Cutting Edge: Critical Roles for Microbiota-Mediated Regulation of the Immune System in a Prenatal Immune Activation Model of Autism

The cause of autism spectrum disorder (ASD) currently remains poorly understood; however, emerging clinical and experimental evidence suggests central roles for immune dysregulation in autism pathogenesis (1–3). For instance, reports have increasingly linked maternal immune activation (MIA) during pregnancy to ASD (3–6). In particular, pregnant mothers who suffer from autoimmune disorders, obesity-associated inflammation, or infectious disease are at a significantly higher risk of having children with ASD (3–6). It is believed that hyperactive immune responses are responsible for the elevated ASD risk that is associated with each of these inflammatory conditions. ASD-like phenotypes can be modeled in mice by treating pregnant females with the viral mimetic polyinosinic-polycytidylic acid (PolyI:C) between embryonic day (E) 11.5 and E12.5 of development (1). In this model system, offspring from PolyI:C–treated pregnant mice develop many of the defining features of ASD, including defects in social preference, communicative impairments, and repetitive/stereotyped behaviors (1, 7, 8). Our understanding of the immunological processes underlying neurodevelopmental abnormalities in the MIA model is still limited. Recent evidence indicates roles for IL-17a and IL-6 in promoting MIA-induced neurodevelopmental disease (1, 9–11); however, there are likely other immune pathways that contribute to altered neurodevelopment in this model.

In addition to immune dysfunction, ASD in humans has also been associated with dysbiosis and gastrointestinal inflammation (12–14), which has led to increasing speculation about a role for the microbiome in ASD. Furthermore, recent studies have identified pivotal roles for the microbiome in the regulation of neurologic disease progression, brain function, and neurodevelopment (15, 16). Given the extensive clinical evidence of dysbiosis in ASD (12, 13) and emerging data implicating key roles for the microbiome–gut–brain axis in neurologic disorders (15, 16), we were interested in determining whether differences in maternal microbiota composition affect the induction of ASD-related phenotypes in the MIA model. To investigate this, we capitalized on the well-described differences in intestinal microbial landscape that exist between C57BL/6 mice originating from The Jackson Laboratory (Jax) and Taconic Biosciences (Tac) (17, 18). Previous reports have shown that C57BL/6 mice from these vendors harbor distinct intestinal microflora and that these differences in microbiota landscape uniquely modify aspects of the immune response (17, 18). A well-described example of this is the skewing of T cell responses toward IL-17a production by the commensal segmented filamentous bacteria (SFB) in Tac mice (17).

In this study, we report that preconception microflora landscape dictates neurodevelopmental disease susceptibility in a gestational inflammation-based model of ASD. Moreover, we identify the microbiome as a pivotal modulator of maternal immune responses and demonstrate that blockade of IL-17a signaling during gestation ameliorates the development of neurodevelopmental abnormalities in MIA offspring.

All mouse experiments were performed in accordance with the relevant guidelines and regulations of the University of Virginia and approved by the University of Virginia Animal Care and Use Committee. C57BL/6 mice were obtained from either Jax or Tac. Mice were housed and behavior was conducted in specific pathogen-free conditions under standard 12-h light/dark cycle conditions in rooms equipped with control for temperature (21 ± 1.5°C) and humidity (50 ± 10%). Mice matched for sex and age were randomly assigned into experimental groups. All presented behavior data include combined results from 6 to 12 independent litters with between one and four mice per litter.

Tac bedding containing fecal samples was transferred three times per week to cages of Jax mice who were 3–6 wk old for a total of 2 wk. Mice were then mated and subjected to MIA, as described in detail below.

Genomic DNA was isolated from fecal pellets according to the manufacturer’s instructions (QIAamp DNA Stool Mini Kit; Qiagen). Quantitative PCR was performed as previously described using primers for SFB, Bacteroides, Prevotellaceae, Bifidobacterium spp., and total bacterial 16S rRNA genes (17, 19). Relative quantity of SFB was calculated using the Δ threshold cycle method and was normalized to the amount of total bacterial 16S rRNA in each sample.

Mice were mated overnight and the presence of a vaginal plug was designated as E0.5. Each pregnant dam was weighed and administered 20 mg/kg PolyI:C potassium salt (Millipore Sigma) or saline by i.p. injections on both E11.5 and E12.5. Pups were weaned from their mothers at postnatal day 21 and housed with same-sex littermates with two to five mice per cage.

Blood was collected by submandibular venipuncture from dams 48 h following the last injection with either 20 mg/kg PolyI:C or saline. Blood was allowed to clot for 60 min at room temperature. Serum was collected after centrifugation, and IL-17a serum cytokine levels were measured by ELISA according to the manufacturer’s instructions (eBioscience).

