Shifts in the bacterial community of supragingival plaque related to metabolic associated fatty liver disease


 Background

Metabolic-associated fatty liver disease (MAFLD), also known as the hepatic manifestation of metabolic disorders, has become one of the most common chronic liver diseases worldwide. The associations between some oral resident microbes and MAFLD have been described. However, changes to the oral microbial community in patients with MAFLD remain unknown.
Methods

In this study, variations to the supragingival microbiota of MAFLD patients were identified. The microbial genetic profile of supragingival plaque samples from 24 MAFLD patients and 22 healthy participants were analyzed by 16S rDNA sequencing and bioinformatics analysis. Clinical variables, including indicators of insulin resistance, obesity, blood lipids, and hepatocellular damage, were evaluated with laboratory tests and physical examinations.
Results

The results showed that the diversity of the supragingival microbiota in MAFLD patients was significantly higher than that in healthy individuals. Weighted UniFrac principal coordinates analysis and partial least squares discriminant analysis showed that the samples from the MAFLD and control groups formed separate clusters (Adonis, P = 0.0120). There were 27 taxa with differential distributions (linear discriminant analysis, LDA༞2.0) between two groups, among which Actinomyces and Prevotella 2 spp. were over-represented in the MAFLD group with highest LDA score, while Neisseria and Bergeyella spp. were more abundant in the control group. Co-occurrence networks of the top 50 abundant genera in the two groups suggested that the inter-genera relationships were also altered in the supragingival plaque of MAFLD patients. In addition, as risk factors for the development of MAFLD, insulin resistance was positively correlated with the abundances of Granulicatella, Veillonella, Streptococcus, and Scardovia spp., while obesity was positively correlated to the abundances of Streptococcus, Oslenella, Scardovia, and Selenomonas spp. Metagenomic predictions based on Phylogenetic Investigation of Communities by Reconstruction of Unobserved States revealed that pathways related to sugar (mainly free sugar) metabolism were enriched in the supragingival plaque of the MAFLD group.
Conclusions

In conclusion, as compared to healthy individuals, component and interactional dysbioses were observed in the supragingival microbiota of the MAFLD group, which could be further explored as potential oral biomarkers of the MAFLD.

source of systemic in ammation, dysbiosis in the oral microbiome had been closely associated with metabolic diseases, including MAFLD [6]. A previous study found that the frequency of Porphyromonas gingivalis in the oral cavity is signi cantly higher in MAFLD patients as compared with health subjects [7]. In another study, intravenous injection of sonicated P. gingivalis caused impaired glucose tolerance, IR, and liver steatosis in C57BL/6J mice fed a high-fat diet [8]. However, because relatively few oral resident microbes have been the focus of previous studies, changes to the oral microbial community in patients with MAFLD remain unknown.
As a common site of oral sampling, the microbial pro le of supragingival plaque is thought to re ect the health status of the host, such as gestation [9], type 2 diabetes [10], and other conditions [11]. As compared with subgingival plaque, the supragingival microbiota can be acquired more easily and with less discomfort to the host [9]. Therefore, sampling of the supragingival microbiota presents a promising method to assess oral and systematic health, especially for a large-scale health census. To the best of our knowledge, the present study is the rst to assess the association between supragingival microbiota and MAFLD.
In the present study, supragingival plaques were obtained from 24 patients with MAFLD and 22 healthy controls. Illumina MiSeq PE300 sequencing and bioinformatics analysis were employed to identify changes to the microbial pro les and inter-taxa relationships of the supragingival plaque of MAFLD patients. Speci c microbial genera correlated with MAFLD and related clinical indexes were also identi ed. Furthermore, potential functional alterations to the supragingival microbiome were predicted based on the sequencing results. These ndings would provide a deeper understanding of the oral ecological dysbiosis associated with MAFLD.

