Exploring the causal link between microbiota and tic disorders: a gene sequencing and Mendelian randomization approach

Abstract

Background

Tic disorders (TD), including Tourette syndrome (TS), are common childhood-onset neurodevelopmental conditions with unclear etiology. Emerging observational data suggest that gut-microbiota (GM) dysbiosis accompanies TD, but causality is unresolved. We aimed to determine whether specific bacterial genera are causally implicated in TD susceptibility.

Methods

Two-sample Mendelian randomization (MR) was performed by integrating the largest available GM genome-wide association study (GWAS) (18,340 Europeans, 211 taxa) with the PGC-TS-2019 GWAS (4,819 cases/9,488 controls). Inverse-variance-weighted estimates were complemented with sensitivity analyses and reverse-MR. Findings were validated in an independent pediatric case–control cohort (10 TD vs seven healthy children) profiled by 16S rRNA V3–V4 sequencing; between-group differences were tested with the Mann–Whitney U test.

Results

Genetically predicted abundance of Anaerotruncus, Butyrivibrio and Ruminococcaceae UCG-002 conferred protection against TS (OR 0.69–0.86, p = 0.014–0.016), whereas Dialister and Ruminiclostridium 6 increased risk (OR 1.28–1.32, p = 0.030–0.041); Sutterella showed no causal effect (p = 0.103). No heterogeneity, directional pleiotropy or reverse causation was detected. Sequencing analyses mirrored MR directions: TD cases exhibited significantly lower relative abundance of the protective genera and higher levels of risk taxa compared with controls (p < 0.05).

Conclusions

By integrating unbiased genetic instrumentation with targeted microbiome profiling, this study offers exploratory evidence suggesting that specific gut bacteria may be associated with TD pathogenesis. Ruminococcaceae UCG-002, Anaerotruncus, and Butyrivibrio emerge as potentially protective taxa, while Dialister and Ruminiclostridium 6 may represent candidate risk markers. These preliminary, mechanistically grounded insights should be considered exploratory and may inform future, larger-scale microbiome-directed precision interventions in TD.

Introduction

Tic disorders (TD), neurodevelopmental disorders characterized by involuntary, rapid, and recurrent movements (motor tics) or vocalizations (vocal tics), often emerge during childhood. TD include Tourette syndrome (TS), chronic tic disorder (CTD), and transient tic disorder (TTD) (Liu et al., 2020). Family- and genome-wide studies indicate that TD and TS share a substantial and largely overlapping polygenic architecture, supporting their conceptualization as different points on a continuous tic spectrum rather than as distinct disease entities (Lin, Tsai & Chou, 2022). Epidemiological studies have shown that TD, even transient and mild tics, may affect up to one-fifth of school-aged children and are common in boys (Malik & Hedderly, 2018). Children with TD frequently experience cooccurring psychiatric and developmental conditions, including attention deficit hyperactivity disorder (ADHD), obsessive−compulsive disorder (OCD) and anxiety (Malik & Hedderly, 2018). At present, while it has been widely accepted that a combination of genetic predispositions and environmental influences play significant roles in the development of this condition, the exact origins of TD remain unknown (Bao et al., 2023). Therefore, investigating the etiological factors and pathophysiological processes associated with TD is crucial for the development of innovative treatment strategies.

The core gut microbiota (GM) includes the foundation guild and patobiont guild, which form a dynamic competitive relationship that impacts human health (Wu et al., 2024). The GM plays a significant role in the development and function of neurological disorders (Jabbari Shiadeh et al., 2025). Bidirectional communication between the GM and the brain occurs via the microbiota−gut−brain axis (MGBA) (Sorboni et al., 2022). Disruptions in the GM can affect brain physiology, cognitive functions, and behavior (Jabbari Shiadeh et al., 2025). Research has consistently revealed that the GM of those afflicted with TD is markedly different from that of healthy controls (HCs) (Fan et al., 2022). A robust correlation was identified by researchers, connecting the intensity of tic symptoms to the abundance of diverse bacterial species and the metabolic processes of the gut microbiome (Xi et al., 2021). This evidence points to a possible role of the GM in the development of TD. Nevertheless, the existence of a causal link between the GM and TD remains uncertain.

