Altered gut microbiome and autism like behavior are associated with parental high salt diet in male mice

Neurodevelopmental disorders are conditions caused by the abnormal development of the central nervous system. Autism spectrum disorder (ASD) is currently the most common form of such disorders, affecting 1% of the population worldwide. Despite its prevalence, the mechanisms underlying ASD are not fully known. Recent studies have suggested that the maternal gut microbiome can have profound effects on neurodevelopment. Considering that the gut microbial composition is modulated by diet, we tested the hypothesis that ASD-like behavior could be linked to maternal diet and its associated gut dysbiosis. Therefore, we used a mouse model of parental high salt diet (HSD), and specifically evaluated social and exploratory behaviors in their control-fed offspring. Using 16S genome sequencing of fecal samples, we first show that (1) as expected, HSD changed the maternal gut microbiome, and (2) this altered gut microbiome was shared with the offspring. More importantly, behavioral analysis of the offspring showed hyperactivity, increased repetitive behaviors, and impaired sociability in adult male mice from HSD-fed parents. Taken together, our data suggests that parental HSD consumption is strongly associated with offspring ASD-like behavioral abnormalities via changes in gut microbiome.


Scientific Reports
| (2021) 11:8364 | https://doi.org/10.1038/s41598-021-87678-x www.nature.com/scientificreports/ is a worldwide well-established cause of morbidity and mortality associated with hypertension, cardiovascular diseases as well as kidney failure 22 . However, HSD has not been fully addressed in the context of gut microbiomemediated prenatal effects on offspring neurodevelopment. Therefore, we report that 8-week chronic consumption of HSD alters the parental gut microbiome composition. Specifically, we found that HSD was associated with depletion of Lactobacillus spp. and enrichment of Akkermansia in mice fecal samples. We also observed that this altered gut microbiome is transferred to the offspring, which shared the same gut microbial composition with their parents until (at least) their weaning age. Finally, we found that this "inherited" modified gut microbiome is associated with abnormal social behaviors and hyperactivity in adult male mice only, not in females. Together, our results support the notion that ASD-like behaviors can be linked to parental HSD-induced gut dysbiosis.

Methods
Animals. Male and female wild-type C57Bl/6 mice (3-4 weeks-old) were purchased from The Jackson Laboratory (Bar Harbor, ME, US), and housed at Texas Tech University' animal facility. Mice were group housed (4-5 same-sex mice/cage) in ventilated cages in a temperature-controlled room (21-23 °C) with 12:12 h light/ dark light cycle (lights on 07:00-19:00), humidity of 40-60%, and food and water available ad libitum. All experiments and procedures were performed observing the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines and following relevant guidelines and regulations approved by the Texas Tech University Institutional Animal Care and Use Committee (IACUC). Mice were randomly assigned to control or HSD groups and were fed chow supplemented with 0.1% NaCl (Control, Teklad Custom Diet TD-94268) or 8% NaCl (HSD, Teklad Custom Diet TD-180241) respectively for 8 weeks. Both diets contained ~ 23% calories from protein, 62% from carbohydrate and ~ 15% from fat. The HSD groups were also given 1% NaCl in their drinking water whereas the control groups drank regular tap water. After 8 weeks male and female mice were used for behavioral assays while others were paired for breeding. All the experiments were performed in a separate room adjacent to the room where mice were housed.

Behavioral analysis.
Open field test (OFT). The open field test was done using a large plastic container measuring 12in width × 18in length × 12in height. Individual mice were placed inside and allowed to freely explore the arena for 5 min. The session was videotaped and then analyzed using Ethovision XT tracking system, which was used to analyze all behavioral tests (Noldus, Leesburg, VA). Several parameters were quantified: Total distance traveled, speed of movement and time spent in the center versus periphery regions. Additionally, the number of rearing events and cumulative grooming time were measured. This test was done on both parents and offspring.
Elevated plus maze test (EPM). The EPM test was performed using a four-arm plus shaped maze elevated 30 inches from the ground, with two arms with walls and two arms open. The arm length was ~ 20 inches with a 4-inch center square between all four arms. Individual mice were placed in the center part and then were allowed to freely explore for 5 min while the session was videotaped. The time the mouse spent in the closed arms versus the open arms was quantified for detecting avoidance behavior 23 .