Study Population
The study protocol was approved by the Ethics Committee of Shanghai Ninth People's Hospital a liated with Shanghai Jiao Tong University, School of Medicine (Shanghai, China) (approval no. SH9H-2019-T295-1) and conducted in accordance with the tenets of the Declaration of Helsinki. Written informed consent was obtained from all participants prior to enrollment.
The study participants were recruited from a health census and assigned to one of two groups: the MAFLD group, consisting of 24 persons diagnosed with MAFLD via upper abdomen ultrasonography and other clinical examinations [1], or a control group, consisting of 22 persons with normal ndings by upper abdomen ultrasonography. Age and sex in the participants of the two groups were matched. For each participant, upper abdomen ultrasonography was performed successively by two experienced sonographers and those with the same diagnosis were enrolled.

Acquisition of Clinical Variables
All participant demographics were retrieved from self-reported questionnaires and fasting blood samples were collected to detect clinical levels of total cholesterol (TC), total triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma glutamyl transpeptidase (GGT), fasting blood glucose (FBG), and fasting serum insulin (FSI). As an approximation of IR, the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) equation was calculated as HOMA-IR = FBG (mmol/L) × FSI (mU/L) / 22.5 [12]. The waist circumference, weight, height, and body mass index (BMI) of each participant were acquired via physical examination. The unpaired Student's t-test was applied for analysis of all clinical variables with an exception of "sex," which was analyzed using the chi-squared test.

Supragingival Plaque Collection
Supragingival plaque was collected before eating in the morning in accordance with the methods described in the Manual of Procedures for the Human Microbiome Project (https://www.hmpdacc.org/hmp/doc/HMP_MOP_Version12_0_072910.pdf) with minor modi cations.
The index teeth (#3, #9, #12, #19, #25, and #28) were isolated with cotton rolls and dried under a gentle stream of air. A sterile sickle scaler was used to collect the supragingival plaque from the buccal surfaces of the index teeth. Then, the scaler tips were immersed in 300 µL of sterile normal saline contained in a sterile Eppendorf tube for 5-10 s with slight shaking and the surface of the scaler was wiped off on the inside edge of the tube. When the supragingival plaques of all index teeth were obtained, the Eppendorf tubes were sealed, marked, and kept frozen in liquid nitrogen until DNA extraction. DNA Extraction, Ampli cation, and High-throughput Sequencing Total bacterial genomic DNA was extracted from the collected supragingival samples using the QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA) in accordance with manufacturer's protocols. The concentration and puri cation of the extracted DNA were determined using a NanoDrop 2000 UV-vis spectrophotometer (Thermo Scienti c, Wilmington, DE, USA), while DNA quality was checked by 1% agarose gel electrophoresis.
The 16S rDNA hypervariable V3-V4 region was PCR-ampli ed with the forward primer 338F (5'-ACTCCTACGGGAGGCAGCAG-3') and the reverse primer 806R (5'-GGACTACHVGGGTWTCTAAT-3') using a GeneAmp™ PCR System 9700 (Applied Biosystems, Carlsbad, CA, USA). The parameters of the PCR reactions have been described in a previous study [13]. Each PCR reaction was performed in triplicate and all resulting PCR products were extracted from 2% agarose gels and then further puri ed using the

