Emotional well-being and gut microbiome profiles by enterotype

The study investigates whether emotional well-being is associated with gut microbiome profiles in healthy adults and whether this association differs by gut enterotype. Context: The brain-gut-microbiota axis links neuroendocrine and immune pathways, with growing evidence that gut microbial communities relate to complex human behaviors and psychiatric conditions. Emotional disorders such as depression and anxiety often co-occur with gastrointestinal disorders, and longitudinal evidence suggests intestinal infections can predict later anxiety. Prior clinical and animal studies indicate that alterations in gut microbiota can modulate emotional behaviors and that patients with major depressive disorder show altered diversity and specific taxa differences. Purpose: To assess associations between gut microbiota composition and diversity with positive and negative affect and to test whether enterotypes (Bacteroides- vs Prevotella-dominant) moderate these associations. Hypothesis: Gut microbiome profiles are associated with emotional well-being, with differential patterns by enterotype.

Background work highlights: (1) Brain-gut-microbiota axis research links gut microbes with mental health, with comorbidity between emotional disorders and functional gastrointestinal disorders. (2) Longitudinal evidence shows intestinal infections predict future anxiety. (3) Clinical findings report altered fecal microbial diversity (Shannon index) and lower Faecalibacterium in major depressive disorder. (4) In healthy adults, personality traits (openness, conscientiousness) and higher quality of life correlate with specific genera (Faecalibacterium, Coprococcus, Lachnospiraceae) and greater diversity. (5) Psychobiotics research indicates probiotics and prebiotics can influence mood and emotion-related brain activation; fermented foods and probiotic supplements have reduced social anxiety, depressive symptoms, and negative cognitive reactivity in studies. (6) Enterotypes—Bacteroides, Prevotella, Ruminococcus—relate to long-term diet and have been linked to brain responses to emotional stimuli and to quality of life and depression scores, suggesting a moderating role in mental well-being. These studies motivate examining enterotype-specific associations between gut microbiota and emotion.

Design and participants: Cross-sectional analysis of a subset of the Korean Adult Longitudinal Study (KALS). KALS recruited 561 Seoul residents (2017–2018) via random-digit dialing with balanced demographics. A subset (n=85) consented to provide stool samples and completed surveys including PANAS and health questionnaires. Two samples were excluded for low read counts, yielding n=83 (mean age 48.9, SD 13.2). Measures: Emotional well-being assessed using the Positive and Negative Affect Schedule (PANAS), 20 items (10 positive, 10 negative), 5-point Likert scale over the preceding week (positive alpha=0.86; negative alpha=0.92). PANAS at stool collection was compared with PANAS measured approximately 1.5 years earlier (Wave 1). Behavioral/clinical covariates: self-reported abdominal pain (0–10), Bristol Stool Scale (1–7), antibiotic use in preceding month (yes/no), and dietary preference (1 meat–10 vegetables). Stool collection: Participants collected ~0.2 g stool in SB-01 kits (lysis buffer provided), shipped at room temperature (<3 days), then stored at -80 °C until analysis. Microbiome profiling: DNA extracted after homogenization, bead-beating, centrifugation; 16S rRNA V3–V4 regions amplified with primers 341F/805R; sequencing on Illumina MiSeq (ChunLab, Korea). Bioinformatics: Processing in EzBioCloud MTP pipeline. Quality filtering (Trimmomatic v0.32, Q<25 removed), read merging (VSEARCH v2.13.4), trimming barcodes/primers/linkers, removal of non-16S sequences (HMMER v3.2.1), clustering redundant reads (VSEARCH), taxonomic assignment (VSEARCH against EzBioCloud DB), chimera removal (UCHIME with EzBioCloud non-chimeric DB), de novo clustering for reads <97% similarity. All 83 samples had Good’s coverage >99%. Copy number adjusted; count normalization to minimum 11,289 reads. Diversity metrics: Alpha diversity via QIIME included Shannon (evenness) and species richness indices (Observed, Chao1, ACE). Observed, Chao1, ACE were log-transformed due to skew. Enterotype analysis: Partitioning Around Medoids (PAM) clustering using Jensen–Shannon Distance computed on genus-level relative abundances (R ‘cluster’ package). Optimal cluster number set to two by maximum Calinski–Harabasz index (CH=60.450). LEfSe used for differentially abundant taxa between clusters. Statistical analysis: Group comparisons by t-tests or chi-square. Generalized linear models tested associations between diversity indices and PANAS scores, controlling for age, sex, and antibiotic use; interaction terms tested moderation by enterotype. MaAsLin2 multivariate models (R package) examined associations between genus-level taxa (present in ≥80% of participants; n=52 taxa) and PANAS scores, controlling for age, sex, antibiotic use; Benjamini–Hochberg FDR applied; significance set at p<0.05 and q<0.25. Ethics: Approved by Seoul National University IRB (10-2018-21); informed consent obtained.