Novel object recognition (NOR) test. The NOR test was performed in the same plastic container used for OFT. The test consisted of two consecutive sessions, a first session used for familiarization and a second session used as a test for memory acquisition. In the familiarization session, two identical objects were placed in the bottom of the container, equidistant from each other and the side walls. Then a single mouse was placed in the arena and allowed to freely explore for 5 min. After this session, the mouse was returned to its home cage. For the test session one of the identical objects was replaced with a novel unfamiliar one. The test session was performed in two phases. For evaluating short term memory, the mouse was placed inside the container 1 h after familiarization. For long term memory, the mouse was placed inside the container 24 h after the familiarization session. In both phases, each mouse was allowed to explore the objects for 5 min while the session was videotaped. The time each mouse spent exploring the familiar and novel objects was quantified and a preference index was calculated using the following equation PI = [time spent exploring novel object/(time exploring familiar object + time exploring novel object)] × 100% 24 .
T-maze test. The T-maze test was performed using a T shaped maze, 30 inches elevated from the ground. The total length of the top arm was 44 inches and the bottom arm was 20 inches. All arms had walls. For the test, a single mouse was placed in the outermost corner of the bottom arm and was allowed to explore the whole maze for 5 min while videotaping the session. Working memory was evaluated by quantifying the number of spontaneous alternations between entering the three different arms 25 .
3-chamber sociability test. The 3-chamber sociability test was done in the offspring generation to evaluate possible sociability deficits 26 . The test was done in a plastic container (15in × 19in × 10in) divided in three separate chambers by two inner partitions which had a single small opening to allow mice to move between chambers. The experiment was performed in two consecutive sessions, the first for training and the second for testing. In the training session, a small metal mesh box was placed in both the left and right chambers. An individual subject mouse was placed in the middle chamber and was allowed to explore the entire container for 5 min, then removed and kept in an empty cage while the container was prepared for the following session (test). For the testing session, an unfamiliar mouse (i.e. not a cage mate) of the same sex was placed inside one of the metal mesh boxes and kept there as "social (S) chamber". The mesh box in the other lateral chamber was kept empty Fecal sample collection and 16s rRNA Illumina sequencing. Colonic fecal samples were collected from the dams of both diet groups before weaning. The fecal samples from the offspring groups were collected before weaning at 2-3 weeks of age (i.e. while they were still housed with their mothers). To reduce stress, mice were placed inside autoclaved plastic boxes where they freely defecated. Feces were then collected and placed in autoclaved cryotubes, flash frozen in liquid nitrogen and immediately stored at − 80 °C. DNA extraction from the samples, Polymerase Chain Reaction (PCR) amplification of the variable 4 (V4) region of the 16s rRNA gene and sequencing using the Illumina MiSeq platform were performed by the Molecular Research LP laboratories (MR DNA, Shallowater, TX, USA).
Sequencing data processing and analysis. The data obtained from sequencing was processed using the MR DNA data processing pipeline (MR DNA, Shallowater, TX, USA). Sequences shorter than 150 bp, sequences with ambiguous base call and sequences depleted of barcodes and primers were removed. Operational taxonomic units (OTUs) were defined by clustering at 97% similarity followed by the removal of chimeras. OTUs were taxonomically classified using BLASTn against a curated database derived from RDPII and NCBI. All samples were run concurrently by blinded experimenters. Raw data have been deposited under the umbrella Bio-Project: PRJNA713927. Data are publicly accessible at https:// www. ncbi. nlm. nih. gov/ biopr oject/ PRJNA 713927.
Blood sample preparation. After weaning, female mice from the parental generation were anesthetized with isoflurane and euthanized by decapitation. Trunk blood samples were rapidly collected in autoclaved tubes, and the serum was separated by centrifugation (3000 rpm for 30 min). Proinflammatory cytokines IL-1β, IL-10, IL-17, IL-23, and TNF-α were detected using a LEGENDplex Mouse Inflammation Panel (Biolegend, CA, USA).
Statistical analysis and data availability. All behavioral analysis was done using Ethovision XT software (Noldus, Leesburg, VA). Data analysis and statistical analysis was done using Origin 2017 software (Origin-Lab Corporation). For determining statistical differences between two groups, paired and unpaired Student's t-test was used. Whereas difference between multiple groups was performed by one-way ANOVA followed by Tukey's post-hoc test. Differences between groups were considered statistically significant at p < 0.05. The datasets from behavioral experiments generated during the current study are available in the DRYAD repository https:// datad ryad. org. The following link can be used to access it: https:// datad ryad. org/ stash/ share/ PNwbY bu9Lh eZuwI Cm11V 91LCi btfEq jnR1a n1VIR PCE.
