Evaluating the Genetic Effects of Gut Microbiota on the Development of Neuroticism and General Happiness: A Polygenic Score Analysis and Interaction Study Using UK Biobank Data

Limited efforts have been invested in exploring the interaction effects between genetic factors and gut microbiota on neuroticism and general happiness. The polygenic risk scores (PRS) of gut microbiota were calculated from individual-level genotype data of the UK Biobank cohort. Linear regression models were then used to assess the associations between individual PRS of gut microbiota and mental traits and interaction analysis was performed by PLINK2.0. KOBAS-i was used to conduct gene ontology (GO) enrichment analysis of the identified genes. We observed suggestive significant associations between neuroticism and PRS for the genus Bifidobacterium (rank-normal transformation, RNT) (beta = −1.10, P = 4.16 × 10−3) and the genus Desulfovibrio (RNT) (beta = 0.54, P = 7.46 × 10−3). PRS for the genus Bifidobacterium (hurdle binary, HB) (beta = 1.99, P = 5.24 × 10−3) and the genus Clostridium (RNT) (beta = 1.26, P = 9.27 × 10−3) were found to be suggestive positively associated with general happiness. Interaction analysis identified several significant genes that interacted with gut microbiota, such as RORA (rs575949009, beta = −45.00, P = 1.82 × 10−9) for neuroticism and ASTN2 (rs36005728, beta = 19.15, P = 3.37 × 10−8) for general happiness. Our study results support the genetic effects of gut microbiota on the development of neuroticism and general happiness.


Introduction
Mental disorders are highly prevalent, placing a heavy burden on global public health [1]. The estimated prevalence of mental disorders was 0.97 billion globally in 2017 [2]. According to a recent report, mental disorders have consistently accounted for more than 14% of age-standardized years lived with disability for nearly three decades [2]. Neuroticism and positive emotion are common personality traits that are often of concern [3,4]. Mental traits are the product of the interplay between genetic and environmental influences. The estimated heritability was 40~60% for neuroticism and 37~64% for positive emotion [5,6]. With regards to environmental influences, adversities during development, such as "emotional neglect and sexual abuse", were found to be positively associated with these two common mental traits [5,7,8]. However, the pathogenesis of mental traits remains elusive even now.
Gut microbiota co-develops with the host from birth and changes throughout growth depending on different physiological, pathological, dietary patterns, and environmental conditions [9]. Increasing evidence has indicated that the gut microbiota is closely related to host health and plays an important role in the etiology of a variety of complex human disorders, including mental illness [10][11][12]. Cryan et al. suggested that the gut microbiota is a pivotal part of the signaling along the microbiota-gut-brain axis, a bi-directional

Phenotypes Definition
Neuroticism score is derived based on 12 domains of neurotic behavior as reported from UK Biobank data fields 1920,1930,1940,1950,1960,1970,1980,1990,2000,2010,2020, and 2030 from the touchscreen questionnaire at baseline. Participants were assessed for the 12 domains of neurotic behaviors via the touchscreen questionnaire. Neuroticism score summarizes the number of "Yes" answers across the 12 questions into a single integer score for each participant. The detailed questionnaire is shown in Supplementary File S1. General happiness was collected from the response to the UK Biobank online "Thoughts and Feelings" mental health questionnaire: "In general how happy are you?" by choosing "Do not know (−121)", "Extremely happy (1)", "Very happy (2)", "Moderately happy (3)", "Moderately unhappy (4)", "Very unhappy (5)", "Extremely unhappy (6)", and "Prefer not to answer (−818)". The subjects whose answers are "Do not know (−121)" and "Prefer not to answer (−818)" were excluded from this study. The detailed definition of phenotypes is shown in Supplementary File S1.

GWAS Summaries of Gut Microbiota
The gut microbiota-associated SNPs were derived from a recently published GWAS [16]. Briefly, using fecal 16S ribosomal RNA gene sequences and host genotype data from the Flemish Gut Flora Project (n = 2223) and two German cohorts (n = 1667), Hughes et al. [16] identified several genetic associations involving multiple microbial traits. Sequencing was carried out on the Illumina HiSeq platform at the VIB Nucleomics core laboratory (Leuven, Belgium), with 500 cycles (sequencing kit HiSeq-Rapid SBS v.2). After merging paired sequences and the removal of chimeras, compositional matrices for each taxonomical level were carried out using the Ribosomal Database Project (RDP) training set 'rdp_train_set_16'. Microbial taxa were described as relative abundance profiles using the rank normal transformed (RNT) model, while those with zero-inflated abundance distributions were described using a hurdle binary (HB) model. For association analysis, any taxa that met the following two criteria were included: (1) comprise ≥ 5% of the reads for at least one individual; and (2) have ≥15% of individuals with non-zero data were included. In total, 114 taxa across all phylogenetic levels met these criteria. We selected microbial traits associated with SNPs with a significant threshold of P < 5 × 10 −6 to calculate the PRSs. A detailed description of sample characteristics, array design, quality control, and statistical analysis can be found in the previous study [16].