Mendelian randomization (MR) is an epidemiological method that uses genetic variants as instrumental variables (IVs) to assess the causal effects of exposures on outcomes (Yang et al., 2024). This approach is grounded in the principles of Mendel’s laws, which state that alleles segregate randomly during gamete formation, thus reducing the likelihood of confounding from environmental factors that might influence both the exposure and outcome. MR analyses serves as an exemplary approach to investigate the causal links between the GM and neuropsychiatric conditions (Chen et al., 2024).

In this research, we utilized 16S rRNA gene sequencing to identify the differences in GM composition among individuals in the TD group and those in the control group. Following this, we applied the two-sample MR approach to investigate the potential causal impact of the GM on the development of TD. We subsequently corroborated these findings by analysing fecal sequencing data from both TD patients and healthy control subjects. Our aim was to generate preliminary insights into the possible relationship between GM and TD, thereby contributing to a better understanding of TD etiology for future prevention and treatment strategies.

Materials & Methods

This study received clearance from the Ethics Committee at Cangzhou People’s Hospital in Hebei Province (approval no. K2022-Approval-020; 27 June 2022). Consent forms were given to the legal guardians of all participants and duly signed.

Study design

Initially, we performed 16S rRNA sequencing to compare GM profiles between individuals with TD and HC. To test for potential causal relationships, we conducted two-sample MR analyses—forward MR to estimate the effect of GM on TD risk and reverse MR to assess whether TD might reshape the GM. To determine genetic variants that may predict causal effects, it is crucial to adhere to three key assumptions: (1) there must be a substantial association between IVs and the gut microbiota, (2) IVs should be free from the influence of confounding factors, and (3) IVs should not exert a direct influence on the manifestation of TD (Yang et al., 2024). Finally, we corroborated the MR findings by profiling the fecal microbiota of TD patients and matched healthy controls with 16S rRNA gene sequencing and comparing the observed differences with the genetically predicted effects. The methodology is depicted in Fig. 1.

Figure 1. The study’s flow diagram.

Participant recruitment

For this study, we recruited 20 boys aged 5–11 years: 10 with newly diagnosed TD and 10 healthy controls (HC). All TD patients were newly diagnosed at Cangzhou People’s Hospital, Hebei, between February 2023 and April 2024 by attending-level pediatricians using DSM-V criteria; age-matched HC were recruited from the same hospital’s pediatric health clinic. Within a 4-week period prior to the study, both the TD and HC groups had no history of acute or chronic respiratory or gastrointestinal diseases. Furthermore, neither group had records of prior medication intake, particularly the use of antibiotics and microbiome-based treatments. Written consent had been obtained from the patients before the study.

Fecal sample collection and 16S rRNA sequencing analyses

Fecal samples, the size of a soybean grain and two per collection, were obtained from each child by their guardians either at home or in the hospital and immediately placed into cell preservation solution (Chengge Biotechnology, Xiamen, China) within 3 min of defecation. One tube was used for DNA extraction, and the other tube was used for quantifying the relative abundance of a specific GM. They were transported at temperatures below 18 °C to our hospital within a two-day timeframe for preservation in a −80 °C freezer for subsequent intestinal microbiota analyses.

Genomic DNA was extracted from microbiome samples via the Human Fecal Intestinal Microbiota Nucleic Acid Extraction Reagent (Magnetic Bead Method; Xiamen Chengge Biotechnology Co., Ltd.). The V4 region of the 16S rRNA gene was subsequently amplified via PCR with the primers 515F (GTGCCAGCMGCCGCGGTAA) and 806R (GGACTACHVGGGTWTCTAAT). Following PCR amplification, the library was constructed via the Human Intestinal Microbiota Gene Detection Kit (High-throughput Sequencing Method, Xiamen Chengge Biotechnology Co., Ltd.). Libraries that met the quality standards were then subjected to paired-end sequencing on the DNBSEQ-G99 platform (MGI Tech, Shenzhen, China), adhering to the detailed procedures outlined in previous studies on intestinal microbiology (Liu et al., 2024). Raw reads were demultiplexed; samples failing any pre-specified cut-off (<20,000 merged paired-end reads, mean Phred score <25, or >10% chimeric sequences) were excluded from downstream analyses.