After our 8-week feeding protocol a subset of mice was paired to breed, control-fed males and females (n = 14 pairs), or HSD-fed males and females (n = 15 pairs). We kept each breeding pair fed with the same food of their feeding protocol until their offspring reached weaning age. We did not observe any significant difference in fertility (Fig. 1c, 92% pregnancy rate in control-fed vs. 100% in HSD-fed pairs, p = 0.37, unpaired T-test, t-score = 1.414), as well as number of pups born per litter (Fig. 1d In addition, we evaluated working memory using the T-maze alternation test, and spatial memory using the NOR test (both short-and long-term). Figure 2d shows that control and HSD-fed mice displayed a similar performance in the T-maze. On average, the spontaneous alternation in the control-fed male group was 57.15 ± 3.32% versus 55.03 ± 3.31% in the HSD-fed male group (p = 0.98, q-Value = 0.58; F (3,29) Value = 0.58, Prob > F = 0.63). Comparably, the control-fed females showed 50.78 ± 4.33% alternation versus 52.65 ± 3.81% in the HSD-fed     Altered gut microbiome found in HSD-fed parental female mice is shared with their offspring. Maternal microbiome can transfer to the offspring during birth 6 . It has also been shown that elevated salt consumption can lead to changes in the maternal gut microbiome composition 17,18 . Therefore, we reasoned that it is possible that our feeding protocol could be altering the maternal gut microbiome and that this microbiome could be then transferred to the offspring. To test this hypothesis, we collected fecal samples from females in the parental group (i.e. from both control and HSD-fed mothers or dams) and their offspring mice (i.e. at 2-3 weeks old, before weaning). We then performed 16 s rRNA amplicon sequencing (see "Methods" section for details) to obtain an accurate description of their gut microbiome composition. Results in Fig. 3a show that alpha-diversity (calculated based on observed operation taxa units or OTUs) was not significantly different, although there was a trend towards reduction in HSD-fed dams. The average OTUs observed in the control-fed group of dams was 2508.2 ± 104.74 versus 2129.6 ± 139.41 in HSD-fed counterparts (n = 5 in both control/HSD-fed, p = 0.11, q-Value = 3.34, F (3,26) Value = 15.55, Prob > F = 5.41E−6). Similarly, we found no differences between offspring mice. The average OTUs in the offspring from control-fed parents was 1775.2 ± 72.72 versus 1630.9 ± 80.50 in offspring from HSD-fed groups (total n = 10 for both groups, 5 females and 5 males in each; p = 0.59, q-Value = 1.801). However, we found a significant reduction in observed OTUs when comparing dams to their respective offspring (Fig. 3a, Control-fed dam vs. offspring from control-fed dam, ***p = 8.80E−5, q-Value = 7.47; HSD-fed dams vs. offspring from HSD-fed parents **p = 0.006, q-Value = 5.08). Further, our microbiome data analysis showed a significant difference in beta diversity between parental HSD and control-fed samples as well as their offspring (Fig. 3b), based on the non-phylogenetic Bray Curtis distance matrix (ADONIS of Bray Curtis: R 2 = 0.48, F (3,26) Value = 7.86, **Prob > F = 0.001) 29 . The PCoA graph shows that data points from HSD-fed dams were closely clustered, indicating comparable microbial diversity. Moreover, when comparing this group with control-fed dams, we observed that while control-fed mice also clustered together, they did so significantly far from the HSD-fed parental group. Lastly, the offspring from HSD and control-fed dams were also clustered in distance from each other whereas HSD-fed parents and their offspring are clustered in closer proximity (Fig. 3b).