PRS Analysis
Using the genotype data of the UK Biobank cohort, PRS analysis was performed by using the PLINK's "-score" command [24]. Briefly, let PRSg denote the PRS value of each microbial trait for the gth subjects, where i (i = 1, 2, 3, . . . , l) and g (g = 1, 2, 3, . . . , k) denote the number of genetic markers and sample size, respectively. βi is the effect parameter of the risk allele of the ith significant SNP related to each microbial trait obtained from the previously published study. SNPig is the dosage (0 to 2) of the risk allele of the ith SNP for the gth subject. The PRS values were standardized to have a mean of 0 and a variance of 1 before further analyses. Using computed PRSs as the instrumental variables of gut microbiota, a linear regression model was finally used to detect the potential associations between gut microbiota and target traits. In this study, the significant association thresholds should be P < 2.19 × 10 −4 [0.05/(114 × 2)] after strict Bonferroni correction. The suggestive significance threshold was set as P < 0.05. All statistical analyses were performed using R3.5.3 (https://www.r-project.org/ (accessed on 5 February 2021)). Additionally, the sex, age, and 10 principal components of the population structure were used as covariates in the regression model.

Statistical Analysis
Based on the result of the regression model, interaction analysis was then conducted to explore the impact of the interaction between genetic factors and gut microbiota traits related to PRS on mental traits in the UK Biobank cohort. The outcome variables, including neuroticism and general happiness, were adjusted by sex, age, and 10 principal components of population structure. The instrumental variables are the PRS of gut microbiota traits. Based on the previous study, the genetic additive (ADD) model of PLINK2.0 was used in this study [25]. SNPs with a call rate <0.95, Hardy-Weinberg equilibrium testing of P < 0.001, and minor allele frequencies (MAFs) < 0.01 were excluded for quality control [25]. A significant threshold was set at P = 5.0 × 10 −8 . Circular Manhattan plots were generated using the "CMplot" R script (https://github.com/YinLiLin/R-CMplot (accessed on 3 April 2021)).

Gene Set Enrichment Analyses
To explore the functional relevance of identified genes interacting with individual PRS of gut microbiota for neuroticism and general happiness, gene ontology (GO) enrichment analyses of the identified target genes were performed by using KOBAS-i (short for KOBAS intelligent version) (http://kobas.cbi.pku.edu.cn/ (accessed on 29 December 2022) [26].

Basic Characteristics of Study Samples
The general characteristics of the subjects are presented in Table 1. For the association analysis of neuroticism and G_Bifidobacterium_RNT, 306,161 participants were selected; 161,977 of them were female, with mean age (SD) was 56.81 (7.93) years. For the association analysis of neuroticism and G_Desulfovibrio_RNT, 153,483 participants were selected; 81,310 of them were female, with mean age (SD) was 56.80 (7.93) years. For the association analysis of general happiness and G_Bifidobacterium_HB, 89,206 participants were selected; 49,806 of them were female, with mean age (SD) was 56.27 (7.62) years. For the association analysis of general happiness and G_Clostridium_sensu_stricto_RNT, 115,013 participants were selected; 64,229 of them were female, with mean age (SD) was 56.28 (7.61) years.

GO Enrichment Analysis Results
GO enrichment analysis identified five GO terms enriched in the significant genes interacting with the genus Bifidobacterium (RNT), such as GO:0005515-protein binding (P = 8.12 × 10 −3 ), and 15 GO terms enriched in the significant genes interacting with the genus Desulfovibrio (RNT) for neuroticism, such as GO:0005925-focal adhesion (P = 1.56 × 10 −2 ), GO:0045202-synapse (P = 1.57 × 10 −2 ). Fourteen GO terms enriched in the significant genes interacting with the genus Clostridium (RNT) for general happiness were identified, such as GO:0045666positive regulation of neuron differentiation (P = 2.46 × 10 −4 ). The detailed enrichment analysis results are summarized in Tables S3 and S4. Note: SNP, single-nucleotide polymorphism; G, genus; RNT, rank-normal transformation.