Bioinformatics analyses

Reads were assembled and analyzed using Vsearch (version 2.14.1) and USEARCH10. After quality filtering and chimera removal, clean sequences and exact amplicon sequence variants (ASVs) were resolved using Unoise3. Taxonomy was assigned to ASVs using USEARCH10 -sintax with the parameters -strand_both and -sintax_cutoff 0.6. ASV abundance was normalized using a standard of sequence number corresponding to the sample with the least sequences. Alpha diversity was assessed via the Chao 1, Shannon, and Simpson indices, whereas beta diversity was analysed via PCoA with the Bray–Curtis distance metric. LEFSe was used to identify significant biomarkers between groups.

Data source for exposure

The MiBioGen consortium was established to explore the impact of human genetics on the composition of the gut microbiota. Our study leveraged the latest genome-wide association study (GWAS) data from a cohort comprising 18,340 participants. This GWAS revealed genetic variants associated with 211 distinct gut microbial taxa, spanning nine phyla, 16 classes, 35 families, and 131 genera. In this study, genus-level taxa were examined via MR analyses, as they represent the highest quantity of microbial groups (Xu et al., 2024). The GWAS summary statistics for the gut microbiota were accessed via https://mibiogen.gcc.rug.nl/.

Data sources for outcomes

We used the most recent Tourette syndrome GWAS meta-analysis from the Psychiatric Genomics Consortium (PGC-TS 2019; 4,819 cases/9,488 controls of European descent; PMID 30818990; https://pgc.unc.edu) as the source of outcome summary statistics. This dataset has been shown to capture the broader tic-spectrum genetic architecture, as TD and TS occupy a continuous polygenic spectrum with extensively overlapping risk loci (Sanna et al., 2019).

Identification of IVs

To confirm the causal link between TD and GM, we carefully chose instrumental variables (IVs) after thorough quality control. We selected the most prevalent gut bacteria at the genus level for analyses. Following recent practice in gut-microbiota Mendelian-randomization analyses, we relaxed the SNP-selection threshold to P < 1 × 10⁻⁵ to obtain a sufficient set of genetic instruments (Zhao et al., 2023). To manage linkage disequilibrium (LD), we used an LD correlation coefficient threshold of r2 < 0.001 and a clumping window exceeding 10,000 kb (Ni et al., 2022). We removed palindromic SNPs from our IV pool. Finally, to avoid weak IV bias, we included SNPs with an F statistic ≥ 10 for further analyses.

MR analyses

To assess the impact of the GM on TD, we employed the inverse variance-weighted (IVW) method as our primary analytical strategy. Additionally, we integrated other techniques, such as MR Egger, weighted median, simple mode, weighted mode, and leave-one-out analyses, to bolster the robustness of our results (Grover et al., 2017). To account for multiple testing across all genus–outcome pairs, we applied Benjamini–Hochberg FDR (BH-FDR) correction; q < 0.05 was considered significant. The study employed reverse MR analyses to ascertain the direction of causality, considering TD as the exposure and utilizing single nucleotide polymorphisms (SNPs) associated with TD as IVs, with a significance threshold set at p < 5 × 10⁻⁸. Missing values were handled by removing rows with na.omit(dat) in R. All analyses were conducted via the ggplot2, TwoSampleMR, and MRPRESSO packages within R software (version 4.3.3).

Validation of MR

Following the MR causal estimates, we corroborated these findings with 16S rRNA sequencing data and used GraphPad Prism 8.0.2 for Mann–Whitney U inter-group comparisons.

Results

Characteristics of the study cohort

Quality control removed three of 10 HC libraries (failures: 1 < 20,000 reads, one Phred < 25, 1 > 10% chimeras), leaving 10 TD and seven HC samples for downstream analyses (retention rate 85%). Mean age did not differ between HC (7.6 ± 2.1 years) and TD (7.9 ± 1.9 years) children (P = 0.78). Likewise, distribution of diet patterns and daily milk/yogurt intake was comparable across groups (all P > 0.60, Table S1).