Monoclonal anti-mouse IL-17A neutralizing Ab (clone 17F3; Bio X Cell, West Lebanon, NH) was administered via i.p. injections (500 μg/mouse) 6 h prior to the first PolyI:C injection of MIA induction.

All behavioral testing was performed according to previously established behavioral methodologies (8, 9, 15) with minor modifications. Behavioral experiments were carried out during daylight hours in a blinded fashion.

On postnatal day 10, male pups were removed from their cages and habituated to the room away from their mother for 10 min. After the habituation period, mice were placed in a clean 1-l plastic flask. Ultrasonic vocalizations (USVs) were measured for 3 min using an UltraSoundGate GM16/CMPA microphone (Avisoft Bioacoustics) and recorded with SASLab Pro software (Avisoft Bioacoustics). USVs were measured between 25 and 125 kHz, and background recordings shorter than 0.02 ms were excluded.

Adult male mice (8–10 wk of age) were assessed for social preference using the three-chamber social approach test, which uses a three-chamber arena where chambers containing either a novel age- and sex-matched mouse (male C57BL/6) or a novel object (plastic blue ball) in wire cups are separated by an empty center chamber. Experimental mice were habituated to the three-chamber arena with empty wire cages for two 5-min sessions. One day later, mice were placed in the empty center chamber without access to other test arenas for 5 min. Following this exploration period, the barriers were removed and mice were allowed to freely roam between the three chambers for 10 min and were observed for interaction time with the targets in each chamber. Sessions were video recorded, and investigation time and distance traveled were tracked and analyzed using TopScan 3.00. The social preference index was calculated as the percentage of time investigating the novel mouse out of the total time investigating both the object and the mouse.

One week following the three-chamber social approach test, male mice were acclimated overnight to caging containing wood chip bedding. One day later, experimental mice were placed in a clean cage (arena size: 12 × 7 × 5 inches) that was filled with 2 inches of wood chip bedding. Twenty glass marbles were arranged on the top of the bedding in five rows of four marbles equidistant from one another. After a 15-min exploration period, mice were carefully removed from the cages and a marble burying index score was calculated based on the following scale: 1 for marbles covered >50% by bedding, 0.5 for ∼50% covered, or 0 for anything less.

All statistical analyses were performed using GraphPad Prism. Statistical significance was calculated by one-way ANOVA with Tukey post hoc tests or two-way ANOVA with Sidak or Tukey post hoc tests. Outliers were excluded if they fell more than two SD from the mean. All p values <0.05 were considered significant. The p values are denoted by *p < 0.05, **p < 0.01, and ***p < 0.001.

To better understand how alterations in maternal microbiota diversity impact the development of autism-related phenotypes, we treated C57BL/6 Jax and Tac mice with either PolyI:C to induce prenatal MIA or saline as a control at E11.5 and E12.5. We first assessed communicative irregularities by recording USVs in control and MIA pups following separation from their mothers. Offspring from PolyI:C–treated Tac dams made substantially fewer calls and vocalized for shorter total durations than pups from saline-treated Tac mothers (Fig. 1A, 1B). In contrast, MIA offspring from Jax dams did not display communicative deficits and vocalized as frequently and for the same duration of time as control offspring from both Jax and Tac mothers (Fig. 1A, 1B). Next, we evaluated the development of ASD-related abnormalities in social behavior using the three-chamber social preference test. Adult offspring from PolyI:C–treated Tac dams exhibited irregular social behavior, as indicated by the lack of preference for the novel mouse over the novel object in this test (Fig. 1C, 1D). In contrast, prenatal exposure to MIA did not appreciably influence sociability in Jax mice (Fig. 1C, 1D). Distance traveled in the sociability test was similar between all groups (Fig. 1E), suggesting that the abnormal social behavior detected in MIA offspring from Tac dams was likely not due to differences in overall activity or arousal. Repetitive and stereotyped behaviors are also hallmarks of autism; therefore, we next assessed repetitive/stereotyped behaviors in our experimental mice using the marble-burying test. In these studies, we found that Tac MIA mice exhibited excessive repetitive behaviors and buried significantly more marbles than control offspring from saline-treated mothers (Fig. 1F). In contrast, no significant differences in marble-burying behavior were detected between MIA and control Jax offspring (Fig. 1F). Taken together, these findings indicate that vendor-specific differences in C57BL/6 mouse colonies prominently influence the development of autism-related behaviors in the MIA mouse model of ASD.

