Gut Microbiota and Metabolome Alterations Associated with Chinese Monks’ Practice

of include which the gut and We of microbiotas and function by affecting the gut microbial composition and metabolites. The appropriate practice of monks makes the intestine younger and increases immunity, but long-term practice may cause adverse physical and mental events.


Introduction
Gut microbes have evolved with humans since the emergence of humans and have developed symbiotic relationships to ensure the survival of microbes and hosts [1]. Many studies have con rmed that the intestinal ora is closely related to the regulation of host immunity and in ammation [2,3]. Therefore, doctors treat a variety of diseases and improve human immunity by regulating intestinal microorganisms and intestinal metabolites, such as fecal bacteria transplantation [4,5].
In recent years, many types of studies have shown that vegetarianism and meditation are bene cial for human physical and mental health. Different dietary habits affect the composition of intestinal ora and then in uence physiological, metabolic, nutritional, and immune functions [7][8][9][10][11]. Vegetarianism effectively prevents metabolic diseases such as cardiovascular disease [9], diabetes [11], and obesity [8], as well as some cancers [12,13]. Meditation by reducing stress and improving mental health has been widely studied for the treatment of various psychological, cardiovascular and digestive diseases, as well as relieving chronic pain [14,15]. Additionally, meditation regulates overall health through the neuroendocrine-immune network [16][17][18][19][20], thus affecting the characteristics of the intestinal ora.
As a special group, the practices of monks mainly includes the long-term vegetarian diet and focusing on meditation.
Therefore, feces of monks are excellent data to study the synthetic effects of vegetarianism and meditation on intestinal changes. We aimed to investigate the effects of monks' long-term meditation and vegetarianism on the regulation of intestinal ora and metabolites.

