diagnosed four times more often in males than in females.” Consistent with much of the human literature, our group²³ and others have reported increased severity of autism-like phenotypes among male animals in preclinical models for ASD. However, in a 2009 meta-review of 43 studies on the prevalence of ASD published since 1966,²⁴ the CDC concluded that this gap is narrowing, as the ratio between boys and girls has decreased in the autism and developmental disabilities monitoring (ADDM) network overall. Girls affected by ASD are more likely to be misdiagnosed²⁵ or late-identified²⁶ given that female autism phenotypes diverge from the classically established diagnostic criteria for ASD.^27,28 More recent studies implementing improved diagnostic assessments of social communication and restricted and repetitive behaviors that adjust for bias in sex-related measurements and account for sex-specific symptom trajectories identify a more equivalent ASD prevalence among boys and girls.²⁹ Similarly, classical behavioral analyses in preclinical models for ASD were designed to assess male-dominant behavioral traits;³⁰ thus, there is a historical gap assessing and reporting on female-specific behaviors, like maternal behavior and alloparenting,³¹ in the preclinical literature. Nonetheless, biological factors such as sex hormones and sex-based genetic risk, areas of active investigation, might be responsible for the apparent increase in vulnerability to ASD among males.^32–34

The etiology of non-syndromic ASD, like other neurodevelopmental and neuropsychiatric disorders, is now thought to be multifactorial, involving complex interactions between risk genes and environmental insults^35,36 incurred in early development. ASD etiology has been under intense investigation for decades. In the 1960s, a popular, but errant, theory that lack of parental warmth as a determinant of ASD³⁷ emerged in opposition to competing theories focused on biological factors like brain development.³⁸ In the following decades, epidemiologic, genetic, cytogenetic, and neuroimaging studies^39,40 provided causal links between ASD and altered brain development. The resulting data definitively characterized autism as a disorder of biological origin. This ushered in a new era of investigation into ASD anchored by the concept that gene mutations were the primary driver of autistic phenotypes, leading to the classification of ASD into syndromic and non-syndromic forms.⁴¹ Syndromic autism occurs in subjects with other neurological conditions, such tuberous sclerosis complex (TSC),⁴² Rett syndrome (RTT),⁴³ fragile X syndrome (FXS),⁴⁴ and phosphatase and tensin homologue (PTEN) macrocephaly syndrome.⁴⁵ It is determined by a mutation in a specific gene or group of genes. Non-syndromic autism, which accounts for most cases, is not linked to other defined conditions and cannot be traced to mutations in a single gene or specific chromosomic aberrations. Nonetheless, investigation of syndromic ASD can aid in understanding non-syndromic autism.⁴⁶

Several nongenetic – or, environmental – factors can interfere with fetal brain development in ways that converge onto the molecular and cellular pathways implicated in syndromic autism and are, thereby, proposed to contribute to either risk for or the severity of autistic phenotypes.^47,48 While there is a significant amount of evidence to support that the early life exposome is relevant to ASD pathology, the causal relationship between environmental exposures and ASD remains an area of intense investigation and equally intense debate in the field.

Pregnancy and the periconceptional period have been identified as developmental critical periods during which environmental exposures can influence child health outcomes by longitudinal population-based birth cohort studies, such as the Human Early Life Exposome (HELIX) study.⁴⁹ The environmental factors – chemical or physical agents, nutrition, psychological and social conditions, and infectious disease, among others – to which we are exposed throughout life constitute the ‘exposome.^50,51 The idea that the pregnancy exposome exerts a considerable impact on fetal development originates from David Barker’s “Fetal Programming Hypothesis”⁵² that chronic disorders manifesting in postnatal/adult life may, in part, result from environmental insults in utero. This influential theory is also known as the “developmental origins of health and disease (DOHaD)” hypothesis.

In 2003, the CHARGE (Childhood Autism Risks from Genetics and the Environment) study was launched to comprehensively assess the contribution of environmental factors to ASD.⁵³ CHARGE identified increased risk for ASD associated with maternal occupational exposure to solvents and postulated that other chemicals might have a similar effect on neurodevelopment.⁵⁴ Human epidemiological⁵⁵ and animal studies suggest that other maternal factors, such as infection and diet/metabolic status during pregnancy can likewise contribute to ASD risk in offspring. The underlying mechanisms by which discrete environmental exposures increase risk for ASD, among other non-communicable chronic disorders, are an area of active investigation. One unifying theory is that they induce epigenetic reprogramming of various systems (e.g., the immune and/or nervous systems) during developmental critical periods, which depend on temporally strict and spatially precise regulation of gene expression. The hypothesis that environmentally induced reprogramming of the fetal epigenome can trigger fetal programming of disease risk has gained ground in the scientific and medical communities, setting the stage for large-scale epigenomic studies which have the potential to greatly expand upon current screening and therapeutic approaches for ASD, among other diseases.^56,57

Reimagining ASD as a disease at the crossroad between genes and the environment, in which environmental exposures can serve as a “tipping point” toward disease manifestation in genetically susceptible individuals has facilitated the development of new and innovative theories and investigation into the determinants of ASD. In this context, a growing body of evidence suggests that gut microbiome-immune interactions play a critical role in mediating the effects of early life environmental exposures on long-term neurodevelopmental health outcomes.

Host and microbial genetics in ASD

In recent decades, rodent and non-human primate models for ASD have been developed to facilitate mechanistic understanding of ASD pathology and spur the discovery of novel therapeutics. These models target human risk variants, obtained by genetic engineering,^58,59 and include models for single-gene mutations associated with syndromic ASD or non-syndromic ASD and models of copy number variations (CNVs).

Mecp2 mutant mice⁶⁰ and macaques,⁶¹ for instance, reproduce genetic mutations associated with Rett syndrome (RTT), a neurodevelopmental disorder caused by mutation of the gene encoding X-linked methyl-CpG binding protein 2 (MECP2), which regulates transcription⁶² and RNA splicing.⁶³ Similarly, Tsc1/Tsc2 mutant mice model Tuberous sclerosis complex (TSC), an autosomal dominant neurodevelopmental syndrome resulting from mutation in TSC1 or TSC2, which encode proteins hamartin (Tsc1) and tuberin (Tsc2), respectively, inhibitors of the mTORC1 translational control pathway.⁶⁴ About 50% of TSC individuals are also diagnosed with ASD, and TSC genetic alterations account for about 1–4% of ASD cases.⁶⁵ Tsc1 heterozygous mice show increased anxiety-like behavior, impaired learning and memory, and reduced social interaction, which is rescued by rapamycin administration.^66,67

Mouse models of single-gene mutations associated with non-syndromic ASD, like those encoding neurexins (NRXN), neuroligins (NLGN), and SHANK proteins, have uncovered a significant and valuable portion of ASD biology.⁶⁸ However, modeling the contribution of a single gene in an animal model, particularly when using Cre-drivers to investigate cell type-specific roles of the gene, does not yield full insight into the role of that gene in the pathophysiology of the disease in genetically heterogeneous human patient populations. To overcome these limitations, functional studies have been developed to identify and manipulate one or more cellular and molecular pathways on which common and rare variants associated with ASD converge.^69,70

Most human ASD risk genes belong to at least two main clusters: (1) regulation of mRNA translation and protein synthesis⁷¹ and (2) regulation of synaptic structure and function.⁷² Investigation of syndromic ASD strongly supports a mechanistic link between synaptic dysfunction and dysregulated translational control. An intriguing hypothesis is that non-syndromic ASD is driven by a similar convergence.⁴⁶ We propose that this could also true of some cases associated with specific environmental exposures, particularly those that affect the intrauterine environment.

