Article Integrative multi-omics analysis reveals oral microbiome-metabolome signatures of obesity Graphical abstract Highlights • Oral microbiome composition and functions differ significantly in obesity • Obesity is linked to proinflammatory and lactate-producing oral bacteria • Obese individuals show disrupted oral metabolism and altered energy balance • Obesity-linked metabolites correlate with cardiometabolic disease markers Authors Ahmed A. Shibl, Tsedenia W. Denekew, Anique R. Ahmad, ..., Shady A. Amin, Youssef Idaghdour, Aashish R. Jha Correspondence jhaar@nyu.edu In brief Shibl et al. conduct a comprehensive multi-omics analysis of the oral microbiome in obesity, revealing distinct microbial and metabolic profiles. Obese individuals show enrichment of proinflammatory, lactate-producing bacteria and obesogenic metabolites, robustly linking oral microbiome disruption to cardiometabolic risk, highlighting the oral cavity as a potential obesity prevention target. Shibl et al., 2026, Cell Reports 45, 116819 February 24, 2026 © 2025 The Authors. Published by Elsevier Inc. https://doi.org/10.1016/j.celrep.2025.116819 ll mailto:jhaar@nyu.edu https://doi.org/10.1016/j.celrep.2025.116819 http://crossmark.crossref.org/dialog/?doi=10.1016/j.celrep.2025.116819&domain=pdf Article Integrative multi-omics analysis reveals oral microbiome-metabolome signatures of obesity Ahmed A. Shibl,1,2,9 Tsedenia W. Denekew,1,9 Anique R. Ahmad,1,9 Salah Abdelrazig,3 Christopher E. Leonor,2 Lina Utenova,1 Guihao Zhang,1 Mamoun AbdelBaqi,2 Yashaswi Malla,1,4 Muhammad Arshad,5 Marc Arnoux,6 Nizar Drou,5 Abdishakur Abdulle,2 The UAE Healthy Future Study Investigators Group, Raghib Ali,2 Shady A. Amin,3,5,7 Youssef Idaghdour,2,5,8 and Aashish R. Jha1,2,5,10,* 1Genetic Heritage Group, Program in Biology, New York University Abu Dhabi, Abu Dhabi, UAE 2Public Health Research Center, New York University Abu Dhabi, Abu Dhabi, UAE 3Marine Microbiomics Lab, Program in Biology, New York University Abu Dhabi, Abu Dhabi, UAE 4Computer Science Department, New York University Abu Dhabi, Abu Dhabi, UAE 5Center for Genomics and Systems Biology, New York University Abu Dhabi, Abu Dhabi, UAE 6Core Technology Platform, New York University Abu Dhabi, Abu Dhabi, UAE 7Mubadala ACCESS Center, New York University Abu Dhabi, Abu Dhabi, UAE 8Environmental Genomics Lab, Program in Biology, New York University Abu Dhabi, Abu Dhabi, UAE 9These authors contributed equally 10Lead contact *Correspondence: jhaar@nyu.edu https://doi.org/10.1016/j.celrep.2025.116819 SUMMARY Obesity is a leading global health challenge and risk factor for cardiometabolic disorders, driven in part by industrialization and low-fiber, ultra-processed diets. While the gut microbiome has been implicated in obesity, the contribution of the oral microbiome—the body’s second largest microbial ecosystem—remains underexplored. We analyze a prospective cohort of 628 Emirati adults, including multi-omics profiling of 97 obese individuals and 95 matched controls, generating the most comprehensive oral microbiome analysis to date. Obese participants show altered microbial diversity, composition, functions, and metabolites with enrichment of proinflammatory Streptococcus parasanguinis, Actinomyces oris, and lactate-producing Ori- bacterium sinus. Pathways for carbohydrate metabolism, histidine degradation, and obesogenic metabolites are upregulated, whereas B-vitamin and heme biosynthesis are depleted. Corresponding metabolites— including lactate, histidine derivatives, choline, uridine, and uracil—are elevated and correlate with obesity- linked cardiometabolic markers. These findings reveal mechanistic oral microbiome-metabolite shifts, high- lighting oral microbiome-host interactions as novel targets for obesity prevention and intervention. INTRODUCTION The World Health Organization reports that nearly half of the global adult population (43%) is overweight, with 890 million adults classified as obese.1 Obesity, characterized by excessive fat accumulation, is a major risk factor for metabolic disorders, including diabetes, hypertension, and cardiovascular disease.2 While genetics influence weight gain,3 the global rise in obesity largely reflects dietary and lifestyle changes accompanying industrialization, including increased consumption of ultra-pro- cessed, low-fiber foods.2,3 This trend is pronounced in the United Arab Emirates (UAE), where rapid urbanization has cata- lyzed significant dietary and lifestyle shifts,4 contributing to a marked rise in obesity and cardiovascular mortality.5 Disruption of the gut microbiome contributes to obesity devel- opment and progression6–9 by modulating carbohydrate meta- bolism, energy homeostasis, vitamin production, and body weight regulation.10,11 By contrast, the role of the oral micro- biome in systemic health and disease remains underexplored.12 The oral cavity harbors a diverse array of microorganisms such as bacteria, archaea, viruses, and unicellular eukaryotes, which produce a variety of biomolecules.13 This collective community of microorganisms and their products constitutes the oral micro- biome—the second largest microbial ecosystem in the human body after the gut.14 Compounds generated by oral microbes can interact locally with oral tissues or enter circulation, trig- gering various signaling mechanisms in distant organs.13 Life- style factors, such as diet, smoking, oral hygiene, and antibiotic use, can shape oral microbiome composition.15,16 Emerging evidence links oral dysbiosis to several metabolic diseases, including obesity.17 Obese individuals often show decreased overall microbial diversity, along with increased rela- tive abundances of Bacillota (previously known as Firmicutes), periodontal pathogens, and pro-inflammatory bacteria,17,18 which may promote systemic inflammation and metabolic dysfunction.19 Furthermore, obesity-related shifts in oral micro- bial community composition20 may induce alterations in the gut microbiome and, collectively, they can influence systemic Cell Reports 45, 116819, February 24, 2026 © 2025 The Authors. Published by Elsevier Inc. 1 This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). ll OPEN ACCESS http://creativecommons.org/licenses/by-nc/4.0/ mailto:jhaar@nyu.edu https://doi.org/10.1016/j.celrep.2025.116819 http://crossmark.crossref.org/dialog/?doi=10.1016/j.celrep.2025.116819&domain=pdf http://creativecommons.org/licenses/by-nc/4.0/ changes in metabolic disease.21 Also, oral microbiome sampling is practical for large-scale and longitudinal studies, yet prior research has been constrained by small sample sizes and 16S ribosomal RNA (16S rRNA) sequencing.22 Whole metagenomics shotgun sequencing integrated with metabolomics enables comprehensive profiling of species, functions, and metabo- lites.13,22 Integrating them provides a promising strategy for gaining a comprehensive understanding of how oral microbes contribute to diseases.23 Here, we investigate oral microbiome alterations in obesity in a large cohort. Using 16S rRNA sequencing in 628 individuals, we identify obesity as a key factor shaping oral microbial variation. Next, we conduct multi-omics profiling—shotgun metagenom- ics and untargeted liquid chromatography-mass spectrometry (LC-MS) metabolomics—on mouthwash samples from 192 indi- viduals (97 obese and 95 matched healthy-weight controls), inte- grating microbial, functional, and metabolic data with clinical biomarkers derived from blood and urine. This approach estab- lishes robust links between oral microbiome shifts and obesity, providing initial insights into how compositional, functional, and metabolic shifts in the oral microbiome may contribute to obesity. RESULTS Study design and cohort description The UAE Healthy Future Study (UAEHFS) is a large, ongoing population-based prospective cohort of 20,000 Emirati nationals aimed at identifying factors contributing to cardiometabolic dis- eases.24 In this study, we analyzed a cross-sectional, baseline cohort of 669 consenting Emirati nationals aged 18–43 years (mean ± SD = 24 ± 5.3 years), from whom mouthwash samples were collected. Participants completed standardized survey questionnaires capturing demographic (age, sex, marital status, etc.) and lifestyle (smoking, exercise, etc.) factors and underwent clinical assessments for anthropometric and physiological traits (BMI, blood pressure, heart rate, etc.). Based on self-reported questionnaires, 13% of participants reported a prior history of periodontal disease. However, ∼95% reported daily brushing and no current oral health issues (e.g., mouth ulcers, loose teeth, or painful gums), suggesting generally healthy oral conditions (Table S1). Additionally, we measured 58 clinical parameters from blood and urine (Figure 1A; Table S1). The average BMI and cholesterol in this cohort were 26.8 kg/m2 (SD = 6, range = 14.4–52.2 kg/m2) and 179 mg/dL (SD = 38.3 mg/dL, range = 93–445 mg/dL), respectively. Consistent with previous reports on the broader Emirati population,5,25 obesity (BMI ≥ 30, 30.4%) and hypercholesterolemia (blood cholesterol ≥ 200 mg/dL, 28.4%) were the most prevalent medical conditions in this cohort (Figure S1A). Effect of lifestyle and systemic health on the oral microbiome in the Emirati population The oral microbiome plays a crucial role in oral health.13,26 How- ever, its relationship with broader lifestyle and systemic health re- mains underexplored.22 Identifying lifestyle and health-related variables contributing to the oral microbiome is essential before conducting association studies, as they may act as confounders. To address this, we initially characterized the oral microbiome of 669 participants using 16S rRNA amplicon sequencing, of which 628 passed quality control. Consistent with previous studies,13 Bacillota (previously known as Firmicutes, 14.9%–87.3%), Bac- teroidota (previously known as Bacteroidetes, 2.7%–61.5%), and Pseudomonadota (previously known as Proteobacteria, 0.09%–63.6%) were the predominant phyla, collectively ac- counting for ∼93% of reads (Figure 1B). Oral microbial diversity, calculated using Shannon’s diversity index, was significantly associated with postprandial intervals (p = 4 × 10− 5, GLM, Table S2) and negatively correlated with the Bacillota- to-Bacteroidota ratio (BBR) (Figure 1C, ρ = − 0.71, p < 2.2 × 10− 16, Spearman’s rank correlation test). Notably, we did not find consistent associations between oral microbial diver- sity and other anthropometric, demographic, lifestyle, or health factors examined (Table S2). Principal coordinate analysis (PCoA) of the amplicon sequence variants (ASVs) using weighted UniFrac distances re- vealed that oral bacterial composition was significantly associ- ated with BMI categories (p = 0.018, EnvFit), postprandial inter- val, and smoking (Figures 1D, S1B, and S1C; Table S2). We assessed the relationship between oral microbiome composition and 27 non-collinear metadata variables—13 lifestyle and 14 health-related (see STAR Methods)—using two complementary multivariable approaches. PERMANOVA evaluated differences in weighted UniFrac distances to test the overall effect of each variable, while EnvFit assessed the contributions of these factors along the top three principal coordinate axes, helping to identify which factors most strongly contribute to community variation. We first applied a full model to 312 individuals with complete data for all 27 variables, identifying significant factors contrib- uting to the oral microbiome after adjusting for demography and lifestyle. To test robustness, we then applied a reduced model, retaining only the significant variables from the full model to a larger cohort of 618 individuals with available data (Table S2). Across all these analyses, BMI consistently emerged as a robust determinant of oral microbiome variation. Given its high incidence in the UAE and clinical relevance to cardiometa- bolic conditions,27 we further investigated obesity’s impact us- ing multi-omics approaches. Matched analysis highlights biomarkers of metabolic dysregulation in obesity To minimize lifestyle-related confounding, we identified 97 obese individuals (BMI ≥ 30) and 95 individuals within a healthy BMI range (controls, 18.5–24.9) matched for demographic, life- style, and oral health variables (Methods and Table S3). Principal component analysis (PCA) of clinical biomarkers, followed by multiple regression (EnvFit), revealed that only obesity was significantly associated with variation in PCA space (p = 2.7 × 10− 3, EnvFit, Figures 2A and 2B), confirming effective matching. A random forest classifier distinguished obese from control participants with moderate accuracy (out-of-bag [OOB] error = 34%, AUC = 0.84). Variable impor- tance scores identified elevated obesity-related serum markers in obese individuals—including triglycerides (TGs), gamma-glutamyl transferase (GGT), alanine aminotransferase (ALT), alkaline phosphatase (ALP), blood glucose (GLUH), and 2 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS HbA1c (Figures 2C and 2D). Elevated TG reflect insulin resis- tance and increased hepatic lipogenesis,28 while increased liver enzymes (GGT, ALT, and ALP) indicate hepatic stress,29–31 and higher glucose and HbA1c indicate impaired glycemic control and heightened type 2 diabetes risk.32 These findings under- score obesity’s central role in metabolic dysregulation and cardi- ometabolic risk. Obesity-associated alterations in oral microbiome composition To characterize oral bacteria at the species level and compare functional potentials between obese and control groups, we per- formed shotgun metagenomics sequencing on 192 samples, generating ∼6.1 billion reads. After removing one low-depth sample, 191 samples remained (obese n = 97, controls n = 94), with an average of 31.7 million reads per sample, of which ∼3.7 million mapped to oral bacteria (Table S4). Analysis of the microbial community structure, inferred by mapping individual reads to species-specific marker genes using MetaPhlAn4,33 fol- lowed by PCoA, revealed compositional differences between obese and control participants (Figure 3A; Table S5). Multivariate modeling with MaAsLin234 identified 26 of 366 bacterial species as differentially abundant (false discovery rate [FDR] adjusted p < 0.05, Figure 3B). Random forest analysis using these species distinguished obese from controls with appreciable accuracy (Figure 3C, OOB = 33%, AUC = 0.75). Bacteria enriched in obesity included pro-inflammatory Streptococcus species (S. parasanguinis, S. gordonii, S. infantis, and S. cristatus),35 A B C D Figure 1. Study design, cardiometabolic conditions, and oral bacteria in the full cohort (A) A total of 669 mouthwash samples were collected from the Emirati nationals enrolled in the ongoing UAEHFS for 16S rRNA gene sequencing (blue boxes). A total of 58 clinical biomarkers from blood and urine samples were assessed from these individuals (pink boxes), and a total of 145 demographic, lifestyle, and anthropometric variables were collected from self-completed questionnaires or physical exams (green boxes). In addition, a subset of 192 individuals that included 97 obese and 95 matched healthy-weight individuals (controls) was used for shotgun metagenomics sequencing and untargeted LC-MS metabolomics. (B) Relative abundances of the bacterial phyla in the mouthwash samples in 628 participants retained after quality control show the interindividual variation in oral microbiota. (C) Alpha diversity calculated using Shannon’s diversity index of 628 individuals is strongly correlated with the BBR, Spearman’s ρ = − 0.71, p < 2.2 × 10− 16, Spearman’s rank correlation test). (D), Postprandial interval, BMI categories, and smoking frequency were most strongly associated with oral microbiome composition. Tachycardia and anemia were also associated, but their effect sizes were relatively small. p values < 0.05 are shown on top of the bars. For (B) and (C), participants are sorted by the relative abundance of Bacteroidota on x axis. See Table S2 for a full list of coefficients and p values. Cell Reports 45, 116819, February 24, 2026 3 Article ll OPEN ACCESS Actinomyces oris, and A. massiliensis.36 The elevated presence of these taxa may create a microenvironment conducive to the growth of cariogenic bacteria, including Lactobacillus gasseri and Limosilactobacillus fermentum,37 which—although less frequently detected–were more abundant in obese participants. Species typically found in the oral cavity, such as Capnocyto- phaga SGB2499, SR1 HOT 345 SGB6893, and Tannerella HOT 286 SGB2048, were depleted (Figure S2; full list Table S6). Amplicon sequencing confirmed these patterns (Figure S3). Specifically, obese participants showed higher BBR (p = 5 × 10− 4, GLM), lower diversity (p = 0.031, GLM), and distinct composition compared to controls (p < 0.001, EnvFit). ASVs associated with poor oral health, such as Streptococcus, Actino- myces, and Lactobacillus were enriched in obesity, while those linked to oral health, such as Haemophilus, Capnocytophaga, and Veillonella,18,38 were depleted (Table S6). These findings un- derscore the distinct differences in the oral microbial ecosystem in obesity, which are consistent with previous reports.17,18,20 Obesity-associated functional changes correspond to metabolic alterations Beyond compositional shifts, we identified significant shifts in microbial metabolic capacity associated with obesity. Of 321 identified pathways, 94 differed significantly between obese and control groups (Figure 4A; Table S7, FDR-corrected p < 0.05, MaAsLin2). At the enzyme level, 457 of 1,824 microbial enzymes showed differential abundance between groups (FDR-corrected p < 0.05, Table S8), mirroring the observed pathway-level changes (Figure 5). Fifty-three pathways were en- riched in obesity, including carbohydrate fermentation to lactate, A B C D Figure 2. Obesity-associated cardiometabolic biomarkers in the matched cohort (A) PCA of clinical markers revealed significant differences between the obese (n = 97) and controls (n = 94) along the top two principal components (p = 0.03 and 0.001, respectively, GLM). (B) p values from EnvFit analysis of demographic and lifestyle variables independent of the clinical markers. (C) ROC curves for random forest classifier of obese (purple) and control (orange) individuals. (D) Clinical markers with high VIF in the random forest analysis differ significantly between the obese and control groups. TGs, triglycerides; GGT, gamma- glutamyl transferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GLUH, blood glucose level; and HbA1c, hemoglobin A1c. p values are mentioned for each clinical marker. 