Results
The altered microbiota composition of monks compared with healthy omnivorous subjects We summarized the demographic information, including age, BW, BMI, and the practice lasting time of samples in the monk and control groups, as shown in Table 1. Eleven phyla, 221 genera, and 842 OTUs were obtained from 24 monk and 48 control stool samples. We con rmed that gut community structures signi cantly differed between the two groups in beta diversity (p = 0.001, Figure 1A), and the monk group had a clearer structural separation of microbial communities by PCoA analysis ( Figure 1B).
At the phylum level, the composition analysis of the intestinal microbial communities showed that both groups were dominated by four phyla; among them, the abundance of Bacteroides increased in monks, while the Firmicutes and Actinobacteria were increased in controls ( Figure 1C, D, E). Interestingly, a signi cant difference in the Firmicutes/Bacteroidetes (F/B) ratio was observed between the two groups (p < 0.05, Figure 1F). Furthermore, the type analysis according to the distribution of dominant organisms on JSD distance demonstrated that the samples were clustered into seven types (Figure 2A-B). Speci cally, monks were clustered into types 1-3, while controls were clustered into types 1-7, suggesting that monks have a lower microbiota diversity of phyla. There were a greater number of type 1-2 cases in the monks, mainly including Firmicutes and Bacteroidetes ( Figure 2B). Of note, Bacteroidetes accounted for the largest proportion in types 1-2 ( Figure 2C), which was consistent with the above results.
At the genus level, the Wilcoxon rank-sum test (mean abundance level of monk group > 1%, p < 0.05, Figure 2D) indicated 13 differentially expressed genera between the two groups. LEfSe analysis (LDA > 3.0, p < 0.05, Figure 2E) found that 33 genera could signi cantly distinguish the two groups. Speci cally, the gut microbial composition characteristics in the two groups are shown in Figure 2F. Among 33 differential genera, we conducted random forest analysis to assess the value of these genera and selected the top nine important genera to serve as diagnostic biomarkers. The AUC of the ROC curve was 0.7795 [95% con dence interval (CI): 0.6686-0.8904], Figure 3A), suggesting that these 9 genera could be employed as combined predictors. For the 9 genera, Eubacterium__eligens_group, Lachnospira, Roseburia, and norank_f__Lachnospiraceae were increased in monks, while Blautia, Dorea, Eubacterium__hallii_group, and Streptococcus were overexpressed in controls.
We analyzed the Spearman correlation between 33 differential genera and monks' general characteristics ( Figure 3B). It is worth mentioning that the ora related to the monk's age and the practice duration was different. The results demonstrated that monk age was negatively correlated with Klebsiella (Rho = -0.461, p = 0.023). However, practice duration was positively correlated with norank_f_Lachnospiraceae (Rho = 0.542, p = 0.006) but negatively correlated with Eubacterium__hallii_group (Rho = -0.477, p = 0.018). Therefore, the in uence of monk practice duration on intestinal ora was different from that of age. Monk practice duration is an independent factor that in uences intestinal ora.
Tax4Fun and PICRUSt2 were used to reveal the functional and metabolic-related changes in the gut microbiota in monks. We found that 308 KEGG categories were obtained using PICRUSt2, while 268 KEGG categories were obtained using Tax4Fun, indicating that the functional prediction of Tax4Fun was relatively conservative. The LEfSe analysis (LAD > 2.5, p < 0.05, Figure 3C-D) showed that 35 categories of Tax4Fun and 20 categories of PICRUSt2 were signi cantly different between the groups. For the eight identi ed common categories in both Tax4Fun and PICRUSt2 (Table 2 and Figure 3E), the majority were related to metabolism (6/8, 75.0%), including carbohydrate metabolism, energy metabolism, and glycan biosynthesis and metabolism, while the minority were related to environmental information processing (2/8, 25.0%), such as membrane transport. Notably, at KEGG pathway level 3, citrate cycle (TCA cycle), carbon xation pathways in prokaryotes, other glycande gradation, and lipopolysaccharide biosynthesis were overrepresented in monks, while ABC transporters and phosphotransferase system (PTS) were overrepresented in controls. Spearman correlation was analyzed between the eight common KEGG categories and 33 differential genera ( Figure 3F). The majority of the genera with the highest abundance in monks were positively correlated with the functional pathway enriched in the monk group, indicating that these ascendent genera in uence the alteration of gut metabolism function.
Analysis of metabolic spectrum and differences between the monks and the controls The nontargeted LC-MS metabolomics results demonstrated that a total of 34 differential metabolites (VIP >2, p < 0.05) were successfully identi ed. A majority of differential metabolites were lipids and lipid-like molecules according to the HMDB database, as shown in Table S1. We used a PCA model with good performance to identify the distribution characteristics of metabolites between the groups ( Figure 4A). VIP analysis was performed to visually display the importance of differential metabolites in the groups and the trend of expression ( Figure 4B). The Kruskal test ( Figure 4C) showed that 26 metabolites were signi cantly increased in monks, such as L-dopa and berberine, while only eight metabolites were signi cantly decreased, such as indoleacrylic acid, valyl−valine, and cadaverine.