Epigenetic modifications in ASD

Despite its multifactorial etiology, ASD aggregates in families and is highly heritable.⁷³ However, the concordance rate for ASD in MZ twin pairs ranges from 36% to 96%,^74,75 implicating nongenetic disease liability.^76,77 The emerging hypothesis that epigenetic mechanisms can causally contribute to ASD risk^78,79 provides a rational explanation for disease-discordant MZ twin pairs. Epigenetic modifications, such as DNA methylation, histone modification and regulation, and transcriptional gene silencing by means of long non-coding RNAs (lncRNAs) and small non-coding RNAs (sncRNAs), influence chromatin architecture and conformation, the accessibility of genes to transcriptional complexes, and gene expression.⁸⁰ Epigenetic programming is complex. It can be the result of primary stochastic phenomena, environmental factors, or DNA mutations.⁸¹ Given that precise spatial and temporal regulation of gene expression is crucial for the establishment of proper excitatory and inhibitory synaptic connections, activity-dependent responses, and neuronal specification, epigenetic reprogramming driven by environmental exposures has significant ramifications for brain development, function, and disease risk.

Mutations in genes encoding chromatin remodeling enzymes are implicated in ASD and other neurodevelopmental disorders.^82–84 Significant epigenetic variability has been reported between MZ twins.^85,86 Consistently, considerable differences in DNA methylation have been found in MZ twins discordant for phenotypically complex disorders, like schizophrenia and bipolar disorder.⁸⁷ Analysis of lymphoblastoid cell lines derived from ASD-discordant MZ twin pairs’ peripheral blood lymphocytes revealed many ASD-relevant loci with differential methylation profiles.⁸⁸ Moreover, a genome-wide analysis revealed that DNA methylation at specific CpG sites varied significantly within ASD-discordant MZ twin pairs.⁸⁹ Further studies^90–92 identified differential DNA methylation patterns in autistic versus non-affected individuals. Some autistic individuals carry mutations in genes encoding proteins involved in epigenetic modification. Indeed, a de novo mutation in the HIST1H1E gene, which encodes for the linker histone H, was reported to disrupt chromatin organization and downregulate protein expression in ASD patients.⁹³ Intriguingly, in the same study, a review of SFARI GENE (https://gene.sfari.org/),⁹⁴ a curated database of candidate ASD risk genes, determined that almost 20% of risk genes encode proteins involved in epigenetic regulation and chromatin remodeling. For instance, SETD5, a member of the SET-domain family encoding for histone methyltransferase (HMT) which regulates gene expression during early development and is implicated in both synaptic plasticity and cell fate determination, is linked to both ASD and ID.^95–97

Beyond non-coding RNAs, DNA methylation, and chromatin remodeling, short stretches of RNA known to regulate mRNA translation and degradation, microRNAs (miRNA), are implicated in ASD pathology. The contributions of miRNAs to ASD pathoetiology have been recently well reviewed elsewhere.⁹⁸

The gut microbiome in ASD

Large-scale studies aimed at characterizing the human microbiome (i.e., the NIH-funded Human Microbiome Project (HMP)⁹⁹ and the European Metagenomics of the Human Intestinal Tract (MetaHIT))¹⁰⁰ contributed to technological and computational advances that dramatically increased accessibility to the field and, together, drove a consequent boom in the number of publications on the gut microbiome in human physiology and pathology. It is now well established that the gut microbiome plays key roles in both health and disease, from cancer to autoimmune disorders and neuropsychiatric disorders.¹⁰¹

The gut microbiome is a dynamic ecological system that varies between individuals and within the same individual across the lifespan. Community dynamism occurs over even shorter timescales, as the relative abundance of certain taxa can ebb and flow according to circadian rhythms.¹⁰² Host factors, including diet, drugs, toxins, pathogens, the immune system, and physical and psychological conditions together drive a constant reshaping of gut microbiome composition (Figure 1), which can evolve into a detrimental state – here referred to as, “dysbiosis.”¹⁰³ Powerful stressors can trigger a cascade of events that ultimately lead to a decrease in microbial diversity while simultaneously creating a permissive environment for the growth of opportunistic pathogenic taxa.^104,105 These events, in turn, alter the pool of metabolites produced by the microorganisms, and consequently, interactions between the microbiome and host, often with adverse consequences for the host.

Figure 1.

As a key intermediary between the exposome and genetic susceptibility, the gut microbiome is poised to influence risk for neurodevelopmental disorders.

Classically, mutations in one or more genes involved in regulating brain development and function were thought to be the exclusive drivers of neurodevelopmental disorders; however, a growing body of preclinical and clinical research is revealing a critical role for gene X environment interactions in determining predisposition to and the severity of neurodevelopmental disorders. Environmental exposures – from infection to diet to air quality – that influence the functional composition of the maternal and early life infant gut microbiome can alter the diverse pool of microbially associated metabolites available to the host, which can then affect maturation and function of the immune system and drive cell type-specific epigenetic reprogramming that influences neurobehavioral outcomes. Notably, environmental exposures incurred across multiple generations can affect early life neurodevelopment and disease risk through inherited patterns of microbiome alterations and epigenetic modifications. Given its strategic position between the host and its environment, the gut microbiome presents an intriguing duality as both a potential contributor to and a therapeutic target for reducing risk for neurodevelopmental disorders in children.

Investigations at the intersection of neuroscience and microbiology have begun to unravel the contributions of gut resident microorganisms to the development and homeostasis of host brain structure and function. The microbiome-gut-brain axis (MGBA) encompasses the mechanisms involved in mediating the interplay between the gut and the brain, which include anatomical (e.g., the vagus nerve), immunological, metabolic, neuronal, and chemical pathways.¹⁰⁶ Early evidence for the existence of the MGBA was provided by experiments using germ-free (GF) animals,¹⁰⁷ fecal microbiota transplantation (FMT), and antibiotic-driven changes in the gut microbiome composition¹⁰⁸ in the context of brain disorders.^109–113 These pioneering studies identified key contributions of gut microbiota to brain function and behavior,¹¹⁴ making evident that elucidating the mechanisms by host–microbe interactions affect brain development, function, and behavior is imperative to understanding the physiopathology of many neuropsychiatric conditions and to the development of new microbiota-based and -targeted treatments.¹¹⁵

Among the multiple comorbidities associated with ASD, gastrointestinal (GI) dysfunction – manifesting as diarrhea, constipation, and abdominal pain¹¹⁶ – is one of the most frequently reported,^117,118 though different studies show varying prevalence,¹¹⁹ likely due to the genetic and epigenetic heterogeneity of ASD. The presence of GI symptoms is associated with more severe ASD¹²⁰ and is heavily cited in surveys of parents of autistic children.¹¹⁸ Several studies report substantial differences in the composition of the gut microbiome of autistic children compared to neurotypical controls,^121–123 although a direct causal relationship remains to be demonstrated. Furthermore, the directionality of a relationship – whether ASD pathology alters the gut microbiome or changes in the gut microbiome could causally contribute to ASD – is an area of debate, with limited investigations demonstrating that divergent and limited dietary preferences in autistic individuals may be the driver of any microbiome changes observed in autistic populations when compared to neurotypical controls¹²⁴ (but see,¹²⁵ with which we side). Despite limited reproducibility of taxa-specific signatures across studies of gut microbiome composition among children with ASD, recent investigations coupling intestinal metagenomics with metabolomic analyses have identified metabolic signatures reflecting functional alterations in gut microbial ecology of affected individuals.^126–128 In an effort to reconcile the notorious inconsistency comparing microbiome composition among an autistic population versus controls, Morton et al.¹²⁷ applied a Bayesian differential ranking algorithm to identify commonalities among 10 cross-sectional microbiome datasets and 15 others. The authors identified distinct patterns between children with ASD and age- and sex-matched counterparts when accounting for microbiome (16S rRNA and whole-genome shotgun metagenomic sequencing) and the human transcriptome (RNA-seq). They also revealed positive correlations between pro-inflammatory cytokines, read immune dysregulation, and the global microbial log fold changes between ASD and control pairs – interestingly, canonical pro-inflammatory cytokine IL-6 was only linked to a few taxa, whereas TGF-β levels were linked to many taxa. Moreover, a recent study¹²⁹ comparing microbiome composition between children and adolescents diagnosed with ASD, attention deficit hyperactivity disorder (ADHD), and comorbid ASD/ADHD,^130,131 was the first to show the presence of shared microbiome signature in children with ASD and ADHD which are distinct from non-related controls as well as similarities in altered immune markers and an increase in gut permeability indicators. Here, the authors report lower bacterial richness among youth with both disorders compared to non-related controls, including a specific decrease in the relative abundance of Coprobacter and Howardella; in contrast, Eggerthella – a taxa previously associated with developmental delays in children¹³² – Hungatelle, and Ruminococcus gnavus group were found to be enriched in both the ASD and ADHD groups. While greater microbial variability was reported between children with ASD, as compared to ADHD, and controls, relatively few variations were observed between the gut microbiota of youth with ASD and ADHD. Together, these findings suggest a common, microbiome-mediated mechanism might contribute to the overlapping clinical features of ASD and ADHD.