4 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS A B C D Figure 3. Obesity-associated variation in the oral microbiota assessed using shotgun metagenomic sequencing (A) PCoA performed using Bray-Curtis distances at the species level for the 191 individuals (obese n = 97, controls n = 94) reveals small but significant differences in the oral microbiome composition between the obese and control groups. Each dot is an individual, and color denotes obesity status (purple: obese, orange: controls). The size of the dot is scaled to the BMI of the individual. Boxplots reveal differences in the distributions of obese and control individuals across the two primary PCo axes. Both PCo 1 and 2 were associated with obese status (p = 0.035 and 0.0016, respectively, GLM). (B) Volcano plot showing 26 differentially abundant species between obese and controls. Key species are labeled. The list of coefficients and p values for all 366 taxa is shown in Table S6. (C) Receiver operating characteristics (ROC) curve (AUC=0.75) for a random forest classifier trained using the 26 differentially abundant species differentiates the obese (purple) from control (orange) individuals. (D) Boxplots demonstrating differences in relative abundances of key obesity enriched species with FDR-adjusted p values. Relative abundances of the re- maining 20 differentially abundant species are shown in Figure S2. Cell Reports 45, 116819, February 24, 2026 5 Article ll OPEN ACCESS A B C Figure 4. Obesity-associated oral microbial functional differences (A) Abundance of the 94 bacterial pathways significantly altered in obesity (obese n = 97, controls n = 94). Coefficients and p values for all 321 pathways are shown in Table S7. The differentially abundant pathways are in rows, while each column represents an individual. The barplot below the heatmap indicates the obesity status of the individual. The numbers in the left column show the R2 values obtained from MaAsLin2. Cells in blue indicate pathways enriched in obesity and those in yellow show obesity-depleted pathways. The dendrograms from unsupervised hierarchical clustering on the right cluster pathways according to obesity status. (legend continued on next page) 6 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS histidine degradation to imidazole propionate (IMP), uracil/uri- dine biosynthesis, and production of obesogenic metabolites, all previously implicated in obesity39–47 (Figure 5). Plasma lactate is often elevated in obesity and linked to inflammation and insulin resistance40–42; histidine degradation leads to biosynthesis of urocanic acid and IMP, which impairs insulin signaling43,44; and amino acids, such as cystine, methionine, and leucine promote adipogenesis.45–47 In contrast, 41 pathways, including several B vitamin biosynthesis pathways, were depleted. Sensitivity analysis with matched subsets (94 cases and 94 matched con- trols) yielded consistent results (Figure S4). To investigate whether these microbial functional shifts contribute to metabolic changes in the oral environment, we per- formed untargeted LC-MS metabolomics on 191 mouthwash su- pernatants, enriching for bacterial-derived metabolites, which yielded 3,551 high-quality metabolite features after quality con- trol (methods). Multivariate orthogonal partial least squares- discriminant analysis (OPLS-DA) analysis revealed clear separa- tion between obese and control groups (Figure 4B). Annotation of metabolite features using curated bacterial da- tabases identified 285 metabolites across various chemical clas- ses, including organic acids, energy molecules, nucleotides, and carbohydrates (Figures S5A; Table S9; Methods). These were enriched for compounds commonly found in saliva and feces and showed no enrichment consistent with dietary exposure (Figures 5B and 5C). Histidine metabolism and arginine biosyn- thesis were the two most significantly enriched pathways (Figure 5D), both canonical bacterial processes not active in oral epithelial cells,48 supporting their microbial origin. Procrustes analysis, comparing microbial functional and metabolite distance matrices (using the 285 annotated metabo- lites), revealed a strong correlation (ρ = 0.69, p = 0.001, permu- tation test, Figure S6), indicating that oral microbial function con- tributes to salivary metabolic differences. Consistent with this, 29 metabolites differed significantly between groups, all en- riched in obesity (Figures 4C; Table S10). Notably, nine corre- sponded to the end products of microbial pathways also en- riched in the obese group (Figure 5), reinforcing the functional link between microbial activity and metabolomic shifts. The obese group showed enrichment of the superpathway of pyrimidine deoxyribonucleosides degradation (PWY0-1298), key enzymes involved in uracil and uridine production—pyrimidine- nucleoside phosphorylase (E.C. 2.4.2.2) and uracil phosphoribo- syltransferase (E.C. 2.4.2.9)—and higher levels of salivary uracil and uridine (Figure 5). Lactobacilli, enriched in the obese group, can convert uracil and 5-phospho-α-D-ribose 1-diphosphate (PRPP) into uridine 5-monophosphate,49 which increases circu- lating uridine, enhancing hunger and caloric intake.39 Obese in- dividuals also showed enrichment in pathways converting nucleosides to PRPP, with key enzymes—ribose-phosphate di- phosphokinase (E.C. 2.7.6.1), amidophosphoribosyltransferase (E.C. 2.4.2.14), and 5-(carboxyamino)imidazole ribonucleotide synthase (E.C. 6.3.4.18) implicated in insulin resistance50 and non-alcoholic fatty liver disease.51 PRPP additionally feeds into histidine biosynthesis. While histidine itself is inversely associ- ated with BMI52 and improves glycemic control,53 its derivatives, 1-methylimidazole acetic acid54 and urocanic acid55— both en- riched in obese participants are linked to obesity comorbidities (Figure 5). Other obesity-enriched pathways included lactate production, methionine metabolism, and farnesol biosynthesis (Figure 5). Increased L-lactase dehydrogenase (E.C. 1.1.1.27) in the obese group correlated with elevated salivary lactate, a key mediator of adipose inflammation and insulin resistance,56,57 highlighting the potential role of adipose lactate as a signaling molecule in obesity.40 Methionine metabolism has been associated with adiposity and weight gain,58 while farnesol metabolism has been linked to sleep disturbances in obesity.59 These findings collectively underscore the intricate connections between oral microbial pathways and metabolic dysregulation in obesity. Additionally, we detected an elevated salivary retinol in the obese group, consistent with enrichment of pathways and en- zymes related to the biosynthesis and catalysis of retinal (Figure 5). We also detected elevated diaminopimelate (DAP) in the obese group, which results from catalysis of aspartate by en- zymes (e.g., E.C. 1.17.1.8, 2.3.1.89, and 3.5.1.47) within the L-lysine biosynthesis pathway (PWY-2941). In contrast, controls were enriched for diaminopimelate epimerase (E.C. 5.1.1.7), which converts DAP into lysine. While the role of retinol in meta- bolic disease is emerging60 and the relationship between DAP and obesity remains unclear, these molecules represent potential novel mediators of host-microbe metabolic interactions.61 The roles of these molecules in obesity warrant further investigation. Obese participants also showed a depletion of 41 microbial pathways, many involved in the biosynthesis of essential B vita- mins (flavin/B2, pyridoxal 5-phosphate/B6, biotin/B7, and cobal- amin/B12), heme synthesis, and nitrate reduction (Figure 5). B vi- tamins are crucial for both human physiology and microbial ecosystem function.62 Notably, several enzymes required for B12 biosynthesis were depleted, including riboflavin synthase (E.C. 2.5.1.9), nicotinate nucleotide dimethyl benzimidazole phos- phoribosyl transferase (E.C. 2.4.2.21), and enzymes producing adeonsylcobinamide-GDP from uroporphyrinogen (e.g., uropor- phyrinogen-III C-methyltransferase: E.C. 2.1.1.107 and adenosyl- cobinamide-phosphate guanylyltransferase: E.C. 2.7.7.62). Vitamin B12 and heme biosynthesis pathways share several en- zymes,63 and two involved in converting uroporphyrinogen to heme (e.g., protoporphyrinogen IX dehydrogenase: E.C. 1.3.5.3 and coproporphyrinogen dehydrogenase: E.C. 1.3.98.3), were also depleted in obese participants. Deficiencies in these microbial functions have important meta- bolic implications: B12 deficiency can alter host lipid metabolism, reduced heme promotes adipogenesis,64,65 and loss of vitamins B6 and B7 impairs carbohydrate and lipid metabolism,66 (B) An orthogonal partial least squares-discriminant analysis (OPLS-DA) using the normalized abundances of all metabolites. Each circle is an individual, color coded by their obesity status. The orthogonal component and the top predictive component of the OPLS-DA are shown. The predictive component (t[1]) dis- criminates obese and control individuals. OPLS-DA cross-validation values were: R2X = 0.28, R2Y = 0.96, and Q2 = 0.38. (C) Bar graph depicting the log2 fold change of 29 metabolites enriched in obesity. p values of all metabolites are listed in Table S9. Metabolites resulting from obesity enriched pathways and enzymes as shown in Figure 5 are labeled in red. Cell Reports 45, 116819, February 24, 2026 7 Article ll OPEN ACCESS potentially worsening obesity and its comorbidities.67 Since hu- mans cannot synthesize most vitamins, they rely on microbial production from dietary sources.62 Thus, the depletion of B-vitamin-producing oral bacteria may alter host energy meta- bolism and overall health.67,68 To our knowledge, no existing oral microbiome datasets integrate both metagenomic and me- tabolomic data, limiting external validation. Nevertheless, our findings lay a strong foundation for future functional validation via strain isolation and experimental assays. Oral bacterial species contributing to obesity- associated functional differences To evaluate species-level contributions to obesity-associated microbial functions, we correlated differentially abundant path- ways with the top 20% most abundant species. While no single species was exclusively associated with any of the 94 obesity- associated pathways, 7 of the 26 species enriched in obesity were positively correlated with obesity-enriched pathways and negatively correlated with those depleted in obesity (FDR- adjusted p < 0.05, Spearman’s rank correlation test, Figure S7; Table S11). For instance, pathways related to lactate production and purine or pyrimidine metabolism were positively correlated with obesity-enriched bacteria, including Gemella sanguinis, A. oris, S. parasanguinis, S. cristatus, and S. gordonii. Conversely, these same species were negatively correlated with B vitamin biosynthesis pathways (Figure S7). Next, we reconstructed 1,576 metagenome-assembled ge- nomes (MAGs) from 191 samples, which included 810 MAGs spanning 179 taxa from the obese individuals and 766 MAGs spanning 192 taxa from the controls (Table S12). After dereplica- tion, we retained 221 unique MAGs (uniMAGs, Average Nucleo- tide Identity [ANI] ≥ 95%) of medium or higher quality (based on Minimum Information about a Metagenome-Assembled Genome [MIMAG] standards,69 >50% completeness and <10% contami- nation determined with CheckM70). Forty-one uniMAGs could not be assigned to known species in the Genome Taxonomy Database (GTDB, release 226)71 or matched to the Unified Human Gastrointestinal Genome collection (UHGG ver2),72 the Chinese Gut Microbial Reference (CGMR),73 and Weizmann Institute of Science catalogue (WIS),74 suggesting they represent novel bac- terial species (novMAGs; Figure 6A). These novMAGs, primarily from Bacillota, Pseudomonadota, and Patescibacteria, formed monophyletic, distant branches relative to reference genomes, suggesting distinct gene content (Figure 6B). Pangenome analysis of the uniMAGs along with their closest taxonomic representatives for gene clusters obtained from Clus- ters of Orthologous Groups (COG) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) databases75,76 further revealed enrichment of coenzyme transport and metabolism genes in con- trol MAGs (FDR-corrected p < 0.05, Figures 6C; Table S13). Spe- cifically, Pseudomonadota MAGs from controls were enriched for genes involved in cobalamin (vitamin B12) biosynthesis, while Figure 5. Obesity-associated oral microbial metabolic reprogramming Of the 94 obesity-associated pathways, 67 were involved in seven categories shown in the boxes above. These microbial pathways (BioCyc; black text) and obesity-associated enzymes (E.C. numbers; green text) are shown. Pathways and enzymes in purple boxes are enriched in the obese group, and those in orange boxes are depleted. Metabolites enriched in the obese are in red; their fold changes in obese relative to controls are shown in Figure 4C. Blue indicates me- tabolites detected in the dataset, but their abundances did not differ significantly between the obese and control groups. Intermediate metabolites potentially involved in these pathways but not detected in our dataset are shown in gray. TCA, tricarboxylic acid cycle; DMAPP, dimethylallyl pyrophosphate; GPP, geranyl diphosphate; GGPP, geranylgeranyl diphosphate; PRPP, 5-phospho-α-D-ribose 1-diphosphate; AIR, 5-aminoimidazole ribonucleotide; ppGpp, guanosine tetraphosphate; FMN, flavin mononucleotide; DMB, dimethylbenzimidazole; GMP, guanosine monophosphate; GTP, guanosine triphosphate; IMP, inosine monophosphate. 