The correlation analysis between the differential metabolites and general characteristics of monks ( Figure 4D) found that valyl-hydroxyproline, putreanine, lotaustralin, and N-docosanoyl taurine levels inversely correlated with practice duration, while N-arachidonoyl tyrosine levels positively correlated with practice duration.
KEGG functional annotation analysis indicated that a majority of the KEGG pathways at level 2 (6/12, 50%) belong to the metabolism pathway ( Figure 5A and Table S2). Surprisingly, L-Dopa was involved in seven of the twelve KEGG metabolic pathways. KEGG pathway enrichment analysis found that differential metabolites were enriched in thirteen metabolic pathways (p < 0.05, Figure 5B and Table S3), including biosynthesis of plant secondary metabolites, lysine degradation, tryptophan metabolism, cocaine addiction, amphetamine addiction, alcoholism, dopaminergic synapse, Parkinson disease, AMPK signaling pathway, and betalain biosynthesis. Interestingly, L-dopa was still involved in eleven of the thirteen enriched metabolic pathways. These results provide a valuable molecular basis for revealing the changes in intestinal function affected by monk practice.
The monks' practice time signi cantly affected the composition of the microbiota and gut metabolome Based on the above analysis, we found that the practice lasting time of vegetarianism and meditation was of great signi cance to the alterations of intestinal ora and metabolites in monks. To further explore the impact of the practice duration on intestinal changes, the monks were divided into H groups and L group using the median 7.5 of the practice duration (Table 3). PCoA analysis showed that the microbial composition of the H group was signi cantly different from that of the L group ( Figure 6A). The Wilcoxon rank-sum test (p < 0.05) and LEfSe analysis (LDA > 2.0, p < 0.05) found that norank_f__Lachnospiraceae and norank_c__Cyanobacteria were overrepresented in the H group, while Prevotella_9, Bi dobacterium, unclassi ed_f__Peptostreptococcaceae, Gordonibacter and Enterococcus were overrepresented in the L group ( Figure 6B-C).
To reveal the effects of practice duration on the functional and metabolic-related changes of gut microbiota, PICRUSt2 (321 KEGG categories) and Tax4Fun (268 KEGG categories) identi ed a total of 227 common KEGG pathways. Based on the LEfSe analysis (LAD>2.0, P <0.05, Figure 6D-E and Table 4-5), Tax4Fun or PICRUSt2 screened 12 KEGG pathway level 3 pathways with signi cant differences between the two groups. Notably, valine leucinean disoleucine degradation (ko00280) and peptidoglycan biosynthesis (ko00550) were identical in Tax4Fun and PICRUSt2. The former belonged to amino acid metabolism and was highly expressed in the H group, and the latter belonged to glycan biosynthesis and metabolism and was highly expressed in the L group. Additionally, dopaminergic synapses (ko04728) of the nervous system were enriched in the L group.
Spearman correlation was utilized to analyze the potential connections between the seven differential genera and twelve KEGG pathway level 3 of Tax4Fun or PICRUSt2 ( Figure 6F-G). In Tax4Fun, Prevotella_9 was positively correlated with the highly expressed KEGG pathways in the L group but negatively correlated with the highly expressed KEGG pathways in the H group. Interestingly, Prevotella_9 was negatively correlated with all KEGG pathways in PICRUSt2.
A total of 23 differential metabolites (VIP > 2, p < 0.05) were successfully identi ed in the H and L groups, which were classi ed into organoheterocyclic compounds, organic oxygen compounds, organic acids and derivatives, lipids and lipid-like molecules, and benzenoids according to the HMDB database, as shown in Table S4. The PCA model indicated that the two groups were different from each other ( Figure 7A). The Kruskal-Wallis test ( Figure 7B) showed that fteen metabolites were upregulated in the H group, such as DG (16:0/18:0/0:0), 5-Hydroxy avone, 2-Linoleoyl Glycerol, and Glycocholic Acid, while Incensole oxide, Valyl-Hydroxyproline, Ginkgoic acid, and Acetyl-L-tyrosine, were signi cantly reduced.
For KEGG functional pathways, differential metabolites were involved in the ve primary KEGG pathways, containing 19 secondary pathways ( Figure 7C). Notably, DG (16:0/18:0/0:0) participated in eight organic systems pathways, six human diseases pathways, one environmental information processing pathway and one cellular process, as shown in Table S5.
Based on KEGG enrichment pathway analysis, differential metabolites were enriched in 67 metabolic pathways (p < 0.05), and the top 20 enrichment pathways in ER values are shown in Figure 7D. Many enrichment pathways were related to immunity, such as Th1 and Th2 cell differentiation, Th17 cell differentiation, B cell receptor signaling pathway, natural killer cell-mediated cytotoxicity, and T cell receptor signaling pathway. Surprisingly, DG (16:0/18:0/0:0) was also involved in these immune-related pathways.