Gut microbiota produce metabolites that can have a strong impact on behavior and the underlying neural correlates via the gut-brain-axis.^133–135 Consequently, modulating gut microbiome composition represents a novel and innovative strategy for treating ASD. An open-label study of Microbiota Transfer Therapy (MTT)¹³⁶ was among the first to investigate the effects of fecal microbiota transplant (FMT) in children with ASD and chronic gastrointestinal disturbances. A two-year follow-up study¹³⁷ aimed at assessing the long-term results of MTT showed significant improvements not only gastrointestinal symptoms but also in core autism symptoms for all 18 subjects. MTT specifically increased microbial diversity and restored microbial metabolic capability to a similar level to the typically developing (TD) children.¹³⁸ These promising results prompted the FDA to grant fast track status to MTT for autistic children in 2019. More recently, oral delivery of AB-2004, a small-molecule sequestrant targeting microbially derived metabolites, was shown to significantly reduce irritability in children with ASD.¹³⁹ Additionally, a double-blind randomized placebo-controlled trial of precision treatment with a microbe shown to reverse social dysfunction in multiple mouse models for ASD^133,140,141 was found to specifically enhance social behavior in children with ASD, consistent with its behavior-specific effects in the preclinical experiments.¹⁴²

Microbial communities can serve as a source of epigenetic modifiers influencing gene expression.¹⁴³ Microbially derived metabolites have been shown to both directly and indirectly impact the activity of enzymes involved in regulating epigenetic pathways, including those orchestrating DNA methylation and histone modification.¹⁴⁴ Microbial regulation of host chromatin modification states and associated transcriptional responses is strictly dependent on host dietary patterns, in particular fiber content. Microbial anaerobic fermentation of insoluble dietary fiber produces short-chain fatty acids (SCFAs),¹⁴⁵ organic monocarboxylic acids that can cross the intestinal barrier through monocarboxylate transporters (MCTs), travel through systemic circulation, and reach distal organs.¹⁴⁶ Here, they can be metabolized as an energy source via the Krebs cycle but also play multiple signaling roles as SCFAs bind to the G protein-coupled receptors (GPCR) GPR43 and GPR41, later renamed free fatty acid receptor 2 (FFAR2) and 3 (FFAR3).¹⁴⁷ SCFAs promote intestinal barrier integrity,¹⁴⁸ counteract intestinal inflammation,¹⁴⁹ and modulate gastrointestinal motility, adipogenesis, and glucose homeostasis.¹⁵⁰ They have been shown to modulated intestinal mucosal immunity¹⁵¹ and are thought to also affect the peripheral immune system. CD4⁺ regulatory T cells, in particular Th17 cells,¹⁵² are regulated by SCFAs and their differentiation is impaired in GF mice.¹⁵³ Similarly, SCFAs are required for CD8⁺ cytotoxic T cell transition in memory cells.¹⁵⁴ SCFA oral administration promoted peripheral regulatory T cell differentiation in mice,¹⁵⁵ and FFAR agonists have been shown to modulate the human monocyte inflammatory pathway by decreasing the release of pro-inflammatory cytokines.¹⁵⁶ Such regulation of peripheral immunity might be key to their impact on brain function.

SCFAs facilitate MGBA communication and affect brain physiology through multiple mechanisms, as reviewed in Dalile et al. (2019).¹⁵⁷ SCFA receptors are expressed by neurons in both the peripheral and central nervous systems.¹⁵⁸ Recent work showed an association between SCFA production in the gut and regulation of feeding behavior via direct hypothalamic regulation.¹⁵⁹ Additionally, SCFAs maintain blood–brain barrier (BBB) integrity.¹⁶⁰ Propionate-induced FFAR3 activation of vagal fibers has been shown to increase the activity of the dorsal vagal complex, parabrachial nuclei, and hypothalamus,¹⁶¹ thus suggesting that SCFAs can directly influence brain activity through vagal signaling. Another mechanism by which SCFAs influence brain activity is via enteroendocrine signaling. FFAR activation in gut enteroendocrine L cells determines the production of hormones released in response to food, GLP-1 and PYY, into circulation^162,163 to regulate appetite and nutrient intake.^164,165 Animal studies have shown GLP-1 can further promote learning and memory¹⁶⁶ and improve neuroplasticity while reducing microglial activation in the hippocampus.¹⁶⁷ Furthermore, microbially derived SCFAs influence hippocampal neurogenesis by acting on monocytes.¹⁶⁸ SCFAs are also implicated in the modulation of anxiety- and depressive-like behavior following psychosocial stress, which is associated with alterations in the gut microbiome.¹⁶⁹

SCFAs can directly modulate the activity of histone deacetylase enzymes (HDACs), which regulate chromatin accessibility and gene expression. Acetate, butyrate, and propionate exert an inhibitory effect on HDACs,¹⁷⁰ dysregulation of which is linked to neurodegenerative and neuropsychiatric disorders, including schizophrenia.¹⁷¹ They are proposed to mediate hippocampal long-term potentiation by enhancing histone acetylation, a process required for long-term memory formation.¹⁷² Additionally, SCFAs modulate epigenetic modifications in neuro-, peripheral, and enteric immune systems.^173,174 Acetate was recently shown to regulate microglial metabolism and function through modifying histone methylation on genes related to microglial proliferation, morphology, and activation.¹⁷⁵ Finally, SCFAs, particularly propionate, decrease IL-17 production by human and mouse γδ intraepithelial lymphocytes in a HDAC-dependent manner.¹⁷⁶

Several recent studies investigated the association between altered SCFAs and neurodevelopmental disorders in children, including ASD;^177,178 however, whether SCFAs contribute to or relieve ASD pathology and symptom severity is controversial. While some studies report elevated SCFAs in fecal samples isolated from autistic individuals compared to controls, others report lower levels. Serum SCFA concentration is less commonly reported but this could be an important data point for teasing out the effects of SCFA concentration on host brain function and behavior in the context of ASD.^179,180 Higher levels of SCFAs have been found in a valproic acid mouse model for ASD,¹⁸¹ while a beneficial effect of butyrate on social dysfunction has been reported in the BTBR mouse model for idiopathic ASD.¹⁸² Hence, it is likely that the impact of SCFAs on host health is context-dependent.

Taken together, the studies above demonstrate that gut microbiota influence host gene expression and can thereby contribute to disease risk and outcomes. This intricate interplay between host and microbial factors could contribute to the huge variability characteristic of nonsyndromic ASD. Heritability of the gut microbiome, another form of genetic inheritance, could furthermore influence disease risk across generations, as we have found in an animal model for maternal overnutrition.²³ However, such heritability appears to be responsive to environmental forces.¹⁸³ Further studies with large sample sizes and high resolution longitudinal multi-OMICs-based assessments are required to elucidate the extent to which gut microbiota impact health outcomes across multiple generations. Nonetheless, increasing evidence suggests that antenatal maternal gut microbial communities play a particularly pivotal role in offspring neurodevelopment, especially during critical developmental stages such fetal and early-post natal life.