Data Processing and Bioinformatics Analysis
Raw fastq les were quality-ltered using the Trimmomatic read trimming tool (https://kbase.us/) and merged using FLASH software (version 1.2.11; https://ccb.jhu.edu/software/FLASH/index.shtml) in accordance with the criteria described in a previous report [13]. After trimming, operational taxonomic units (OTUs) were clustered at a similarity cutoff value of 97% using the UPARSE algorithm (version 7.1; http://drive5.com/uparse/). Alpha diversity indexes were calculated using MOTHUR software for describing and comparing microbial communities (version 1.30.2; https://www.mothur.org/wiki/ Download_mothur) and rarefaction curves were constructed at an inter-sequence similarity value of 97% using the QIIME bioinformatics pipeline (version 1.9.1; http://qiime.org/install/index.html). Bar plots were generated to visualize the species composition of all samples at the phylum and genus levels, and heat maps at the genus level were constructed using the R platform (version 3.6.1). Weighted UniFrac principal coordinates analysis (PCoA), nonparametric multivariate analysis of variance (Adonis), and partial least squares discriminant analysis (PLS-DA) were performed to identify differences in species composition between the MAFLD and control groups using QIIME. The linear discriminant analysis (LDA) effect size (LEfSe; http://huttenhower.sph.harvard.edu/galaxy) was applied to identify the most discriminatory taxa between the groups at the phylum and genus levels. Taxa with logarithmic LDA scores of > 2.0 were regarded as discriminative species. Co-occurrence networks of the 50 most abundant genera of each group were demonstrated using MOTHUR. Spearman's correlation coe cients were calculated and those with a |ρ| value of > 0.5 and a probability (P) value of < 0.05 were visualized via the Cytoscape software platform (https://cytoscape.org/). Spearman's correlation coe cients among the clinical variables and the top 50 abundant genera of the supragingival microbiome were calculated, and the results were visualized as heat maps via the R platform. Furthermore, the bioinformatics software package Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2, version 1.1.0; http://picrust.github.io/picrust/) was used to predict the functional pathways of each group according to the Kyoto Encyclopedia of Genes and Genomes (KEGG).

Subject Characteristics and Sequencing Results
The demographic and clinical characteristics of the participants are summarized in Table 1. There were no signi cant differences in age, sex, blood pressure, and heart rate between the two groups. Subjects in the MAFLD group had relatively higher TG levels and lower HDL-C levels, and thus higher TG/HDL-C ratios than the control group [14,15]. Moreover, ALT and GGT levels were relatively higher in the MAFLD group than the control group, revealing hepatocellular damage in MAFLD patients [2]. Because subjects with diabetes were excluded, there was no signi cant difference in FPG levels between the two groups, although FSI and HOMA-IR were signi cantly increased in the MAFLD group, indicating the prevalence of IR [12]. Waist circumference was also signi cantly higher in the MAFLD group, illustrating visceral adiposity was ubiquitous in MAFLD patients [16].
A total of 2,318,404 high quality sequences were produced, with an average of 50,507 ± 11,212 sequences per sample. In total, 23 phyla, 37 classes, 92 orders, 163 families, 339 genera, 628 species, and 1,021 OTUs were identi ed after taxonomic assignment of the sequences. Data are presented as the mean ± standard deviation. * P < 0.05. All clinical and demographic data were analyzed using the unpaired t-test with an exception of "sex," which was analyzed with the χ 2 test. The Chao1 index and the Ace index were calculated to identify the community richness of the supragingival microbiota [17]. The Simpson and Shannon indexes, which are also alpha diversity estimators, were employed to evaluate the community diversity of the supragingival microbiota [17]. As shown in Table 2, the MAFLD group had a higher Shannon index and lower Simpson index, suggesting the diversity of supragingival microbiota was higher in the MAFLD group than the control group (Table 2). Good's coverage indexes, which were close to 1 (Table 2), and rarefaction curves based on OTU levels ( Figure S1) reached saturation plateaus, indicating that the sequencing depths were su cient to represent the majority of the microbiota in both groups [18]. Results are presented as the mean ± standard deviation. * P < 0.05. All α-diversity estimators were analyzed using the Wilcoxon rank-sum test.
The taxa abundance of the supragingival microbiome from the two groups at the phylum and genus levels are depicted in Fig. 1A and 1B, respectively. In general, the dominant taxa of the two communities were similar and consistent with the core species of the supragingival microbiome [9]. The core phyla of all samples from both groups consisted of Bacteroidetes, Proteobacteria, Actinobacteria, Firmicutes, Fusobacteria, Patescibacteria, Epsilonbacteraeota, and Spirochaetes. Of the top 10 abundant genera, the MAFLD group had higher proportions of Capnocytophaga, Leptotrichia, Corynebacterium, Actinomyces, Streptococcus, Fusobacterium, Prevotella, and Veillonella; while Neisseria and Comamonas were more prevalent in the control group. The abundances of the top 50 abundant genera in each sample are displayed in the heat map presented in Figure S2. Weighted UniFrac PCoA at the OTU level was employed to evaluate the similarity of the bacterial communities between the two groups. The results indicated that although samples from the two groups partly overlapped, there was a tendency of separation along the PC1 axis (Fig. 1C). The results of Adonis (P = 0.0120) based on weighted UniFrac distances further veri ed the existence of signi cant differences in the overall structures of the supragingival microbiota of the two groups. PLS-DA, a supervised analysis suitable for high dimensional data [19], showed separate clustering of the samples from the MAFLD and control groups (Fig. 1D), further demonstrating remarkable differences in the supragingival microbiota between the two groups.