Discussion
We aimed to study the regulation of Chinese monks' gut microbiota and metabolome alterations from the practices of meditation and vegetarianism, displayed an outline of the complex alterations interrelated with Chinese monks in the gut microbiota as well as the intestinal metabolites in Chinese monks, and explored the correlation between the gut changes and the practice duration of monks. We obtained many interesting results, which display the characterizations of microbial community composition and metabolites in human feces and provide a brand and new idea to understand the molecular phenotypic differences between Chinese monks with meditation and vegetarian lifestyles and controls with omnivorous and nonmeditation lifestyles.
We found that Firmicutes were lower and Bacteroides were higher in monks, and the Firmicutes/Bacteroidetes ratio was decreased in monks. Previous studies have reported that Firmicutes increased in obese people, while Bacteroides increased in thin people [21,22]. Some studies have suggested an increased F/B ratio as one marker of obesity [23]. Moreover, the F/B ratio was lower in young people [24]. Monks are long-term vegetarians and rarely eat fatty food. Additionally, monks meditate many times per day, which activates thinking and relaxes the body; thus, monks have less pressure than peers without practice lifestyles, making bodies and minds younger. We believe that the above factors cause a decline in the F/B ratio in monks. However, monks never eat meat in their diets, resulting in the lack of su cient protein and some essential trace elements, which lead to a decline in intestinal microbial diversity.
Previous research found that Roseburia, Eubacterium_eligens_group, norank_f__Lachnospiraceae, and Lachnospiraceae_NK4A136_group were dominant bacterial genera in the vegan meditation group [6], which was consistent with our research showing that these genera were signi cantly increased in monks. Roseburia is a common butyric acidproducing bacterium and one key bacterium degrading dietary ber xylan in the human intestine [25,26]. Studies have shown that colonization of butyric acid-producing intestinal Roseburia bacteria in mice can reduce in ammatory markers and improve atherosclerosis under the in uence of a diet rich in plant polysaccharides [25], indicating that Roseburia has a bene cial effect on the cardiovascular system. Faecobacteria is one of the most important butyric-producing bacteria in the human colon and is considered a bioindicator of human health [27]. In ammatory process development promoted by Faecobacteria can be reduced. Roseburia and Faecalibacterium signi cantly increased in monks, showing that long-term vegan meditation promotes the growth of butyric acid-producing bacteria and plays a positive role in health. Lachnospira promotes the e cacy of antihistamines in chronic spontaneous urticaria (CSU), which is a signature to predict the e cacy of antihistamines in CSU patients [28]. Lachnospira increased in monks, suggesting that CSU patients may improve the e cacy of antihistamines by the practices of vegetarians and meditation. It has been reported that the Eubacterium_hallii_ group is the dominant bacteria in chronic pancreatitis [29], and pancreatitis is closely related to overeating and hyperlipidemia.
Eubacterium_hallii__group signi cantly decreased in monks, which is consistent with the healthy practices of long-term vegetarian monks. However, long-term vegetarian diets inevitably caused a decrease in some bene cial bacteria, such as Bi dobacteria and Blautia. Bi dobacterium is a bene cial bacteria physiologically with various functions, such as nutrition [30], enhancing immunity [31], improving gastrointestinal function [32], preventing cancer [33], and antiaging. Blautia can prevent intestinal in ammation, promote the production of short-chain fatty acids and maintain intestinal steady-state activity, which has potential probiotic properties [35]. Thus, the reduction of Bi dobacteria and Blautia led to a negative impact on intestinal balance in monks.
Citrate cycle, carbon xation pathways in prokaryotes, other glycande gradation, and lipopolysaccharide biosynthesis were enriched in monks, showing that the energy metabolism of monks was higher than that of controls. This suggests that vegetarianism and meditation affect changes in the intestinal ora and metabolites, thus affecting the absorption of intestinal nutrients and nally causing changes in monk body metabolism. Of note, ABC transporters and PTS were inhibited in monks, which is bene cial to health. Some studies have proven that PTS plays a role in antimicrobial resistance. For example, glycocins are antimicrobial glycopeptides that are regarded as novel targets, and PTS affects the antibacterial activity of glycocins [36]. Another study showed that cajanin stilbene acid killed Enterococcus by inhibiting the PTS of Enterococcus [37], thus hindering the normal growth and metabolic functions of bacteria. As a drug transporter in a variety of tumors, ABC transporters are considered to be the major participant in tumor chemoresistance and play a key role in the development of tumors [38][39][40]. Therefore, monks may be less likely to develop drug resistance than controls.
We identi ed twelve lipids and lipid-like molecules of differential metabolites, which were increased in monks. Previous research has found that lipids and lipid-like molecules were signi cantly increased in type 2 diabetes patients [41]. We believe that monks have lipid dysregulation due to a long-term carbohydrate-based vegetarian diet, which may require early intervention and treatment. It should be noted that L-dopa, a neurotransmitter involved in happiness and the most commonly used drug to treat Parkinson's disease [42][43][44], was signi cantly increased in monks. L-Dopa plays an important role in multiple KEGG pathways, which may be related to the long-term meditation of monks' practice. Meditation can reduce stress and anxiety and enhance memory and concentration. Thus, we suggest that high-pressure people be happy by increasing L-dopa via the lifestyle of monks' practice. Additionally, Parkinson's disease patients might improve their condition via the monks' practice lifestyle. Kynurenic acid, as the major metabolite of tryptophan oxidative metabolism, regulates neuronal excitability and plasticity, brain development, and behavior and plays a protective role in Huntington's disease [45,46]. Therefore, we believe that the long-term meditation of monks promotes the brain to think, protects the function of the brain, and prevents neurodegenerative diseases. Many studies have reported that solasodine [47], peimine [48], berberine [49], ginkgoic acid [50], and 6-pentadecyl salicylic acid [51] have antimetastatic effects and inhibit tumor cell proliferation in various cancers. Therefore, the practices of meditation and vegetarianism may help monks prevent cancer.
A signi cant correlation between dominant microbiotas and enriched metabolic pathways was observed in monks, indicating that gut metabolic phenotype alterations were associated with microbiota changes. Lysine degradation is an important link in human central metabolism and is associated with human cancer [52,53]. The AMPK signaling pathway is related to cell metabolism and tumor autophagy [54,55]. These results coincided with the functions of the identi ed microbiota and metabolites in monks; that is, meditation and vegetarianism of monks have an impact on metabolism and cancers. The pathways of cocaine addiction [56], amphetamine addiction, alcoholism, dopaminergic synapses, and Parkinson's disease were related to mental states or mental illnesses, indicating that meditation of monks plays a part in the "gut-brain" axis, but further experimental veri cation is still needed to con rm this hypothesis.
The practice duration among monks has a signi cant impact on the composition of intestinal ora. Interestingly, we found that norank_f__ Lachnospiraceae was signi cantly increased not only in the monk group but also in the H group, indicating that norank_f__ Lachnospiraceae is an important genus that affects intestinal changes through monks' practice lifestyle. It has been reported that norank_f__ Lachnospiraceae is one of the dominant bacteria in autism spectrum disorder [57], so we speculated that norank_f__ Lachnospiraceae may be one of the signi cant bacteria on the "gut-brain" axis. In addition, norank_f__ Lachnospiraceae could alleviate body weight gain, visceral fat accumulation, dyslipidemia, and impaired glucose tolerance induced by a high-fat diet [58]. This conformed to the general point that the long-term practice of monks prevents the body from harming the fatty diet. Moreover, Prevotella_9 and Bi dobacterium were reduced in the long-term practice group, showing that Provotella_9 and Bi dobacterium lose their initial advantage in the monk gut with the extension of practice time. Many studies have reported that Prevotella is a common human intestinal bacterium that is related to vegetarian diets and is a dominant ora in carbohydrate diets [59,60].
The majority of differential metabolites of the H and L groups were lipids and lipid-like molecules, indicating that the longer monks practiced, the higher the risk of lipid metabolism disorder. Notably, DG (16:0/18:0/0:0) was involved in the metabolic pathway of long-term depression enrichment, which was highly expressed in the H group. Therefore, we further speculated that the appropriate practice of monks could improve the sense of happiness, reduce life pressure, and improve the mental state, but long-term practice may lead to depression. Additionally, DG (16:0/18:0/0:0) is involved in many immune-related pathways, such as natural killer cell-mediated cytotoxicity, Th17 cell differentiation, the B cell receptor signaling pathway, the T cell receptor signaling pathway, and Th1 and Th2 cell differentiation. Thus, we inferred that the long-term practice of monks might stimulate some immune responses in the intestines to protect or harm the body, which is related to long-term vegetarianism or meditation. DG (16:0/18:0/0:0) participated in the circular system, digestive system, endocrine system, environmental adaptation, excess system, immune system, nervous system, and sensory system in the KEGG functional pathway, suggesting that the organic systems of the whole body would be affected with prolonged practice. The above results re ected that DG (16:0/18:0/0:0) plays an important role in the intestinal metabolism of monks, but the speci c molecular mechanism is unknown, which is the focus of further research.
In summary, our data highlighted that gut microbiota and metabolite alterations are associated with the practices of monks, suggesting that vegetarianism and meditation have an important impact on intestinal stability. We hold the view that proper practice of monks is conducive to physical and mental health, but long-term practices might lead to physical and mental adverse events.