Neurodevelopmental disorders and the maternal exposome

In recent years, the pregnancy exposome ¹⁸⁴ has garnered increased attention in the investigation of risk factors occurring in early development that may permanently affect vulnerability to disease later in life. Given that the maternal environment is, to an extent, modifiable, it has also become a target for the development of therapeutic interventions. Multiple investigations highlight the connection between the pregnancy exposome and fetal programming of disease, particularly metabolic impairments and cardiovascular disorders.^185–188 We and others propose that early life programming is likely to influence risk for neurodevelopmental disorders, including ASD.^189–191

Intrauterine brain development involves a precise succession of events orchestrated by lineage-specific gene expression programs.^192–194 Environmental insults can determine cell type-specific epigenetic programming at different stages of neurodevelopment,^195,196 including of neurons, glia, and immune cells.¹⁹⁷ Epigenetic reprogramming is implicated in synaptic formation,^198,199 and environmental exposures incurred during the third trimester of human development can disrupt synaptogenesis, a key process dysregulated in ASD. Multiple nongenetic factors, such as maternal infection during pregnancy, maternal diet/metabolic status, and maternal chemical exposure, incurred at various pregnancy stages have been proposed to interfere with the developing human brain and, thereby, contribute to autistic phenotypes.^47,48 Below, we consider the relationship between environmental exposures, the maternal gut microbiome, and offspring outcomes.

Maternal infection

In utero exposure to maternal viral infections, particularly those requiring hospitalization, is associated with increased risk for ASD. In the 1970s, an American child psychiatrist, Stella Chess, diagnosed symptoms of ASD in a group of pediatric patients with congenital rubella syndrome, resulting from the 1963–1964 rubella epidemic in New York. The reported prevalence was 200 times higher than that of the general population in the US.^200,201 Subsequent studies revealed similar findings, not only in relation to rubella infection but also in response to maternal infection during pregnancy with other viruses, such as cytomegalovirus^202,203 and influenza.^204,205 A great body of work now demonstrates that damage to the developing brain results from maternal immune activation (MIA) and related inflammatory responses independent of the specific class of pathogen (e.g., viral versus bacterial).^206–208 Relatedly, multiple studies identify an association between ASD and dysregulation of the inflammatory response: increased activation of microglia and astroglia,²⁰⁹ upregulation of markers of inflammation,²¹⁰ and alterations in genes involved in immune function²¹¹ have been identified in autistic patient populations. Consequently, cytokine profiles have been proposed as biomarkers of immune dysfunction in autistic individuals.²¹² Indeed, human studies have identified a significant increase in the levels of pro-inflammatory cytokines, including IL-6,²¹⁰ TNFα,²¹³ IFN-γ,²¹⁴ IL-17,²¹⁵ in the brain and in biological fluids, including serum and cerebrospinal fluid, of autistic individuals compared to controls. In contrast, levels of anti-inflammatory, regulatory cytokines such as IL-23²¹⁶ and TGF-β²¹⁷ have been found to be downregulated in autistic individuals. Similarly, analyses of serum^218,219 and amniotic fluid^206,220 from mothers who gave birth to autistic children reveal an increase in proinflammatory cytokine and chemokine levels, when compared to control subjects.

The causal relationship between gestational MIA and ASD and the biological mechanisms by which MIA interferes with fetal neurodevelopment has been investigated using preclinical animal models²²¹ including: (1) maternal exposure to pathogens during pregnancy, such as influenza virus,^205,222,223 (2) exposure to agents which stimulate the innate immune system, such as the bacterial endotoxin lipopolysaccharide (LPS), polyinosinic-polycytidylic acid (poly(I:C)),²²⁴ a synthetic analog of double-stranded RNA (dsRNA), or the soluble tachyzoite antigen of the protozoan Toxoplasma gondii ²²⁵ (3) stimulation of the immune system by pro-inflammatory cytokines, such as IL-6,²²⁶ and (4) administration of immunological factors linked with the pathogenesis of the disorder,²²⁷ as in the case of ASD-associated maternal autoantibodies.²²⁸

The effects of MIA on fetal neurodevelopment differ according to pregnancy stage at the time of exposure. Studies in LPS/Poly(I:C)-stimulated models have shown that the development of the dopaminergic (DA) system, a neuronal network strongly implicated in ASD,²²⁹ is selectively impaired when MIA is triggered during early gestation.²³⁰ In LPS/Poly(I:C) models, stimulants do not reach the fetus directly; instead, their effects on neurodevelopment are mediated by maternal cytokines released in the circulation and transmitted to the fetus through the placenta.²³¹ Cytokines modulate neurodevelopment,^232,233 however an imbalance in maternal pro- and anti-inflammatory cytokines can have detrimental effects on the fetal brain. The relationship between maternal cytokine imbalance and disruption of neurodevelopment has also been investigated in the context of ASD.²³⁴ Offspring of mice stimulated with Poly(I:C) or LPS during gestation display core ASD-like behavioral impairments:²³⁵ repetitive self-grooming and stereotypies, restricted interests/cognitive inflexibility, and decreased sociability.^{134,236–240} In this context, IL-6,²⁴¹ IL-17a,²⁴⁰ and IL-1β²⁴² have been implicated in MIA phenotypes. However, the precise mechanisms by which pro-inflammatory cytokines impact fetal brain development, brain function, and behavior remain mostly unknown.

Another mechanism linking immune and brain function (or dysfunction, in the case of MIA) converges on mammalian target of rapamycin (mTOR) complexes one and two (mTORC1, mTORC2), powerful regulators of mRNA translation and actin (and, thereby, synaptic) remodeling, respectively.²⁴³ mTORC1 and mTORC2 are involved in the differentiation of Th1 and T helper 17 (Th17) and Th2 cells, respectively.²⁴⁴ mTOR complex inhibition drives T cells to differentiate into Treg cells.²⁴⁵ Thus, an interesting hypothesis suggests that mTOR hyperactivity along the gut-brain-immune axis, one of the pathological mechanisms involved in ASD, might lead to a decrease in Treg cell-associated anti-inflammatory cytokines, including IL-10 and TGF-β.²⁴⁶

Multiple studies highlight a role for the gut microbiome in the interplay between the immune system and neurodevelopment in the context of MIA.²⁴⁷ A seminal study in the field demonstrated that supplementation of MIA offspring with a probiotic species shown to contribute to host immune maturation and protect against Helicobacter hepaticus-driven colitis in mice,²⁴⁸ Bacteroides fragilis, could rescue many ASD-like behaviors including aberrant communication, stereotypy, anxiety-like, and hyperactivity, but notably did not rescue social dysfunction.¹³⁴ While the gut microbiome influences the development of both the adaptive and innate immune systems both locally and systemically, the immune system finely tunes the symbiotic host–microbe relationship to avoid microorganism overgrowth while simultaneously allowing tolerance.^249,250 For instance, a MIA-associated spike in IL-17a in maternal circulation is a consequence of Th17 cell expansion in the gut and depends on segmented filamentous bacteria (SFB). Mice devoid of SFB are protected from the pathogenic release of IL-17a and do not display the associated phenotypical aberrations typical of the MIA model.^135,240 Immunotherapy-mediated blockade of pathological IL-17a pathway activation likewise prevented ASD-like symptoms in the poly(I:C) MIA model.²⁴⁰

Maternal infections during gestation could directly affect fetal microglia, which could contribute to behavioral dysfunction observed in MIA offspring.²⁵¹ Microglia are innate sentinel immune cells which regulate inflammatory processes in the brain by the release of pro- and anti-inflammatory cytokines and chemokines. In addition to immune surveillance, CNS microglia also regulate CNS maturation and synaptic plasticity. Microglial abnormalities have been reported in postmortem analysis of brains of autistic individuals.²⁰⁹ Microglial maturation and differentiation of microglia are partially regulated by gut microbes.^174,252 The maternal microbiome in particular influences the activity of microglia during prenatal life, as microglial homeostasis was shown to be disrupted in mice born to germ-free dams.²⁵³ Microglia also mobilize monocytes from the periphery to enter the brain, a process mediated by systemic TNF-α signaling, which leads to microglial activation and subsequent recruitment of activated monocytes.^254,255 Interestingly, the trafficking of monocytes from the spleen might be modulated by microbiota-produced SCFAs, which bind to free fatty acid receptor type 2 (FFAR2) expressed on peripheral lymphocytes.²⁵⁶

Recent work highlighted the effect of SCFA supplementation, via high-fiber diet, on the inhibition of inflammatory microglia by downregulation of HDAC activity, NF-κB activity, and inflammation caused by LPS stimulation.²⁵⁷ Inhibition of HDACs, which leads to transcriptional repression, has been proposed as a primary downstream action of SCFAs toward prevention of neuroinflammation.²⁵⁸ A recent study conducted in the BTBR mouse model for idiopathic autism, which is characterized by systemic immune dysregulation and comorbid gut dysbiosis, traced the immune abnormalities back to the embryonic stages of the yolk sac where macrophages (microglia) and peripheral immune cells differentiate.²⁵⁹ The underlying mechanism involved transcriptional regulation by HDAC1. The epigenetic abnormalities, associated with increased proinflammatory cytokines and microglia activation, were successfully reversed upon administration of sodium butyrate, which inhibits HDAC1 activity. This study reveals a key role for epigenetic reprogramming of immune function as common etiology between environmental risk factors for ASD and highlights the potential for correcting postnatal immune dysregulation at the embryonic stage through maternal microbiome-targeted therapies.