Figure 2.
Interaction analysis between SNP and gut microbiota for general happiness. Chromosomal regions interacting with G_Bifidobacterium_HB (a) and G_Clostridium_sensu_stricto_RNT (b) for general happiness. From the center, the first circos depicts the -log10 P of each variant owing to double exposure, i.e., the effect of both single nucleotide polymorphism allele and gut microbiota. The second circos shows chromosome density. Red plots indicate the P-value threshold for genomewide significance (P < 5 × 10 −8 ), while the green plots indicate the P-value threshold for suggestive significance (P < 5 × 10 −7 ). The plots were generated using the "CMplot" R script (https://github.com/YinLiLin/R-CMplot (accessed on 3 April 2021)). SNP, single-nucleotide polymorphism; G, genus; HB, hurdle binary; RNT, rank-normal transformation.

Discussion
In this study, we conducted PRS and interaction analyses to explore the relationship between gut microbiota and two common mental traits, neuroticism and general happiness. We found significant associations between gut microbiota and the risk of neuroticism and general happiness. In addition, we identified multiple genes which interacted with the gut microbiota influencing neuroticism and general happiness.
It has been demonstrated by previous studies that the gut microbiota is associated with certain behaviors and psychiatric disorders, which are consistent with our findings. For example, according to an observational and cross-sectional study in 672 adults, high neuroticism was correlated with a high abundance of Gammaproteobacteria and Proteobacteria, respectively, when age, sex, BMI, and nutrient intake were controlled as covariates. The results of the beta diversity analysis indicate a lower diversity and closer similarity Figure 2. Interaction analysis between SNP and gut microbiota for general happiness. Chromosomal regions interacting with G_Bifidobacterium_HB (a) and G_Clostridium_sensu_stricto_RNT (b) for general happiness. From the center, the first circos depicts the -log 10 P of each variant owing to double exposure, i.e., the effect of both single nucleotide polymorphism allele and gut microbiota. The second circos shows chromosome density. Red plots indicate the P-value threshold for genomewide significance (P < 5 × 10 −8 ), while the green plots indicate the P-value threshold for suggestive significance (P < 5 × 10 −7 ). The plots were generated using the "CMplot" R script (https://github. com/YinLiLin/R-CMplot (accessed on 3 April 2021)). SNP, single-nucleotide polymorphism; G, genus; HB, hurdle binary; RNT, rank-normal transformation.