Diversity of the GM in TD individuals

We assessed gut microbial community variation through 16S rRNA gene sequencing and analysed alpha diversity with the Shannon, Simpson, and Chao1 indices. The Chao1 index indicated significant group differences, unlike the Shannon and Simpson indices (Figs. 2A–2C). Beta diversity, analysed by principal component analysis (PCoA), revealed marked compositional distinctions in the gut microbiota between the TD and HC groups (Fig. 2D). At the phylum level, Bacteroidetes and Actinobacteria were more abundant in the TD group, with a concurrent decrease in Firmicutes, leading to a lower Firmicutes-to-Bacteroidetes ratio (Fig. 2E). At the genus level, Megamonas, Bifidobacterium, and Bacteroides were more prevalent in the TD group, and Prevotella was uniquely identified in this group. Conversely, Phascolarctobacterium, Dialister, Faecalibacillus, and Streptococcus were more dominant in the HC group (Fig. 2F).

Figure 2. Comparison of GM diversity and composition between HC and TD groups.

(A–C) Alpha diversity was assessed using Chao1, Shannon, and Simpson indices. (D) Beta diversity was analysed via PCoA. (E–F) Relative abundance histograms show phyla and genera in both groups, with top 10 abundant genera are color-coded, with less abundant ones grouped as ‘other’. NS, not significant; *: P < 0.05.

Marker genera in TD children

Given the limitations inherent in 16S rRNA amplicon pyrosequencing, our study focused on downstream analyses at the genus level. We utilized the Wilcoxon rank sum test (LEfSe) (LDA > 2) to identify differentially abundant marker genera between the two groups. Compared with the HC group, the TD group presented a reduced abundance of marker bacteria (Fig. 3A). The genera that were indicative of the TD group included Lachnospira, Gemmiger, Unassigned, Lachnospiracea incertae sedis, Faecalibacterium, and Granulicatella. Conversely, the HC group was dominated by marker genera such as Holdemania, Hungatella, Coprobacillus, Eggerthella, Romboutsia, Akkermansia, Streptococcus, Faecalibacillus, and Dialister (Fig. 3B).

Figure 3. Distinctive gut microbiota markers in HC and TD groups.

(A) Phylogenetic tree illustrating hierarchical relationships of marker taxa from phylum to species. (B) Significantly altered phylotypes at various taxonomic levels between groups.

IV selection

On the basis of the established criteria, a total of 1,531 SNPs associated with 131 different genera were chosen as the IVs for further MR analyses. The F statistics of these chosen SNPs varied between 14.58 and 88.42, suggesting a lower likelihood of bias from weak instruments.

Two-sample MR analyses

IVW-MR linked four of the six genera to TS after BH-FDR correction (q ≤ 0.05, Table 1).

Table 1. Results of MR analyses between GM and TS.

Exposure	Method	nsnp	β	SE	p-val	OR	95% CI	q-val
Anaerotruncus	MR Egger	12	−0.685	0.499	0.200	0.504	0.190–1.340	0.083
	Weighted median	12	−0.486	0.205	0.017	0.615	0.412–0.919
	IVW	12	−0.378	0.154	0.014	0.685	0.507–0.926
	Simple mode	12	−0.680	0.349	0.077	0.507	0.256–1.004
	Weighted mode	12	−0.601	0.307	0.076	0.548	0.301–1.000
Butyrivibrio	MR Egger	15	−0.133	0.292	0.656	0.875	0.494–1.550	0.032
	Weighted median	15	−0.160	0.085	0.059	0.852	0.722–1.006
	IVW	15	−0.150	0.062	0.016	0.860	0.761–0.972
	Simple mode	15	−0.184	0.148	0.234	0.832	0.622–1.112
	Weighted mode	15	−0.166	0.137	0.245	0.847	0.647–1.108
Dialister	MR Egger	11	0.790	0.571	0.200	2.203	0.719–6.750	0.050
	Weighted median	11	0.225	0.172	0.190	1.253	0.895–1.755
	IVW	11	0.276	0.135	0.041	1.318	1.011–1.719
	Simple mode	11	0.235	0.275	0.412	1.265	0.738–2.168
	Weighted mode	11	0.231	0.275	0.420	1.260	0.735–2.161
Ruminiclostridium 6	MR Egger	15	0.010	0.274	0.971	1.010	0.590–1.729	0.045
	Weighted median	15	0.160	0.147	0.276	1.174	0.880–1.566
	IVW	15	0.244	0.112	0.030	1.276	1.024–1.591
	Simple mode	15	0.084	0.222	0.711	1.087	0.704–1.680
	Weighted mode	15	0.109	0.200	0.593	1.115	0.7540–1.650
Ruminococcaceae UCG-002	MR Egger	22	−0.031	0.305	0.919	0.969	0.533–1.763	0.047
	Weighted median	22	−0.245	0.150	0.102	0.783	0.583–1.050
	IVW	22	−0.258	0.107	0.016	0.773	0.627–0.952
	Simple mode	22	−0.238	0.263	0.376	0.788	0.471–1.320
	Weighted mode	22	−0.224	0.249	0.378	0.799	0.490–1.302
Sutterella	MR Egger	11	−0.457	0.808	0.585	0.633	0.130–3.083	0.103
	Weighted median	11	0.089	0.204	0.662	1.093	0.733-1.631
	IVW	11	0.274	0.168	0.103	1.316	0.946–1.381
	Simple mode	11	−0.101	0.317	0.757	0.904	0.486–1.682
	Weighted mode	11	−0.070	0.328	0.836	0.933	0.490–1.775