Maternal diet

Maternal nutrition and metabolism exert a major impact on offspring fetal and early-postnatal development, including on gamete maturation and placental growth.²⁶⁰ Furthermore, maternal intake of micro- and macronutrients-alike has been shown to be crucial for successful development of offspring organs and systems, including the nervous and immune systems, which has significant implications for neurodevelopment and risk for neurodevelopmental disorders.^47,261

Recent epigenome-wide studies of low-income populations^262,263 suggest that epigenetic alterations contribute to the detrimental consequences of micronutrient deficiency on neurodevelopment. Vitamin D and folate deficiency have been extensively investigated in the context of ASD and other NDDs.²⁶⁴ Vitamin D plays a crucial role in many biological processes,²⁶⁵ with numerous studies highlighting its importance in pregnancy and fetal growth and development.²⁶⁶ Suboptimal levels of circulating vitamin D in pregnant women²⁶⁷ are associated with increased pregnancy complications, such as miscarriage,²⁶⁸ hypertensive disorders,²⁶⁹ and gestational diabetes.²⁷⁰ Notably, low maternal vitamin D also increases risk for developmental deficits, including ASD, in offspring.^271–273 To this end, vitamin D supplementation is indicated when vitamin D deficiency is identified during pregnancy.²⁷⁴ Interestingly, it has been reported that maternal depletion of vitamin D is associated with alterations of the epigenetic landscape, specifically in DNA methylation, across multiple generations.²⁷⁵ Recent studies in human populations²⁷⁶ have reported significant associations between vitamin D and changes microbiome composition in the context of autoimmune disorders. These findings suggest a link between maternal vitamin D, microbiome composition, and immune function with potential effects on fetal epigenetic programming. However, further studies are required to test this hypothesis. Similarly, folate (vitamin B-9) is a crucial nutrient during early pregnancy for reducing risk for birth defects, notably neural tube defects (NTDs).²⁷⁷ The neuroprotective effect of folate is likely mediated by genome-wide modification of methylation patterns in neural target genes.^278,279 While mammalian cells are unable to produce folate, and therefore exogenous intake is required, there is substantial evidence that colonic bacteria produce a considerable amount of folate, as well as other B-vitamins. While their production can be enhanced by prebiotic supplementation,²⁸⁰ there is no evidence for an essential role for bacterial folate biosynthesis in early development.

Beyond micronutrient supplementation, large preclinical and clinical studies on the effects of macronutrient intake – general dietary habits considering the ratio of protein to fat to carbohydrates consumed – such as Western Pattern Diet (WPD),²⁸¹ also known as the Standard American Diet (SAD), Mediterranean Diet (MedDiet),²⁸² and ketogenic diet (KD),²⁸³ and associated metabolic conditions, are beginning to reveal how maternal diet could contribute to risk for ASD and other NDDs. As introduced above, the developmental origins of health and disease (DOHaD) hypothesis suggests that maternal diet, in particular maternal undernutrition, has a causal role in the incidence of various disorders in adulthood. While initial observations focused on alterations in cardiovascular and metabolic function in subsequent generations of populations impacted by famine, this theory was later extended to include risk for mental disorders.²⁸⁴ Increased prevalence of major affective disorder,²⁸⁵ antisocial personality disorder,²⁸⁶ schizophrenia spectrum disorder,²⁸⁷ impaired cognitive performances,²⁸⁸ and substance addiction²⁸⁹ have each been correlated with prenatal exposure to extreme caloric restriction in epidemiological studies. Notably, further epidemiological data suggests specific correlations between the trimester in which mothers are exposed to dietary restrictions and the specific disorder induced in the offspring,^290,291 thus highlighting the relevance of timing in developmental programming of future disorders, and suggest a critical role for exposure during early gestation.

Operating as the interface between the maternal and fetal blood circulation and regulating the nutrient and oxygen transfer from the mother to the fetus, the placenta directly contributes to intrauterine fetal programming. It integrates maternal and fetal signals and constantly balances fetal needs with maternal supply. Perturbations in the maternal compartment are sensed by the placenta, which in turn modulates blood flow and nutrient supply and adaptively modifies hormonal release through epigenetic changes in placental cells.^260,292 Excessive deprivation (undernourishment) or abnormal increase (overnutrition) in maternal nutrient intake at conception and throughout pregnancy impact the ability of the placenta to properly allocate necessary resources for fetal growth.²⁹³ Nutrient sensing in the placenta occurs by means of multiple mechanisms, including one involving the mTORC1 translational control pathway²⁹⁴ in the syncytiotrophoblast, which is regulated by several factors, including maternal adipokines,²⁹⁵ IL-6,²⁹⁶ TNF-α,²⁹⁷ leptin,²⁹⁸ adiponectin.²⁹⁹ Maternal undernourishment causes a decrease placental amino acid transport, which in turn drives intrauterine growth restriction (IUGR).^300,301 Maternal obesity and diabetes drive excess of nutrient supply to the placenta which results in fetal overgrowth^302,303 and increased risk for the infants to develop obesity and metabolic dysfunction in adulthood.^304,305 This evidence suggests a U-shaped relationship between maternal nutritional imbalances, either maternal malnutrition^306,307 or obesity,^306–308 with offspring risk for metabolic disorders.

Maternal diet and metabolism are also implicated in offspring mental health outcomes. Preclinical and human studies suggested a role for maternal nutrition in the etiology of neuropsychiatric disorders,^309–312 neurodevelopmental disorders,^313,314 and cognitive function.^315,316 As we recently reviewed in Di Gesu et al. (2021),²⁸¹ maternal obesity, overweight, and associated metabolic disorders increase odds ratios for neurodevelopmental disorders, including ASD and ADHD, in children.^317–319 Accumulating evidence provided by epidemiological studies suggests a strong correlation between maternal obesity^320–323 and diabetes^323,324 and an increased risk for ASD among children exposed to maternal obesity and diabetes in utero. Animal models³²⁵ of diet-induced obesity have been used to investigate the mechanisms underlying the detrimental effects maternal obesity on offspring neurodevelopment in neuropsychiatric disorders. Maternal high-fat diet (MHFD) has been shown to impair synaptic plasticity,³²⁶ social behavior,^23,133,327 learning and memory^328,329 and neurogenesis^330,331 in offspring. Yet, the precise mechanisms underlying the impact of maternal diet on offspring neurodevelopment remain to be determined.