Alterations of the Supragingival Microbial Phylotypes/intergenera relationship Associated with MAFLD
A circular cladogram based on the LEfSe results demonstrated differentially abundant taxa between the two groups ( Fig. 2A). Taxa with logarithmic LDA scores of > 2.0 (P < 0.05) are plotted in Fig. 2B. Brie y, there were 27 taxa with differential distributions between the two groups. At the genus level, Actinomyces, Prevotella 2, Scardovia, Megasphaera, and Alysiella were more abundant in the NAFLD group, while Neisseria, Bergeyella, Sphingomonas, and H1 were over-represented in the control group. Co-occurrence networks of the top 50 abundant genera of the two groups were depicted via Cytoscape (Fig. 3A, B). In general, the taxa within the main bacterial cluster (> 5 nodes and connected with intense lines) had stronger interrelationships in the MAFLD group.

Correlations between Clinical Variables and Supragingival Microbiota
The heat map presented in Fig. 4 depicts the correlations between the 50 most abundant bacterial genera and single clinical variables based on the Spearman's correlation coe cients. Of the top 50 abundant genera, HOMA-IR showed positive correlations with Streptococcus, Scardovia, Granulicatella, and Veillonella, but a negative correlation with Neisseria. Alloprevotella and Peptostreptococcus were extremely signi cantly negatively correlated with TC and LDL-C levels. Aggregatibacter was negatively correlated with TG levels, but positively correlated with HDL-C levels. Streptococcus, Oslenella, Scardovia, and Selenomonas were signi cantly positively correlated with BMI and waist circumference. AST/ALT and GGT, two indicators of hepatocellular damage, were negatively correlated with Capnocytophaga.

Predictive Metagenome Functional Pro ling of the Supragingival Microbiomes of the MAFLD and Control groups
To detect functional differences in the supragingival microbiomes of the MAFLD and control groups, PICRUSt 2 was employed to predict the metagenome functional contents based on the 16S rRNA datasets [20]. Statistically signi cant differences (P < 0.05) in KEGG pathways were calculated with the Wilcoxon rank-sum test. As shown in Fig. 5A phosphorylation, and two-component system, were signi cantly decreased in the MAFLD group (Fig. 5B).

