ABSTRACT

   Autism spectrum disorder (ASD) is a heterogeneous neurodevelopmental
   condition characterized by social communication deficiencies and
   stereotypic behaviors influenced by hereditary and/or environmental
   risk factors. There are currently no approved medications for treating
   the core symptoms of ASD. Human fecal microbiota transplantation (FMT)
   has emerged as a potential intervention to improve autistic symptoms,
   but the underlying mechanisms are not fully understood. In this study,
   we evaluated the effects of human-derived FMT on behavioral and
   multi-omics profiles of the BTBR mice, an established model for ASD.
   FMT effectively alleviated the social deficits in the BTBR mice and
   normalized their distinct plasma metabolic profile, notably reducing
   the elevated long-chain acylcarnitines. Integrative analysis linked
   these phenotypic changes to specific Bacteroides species and vitamin
   B[6] metabolism. Indeed, vitamin B[6] supplementation improved the
   social behaviors in BTBR mice. Collectively, these findings shed new
   light on the interplay between FMT and vitamin B[6] metabolism and
   revealed a potential mechanism underlying the therapeutic role of FMT
   in ASD.

IMPORTANCE

   Accumulating evidence supports the beneficial effects of human fecal
   microbiota transplantation (FMT) on symptoms associated with autism
   spectrum disorder (ASD). However, the precise mechanism by which FMT
   induces a shift in the microbiota and leads to symptom improvement
   remains incompletely understood. This study integrated data from
   colon-content metagenomics, colon-content metabolomics, and plasma
   metabolomics to investigate the effects of FMT treatment on the BTBR
   mouse model for ASD. The analysis linked the amelioration of social
   deficits following FMT treatment to the restoration of mitochondrial
   function and the modulation of vitamin B[6] metabolism. Bacterial
   species and compounds with beneficial roles in vitamin B[6] metabolism
   and mitochondrial function may further contribute to improving FMT
   products and designing novel therapies for ASD treatment.

   KEYWORDS: fecal microbiota transplantation (FMT), autism spectrum
   disorder (ASD), BTBR, social deficit, vitamin B[6 ]metabolism, fatty
   acid metabolism, Bacteroides, pyridoxal 5′-phosphate (PLP)

INTRODUCTION

   Autism spectrum disorder (ASD) constitutes a group of
   neurodevelopmental disorders characterized by impaired social
   communication and restricted, repetitive behaviors ([60]1). The
   prevalence of ASD in children in the United States has been steadily
   increasing, rising from 1 in 150 in 2000 to 1 in 36 in 2020, with males
   affected four times more frequently than females ([61]2). ASD was
   assumed to be caused by environmental factors ([62]3, [63]4) and
   genetic abnormalities ([64]5[65]–[66]7). The U.S. Food and Drug
   Administration (FDA) has approved only risperidone and aripiprazole for
   treating irritability in ASD patients ([67]8). There is an urgent need
   for innovative medications targeting the core symptoms of ASD,
   especially social deficits.

   Recent research has bolstered the concept of a functional gut-brain
   axis (GBA), establishing a connection between gut microbiota and brain
   activity in autism ([68]9, [69]10) as well as other neuropsychiatric
   disorders ([70]11, [71]12). Microbiome compositions differed between
   individuals with ASD and those with typically developing (TD) brains
   ([72]13[73]–[74]16), evolving as ASD individuals aged ([75]17), though
   inconsistencies exist between studies ([76]18). The beneficial impact
   of microbiome therapeutics on ASD-related behaviors has been explored
   in animal models ([77]19, [78]20) and ASD children ([79]21). Microbes
   may either trigger or alleviate ASD symptoms by interacting with
   intestinal epithelium ([80]22), immune and neural systems ([81]23),
   regulating neurotransmitters ([82]16, [83]24), and modulating amino
   acid metabolism ([84]25). Notably, fecal microbiota transplantation
   (FMT) has been applied to ASD children in a few clinical studies
   ([85]26[86]–[87]29), achieving promising results in both
   gastrointestinal (GI) symptoms and behaviors, with an efficacy rate of
   64.6% in some studies ([88]30). Due to the promising preliminary
   results of FMT in ASD ([89]26, [90]27), the FDA has fast-tracked an FMT
   therapy for children with ASD in 2019 ([91]31). Despite the clinical
   progress, the mechanisms and efficacy of FMT are likely donor- and
   recipient-dependent. Using genetically identical mice with
   human-derived FMT allows us to focus only on the donor part in the
   context of ASD. Understanding the mechanisms and key components of FMT
   could aid in refining FMT treatment and products for ASD.

   To investigate the efficacy of FMT in alleviating ASD symptoms,
   microbiota from a healthy adult male donor was administered to BTBR
   mice. FMT is commonly used in ASD mouse model studies to establish a
   causal relationship between the microbiome and host phenotype
   ([92]32[93]–[94]35). In this study, FMT successfully rescued social
   deficits while leaving stereotypic behavior unaffected in mice.
   Post-FMT treatment, significant changes occurred in both metabolome and
   metagenome profiles, with many alterations associated with disturbed
   fatty acid metabolism, indicative of mitochondrial dysfunction in the
   host. The multi-omics analysis also hinted at the potential roles of
   vitamin B[6] metabolism and associated Bacteroides species in the
   host-microbe interaction. In a subsequent proof-of-concept study, the
   effectiveness of vitamin B[6] in mitigating social behavior deficits in
   BTBR mice was confirmed. These findings contribute to our understanding
   of the mechanisms behind FMT treatment for ASD and propose vitamin B[6]
   metabolism as a potential therapeutic target in ASD management.

RESULTS

FMT ameliorated the impaired social behaviors of the BTBR mice

   FMT of the human microbiome was applied to the BTBR model to
   investigate the mechanism underlying FMT treatment for ASD. The
   selection of BTBR mice as an ASD model is based on the resemblance of
   their abnormal behaviors to the core deficits observed in ASD ([95]36,
   [96]37). Previous research has indicated that BTBR mice have a unique
   gut microbial community ([97]38, [98]39), and cohousing with C57BL/6J
   (B6) mice can ameliorate their social deficits ([99]40). In our study,
   the BTBR mice were administered either FMT from a male adult donor
   (BTBR+FMT group) or saline (BTBR+Saline group) five times within 3
   weeks ([100]Fig. 1a). As a control for ASD-like behaviors, B6 mice were
   given saline (B6+Saline) according to the same protocol. Two weeks
   after FMT administration, the three-chamber social test (3CST), open
   field test (OFT), and grooming test were performed to evaluate the
   efficacy of the treatment on the social, anxiety-like, and stereotypic
   behaviors of the BTBR and B6 mice ([101]41, [102]42), respectively.

Fig 1.