Higher abundance of Ruminococcaceae UCG-002 and Butyrivibrio protected against TS (OR 0.773, 95% CI [0.627–0.952], q = 0.047; OR 0.860, 95% CI [0.761–0.912], q = 0.032), whereas Dialister and Ruminiclostridium 6 increased risk (OR 1.318, 95% CI [1.011–1.719], q = 0.050; OR 1.276, 95% CI [1.024–1.591], q = 0.045). Anaerotruncus showed a suggestive protective effect (OR 0.685, 95% CI [0.507–0.926], q = 0.083); Sutterella was null (OR 1.316, 95% CI [0.946–1.381], q = 0.103) and served as a negative control (Fig. 4 and Table 1).

Figure 4. (A) Forest plot of the association between GM taxa and TS risk based on IVW analyses.

Red indicates potentially harmful bacteria (enriched in TS), and green indicates potentially protective bacteria (enriched in HC). (B–G) Scatter plots indicating possible connections between GM and TS risk. Each dot represents one SNP. The five colors correspond to the five MR methods. Upward-sloping lines indicate positive associations (potentially harmful bacteria), and downward-sloping lines indicate inverse associations (potentially protective bacteria).

Sensitivity analyses

Cochran’s Q, MR-Egger intercept and MR-PRESSO global tests showed no indication of heterogeneity or directional pleiotropy for any genus (Egger intercept p > 0.100, Cochran’s Q p > 0.050, Table 2). Leave-one-out plots revealed that the IVW estimates for Anaerotruncus, Butyrivibrio, Dialister, Ruminiclostridium 6, Ruminococcaceae UCG-002 and Sutterella remained essentially unchanged after sequential removal of individual SNPs (Fig. 5), supporting the robustness of the causal inferences.

Table 2. Heterogeneity and pleiotropy tests for MR analyses of GM.

Expose	Heterogeneity			Pleiotropy		MR-PRESSO
	Methods	Cochran’s Q	p-value	MR-Egger intercept	p-value	p-value
Anaerotruncus	MR Egger	12.108	0.278	0.220	0.532	0.394
	IVW	12.617	0.319
Butyrivibrio	MR Egger	6.418	0.930	−0.002	0.952	0.963
	IVW	6.422	0.955
Dialister	MR Egger	5.316	0.806	−0.039	0.379	0.795
	IVW	6.173	0.801
Ruminiclostridium 6	MR Egger	6.405	0.930	0.023	0.367	0.924
	IVW	7.279	0.923
Ruminococcaceae UCG-002	MR Egger	21.043	0.395	−0.018	0.437	0.422
	IVW	21.705	0.417
Sutterella	MR Egger	13.331	0.148	0.050	0.378	0.154
	IVW	14.603	0.147

Figure 5. (A–F) The leave-one-out plots of the MR results of six genera associated with TS.