Discussion
The oral microbiota is attracting increased attention because of probable associations with systemic and metabolic disorders [21][22][23]. The associations between the oral microbiota and systemic/metabolic disorders can be explained in the following two aspects. On the one hand, microbial dysbiosis in the oral cavity is a source of systemic in ammation, which could lead to chronic low-grade in ammation and adversely affects the systemic health of the host [6]. On the other hand, the oral microbiota can in uence the composition of the gut microbiome, which plays important roles in systemic health [24][25][26][27]. Due to the anatomical position, about 10 11 bacteria are swallowed from the oral cavity to the stomach every day [25], cultivation and sequencing techniques have also substantiated the association between the oral and gut microbiomes: Arimatsu et al. reported that oral administration of P. gingivalis signi cantly altered the Firmicutes/Bacteroidetes ratio, a signi cant index to evaluate the health status of the gut microbiome [26]; Li et al. found that the oral microbiota could overcome physical barriers and colonize the gut in gnotobiotic mice [27]. These ndings acknowledged that the oral microbiota plays an important role in the development of systemic and metabolic diseases via "oral-gut axis". As for MAFLD, although some oral resident microbes have been associated with the development of it [7,8], there has been no microbiome-wide association study of the association between the development of MAFLD and oral microbial ecology. Shaped by the health status of the host, supragingival plaque has been related to various metabolic disorders. For example, Hintao et al. reported signi cant differences in the microbial pro les of supragingival plaque between subjects with and without diabetes [28]; La Monte et al. found that metabolic syndrome was signi cantly associated with supragingival plaque (odds ratio = 1.74; 95% con dence interval = 1.22-2.50) [29]. Considering supragingival plaque can be obtained with minimal discomfort and risk [9], it was collected in this study to explore the ecological shifts of oral microbiota in MAFLD patients. By screening with strict inclusion/exclusion criteria and matching of confounding factors, the differences among the participants were minimized as much as possible in order to focus on compositional and structural differences of the supragingival microbiota in MAFLD patients.
The diversity of the supragingival microbiota of each group was determined using alpha diversity estimators. It is generally acknowledged that microbial diversity re ects the health status of the host. For example, decreased diversity of gut microbiota indicates functional or metabolic disorders in the host [30], while increased diversity of oral microbiota is reported to imply poor oral [31,32] and holistic health [21][22][23][24][25][26][27][28][29][30][31][32][33] because in a state of poor oral health, gingival bleeding provides a richer nutrient source [25]. In the present study, increased diversity (lower Simpson index and higher Shannon index) of the supragingival microbiota in the MAFLD group was observed, suggesting possible alterations to the nutritional status of supragingival plaque in MAFLD patients.
Consistent with previous studies [9,34], the core phyla identi ed in the present study included Proteobacteria, Bacteroidetes, Firmicutes, Actinobacteria, and Fusobacteria, which accounted for 93.83% and 92.37% of the supragingival microbiomes of the control and MAFLD groups, respectively. Similarly, although the proportions differed, the majority of the observed genera (including Capnocytophaga, Leptotrichia, Corynebacterium, Actinomyces, Streptococcus, Fusobacterium, Prevotella, Veillonella, Neisseria, and Comamonas) existed in both groups, thereby also supporting the core genera of the supragingival microbiota [9]. A lower Firmicutes/Bacteroidetes ratio is considered as a healthy trait in both the oral cavity and gut [22]. In the present study, the Firmicutes/Bacteroidetes ratio was lower in the supragingival plaque of the control group as compared to the MAFLD group (61.41% vs. 72.38%, respectively), indicating dysbiosis of the supragingival microbiota of the MAFLD group.
The PCoA and PLS-DA results demonstrated differences in the community compositions between the two groups (Adonis, P = 0.0120). The discriminatory taxa between two groups were identi ed using LEfSe. At the genus level, Actinomyces and Prevotella 2 had the highest LDA scores in the MAFLD group. Actinomyces spp. are normal resident bacteria of the oral cavity, which exert important roles in bio lm formation [35]. Actinomyces spp. have been associated with the severity of chronic periodontitis [36]. Prevotella 2 is a genus of Gram-negative, anaerobic bacteria that exist in the gut and are relevant to multiple disease states, including an increased lifetime risk of cardiovascular disease [30], ankylosing spondylitis [37], and increased levels of C-reactive protein [38]. Considering the consistency between the oral and gut microbiotas [25][26][27], the prevalence of Prevotella 2 in the oral cavity is proposed as a potential marker of systematic diseases including MAFLD. In healthy participants, the genera Neisseria and Bergeyella had the highest LDA scores. Neisseria spp. are among the most abundant taxa in the oral cavity [39]. A predominance of Neisseria spp. in the oral cavity indicates healthy conditions of the oral cavity [40,41,42]. Bergeyella spp. are Gram-negative, aerobic bacteria [43]. In the present study, Bergeyella spp. were more prevalent in the control group, suggesting a negative correlation to MAFLD.
Co-occurrence networks were used to predict inter-genera correlations of supragingival plaque between the two groups. As shown in Fig. 3, there were signi cant differences in the interaction patterns of the two groups. In the MAFLD group, there were stronger and more complex interactions within the main cluster, but weaker and sparser correlations among the genera outside of the main cluster. Reportedly, an increase in interaction strength among taxa not only excludes other taxa, but decreases the stability of the microbial community [44]. Therefore, it could be speculated that the supragingival microbial community of the NAFLD group was more unstable.