8 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS A B C D E Figure 6. Taxonomic and functional diversity of MAGs recovered from obese and control individuals (A) A total of 221 uniMAGs of medium quality or higher (>50% completion, <10% contamination) and according to MIMAG standards were recovered. The bars indicate uniMAGs obtained from each phylum, and their percentages per phyla are shown in parentheses. Blue bars denote novel MAGs, and their percentages are listed to the right. Bacillota, Bacteroidota, and Actinomycetota had the most uniMAGs, while Bacillota had the highest percentage of novMAGs (31%, n = 25). (B) The phylogenomic relationship between the 1,576 MAGs based on 74 single-copy gene-set Hidden Markov Models. The outer bars indicate completeness (green) and contamination (red) for each MAG, and the red dots indicate uniMAGs. Phyla are colored according to the legend. Branch colors indicate MAGs obtained from obese (purple) and control (orange) participants. (C) COG and KEGG functions enriched in the MAGs obtained from the obese (top) and control (bottom) groups. In the obese group, a notable number of enriched genes with unknown functions came from Bacteroidota MAGs, and MAGs from Bacillota were enriched for genes involved in energy metabolism, whereas MAGs from controls showed enrichment of coenzyme transport and metabolism category, which included genes involved in cobalamin (vitamin B12) biosynthesis. The (legend continued on next page) Cell Reports 45, 116819, February 24, 2026 9 Article ll OPEN ACCESS Patescibacteria MAGs from obese participants were enriched for genes involved in histidine degradation and uridine monophos- phate biosynthesis (FDR-corrected p < 0.05, Figures 6C–E). Pre- vious studies have reported cobalamine synthesis genes across diverse gut77 and oral bacteria,63 supporting our findings. Together, these results highlight species-level functional differ- ences in the oral microbiome that may contribute to obesity- related metabolic perturbations. Integrative multi-omics analysis reveals microbe- metabolomic signatures linked to obesity-associated clinical markers Associations between shifts in the microbiome, metabolites, and clinical phenotypes indicate that microbial shifts may signifi- cantly impact host physiology and vice versa. A comprehensive association analysis integrating microbiome, metabolite, and serum biomarkers revealed plausible biological mechanisms by which oral bacteria may contribute to obesity (Figure S8; Table S14). Obesity-enriched species such as O. sinus corre- lated positively with acidic metabolites, including lactate, a key signaling molecule in obesity40 linked to metabolic disorders.57 Accordingly, lactate and these bacterial species were also posi- tively correlated with serum TGs (Figure 7A; Table S14). Acidic metabolites may lower oral pH, favoring lactic acid bacteria growth. Streptococcus species, including S. parasanguinis, har- bor a glycerophosphodiesterase enzyme (glpQ), which metabo- lizes dietary glycerophosphocholine into choline and glycerol-3- phosphate.78 Choline is essential for survival and is incorporated into the bacterial cell wall by biofilm-producing bacteria in the oral cavity,79 while glycerol-3-phosphate is a TG precursor. Consistent with these, S. parasanguinis and other obesity-en- riched bacterial species correlated with glycerophosphocholine, choline, and serum TGs (Figures 7A and 7B). Lactic acid bacteria also express histidine decarboxylases, breaking down histidine into imidazoles.80 Accordingly, S. para- sangiunis, A. oris, L. gasseri, and L. fermentum were also posi- tively associated with histidine and pyrimidine degradation path- ways as well as their metabolic products such as imidazoles, urocanic acid, uracil, and uridine (Figures 7A and 7D; Table S14). Uracil and uridine regulate liver function,81 while uro- canic acid and related histidine-derived metabolites are linked to obesity comorbidities.54,55 Consequently, both uridine and uracil were associated with elevated ALP, and urocanic acid with TGs (Figure S9). Conversely, obesity-depleted taxa, such as Neisse- ria subflava, C. gingivalis, and SGB1455 inversely correlated with obesity-associated metabolites and markers (Figures 6B–6E). Instead, they were positively correlated with asparagine, a metabolite inversely associated with obesity in Middle-Eastern adults82 (Table S14). We also evaluated bacteria-metabolite associations using hier- archical-all-against-all association testing (HAllA).83 This machine learning framework identifies covarying species-metabolite pairs in high-dimensional datasets and is well-suited for detecting non- linear associations and handling sparsity. HAllA corroborated positive associations between O. sinus and Streptococcus spe- cies and acidic metabolites (Figure 7E), while also identifying several additional bacteria-metabolite associations relevant to oral microbial ecology (Figures 7E and S10). For example, several obesity-depleted bacteria correlated with pyroglutamyl glycine, a bacterial nutrient and signaling molecule.84 Phosphoe- thanolamine was negatively associated with both obesity-en- riched and depleted taxa, likely reflecting its broad utilization by oral bacteria in membrane biosynthesis and cell envelope remodeling. Collectively, these results indicate that obesity-enriched oral bacteria generate metabolites that change the oral environment and influence systemic physiology. In addition to these, we de- tected many more bacteria-metabolite-serum metabolic marker associations in this dataset, which may be relevant to obesity and warrant further exploration (Table S14), and a complete list of HAllA identified bacteria-metabolite clusters is shown in Figure S10. Random forest models confirmed the predictive value of these features: clinical markers alone yielded an AUC of 0.81, which increased to 0.91 with microbial species and up to 0.95 when microbial pathways and salivary metabolites were included (Figure S11). These results indicate that oral microbiome features provide complementary information to clinical markers, although findings are exploratory given the cross-sectional design and require replication in independent cohorts. DISCUSSION This study provides the most comprehensive analysis of oral mi- crobial perturbations in obesity to date, addressing critical gaps in our understanding of how the oral microbiome influences sys- temic health. Previous studies have been limited by small sample sizes, inconsistent sequencing methods, and limited taxonomic resolution, resulting in conflicting findings. For example, smaller cohorts in the US (14 obese, 19 controls) and Qatar (37 obese, 36 lean) reported no significant differences in oral microbial diver- sity and composition between obese and lean individuals.85,86 While a larger US cohort (647 obese and 969 non-obese Ameri- cans) corroborated these results,38 a Chinese cohort (n = 659) identified significant associations between BMI and beta diver- sity.87 These inconsistencies underscore the need for large, deeply phenotyped cohorts with high-resolution multi-omics profiling. Additionally, lifestyle factors such as diet may contribute to discrepancies in findings, although such differ- ences tend to be small compared to the gut microbiomes.16 Furthermore, because of the use of 16S rRNA, oral microbiome analysis of most of these studies is limited to broad taxonomic categories translation, ribosomal structure and biogenesis, transcription, replication, recombination and repair, and cell wall/membrane/envelope biogenesis were grouped together into one category called translation, transcription, replication, and cell membrane biogenesis. (D and E) Pangenomic analysis for Pseudomonadota (D) and Patescibacteria (E), including dereplicated MAGs from obese (purple), controls (orange), and reference genomes (black). Gene clusters are organized according to a hierarchical clustering of their presence or absence (central dendrogram). Genomes are organized based on the presence or absence of gene clusters (top right dendrogram). Opacity denotes the presence or absence of the COG function in each genome. 10 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS U ro can ic acid Tannerella_sp_oral_taxon_HOT_286_SGB2048 Bacteroidaceae_bacterium_SGB1469 GGB3385_SGB4471_SGB4471 Campylobacter_showae_SGB19322 SR1_bacterium_HOT_345_SGB6893 Alloprevotella_sp_Lung230_SGB1467 GGB1138_SGB1455_SGB1455 Eikenella_corrodens_SGB9434 Capnocytophaga_gingivalis_SGB2479 Streptococcus_gordonii_SGB8053 Oribacterium_sinus_SGB5273 Streptococcus_parasanguinis_SGB8071 Eubacterium_brachy_SGB3931 P E A γ- G lu -B U tA L A sp ar ag in e pE H G G P C T hr eo ni ne L- A la ny l-L -P ro lin e N -A ce ty l-L -L eu ci ne 3- M et hy lth ym in e N ic ot in am id e 9- ox o- O D A P ic ol in ic a ci d N A cG lu A sc or bi c ac id M IA A 4- H P LA P P A C yt id in e L ac ti c ac id 5- M et hy l c yt os in e 4- M et ho xy -c in na m al de hy de B en zy lid en e ac et on e 1, 4, D im et hy l-i m id az ol e Im id az ol in on e Im id az ol e- 4- ac et al de hy de 2- M et hy l-4 -p en te no ic ac id Im idazo le Coniferol 1-Methylimidazole-acetic acid GPC A B C D E Figure 7. Microbiome-metabolite shifts are associated with serum cardiometabolic markers (A) Spearman’s rank correlations between 26 obesity-associated microbial species, 29 metabolites, and 5 cardiometabolic serum markers calculated for 191 individuals. Red lines indicate significant positive correlations, and blue lines represent negative correlations. Several obesity-enriched bacterial species (green outer segments) are positively associated with obesity enriched metabolites (blue outer segments) as well as serum cardiometabolic markers (pink outer seg- ments). Obese depleted species (brown outer segments) are negatively associated with these metabolites and serum markers. (B) Scatterplots showing positive associations between S. parasanguinis, choline, and serum TGs (top row), while C. gingivalis is negatively associated with both (bottom). (C) O. sinus is positively associated with lactate and TGs (top). N. subflava, a common oral bacterium, has negative associations with both of these features (bottom). (D) A. oris, urocanic acid, and TGs show positive correlations, while SGB1455, an uncharacterized species, is depleted in the obese and is negatively associated with both (bottom). Violin plots demonstrate differential abundances of these metabolites and TGs between obese and controls. Correlation coefficients and p values of the associations shown in (A–D) are listed in Table S14. (E) Key bacteria-metabolite associations identified by HAllA. PEP, phosphoethanolamine, γ-Glu-ButAL, N-(4-oxobutyl)-L-glutamine; pEHG, pyroglutamyl glycine; GPC, glycerophosphocholine; 9-oxo-ODA, 9-oxo-10(e) 12(e)-octadecadienoic acid; NAcGlu, N-Acetyl-L-glutamic acid; MIAA, methylimidazoleacetic acid; 4-HPLA, 4-hydroxyphenyllactic acid; and PPA, phenylpyruvic acid. Cell Reports 45, 116819, February 24, 2026 11 Article ll OPEN ACCESS levels. At the phylum level, BBR is positively associated with obesity86,88 along with higher relative abundances of Strepto- coccus, Gamella, Oribacterium, Corynebacterium, Bifidobacte- rium, Lactobacillus, and Actinomyces.38,85 Leveraging the UAEHFS, we identified significant shifts in the oral microbiome of obese individuals, including increased BBRs, along with shifts in microbial composition, and enrich- ment of pro-inflammatory species, such as Streptococcus parasanguinis, S. gordonii, S. cristatus, Gamella sanguinis, Oribacterium sinus, Corynebacterium durum, L. gasseri, and Actinomyces oris. In contrast, several novel bacterial species, such as SR1 HOT 345 SGB6893; Capnocytophaga SGB2499; Tannerella HOT 286 SGB2048; and Bacteroidaceae bacterium SGB1469, GGB3385 SGB4471, and GGB1138 SGB1455, were depleted. Many of these microbes were strongly associ- ated with serum metabolic markers of obesity, including TGs and ALP, linking oral dysbiosis to systemic metabolic reprogramming. All three obesity enriched Streptococcus species are natural members of the oral flora, playing a key role in oral biofilm forma- tion and shaping the oral microbial ecosystem through quorum sensing. Their metabolic activities include fermenting dietary car- bohydrate, which produces lactic acid and acidifies the oral cav- ity, creating an environment where opportunistic bacteria like S. parasanguinis can thrive. Consistent with this, we found an enrich- ment of Oribacterium sinus, L. fermentum, and L. gasseri in the obese participants. These bacteria are known to produce lactic acid as an end product.89 Enrichment of lactic-acid-producing bacteria was linked to elevated lactate in saliva in obese partici- pants. Lactate not only acidifies the oral cavity favoring barrier disruption but also enters blood circulation,13,90 where it promotes TG accumulation and adipogenesis,40,56 which are key features of obesity.57,91 Although part of the normal oral flora, Streptococcus parasanguinis, S. gordonii, and S. cristatus have been linked to systemic inflammation in obesity and metabolic syndromes.35 These species metabolize glycerophosphocholine into choline and TGs, both of which are key drivers of obesity-related meta- bolic disturbances and poor cardiovascular outcomes.92 The Emirati diet, rich in animal products such as eggs, meat, fish, and dairy, provides phosphatidylcholine and histidine, which can be converted by microbial enzymes into trimethylamine (TMA), trimethylamine N-oxide (TMAO), glycerophosphocholine, and choline.93 While such processes are typically attributed to gut microbes, we found several obesity enriched oral bacteria positively correlated with glycerophosphocholine, choline, and serum TGs, implicating them in host lipid metabolism. Likewise, obesity-enriched bacteria such as L. fermentum encode histidine decarboxylases that convert histidine into histamine, urocanic acid, and imidazoles—metabolites associated with systemic inflammation, appetite dysregulation, circadian rhythm disrup- tion, and impaired insulin signaling.92,94,95 Consistent with this, we detected enrichment of histidine metabolism pathways and enzymes in the oral microbbiome of obese participants, which corresponded with elevated imidazole metabolites, reflecting increased histamine turnover and metabolic dysregulation often seen in obesity.96 Furthermore, we found carbohydrate metabolism in obese participants was redirected toward ribose-5-phosphate and PRPP synthesis, precursors for uracil and uridine. Several path- ways contributing to PRPP synthesis—including PRPP synthe- tase (E.C. 2.7.6.1), the key enzyme catalyzing PRPP forma- tion—were elevated in the obese group. Additionally, the superpathway of pyrimidine deoxyribonucleosides degradation (PWY0-1298), which potentially produces uracil and uridine, was also elevated in obese participants. Both uracil and uridine play a critical role in energy homeostasis.81 Their elevation is associated with increased hunger, higher caloric intake,39 hyper- glycemia, and oxidative stress,97,98 suggesting that the oral mi- crobiome contributes significantly to energy homeostasis, typi- cally attributed to the gut microbiome.99 Thus, disruptions in the oral microbiome may trigger inflammatory responses and contribute to obesity-related comorbidities.94–96 In addition, several microbial pathways involved in B vitamin synthesis were depleted in obesity, potentially contributing to metabolic dysfunction.100 While previous studies show an in- verse relationship between vitamin B12 and obesity101 and the gut microbiome to vitamin synthesis,102 cobalamin biosyn- thesis genes and salvage pathways have been detected in a wide range of bacteria.63 We observed that control-derived Pseudomonadota MAGs, including Lautropia mirabilis, Lautro- pia dentalis, and Ottowia spp., were enriched for B vitamin- related functions, whereas the adenosylcobalamin salvage pathway was depleted in the obese group, suggesting reduced B12 bioavailability. Depletion of these bacteria in obesity may impair heme synthesis, which is essential for oxygen sensing and electron transfer,66,67 with downstream effects, such as accumulation of fatty acids and exacerbation of obesity.68 While this warrants further investigation, these findings suggest that oral bacteria may contribute to local B12 production and that their depletion in obesity could impact host metabolism. Beyond these mechanistic insights, our results suggest that oral microbiome features may serve as predictive markers with potential diagnostic utility, warranting confirmation in inde- pendent cohorts. While gut microbiome has been extensively linked to obesity,6,103–105 the contribution of the oral microbiome in obesity and systemic diseases remains underexplored. We leveraged the mouthwash samples paired with blood and urine samples and detailed lifestyle information from the UAEHFS— a result of years of efforts in overcoming cultural barriers through extensive community engagement—and integrated high-dimen- sional multi-omics to investigate oral bacteria, their metabolic activities, and interactions with cardiometabolic markers. We focused on obesity due to its global relevance as a major risk fac- tor for cardiovascular diseases1,2 and its high prevalence among young Emirati nationals.5 Collectively, our findings offer initial in- sights into oral bacterial crosstalk in influencing systemic health beyond the oral diseases and expand current understanding of obesity-related microbial perturbations. Further validations across diverse populations and mechanistic insights through longitudinal studies will be critical to establish causality and offer a more comprehensive understanding of host-microbiome inter- actions in obesity. These findings also suggest opportunities for microbiome-based prevention and therapeutic strategies against obesity and underscore the importance of investigating the oral microbiome’s contributions to other complex diseases. 