Based on relative abundance, we listed the 14 most represented bacterial genera in the gut microbiome of both parental and offspring generations (Fig. 3c) and observed that the abundance of specific bacterial genera was significantly altered. Further, the genus Lactobacillus showed the greatest decrease in dams after 8 weeks on HSD compared to control-fed females, thus confirming previous findings 17,18,30 . On the contrary, the genus Akkermansia showed a robust increase in HSD-fed groups. Our data shows two important results. First, there were clear differences in the abundance of specific bacterial genera in dams fed with either control or HSD. Second, those differences were clearly maintained between parents and offspring mice fed with the same diet, which supports the notion of mother-to-infant microbial transmission 31,32 . Furthermore, we specifically analyzed the presence of Lactobacillus spp. and found that these bacteria were significantly reduced in both HSD-fed dams and their offspring (Fig. 3d). Our analysis showed that 25.54 ± 2.04% of the total bacterial abundance in the control-fed dams was Lactobacillus spp whereas in the HSD-fed parental groups this was reduced to 7.47 ± 2.62% (F (3,26) Value = 16.41, Prob > F = 3.47E−6, ***p = 4.96E−4, q-Value = 6.54). The offspring groups showed similar differences in bacterial abundance: % of Lactobacillus spp in offspring from control-fed dams = 17.08 ± 2.8% whereas in the offspring from HSD-fed dams this % was 4.18 ± 0.61% (***p = 4.41E−4, q-Value = 6.60). On the contrary, Akkermansia spp. was found significantly increased in the HSD-fed dams and their offspring. As shown in Fig. 3e, Akkermansia spp % in control-fed dams was 9.19 ± 3.21% compared to 25.79 ± 0.95% in the HSD-fed parental group (F (3,26) Value = 12.76, Prob > F = 2.55E−5, ***p = 0.001, q-Value = 6.90). The offspring groups showed a similar increase in Akkermansia spp: % in offspring from control-fed dams = 3.52 ± 1.02% versus 27.52 ± 4.84% found in offspring mice from HSD-fed dams (***p = 4.44E−5, q-Value = 7.85). Our microbiome analysis also showed no difference in the bacterial abundance content of Bacteroides spp (Fig. 3f). The % of Bacteroides spp in control-fed dams was 7.42 ± 1.59% whereas in the HSD-fed females this % = 14.84 ± 4.90% (F (3,26)  www.nature.com/scientificreports/ Overall, our data also suggests that as the maternal gut microbiome is likely transferred to the offspring during the first postnatal weeks, it provides a route for shaping the initial microbial population in the offspring that could be altered according to the maternal diet. [19][20][21] . Therefore, we collected trunk blood samples from a group of dams after several weeks in HSD or control diet and measured specific cytokine levels. Blood collection was carried out at the time litters were weaned at postnatal day 28. Figure 4a shows that in HSD-fed mice the IL-17 concentration was not different from control-fed mice. On average we measured 19.68 ± 4.58 pg/ml in control versus 17.33 ± 3.77 pg/ml in HSD-fed mice (control, n = 9, HSD, n = 9, p = 0.60, t-score = 0.39, unpaired t-test). In addition, we measured the concentration of IL-1β, IL-23, TNF-α, and IL-10. Similar to IL-17, neither TNF-α nor IL-10 were different between diet treatments (Fig. 4b-c). We measured 32.37 ± 10.84 pg/ml of TNF-α in control versus 24.04 ± 10.39 pg/ml in HSD-fed mice (control n = 9, HSD n = 8, p = 0.50, t-score = 0.55). IL-10 was 64.91 ± 30.38 pg/ml in control versus 114.83 ± 31.06 pg/ml in HSD-fed female mice (control, n = 6, HSD, n = 8, p = 0.28, t-score = − 1.12). Furthermore, we found that while there was trend towards reduced values, IL-1β and IL-23 were not statistically reduced in HSD-fed dams (Fig.  d-e). In fact, the concentration of IL-1β was 156.62 ± 41.35 pg/ml in control versus 78.73 ± 26.04 pg/ml in HSDfed female mice (control, n = 9, HSD, n = 9, p = 0.13, t-score = 1.60). IL-23 concentration was 156.43 ± 40.00 pg/ml in control and 90.65 ± 27.73 pg/ml in HSD-fed female mice (control, n = 8, HSD, n = 9; p = 0.19, t-score = 1.36). Taken together, these results indicate that while our HSD protocol resulted in altered gut microbiome composition, it did not increase the serum concentration of proinflammatory molecules.

Adult male offspring from HSD-fed mice show hyperactivity, reduced sociability and increased repetitive behaviors.
We investigated potential long-term effects of parental HSD on their offspring behavior, focusing on exploration, repetitive and social behaviors. At the time of weaning, we transitioned all offspring mice to control diet and regular water (i.e. no added NaCl) to avoid direct effects of diet on their behavior.