   [103]Fig 1
   [104]Open in a new tab

   Human FMT rescued the social deficit of the BTBR mice. (a) The timeline
   of fecal microbiota transplantation in mice and scheme of study design.
   (b) Schematic of the sociability task in a three-chamber social
   behavior test. (c) Interaction time of experiment subject with a
   stranger mouse or empty cage in sociability task. (d) Sociability index
   of the BTBR+FMT group, the BTBR+Saline group, and the B6+Saline group
   (n = 10 mice in each group). (e) Schematic of the social novelty task
   in a three-chamber social behavior test. (f) Interaction time of
   experiment subject with a familiar mouse or unfamiliar one in
   sociability task. (g) Social novelty index of the BTBR+FMT group, the
   BTBR+Saline group, and the B6+Saline group (n = 10 mice in each group).
   Data represent the mean ± SEM. Error bars indicate SEM. *P < 0.05, **P
   < 0.01, ***P < 0.001, ****P < 0.0001. Two-way ANOVA with Tukey’s post
   hoc test was used in panels c and f, and one-way ANOVA with Dunnett’s
   post hoc test was used in panels d and g. The graphical element in
   panels b and e were created using BioRender
   ([105]https://biorender.com).

   To validate the autism model, we compared the behavior test results of
   the BTBR+Saline and B6+Saline groups. In the sociability test, the
   duration of interaction for the experimental mouse was assessed, either
   with a stranger mouse (Mouse 1) or an empty wired cup ([106]Fig. 1b).
   As expected, B6+Saline mice exhibited normal social behavior, as
   evidenced by their preference for interacting with the unfamiliar
   mouse. In contrast, the BTBR+Saline mice did not prefer mouse 1,
   indicating impaired sociability (two-way ANOVA F[(2,54)] = 11.07, P <
   0.0001, Tukey’s post hoc test: P = 0.1180, [107]Fig. 1c). Meanwhile,
   the sociability index of the BTBR+Saline mice was significantly lower
   than that of the B6+Saline group (one-way ANOVA F[(2, 27)] = 23.98, P <
   0.0001, Dunnett’s post hoc test: P < 0.0001, [108]Fig. 1d). In
   addition, the social novelty analysis ([109]Fig. 1e) demonstrated that
   mice in the B6+Saline group spent significantly more time interacting
   with the novel mouse (Mouse 2) than in the BTBR+Saline group (two-way
   ANOVA F[(2,54)] = 4.536, P = 0.0151, Tukey’s post hoc test: P = 0.0028,
   [110]Fig. 1f; one-way ANOVA F[(2, 27)] =1.496, P = 0.0075, Dunnett’s
   post hoc test: P = 0.0134, [111]Fig. 1g). Compared to the B6 mice, mice
   in the BTBR+Saline group spent less time in the center zone and more
   time grooming themselves during the open field test (one-way ANOVA
   F[(2, 27)] = 1.063, P = 0.0413, Dunnett’s post hoc test: P = 0.0231,
   [112]Fig. S1a; one-way ANOVA F[(2, 27)] = 6.500, P < 0.0001, Dunnett’s
   post hoc test: P < 0.0001, [113]Fig. S1b), which indicated their
   anxiety-like and stereotypic behaviors, respectively.

   The study proceeded to assess the impact of FMT by comparing the
   behavioral improvements between the BTBR+FMT group and the BTBR+Saline
   group. Intriguingly, treatment with human FMT rescued the sociability
   of BTBR mice, as measured by the time spent with mouse 1 (two-way ANOVA
   F[(2,54)] = 11.07, P < 0.0001, Tukey’s post hoc test: P < 0.0001,
   [114]Fig. 1c), and elevated the sociability index in the BTBR+FMT group
   to the level of the control group (B6 mice) (one-way ANOVA F[(2, 27)] =
   23.98, P < 0.0001, Dunnett’s post hoc test: P < 0.0001, [115]Fig. 1d).
   Meanwhile, mice in the BTBR+FMT group showed a usual preference for
   social novelty, as they spent more time interacting with Mouse 2 (novel
   one) than Mouse 1 (familiar one) (two-way ANOVA F[(2,54)] = 4.536, P =
   0.0151, Tukey’s post hoc test: P = 0.0015, [116]Fig. 1f). The social
   novelty index of the BTBR+FMT group also increased significantly
   compared to the BTBR+Saline group one-way ANOVA F[(2, 27)] = 1.496, P =
   0.0075, Dunnett’s post hoc test: P = 0.0100, [117]Fig. 1g).

   Besides, mice in the BTBR+FMT group still displayed abnormal anxiety
   after FMT treatment, in which mice spent more time in the periphery
   than the center in the open field test (one-way ANOVA F[(2, 27)] =
   1.063, P = 0.0413, Dunnett’s post hoc test: P = 0.2987, [118]Fig. S1a).
   Meanwhile, FMT failed to alleviate the stereotypic behaviors of BTBR
   mice, as seen by the unchanged accumulated duration of self-grooming
   with FMT treatment (one-way ANOVA F[(2, 27)] = 6.500, P < 0.0001,
   Dunnett’s post hoc test: P = 0.9975, [119]Fig. S1b). This finding is
   unsurprising since social activity and stereotypic behaviors may
   undergo disparate mechanisms ([120]43, [121]44), and the FMT efficacy
   is also donor-dependent ([122]45).

   In summary, treatment with human gut microbiota selectively reversed
   ASD-like social deficit symptoms in BTBR mice.

The FMT-treated BTBR mice harbored specific bacterial species potentially
associated with enhanced social behaviors