A growing body of evidence suggests that dietary challenges, as in the case of WPD, and associated dysbiosis of the gut microbiome have the potential to reprogram the host epigenome in a tissue-specific fashion due to alterations in the intestinal metabolite pool either produced or transformed by microbiome enzymatic activity.^145,332 Epigenetic alterations are associated with other maternal nutritional conditions, such as maternal starvation³³³ and maternal deficiency of vitamins and cofactors,^334,335 and are associated with ASD and other NDDs.^79,336 Modifications in the fetal epigenome are associated with maternal obesity^337,338 and maternal diabetes.^339–341 For instance, MHFD offspring display alterations in histone binding and expression of the oxytocin receptor (OXT-R) in the hippocampus³⁴² which plays an important role in social behavior,³⁴³ hypermethylation in the regulatory regions of hypothalamic POMC gene^344,345 which is involved in regulation of food intake,³⁴⁶ and alteration of histone modifications and expression in the hippocampal leptin receptor (Lepr),³⁴⁷ which is involved in synaptogenesis and neural circuit maturation.³⁴⁸

Maternal high-fat diet (MHFD)-induced changes in both the maternal and offspring immune system could, likewise, causally contribute to increased risk for NDDs. Interestingly, in perinatal MHFD mouse models, changes in the expression of several epigenetic regulators in the offspring developing brains were found in association with an anxiety-like phenotype.³⁴⁹ One intriguing possibility is that, similar to what has been observed in MIA offspring, one of the causal mechanisms underlying social dysfunction in MHFD offspring might depend on maternal microbiota-dependent pathological activation of pro-inflammatory pathways, such as the IL-17a pathway, given that HFD regimen also results in a TH17 bias,³⁵⁰ and MHFD drives microbiota-dependent IL-17–producing type 3 innate lymphoid cell expansion (ILC3s).³⁵¹ Of note, L. reuteri supplementation has been shown to promote differentiation of CD4⁺CD8aa⁺ T precursor cells into immunoregulatory T cells (Tregs), as opposed to a TH17 cell fate,³⁵² an effect which might be mediated by probiotic-dependent increase in SCFA levels.

More recently, birth cohort studies have investigated the potential beneficial effects of a Mediterranean-style diet, rich in fruits, vegetables, and polyunsaturated fatty acids (PUFAs), and low in ultra-processed foods and saturated fatty acids, during pregnancy in reducing risk for neurodevelopmental disabilities in offspring.^282,353,354 While adherence to a MedDiet has been linked to lower mortality and decreased prevalence of obesity, diabetes, low-grade inflammation, cancer, neurodegenerative disorders, and depression,^355,356 the underlying molecular and cellular mechanisms of action are not yet fully elucidated. Recently, a link between the MedDiet and the gut microbiota in disease risk was proposed.³⁵⁷ In randomized controlled studies, a MedDiet regimen in obese subjects lowered plasma cholesterol and was associated with changes in the gut microbiome and systemic metabolome independent of caloric intake.³⁵⁸ A similar study in elderly populations showed that MedDiet increased the abundance of microbial taxa associated with SCFA production, lowered inflammation, and improved cognitive performance.³⁵⁹

The effects of adherence to Mediterranean dietary patterns during gestation on both maternal and offspring health outcomes have been investigated in contrast to obesity-associated WPD,³⁶⁰ however the relationship between maternal MedDiet and offspring neurodevelopment has yet to be fully explored. Interestingly, a recent study including mother and infant dyads enrolled in the Newborn Epigenetics STudy (NEST), showed that MedDiet in early gestation was associated with favorable neurobehavioral outcomes and sex-dependent changes in methylation patterns of imprinted genes in offspring.³⁶¹ As mentioned above, the MedDiet is described as rich in polyunsaturated fatty acids (PUFAs), which are classified as n-6 PUFAs and n-3 PUFAs and considered essential nutrients given the absence of specific enzymes required for their synthesis in mammals.³⁶² Intriguingly, low levels of n-3 PUFAs (mainly eicosapentaenoic acid (EPA) and docosahexaenoic (DHA) acids) in the plasma and the brain of autistic individuals have reported in epidemiological studies.^363,364 Randomized controlled trials reported behavioral improvements in ASD children treated with n-3 fatty acid dietary supplementation.³⁶⁵ Animal studies corroborated epidemiological evidence showing that n-3 PUFA deficiency to be associated with alterations in GABAergic and dopaminergic neurotransmission as well as ASD-like behavioral impairment in rodents.^366,367

Studies in MIA models report a positive effect of n-3 PUFA-enriched diet in reducing ASD-like symptoms.³⁶⁸ Given that n-3 PUFAs can regulate neuroinflammatory processes and microglial activity,³⁶⁹ their protective effects stem from anti-inflammatory activity in the developing brain. In further support of the beneficial role of these fat molecules in preventing ASD-like symptoms in animal models, a recent study showed that n-3 PUFA supplementation prevented the behavioral, cellular, and molecular ASD-related disturbances in the VPA mouse model for ASD, especially in female offspring.³⁷⁰ Interestingly, n-3 PUFAs have been shown to modulate the composition of the gut microbiome,³⁷¹ and increase the abundance of probiotic taxa, such as Lactobacilli and Bifidobacterium, and SCFA-producing species.³⁷² Conversely, gut microbiota can also affect the metabolism and absorption of n-3 PUFAs.³⁷³ While the interplay between gut microbes and this class of polyunsaturated fats remains to be fully understood, recent research demonstrated PUFA supplementation ameliorated autistic phenotypes and GI dysfunction in Fmr1 knockout (KO) mice, a genetic model for fragile X syndrome (FXS), by altering the gut microbiome.³⁷⁴ Epidemiological studies show that maternal consumption of n-3 PUFAs is associated with lower risk of impairments in social development scores and motor and communication skills in children.³⁷⁵ While preclinical studies showed that exposure to a diet rich in n-3 PUFAs during pregnancy modulate the gut microbiome of the offspring and prevents metabolic alterations induced by HFD, further investigation is required to elucidate how dietary PUFAs act on maternal microbial ecology and contribute to the development of ASD in offspring.³⁷⁶ Given the ability of PUFAs to positively modulate epigenetic modifications in both the placenta and the fetal brain,^377–379 it is possible that maternal gut microbiota mediate the effects of PUFAs in offspring via epigenetic programming of neurodevelopment. Consistently, the relationship between maternal PUFA consumption, associated changes in the gut microbiome, and offspring risk for neurodevelopmental disorders, including ASD, is of great interest in the research community and we anticipate increasingly mechanistic work to come out on this topic.³⁸⁰

Another dietary pattern which has gained popularity in recent years is the ketogenic diet (KD), which is characterized by a high proportion (modeled at 74% kcal from fat) of fats and proteins with low intake of carbohydrates. KD enhances the production of ketone bodies (KBs), which can substitute for glucose as the primary energy source, especially in the brain where KBs are used as substrates for oxidative metabolic processes. KBs are considered beneficial in certain neurological conditions given their role in various in multiple brain processes, including neuroinflammation, neuroplasticity, synaptic transmission, and cellular energetics and metabolism.³⁸¹ The beneficial effects of KD have been proposed to be mediated by changes in the gut microbiome. Importantly, despite the high intake of fats characterizing the KD, these alterations seem to be distinct from those induced by HFD regimen, probably due to the concomitant production of KBs by the host. KBs have been shown to reduce the abundance of Bifidobacterium, with a concomitant decrease in Th17-mediated immune response,³⁸² a pathway implicated in ASD pathogenesis.^240,383 Multiple studies have demonstrated the ability of a KD to mitigate some of the behavioral symptoms displayed in animal models of ASD.^384,385 Similar effects were also observed in autistic children following a KD regimen, including amelioration of hyperactivity, social deficits,³⁸⁶ and seizure frequency.³⁸⁷ Preclinical evidence suggests that the antiseizure effects of KD occurs via increases in the relative abundance of Akkermansia and Parabacteroides in mice.³⁸⁸ Additionally, studies in the BTBR model of ASD showed that microbiome remodeling was crucial in the modulation of neurobehavioral symptoms.^389,390 While these data suggest that the KD and associated effects on the gut microbiome may be a new therapeutic approach in autistic patients, there is a paucity of investigations on alterations of the gut microbiota in children treated with a KD and caution is warranted when proposing KD in autistic populations. KD is an extreme dietary regimen which could be difficult to implement in children affected with ASD, who often display signs of food aversion and might therefore lead to macronutrient and micronutrient deficiencies. Similar considerations should be made when contemplating maternal nutritional interventions based on KD, given limited evidence for beneficial effects on offspring neurodevelopment. Notably, a recent study investigating the effects of KD on the course of gestation and fetal development in rats showed both metabolic impairments and delays in neurological development, particularly in female offspring.³⁹¹

The studies summarized here highlight the profound effects of maternal nutrition – from the periconceptional period throughout gestation and lactation – on offspring neurodevelopment and provide compelling evidence for a complex interplay between maternal nutrition, the gut microbiome, and the immune response, with epigenetic modifications as a consequential bridge between these environmental cues and early life programming of neurodevelopmental disorders.