Inhibition of hepatic glucose production, increased accumulation of lipids in the liver, and IR are vital to the development of MAFLD [45]. It is currently believed that IR is an independent risk factor for the severity of MAFLD [45]. As rst proposed by Matthews et al. in 1985, HOMA-IR is both practical and highly e cient for the evaluation of IR in both clinical and scienti c studies [12,46]. In the present study, HOMA-IR was beyond the normal range (normal range ≤ 1) in the MAFLD group and signi cantly higher than that in the control group (P = 0.0013) suggesting that IR is prevalent in patients with MAFLD. It was believed that chronic low-grade in ammation resulting from dysbiosis of the oral microbiota can reportedly aggravate IR [47]. In this study, Spearman's correlation analysis revealed that the presence of Granulicatella, Veillonella, Streptococcus, and Scardovia spp. was positively correlated with HOMA-IR. The congregation of Granulicatella spp. with Aggregatibacter actinomycetemcomitans [48] has been positively correlated to periodontitis [49], as well as serious infections outside of the oral cavity, such as infective endocarditis [50]. Veillonella is a genus of Gram-negative anaerobic bacteria mainly found in the oral and gastrointestinal tracts. The presence of Veillonella spp. in the oral cavity has been correlated to increased production of pro-in ammatory cytokines [47,51] and periodontal infections [41]. Streptococcus and Scardovia spp. are resident bacteria of the oral cavity that are closely related to caries formation [52]. Although relatively few studies have investigated the relationship between caries-related bacteria and IR, patients with IR tend to have more decayed teeth [53].
In a state of chronic low-grade in ammation [54], obesity is a contributor to various metabolic dysfunctions, such as MAFLD and type 2 diabetes [55]. As compared to BMI, visceral adiposity, as measured by waist circumference, has been closely linked to the severity of lipid deposition in the liver [54], which is consistent with the results of the present study, which found an increase in waist circumference in MAFLD patients (P = 0.0020). In addition, multiple studies have veri ed the in uence of obesity on the microbial pro le of the oral cavity [56,57]. In this study, genera positively correlated with obesity mainly included Streptococcus, Oslenella, Scardovia, and Selenomonas. Streptococcus and Scardovia spp. were also positively correlated to IR, supporting the positive association between obesity and IR [54]. The involvement of Oslenella spp. in endodontic infections [58] and periodontal in ammation [22] have been well documented. In the Veillonellaceae family, Selenomonas is a genus of Gram-negative anaerobic bacteria. Members of Veillonellaceae family are considered to act as pro-in ammatory mediators [59] and putative periodontal pathogens [60]. These results support the presumption that obesity is positively correlated to the abundance of bacteria associated with infectious diseases of the oral cavity [61].
Dyslipidemia is a common clinical manifestation of MAFLD, especially hypertriglyceridemia and low serum HDL-C [14,15,62], which were also veri ed in this study (P = 0.0083 for TG; P = 0.0011 for HDL-C). Reportedly, oral infectious diseases and dyslipidemia could have a two-way relationship without a clear cause-and-effect relationship [63]. Actinomyces spp. have been positively correlated to TG levels as a potential indicator of MAFLD-related metabolic dysfunction. A surprising result was that the presence of Aggregatibacter spp. was negatively correlated with TG levels, but positively correlated with HDL-C levels, which might indicate good health, challenging the mainstream concept that the presence of Aggregatibacter spp. (especially A. actinomycetemcomitans) is related to dyslipidemia and other metabolic diseases [63]. Sampling sites may explain this discrepancy because Aggregatibacter spp. are anaerobic bacteria with growth behaviors that may change in response to aerobic conditions (supragingival habitats). However, the exact reasons for this paradox remain unclear.
Known as indicators of hepatocellular damage, elevated serum levels of transaminases and transpeptidases are also main clinical manifestations of MAFLD [2]. Moreover, a decreased AST/ALT ratio is regarded as biomarker of progressive MAFLD [2]. In this study, a decreased AST/ALT ratio (P = 0.0142) as well as elevated GGT (P = 0.0084) were prevalent in MAFLD patients, suggesting the enrolled MAFLD patients had different degrees of hepatocellular damage. Capnocytophaga is a genus of Gramnegative anaerobic bacteria reportedly associated with periodontitis [36] and hyperglycemia [64]. In this study, an abundance of Capnocytophaga spp. was negatively correlated to the AST/ALT ratio, suggesting it could be a potential biomarker of MAFLD progression.
Metagenomic predictions based on PICRUSt2 revealed that functional changes between the control and MAFLD groups mainly involved metabolism (KEGG pathway level 2). Among the KEGG pathways level 3, metabolism of sugars (mainly free sugars, including starch and sucrose, fructose and mannose, and galactose) was more prevalent in subjects with MAFLD, revealing that supragingival plaque in MAFLD patients can easily obtain nutrients, which could explain the increased microbial diversity observed in the supragingival plaque of the MAFLD group (Table 2). Pathways related to aerobic respiration (including oxidative phosphorylation, pyruvate metabolism, and the citrate cycle) were more abundant in the supragingival plaque of the control group, suggesting that the proportion of aerobic bacteria in the supragingival plaque is higher in healthy people. Produced by oral anaerobic bacteria, volatile sulfur butanoate compounds (VSCs) have been positively correlated with halitosis [65] and as indicators of the severity of oral infectious diseases and other disorders of the digestive system [66]. The results of the present study showed that the metabolic pathways of VSC precursors (cysteine and methionine) were signi cantly over-represented in healthy individuals, which may alleviate halitosis and maintain good oral health. However, a de ciency of predicting functions based on taxa composition is that bacterial functions can change with the health status of the host [65]. Consequently, metatranscriptomics and metabolomics of the microbiota may provide more realistic functional pro les.
As this was a pilot study with matching of confounding factors, some intriguing ndings surfaced, but still need to be veri ed in future studies with larger sample sizes. In addition, with the increasing attention to the functions of the oral microbial community, it is essential to identify changes to the actual functional pro les of the supragingival microbiota in MAFLD by metatranscriptomics and metabolomics.