   Investigating the impact of FMT on the colon-content microbiome profile
   in BTBR mice, we observed no significant difference in α-diversity
   between the BTBR+Saline and B6+Saline groups ([123]Fig. 2a). In
   contrast, BTBR mice treated with FMT showed significantly higher
   α-diversity values compared to both the B6 and BTBR mice treated with
   saline ([124]Fig. 2a). β-diversity analysis based on Bray-Curtis
   distance showed that samples from the BTBR+FMT group were shifted away
   from the other two groups (PERMANOVA test, R^2 = 0.441, P = 0.001,
   [125]Fig. 2b) and the BTBR+Saline samples separated from the B6+Saline
   group (PERMANOVA test, R^2 = 0.301, P = 0.002, [126]Fig. S2a). This
   suggested that the human-derived FMT caused substantial changes in the
   microbiome of the BTBR mice, which was also distinct from the normal
   microbiome profile of the B6 mice. Comparison between the BTBR+Saline
   and B6+Saline groups identified 13 differential species using LEfSe
   ([127]46) (LDA > 2, P < 0.05) ([128]Fig. S2b). Akkermansia muciniphila
   and Olsenella unclassified were significantly increased in the B6
   group. We then compared the taxonomic compositions between the BTBR+FMT
   and BTBR+Saline groups. Multiple species from Parabacteroides,
   Bacteroides, Alistipes, Bilophila, and so on, were present in the
   BTBR+FMT samples and the donor sample but depleted in the BTBR+Saline
   samples ([129]Fig. 2c; [130]Table S1), suggesting the donor origin of
   these species in the BTBR+FMT group. We performed differential analysis
   between the BTBR+FMT and BTBR+Saline groups using LefSe ([131]46) (LDA
   > 2, P < 0.05) and identified 23 species ([132]Fig. 2d). Most species
   enriched in the BTBR+FMT group belonged to the genera Bacteroides,
   Parabacteroides, and Alistipes, although species from genera Bilophila,
   Paraprevotella, and Desulfovibrio were also increased in the group
   ([133]Fig. 2d). The results were consistent with previous animal and
   clinical studies: B. ovatus and P. merdae were absent in offspring ASD
   mice ([134]32); Parabacteroides, Alistipes, and Bilophila were
   decreased in ASD individuals ([135]14); Paraprevotella and Bacteroides
   were found to decrease in ASD children ([136]47); and D. piger was
   found to increase in ASD children after receiving FMT treatment
   ([137]48). The enriched species have been recognized for their
   beneficial effects in ASD, including correction of gut permeability
   ([138]19), and modulation of neurotransmitters such as γ-aminobutyric
   acid (GABA) ([139]49, [140]50), glutamate ([141]50), and tryptophan
   ([142]51). The species decreased in the BTBR+FMT group included
   Eubacterium plexicaudatum, Parabacteroides goldsteinii, and
   unclassified species from Candidatus Arthromitus, Alistipes, and
   Oscillibacter ([143]Fig. 2d). C. Arthromitus was enriched in ASD mice
   and significantly decreased after FMT treatment ([144]52). P.
   goldsteinii has been reported to be beneficial to ASD-like behaviors in
   MIA offspring ([145]53), though it was more abundant in our study in
   the BTBR+Saline group ([146]Fig. 2c and d). The exact roles of these
   species in ASD require further investigation.

Fig 2.

   [147]Fig 2
   [148]Open in a new tab

   Human FMT altered the colon-content microbiome in the BTBR mice. (a)
   Alpha diversity (Shannon index) between mice in the B6+Saline,
   BTBT+Saline, and BTBR+FMT groups were compared (n = 10 per group).
   Horizontal lines represent the third quartile, median, and first
   quartile, respectively. The whiskers above and below the box show the
   maximum and minimum. (b) PCoA analysis of the microbiome profiles of
   the three groups B6+Saline, BTBR+Saline, and BTBR+FMT (R^2 = 0.279, P =
   0.006, PERMANOVA analysis). (c) Comparison of the genus-level taxonomic
   composition of the microbiome between the B6+Saline, BTBR+Saline, and
   BTBR+FMT groups. (d) LEfSe analysis showing differentially abundant
   bacterial species between the BTBR+FMT and the BTBR+Saline groups (LDA
   score cutoff set as 2.0). (e) Bacteria species contributing to the
   pathway of pyridoxal 5′-phosphate Biosynthesis I in the BTBR+Saline and
   the BTBR+FMT. The stacked bar indicates the bacteria-stratified
   abundance of the pathway within a single sample.

   Functional changes introduced to the microbiome by the FMT treatment
   were further examined using HUMAnN 2 ([149]54). From the pathway
   analysis results, we identified 24 differentially regulated pathways
   (|log[2](FC)| > 1, P < 0.05) ([150]Fig. S3a). HUMAnN 2 ([151]54) also
   provided a pathway abundance table stratified by taxa to illustrate the
   contribution of specific taxa in the pathways. Notably, the pathway
   “pyridoxal 5'-phosphate biosynthesis I” (pathway #11 in [152]Fig. S3a)
   was enriched in the BTBR+FMT group with multiple Bacteroides species
   (B. uniformis, B. eggerthii, B. xylanisolvens, B. thetaiotaomicron, and
   B. ovatus) involved ([153]Fig. 2e). Pyridoxal 5′-phosphate (PLP) is the
   active form of vitamin B[6], which acts as a co-factor involved in the
   biosynthesis of multiple neurotransmitters, including serotonin,
   dopamine, norepinephrine, epinephrine, and GABA ([154]55). Gut
   microbiota can mediate vitamin B[6] homeostasis and regulate
   autism-like behavior in a mice model ([155]56), but the
   generalizability of the model and the underlying mechanisms in ASD are
   not yet fully understood.

   We correlated the differential species with behavior indices across the
   BTBR mice samples to investigate the connection between ASD-like
   behaviors and microbiome members ([156]Fig. S4a). The results showed
   that the sociability index was positively correlated with most species
   enriched in the BTBR+FMT group and negatively correlated with those
   depleted in the BTBR+FMT group, suggesting strong variation of both
   microbiome abundance and sociability post-FMT treatment. In addition,
   the social novelty index was negatively associated with Alistipes
   unclassified, and positively correlated with Alistipes senegalensis.
   The anxiety behavior index (time in the center of open field test) was
   positively associated with Paraprevotella unclassified, and the
   stereotypic behavior index (grooming duration time) was negatively
   associated with Ca. Arthromitus.

Metabolomics of colon contents linked FMT treatment to lowered toxins,
increased protective compounds, and altered neurotransmitter regulation

   Principle component analysis (PCA) revealed that the samples from the
   BTBR+FMT and BTBR+Saline groups could not be distinguished based on
   their colon contents metabolite profiles; while samples from the
   B6+Saline group clustered separately ([157]Fig. S2c). Subsequent
   statistical and pathway analyses using MetaboAnalyst ([158]57)
   identified 26 differential metabolites (|log[2](FC)| > 1, P < 0.05, VIP
   > 1) between the BTBR+FMT and the BTBR+Saline groups ([159]Fig. 3a).
   Pathway enrichment analysis with MetaboAnalyst identified 16 pathways,
   where pathway “D-glutamine and D-glutamate metabolism” had the highest
   impact value (0.5) and a small P value (P = 0.024) ([160]Fig. 3b).

Fig 3.

   [161]Fig 3
   [162]Open in a new tab

   The effect of FMT treatment on the colon-content metabolome in the BTBR
   mice. (a) The volcano plot showing the variations of colon-content
   metabolites between the BTBR+FMT group and the BTBR+Saline group. The
   dots represented in green (downregulated) and red (upregulated) are
   differentially regulated metabolites with |log[2](fold change) | > 1, a
   P value < 0.05, and VIP value > 1.0. VIP value represents the
   importance projection value of the metabolite obtained from the PLS-DA
   model. (b) Bubble plot showing the metabolic pathways altered by FMT
   treatment in BTBR mice. The bubble size corresponds to the impact of
   the pathway, while the color gradient represents the levels of
   significance, transitioning from the highest (red) to the lowest
   (white). (c) Heatmap depicting the differential metabolites between the
   BTBR+FMT and BTBR+Saline groups. Metabolite abundances were normalized
   to a constant sum and scaled by Pareto scaling. The color gradient,
   ranging from red (high abundance) to blue (low abundance), represents
   metabolite levels on a log[2] scale.