We were next interested in ascertaining whether the observed differences in MIA-induced neurodevelopmental disorder susceptibility between Tac and Jax C57BL/6 mice were indeed due to differences in microbiota diversity and not the potential effects of genetic drift. To this end, we performed fecal transplantation studies in which Jax mice were exposed to the Tac microbiome for at least 2 wk prior to MIA induction. Following fecal transplantation of Tac microbiota into Jax mice, MIA was induced with gestational PolyI:C treatment, and then the development of ASD-related phenotypes, including social preference abnormalities and communicative deficits, were evaluated in the offspring. Interestingly, MIA offspring from Jax dams that were previously cohoused with Tac microbiota (referred to in this article as cohoused Jax [Co Jax] mice) exhibited abnormalities in social preference, whereas MIA did not promote social interaction deficits in conventionally raised Jax mice (Fig. 2A, 2B). Distance traveled during the sociability test was similar between all experimental groups, which suggests that the effects of fecal transplantation on autism-like behaviors were not due to altered mobility or arousal (Fig. 2C). Likewise, cohousing Jax mice with Tac microbiota before PolyI:C treatment was sufficient to confer communicative defects in MIA Jax offspring (Fig. 2D, 2E). Notably, we observed reduced numbers and duration of USVs in MIA offspring from Jax mice supplemented with Tac microbiota (Fig. 2D, 2E, data not shown). Collectively, these findings indicate key roles for the prenatal microbiome in shaping the development of autism-associated behaviors in the MIA model.

The commensal SFB in the Tac microbiome is known to promote induction of IL-17a inflammatory responses (17), which provides one plausible immune-based route through which the Tac microbiome can influence neurodevelopmental disease. Interestingly, recent studies demonstrate that IL-17a can modulate neurodevelopment, neural circuit function, and behavior (9–11, 20). Therefore, we were particularly interested in exploring a potential role for microbiota-mediated regulation of IL-17a inflammatory responses in driving autism-related phenotypes.

To first validate that our microbiota transplantation methodology results in the transfer of SFB to Jax mice, we evaluated the colonization of SFB in Co Jax mice. Consistent with previous reports (17), we observed levels of SFB colonization in Jax fecal transplantation mice (Co Jax mice) that were similar to those found in conventionally raised Tac mice (Fig. 3A). In contrast, we did not detect major changes in the relative abundance of Prevotellaceae, Bacteroides, or Bifidobacterium spp. in Jax mice following cohousing (Supplemental Fig. 1). Changes in microbiota composition in Co Jax dams also promoted enhanced IL-17a secretion following PolyI:C injection (Fig. 3B). These findings suggest that maternal microbiota landscape centrally impacts MIA-induced inflammatory cytokine production. Moreover, these results provide further rationale for testing a causal role for dysregulated gestational IL-17a production in driving the development of autistic phenotypes in offspring from Jax dams that underwent microbiota transplantation. To this end, we first sought to confirm a role for IL-17a in driving neurodevelopmental disorders in the MIA model. Consistent with a recently published study (9), we found that blockade of IL-17a in MIA mothers prevented Tac offspring from developing social preference abnormalities, stereotyped/repetitive behaviors, and communicative deficits (Supplemental Fig. 2).

Next, we investigated whether neutralization of IL-17a impacts the development of autism-related behaviors in Jax mice that were previously cohoused with microbiota from Tac mice. We found that inhibition of IL-17a signaling during gestation rescues social abnormalities in MIA offspring from Jax dams that previously underwent fecal transplantation with Tac microflora (Fig. 4A, 4B). Importantly, the amelioration of social impairments with anti-IL-17a treatment was not due to enhanced arousal or activity, as distances traveled were similar between the experimental groups (Fig. 4C). Neutralization of IL-17a during gestation in Co Jax dams was also effective in restoring normal communicative behavior in MIA offspring (Fig. 4D–F). In contrast, we observed substantial reductions in the number and duration of USVs in offspring from Co Jax dams that received sham treatment prior to MIA induction (Fig. 4D–F).

In summary, our findings implicate key roles for microbiota-mediated regulation of immunity in a prenatal inflammation model of autism. Moreover, our results identify IL-17a as a specific immune regulator that contributes to the effects of the microbiome on the development of ASD-related phenotypes in the MIA model. These findings suggest that targeting the maternal microbiome and/or immune system during pregnancy may offer therapeutic strategies to prevent some forms of neurodevelopmental disorders.

We thank members of the Lukens laboratory and the Center for Brain Immunology and Glia for valuable discussions.

The authors have no financial conflicts of interest.