Study design
To investigate the gut microbiota and metabolites in monks, we recruited a total of 24 monks with the practices of long-term vegetarianism and meditation from Zhengfa Temple of Wuxi City, Jiangsu Province. Other key information, including age, body weight (BW), body mass index (BMI), and the duration of monk use, was obtained when collecting monk fecal samples.
To avoid the in uence of gender on the results, we selected 48 healthy male omnivores who did not mediate as the control group. This protocol in this study was approved by the ethics committee of Shanghai Tenth People's Hospital (ethics No.: shsy -IEC -ki -4.0/16-18/01). All subjects signed informed consent forms and were informed of the nature of the study prior to the collection of stool samples.
DNA extraction, PCR ampli cation, and MiSeq sequencing Fecal samples of participants were collected using sterile tubes and stored in liquid nitrogen at -80 ℃ until they were used for the study of extracting microbial DNA and metabolites. 16S rRNA gene sequencing was used to analyze the fecal microbiota. The E.Z.N.A. ® Soil DNA Kit (Omega Bio-Tek, Norcross, GA) was used to extract bacterial DNA from fecal samples. A Nanodrop2000 was used to determine the concentration of bacterial DNA. The extracted DNA was stored at -20°C to prevent degradation. The V3-V4 variable region of the 16S rRNA gene was ampli ed by PCR with speci c primers 338F (5'-ACTCCTACGGGAGGCAGCAG-3') and 806R (GGACTACHVGGGTWTCTAAT-3'). Each sample was repeated three times to ensure the reliability of the subsequent microbiological analysis. The ampli ed products were puri ed by an AxyPrep DNA Extraction Kit (AxyPrep Biosciences, USA) and quanti ed by quantitative TM st (Promega, USA). Then, the puri ed amplicons were combined and sequenced on the Illumina MiSeq PE300 platform (Illumina San Diego, USA).