12 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS Limitations of the study Despite our best efforts, this study has several limitations that must be noted. First, we were unable to collect reliable dietary data and could not fully control for smoking or post-prandial inter- vals, all of which influence both microbiome and metabolome pro- files. Nevertheless, we detected consistent bacterial species, pathways, and enzyme differences between obese and control participants that correlated with respective metabolites and clin- ical markers in the expected directions. Second, a large fraction of our sequencing reads mapped to the human genome, which likely hinders large-scale oral metagenomics research. Only 10% of the metagenomics reads were bacterial, limiting assembly of representative MAGs. Despite these challenges, we recovered ∼1,500 metagenomes from∼200 bacterial species. Although this is fewer than expected based on 16S rRNA profiling from the same samples, we identified 41 MAGs (<0.5% of all MAGs) with no close matches in public databases, highlighting gaps in our understand- ing of the oral microbiome taxonomic and functional diversity, particularly in underrepresented populations.106,107Third, our me- tabolome profiles likely capture only a subset of stable metabolites due to sample collection constraints that may not have preserved volatile or less-stable metabolites. Notably, we did not detect B vi- tamins, likely because salivary proteins sequester free vitamins.108 While these proteins may protect it from degradation in the stom- ach, protein-bound B vitamins in saliva samples are not detectable in the metabolome using LC-MS. Finally, as a cross-sectional study, we cannot infer causality between oral microbiome features and obesity. Longitudinal or interventional designs, coupled with detailed dietary data, will be needed to disentangle dietary effects and establish causality. However, the oral metabolomic profiles we generated align closely with those from two previous studies involving small cohorts of patients with type 2 diabetes mellitus (n = 20)109 and depression-obesity comorbidity, though those studies lacked metagenomic data.110 Future work focusing on improved sample preservation, depletion of human DNA, and deeper sequencing for better characterization of the oral micro- biome will be essential to overcoming these limitations. RESOURCE AVAILABILITY Lead contact For further information, resources, or requests, please contact the lead inves- tigator, Dr. Aashish R. Jha (jhaar@nyu.edu). Materials availability This study did not generate new, unique reagents. Data and code availability Controlled-access data including raw amplicon data, human-depleted microbial shotgun metagenomic sequence data, metabolomics data, and phenotypic data that support the findings of this study are available at National Center for Biotech- nology Information’s controlled-access dbGaP (accession ID dbGaP: phs004453.v1.p1). Investigators may contact NYUAD Public Health Research Center (nyuad.phrc@nyu.edu) or Aashish R. Jha (jhaar@nyu.edu) to facilitate data access for non-profit research purposes. Summary data necessary to reproduce all analyses in this manuscript are included in the supplementary tables. All software used to perform these analyses is publicly available. Software tools used are listed in the main text and methods section. Any additional information required to reanalyze the data reported in this pa- per is available from the lead contact upon request. CONSORTIA The UAE Healthy Future Study Investigators Group: Scott Sher- man, Muna Tahlak, May Raouf, Mahera Abdul-Rahman, Hamda Khansaheb, Abdullah Al Naeemi, Abdullah Al Junaibi, Eiman Al Zaabi, Naima Oumeziane, Marina Bastaki, Huda Al Shamsi, Mo- hammed Al Houqani, Fatma Al Maskari, Juma Al Kaabi, Ayesha Al Dhaheri, Luai Ahmed, Fatme Al Anouti, Tom Loney, Mo- hammed Hag-Ali, Habiba Al Safar, Habiba Ali, Wael Al Mahmeed, Asma Al Mannaei, Omniyat Al Hajeri, Erik Koornneef, Laila Abdel Wareth, Mai Ahmed Sultan Essa Aljaber, Basema Saddik, Walid Zaher, Hayfa Hamed, Buthaina Abdulla, Jamila Yacoub, Ali Abdul Kareem Al Obaidli, Andrea Jabari, Amar Ahmad, Yvonne Valles, Fatima Al Maisary, Imane Morjane, Sara Al Balushi, Mitha Alba- lushi, Manal Taimah Kayed, Vinu Manikandan, Manal Al Balooshi, Ayesha Al Hosani, Khaloud Alremeithi, Thekra Al Zaabi, Tamadher Alameri, Maryam Abed Al Marri, Mariem Elhadj, Fatma Al Shu- hoomi, Ghausia Begum, Klaithem Mohamed, Manar T. Al Shaikh, Hassan Sabour Abdrabo, and Judalyn Del Monte. ACKNOWLEDGMENTS This study is based on the work supported by Tamkeen under the Research Institute, New York University Abu Dhabi (grant no. G1206). We express our sincere gratitude to all volunteer participants, members of the Public Health Research Center (PHRC), and the NYUAD Core Technology Platform Sequencing Center. We would also like to acknowledge the NYUAD High-Per- formance Computing Core team for providing computational resources. We are grateful to Jayaram Radhakrishnan in the NYUAD Bioinformatics Core for assistance with data preprocessing and bioinformatics, and the members of the Genetic Heritage Group for helpful discussions and constructive critique throughout this study. We thank Ms. Siya Adhikari of Basis High School, Tuc- son, Arizona, for the graphical abstract. AUTHOR CONTRIBUTIONS Conceptualization, A.R.J., sample collection, A.A., R.A., Y.I., and The UAE Healthy Future Study Investigators Group; experiments, T.W.D., C.E.L., L.U., G.Z., M. AbdelBaqi, and M. Arnoux.; data analysis, A.A.S., T.W.D., A.R.A., S.A., S.A.A., Y.M., M. Arshad., N.D., and A.R.J.; interpretation, A.A.S., T.W.D., A.R.A., S.A., S.A.A., Y.I., and A.R.J.; original draft, A.A.S., T.W.D., A.R.A., S.A., and A.R.J.; writing and editing subsequent versions, A.A.S., T.W.D., A.R.A., S.A., S.A.A., Y.I., and A.R.J.; supervision, A.R.J, S.A.A., Y.I., and R.A.; funding, A.R.J., Y.I., and R.A.; and all authors discussed the results, contributed to the final manuscript, and have read and approved the submitted version. DECLARATION OF INTERESTS The authors declare no competing interests. STAR★METHODS Detailed methods are provided in the online version of this paper and include the following: • KEY RESOURCES TABLE • EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS ○ Study design and mouthwash sample collection ○ Ethics statement • METHOD DETAILS ○ Blood and urine biomarker measurements ○ DNA extraction ○ 16S rRNA gene sequencing Cell Reports 45, 116819, February 24, 2026 13 Article ll OPEN ACCESS mailto:jhaar@nyu.edu mailto:nyuad.phrc@nyu.edu mailto:jhaar@nyu.edu ○ 16S rRNA amplicon processing ○ Identification of matched cohort and lifestyle and health-associated variable selection ○ Blood and urine marker analysis ○ Metagenomics library preparation and sequencing ○ Metagenomic processing, taxonomic profiling, and functional char- acterization ○ Metagenomic assembly and functional analyses ○ Untargeted metabolomics using LC-MS and LC-MS/MS • QUANTIFICATION AND STATISTICAL ANALYSIS ○ Oral microbial diversity analyses using 16S rRNA amplicons ○ Random forests classifier ○ Oral microbial composition ○ Oral microbiome-obesity associations ○ Differential abundance ○ Metabolite abundances ○ Procrustes analyses ○ Microbiome blood marker associations SUPPLEMENTAL INFORMATION Supplemental information can be found online at https://doi.org/10.1016/j. celrep.2025.116819. Received: March 2, 2025 Revised: September 11, 2025 Accepted: December 5, 2025 Published: January 22, 2026 REFERENCES 1. WHO (2024). Obesity and overweight (WHO 2024). (https://www.who.int/ news-room/fact-sheets/detail/obesity-and-overweight. 2. Bluher, M. (2019). Obesity: global epidemiology and pathogenesis. Nat. Rev. Endocrinol. 15, 288–298. https://doi.org/10.1038/s41574-019- 0176-8. 3. Loos, R.J.F., and Yeo, G.S.H. (2022). The genetics of obesity: from dis- covery to biology. Nat. Rev. Genet. 23, 120–133. https://doi.org/10. 1038/s41576-021-00414-z. 4. Musaiger, A.O., and Abuirmeileh, N.M. (1998). Food consumption pat- terns of adults in the United Arab Emirates. J. R. Soc. Promot. Health 118, 146–150. https://doi.org/10.1177/146642409811800304. 5. Mezhal, F., Oulhaj, A., Abdulle, A., AlJunaibi, A., Alnaeemi, A., Ahmad, A., Leinberger-Jabari, A., Al Dhaheri, A.S., AlZaabi, E., Al-Maskari, F., et al. (2023). High prevalence of cardiometabolic risk factors amongst young adults in the United Arab Emirates: the UAE Healthy Future Study. BMC Cardiovasc. Disord. 23, 137. https://doi.org/10.1186/s12872- 023-03165-3. 6. Turnbaugh, P.J., Ley, R.E., Mahowald, M.A., Magrini, V., Mardis, E.R., and Gordon, J.I. (2006). An obesity-associated gut microbiome with increased capacity for energy harvest. Nature 444, 1027–1031. https:// doi.org/10.1038/nature05414. 7. Van Hul, M., and Cani, P.D. (2023). The gut microbiota in obesity and weight management: microbes as friends or foe? Nat. Rev. Endocrinol. 19, 258–271. https://doi.org/10.1038/s41574-022-00794-0. 8. Nicholson, J.K., Holmes, E., Kinross, J., Burcelin, R., Gibson, G., Jia, W., and Pettersson, S. (2012). Host-gut microbiota metabolic interactions. Science 336, 1262–1267. https://doi.org/10.1126/science.1223813. 9. Zhao, L. (2013). The gut microbiota and obesity: from correlation to cau- sality. Nat. Rev. Microbiol. 11, 639–647. https://doi.org/10.1038/ nrmicro3089. 10. Miyamoto, J., Igarashi, M., Watanabe, K., Karaki, S.I., Mukouyama, H., Kishino, S., Li, X., Ichimura, A., Irie, J., Sugimoto, Y., et al. (2019). Gut mi- crobiota confers host resistance to obesity by metabolizing dietary poly- unsaturated fatty acids. Nat. Commun. 10, 4007. https://doi.org/10. 1038/s41467-019-11978-0. 11. Sarmiento-Andrade, Y., Suárez, R., Quintero, B., Garrochamba, K., and Chapela, S.P. (2022). Gut microbiota and obesity: New insights. Front. Nutr. 9, 1018212. https://doi.org/10.3389/fnut.2022.1018212. 12. Hajishengallis, G., and Chavakis, T. (2021). Local andsystemic mechanisms linking periodontal disease and inflammatory comorbidities. Nat. Rev. Im- munol. 21, 426–440. https://doi.org/10.1038/s41577-020-00488-6. 13. Baker, J.L., Mark Welch, J.L., Kauffman, K.M., McLean, J.S., and He, X. (2024). The oral microbiome: diversity, biogeography and human health. Nat. Rev. Microbiol. 22, 89–104. https://doi.org/10.1038/s41579-023- 00963-6. 14. Peterson, J., Garges, S., Giovanni, M., McInnes, P., Wang, L., Schloss, J.A., Bonazzi, V., McEwen, J.E., and Wetterstrand, K.A. (2009). The NIH Human Microbiome Project. Genome Res. 19, 2317–2323. https:// doi.org/10.1101/gr.096651.109. 15. Santonocito, S., Giudice, A., Polizzi, A., Troiano, G., Merlo, E.M., Scla- fani, R., Grosso, G., and Isola, G. (2022). A Cross-Talk between Diet and the Oral Microbiome: Balance of Nutrition on Inflammation and Im- mune System’s Response during Periodontitis. Nutrients 14, 2426. https://doi.org/10.3390/nu14122426. 16. Ryu, E.P., Gautam, Y., Proctor, D.M., Bhandari, D., Tandukar, S., Gupta, M., Gautam, G.P., Relman, D.A., Shibl, A.A., Sherchand, J.B., et al. (2024). Nepali oral microbiomes reflect a gradient of lifestyles from tradi- tional to industrialized. Microbiome 12, 228. https://doi.org/10.1186/ s40168-024-01941-7. 17. Schamarek, I., Anders, L., Chakaroun, R.M., Kovacs, P., and Rohde-Zim- mermann, K. (2023). The role of the oral microbiome in obesity and meta- bolic disease: potential systemic implications and effects on taste percep- tion. Nutr. J. 22, 28. https://doi.org/10.1186/s12937-023-00856-7. 18. Wu, Y., Chi, X., Zhang, Q., Chen, F., and Deng, X. (2018). Characteriza- tion of the salivary microbiome in people with obesity. PeerJ 6, e4458. https://doi.org/10.7717/peerj.4458. 19. Lamont, R.J., Koo, H., and Hajishengallis, G. (2018). The oral microbiota: dynamic communities and host interactions. Nat. Rev. Microbiol. 16, 745–759. https://doi.org/10.1038/s41579-018-0089-x. 20. Gasmi Benahmed, A., Gasmi, A., Dos‚a, A., Chirumbolo, S., Mujawdiya, P.K., Aaseth, J., Dadar, M., and Bjørklund, G. (2021). Association be- tween the gut and oral microbiome with obesity. Anaerobe 70, 102248. https://doi.org/10.1016/j.anaerobe.2020.102248. 21. Arimatsu, K., Yamada, H., Miyazawa, H., Minagawa, T., Nakajima, M., Ryder, M.I., Gotoh, K., Motooka, D., Nakamura, S., Iida, T., and Yama- zaki, K. (2014). Oral pathobiont induces systemic inflammation and meta- bolic changes associated with alteration of gut microbiota. Sci. Rep. 4, 4828. https://doi.org/10.1038/srep04828. 22. Baker, J.L. (2023). Illuminating the oral microbiome and its host interac- tions: recent advancements in omics and bioinformatics technologies in the context of oral microbiome research. FEMS Microbiol. Rev. 47, fuad051. https://doi.org/10.1093/femsre/fuad051. 23. Ma, W., Wang, Y., Nguyen, L.H., Mehta, R.S., Ha, J., Bhosle, A., Mclver, L.J., Song, M., Clish, C.B., Strate, L.L., et al. (2024). Gut microbiome composition and metabolic activity in women with diverticulitis. Nat. Commun. 15, 3612. https://doi.org/10.1038/s41467-024-47859-4. 24. Abdulle, A., Alnaeemi, A., Aljunaibi, A., Al Ali, A., Al Saedi, K., Al Zaabi, E., Oumeziane, N., Al Bastaki, M., Al-Houqani, M., Al Maskari, F., et al. (2018). The UAE healthy future study: a pilot for a prospective cohort study of 20,000 United Arab Emirates nationals. BMC Public Health 18, 101. https://doi.org/10.1186/s12889-017-5012-2. 25. Azizi, F., Hadaegh, F., Hosseinpanah, F., Mirmiran, P., Amouzegar, A., Abdi, H., Asghari, G., Parizadeh, D., Montazeri, S.A., Lotfaliany, M., et al. (2019). Metabolic health in the Middle East and north Africa. Lancet Diabetes Endocrinol. 7, 866–879. https://doi.org/10.1016/S2213- 8587(19)30179-2. 14 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS https://doi.org/10.1016/j.celrep.2025.116819 https://doi.org/10.1016/j.celrep.2025.116819 https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight https://doi.org/10.1038/s41574-019-0176-8 https://doi.org/10.1038/s41574-019-0176-8 https://doi.org/10.1038/s41576-021-00414-z https://doi.org/10.1038/s41576-021-00414-z https://doi.org/10.1177/146642409811800304 https://doi.org/10.1186/s12872-023-03165-3 https://doi.org/10.1186/s12872-023-03165-3 https://doi.org/10.1038/nature05414 https://doi.org/10.1038/nature05414 https://doi.org/10.1038/s41574-022-00794-0 https://doi.org/10.1126/science.1223813 https://doi.org/10.1038/nrmicro3089 https://doi.org/10.1038/nrmicro3089 https://doi.org/10.1038/s41467-019-11978-0 https://doi.org/10.1038/s41467-019-11978-0 https://doi.org/10.3389/fnut.2022.1018212 https://doi.org/10.1038/s41577-020-00488-6 https://doi.org/10.1038/s41579-023-00963-6 https://doi.org/10.1038/s41579-023-00963-6 https://doi.org/10.1101/gr.096651.109 https://doi.org/10.1101/gr.096651.109 https://doi.org/10.3390/nu14122426 https://doi.org/10.1186/s40168-024-01941-7 https://doi.org/10.1186/s40168-024-01941-7 https://doi.org/10.1186/s12937-023-00856-7 https://doi.org/10.7717/peerj.4458 https://doi.org/10.1038/s41579-018-0089-x https://doi.org/10.1016/j.anaerobe.2020.102248 https://doi.org/10.1038/srep04828 https://doi.org/10.1093/femsre/fuad051 https://doi.org/10.1038/s41467-024-47859-4 https://doi.org/10.1186/s12889-017-5012-2 https://doi.org/10.1016/S2213-8587(19)30179-2 https://doi.org/10.1016/S2213-8587(19)30179-2 26. Baker, J.L., Morton, J.T., Dinis, M., Alvarez, R., Tran, N.C., Knight, R., and Edlund, A. (2021). Deep metagenomics examines the oral microbiome during dental caries, revealing novel taxa and co-occurrences with host molecules. Genome Res. 31, 64–74. https://doi.org/10.1101/gr. 265645.120. 