Validation of MR

Following the MR estimates, we validated the six selected genera by comparing their relative abundances between TD patients and controls with 16S rRNA sequencing data and presented the results in box plots. Interestingly, the prevalence of Anaerotruncus, Butyrivibrio and Ruminococcaceae UCG-002 were considerably greater, whereas the prevalence of Dialister and Ruminiclostridium 6 were lower in the healthy control group than that of the TD group (Fig. 6). This alignment between the MR analyses outcomes and the sequencing data underscores the possibility that these particular microbial genera may possess either protective or detrimental properties concerning the progression of TD.

Figure 6. Box plots.

(A–F) show the relative gene abundance of Anaerotruncus (A), Butyrivibrio (B), Dialister (C), Ruminiclostridium 6 (D), Ruminococcaceae UCG-002 (E) and Sutterella (F) in HCs and TDs. *: p < 0.05, **: p < 0.01, ***: p < 0.001, ns: not significant.

Reverse MR analyses

Reverse MR analysis was conducted through the IVW method to evaluate whether genetic liability to Tourette syndrome (TS) causally influences the six target gut microbial genera. As shown in Table 3, no significant reverse causal association was observed between TS and any of the investigated taxa (all p ≥ 0.069).

Table 3. Reverse MR: effect of TS liability on GM.

Expose	Outcome	Method	nsnp	β	SE	pval
Tourette Syndrome	Anaerotruncus	IVW	18	0.006	0.021	0.763
	Butyrivibrio	IVW	13	−0.096	0.053	0.069
	Dialister	IVW	16	−0.043	0.026	0.099
	Ruminiclostridium 6	IVW	17	−0.040	0.024	0.098
	Ruminococcaceae UCG-002	IVW	17	−0.038	0.022	0.079
	Sutterella	IVW	16	−0.019	0.025	0.440

Discussion

In this study, we employed an integrated pipeline that combined two-sample MR with 16S rRNA gene sequencing to explore whether gut microbiota composition might be associated with TD susceptibility, leveraging publicly available GWAS summary statistics as genetic instruments. We found that genetically predicted higher abundance of Ruminococcaceae UCG-002 and Butyrivibrio exerted a protective effect on TD, whereas Dialister and Ruminiclostridium 6 increased TD risk. Anaerotruncus showed a suggestive protective association with TD. Sutterella displayed no causal effect and was therefore retained as a negative control. These MR predictions were independently validated in a pediatric 16S rRNA cohort: TD cases displayed GM shifts concordant with MR directions. Collectively, these data provide preliminary evidence suggesting that GM may be linked to TD susceptibility.

Emerging evidence underscores the presence of a bidirectional communication system involving the central nervous system, GM, and digestive tract, collectively termed the microbiota−gut−brain axis (MGBA) (Sorboni et al., 2022). Disruptions in the GM can potentially alter brain function, cognition, and behavior (Sorboni et al., 2022). Research has consistently identified variations in GM composition between individuals with TD and HC, although these findings are not entirely consistent across studies (Fan et al., 2022). A significant correlation has been established between TD severity and both the abundances and metabolic functions of the GM (Xi et al., 2021). Studies suggest that fecal microbiota transplantation (FMT) may mitigate symptoms in children with TD (Vendrik et al., 2020). These preliminary associations suggest that the GM may be involved in TD pathogenesis and that targeting GM imbalances warrants further investigation as a potential therapeutic direction.

The GM primarily interacts with the central nervous system through the nervous, endocrine, immune and metabolic systems, which work together to affect brain function (Li et al., 2017). First, the vagus nerve, a key part of the MGBA, significantly influences emotional disorders and stress-induced and inflammatory diseases. Initial findings indicate that gut bacteria may influence psychiatric illnesses, in part through their impact on vagus nerve activity (Breit et al., 2018). For example, short-chain fatty acids (SCFAs), such as butyric acid, which are produced by the gut microbiome, can directly influence the terminals of vagal afferents (Liu et al., 2021; Li et al., 2022a). Butyrivibrio and Anaerotruncus are butyrate producers (O’Hara et al., 2018; Eslabão et al., 2022). Butyrate enhances vagal afferent signaling, increases cholinergic anti-inflammatory output, and strengthens the intestinal barrier, indirectly stabilizing the cortical-striatum-thalamo-cortical (CSTC) circuit activity (Chakrabarti & Chattopadhyay, 2025). Ruminiclostridium 6, though also a butyrate producer, harbors LPS synthesis gene clusters (Li et al., 2019). Elevated lipopolysaccharide (LPS) shifts vagal signaling toward a pro-inflammatory mode, which may dysregulate CSTC loops and worsen tics (Rojas-Colón et al., 2021; Bao et al., 2023). Collectively, these microbiota-driven vagal, inflammatory and barrier mechanisms converge on CSTC-circuit modulation, providing a plausible biological substrate for the observed link between gut dysbiosis and tic exacerbation.