We first monitored body weight weekly from their weaning age until 8-10 weeks old when behavioral experiments were conducted. At 4-weeks old, both male and female offspring from HSD-fed parents were significantly heavier than offspring from control-fed pairs. However, the difference in body weight persisted only in males (Fig. 5a). At 4 weeks old the average weight in males from control-fed parents was 18.39 ± 2.02 g versus 20.44 ± 1.79 g in male offspring from HSD-fed mice ( Fig. 5a and Supplementary Figure 1a Previous research has shown that altered gut microbiome is associated with behavioral signs of ASD 33,34 . The offspring from our HSD-fed parents showed altered gut microbiome during juvenile stages, thus we hypothesized that these changes might lead to abnormal behavior. We therefore characterized social and exploratory behaviors in offspring from control and HSD-fed mice using the OFT. Figure 5b shows that male offspring from HSD-fed parents explored significantly larger area. On average, male mice from control-fed parents moved 1293.94 ± 77.99 cm versus 1640.09 ± 105.85 cm covered by male offspring from HSD-fed parents (n = 13/14 for male offspring from control/HSD-fed parents; F (3,48) = 3.06, Prob > F 0.03, *p = 0.04, q-Value = 3.85). These differences, however, were not observed in females. The area covered by females from control-fed parents was 1328.07 ± 107.35 cm versus 1468.10 ± 67.55 cm in females from HSD-fed parents (n = 12/13 for female offspring from control/HSD-fed parents; p = 0.71, q-Value = 1.49). Furthermore, male offspring from HSD-fed parents moved faster (Fig. 5c). The average speed in males from control-fed parents was 4.14 ± 0.90 cm/s versus 5.52 ± 1.22 cm/s in male from HSD-fed parents (F (3,48) = 3.58, Prob > F = 0.02, *p = 0.04, q-Value = 3.83). These differences were again not statistically different between female offspring (speed in females from control-fed parents = 4.43 ± 1.24 cm/s; speed in females from HSD-fed parents = 5.41 ± 1.78 cm/s; p = 0.27, q-Value = 2.58). We also quantified other behaviors such as avoidance (i.e. time in the center of the OFT arena), rearing and grooming (Supplementary Figure 2) and found no significant differences between offspring from control and HSD-fed parents. These results indicate that male offspring from HSD-fed mice show hyperactivity, one of the prominent signs of ASD-like behavior in mouse models 35 .
To characterize repetitive behavior we used the marble burying test 27 . We found that only male offspring from the HSD-fed parents buried a significantly higher % of marbles compared to male mice from control-fed parents. Figure 5d shows that, on average, male mice from control-fed parents buried 59.81 ± 4.10% of marbles versus 77.14 ± 3.14% in male mice from HSD-fed parents (F (3,46) = 5.03, Prob > F = 0.004; *p = 0.04, q-Value = 3.93). Female mice from the control-fed group buried 65 ± 4.99% of marbles compared to 54.42 ± 5.87% in female mice from HSD-fed parents (p = 0.41, q-Value = 2.19).
We also evaluated social behavior in the offspring generation because it has been shown to be modulated by the commensal microbiome, and deficits in social behavior are key signs of ASD 36,37 . Therefore, we subjected our mice to a 3-chamber sociability test (Fig. 5e)  www.nature.com/scientificreports/ (Fig. 5f). In the 3-chamber test we quantified the time mice spent exploring the social (S) versus the non-social (Non-S) chamber. Figure 5e shows that the male offspring from HSD-fed parents showed no preference for the social chamber compared to the male offspring from the control-fed group that spent more time in the S chamber versus NonS. On average, the male offspring from control-fed parents spent 147.31 ± 9.85 s in the S   , c) Male offspring from HSD-fed parents show higher distance traveled and moving speed in the OFT (n = 13 male offspring from control-fed parents, 14 male offspring from HSD-fed parents, 12 female offspring from control-fed parents, 13 female offspring from HSDfed parents). In the marble burying test, male offspring from HSD-fed parents show higher repetitive behavior (d). Offspring from HSD-fed parents show lower preference for the Social (S) chamber in the 3-chamber test compared to offspring from control-fed parents, (e) but no difference was observed in the urinary pheromone test (f). Data are described as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA and Tukey's test. In the urinary pheromone test (Fig. 5f), all groups behaved comparably, and no statistically significant difference was observed. On average, the % of time interacting with the urine-soaked swab in male mice from controlfed parents was 34.08 ± 5.38% versus 42.51 ± 5.69% in male mice from HSD-fed parents (p = 0.69, F (3,48) = 0.43, Prob > F = 0.73; q-Value = 1.54). Likewise, the time interacting with the urine-soaked swab in female offspring from control-fed parents was 38.04 ± 5.96% versus 36.22 ± 4.83% in female offspring from HSD-fed parents (p = 0.99, q-Value = 0.29).