27. Bakhtiyari, M., Kazemian, E., Kabir, K., Hadaegh, F., Aghajanian, S., Mardi, P., Ghahfarokhi, N.T., Ghanbari, A., Mansournia, M.A., and Azizi, F. (2022). Contribution of obesity and cardiometabolic risk factors in developing cardiovascular disease: a population-based cohort study. Sci. Rep. 12, 1544. https://doi.org/10.1038/s41598-022-05536-w. 28. Smith, G.I., Shankaran, M., Yoshino, M., Schweitzer, G.G., Chondroni- kola, M., Beals, J.W., Okunade, A.L., Patterson, B.W., Nyangau, E., Field, T., et al. (2020). Insulin resistance drives hepatic de novo lipogenesis in nonalcoholic fatty liver disease. J. Clin. Investig. 130, 1453–1460. https://doi.org/10.1172/JCI134165. 29. Kashyap, P., and Farrugia, G. (2011). Oxidative stress: key player in gastrointestinal complications of diabetes. Neuro Gastroenterol. Motil. 23, 111–114. https://doi.org/10.1111/j.1365-2982.2010.01659.x. 30. Son, D.H., Ha, H.S., and Lee, Y.J. (2021). Association of Serum Alkaline Phosphatase with the TG/HDL Ratio and TyG Index in Korean Adults. Biomolecules 11, 882. https://doi.org/10.3390/biom11060882. 31. Johansen, M.J., Gade, J., Stender, S., Frithioff-Bøjsøe, C., Lund, M.A.V., Chabanova, E., Thomsen, H.S., Pedersen, O., Fonvig, C.E., Hansen, T., and Holm, J.C. (2020). The Effect of Overweight and Obesity on Liver Biochemical Markers in Children and Adolescents. J. Clin. Endocrinol. Metab. 105, dgz010. https://doi.org/10.1210/clinem/dgz010. 32. Deng, L., Jia, L., Wu, X.L., and Cheng, M. (2025). Association Between Body Mass Index and Glycemic Control in Type 2 Diabetes Mellitus: A Cross-Sectional Study. Diabetes Metab. Syndr. Obes. 18, 555–563. https://doi.org/10.2147/DMSO.S508365. 33. Blanco-Miguez, A. (2023). Extending and improving metagenomic taxo- nomic profiling with uncharacterized species using MetaPhlAn 4. Nat. Bio- technol. 41, 1633–1644. https://doi.org/10.1038/s41587-023-01688-w. 34. Mallick, H., Rahnavard, A., McIver, L.J., Ma, S., Zhang, Y., Nguyen, L.H., Tickle, T.L., Weingart, G., Ren, B., Schwager, E.H., et al. (2021). Multivari- able association discovery in population-scale meta-omics studies. PLoS Comput. Biol. 17, e1009442. https://doi.org/10.1371/journal. pcbi.1009442. 35. Bloch, S., Hager-Mair, F.F., Andrukhov, O., and Schäffer, C. (2024). Oral streptococci: modulators of health and disease. Front. Cell. Infect. Micro- biol. 14, 1357631. https://doi.org/10.3389/fcimb.2024.1357631. 36. Kononen, E., and Wade, W.G. (2015). Actinomyces and related organ- isms in human infections. Clin. Microbiol. Rev. 28, 419–442. https://doi. org/10.1128/CMR.00100-14. 37. Caufield, P.W., Schön, C.N., Saraithong, P., Li, Y., and Argimón, S. (2015). Oral Lactobacilli and Dental Caries: A Model for Niche Adaptation in Humans. J. Dent. Res. 94, 110S–118S. https://doi.org/10.1177/ 0022034515576052. 38. Yang, Y., Cai, Q., Zheng, W., Steinwandel, M., Blot, W.J., Shu, X.O., and Long, J. (2019). Oral microbiome and obesity in a large study of low-in- come and African-American populations. J. Oral Microbiol. 11, 1650597. https://doi.org/10.1080/20002297.2019.1650597. 39. Hanssen, R., Rigoux, L., Albus, K., Kretschmer, A.C., Edwin Thanarajah, S., Chen, W., Hinze, Y., Giavalisco, P., Steculorum, S.M., Cornely, O.A., et al. (2023). Circulating uridine dynamically and adaptively regulates food intake in humans. Cell Rep. Med. 4, 100897. https://doi.org/10. 1016/j.xcrm.2022.100897. 40. Feng, T., Zhao, X., Gu, P., Yang, W., Wang, C., Guo, Q., Long, Q., Liu, Q., Cheng, Y., Li, J., et al. (2022). Adipocyte-derived lactate is a signalling metabolite that potentiates adipose macrophage inflammation via tar- geting PHD2. Nat. Commun. 13, 5208. https://doi.org/10.1038/s41467- 022-32871-3. 41. Juraschek, S.P., Selvin, E., Miller, E.R., Brancati, F.L., and Young, J.H. (2013). Plasma lactate and diabetes risk in 8045 participants of the atherosclerosis risk in communities study. Ann. Epidemiol. 23, 791– 796.e4. https://doi.org/10.1016/j.annepidem.2013.09.005. 42. Jansson, P.A., Larsson, A., Smith, U., and Lönnroth, P. (1994). Lactate release from the subcutaneous tissue in lean and obese men. J. Clin. In- vestig. 93, 240–246. https://doi.org/10.1172/JCI116951. 43. Molinaro, A., Bel Lassen, P., Henricsson, M., Wu, H., Adriouch, S., Belda, E., Chakaroun, R., Nielsen, T., Bergh, P.O., Rouault, C., et al. (2020). Imidazole propionate is increased in diabetes and associated with die- tary patterns and altered microbial ecology. Nat. Commun. 11, 5881. https://doi.org/10.1038/s41467-020-19589-w. 44. Koh, A., Molinaro, A., Ståhlman, M., Khan, M.T., Schmidt, C., Mannerås- Holm, L., Wu, H., Carreras, A., Jeong, H., Olofsson, L.E., et al. (2018). Mi- crobially Produced Imidazole Propionate Impairs Insulin Signaling through mTORC1. Cell 175, 947–961.e17. https://doi.org/10.1016/j. cell.2018.09.055. 45. Saenz de Urturi, D. (2022). Methionine adenosyltransferase 1a antisense oligonucleotides activate the liver-brown adipose tissue axis preventing obesity and associated hepatosteatosis. Nat. Commun. 13, 1096. https://doi.org/10.1038/s41467-022-28749-z. 46. Newgard, C.B., An, J., Bain, J.R., Muehlbauer, M.J., Stevens, R.D., Lien, L.F., Haqq, A.M., Shah, S.H., Arlotto, M., Slentz, C.A., et al. (2009). A branched-chain amino acid-related metabolic signature that differenti- ates obese and lean humans and contributes to insulin resistance. Cell Metab. 9, 311–326. https://doi.org/10.1016/j.cmet.2009.02.002. 47. Elshorbagy, A.K., Smith, A.D., Kozich, V., and Refsum, H. (2012). Cysteine and obesity. Obesity 20, 473–481. https://doi.org/10.1038/ oby.2011.93. 48. Jorth, P., Turner, K.H., Gumus, P., Nizam, N., Buduneli, N., and Whiteley, M. (2014). Metatranscriptomics of the human oral microbiome during health and disease. mBio 5, e01012–e01014. https://doi.org/10.1128/ mBio.01012-14. 49. Arsene-Ploetze, F., Nicoloff, H., Kammerer, B., Martinussen, J., and Bringel, F. (2006). Uracil salvage pathway in Lactobacillus plantarum: Transcription and genetic studies. J. Bacteriol. 188, 4777–4786. https://doi.org/10.1128/JB.00195-06. 50. Ye, J. (2021). Mechanism of insulin resistance in obesity: a role of ATP. Front. Med. 15, 372–382. https://doi.org/10.1007/s11684-021-0862-5. 51. Testerman, T., Li, Z., Galuppo, B., Graf, J., and Santoro, N. (2022). In- sights from shotgun metagenomics into bacterial species and metabolic pathways associated with NAFLD in obese youth. Hepatol. Commun. 6, 1962–1974. https://doi.org/10.1002/hep4.1944. 52. Li, Y.C., Li, C.L., Qi, J.Y., Huang, L.N., Shi, D., Du, S.S., Liu, L.Y., Feng, R.N., and Sun, C.H. (2016). Relationships of Dietary Histidine and Obesity in Northern Chinese Adults, an Internet-Based Cross-Sectional Study. Nutrients 8, 420. https://doi.org/10.3390/nu8070420. 53. Warmbrunn, M.V., Attaye, I., Aron-Wisnewsky, J., Rampanelli, E., van der Vossen, E.W.J., Hao, Y., Koopen, A., Bergh, P.O., Stols-Gonçalves, D., Mohamed, N., et al. (2024). Oral histidine affects gut microbiota and MAIT cells improving glycemic control in type 2 diabetes patients. Gut Mi- crobes 16, 2370616. https://doi.org/10.1080/19490976.2024.2370616. 54. Rekha, B., Velmurugan, G., Freddy, A.J., Anusha, S., Ramprasath, T., Karthik, K.V., Suresh, S., Kulshrestha, P., Mithieux, G., Lyon, A.R., et al. (2018). Chronic intake of 4-Methylimidazole induces Hyperinsuline- mia and Hypoglycaemia via Pancreatic Beta Cell Hyperplasia and Glucose Dyshomeostasis. Sci. Rep. 8, 17037. https://doi.org/10.1038/ s41598-018-35071-6. 55. Liu, Y., Chen, L., Liu, L., Zhao, S.S., You, J.Q., Zhao, X.J., Liu, H.X., Xu, G.W., and Wen, D.L. (2022). Interplay between dietary intake, gut micro- biota, and metabolic profile in obese adolescents: Sex-dependent differ- ential patterns. Clin. Nutr. 41, 2706–2719. https://doi.org/10.1016/j.clnu. 2022.10.009. Cell Reports 45, 116819, February 24, 2026 15 Article ll OPEN ACCESS https://doi.org/10.1101/gr.265645.120 https://doi.org/10.1101/gr.265645.120 https://doi.org/10.1038/s41598-022-05536-w https://doi.org/10.1172/JCI134165 https://doi.org/10.1111/j.1365-2982.2010.01659.x https://doi.org/10.3390/biom11060882 https://doi.org/10.1210/clinem/dgz010 https://doi.org/10.2147/DMSO.S508365 https://doi.org/10.1038/s41587-023-01688-w https://doi.org/10.1371/journal.pcbi.1009442 https://doi.org/10.1371/journal.pcbi.1009442 https://doi.org/10.3389/fcimb.2024.1357631 https://doi.org/10.1128/CMR.00100-14 https://doi.org/10.1128/CMR.00100-14 https://doi.org/10.1177/0022034515576052 https://doi.org/10.1177/0022034515576052 https://doi.org/10.1080/20002297.2019.1650597 https://doi.org/10.1016/j.xcrm.2022.100897 https://doi.org/10.1016/j.xcrm.2022.100897 https://doi.org/10.1038/s41467-022-32871-3 https://doi.org/10.1038/s41467-022-32871-3 https://doi.org/10.1016/j.annepidem.2013.09.005 https://doi.org/10.1172/JCI116951 https://doi.org/10.1038/s41467-020-19589-w https://doi.org/10.1016/j.cell.2018.09.055 https://doi.org/10.1016/j.cell.2018.09.055 https://doi.org/10.1038/s41467-022-28749-z https://doi.org/10.1016/j.cmet.2009.02.002 https://doi.org/10.1038/oby.2011.93 https://doi.org/10.1038/oby.2011.93 https://doi.org/10.1128/mBio.01012-14 https://doi.org/10.1128/mBio.01012-14 https://doi.org/10.1128/JB.00195-06 https://doi.org/10.1007/s11684-021-0862-5 https://doi.org/10.1002/hep4.1944 https://doi.org/10.3390/nu8070420 https://doi.org/10.1080/19490976.2024.2370616 https://doi.org/10.1038/s41598-018-35071-6 https://doi.org/10.1038/s41598-018-35071-6 https://doi.org/10.1016/j.clnu.2022.10.009 https://doi.org/10.1016/j.clnu.2022.10.009 56. Lin, Y., Bai, M., Wang, S., Chen, L., Li, Z., Li, C., Cao, P., and Chen, Y. (2022). Lactate Is a Key Mediator That Links Obesity to Insulin Resistance via Modulating Cytokine Production From Adipose Tissue. Diabetes 71, 637–652. https://doi.org/10.2337/db21-0535. 57. Leija, R.G., Curl, C.C., Arevalo, J.A., Osmond, A.D., Duong, J.J., Huie, M.J., Masharani, U., and Brooks, G.A. (2024). Enteric and systemic post- prandial lactate shuttle phases and dietary carbohydrate carbon flow in humans. Nat. Metab. 6, 670–677. https://doi.org/10.1038/s42255-024- 00993-1. 58. Elshorbagy, A.K., Kozich, V., Smith, A.D., and Refsum, H. (2012). Cysteine and obesity: consistency of the evidence across epidemiologic, animal and cellular studies. Curr. Opin. Clin. Nutr. Metab. Care 15, 49–57. https://doi.org/10.1097/MCO.0b013e32834d199f. 59. Lee, J., Kang, J., Kim, Y., Lee, S., Oh, C.M., and Kim, T. (2023). Integrated analysis of the microbiota-gut-brain axis in response to sleep deprivation and diet-induced obesity. Front. Endocrinol. 14, 1117259. https://doi. org/10.3389/fendo.2023.1117259. 60. Blaner, W.S. (2019). Vitamin A signaling and homeostasis in obesity, dia- betes, and metabolic disorders. Pharmacol. Ther. 197, 153–178. https:// doi.org/10.1016/j.pharmthera.2019.01.006. 61. Lin, H.Y., Chen, C.C., Chen, Y.J., Lin, Y.Y., Mersmann, H.J., and Ding, S.T. (2014). Enhanced amelioration of high-fat diet-induced fatty liver by docosahexaenoic acid and lysine supplementations. BioMed Res. Int. 2014, 310981. https://doi.org/10.1155/2014/310981. 62. Tarracchini, C., Lugli, G.A., Mancabelli, L., van Sinderen, D., Turroni, F., Ventura, M., and Milani, C. (2024). Exploring the vitamin biosynthesis landscape of the human gut microbiota. mSystems 9, e0092924. https://doi.org/10.1128/msystems.00929-24. 63. Shelton, A.N., Seth, E.C., Mok, K.C., Han, A.W., Jackson, S.N., Haft, D.R., and Taga, M.E. (2019). Uneven distribution of cobamide biosyn- thesis and dependence in bacteria predicted by comparative genomics. ISME J. 13, 789–804. https://doi.org/10.1038/s41396-018-0304-9. 64. Moreno-Navarrete, J.M., Rodrı́guez, A., Ortega, F., Becerril, S., Girones, J., Sabater-Masdeu, M., Latorre, J., Ricart, W., Frühbeck, G., and Fer- nández-Real, J.M. (2017). Heme Biosynthetic Pathway is Functionally Linked to Adipogenesis via Mitochondrial Respiratory Activity. Obesity 25, 1723–1733. https://doi.org/10.1002/oby.21956. 65. Boachie, J., Adaikalakoteswari, A., Samavat, J., and Saravanan, P. (2020). Low Vitamin B12 and Lipid Metabolism: Evidence from Pre- Clinical and Clinical Studies. Nutrients 12, 1925. https://doi.org/10. 3390/nu12071925. 66. Parra, M., Stahl, S., and Hellmann, H. (2018). Vitamin B(6) and Its Role in Cell Metabolism and Physiology. Cells 7, 84. https://doi.org/10.3390/ cells7070084. 67. Belda, E., Voland, L., Tremaroli, V., Falony, G., Adriouch, S., Assmann, K.E., Prifti, E., Aron-Wisnewsky, J., Debédat, J., Le Roy, T., et al. (2022). Impairment of gut microbial biotin metabolism and host biotin sta- tus in severe obesity: effect of biotin and prebiotic supplementation on improved metabolism. Gut 71, 2463–2480. https://doi.org/10.1136/ gutjnl-2021-325753. 68. Cox, L.M., and Blaser, M.J. (2013). Pathways in microbe-induced obesity. Cell Metab. 17, 883–894. https://doi.org/10.1016/j.cmet.2013. 05.004. 69. Bowers, R.M., Kyrpides, N.C., Stepanauskas, R., Harmon-Smith, M., Doud, D., Reddy, T.B.K., Schulz, F., Jarett, J., Rivers, A.R., Eloe-Fa- drosh, E.A., et al. (2017). Minimum information about a single amplified genome (MISAG) and a metagenome-assembled genome (MIMAG) of bacteria and archaea. Nat. Biotechnol. 35, 725–731. https://doi.org/10. 1038/nbt.3893. 70. Parks, D.H., Imelfort, M., Skennerton, C.T., Hugenholtz, P., and Tyson, G.W. (2015). assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043– 1055. https://doi.org/10.1101/gr.186072.114. 71. Chaumeil, P.A., Mussig, A.J., Hugenholtz, P., and Parks, D.H. (2022). GTDB-Tk v2: memory friendly classification with the genome taxonomy database. Bioinformatics 38, 5315–5316. https://doi.org/10.1093/bioin- formatics/btac672. 72. Almeida, A., Nayfach, S., Boland, M., Strozzi, F., Beracochea, M., Shi, Z.J., Pollard, K.S., Sakharova, E., Parks, D.H., Hugenholtz, P., et al. (2021). A unified catalog of 204,938 reference genomes from the human gut microbiome. Nat. Biotechnol. 39, 105–114. https://doi.org/10.1038/ s41587-020-0603-3. 73. Huang, P., Dong, Q., Wang, Y., Tian, Y., Wang, S., Zhang, C., Yu, L., Tian, F., Gao, X., Guo, H., et al. (2024). Gut microbial genomes with paired iso- lates from China illustrate probiotic and cardiometabolic effects. Cell Ge- nom. 4, 100559. https://doi.org/10.1016/j.xgen.2024.100559. 74. Leviatan, S., Shoer, S., Rothschild, D., Gorodetski, M., and Segal, E. (2022). An expanded reference map of the human gut microbiome re- veals hundreds of previously unknown species. Nat. Commun. 13, 3863. https://doi.org/10.1038/s41467-022-31502-1. 75. Kanehisa, M., Furumichi, M., Sato, Y., Kawashima, M., and Ishiguro-Wa- tanabe, M. (2023). KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. 51, D587–D592. https://doi.org/10.1093/ nar/gkac963. 76. Galperin, M.Y., Wolf, Y.I., Makarova, K.S., Vera Alvarez, R., Landsman, D., and Koonin, E.V. (2021). COG database update: focus on microbial diversity, model organisms, and widespread pathogens. Nucleic Acids Res. 49, D274–D281. https://doi.org/10.1093/nar/gkaa1018. 77. Magnusdottir, S., Ravcheev, D., de Crecy-Lagard, V., and Thiele, I. (2015). Systematic genome assessment of B-vitamin biosynthesis sug- gests co-operation among gut microbes. Front. Genet. 6, 148. https:// doi.org/10.3389/fgene.2015.00148. 78. Geng, J., Chiu, C.H., Tang, P., Chen, Y., Shieh, H.R., Hu, S., and Chen, Y.Y.M. (2012). Complete genome and transcriptomes of Streptococcus parasanguinis FW213: phylogenic relations and potential virulence mechanisms. PLoS One 7, e34769. https://doi.org/10.1371/journal. pone.0034769. 79. Tettelin, H., Nelson, K.E., Paulsen, I.T., Eisen, J.A., Read, T.D., Peterson, S., Heidelberg, J., DeBoy, R.T., Haft, D.H., Dodson, R.J., et al. (2001). Complete genome sequence of a virulent isolate of Streptococcus pneu- moniae. Science 293, 498–506. https://doi.org/10.1126/science.1061217. 80. Rossi, F., Gardini, F., Rizzotti, L., La Gioia, F., Tabanelli, G., and Torriani, S. (2011). Quantitative analysis of histidine decarboxylase gene (hdcA) transcription and histamine production by Streptococcus thermophilus PRI60 under conditions relevant to cheese making. Appl. Environ. Micro- biol. 77, 2817–2822. https://doi.org/10.1128/AEM.02531-10. 81. Jain, S., and Jacobson, K.A. (2022). Adipocyte purinergic receptors acti- vated by uracil nucleotides as obesity and type 2 diabetes targets. Curr. Opin. Pharmacol. 63, 102190. https://doi.org/10.1016/j.coph.2022.102190. 