Conclusions
In the present study, dysbiosis of supragingival microbiota associated with MAFLD was characterized.
Brie y, the results revealed increases in the abundances of bacteria associated with oral infections, decreases in the abundances of potential bene cial aerobic bacteria, and changes in the interactions of the core micro ora with the supragingival microbiota in patients with MAFLD. Moreover, as risk factors for the development of MAFLD, IR was positively correlated to the abundances of Granulicatella, Veillonella, Streptococcus, and Scardovia spp., while obesity was positively correlated with the abundances of Streptococcus, Oslenella, Scardovia, and Selenomonas spp. The increased free sugar metabolic pathways suggested that supragingival bacteria related to the metabolism of free sugars were associated with MAFLD. These ndings provide a deeper understanding of the association between the oral microbiome and MAFLD, although further studies are needed to explore potential causal relationships.

Consent for publication
Not applicable.

Availability of data and materials
The raw reads of all the 16S rDNA sequencing results analyzed in the current study were deposited into the NCBI Sequence Read Archive (SRA) database (Accession Number: PRJNA645880).
Competing interests bioinformatics analysis and data acquisition. Ning-Jian Wang and Fang-Zhen Xia performed the data analysis and acquisition. Ying-Li Lu, Di-Liu, Zhi-Min Wang and Zheng-Wei Huang performed the study design, clinical sample collection, data analysis, drafting, revising, and nal approval and were also involved in the accountability of all aspects of the work. All authors had approved the nal version of the work.

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