Altogether our results show important sex-dependent long-term effects associated with parental HSD. Specifically, male offspring from HSD-fed parents showed hyperactivity, social preference deficits and increased repetitive behavior. Interestingly, these alterations closely resemble other animal models of ASD 36 suggesting the possibility that HSD could be an environmental risk factor for neurodevelopmental disorders.

Discussion
In this study we showed that: (1) the altered parental microbiome is shared with their offspring (likely transmitted during birth and first postnatal weeks), (2) these changes can be detrimental for neurodevelopment, and (3) altered gut microbiome is associated with ASD-like behavioral abnormalities in the offspring adulthood, changes that are only visible in male mice.
Previous studies have shown that HSD can modulate body weight, however, the effects have varied from no change, to either increase or decrease [38][39][40] . These discrepancies have been mostly attributed to differences in experimental design, i.e. salt content (4%, 7% or 8% in different studies), feeding period length (2, 4, 7 or 8 weeks) or nutritional differences between diets, most notably different fat content (10, 14 or 60%) [40][41][42][43] . Our results showing that HSD-fed male mice had a lower body weight replicate previous findings and agree with other studies using similar fat content. Our experiments also expanded these results to females, which were found to not be affected, in another example of sex-dependent differences. Unfortunately, there is limited literature addressing response to HSD in female mice. A recent study in female rats showed that HSD prevented weight gain caused by a high-fat diet 40 , while causing no effect in low-fat conditions. This reduction in weight gain was associated with adipocyte size and reduction in leptin levels 40 . Our results further emphasize the need for including sex as a biological variable in future studies examining how environmental factors (such as food) influence metabolism and neurodevelopment.
HSD has been associated with cognitive impairment due to several mechanisms including increasing oxidative stress, reducing the expression of synaptic proteins such as synapsin and CamK (Calcium-calmodulin dependent protein kinase), as well as by reducing the resting cerebral blood flow and nitric oxide production in cerebral endothelial cells 19,44 . While these studies have implemented HSD of varied duration (i.e. 4 to 24 weeks long), they have reported short-term memory impairment after 7 weeks of HSD 19,44 , and long-term memory impairment even after 4 weeks on HSD 44 . Moreover, another study showed that HSD can impair cognitive function in an age dependent manner, namely HSD was associated with impaired cognitive function only in 20-months old rats but not in young adults (2-months old) 45 . In our study, we started HSD feeding at 4-weeks old and after 8 weeks (i.e. at 12 weeks of age), we only observed a trend towards reduced short-term memory function in the NOR test in male mice, but the difference was not statistically significant from the control-fed group. The discrepancy with previous findings might be related to the much younger starting age in our experiments. We also investigated stress and anxiety-related behaviors in HSD-fed mice and found no significant differences, which is in line with previous findings 45 . Taken together, our data indicates that HSD-fed mice were not significantly different from control-fed after 8 weeks of HSD. Importantly, our study also included female mice and showed that females were not affected by HSD (at least under our experimental conditions). Nevertheless, our findings warrant more experiments using female mice.