82. Bagheri, M., Djazayery, A., Farzadfar, F., Qi, L., Yekaninejad, M.S., Asli- bekyan, S., Chamari, M., Hassani, H., Koletzko, B., and Uhl, O. (2019). Plasma metabolomic profiling of amino acids and polar lipids in Iranian obese adults. Lipids Health Dis. 18, 94. https://doi.org/10.1186/ s12944-019-1037-0. 83. Ghazi, A.R., Sucipto, K., Rahnavard, A., Franzosa, E.A., McIver, L.J., Lloyd-Price, J., Schwager, E., Weingart, G., Moon, Y.S., Morgan, X.C., et al. (2022). High-sensitivity pattern discovery in large, paired multiomic datasets. Bioinformatics 38, i378–i385. https://doi.org/10.1093/bioinfor- matics/btac232. 84. Bochtler, M., Mizgalska, D., Veillard, F., Nowak, M.L., Houston, J., Veith, P., Reynolds, E.C., and Potempa, J. (2018). The Bacteroidetes Q-Rule: Pyroglutamate in Signal Peptidase I Substrates. Front. Microbiol. 9, 230. https://doi.org/10.3389/fmicb.2018.00230. 85. Janem, W.F., Scannapieco, F.A., Sabharwal, A., Tsompana, M., Berman, H.A., Haase, E.M., Miecznikowski, J.C., and Mastrandrea, L.D. (2017). Salivary inflammatory markers and microbiome in normoglycemic lean 16 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS https://doi.org/10.2337/db21-0535 https://doi.org/10.1038/s42255-024-00993-1 https://doi.org/10.1038/s42255-024-00993-1 https://doi.org/10.1097/MCO.0b013e32834d199f https://doi.org/10.3389/fendo.2023.1117259 https://doi.org/10.3389/fendo.2023.1117259 https://doi.org/10.1016/j.pharmthera.2019.01.006 https://doi.org/10.1016/j.pharmthera.2019.01.006 https://doi.org/10.1155/2014/310981 https://doi.org/10.1128/msystems.00929-24 https://doi.org/10.1038/s41396-018-0304-9 https://doi.org/10.1002/oby.21956 https://doi.org/10.3390/nu12071925 https://doi.org/10.3390/nu12071925 https://doi.org/10.3390/cells7070084 https://doi.org/10.3390/cells7070084 https://doi.org/10.1136/gutjnl-2021-325753 https://doi.org/10.1136/gutjnl-2021-325753 https://doi.org/10.1016/j.cmet.2013.05.004 https://doi.org/10.1016/j.cmet.2013.05.004 https://doi.org/10.1038/nbt.3893 https://doi.org/10.1038/nbt.3893 https://doi.org/10.1101/gr.186072.114 https://doi.org/10.1093/bioinformatics/btac672 https://doi.org/10.1093/bioinformatics/btac672 https://doi.org/10.1038/s41587-020-0603-3 https://doi.org/10.1038/s41587-020-0603-3 https://doi.org/10.1016/j.xgen.2024.100559 https://doi.org/10.1038/s41467-022-31502-1 https://doi.org/10.1093/nar/gkac963 https://doi.org/10.1093/nar/gkac963 https://doi.org/10.1093/nar/gkaa1018 https://doi.org/10.3389/fgene.2015.00148 https://doi.org/10.3389/fgene.2015.00148 https://doi.org/10.1371/journal.pone.0034769 https://doi.org/10.1371/journal.pone.0034769 https://doi.org/10.1126/science.1061217 https://doi.org/10.1128/AEM.02531-10 https://doi.org/10.1016/j.coph.2022.102190 https://doi.org/10.1186/s12944-019-1037-0 https://doi.org/10.1186/s12944-019-1037-0 https://doi.org/10.1093/bioinformatics/btac232 https://doi.org/10.1093/bioinformatics/btac232 https://doi.org/10.3389/fmicb.2018.00230 and obese children compared to obese children with type 2 diabetes. PLoS One 12, e0172647. https://doi.org/10.1371/journal.pone.0172647. 86. Sohail, M.U., Elrayess, M.A., Al Thani, A.A., Al-Asmakh, M., and Yassine, H.M. (2019). Profiling the Oral Microbiome and Plasma Biochemistry of Obese Hyperglycemic Subjects in Qatar. Microorganisms 7, 645. https://doi.org/10.3390/microorganisms7120645. 87. Lin, S.W., Freedman, N.D., Shi, J., Gail, M.H., Vogtmann, E., Yu, G., Kle- pac-Ceraj, V., Paster, B.J., Dye, B.A., Wang, G.Q., et al. (2015). Beta-di- versity metrics of the upper digestive tract microbiome are associated with body mass index. Obesity 23, 862–869. https://doi.org/10.1002/ oby.21020. 88. Koliada, A., Syzenko, G., Moseiko, V., Budovska, L., Puchkov, K., Pere- deriy, V., Gavalko, Y., Dorofeyev, A., Romanenko, M., Tkach, S., et al. (2017). Association between body mass index and Firmicutes/ Bacteroidetes ratio in an adult Ukrainian population. BMC Microbiol. 17, 120. https://doi.org/10.1186/s12866-017-1027-1. 89. Sizova, M.V., Muller, P.A., Stancyk, D., Panikov, N.S., Mandalakis, M., Hazen, A., Hohmann, T., Doerfert, S.N., Fowle, W., Earl, A.M., et al. (2014). Oribacterium parvum sp. nov. and Oribacterium asaccharolyti- cum sp. nov., obligately anaerobic bacteria from the human oral cavity, and emended description of the genus Oribacterium. Int. J. Syst. Evol. Microbiol. 64, 2642–2649. https://doi.org/10.1099/ijs.0.060988-0. 90. Senel, S. (2021). An Overview of Physical, Microbiological and Immune Barriers of Oral Mucosa. Int. J. Mol. Sci. 22, 7821. https://doi.org/10. 3390/ijms22157821. 91. Li, X., Yang, Y., Zhang, B., Lin, X., Fu, X., An, Y., Zou, Y., Wang, J.X., Wang, Z., and Yu, T. (2022). Lactate metabolism in human health and dis- ease. Signal Transduct. Target. Ther. 7, 305. https://doi.org/10.1038/ s41392-022-01151-3. 92. Skonieczna-Zydecka, K. (2020). Gut Biofactory-Neurocompetent Me- tabolites within the Gastrointestinal Tract. A Scoping Review. Nutrients 12, 3369. https://doi.org/10.3390/nu12113369. 93. Tacconi, E., Palma, G., De Biase, D., Luciano, A., Barbieri, M., de Nigris, F., and Bruzzese, F. (2023). Microbiota Effect on Trimethylamine N-Oxide Production: From Cancer to Fitness-A Practical Preventing Recommen- dation and Therapies. Nutrients 15, 563. https://doi.org/10.3390/ nu15030563. 94. Kimura, K., Nakamura, Y., Inaba, Y., Matsumoto, M., Kido, Y., Asahara, S.I., Matsuda, T., Watanabe, H., Maeda, A., Inagaki, F., et al. (2013). His- tidine augments the suppression of hepatic glucose production by cen- tral insulin action. Diabetes 62, 2266–2277. https://doi.org/10.2337/ db12-1701. 95. Hasegawa, S., Ichiyama, T., Sonaka, I., Ohsaki, A., Okada, S., Wakigu- chi, H., Kudo, K., Kittaka, S., Hara, M., and Furukawa, S. (2012). Cysteine, histidine and glycine exhibit anti-inflammatory effects in human coronary arterial endothelial cells. Clin. Exp. Immunol. 167, 269–274. https://doi. org/10.1111/j.1365-2249.2011.04519.x. 96. Ratliff, J.C., Barber, J.A., Palmese, L.B., Reutenauer, E.L., and Tek, C. (2010). Association of prescription H1 antihistamine use with obesity: re- sults from the National Health and Nutrition Examination Survey. Obesity 18, 2398–2400. https://doi.org/10.1038/oby.2010.176. 97. Chen, M., Lu, B., Li, Y., Wang, Y., Zheng, H., Zhong, D., Liao, Z., Wang, M., Ma, F., Liao, Q., and Xie, Z. (2018). Metabolomics insights into the modulatory effects of long-term compound polysaccharide intake in high-fat diet-induced obese rats. Nutr. Metab. 15, 8. https://doi.org/10. 1186/s12986-018-0246-2. 98. Kim, E.K., Lee, K.A., Hyeon, D.Y., Kyung, M., Jun, K.Y., Seo, S.H., Hwang, D., Kwon, Y., and Lee, W.J. (2020). Bacterial Nucleoside Catab- olism Controls Quorum Sensing and Commensal-to-Pathogen Transition in the Drosophila Gut. Cell Host Microbe 27, 345–357.e6. https://doi.org/ 10.1016/j.chom.2020.01.025. 99. Montenegro, J., Armet, A.M., Willing, B.P., Deehan, E.C., Fassini, P.G., Mota, J.F., Walter, J., and Prado, C.M. (2023). Exploring the Influence of Gut Microbiome on Energy Metabolism in Humans. Adv. Nutr. 14, 840–857. https://doi.org/10.1016/j.advnut.2023.03.015. 100. Thomas-Valdes, S., Tostes, M., Anunciacao, P.C., da Silva, B.P., and Sant’Ana, H.M.P. (2017). Association between vitamin deficiency and metabolic disorders related to obesity. Crit. Rev. Food Sci. Nutr. 57, 3332–3343. https://doi.org/10.1080/10408398.2015.1117413. 101. Sun, Y., Sun, M., Liu, B., Du, Y., Rong, S., Xu, G., Snetselaar, L.G., and Bao, W. (2019). Inverse Association Between Serum Vitamin B12 Con- centration and Obesity Among Adults in the United States. Front. Endo- crinol. 10, 414. https://doi.org/10.3389/fendo.2019.00414. 102. Guetterman, H.M., Huey, S.L., Knight, R., Fox, A.M., Mehta, S., and Fin- kelstein, J.L. (2022). Vitamin B-12 and the Gastrointestinal Microbiome: A Systematic Review. Adv. Nutr. 13, 530–558. https://doi.org/10.1093/ad- vances/nmab123. 103. Turnbaugh, P.J., Hamady, M., Yatsunenko, T., Cantarel, B.L., Duncan, A., Ley, R.E., Sogin, M.L., Jones, W.J., Roe, B.A., Affourtit, J.P., et al. (2009). A core gut microbiome in obese and lean twins. Nature 457, 480–484. https://doi.org/10.1038/nature07540. 104. Karlsson, F.H., Tremaroli, V., Nookaew, I., Bergström, G., Behre, C.J., Fagerberg, B., Nielsen, J., and Bäckhed, F. (2013). Gut metagenome in European women with normal, impaired and diabetic glucose control. Nature 498, 99–103. https://doi.org/10.1038/nature12198. 105. Le Chatelier, E., Nielsen, T., Qin, J., Prifti, E., Hildebrand, F., Falony, G., Almeida, M., Arumugam, M., Batto, J.M., Kennedy, S., et al. (2013). Rich- ness of human gut microbiome correlates with metabolic markers. Na- ture 500, 541–546. https://doi.org/10.1038/nature12506. 106. Carter, M.M., Olm, M.R., Merrill, B.D., Dahan, D., Tripathi, S., Spencer, S.P., Yu, F.B., Jain, S., Neff, N., Jha, A.R., et al. (2023). Ultra-deep sequencing of Hadza hunter-gatherers recovers vanishing gut microbes. Cell 186, 3111–3124.e13. https://doi.org/10.1016/j.cell.2023.05.046. 107. Olm, M.R., Dahan, D., Carter, M.M., Merrill, B.D., Yu, F.B., Jain, S., Meng, X., Tripathi, S., Wastyk, H., Neff, N., et al. (2022). Robust variation in infant gut microbiome assembly across a spectrum of lifestyles. Science 376, 1220–1223. https://doi.org/10.1126/science.abj2972. 108. Blakeley, M., Sobczynska-Malefora, A., and Carpenter, G. (2020). The Origins of Salivary Vitamin A, Vitamin B(12) and Vitamin D-Binding Pro- teins. Nutrients 12, 3838. https://doi.org/10.3390/nu12123838. 109. Li, Y., Qian, F., Cheng, X., Wang, D., Wang, Y., Pan, Y., Chen, L., Wang, W., and Tian, Y. (2023). Dysbiosis of Oral Microbiota and Metabolite Pro- files Associated with Type 2 Diabetes Mellitus. Microbiol. Spectr. 11, e0379622. https://doi.org/10.1128/spectrum.03796-22. 110. Aleti, G., Kohn, J.N., Troyer, E.A., Weldon, K., Huang, S., Tripathi, A., Dor- restein, P.C., Swafford, A.D., Knight, R., and Hong, S. (2022). Salivary bacterial signatures in depression-obesity comorbidity are associated with neurotransmitters and neuroactive dipeptides. BMC Microbiol. 22, 75. https://doi.org/10.1186/s12866-022-02483-4. 111. B. Callahan, ed. (2024). (Zenodo). 112. Wright, E.S. (2015). DECIPHER: harnessing local sequence context to improve protein multiple sequence alignment. BMC Bioinf. 16, 322. https://doi.org/10.1186/s12859-015-0749-z. 113. Schliep, K.P. (2011). phangorn: phylogenetic analysis in R. Bioinformat- ics 27, 592–593. https://doi.org/10.1093/bioinformatics/btq706. 114. McMurdie, P.J., and Holmes, S. (2013). phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One 8, e61217. https://doi.org/10.1371/journal.pone. 0061217. 115. Callahan, B.J., McMurdie, P.J., Rosen, M.J., Han, A.W., Johnson, A.J.A., and Holmes, S.P. (2016). DADA2: High-resolution sample inference from Illumina amplicon data. Nat. Methods 13, 581–583. https://doi.org/10. 1038/nmeth.3869. 116. Ho, D.E., Imai, K., King, G., and Stuart, E.A. (2011). MatchIt: Nonpara- metric Preprocessing for Parametric Causal Inference. J. Stat. Softw. 42, 1–28. https://doi.org/10.18637/jss.v042.i08. Cell Reports 45, 116819, February 24, 2026 17 Article ll OPEN ACCESS https://doi.org/10.1371/journal.pone.0172647 https://doi.org/10.3390/microorganisms7120645 https://doi.org/10.1002/oby.21020 https://doi.org/10.1002/oby.21020 https://doi.org/10.1186/s12866-017-1027-1 https://doi.org/10.1099/ijs.0.060988-0 https://doi.org/10.3390/ijms22157821 https://doi.org/10.3390/ijms22157821 https://doi.org/10.1038/s41392-022-01151-3 https://doi.org/10.1038/s41392-022-01151-3 https://doi.org/10.3390/nu12113369 https://doi.org/10.3390/nu15030563 https://doi.org/10.3390/nu15030563 https://doi.org/10.2337/db12-1701 https://doi.org/10.2337/db12-1701 https://doi.org/10.1111/j.1365-2249.2011.04519.x https://doi.org/10.1111/j.1365-2249.2011.04519.x https://doi.org/10.1038/oby.2010.176 https://doi.org/10.1186/s12986-018-0246-2 https://doi.org/10.1186/s12986-018-0246-2 https://doi.org/10.1016/j.chom.2020.01.025 https://doi.org/10.1016/j.chom.2020.01.025 https://doi.org/10.1016/j.advnut.2023.03.015 https://doi.org/10.1080/10408398.2015.1117413 https://doi.org/10.3389/fendo.2019.00414 https://doi.org/10.1093/advances/nmab123 https://doi.org/10.1093/advances/nmab123 https://doi.org/10.1038/nature07540 https://doi.org/10.1038/nature12198 https://doi.org/10.1038/nature12506 https://doi.org/10.1016/j.cell.2023.05.046 https://doi.org/10.1126/science.abj2972 https://doi.org/10.3390/nu12123838 https://doi.org/10.1128/spectrum.03796-22 https://doi.org/10.1186/s12866-022-02483-4 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref111 https://doi.org/10.1186/s12859-015-0749-z https://doi.org/10.1093/bioinformatics/btq706 https://doi.org/10.1371/journal.pone.0061217 https://doi.org/10.1371/journal.pone.0061217 https://doi.org/10.1038/nmeth.3869 https://doi.org/10.1038/nmeth.3869 https://doi.org/10.18637/jss.v042.i08 117. Ritchie, M.E., Phipson, B., Wu, D., Hu, Y., Law, C.W., Shi, W., and Smyth, G.K. (2015). limma powers differential expression analyses for RNA- sequencing and microarray studies. Nucleic Acids Res. 43, e47. https://doi.org/10.1093/nar/gkv007. 118. Oksanen, J., Simpson, G., Blanchet, F., Kindt, R., Legendre, P., Minchin, P., O’Hara, R., Solymos, P., Stevens, M., Szoecs, E., et al. (2025). vegan: Community Ecology Package (R package version 2), pp. 6–10. 119. Kuhn, M. (2008). Building Predictive Models in R Using the caret Pack- age. J. Stat. Softw. 28, 1–26. https://doi.org/10.18637/jss.v028.i05. 120. Sing, T., Sander, O., Beerenwinkel, N., and Lengauer, T. (2005). ROCR: visualizing classifier performance in R. Bioinformatics 21, 3940–3941. https://doi.org/10.1093/bioinformatics/bti623. 121. Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis (Springer-Verlag). 122. Csardi, G., and Nepusz, T. (2006). The igraph software. Complex Syst. 1695, 1–9. 123. Bolger, A.M., Lohse, M., and Usadel, B. (2014). Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114–2120. https://doi.org/10.1093/bioinformatics/btu170. 124. Chen, S., Zhou, Y., Chen, Y., and Gu, J. (2018). fastp: an ultra-fast all-in- one FASTQ preprocessor. Bioinformatics 34, i884–i890. https://doi.org/ 10.1093/bioinformatics/bty560. 125. FastQC: A quality control tool for high throughput sequence data (Avail- able online: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/, Babraham Bioinformatics, 2010). 126. Ewels, P., Magnusson, M., Lundin, S., and Käller, M. (2016). MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32, 3047–3048. https://doi.org/10.1093/bioinfor- matics/btw354. 127. Uritskiy, G.V., DiRuggiero, J., and Taylor, J. (2018). MetaWRAP-a flexible pipeline for genome-resolved metagenomic data analysis. Microbiome 6, 158. https://doi.org/10.1186/s40168-018-0541-1. 128. Beghini, F., McIver, L.J., Blanco-Mı́guez, A., Dubois, L., Asnicar, F., Ma- harjan, S., Mailyan, A., Manghi, P., Scholz, M., Thomas, A.M., et al. (2021). Integrating taxonomic, functional, and strain-level profiling of diverse microbial communities with bioBakery 3. eLife 10, e65088. https://doi.org/10.7554/eLife.65088. 129. Li, D., Luo, R., Liu, C.M., Leung, C.M., Ting, H.F., Sadakane, K., Yama- shita, H., and Lam, T.W. (2016). MEGAHIT v1.0: A fast and scalable meta- genome assembler driven by advanced methodologies and community practices. Methods 102, 3–11. https://doi.org/10.1016/j.ymeth.2016. 02.020. 130. Gurevich, A., Saveliev, V., Vyahhi, N., and Tesler, G. (2013). QUAST: qual- ity assessment tool for genome assemblies. Bioinformatics 29, 1072– 1075. https://doi.org/10.1093/bioinformatics/btt086. 131. Alneberg, J., Bjarnason, B.S., de Bruijn, I., Schirmer, M., Quick, J., Ijaz, U.Z., Lahti, L., Loman, N.J., Andersson, A.F., and Quince, C. (2014). Binning metagenomic contigs by coverage and composition. Nat. Methods 11, 1144–1146. https://doi.org/10.1038/nmeth.3103. 132. Wu, Y.W., Simmons, B.A., and Singer, S.W. (2016). MaxBin 2.0: an auto- mated binning algorithm to recover genomes from multiple metagenomic datasets. Bioinformatics 32, 605–607. https://doi.org/10.1093/bioinfor- matics/btv638. 133. Kang, D.D., Li, F., Kirton, E., Thomas, A., Egan, R., An, H., and Wang, Z. (2019). MetaBAT 2: an adaptive binning algorithm for robust and efficient genome reconstruction from metagenome assemblies. PeerJ 7, e7359. https://doi.org/10.7717/peerj.7359. 134. Schwengers, O., Jelonek, L., Dieckmann, M.A., Beyvers, S., Blom, J., and Goesmann, A. (2021). Bakta: rapid and standardized annotation of bacterial genomes via alignment-free sequence identification. Microb. Genom. 