Fecal metabolite extraction
The 400-µL methanol:water (4:1, V/V) solution was thoroughly mixed with a 50-mg stool sample and then crushed at 50 Hz for 6 min using a WONBio-96C high-throughput tissue crusher (Shanghai Wanbo Biotechnology Co., Ltd.). The mixture was then rotated for 30 seconds and treated with ultrasound for 30 minutes at 40 kHz and 5 °C. The mixture was then incubated at -20 °C for 30 minutes to precipitate the protein. After centrifugation at 4 °C at 13,000 g for 15 min, the supernatant was transferred to the sample ask, and positive and negative ions were analyzed by UPLC QTOF MS. Additionally, 10 μL supernatant was mixed from each sample as a quality control (QC) sample and injected periodically (every 8 samples) to monitor the stability of the analysis.

Microbial analysis of 16S rRNA gene sequencing
We used the Majorbio Cloud Platform (https://cloud.majorbio.com/) to analyze the 16S rRNA gene sequencing data. According to the overlap relationship, the paired end fragments obtained from Illumina MiSeq PE300 high-throughput sequencing were spliced into the target sequence for quality control screening. We eliminated the chimera and optimized the sequence by UCHIME software. The Ribosome Database Project (RDP) classi er (version 2.2, http://sourceforge.net/projects/rdp-classi er/) was in the Silva database (version 128, https://www.arbsilva.de/documentation/release-Use 128/) to classify and label the optimal sequence with a con dence threshold of 70%. UPARSE software (version 7.1, http://drive5. COM/UPARSE/) was used to cluster operational taxa (OTUs) with a similarity of 97%. To avoid pseudosequences, OTUs accounting for less than 0.005% of the total number of sequences were deleted.

Data preprocessing of LC-MS/MS
LC-MS raw data were imported into Progenesis QI 2.3 (Nonlinear Dynamics, Waters, USA) for peak detection and comparison. Then, a data matrix consisting of retention time (RT), mass charge ratio (m/z) value and peak intensity was generated. Qualitative and quantitative results for speci c metabolites were obtained. It is important to note that metabolic datasets must be preprocessed prior to multivariate statistical analysis. The pretreatment methods were as follows: 1) data ltering included removing samples with missing values >50%; 2) missing values were simulated according to half of the minimum metabolite values, and the total peak was normalized; and 3) elimination of QC samples with a relative standard deviation (RSD) >30%. Finally, after logarithmic conversion, a preprocessed dataset is prepared for further analysis.