Dietary habits have been shown to change the gut microbiome composition 16,46,47 . For instance, short-term consumption of entirely animal or plant-based food resulted in completely different gut microbiome community structure in humans 48 . In this context, several recent studies showed that feeding rodents with HSD for 4-8 weeks changed the abundance levels of specific microbial genera 17,18,30 . It has also been found that while beta diversity was significantly different in mice fed with HSD 17 , alpha diversity was not different between HSD and control-fed groups 17,18,30 . Our data were consistent with these results, namely our HSD-fed dams had reduced abundance of Lactobacillus and a significant increase in the abundance of genus Akkermansia compared to control-fed mice. However, we also found that our HSD-fed mice did not show a comparable increase in proinflammatory cytokines. It has been shown that HSD can alter the immune homeostasis by increasing the production of proinflammatory cytokines such as IL-17, IL-23, IL-1β, TNF-α [19][20][21] . This difference could be attributed to methodology employed since we performed flow cytometry-based multiplex assay for serum cytokines detection, whereas other studies used flow cytometry for measuring the differentiation of CD4 + T cells into proinflammatory T-cells 18,20,21 . However, another possible explanation for the observed lack of increased cytokines could be the abundance of Akkermansia spp in the gut of our HSD-fed mice. Akkermansia spp is considered as the next-generation probiotic due to its effects on glucose and lipid metabolism, and specially as potent antiinflammation agent 49,50 . Several studies have shown that Akkermansia spp can reduce inflammation by a number of different mechanisms such as reducing plasma level of lipopolysaccharide (LPS)-binding protein (LBP), www.nature.com/scientificreports/ reducing the expression of inflammatory genes, and reducing the production of pro-inflammatory cytokines such as IL-17, IL-23, IL-8, TNF-α as well as increasing the production of anti-inflammatory cytokines such as IL-10 and IL-12 [51][52][53][54][55][56] . Furthermore, while low levels of Akkermansia were found in obesity and type-2 diabetes mouse models, exogenous administration of Akkermansia increased the control of gut inflammation and permeability by regulating tight-junction-related proteins, and thickening the intestinal mucous layer 57 . Accordingly, the commonly used diabetes treatment Metformin has been reported to increase Akkermansia spp abundance 58 , and to significantly improve glucose metabolism in high-fat diet fed mice while also increasing the number of mucin-producing goblet cells 59 . Moreover, the population of Akkermansia in the gut is negatively modulated by the fat content of the consumed diet 54,60,61 . Our HSD has a fat content of approximately 15% which is considered low-fat 62 . Therefore, the apparent lack of inflammation in our HSD-fed mice could originate from the increased presence of Akkermansia spp in their gut. According to several studies, the maternal gut microbiome during pregnancy is vertically transmitted to the offspring during the childbirth and thus it is uniquely poised to shape the early-life microbiome in the newborn 1,6,63 . Initially after birth, the infant microbiome consists mostly of microbes from maternal skin and/ or birth canal (according to birth conditions), but the microbes from maternal gut remain more constant in the infant gut 31,32 . In our study, we observed similar changes in microbial composition in the offspring of either control or HSD-fed mice, supporting the idea that the original microbiome is first dictated by the mother's microbiome, and that it stays relatively unchanged during the first 2-3 postnatal weeks.
Research using germ-free mice has revealed that the gut microbiome present during early developmental stages is crucially important for the nervous and immune system development, brain function and ultimately behavior of the offspring 64 . Further, alterations in the early microbiome linked to maternal diet have been associated with neurodevelopmental disorders such as ASD 65 . For example, a recent study showed that female mice fed with a high fat diet for 8-weeks produce offspring that display several behaviors indicative of ASD 66 . Further, sequencing of the gut microbiome of both dams and their offspring showed significant alterations due to high fat diet, and the behavioral abnormalities were reversed by co-housing the offspring with offspring from controlfed dams 66 . Our data also showed that HSD-fed dams had a different gut microbiome compared to control-fed females. Therefore, we hypothesized that the offspring of HSD-fed mice could indeed display behavioral abnormalities resembling ASD. Our results supported this idea and showed a significant sex-dependent effect: only males consistently performed differently from offspring from control-fed parents. Specifically, the male offspring from HSD-fed parents showed increased locomotion, excessive marble burying as a compulsive behavior and reduced social interaction, all behavioral abnormalities resembling ASD-like phenotype in mice 36,67 .
Though the initial gut microbiome is essential for normal brain development, how the HSD-associated decrease in Lactobacillus leads to ASD-like behavioral abnormalities, is still unclear. Gut Lactobacillus was found to regulate emotional behavior and GABA expression in mice 68 . Further, the Shank3 KO mice (established ASD model) has reduced gut Lactobacillus abundance as well as reduced expression GABA receptors in hippocampus and prefrontal cortex, changes that were reversed by the probiotic administration of a strain of Lactobacillus spp 69 . In addition, low levels of Lactobacillus were also associated with reduced levels of hypothalamic oxytocin in mice 66 . Our results show that the offspring from HSD-fed parents has reduced level of Lactobacillus, most likely passed from their mother. Therefore, we can speculate that low Lactobacillus abundance might hamper the expression of GABA receptors and secretion of oxytocin hormone in the male offspring from HSD-fed parents, which ultimately results in ASD-like behavior. More extensive studies are needed to dissect the mechanism underlying the association between low gut Lactobacillus abundance and ASD-like behaviors.