   Due to the complex origins of the metabolites in the colon contents,
   they were expected to have a multifaceted influence on ASD. In fact,
   some of the differential metabolites have been reported as biomarkers
   of ASD, including hippuric acid ([163]58), glutamine/glutamate
   ([164]59), and serotonin ([165]60); some are related to
   neurotransmitter metabolism, including serotonin ([166]61), 5-methoxy
   tryptamine (5-MT, serotonin agonist, structurally related to
   melatonin), 3-indole propionic acid (3-IPA; tryptophan metabolite)
   ([167]62), and glutamine/glutamate ([168]63). Besides, some metabolites
   are associated with beneficial biological functions associated to ASD,
   such as cholesterol homeostasis (MVA) ([169]64), mitochondria function
   (tetradecanedioic acid) ([170]65, [171]66), anti-inflammation
   (stearidonic acid) ([172]67), neuroprotection (N-acetyl-tryptophan)
   ([173]68), dihydrocaffeic acid ([174]69), human cannabinoid type 1
   (CB1) receptor-mediated memory, and neurohormone regulation ([175]70,
   [176]71) (2-linoleoyl glycerol, or 2-LG, partial agonist of CB1
   receptor). We also noticed a decrease in toxicants in the BTBR+FMT
   group, including genipin ([177]72), callystatin A ([178]73), and
   glutathionylspermidine ([179]74). This finding is consistent with the
   theory that environmental toxicants contribute to the impairment of
   mitochondrial function in ASD patients who miss detoxification
   capability ([180]75).

   Comparison between the BTBR+Saline and B6+Saline groups revealed 293
   differential metabolites (|log[2](FC)| > 1, P < 0.05, VIP > 1)
   ([181]Fig. S2e). Pathway analysis between the two groups illustrated
   “Alanine, aspartate and glutamate metabolism,” “Linoleic acid
   metabolism,” “Riboflavin metabolism,” “Arginine and proline
   metabolism,” and “Vitamin B[6] metabolism” as top enriched pathways
   (raw P < 0.05, Impact > 0.5) ([182]Fig. S2g), which did not overlap
   with the pathways altered by FMT treatment in the BTBR mice ([183]Fig.
   3b).

   In short, the colon-content metabolome changes in the BTBR+FMT group
   suggested the beneficial roles of FMT in reducing toxicants and
   increasing protective metabolites for mitochondria and neurological
   health. FMT may also contribute to the modulation of
   tryptophan-serotonin metabolism and glutamine/glutamate balance in the
   gut.

Plasma metabolomics analysis identified altered metabolites related to
GABAergic signaling and mitochondrial function

   Compared to the colon-content metabolome, the plasma metabolome was
   closely related to the physiological status of the host. As
   anticipated, PCA analysis of the plasma metabolome revealed a distinct
   separation between samples from the B6+Saline group, the BTBR+Saline
   group, and the BTBR+FMT group ([184]Fig. S2d). The BTBR+FMT group was
   closer to the BTBR+Saline group than to the B6+Saline group ([185]Fig.
   S2d), suggesting that mouse strain is a key factor shaping the profile
   of the plasma metabolome.

   Metabolomics analysis of the data identified five decreased metabolites
   and one increased metabolite in the BTBR+FMT group (VIP > 1, P < 0.05,
   |log[2](FC)| > 1). The decreased metabolites included malonic acid
   (MA), 3-hydroxybutyrate, 2-LG, carnitine C14:2-OH, and
   MG(0:0/18:3(9Z,12Z,15Z)/0:0). Notably, 2-LG was also identified as a
   differential metabolite in the colon-content metabolome. MG
   (0:0/18:3(9Z,12Z,15Z)/0:0), or 2-monolinolenin, is a structural analog
   of 2-LG and may share similar functions. Methylguanidine (MG), the only
   metabolite that increased significantly, is an oxidative stress
   biomarker that reflects the changes in arginine metabolism. It can
   inhibit nitric oxide (NO) formation ([186]76), critical in regulating
   cell survival, differentiation, and proliferation of neurons in autism
   and other neurological disorders ([187]77, [188]78). Perturbation of MG
   level has been observed in the urine of ASD children ([189]79,
   [190]80). To expand the hypothesis space, we relaxed the cutoff of fold
   change to 1.5 (>1.5 or <2/3) and identified 18 increased metabolites
   and 28 decreased metabolites in the BTBR+FMT group ([191]Fig. 4a). The
   following pathway analysis reported 18 pathways in total, including
   pyrimidine metabolism, arginine biosynthesis, β-alanine metabolism,
   amino sugar and nucleotide sugar metabolism, and so on ([192]Fig. 4b).
   Amino acid and urea cycle disorders have been associated with ASD
   pathogenesis ([193]81, [194]82). However, these pathways exhibited
   either significant raw P values (P > 0.05, or equivalently, −log[10](P)
   < 1.3) or shallow impact values (approaching 0), as shown in [195]Fig.
   4b. This diminishes the reliability of functional interpretations
   derived from these pathways.

Fig 4.

   [196]Fig 4
   [197]Open in a new tab

   The effect of FMT treatment on the plasma metabolome in the BTBR mice.
   (a) The volcano plot shows the variations of plasma metabolites between
   the BTBR+FMT group and the BTBR+Saline group. The dots represented in
   green (downregulated) and red (upregulated) are differentially
   regulated metabolites with |log[2](fold change) | > 1, a P value <
   0.05, and VIP value > 1.0. VIP value represents the importance
   projection value of the metabolite obtained from the PLS-DA model. (b)
   Bubble plot showing the alteration of pathways by FMT treatment in the
   BTBR mice. The bubble size corresponds to the impact of the pathway,
   while the color gradient represents the levels of significance,
   transitioning from the highest (red) to the lowest (white). (c) Heatmap
   depicting the differential metabolites between the BTBR+FMT group and
   the BTBR+Saline group mice. Metabolite abundances were normalized to
   constant sum and scaled by Pareto scaling. The color gradient, ranging
   from red (high abundance) to blue (low abundance), represents
   metabolite levels on a log[2] scale.