Disruption of the balance of inhibitory and excitatory signals within the cortical-striatum-thalamus-cortical (CSTC) circuitry is considered the underlying molecular mechanism of TD. Abnormalities in multiple neurotransmitters have been associated with TD, with particular attention given to the dopaminergic, adrenergic, GABAergic, and glutamatergic pathways (Liu et al., 2020). Studies have indicated that the GM influences psychiatric conditions through the production of various neurotransmitters, such as GABA, 5-hydroxytryptamine (HT), histamine, and catecholamines (Li et al., 2022a). Anaerotruncus and Butyrivibrio express genes involved in tryptophan-indole metabolism, producing indole-lactate, a ligand for the aryl hydrocarbon receptor (AhR) (Huang et al., 2024). AhR activation modulates hypothalamic-pituitary-adrenal (HPA) axis reactivity and supports 5-HT and GABA balance, which can reduce tic severity and comorbid anxiety (Aguiniga et al., 2019). Ruminiclostridium 6 diverts tryptophan toward kynurenine, leading to glutamate excitotoxicity and HPA hyper-reactivity—both implicated in TD (Troubat et al., 2021; Spehlmann et al., 2022). Dialister produces propionate, can cross the blood–brain barrier and alter neuroendocrine signaling, potentially disrupting circadian and stress hormone regulation (Wu et al., 2021). Together, these GM-driven shifts in AhR tone, kynurenine flux and propionate signalling point to a possible link with CSTC imbalance and tic severity.

Post-mortem and imaging studies consistently report microglial activation and morphological priming within the basal ganglia of individuals with TD (Martino, Johnson & Leckman, 2020). GM-derived metabolites gate this process: butyrate, GABA and 5-HT produced by commensals regulate microglial maturation, whereas the imbalance in these signals skews microglia toward an M1 phenotype that releases IL-1β, TNF-α and ROS, ultimately compromising striatal synaptic pruning and motor-cognitive circuits. This indicates that butyrate suppresses the activity of histone deacetylase (HDAC) activity, modulating Treg function and influencing mild cognitive impairment progression (Yang et al., 2023). Mechanistically, butyrate from Ruminococcaceae UCG-002, Anaerotruncus and Butyrivibrio crosses the blood–brain barrier, inhibits HDAC2/3, and promotes M2 anti-inflammatory polarization, thereby lowering IL-1β, TNF-α and oxidative stress in the caudate–putamen (O’Hara et al., 2018; Verhoeven et al., 2021; Eslabão et al., 2022; Kang et al., 2025). Conversely, Dialister-generated propionate and LPS-rich outer-membrane vesicles from Ruminiclostridium 6 trigger TLR4–NF-κB signaling, driving M1 polarization and increasing microglial density in the striatum—a pathology repeatedly documented in TD (Li et al., 2019; Rong et al., 2021). Thus, GM-derived butyrate and propionate/LPS act as yin–yang regulators of striatal microglia, offering a preliminary mechanistic explanation for microbiota-associated fluctuations in tic severity.