Discussion
In this study, we conducted PRS and interaction analyses to explore the relationship between gut microbiota and two common mental traits, neuroticism and general happiness. We found significant associations between gut microbiota and the risk of neuroticism and general happiness. In addition, we identified multiple genes which interacted with the gut microbiota influencing neuroticism and general happiness.
It has been demonstrated by previous studies that the gut microbiota is associated with certain behaviors and psychiatric disorders, which are consistent with our findings. For example, according to an observational and cross-sectional study in 672 adults, high neuroticism was correlated with a high abundance of Gammaproteobacteria and Proteobacteria, respectively, when age, sex, BMI, and nutrient intake were controlled as covariates. The results of the beta diversity analysis indicate a lower diversity and closer similarity within paired samples in the neuroticism group [27]. Michels et al. suggested that adjusted and unadjusted taxonomic differences in the gut microbiota were the most pronounced for happiness, which was associated with 24 operational taxonomic units (=11.8% of bacterial counts) in 93 Belgian adolescents [28]. Another research from healthy Korean adults revealed that positive emotion and gut microbiome diversity (Shannon Index) were associated with substantial relationships in the Prevotella-dominating group [29].
We observed a suggestive significant negative correlation between the PRS of the genus Bifidobacterium and neuroticism and a positive correlation between the genus Desulfovibrio and neuroticism. Although evidence on the genus Bifidobacterium and neuroticism is limited, anxiety and depression, typical characteristics among populations with high neuroticism [30], have been shown to be strongly associated with Bifidobacterium. Animal experiments have observed that long-term exposure to a 150 mT static magnetic field improved the abundance of Bifidobacterium and Clostridium and further improved anxiety in C57BL/6J mice [31]. A recent clinical study also confirmed the potential of Bifidobacterium in the treatment of major depression [32]. They found that oral administration of freeze-dried Bifidobacterium breve CCFM1025 significantly reduced the serum serotonin turnover in patients compared to the placebo [32]. The genus Desulfovibrio is anaerobic and Gramnegative, also known as a sulfate-reducing bacterium. Recently, our group identified that Desulfovibrio was significantly related to attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD), bipolar disorder (BD), schizophrenia (SCZ), and major depressive disorder (MDD) [33]. As a potentially harmful bacterium, Desulfovibrio was found to be enriched in diseased individuals in inflammatory disease models [34,35]. The abundance of Desulfovibrio was negatively correlated with short-chain fatty acids (SCFAs), the production of microbial metabolites [34]. SCFAs, such as acetate, butyrate, and propionate, are important immunomodulatory and anti-inflammatory molecules in the gut and have shown promising therapeutic effects on symptoms of depression and anxiety in animal studies [36]. Despite limited research on Desulfovibrio and neuroticism, further multi-omics studies integrating metagenomics, transcriptomics and metabolomics may provide the basis for revealing the mechanisms underlying this interaction.
We also observed suggestive significant positive correlations between the PRS of the genus Bifidobacterium and the genus Clostridium with general happiness. The genus Clostridium is Gram-positive, and most of them are non-pathogenic bacteria. A recent comparative study identified a significantly lower relative abundance of Clostridium in patients with schizophrenia than in healthy controls [37]. Emerging evidence supported the beneficial role of Clostridium in the anti-inflammatory activity, immune protection, and remodeling of the gut microbiome [38,39]. For example, oral administration of Ataining (containing Clostridium butyricum, CGMCC0313.1) to gastrectomized patients resulted in reduced early postoperative inflammation, enhanced immune ability, restored intestinal microbiota eubiosis, and increased intestinal SCFAs [38]. Similarly, Clostridium butyricum SLZX19-05 treatment of weaned piglets inhibited inflammation levels, remodeled the ileal microbiome and increased propionate production [39]. Since inflammation and immune system activation play an important role in brain function and the development of mental health [40], the above evidence also supports to some extent the relationship between Clostridium and general happiness. Collectively, genus Bifidobacterium and Clostridium are promising candidate psychological gut microbiota that may improve emotional health. These mechanisms may be related to the production of neuroactive substances, as well as anti-inflammatory effects and immune system activation. These findings support the future clinical use of psychobiotics to improve emotional health.
Notably, we found that the PRS of the genus Bifidobacterium was suggestive negatively correlated with neuroticism but positively correlated with general happiness. Mechanistically, as a psychobiotic, Bifidobacterium has been reported to secrete γ-aminobutyric acid (GABA, the "happy" chemical) [41]. Moreover, supplementation of Bifidobacterium could increase tryptophan, a precursor of serotonin (another "happy" chemical) [42]. Neuroticism, a health-related personality factor that includes negative emotions such as anxiety, moodiness, and depression [43], is opposed to the positive trait of general happiness. Taken together, the positive correlation between the genus Bifidobacterium and general happiness and the negative correlation between Bifidobacterium and neuroticism can be speculated.
The interaction effects between genetic factors and gut microbiota traits for mental traits remain largely unknown now. The key element of this study is that we conducted an interaction analysis of mental traits and identified multiple loci and candidate genes for the modulation of genetic response to the gut microbiota, which may provide novel insights to help disentangle its underlying etiological mechanisms.