   Further examination of the differential metabolites revealed the
   enrichment of multiple long-chain acylcarnitine derivatives in the
   BTBR+Saline group. These include carnitine C10:0, carnitine C11: DC,
   carnitine C12:1, carnitine 14-OH, carnitine 14:1, carnitine 14:2,
   carnitine 14:2-OH, carnitine 15: DC, carnitine C17:1: DC, carnitine
   C18:2-OH, and carnitine C18:3 ([198]Fig. 4c). This pattern suggests
   impaired fatty acid β-oxidation and has been proposed as a biomarker
   for ASD children with mitochondrial dysfunction ([199]83). Note that
   the pathway analysis did not reveal any pathways related to long-chain
   acylcarnitine derivatives, owing to the missing ID hits of these
   compounds ([200]Table S2). Besides, MA and 3-hydroxybutyrate have also
   been reported to be involved in mitochondrial function ([201]84,
   [202]85). An increase in a subset of metabolites related to GABAergic
   signaling was observed in the BTBR+FMT group, including 5-amino valeric
   acid (5-AV; a methylene homolog of GABA), guanidinoethyl sulfonate
   (GES; an analog of taurine), and O-phosphorylethanolamine (PEA; an
   analog of GABA). Interestingly, 5-AV and taurine are GABA[A] agonists
   and were demonstrated by Sharon et al. to improve social and repetitive
   behaviors in BTBR mice ([203]32). 5-AV was also found to be decreased
   in urine samples from a group of individuals with ASD ([204]86). PEA
   has been proposed as a biomarker for ASD ([205]87). Additionally, an
   increase in 4-pyridoxic acid was observed in the BTBR+FMT group, which
   is the catabolic product of vitamin B[6], reflecting the status of
   vitamin B[6] and renal function in the host ([206]88).

   Comparison between the BTBR+Saline group and the B6+Saline group
   identified 69 differential metabolites (|log[2](FC)| > 1, P < 0.05, VIP
   > 1) ([207]Fig. S2f). Pathway analysis between the BTBR+Saline and
   B6+Saline groups highlighted “Histidine metabolism,” “beta-Alanine
   metabolism,” “Arginine biosynthesis,” “Alanine, aspartate and glutamate
   metabolism,” “Linoleic acid metabolism,” and “Glycine, serine and
   threonine metabolism” as top enriched pathways (raw P < 0.05, Impact >
   0.5) ([208]Fig. S2h).

Integration of multi-omics results: tracing the path from the FMT treatment
to social behavior improvement

   We seek to better understand FMT’s paths to alleviate the social
   behavior deficits in the BTBR mouse model. We performed correlation
   analysis between each omics data set and the behavioral indices
   ([209]Fig. S4a through c) and between the omics data sets themselves
   ([210]Fig. S5). Notably, a significant portion of the differential taxa
   and metabolites were correlated with the sociability index.

   To explore the major contributors to the phenotype change in each omics
   data set, we conducted weighted gene co-expression analysis (WGCNA) to
   extract modules from each data set across the BTBR mice samples (see
   “Integration of multi-omics data” below). The WGCNA analysis extracted
   three modules from the metagenomics data set (MGS), ten from the
   colon-content metabolomics (MIC), and five from the plasma metabolomics
   (PL) ([211]Table S3). We also measured the correlation between these
   modules and the sociability index ([212]Fig. 5a). Significant
   associations were found between the sociability index and several
   modules: Module 1 (MGS), Module 0 (PL), Module 2 (PL), Module 4 (MIC),
   and Module 5 (MIC). Additionally, the social novelty index correlated
   with Module 1 (MGS) and Module 6 (MIC) ([213]Fig. 5a). Module 1 (MGS),
   the only module from the metagenomics data that correlated with the
   sociability index, predominantly included most Alistipes species and
   two Bacteroides species (B. ovatus and B. xylanisolvens) ([214]Table
   S3), suggesting their significant roles in improving social behavior in
   the BTBR mouse model. In plasma metabolomics, Module 2 (PL) was
   dominated by long-chain acylcarnitine derivatives and other compounds
   related to fatty acyls, as shown in [215]Fig. 5b, aligning with the
   results from the previous differential analysis ([216]Fig. 4c). Module
   4 (MIC) and Module 5 (MIC) comprised a variety of oligopeptides
   (classified as “Carboxylic acids and derivatives,” as shown in
   [217]Fig. 5c; [218]Table S3). Module 6 (MIC) included
   glycerophospholipids such as lysophosphatidylcholine (LPCs) and
   phosphatidylethanolamine (PPEs) ([219]Fig. 5c; [220]Table S3). Most
   compounds were not observed in the differential analysis in the
   colon-content metabolomics ([221]Fig. 3c).

Fig 5.

   [222]Fig 5
   [223]Open in a new tab

   Integration and analysis of multi-omics data reveal key contributors to
   social behavior change in BTBR mice. (a) Correlation network between
   sociability and social novelty indices and WGCNA modules derived from
   each omics data set. Orange dots represent plasma metabolites modules,
   blue dots represent the colon-content metabolites modules, and green
   dots represent the metagenome modules. Modules were represented by the
   module eigen genes. Pearson correlation was used for the correlation
   coefficients. Asterisks indicate the significant correlations between
   modules. *P < 0.05, **P < 0.01, ***P < 0.001. (b) The distributions of
   metabolite classes within WGCNA modules derived from the plasma
   metabolomics data set. (c) The distribution of metabolite classes
   within WGCNA modules derived from the colon-content metabolomics data
   set. (d) The STA-Sankey network for [224]R00174 metabolic reaction in
   vitamin B[6] metabolism. Spearman’s rank correlation was used for the
   correlation coefficients. Asterisks indicate taxa with statistical
   significance. Node color: red indicates upregulation, green indicates
   downregulation. (e) The BIO-Sankey network for [225]R00174 metabolic
   reaction in vitamin B[6] metabolism. Band color: red indicates
   positive, and green indicates negative correlations with metabolites.
   Deep red/green on bands denote statistically significant correlations
   (P  <  0.05).

   In parallel, we performed MetOrigin analysis ([226]89) to extract the
   relationships between bacteria species and metabolites (see
   “Integration of multi-omics data” below). MetOrigin is designed to
   trace metabolites to different origins, such as food, host, or
   co-metabolism, and provides statistical and biological association
   information for the connections it identifies. Paired analysis was
   performed between metagenomics and plasma metabolomics (M-P), as well
   as between metagenomics and colon-content metabolomics (M-C),
   separately. In the M-C analysis, MetOrigin identified four pathways as
   originating from “Co-metabolism,” one pathway from “Microbiota,” and
   one pathway from the “Host” ([227]Fig. S6) . Similarly, the M-P
   analysis reported six pathways as “Co-metabolism” origin, one pathway
   as “Microbiota” and one as “Host” ([228]Fig. S7). Notably, the plasma
   compound pyridoxal was traced to multiple Bacteroides (Phocaeicola)
   spp., Parabacteroides spp., and L. gasseri, a connection supported by
   both statistical and biological relationships ([229]Fig. 5d and e).
   Pyridoxal is a natural form of vitamin B[6]. It can be converted to PLP
   by pyridoxal kinase. This finding aligns with the results of stratified
   pathway analysis in metagenomics previously, which indicated that
   multiple Bacteroides species capable of PLP synthesis (a part of
   vitamin B[6] metabolism) were enriched in the BTBR+FMT group ([230]Fig.
   2e). Given that vitamin B[6] is essential for the metabolism of
   long-chain unsaturated fatty acid ([231]90, [232]91), a process crucial
   for normal mitochondria function, we suspected that vitamin B[6]
   metabolism may bridge the gap between the altered Bacteroides species
   brought by FMT and subsequent metabolic and phenotypic changes in the
   BTBR mice.