7, 000685. https://doi.org/10.1099/mgen.0.000685. 135. Jain, C., Rodriguez-R, L.M., Phillippy, A.M., Konstantinidis, K.T., and Aluru, S. (2018). High throughput ANI analysis of 90K prokaryotic ge- nomes reveals clear species boundaries. Nat. Commun. 9, 5114. https://doi.org/10.1038/s41467-018-07641-9. 136. Lee, M.D. (2019). GToTree: a user-friendly workflow for phylogenomics. Bioinformatics 35, 4162–4164. https://doi.org/10.1093/bioinformatics/ btz188. 137. Price, M.N., Dehal, P.S., and Arkin, A.P. (2010). FastTree 2-- approximately maximum-likelihood trees for large alignments. PLoS One 5, e9490. https://doi.org/10.1371/journal.pone.0009490. 138. Letunic, I., and Bork, P. (2016). Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nu- cleic Acids Res. 44, W242–W245. https://doi.org/10.1093/nar/gkw290. 139. Shaffer, M., Borton, M.A., McGivern, B.B., Zayed, A.A., La Rosa, S.L., Solden, L.M., Liu, P., Narrowe, A.B., Rodrı́guez-Ramos, J., Bolduc, B., et al. (2020). DRAM for distilling microbial metabolism to automate the curation of microbiome function. Nucleic Acids Res. 48, 8883–8900. https://doi.org/10.1093/nar/gkaa621. 140. Eren, A.M., Kiefl, E., Shaiber, A., Veseli, I., Miller, S.E., Schechter, M.S., Fink, I., Pan, J.N., Yousef, M., Fogarty, E.C., et al. (2021). Community- led, integrated, reproducible multi-omics with anvi’o. Nat. Microbiol. 6, 3–6. https://doi.org/10.1038/s41564-020-00834-3. 141. Pang, Z., Lu, Y., Zhou, G., Hui, F., Xu, L., Viau, C., Spigelman, A.F., Mac- Donald, P.E., Wishart, D.S., Li, S., and Xia, J. (2024). MetaboAnalyst 6.0: towards a unified platform for metabolomics data processing, analysis and interpretation. Nucleic Acids Res. 52, W398–W406. https://doi.org/ 10.1093/nar/gkae253. 142. Fan, X., Peters, B.A., Min, D., Ahn, J., and Hayes, R.B. (2018). Compar- ison of the oral microbiome in mouthwash and whole saliva samples. PLoS One 13, e0194729. https://doi.org/10.1371/journal.pone.0194729. 143. Walters, W., Hyde, E.R., Berg-Lyons, D., Ackermann, G., Humphrey, G., Parada, A., Gilbert, J.A., Jansson, J.K., Caporaso, J.G., Fuhrman, J.A., et al. (2016). Improved Bacterial 16S rRNA Gene (V4 and V4-5) and Fungal Internal Transcribed Spacer Marker Gene Primers for Microbial Community Surveys. mSystems 1, e00009-15. https://doi.org/10.1128/ mSystems.00009-15. 144. Jha, A.R., Davenport, E.R., Gautam, Y., Bhandari, D., Tandukar, S., Ng, K.M., Fragiadakis, G.K., Holmes, S., Gautam, G.P., Leach, J., et al. (2018). Gut microbiome transition across a lifestyle gradient in Himalaya. PLoS Biol. 16, e2005396. https://doi.org/10.1371/journal.pbio.2005396. 145. Wang, Q., Garrity, G.M., Tiedje, J.M., and Cole, J.R. (2007). Naive Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ. Microbiol. 73, 5261–5267. https://doi.org/10.1128/AEM.00062-07. 146. Mohd Kamal, K., Mahamad Maifiah, M.H., Abdul Rahim, N., Hashim, Y.Z.H.Y., Abdullah Sani, M.S., and Azizan, K.A. (2022). Bacterial Metab- olomics: Sample Preparation Methods. Biochem. Res. Int. 2022, 9186536. https://doi.org/10.1155/2022/9186536. 147. Gardner, A., Parkes, H.G., So, P.W., and Carpenter, G.H. (2019). Deter- mining bacterial and host contributions to the human salivary metabo- lome. J. Oral Microbiol. 11, 1617014. https://doi.org/10.1080/20002297. 2019.1617014. 148. Ciurli, A., Mohammed, Y., Ammon, C., Derks, R.J.E., Olivier-Jimenez, D., Ducarmon, Q.R., Slingerland, M., Neefjes, J., and Giera, M. (2024). Spatially and temporally resolved metabolome of the human oral cavity. iScience 27, 108884. https://doi.org/10.1016/j.isci.2024.108884. 149. Sumner, L.W., Amberg, A., Barrett, D., Beale, M.H., Beger, R., Daykin, C.A., Fan, T.W.M., Fiehn, O., Goodacre, R., Griffin, J.L., et al. (2007). Pro- posed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics 3, 211–221. https://doi.org/10.1007/s11306-007- 0082-2. 18 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS https://doi.org/10.1093/nar/gkv007 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref118 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref118 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref118 https://doi.org/10.18637/jss.v028.i05 https://doi.org/10.1093/bioinformatics/bti623 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref121 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref121 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref122 http://refhub.elsevier.com/S2211-1247(25)01591-8/sref122 https://doi.org/10.1093/bioinformatics/btu170 https://doi.org/10.1093/bioinformatics/bty560 https://doi.org/10.1093/bioinformatics/bty560 http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ https://doi.org/10.1093/bioinformatics/btw354 https://doi.org/10.1093/bioinformatics/btw354 https://doi.org/10.1186/s40168-018-0541-1 https://doi.org/10.7554/eLife.65088 https://doi.org/10.1016/j.ymeth.2016.02.020 https://doi.org/10.1016/j.ymeth.2016.02.020 https://doi.org/10.1093/bioinformatics/btt086 https://doi.org/10.1038/nmeth.3103 https://doi.org/10.1093/bioinformatics/btv638 https://doi.org/10.1093/bioinformatics/btv638 https://doi.org/10.7717/peerj.7359 https://doi.org/10.1099/mgen.0.000685 https://doi.org/10.1038/s41467-018-07641-9 https://doi.org/10.1093/bioinformatics/btz188 https://doi.org/10.1093/bioinformatics/btz188 https://doi.org/10.1371/journal.pone.0009490 https://doi.org/10.1093/nar/gkw290 https://doi.org/10.1093/nar/gkaa621 https://doi.org/10.1038/s41564-020-00834-3 https://doi.org/10.1093/nar/gkae253 https://doi.org/10.1093/nar/gkae253 https://doi.org/10.1371/journal.pone.0194729 https://doi.org/10.1128/mSystems.00009-15 https://doi.org/10.1128/mSystems.00009-15 https://doi.org/10.1371/journal.pbio.2005396 https://doi.org/10.1128/AEM.00062-07 https://doi.org/10.1155/2022/9186536 https://doi.org/10.1080/20002297.2019.1617014 https://doi.org/10.1080/20002297.2019.1617014 https://doi.org/10.1016/j.isci.2024.108884 https://doi.org/10.1007/s11306-007-0082-2 https://doi.org/10.1007/s11306-007-0082-2 STAR★METHODS KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Biological samples Human oral moutwash samples This paper NA Human blood samples This paper NA Human urine samples This paper NA Chemicals, peptides, and recombinant proteins Acetonitrile Honeywell Cat#34967 RapidVap N2 Labconco Cat#7910013 Methanol Sigma-Aldrich Cat#1.06035 Formic acid Sigma-Aldrich Cat#5.43804 Ammonium formate Sigma-Aldrich Cat#516961 Critical commercial assays Qiagen DNeasy Powersoil Pro Kit Qiagen Cat#47014 AMPure XP beads Beckman Coulter Cat#A63881 Qubit BR kit Thermo Scientific Cat#Q33266 Illumina DNA library preparation kit Illumina Cat#20018705 Deposited data Shotgun metagenomics reads This paper dbGaP: phs004453.v1.p1 16S amplicon sequencing reads This paper dbGaP: phs004453.v1.p1 Metabolite abundance table This paper dbGaP: phs004453.v1.p1 Clinical markers table This paper dbGaP: phs004453.v1.p1 Oligonucleotides 16S V4-V5 515F: GTGYCAGCMGCCGCGGTAA Integrated DNA Technologies https://doi.org/10.1128/msystems.00009-15 16S V4-V5 926R: CCGYCAATTYMTTTRAGTTT Integrated DNA Technologies https://doi.org/10.1128/msystems.00009-15 Software and algorithms R (v4.2.0) R Core Team https://www.r-project.org/ RStudio 2022.07.02 Build 576 https://www.rstudio.com/ SILVA v138 DADA2-formatted training fasta Callahan et al.111 https://zenodo.org/records/14169026 ‘‘DECIPHER’’ R package Wright et al.112 https://rdrr.io/bioc/DECIPHER/ ‘‘Phangorn’’ R package Schliep et al.113 https://rdrr.io/bioc/phyloseq/ ‘‘phyloseq’’ R package McMurdie & Holmes114 https://rdrr.io/bioc/dada2/ ‘‘dada2’’ R package Callahan et al.115 https://rdrr.io/bioc/dada2/ ‘‘MatchIt’’ R package Ho et al.116 https://rdrr.io/github/itpir/ MatchIt/man/matchit.html ‘‘limma’’ R package Ritchie et al.117 https://rdrr.io/bioc/limma/ ‘‘vegan’’ R package Oksanen et al.118 https://rdrr.io/cran/vegan/ ‘‘caret’’ R package Kuhn et al.119 https://rdrr.io/rforge/caret/ ‘‘ROCR’’ R package Sing et al.120 https://rdrr.io/cran/ROCR/ ‘‘ggplot2’’ R package Wickam et al.121 https://rdrr.io/cran/ggplot2/ ‘‘Maaslin2’’ R package Mallick et al.34 https://rdrr.io/bioc/Maaslin2/ ‘‘igraph’’ R package Csardi et al.122 https://rdrr.io/cran/igraph/ bcl2fastq2 v2.20.0.422 Illumina NA Trimmomatic v0.36 Bolger et al.123 https://github.com/usadellab/Trimmomatic fastp Chen et al.124 https://github.com/OpenGene/fastp FastQC Andrews125 https://github.com/s-andrews/FastQC MultiQC Ewels et al.126 https://github.com/MultiQC/MultiQC (Continued on next page) Cell Reports 45, 116819, February 24, 2026 19 Article ll OPEN ACCESS https://doi.org/10.1128/msystems.00009-15 https://doi.org/10.1128/msystems.00009-15 https://www.r-project.org/ https://www.rstudio.com/ https://zenodo.org/records/14169026 https://rdrr.io/bioc/DECIPHER/ https://rdrr.io/bioc/phyloseq/ https://rdrr.io/bioc/dada2/ https://rdrr.io/bioc/dada2/ https://rdrr.io/github/itpir/MatchIt/man/matchit.html https://rdrr.io/github/itpir/MatchIt/man/matchit.html https://rdrr.io/bioc/limma/ https://rdrr.io/cran/vegan/ https://rdrr.io/rforge/caret/ https://rdrr.io/cran/ROCR/ https://rdrr.io/cran/ggplot2/ https://rdrr.io/bioc/Maaslin2/ https://rdrr.io/cran/igraph/ https://github.com/usadellab/Trimmomatic https://github.com/OpenGene/fastp https://github.com/s-andrews/FastQC https://github.com/MultiQC/MultiQC EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Study design and mouthwash sample collection The UAEHFS is a prospective cohort study designed to investigate the biological mechanisms underlying chronic diseases in Emirati nationals. Mouthwash samples were collected from baseline participants with written consent between 2016 and 2020, as described previously in detail.24 Participants refrained from eating at least 1 h prior to sample collection. The postprandial interval ranged be- tween 1 and 8 h and is shown in Table S1. Briefly, participants swished 10 mL of 0.9% sterile saline vigorously for 30 s and spit into sterile barcoded tubes. These tubes were immediately transported to the lab and stored at − 80◦C until analysis. We conducted a pilot study to compare the oral microbiome from saliva versus mouthwash samples collected using this protocol (Figure S12). Our results show that mouthwash is as reliable as saliva in capturing the oral microbiome, which is consistent with previous studies.142 A total of 669 moutwash samples were included in this study. In addition to providing mouthwash samples, participants underwent physical and clinical exams, which included anthropometric measurements, blood pressure assessments, blood and urine tests, and self-completed a demographic and lifestyle questionnaire. Data from these exams and questionnaires are summarized in Table S1. Clinical parameters were used to define prevalent risk con- ditions within the cohort based on established criteria (Table S15). Participants were categorized based on body mass index (BMI) as underweight (BMI <18.5), healthy weight (18.5 ≤ BMI ≤24.9), overweight (25 ≤ BMI ≤29.9), or obese (BMI ≥30) categories. Ethics statement The UAE Healthy Future Study (UAEHFS) was conducted according to the guidelines of the Declaration of Helsinki, and the study protocol was approved by the Institutional Review Board of Abu Dhabi Department of Health, reference number DOH/HQD/2020/ 516 and NYUAD’S Research Ethics Committee (REC, ref. 0072017R). All participants read and understood the information leaflet and signed the consent form prior to their recruitment. METHOD DETAILS Blood and urine biomarker measurements Participants provided non-fasting blood and urine samples, which were analyzed for routine chemistry using the Beckman DXC600 (Beckman Coulter, USA) as described previously.24 This provides us with biomarkers indicative of lipids, glycemic levels, liver and kidney function, serum electrolytes, and red blood cell indices. Continued REAGENT or RESOURCE SOURCE IDENTIFIER metaWRAP Uritskiy et al.127 https://github.com/bxlab/metaWRAP MetaPhlAn v4.0.4 Blanco-Miguez et al.33 https://github.com/biobakery/MetaPhlAn HUMAnN v3.5 Beghini et al.128 https://github.com/biobakery/humann MEGAHIT v1.2.9 Li et al.129 https://github.com/voutcn/megahit QUAST v5.0.2 Gurevich et al.130 https://github.com/ablab/quast CONCOCT Alneberg et al.131 https://github.com/BinPro/CONCOCT MaxBin2 Wu et al.132 https://sourceforge.net/projects/maxbin/ MetaBAT2 Kang et al.133 https://bitbucket.org/berkeleylab/metabat/src CheckM v1.2.2 Parks et al.70 https://github.com/Ecogenomics/CheckM GTDB-Tk v2.4.1 Chaumeil et al.71 https://github.com/Ecogenomics/GTDBTk Bakta v1.6.1 Schwengers et al.134 https://github.com/oschwengers/bakta FastANI v1.32 Jain et al.135 https://github.com/ParBLiSS/FastANI GToTree v1.7.06 Lee136 https://github.com/AstrobioMike/GToTree FastTree v2.1.11 Price et al.137 https://github.com/morgannprice/fasttree iTOL Letunic and Bork138 https://itol.embl.de/ DRAM v1.4.6 Shaffer et al.139 https://github.com/WrightonLabCSU/DRAM Anvi’o Eren et al.140 https://anvio.org/ CompoundDiscoverer v3.3 Thermo Fisher Scientific https://www.thermofisher.com/order/ catalog/product/OPTON-31061 Simca P16 Sartorius Stedim Data Analytics https://www.sartorius.com/en/products/ process-analytical-technology/data- analytics-software/mvda-software/simca MetaboAnalyst Pang et al.141 https://www.metaboanalyst.ca/ 20 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS https://github.com/bxlab/metaWRAP https://github.com/biobakery/MetaPhlAn https://github.com/biobakery/humann https://github.com/voutcn/megahit https://github.com/ablab/quast https://github.com/BinPro/CONCOCT https://sourceforge.net/projects/maxbin/ https://bitbucket.org/berkeleylab/metabat/src https://github.com/Ecogenomics/CheckM https://github.com/Ecogenomics/GTDBTk https://github.com/oschwengers/bakta https://github.com/ParBLiSS/FastANI https://github.com/AstrobioMike/GToTree https://github.com/morgannprice/fasttree https://itol.embl.de/ https://github.com/WrightonLabCSU/DRAM https://anvio.org/ https://www.thermofisher.com/order/catalog/product/OPTON-31061 https://www.thermofisher.com/order/catalog/product/OPTON-31061 https://www.sartorius.com/en/products/process-analytical-technology/data-analytics-software/mvda-software/simca https://www.sartorius.com/en/products/process-analytical-technology/data-analytics-software/mvda-software/simca https://www.sartorius.com/en/products/process-analytical-technology/data-analytics-software/mvda-software/simca https://www.metaboanalyst.ca/ DNA extraction The mouthwash samples underwent mechanical lysis using a Bead Ruptor 96 (Omni International, USA) followed by DNA extraction using the QIAGEN DNeasy Powersoil Pro Kit. To minimize cross-contamination, DNA extraction was performed in individual tubes, with a negative control, containing only the extraction reagents but no mouthwash, included after every 4–5 samples. This process yielded 669 mouthwash samples and 134 controls. DNA was eluted in nuclease-free water, and the concentration was measured using NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). The DNA extraction protocol used in this study is available at DOI: https://doi.org/10.17504/protocols.io.eq2ly7dyrlx9/v1. 16S rRNA gene sequencing DNA amplification targeting the V4-V5 region of the 16S rRNA gene was performed on all samples and controls using the universal primer set 515F/926R.143 Forward primers were barcode-tagged for sample multiplexing. PCR products were visualized on an agarose gel, purified using AMPure beads (Beckman, USA), and quantified with the Qubit BR kit (Thermo Scientific, USA). No PCR bands were observed in any of the controls. Thus, these were excluded from further steps. The PCR products from the samples were normalized, pooled, and sequenced on an Illumina MiSeq to generate 2 × 300 paired-end reads. 