Interaction analysis identified several candidate genes interacting with gut microbiota for neuroticism, such as RORA, KCNQ3, and NDUFS1. RORA is a member of the nuclear hormone receptor superfamily. Genetic association studies have identified that RORA is a suggestive gene linked more specifically to the depression facet of neuroticism [44]. In the mouse nervous system, RORA is localized in the cerebellum, thalamus, cerebral cortex, suprachiasmatic nucleus, and other structures [45]. As recently demonstrated by multiple studies of animal models, GWA, and linkage studies converging on variants in the RORA gene, it may be linked to bipolar disorder and trait depression [46,47]. RORA proteins play an important role in the maintenance of circadian rhythms [48]. Clinically, the disruption of circadian rhythms may contribute to anxiety disorders [49]. In addition, RORA proteins are involved in protecting neurons and glial cells from oxidative stress-induced apoptosis [50], which is a potential mechanism implicated in the pathophysiology of depression and anxiety disorders. Direct evidence for RORA and neuroticism, although limited, suggests these physiological roles of RORA protein may support the biological plausibility of its association with neuroticism. Sands et al. revealed that specific gain-of-function variants in KCNQ3 cause neurodevelopmental disability, autism, and ASD [51]. In addition, Kaminsky et al. suggested that epigenetic alterations in the KCNQ3 gene may be important in the aetiopathogenesis of bipolar disorder [52]. As a voltage-gated potassium channel, KCNQ is an important regulator of cell membrane excitability [53]. Preclinical studies suggest that the KCNQ channel is a potential target for the treatment of depression and anhedonia [54]. Specifically, KCNQ2 and KCNQ3 form homologous or heterodimers (such as KCNQ2/3 channels), which constitute M-shaped channels regulating nerve excitability [55]. Up-regulation of the KCNQ3 channel restored the hyperactivity of dopamine neurons and reversed depressive behavior in susceptible mice [56]. A genetic variant in the NDUFS1 gene was reported to be associated with schizophrenia and negative symptoms in Han Chinese subjects [57]. According to a previous study, neuroticism may share common genetic variants with most mental disorders, such as BD and MDD [58]. NDUFS1 knockdown in neurons reduced complex I affiliation into supercomplexes, resulting in decreased oxygen consumption and increased mitochondrial reactive oxygen species [59]. Further functional studies are warranted to validate the role of those genes in neuroticism.
ASTN2 encodes a large vertebrate-specific transmembrane protein that is expressed primarily in developing and adult brains. We observed that it interacted with the genus Clostridium to influence general happiness. In recent studies on ASTN2 and general happiness, although very limited, copy number variants of ASTN2, both deletions and dupli-cations, have been identified in patients with neurodevelopmental disorders, including ASD, SCZ, ADHD, BD, intellectual disability, and global developmental delay [60][61][62][63]. The hippocampus is thought to be involved in emotional and cognitive regulation [64]. A dataset test of 724 twins and siblings found a non-linear correlation between subjective well-being (SWB) and hippocampal volume, characterized by lower SWB in subjects with relatively smaller hippocampal volume compared to those with medium and high hippocampal volume [65]. A previous GWAS study identified nine SNPs at 9p33 in ASTN2 that were significantly associated with hippocampal volume [66]. Another GWAS study also identified one novel independent locus that was significantly associated with hippocampal volume lying within the ASTN2 gene [67]. Emotional or stressful events, on the other hand, are associated with a large release of corticosterone, which rapidly enhances synaptic plasticity [68]. ASTN2 can play a role in emotional events by regulating synaptic strength through endocytic transport and the degradation of surface proteins [69]. Conclusively, the above studies suggest that ASTN2 may influence general happiness by altering the hippocampal volume or regulating synaptic strength, which supports our findings.
GO enrichment analysis identified multiple GO terms associated with neuroticism and general happiness. One notable finding is focal adhesion (GO:0005925). It plays an important role in cell migration during brain development. Previous studies suggest that focal adhesion and extracellular matrix receptor interaction pathways are significantly associated with certain facets of neuroticism (i.e., angry hostility and depression) [70]. It has also been reported that focal adhesion is involved in abnormal neurodevelopment in SCZ patients compared with healthy control subjects [71]. Synapse (GO:0045202) is another notable GO term that we found. A recent finding has raised the importance of glypicans in the development and functions of synapses, and the dysfunctions of glypicans may lead to the malfunction of synapses and abnormal neurodevelopmental disorders, such as neuroticism and SCZ [72].
It is important to emphasize that our study has certain limitations. First, like GWAS, some significant SNPs found by the interaction analysis are located in the non-coding region, which still poses a challenge for us to better illustrate our results. Second, all subjects in this study are of European ancestry. Therefore, one should be careful when applying our study results to other ethnic groups. Next, because the use of antibiotics may disrupt the composition of the microbiota [73], future studies to explore the impact of antibiotic use on mental traits by affecting gut microbiota are warranted. Further experiments are needed to confirm our findings and reveal the potential molecular mechanisms underlying the associations observed in this study. Finally, the gut microbiota-related SNP sets were derived from previous GWAS. The accuracy of our findings may be influenced by the power of the previous study. Further replication studies with other genetic background individuals and experimental studies are required to verify the results of this study.

Conclusions
In conclusion, we observed correlations between gut microbial traits and neuroticism and general happiness in the UK Biobank cohort. Interaction analysis identified multiple candidate genes which may serve as the underlying genetic mechanisms of the observed association. Our study findings could provide novel insights into the impacts of gut microbial traits on the two common mental traits.