Vitamin B6 alleviates social behavior deficits similar to FMT in the BTBR
mouse model

   As a proof of concept, the effectiveness of vitamin B[6] was tested in
   the mouse model ([233]Fig. 6; [234]Fig. S1). Intriguingly, treatment
   with vitamin B[6] remarkably affected the sociability ([235]Fig. 6a)
   and social novelty ([236]Fig. 6d) of the BTBR mice. Indeed, the
   sociability performance of vitamin B[6]-treated BTBR mice resembled
   that of the B6+Vehicle group and the B6+VB[6] group ([237]Fig. 6b and
   c). Compared to the BTBR+Vehicle group, the BTBR+VB[6] group exhibited
   a reversal in the social novelty deficit, showing increased interaction
   time with the novel mouse ([238]Fig. 6e and f). However, vitamin B[6]
   did not alleviate anxiety and repetitive behavior in the BTBR mice
   ([239]Fig. S1c and d), which is consistent with the efficacy
   performance of the BTBR+FMT group ([240]Fig. S1a and b).

Fig 6.

   [241]Fig 6
   [242]Open in a new tab

   Vitamin B[6] alleviated the social behavior of the BTBR mice. (a)
   Representative heatmaps of the three-chamber sociability test performed
   in the B6 and BTBR mice. (b) Interaction time of experiment subject
   with a stranger mouse or empty cage in sociability task. (c)
   Sociability index of B6 and BTBR mice (n = 8 mice in each group). (d)
   Representative heatmaps of the three-chamber social novelty test
   performed in B6 and BTBR mice. (e) Interaction time of experiment
   subject with a stranger mouse or a familiar mouse in social novelty
   task. (F) Social novelty index of B6 and BTRB mice (n = 8 mice in each
   group). Data represent the mean ± SEM. Error bars indicate SEM. *P <
   0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. The test used in (b, c,
   e, f): two-way ANOVA with Tukey’s post hoc test.

DISCUSSION

   Although significant advances have been made in both animal
   ([243]32[244]–[245]34, [246]92, [247]93) and clinical studies
   ([248]26[249]–[250]28), the efficacy of current FMT treatments for ASD
   heavily depends on donor selection and remains a black box. In our
   study, plasma metabolomics analysis revealed a distinct acylcarnitine
   profile in the BTBR+Saline group, suggesting impaired fatty acid
   β-oxidation and mitochondrial dysfunction ([251]94, [252]95) in the ASD
   model ([253]96, [254]97). FMT can normalize the impaired fatty acid
   metabolism ([255]Fig. 3C), with changes in acylcarnitine levels
   negatively correlating with improvements in the social behavior index
   ([256]Fig. S4c). Mitochondrial function is known to be crucial for the
   social behavior of BTBR mice ([257]98) and is a prominent issue in ASD
   children ([258]99). The concept of mitochondrial dysfunction in ASD
   pathogenesis has been extensively discussed in both animal ([259]96,
   [260]97) and human studies ([261]83, [262]100, [263]101). Future
   research will delve deeper into mitochondrial functionality in the
   context of ASD and its connection to social behavior.

   Multiple components in FMT may contribute to the normalization of fatty
   acid metabolism or mitochondrial function. Our colon-content
   metabolomics analysis indicated that FMT increased metabolites
   beneficial for mitochondrial function and neurological health while
   reducing toxicants. The relationship between these metabolites and
   mitochondrial function appears complex and multifaceted, necessitating
   extensive research to elucidate these interactions fully. Nonetheless,
   our results highlighted a potential connection via vitamin B[6]
   metabolism: pathway analysis of the metagenomics data identified an
   enrichment of Bacteroides species (B. uniformis, B. eggerthii, B.
   xylanisolvens, B. thetaiotaomicron, and B. ovatus) with genes involved
   in PLP synthesis in the BTBR+FMT group ([264]Fig. 2e). Differential
   analysis of the plasma metabolomics identified an increase of
   4-pyridoxic acid, the end product of vitamin B[6] metabolism, in the
   BTBR+FMT group ([265]Fig. 4c). Further, integrative multi-omics
   analysis using MetOrigin ([266]89) also associated the Bacteroides
   species with plasma levels of pyridoxal ([267]Fig. 5d and e),
   suggesting their roles in modulating vitamin B[6] metabolism. The level
   of 4-pyridoxic acid was positively correlated with B. ovatus and B.
   eggerthii, while pyridoxal was positively correlated with B. ovatus, B.
   eggerthii and B. xylanisolvens ([268]Fig. S4d). Note that B. ovatus and
   B. xylanisolvens were also present in Module 1 (WGS). In a
   proof-of-concept study, we validated the effectiveness of vitamin B[6]
   in mitigating social behaviors in the BTBR mouse model. Although a
   previous animal study reported the role of gut microbiota in
   alleviating ASD symptoms via vitamin B[6] homeostasis ([269]56), our
   study further suggested the connection between gut microbiota-regulated
   vitamin B[6] metabolism and impaired fatty acid metabolism (suggesting
   mitochondria dysfunction) in the context of ASD mice model. Future
   research should investigate the interplay between these pathways and
   their collective impact on improving social behavior in ASD.

   Additional beneficial elements in FMT may also contribute to the
   improvement of sociability. In the BTBR+FMT group, we noted alterations
   in GABAergic signaling in the plasma metabolomics data. Observing
   altered compounds, such as 5-AV and taurine analog, aligns with
   previous reports that GABA[A] receptor agonists like 5-AV and taurine
   promoted GABAergic signaling and aided in autism symptom alleviation
   ([270]32). Concurrently, metagenomics analysis showed an increase in
   Bacteroides, Parabacteroides, and Alistipes species. B. ovatus can
   module the levels of some intestinal neurotransmitters, for example,
   increasing the intestinal GABA and decreasing tryptophan and glutamine
   ([271]49). B vulgatus and B. dorei can decrease the production of gut
   microbial lipopolysaccharide ([272]102). P. merdae can protect against
   seizure by reducing systemic GABA and elevating hippocampal GABA and
   glutamate levels ([273]50). Many Alistipes species modulate tryptophan
   metabolism by hydrolyzing tryptophan into indole derivatives ([274]51)
   and participate in preserving gut barrier integrity ([275]103). In our
   study, A. senegalensis was negatively associated with undesirable
   metabolites such as fipexide, genipin, and glutathionylspermidine
   ([276]Fig. S5a). Our study did not cover the enriched species' multiple
   effects, necessitating a detailed examination of specific species or
   even individual strains to understand their roles and impacts fully.