Maternal chemical exposure

Exposure to even small amounts of toxic chemicals, such as lead, methylmercury, polychlorinated biphenyls [PCBs], arsenic, and toluene, relatively harmless in adult individuals, during pre- and early postnatal neurodevelopment^392,393 can result in neurodevelopmental toxicity and brain damage.³⁹⁴ The placenta is permeable to a large number of environmental toxins,³⁹⁵ therefore the fetus has little-to-no protection against these agents, many of which can easily cross the blood–brain barrier.^396–398 The detrimental effects of exposure to toxic substances are not limited to the intrauterine period, but instead extend across many years.¹⁹²

PCBs (polychlorinated biphenyls), a class of compounds utilized as additives for pesticides, insulators, paints, and glues, are internationally recognized as hazardous. While industrial exposure to PCBs showed mild toxicity in adult individuals, severe behavioral impairments, as well as hormonal and immune dysfunction were observed in children born to exposed mothers.^399–401 The first evidence for neurotoxicity of PCBs dates back to the 1970s, when babies born to women who ingested PCB-contaminated cooking oil during pregnancy in Japan and Taiwan suffered from significantly higher rates and more severe forms of cognitive and psychomotor impairments.⁴⁰² However, subsequent epidemiological investigations revealed that maternal exposures to even lower, yet environmentally relevant, levels of PCBs poised significant risk for neurotoxicity in offspring,⁴⁰³ and was associated with higher risk of neurodevelopmental disorders, including ADHD and ASD.^404,405 Mechanisms by which PCBs have been proposed to disrupt neurodevelopment include interference with thyroid hormone (TH), altered signaling of γ-aminobutyric acid (GABA), and disruption of intracellular calcium ion (Ca2+) dynamics.⁴⁰⁶

Pesticides, including insecticides and fungicides, are widely used in agriculture and their residues can be found in and on fruits, vegetables, and other food products,⁴⁰⁷ with a large number of these being specifically designed to produce neurotoxic effects in targeted pests.^408,409 Prenatal exposure to organophosphates, which are the most widely used class of pesticides, has been associated with neurodevelopmental deficits, including cognitive dysfunction and attention deficits.^410–413 Interestingly, maternal intake of high dose folic acid preceding pregnancy has been shown to reduce ASD risk arising from prenatal exposure to pesticides,⁴¹⁴ suggesting a protective role for folic acid against harmful chemical exposures.

Fortunately, a growing number of countries, including the US, have begun to implement environmental regulations to limit exposures to known toxicants. Among these, the Toxic Substances Control Act (TSCA) of 1976 provided the US Environmental Protection Agency (EPA) with the authority to require rigorous recording and reporting of chemical substances and/or mixtures released into the environment; however, most food, drugs, cosmetics, and pesticides were notably excluded from the chemicals covered by this act. After these regulatory initiatives, the environmental concentrations of PCBs in commercial mixtures gradually declined. Nevertheless, the persistence of legacy PCBs pose a health hazard to humans,⁴¹⁵ with ongoing utilization of old equipment containing PCBs and their release from aging construction components.

Despite well-intended regulation, the enactment, enforcement, and achievement of strict environmental quality standards remains a challenge, given the inevitable lag between the production and broad commercial application of new classes of synthetic chemicals and research efforts to objectively perform a comprehensive assessment of their impacts on short- and long-term health outcomes. While PCBs came to the forefront of environmental policy when the EPA developed regulations under the TSCA, PFAS (Per- and Polyfluoroalkyl Substances) were subject to relatively less strict regulations, with scientists mainly concentrating on offering guidance on acceptable levels of PFAS in drinking water and groundwater. However, recent studies indicate that PFAS are building up in the environment and exhibit long-lasting stability, to the point that are commonly referred to as “forever” chemicals.⁴¹⁶ This accumulation of PFAS has increasingly adverse effects on both human health and the environment. Consequently, PFAS are becoming as widely recognized among the public as PCBs. PubMed citations including “PFAS” grew from 1 in 2002 to >1,200 in 2023. They are accompanied by a growing number of articles published by popular press outlets, including news and lifestyle magazines, highlighting links between PFAS and health risks, as well as a nuanced consideration of the danger presented by how little is known about them and the consequences of cumulative exposures.

Epidemiological studies have highlighted associations between exposure to specific PFAS, which find extensive application in both industrial and consumer goods, including coatings applied to fabrics, carpets, paper goods, and as nonstick coatings on cookware, and multiple health outcomes, including alterations in immune function, liver and kidney disease, metabolic dysregulation, adverse reproductive and developmental effects, and cancer.⁴¹⁷ As mentioned in previous sections, the CHARGE population-based case–control study investigated the association between environmental factors, including prenatal maternal exposure to PFAS, and risk for autism and developmental delay in 1,800 children and their families. Results from this study revealed that modeled prenatal exposure to perfluorohexane sulfonate (PFHxS) and perfluorooctane sulfonate (PFOS), but not other PFAS, was linked to higher odds of a child being diagnosed with ASD.⁴¹⁸ However, the authors concluded that additional studies in which PFAS concentrations are prospectively measured in mothers and children at multiple developmental stages were required to corroborate these findings. Another study suggested that prenatal serum levels of Perfluorononanoic acid (PFNA), a type of PFAS, might be linked to slight increases in autism-related traits in children.⁴¹⁹ However, future research is warranted to investigate the correlation between maternal exposure to both established and newly emerging PFAS and various quantitative measures of autism-related health outcomes in offspring.

Intriguingly, recent evidence suggests that the gut microbiome might be a major player in the toxicity of environmental pollutants, including pesticides and PFAS.⁴²⁰ Exposure to such toxicants might increase risk for psychiatric or neurological disorders through perturbations of the microbiome-gut-brain axis.⁴²¹ Maternal exposure to environmental pollutants, including metals, PFAS, and pesticides, was associated with alterations in developmental trajectory of the gut microbiome in infants, especially breastfed infants, in a recent birth cohort study.⁴²² Additional evidence of maternal PFAS exposure on microbiota composition in mother-infant dyads in Finland revealed alterations in the maternal gut microbiome and in the levels of certain metabolites, such as bile acid glycoursodeoxycholic acid (GUDCA) and cholic acid (CA).⁴²³ Further studies are required to determine the full extent of consequences of maternal exposures to individual environmental chemicals on offspring neurodevelopment, their impact of the risk neurodevelopmental disorders, and the potential involvement of the gut microbiome in mediating neurotoxic effects of such compounds.

Therapeutic targeting of the maternal gut microbiome to reduce ASD risk

Pregnancy is characterized by significant remodeling of the maternal gut microbiome, even in the absence of environmental insults.^424–426 This remodeling is driven by changes in hormone levels, immunity, and metabolic function required to support fetal development. In parallel, microbiota are poised to modulate immune and metabolic adaptations during pregnancy, as well as gut barrier function, by means of bioactive microbially derived metabolites. In the third trimester, total bacterial load is increased while microbial richness is decreased in the maternal gut microbiome. A concomitant increase in the proportion of opportunistic pathogens is thought to promote in the development of the offspring immune system.⁴²⁷ A surge in maternal progesterone during the third trimester increases the abundance of Bifidobacteria abundance,⁴²⁸ which play an important role in the developing infant microbiota.⁴²⁹ Human studies suggest that pregnancy-specific microbiota rearrangements are most pronounced in the third trimester, regardless of maternal pre-pregnancy BMIs and gestational diabetes status.⁴²⁴ Intriguingly, transplantation of microbial strains from women in the third trimester induced greater weight gain and inflammation in germ-free recipient female mice than those isolated from women in their first trimester, which resemble increased adiposity and insulin resistance⁴³⁰ as well as low-grade inflammation⁴³¹ observed in the latter stages of gestation in healthy women. While such changes are typically associated with metabolic disorders in non-pregnant individuals, in the context of gestation they are thought to be beneficial metabolic adaptations required ensure proper fetal growth and development.⁴³² The precise mechanisms by which pregnancy shapes the maternal microbiome, how microbiota modulate the maternal environment, and the consequences of environmental exposures that disrupt the maternal gut microbiome and its remodeling on offspring development have yet to be fully elucidated. Further studies incorporating multi-OMIC approaches are warranted.⁴³³