• Clustering/enterotypes: Two distinct clusters identified via PCoA and PAM: E1 Bacteroides-dominant (n=49) and E2 Prevotella-dominant (n=34). Bacteroides significantly more abundant in E1 and Prevotella in E2 (Wilcoxon p<0.001 for both). No differences between groups in age, stool form (BSS), food preference, abdominal pain, or PANAS scores; gender distribution differed (more females in Bacteroides group, more males in Prevotella group). - Diversity by enterotype: Prevotella group showed higher species richness. Independent t-tests: Observed index higher in Prevotella (t(81)=-2.12, p=0.037); Chao1 marginal (t(81)=1.94, p=0.056); ACE marginal (t(81)=-1.81, p=0.074). GLMs (controlling age, sex, antibiotics) confirmed higher richness in Prevotella: Chao1 b=0.14 (ref=E1), SE=0.06, p=0.026; ACE b=0.13, SE=0.06, p=0.031; Observed b=0.17, SE=0.07, p=0.011. Shannon index did not differ by enterotype (t(81)=1.17, p=0.24; GLM b=-0.12, SE=0.17, p=0.479). - Diversity and affect (overall): Shannon diversity was positively associated with positive affect (b=0.306, SE=0.15, p=0.045) but not with negative affect (b=0.114, SE=0.112, p=0.312). Observed, Chao1, and ACE were not significantly associated with affect. Interaction: positive affect × enterotype significant (b=0.618, SE=0.30, p=0.042); enterotype main effect also significant (b=2.08, SE=0.98, p=0.037). - Enterotype-specific associations: In the Prevotella group (E2), positive affect was significantly associated with higher Shannon diversity (b=0.61, SE=0.28, p=0.039); negative affect not associated (b=-0.10, SE=0.19, p=0.603). In the Bacteroides group (E1), neither positive (b=0.049, SE=0.172, p=0.775) nor negative affect (b=0.167, SE=0.132, p=0.212) was associated with Shannon. - Longitudinal affect (Wave 1, ~1.5 years earlier): Positive affect from Wave 1 associated with Shannon in Prevotella group (b=0.48, SE=0.27, p=0.087, trend), not in Bacteroides (b=-0.008, SE=0.14, p=0.95). Current and Wave 1 positive affect correlated (r=0.41, p<0.001). - Taxa-level associations (MaAsLin2, controlling covariates; taxa present in ≥80%): Higher Agathobaculum associated with lower negative affect (b=-0.014, SE=0.005, p=0.008, q=0.23). Higher Collinsella associated with lower positive affect (b=-0.051, SE=0.015, p=0.001, q=0.095). A novel Lachnospiraceae genus (PAC001043_g) associated with higher positive affect (b=0.014, SE=0.005, p=0.008, q=0.214) and lower negative affect (b=0.01, SE=0.004, p=0.01, q=0.23).

The findings address the research question by showing an enterotype-specific relationship between gut microbiome diversity and emotional well-being. Enterotypes did not directly differ in mood scores but moderated the association between mood and gut diversity: only the Prevotella-dominant group showed a significant positive link between positive affect and Shannon diversity. This aligns with prior neuroimaging findings of heightened emotional responses in Prevotella-high individuals and with literature associating microbiota diversity with beneficial psychosocial factors. The taxa-level associations suggest potential microbial contributors to affect (e.g., Lachnospiraceae PAC001043_g linked to higher positive and lower negative affect; Agathobaculum linked to lower negative affect; Collinsella linked to lower positive affect). The results support considering enterotypes as a framework to interpret gut-brain relationships and propose that emotional well-being may be more tightly coupled to gut ecosystem evenness/diversity under Prevotella-dominant profiles. Social and psychological factors known to influence both mood and the microbiome may partly mediate these relationships, underscoring the need for longitudinal and mechanistic studies.

This study demonstrates that emotional well-being, particularly positive affect, is associated with gut microbiome diversity and composition in healthy adults, with effects moderated by enterotype. Prevotella-dominant individuals show stronger coupling between positive affect and microbial evenness (Shannon diversity). Specific taxa, including a novel Lachnospiraceae genus, are linked to affective states. These enterotype-specific associations suggest potential for personalized, enterotype-informed interventions targeting the microbiome to support mental health. Future research should employ longitudinal and interventional designs, incorporate comprehensive health and dietary assessments, validate enterotype classifications across methods, and investigate mechanistic pathways and psychobiotic candidates identified here.

• Cross-sectional design precludes causal inference between emotional well-being and gut microbiome features. - Limited sample size, and especially subgroup sizes by enterotype (n=49 Bacteroides, n=34 Prevotella), may reduce statistical power. - Enterotype classification may vary by method (PAM used here; other methods like Dirichlet Multinomial Mixtures could yield different groupings). - Lack of detailed health status (e.g., diabetes, obesity) and dietary intake data limits control over potential confounding variables.