Bioinformatic analysis and multivariate statistical analysis
In the Majorbio Cloud Platform, beta diversity analyses were calculated using the score plot of principal coordinate analysis (PCoA) and analysis of similarities (ANOSIM) to estimate whether intergroup differences were signi cantly greater than intragroup differences. The Wilcoxon rank-sum test was used to compare the differences in gut microbiota between the groups. Linear discriminant analysis effect size (LEfSe) analysis with LDA score was used to determine the differential taxa or functional pathways between the groups [61]. The typing analysis by the Partitioning Around Medoids algorithm was used to cluster all samples into seven types. Random forest model analysis with default settings was performed in Majorbio.
Based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, Tax4Fun and PICRUST2 were also used to perform functional pathway analysis in Majorbio. Venn analysis was performed using the OECloud tools at https://cloud.oebiotech.cn. Spearman correlation analysis with species clusters was performed using the OmicStudio tools (https://www.omicstudio.cn/tool), and the results are presented as a heatmap. Additionally, the receiver operating characteristic (ROC) curve with the area under the curve (AUC) was constructed in OmicStudio.
SIMCA software (V16.0.2, Sartorius Stedim Data Analytics AB, Umea, Sweden) was used to analyze the nontargeted LC-MS metabolomics. Principal component analysis (PCA) was used to visualize the distribution characteristics of metabolites between the groups. We used the PCA model to generate the corresponding VIP values to re ect the contribution of the distribution between groups. The metabolites were considered to differ signi cantly between groups when VIP values > 2 and p < 0.05. Differential metabolites were mapped to their biochemical pathways based on the KEGG database using    Comparison between the two groups at the genus levels. *p < 0.05; **p < 0.01; ***p < 0.001. (E) The LEfSe was used to identify the species that signi cantly differed between the two groups at the genus levels ( LDA score > 3.0, p < 0.05 ). (F) Composition of the gut microbiota at the genus levels.

Figure 3
The relationship between gut microbiota and functional analysis prediction. (A) The ROC curve was used to assess the diagnostic accuracy of the top nine genera based on LEfSe results. (B) Spearson correlation analysis was used to determine the relationships between 33 differential genera and monks' general characteristics. Red represents a positive correlation, and green represents a negative correlation. *p < 0.05; **p < 0.01; ***p < 0.001. The LEfSe analysis ( LAD > 2.0, p < 0.05 ) showed that (C) 35 categories of Tax4Fun and (D) 20 categories of PICRUSt2 were signi cantly differed between the groups. (E) The Venn plot showed the predicted KEGG function counts for the two groups based on Tax4Fun and PICRUSt2. (F) Spearson correlation analysis was used to determine the relationships between 33 differential genera and the eight identi ed common KEGG categories.

Figure 4
Altered composition of intestinal metabolites in monks compared with controls. (A) PCA model was used to identify the distribution characteristics of metabolites between the groups. (B) VIP analysis ( VIP >2, p < 0.05 ) was performed to visually display the importance of differential metabolites in the two groups and the trend of expression. (C) The Kruskal test was used to display the relative abundances of all differential metabolites in monks and controls. (D) Spearson correlation analysis was used to determine the relationships between differential gut metabolites and monks' general characteristics. Figure 5 (A) KEGG functional annotation analysis and (B) KEGG pathways enrichment analysis displayed that 23 differential metabolites were involved in the twelve KEGG annotation pathways at level2 and the thirteen KEGG metabolic pathways ( p < 0.05 ). Spearman correlation was used to determine (C) the relationships between 33 differential genera and 34 differential metabolites, and (D) the relationships between 33 differential genera twelve lipids and lipid-like metabolites.

Figure 6
Altered composition of gut microbiota and the Functional analysis prediction in the two groups (A) PCoA analysis revealed that the microbial composition was clearly different between H group and L group. (B)The LEfSe was used to identify the species that signi cantly differed between the two groups at the genus levels ( LDA score > 2.5, p < 0.05 ). (C) Comparison between H group and L group at the genus levels. **p < 0.01; ***p < 0.001. The LEfSe analysis ( LAD > 2.0, p < 0.05 ) showed that (D) twelve categories of Tax4Fun and (E) twelve categories of PICRUSt2 were signi cantly differed between the two groups. The Spearman correlation was utilized to analyze the potential connections between the seven differential genera and (F) twelve categories of Tax4Fun and (G) twelve categories of PICRUSt2.

Figure 7
Analysis of metabolites, KEGG functional annotation analysis, and KEGG pathways enrichment analysis in H group and L group. (A) PCA model indicated that the H group and the L group were different from each other. (B) The Kruskal test displayed the relative abundances of all differential metabolites in the two groups. (C) KEGG functional annotation analysis and (D) KEGG pathways enrichment analysis showed that twelve differential metabolites were involved in ve primary KEGG annotation pathways containing 19 secondary pathways and the top 20 KEGG enrichment pathways in ER values ( p < 0.05 ).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.