Earlier, we introduced how maternal environmental factors, such as nutrition, shape both the maternal and offspring microbiome, with ramifications for development and long-term health outcomes in offspring. High-fat diet (HFD) exposures leading up to and throughout gestation increase the abundance of taxa involved in the biosynthesis of ketone bodies, fatty acid, vitamins, and bile acids.⁴³⁴ Additionally, HFD reduces the abundance of short-chain fatty acids (SCFAs) and SCFA-producing bacteria, while increasing pro-inflammatory markers lipopolysaccharides (LPS) and tumor necrosis factor (TNF), and promoting loss of intestinal epithelial barrier integrity, and placental hypoxia and inflammation.^435,436 Intriguingly, human studies show that overweight pregnant women harbor distinct microbiota compared to those of normal weight⁴³⁷ associated with alterations in several metabolic hormones and pregnancy metabolism.⁴³⁸ Taken together, these studies highlight the crucial role of the maternal microbiome in pregnancy-associated metabolic adaptations and make the case for more extensive investigation into how environmentally induced alterations in maternal gut microbial ecology influence physiological changes in offspring and their consequences for fetal development and programming of disease, including neurodevelopmental disorders.

We propose that precision targeting the maternal gut microbiome could provide for a healthier in utero environment, thus facilitating typical fetal development and reducing risk for adverse health outcomes. Select probiotics have the potential to remedy dysbiosis of the gut microbiome through a variety of mechanisms⁴³⁹ and are generally regarded as safe (GRAS) to administer during pregnancy.^440,441 In a recent study, maternal probiotic supplementation during gestation and lactation was found to promote intestinal barrier integrity and reduce inflammation.⁴⁴² A limited number of studies have begun to investigate the impacts of daily probiotic administration in obese pregnant women or animal models of diet-induced obesity⁴⁴³ with some of them showing increased gut microbial diversity,⁴⁴⁴ albeit small effects on metabolic parameters.⁴⁴⁵ Promisingly, a randomized double-blinded Danish study investigating the efficacy of a multi-strain probiotic versus placebo-control to regulate blood glucose, gestational weight gain, and reduce risk for gestational diabetes mellitus in obese pregnant women demonstrated >80% adherence to the probiotic regimen and an increase in alpha diversity of the gut microbiome of group receiving the probiotic intervention over time. In contrast, no increase in alpha diversity was observed in the placebo control group.

Both single- and multi-strain probiotic treatments have been used to target gut-brain-behavior interactions and ameliorate or prevent neuropsychiatric outcomes in human and animal studies.^446,447 Clinical trials have reported ameliorations of core and comorbid symptoms in autistic children after rebalancing the composition of the gut microbiome through probiotic interventions.^448,449 However, the efficacy of maternal probiotic supplementation to counteract the detrimental effects of diet-induced dysbiosis of gut microbiome on offspring neurodevelopment has not yet been explored. A recent study in CD-1 IGS mice⁴⁵⁰ showed that multi-strain probiotic (Bio-Kult Advanced® containing Bifidobacterium spp. and Lactobacillus spp.) exposure during the perinatal period reduced anxiety-like behaviors associated with maternal obesity, modulated the expression of genes involved in synaptic plasticity in the prefrontal cortex of offspring, and increased brain lactate and SCFA levels, which are known to regulate gene expression.¹⁴⁶ Additionally, the multi-strain probiotic decreased inflammation, as indicated by a reduction in circulating levels of pro-inflammatory cytokine interleukin-6 (IL-6) and increased SCFA production in obese dams treated with probiotics.⁴⁵⁰ This study also provided evidence of a critical role for probiotic species and their metabolites in the regulation of mood and behavior through changes in the expression of synaptic plasticity-related genes. In another study in which mouse dams were exposed to a pregnancy-specific dietary regimen, single-strain probiotic administration during the second trimester of pregnancy decreased anxiety-like behavior and modified cortical cytoarchitecture, with differential effects on male versus female offspring.⁴⁵¹ In the context of MIA models, a combination of pre- and probiotics (Bifidobacteria and Lactobacillus combined with fructooligosaccharides and maltodextrin) administered to pregnant dams prevented MIA-induced depression-like and ASD-like behaviors in adult offspring, while also reducing the abundance of proinflammatory cytokines levels in the fetal brain.⁴⁵²

The potential for early life probiotic interventions to prevent or treat ASD-associated phenotypes in nonsyndromic autism is further supported by successful intervention in models for genetic and idiopathic ASD. In the BTBR model, administration of probiotic Lactobacillus (L.) rhamnosus rescued social deficits and modulated the composition of the gut microbiome by increasing microbial richness and the abundance of SCFA-producing taxa. Notably, however, in a randomized, placebo-controlled, cross-over study in human males, the JB-1 strain of L. rhamnosus failed to modulate stress or cognitive performance.⁴⁵³ Not long after, treatment with the probiotic Limosilactobacillus reuteri ATTC-PTA-6475 in the Cntnap2^{−/− 140} and Shank3B^−/− mouse models for disorders of social dysfunction was shown to rescue ASD-like behavior and underlying deficits in synaptic plasticity.¹⁴¹ Excitingly, in a recent double-blind, randomized, placebo-controlled clinical trial, this same human-derived strain selectively reversed social deficits in children with ASD when given in combination with its parent strain, Limosilactobacillus reuteri DSM-17938.¹⁴² Thus, the promise of preclinical discoveries that precision targeting of the gut microbiome can relieve some core ASD symptoms is beginning to be realized in human patients. Therapeutic targeting of the gut microbiome for the prevention and treatment of neurodevelopmental disorders is a frontier ripe for discovery, innovation, and implementation.

Concluding perspective

Investigative teams around the world are beginning to elucidate the relative contribution of genetics and the environment, and their interactions, to the etiology of ASD and other neurodevelopmental disorders. Genetics are undoubtedly at the core of ASD pathology, and promising treatment options are emerging for monogenic ASDs, such as anti-sense oligonucleotides (ASOs).¹⁹ However, in many cases, genetic variants could establish baseline risk for ASD, while environmental factors modify the phenotypic expression of genetic determinants, influencing disease severity and clinical manifestation. Therapeutically, this is very exciting given that the environment is modifiable and some changes can even be reversible. Environmental exposure-related epigenetic reprogramming and remodeling of the maternal gut microbiome provide new perspectives from which to investigate the mechanisms underlying some forms of neurodevelopmental disorders and could spur innovation in the development of therapeutic approaches. For instance, integrating microbiome-targeted (prebiotic, probiotic, and/or symbiotic) treatments for expectant mothers and/or infants could influence the maternal immune landscape and thereby facilitate healthy brain development in children. New tools for precision sampling and delivery of gut microbiota will likely be key to advancing diagnostics and developing breakthrough therapeutic approaches in coming years. Multidisciplinary investigations exploiting the power of integrated multi-OMICs in preclinical animal models for ASD and carefully stratified patient populations are warranted to advance our mechanistic understanding of how early life environmental exposures contribute to risk for ASD and could be key to the development of innovative diagnostics and interventions for ASD.

Funding Statement

The work was supported by the National Institute of Child Health and Human Development [R01 HD109095, R01 HD109780]; National Institute of Environmental Health Sciences [P30 ES030285]; Robert and Janice McNair Foundation; The John S. Dunn Foundation.

Disclosure statement

S.A.B. is an inventor on a patent granted to Baylor College of Medicine related to the use of Limosilactobacillus reuteri for treating disorders characterized by social dysfunction (US Patent No. 11135252). The authors declare no other competing interests.