   Our study has several limitations. First, we did not include an
   FMT-treated B6 group as a control. Comparing the FMT-treated BTBR group
   with an FMT-treated B6 group would aid in identifying bacterial
   candidates that uniquely persist in the BTBR+FMT group and may be
   associated with behavior improvements. Additionally, our study only
   inferred the roles of specific bacterial species in alleviating social
   deficits in the BTBR mouse model. Validating the effects experimentally
   using strains isolated from the same donor and replicating the
   experiments with different animal models is crucial to confirm these
   findings. Furthermore, we did not eliminate the impact of bacteria
   originating from the BTBR mice with antibiotic treatment. Although many
   species present in the BTBR mice after FMT treatment were exclusive to
   the human donor samples ([277]Table S1), there remains the possibility
   that the donor’s microbiome may interact with the native microbiome of
   the mice, resulting in a novel community with unique functions.

   Future work should focus on refining the effective components of FMT
   and elucidating their connections with ASD. It’s important to recognize
   that many ASD cases do not involve mitochondrial dysfunction or a
   disorder of vitamin B[6] metabolism. Therefore, it would be valuable to
   compare responses to FMT treatment between ASD children with and
   without mitochondrial dysfunction. Moreover, stereotypic behaviors in
   ASD are mediated through different mechanisms ([278]104). Investigating
   donors or treatments that primarily target stereotypic behaviors could
   lead to the development of combinatory therapies addressing both core
   symptoms of ASD. Such research could significantly enhance our
   understanding of the disease and improve treatment strategies.

MATERIALS AND METHODS

Animals

   Three-month-old male C57Bl/6 mice and BTBRT^+Itpr3^tf/J mice were
   purchased from the Jackson Laboratory (USA). Mice were maintained in a
   pathogen-free SPFII animal facility in a condition-controlled room (23
   ± 1°C, 50 ± 10% humidity). A 12 h light/dark cycle was automatically
   imposed. Mice were housed in groups of four to five per individually
   ventilated cages and given access to food and water ad libitum unless
   otherwise stated in the methods below. Experimenters were blinded to
   the treatment of animals and sample processing throughout the
   subsequent experimentation and analyses.

Fecal microbiota transplantation

   Healthy donors were screened and recruited by Shenzhen Xbiome Biotech
   Co. Ltd. Human fecal samples from a healthy adult male were collected
   at the Shenzhen Xbiome Biotech Co. Ltd according to the study protocol
   approved by. Informed consent was obtained from the subject. Fecal
   samples were suspended in an equal volume (wt/vol) of
   phosphate-buffered saline (PBS) containing 20% glycerol in PBS,
   snap-frozen in liquid nitrogen, and stored at −80°C until use. The
   frozen stocks were thawed, suspended in saline, filtered through a
   100-µm cell strainer, and orally inoculated into BTBR mice
   (approximately 3.64 × 10^10 CFU/mL, 200 µL per mouse). Control groups
   were administered saline.

Mouse behavioral tests

   Transfer rodents to the procedure room (for acclimation) 1 h before
   testing. The apparatus was carefully cleaned by wiping it with 75%
   ethanol after each experiment to remove the olfactory distraction. The
   person performing the test was blind to the animal groups. All mice
   were familiarized with the researchers at least one week in advance.

Open field test

   Mouse spontaneous activity in the open field was tested using a clear
   plastic cube box (L × W × H: 41 × 41 × 38 cm^3) under a camera. The
   center region of the box field was defined as a 20 cm × 20 cm virtual
   area. Each mouse was given 10 min to move freely in the box. All
   traveled distances and the distance trace center were recorded and
   analyzed using Noldus EthoVision XT (V15, Noldus Information
   Technology, Leesburg, VA, USA). The rearing numbers, representing the
   number of times mice lifted the forepaws to explore upwards, were
   recorded by the experimenter.

Self-grooming test

   The mice were placed in a clean observation box (L × W × H: 20 cm × 20
   cm × 50 cm) with bedding on the bottom under a camera. With low light
   conditions (about 40 Lux), the behavior of mice was recorded for 10 min
   after pre-adaptation to the observation box. The experimenter remained
   quiet and away from the test mice during the test. The duration of the
   grooming of mice was recorded and analyzed using captured videos.

Three-chamber social test

   Mice were first habituated for 10 min in an empty (40 cm × 60 cm × 22
   cm) Plexiglass arena divided into three interconnected chambers (left,
   center, and right). Sociability was evaluated during a second 10-min
   period in which the subject could interact either with an empty wire
   cup (Empty) or a wire cup containing a genotype, age, sex, and
   treatment-matched stranger conspecific (Mouse 1). The interaction time
   was determined by measuring the time the subject mouse spent sniffing
   or climbing upon either the empty cup or the cup containing the
   stranger mouse. The position of the empty cup/stranger mouse in the
   left or right chamber during the sociability period was counterbalanced
   between trials to avoid bias. Preference for social novelty was assayed
   in a third 10-min period by introducing a second stranger mouse (Mouse
   2) into the previously empty wire cup. The time spent interacting with
   the empty cup or Mouse 1 or Mouse 2 was recorded and measured using
   Noldus EthoVision XT (V15, Noldus Information Technology, Leesburg, VA,
   USA). Sociability index = (T[mouse 1]− T[empty])/(T[mouse 1] +
   T[empty]), social novelty index = (T[mouse 2] − T[mouse 1])/(T[mouse 2]
   + T[mouse 1]).

Behavior test statistics

   Data were presented as mean ± SEM. Statistical analyses performed
   include one- or two-way ANOVA with Bonferroni post hoc analysis to
   correct for multiple comparisons unless otherwise indicated. P, t, and
   F values were presented in the figure legends. P < 0.05 was considered
   statistically significant: *P < 0.05, **P < 0.01, ***P < 0.001, ****P <
   0.0001. Prism 9 (version 9; GraphPad, CA, USA) software was used to
   perform statistical analyses and generate graphical data
   representations.

Fecal DNA extraction, library preparation, and sequencing

   Colon-content samples were collected from BTBR mice, as previously
   described. Briefly, DNA was extracted using the QIAamp PowerFecal DNA
   Kit (catalog no. 51804, Qiagen), followed by library construction and
   sequencing on an Illumina NovoSeq 6000 platform (Novo Gene). Stool
   samples were collected from BTBR mice, as previously described.
   Briefly, DNA was extracted using the QIAamp PowerFecal DNA Kit (catalog
   no. 51804, Qiagen), followed by library construction and sequencing on
   an Illumina NovoSeq 6000 platform (Novo Gene).

Metagenomics analysis

   Quality control of raw reads and host decontamination were performed
   with KneadData (version 0.6.1, reference genome hg19 for human donor
   data decontamination, and mm10 for mice data decontamination).
   Taxonomic profiling and microbial pathway abundance analysis were
   performed using HUMAnN 2 (version 2.8.1) ([279]54), respectively.
   Shannon indices of α diversity, PCoA analysis with Bray-Curtis
   distance, and taxonomic composition analysis were performed using R
   package microbiome (version 1.21.0). The LEfSe analysis was performed
   by the docker version of LEfSe (version 1.0.0) ([280]46) with the LDA
   cutoff set as 2.