GM-derived short-chain fatty acids (SCFAs)—acetate, propionate, and butyrate—are the dominant metabolites of dietary-fiber fermentation and serve as both energy substrates and signaling molecules within the gut–brain axis (Sorboni et al., 2022; Ou et al., 2023). While low butyrate levels may increase intestinal permeability and the risk of psychiatric disorders, adequate amounts enhance the gut barrier, reduce LPS-driven inflammation, and exert anti-inflammatory effects by lowering TNF-α-induced cytokine and monocyte adhesion (Wang et al., 2019; Ahrens et al., 2021). Butyrate produced by Ruminococcaceae UCG-002, Butyrivibrio and Anaerotruncus strengthens epithelial tight-junction integrity, dampens systemic endotoxemia, and provides β-oxidation fuel for enteric neurons and glia, thereby lowering oxidative stress (Loh et al., 2024; Krauze et al., 2025). Conversely, Dialister-generated propionate can inhibit mitochondrial complex-I activity and elevate ROS (Buchanan et al., 2023). Ruminiclostridium 6 modulates bile-acid composition, a shift linked to increased gut permeability and metabolic dysregulation (Lv et al., 2023). Nevertheless, research examining the link between SCFAs and TD remains scarce.

It is essential to acknowledge the substantial impact of diet on GM. Fiber-rich diets expand Ruminococcaceae UCG-002, whereas plant oils and polyphenols boost Butyrivibrio and Anaerotruncus—three genera whose abundance protects against TD (Li et al., 2022b; Zou et al., 2024). Conversely, high sugar and saturated fat promote Dialister and Ruminiclostridium 6, which MR implicates as TD-risk genera (Ang, 2018). Diet therefore acts as a tunable lever for TD-relevant GM configurations.

There is no direct evidence that Sutterella exerts significant vagal, AhR or endocrine activity, displays appreciable endotoxicity, meaningfully modulates HDAC/TLR4 signaling, or alters microglial polarity; its MR-negative profile and negligible SCFA output are at least consistent with this absence of demonstrated effects (Kaakoush, 2020; Nguyen et al., 2023; Scuto et al., 2024; Dupraz et al., 2025). Associations with TD may therefore be more plausibly attributed to dietary or medication-related confounders rather than direct microbial effects, though further mechanistic clarification is still needed.

Strengths of our study include the use of two-sample Mendelian randomization, which limits both reverse causation and residual confounding by leveraging independent exposure-outcome datasets; genetic instruments were extracted from the largest Tourette-disorder GWAS meta-analysis available (PGC-TS 2019, 4,819 cases/9,488 controls), ensuring robust and well-powered IVs. Explicitly modeling Sutterella as a negative-control genus further guards against false-positive claims and underscores the credibility of the observed causal signals. As a result, our study likely offers more compelling evidence than other observational studies do.

Nevertheless, several limitations merit consideration. All GWAS participants were of European descent, restricting generalizability to other ancestries. The 16S rRNA profiling employed in the replication cohort provides genus-level resolution only, precluding species- or strain-specific insights that might refine mechanistic hypotheses. Finally, the small clinical set limits precision and power, so our findings are exploratory.

Power analyses

Using G*Power (Version 3.1) software, we calculated the required sample sizes for an effect size of 0.7429, a significance level of 0.05, and a target power of 0.80. The results indicated that the TD group needed 44 samples and that the HC group needed 31 samples. However, the actual sample sizes were 10 and 7, respectively, resulting in a power of only 0.74. This shortfall likely explains why some results did not reach statistical significance and suggests that the study may have failed to detect actual effects.

Future research directions

Future studies should prioritize increasing sample sizes to increase the statistical power and reliability of the results. Specifically, the TD group would need 44 samples, and the HC group would need 31 samples to achieve 80% power. Additionally, using more precise measurement tools and improving the study design to assess effect sizes accurately can enhance the scientific rigor and practicality of the research. Multicenter studies and advanced statistical methods to control for confounding factors are also recommended to address these limitations.

Conclusions

By mapping microbial signals onto four major GM-CNS interaction pathways, we provide preliminary, mechanistically informed hypotheses that tentatively link specific gut genera to TD pathophysiology. Ruminococcaceae UCG-002, Anaerotruncus and Butyrivibrio point to a potentially protective association; Dialister and Ruminiclostridium 6 suggest possible risk indicators; while Sutterella appears to be a non-associated bystander. These exploratory findings offer pathway-specific candidate targets for future, larger-scale microbiome-based precision studies in TD.

Funding Statement

This work was supported by grants from the Hebei Provincial Health Commission through the 2023 Hebei Provincial Medical Science Research Project and Cangzhou People’s Hospital (20230259). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.