Plasma and colon-content sample preparation and extraction

   Mice were euthanized by cervical dislocation at the end of behavior
   tests. Plasma and cecum content were collected and transferred to a
   −80°C refrigerator. The sample stored at −80°C refrigerator was thawed
   on ice and vortexed for 10 s. A 150 µL extract solution (ACN: methanol
   = 1:4, vol/vol) containing internal standard was added to a 50 µL
   sample. Then, the sample was vortexed for 3 min and centrifuged at
   12,000 rpm for 10 min (4°C). A 150 µL aliquots of the supernatant were
   collected and placed in −20°C for 30 min, and then centrifuged at
   12,000 rpm for 3 min (4°C). Aliquots of the supernatant at 120 µL were
   transferred for LC-MS analysis.

T3 UPLC conditions

   The sample extracts were analyzed using an LC-ESI-MS/MS system (UPLC,
   ExionLC AD,ˈ[281]https://sciex.com.cn/; MS, QTRAP System,
   [282]https://sciex.com/). The analytical conditions were as follows:
   UPLC: column, Waters ACQUITY UPLC HSS T3 C18 (1.8 µm, 2.1 mm*100 mm);
   column temperature, 40°C; flow rate, 0.4 mL/min; injection volume, 2 or
   5 µL; solvent system, water (0.1% formic acid):acetonitrile (0.1%
   formic acid); gradient program, 95:5 vol/vol at 0 min, 10:90 vol/vol at
   10.0 min, 10:90 vol/vol at 11.0 min, 95:5 vol/vol at 11.1 min, 95:5
   vol/vol at 14.0 min.

QTOF-MS/MS

   The Triple TOF mass spectrometer was used for its ability to acquire
   MS/MS spectra on an information-dependent basis (IDA) during an LC/MS
   experiment. In this mode, the acquisition software (TripleTOF 6600, AB
   SCIEX) continuously evaluates the full scan survey MS data as it
   collects and triggers the acquisition of MS/MS spectra depending on
   preselected criteria. In each cycle, 12 precursor ions whose intensity
   is greater than 100 were chosen for fragmentation at collision energy
   (CE) of 30 V (12 MS/MS events with product ion accumulation time of 50
   ms each). ESI source conditions were set as follows: ion source gas 1
   as 50 psi, ion source gas 2 as 50 psi, curtain gas as 25 psi, source
   temperature 500°C, ion spray voltage floating (ISVF) 5,500 V or −4,500
   V in positive or negative mode, respectively.

ESI-Q TRAP-MS/MS

   LIT and triple quadrupole (QQQ) scans were acquired on a triple
   quadrupole-linear ion trap mass spectrometer (QTRAP), QTRAP LC-MS/MS
   System, equipped with an ESI Turbo Ion-Spray interface, operating in
   positive and negative ion mode and controlled by Analyst 1.6.3 software
   (Sciex). The ESI source operation parameters were as follows: source
   temperature 500°C; ion spray voltage (IS) 5,500 V (positive), −4,500 V
   (negative); ion source gas I (GSI), gas II (GSII), curtain gas (CUR)
   was set at 50, 50, and 25.0 psi, respectively; the collision gas (CAD)
   was high. Instrument tuning and mass calibration were performed with 10
   and 100 µmol/L polypropylene glycol solutions in QQQ and LIT modes,
   respectively. A specific set of MRM transitions was monitored for each
   period according to the metabolites eluted within this period.

Metabolomics analysis

   The qualitative and quantitative determination of metabolites was
   conducted utilizing both a proprietary database, MWDB (Metware
   Biotechnology Co., Ltd., Wuhan, China), and publicly accessible
   databases, including the Human Metabolome Database (HMDB;
   [283]https://www.hmdb.ca), MassBank ([284]https://www.massbank.jp),
   METLIN ([285]https://metlin.scripps.edu), KEGG
   ([286]https://www.kegg.jp), MoNA
   ([287]https://mona.fiehnlab.ucdavis.edu), PubChem
   ([288]https://pubchem.ncbi.nlm.nih.gov), and ChemSpider
   ([289]https://www.chemspider.com). The annotated peak abundance tables
   from plasma metabolomics and colon-content metabolomics were uploaded
   to MetaboAnalyst 5 ([290]57)
   ([291]https://www.metaboanalyst.ca/MetaboAnalyst/) for data
   pre-processing (normalization by sum, log transformation, and
   autoscaling) and downstream analysis: principal component analysis
   (PCA), differential analysis and pathway analysis. 3D PCA was generated
   additionally using the R package scatterplot3d (version 0.3-44).
   Metabolites exhibiting a VIP (variable importance in projection) value
   >1, a P < 0.05, and a fold change >2 or <0.5 were identified as
   differentially expressed between the two groups. The volcano plots were
   generated based on the log[2](fold change), −log[10](P), and VIP values
   of the metabolites. The pathway analysis was performed based on the Mus
   musculus (KEGG) pathway library on the MetaboAnalyst ([292]57) website.
   The taxonomic classes of chemical compounds were annotated by R package
   biodbHMDB (version 1.8.0) and Class II was selected and manually
   curated for final visualization. All other statistical analyses and
   plotting were performed in R (version 4.2.1).

Integration of multi-omics data

   Samples from the BTBR+FMT group and the BTBR+Saline group were used to
   calculate the correlation. The correlation of metabolite abundance,
   microbial species level, and social behavior index was calculated using
   Spearman’s rank correlation method across the BTBR mice samples. The P
   value was corrected by the false positive rate (FDR) method using R
   package psych package (version 2.4.1) ([293]105). All statistical
   analyses and plotting were performed in R (version 4.2.1).

   WGCNA analysis (version 1.72-5) ([294]106) was performed to construct
   significant modules in each omics data. We optimized the soft threshold
   to 10 for the metagenomics data, 14 for the plasma metabolomics data,
   and 12 for the colon-content metabolomics data. The features were
   clustered into modules by hierarchical clustering with Euclidian
   distance, and similar modules were further merged. The correlation
   within modules and correlation between modules and social behavior
   index was also measured by Pearson correlation using raw P value <
   0.05. Modules that were associated with the social behavior index were
   explored for potential patterns of compound classes or taxa
   compositions.

   A paired metabolite abundance table and metagenomics profile table were
   uploaded to MetOrigin ([295]https://metorigin.met-bioinformatics.cn/)
   for metabolite origin tracking analysis. The analysis type was set as
   “Deep MetOrigin Analysis,” the host was set as “Mus musculus (house
   mouse).” For both metagenomics and metabolomics data, the test method
   “Auto” was selected, and P value cutoff was set to 0.05. Sankey diagram
   and network between metabolites and microbes were generated for
   metabolites belonging to the host, bacteria, and both (co-metabolism).

ACKNOWLEDGMENTS