16S rRNA amplicon processing A total of 8,850,022 paired-end reads were obtained from 669 samples. Quality assessment of the sequenced reads using FastQC revealed poor quality in the reverse reads, preventing their merge with the forward reads. Consequently, only forward reads were used for further analyses. Forward reads were processed with an in-house pipeline based on a previous study,144 which implements DADA2115 to learn error rates, dereplicate amplicons, remove chimeras, identify amplicon sequence variants (ASVs), and construct a sequence table. The reads were trimmed to 150 bp to retain high quality sequences (Phred score >30). Reads containing N nucle- otides or >2 expected errors were discarded (maxN = 0, maxEE = 2, truncQ = 2). ASVs were inferred from 5,936,465 high-quality reads, representing 67% of the initial dataset. Taxonomy was assigned using the RDP classifier145 utilizing the SILVA v138 training set111 as the reference. Multiple sequence alignment was conducted with DECIPHER,112 and a maximum likelihood tree was built using the neighbor-joining method in phangorn.113 Subsequent analyses were conducted in R v4.2.0 using the phyloseq package.114 ASVs present in less than 5% of the samples were removed, leaving 5,851,506 reads that were assigned to 384 taxa. Three ASVs were identified as Eukaryota and one was unresolved at the phylum level. These four ASVs were excluded, resulting in 5,712,557 reads across 380 ASVs for subsequent analyses. Identification of matched cohort and lifestyle and health-associated variable selection Out of the 669 participants in our study, we identified 97 individuals with obesity (BMI≥30) and established a comparison group of 95 healthy weight controls (18.5≤BMI≤24.9). To ensure the comparability of these groups, we first conducted a linear regression anal- ysis to identify variables significantly associated with BMI. We removed highly multicollinear variables from this regression (VIF>10). We then applied propensity score matching (using MatchIt v4.5.4116), matching for sex, age, and other variables significantly asso- ciated with oral health. This approach ensured balanced distribution of potential confounding variables that could influence the oral microbiome, such as postprandial interval, brushing frequency, and smoking frequency across both case and control groups as shown in Table S3. Blood and urine marker analysis Principal component analysis (PCA) was performed on a set of 49 non-redundant blood and urine biomarkers, selected based on the availability of complete data across all 191 individuals in the matched cohort. PCA was conducted using the prcomp function in R. In an initial PCA, sex showed the strongest association with biomarker variation (R2 = 0.42, P = 1 × 10− 4, EnvFit); therefore, we re- gressed out the effect of sex using the limma117 package and conducted all downstream analyses on the residualized biomarker data. Associations between metadata and biomarker composition were assessed using the envfit function (Figure 2B). To evaluate the discriminatory power of biomarkers between control and obese individuals, a random forest classifier was trained on the residual- ized data (80% training and 20% testing set), and fit using 10-fold cross validation repeated three times using 500 trees. Performance was assessed using area under the receiver operating characteristic curve (AUC). Predictors with high variance inflation factors (VIF) were identified and retrieved from the model. Metagenomics library preparation and sequencing We performed shotgun whole metagenomics sequencing for the 97 obese and 95 matched controls (n = 192) at New York University Abu Dhabi Core Technology Platform Sequencing Center (Abu Dhabi, UAE) using Illumina DNA Library Preparation kit (Illumina, USA), following the manufacturer’s instructions using 300–450 ng of DNA in 30 μL volume. During the library preparation, a unique 10 base pair (bp) dual-indexed barcode was added to each sample. Sequencing libraries were size selected using AMPure XP beads (Beck- man, USA) targeting a fragment length of 450 bp with insert size of 350 bp and checked using the Agilent 4200 TapeStation system. Paired-end sequencing (2 × 150 bp) was conducted on an Illumina NovaSeq6000 S4 flow cell. We obtained an average sequencing depth of 31,823,102 paired-end reads per sample. One sample producing low reads was excluded from subsequent analysis. Cell Reports 45, 116819, February 24, 2026 21 Article ll OPEN ACCESS https://doi.org/10.17504/protocols.io.eq2ly7dyrlx9/v1 Metagenomic processing, taxonomic profiling, and functional characterization A complete workflow of the metagenomics pipeline used in this study is depicted in Figure S13. Paired-end libraries were demulti- plexed using bcl2fastq2 v2.20.0.422 (Illumina), allowing up to one mismatch in the index barcode sequence (–barcode-mismatches 1). The resulting raw FASTQ reads were quality trimmed with Trimmomatic123 v0.36 and processed with fastp124 in order to remove poly-G tails. After quality trimming, reads were assessed for quality using FastQC125 and a consolidated report was produced using MultiQC.126 High quality reads were mapped to the human genome (hg38) with metaWRAP127 to remove human derived sequences. Quality control information for the samples is shown in Table S4. Taxonomic profiling was conducted using MetaPhlAn v4.0.433 to generate relative abundances of microbial species identified in each sample. Functional characterization of the metagenomic reads was performed using HUMAnN 3.5128 using the ChocoPhlAn and UniRef90 databases (release 201901b). Reads per gene families were normalized for alignment quality and length using copies per million (CPM) units. Metagenomic assembly and functional analyses MEGAHIT v1.2.9129 was used for independent de novo assembly of metagenomic samples and assembly quality was assessed with QUAST v5.0.2.130 Genome bins were initially constructed from resulting scaffolds with >1,000 base pairs using three binning tools; CONCOCT,131 MaxBin2132 and metaBAT2.133 Subsequently, bins were refined and reassembled into metagenomically assembled genomes (MAGs) with metaWRAP.127 CheckM v1.2.270 used to assess the quality of MAGs by estimating several criteria established by MIMAG,69 including completion, contamination, and presence of various rRNA and tRNA genes. Final MAGs fulfilling the criteria of medium-quality (≥50% completeness and ≤10% contamination) bins and other MIMAG standards were used for subsequent ana- lyses. Taxonomic classification of all MAGs was assigned using GTDB-Tk v2.4.1 (GTDB release R226).71 Open reading frame pre- diction and functional annotation of microbial genes was performed with Bakta v1.6.1,134 which resulted in 2,447,523 complete genes. MAGs were dereplicated using FastANI v1.32135 clustering with a cutoff of≥95% Average Nucleotide Identity (ANI), resulting in 221 unique MAGs. MAGs were considered putatively novel if they lacked annotations in GTDB R226, UHGG v2, WIS, and CGMR data- bases, increasing our confidence in the novelty of bacterial species identified in our dataset. A total of 41 dereplicated MAGs showed no species-level classification in GTDB and no clustered representatives in UHGG v2, WIS, and CGMR databases and were consid- ered novel species. The 41 novel MAGs along with reference members of the same taxa in GTDB, were selected to generate align- ments of their single-copy gene (SCGs) using hidden Markov model (HMM) profiles on GToTree v1.7.06.136 Phylogenomic trees were inferred with FastTree v2.1.11137 and visualized on iToL.138 Characteristics of MAGs discussed in this study are presented in Table S12. To better understand the metabolic capacity of the novel MAGs, DRAM v1.4.6139 was used to curate gene annotations and infer functional categories. Pangenomics analysis and identification of enriched functional pathways within the 221 dereplicated MAGs were performed using Anvi’o140 by annotating all protein coding genes using Clusters of Orthologous Genes release 202076 and the Kyoto Encyclopedia of Genes and Genomes (KEGG) modules and classes.75 Untargeted metabolomics using LC-MS and LC-MS/MS Sample Preparation: A volume of 1 mL from the mouthwash samples (n = 192) was added to an equal volume of acetonitrile (1:1 sample:acetonitrile). The sample were ultrasonicated for 5 min, centrifuged at 13,000 g, 4◦C for 10 min and the supernatant (∼800 μL) was collected and dried under nitrogen flow (RapidVap N2, Labconco, Kansas City, MO, USA). The dried extract of the samples were reconstituted in 100 μL of 1:1 methanol:water. Blank samples (n = 9) were included and treated in an identical manner as the samples. The reconstituted samples were stored at − 80◦C until analyzed. A pooled quality control (QC) sample was prepared by pooling equal volumes (20 μL) from each sample in the study, vortexed for 30 s and centrifuged at 13,000 g, 4◦C for five minutes. This protocol closely follows the methods described previously.146–148 Analytical method: Untargeted metabolomics was employed using LC-MS to investigate the microbial metabolic profiles in the oral microbiome samples. Chromatography was performed using Vanquish UHPLC system (Thermo Fisher Scientific, Waltham, MA, USA) on a Zorbax RRHD HILIC Plus column (2.1 × 100 mm, 1.8 μm particle size, 95 Å, Agilent, Santa Clara, CA, USA) maintained at 40◦C. Mobile phases used were: (A) 0.1% formic acid, 10 mM ammonium formate in water and (B) 0.1% formic acid in acetonitrile. The gradient used was as follows: 0–10.5 min (5–60% A, 200 μL/min), 10.5–15 min (at 60% A, 200 μL/min), 15–17 min (60–5% A, 200 μL/min), 17–18 min (5% A, 200 to 300 μL/min), 18–31.50 min (to equilibrate at 5% A, 300 μL/min) and 31.5–32 min (5% A, 300 to 200 μL/min). The injection volume was 5 μL and the samples were maintained at 4◦C during the analysis. A Tribrid Orbitrap Mass Spectrometer (Fusion Lumos, Thermo Fisher Scientific, Waltham, MA, USA) was used in switching ESI+ and ESI- modes for full LC-MS profiling and to generate data dependent MS/MS (ddMS/MS) accurate mass spectra for identification (most intense peaks). The operational parameters for both ESI modes were: spray voltage 3.5 kV, sheath, auxiliary and sweep gas flow rate were: 45, 8 and 1 (arbitrary units), respectively. Ion Transfer Tube and Vaporizer temperatures were maintained at 350◦C and 300◦C, respectively. Data were acquired in full scan mode with resolution 60,000 from m/z 50–1000. The ddMS/MS was performed at a res- olution of 15,000 and stepped-normalized collision energies of 20, 30 and 50 at different mass range: m/z 50–250, m/z 250–500, m/z 500–750 and m/z 750–1000. 22 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS Metabolomics Analysis: The extracted samples were split into three different batches; each batch was randomized and analyzed in a single LC-MS analytical run. The pooled QC samples (n = 6) were analyzed at the beginning of each batch to equilibrate the column prior analysis. The pooled QC sample injections were interspaced throughout the run to check the stability, robustness, repeatability, and performance of the analytical system. The QC sample was also extensively analyzed (n > 6) with ddMS/MS to include different mass ranges, a targeted list of in-house database standards, a list of expected metabolites, and a list of metabolite peaks with no fragmentation. Data pre-processing for untargeted peak-picking, alignment, and deconvolution, and metabolite annotation was car- ried out using Compound Discoverer v3.3 (Thermo Fisher Scientific, USA). The confidence in metabolite annotation was assigned as Level 1–4 (L1-L4) based on the recommendation by Chemical Analysis Working Group, Metabolomics Standards Initiative (MSI).149 QUANTIFICATION AND STATISTICAL ANALYSIS Oral microbial diversity analyses using 16S rRNA amplicons Alpha diversity was calculated using species richness and Shannon’s Diversity Index for all samples (N = 669 individuals), with reads rarefied to depths between 1,000 and 10,000. For each depth, 100 iterations were performed and the mean values were used as the diversity estimates for each sample. Rarefaction curves indicated that 4,000 reads were sufficient to capture most of the bacterial diversity in this dataset. Samples with fewer than 4,000 reads were excluded, resulting in a final dataset with a total of 5,606,877 reads (63% of initial reads) and 380 ASVs from 628 individuals. Comparison of alpha diversity metrics were performed by rarifying all sam- ples to 4,000 reads. A generalized linear mixed-effects model was used to identify covariates associated with Shannon Diversity In- dex, treating sample ID as a random effect (Table S2). Random forests classifier We used a random forest classifier using the randomForest package in R to differentiate between obese and control groups using ASVs as predictors, unless otherwise specified. The data was partitioned into training (80%) and testing (20%) sets, the model was fit using 10-fold cross validation repeated three times using 500 trees. The performance of the classifier was assessed by generating area under the receiver operating characteristic curves (AUC) using ROCR package.120 The top features distinguishing between the two groups were determined using the varImp function in the caret119 package. Oral microbial composition For the amplicon sequencing (N = 628 individuals), beta diversity was assessed at the ASV level using unweighted UniFrac, weighted UniFrac, and Bray–Curtis distances calculated by log+1 transformation of non-rarefied 16S rRNA gene count data. Unsupervised clustering was performed with Principal Coordinate Analyses (PCoA) using the phyloseq package and visualized with the ggplot2121 package in R. For the metagenomics data (N = 191 individuals), microbial composition was assessed by PCoA performed using Bray–Curtis distances using the species level relative abundances obtained from MetaPhlAn. PERMANOVA and EnvFit were per- formed with 10,000 randomizations using the vegan118 package. Covariates with p < 0.05 were considered statistically significant (see Tables S2 and S5 for full list of covariates, effect sizes and p-values). Generalized linear mixed effect models with sample ID as the random effect were used to identify covariates associated with the top 3 Principal Coordinate (PCo) axes (Figures S1C and S3A). Oral microbiome-obesity associations We first performed an association analysis using a full model that included 312 individuals, 27 non-collinear variables (VIF <10), including 13 demography and lifestyle-related and 14 health-associated factors (Table S2). This approach accounted for missing data while minimizing model overfitting, enabling us to evaluate the relationship between the oral microbiome and systemic health while adjusting for lifestyle-related factors. Association analyses were performed using analysis of the weighted UniFrac distances using two complementary approaches: PERMANOVA, which assesses microbiome distance-based differences between individuals, and EnvFit, which evaluates the top three principal coordinate dimensions explaining the greatest variation in the oral microbiome. To ensure the robustness of our findings, we repeated these analyses using a reduced model that included only the significant variables from the full model while incorporating the larger cohort of 618 individuals with available data (Table S2). Across all these analyses, BMI emerged as a consistent factor associated with oral microbiome variation, highlighting it as a key variable for further investigation. Differential abundance MaAsLin234 was used to identify ASVs (Table S6), species (Table S6), microbial pathways (Table S7) and enzymes (Table S8) that were differentially abundant between the obese and control groups (Feature abundance ∼ group). Multiple testing corrections were performed by computing the False Discovery Rates (FDRs) using the Benjamini–Hochberg method. Features with FDR- adjusted p < 0.05 were considered significant. All model coefficients, p-values, adjusted p-values, and N (number of individuals included in the comparisons) are reported in the corresponding Supplementary Tables. Cell Reports 45, 116819, February 24, 2026 23 Article ll OPEN ACCESS Metabolite abundances Raw LC-MS peaks were subjected to batch-to-batch correction and total sum normalization. The dataset was then log transformed to restore normality. The metabolite (tR, m/z) pairs were exported with their normalized abundances for multivariate analysis (MVA) using Simca P16 (Sartorius Stedim Data Analytics AB, Sweden). The imported dataset of the samples in the study was mean- centered and unit-variance scaled. Principal component analysis (PCA), partial least squares - discriminant analysis (PLS-DA) and orthogonal partial least squares - discriminant analysis (OPLS-DA) were used for checking the analytical performance and modeling the differences between samples. Variable importance in the projection (VIP) > 1.0 of the OPLS-DA, correlation scaled loading ab- solute p(corr) value > 0.4, and a Student t test p-value < 0.05 adjusted with false discovery rate using the Benjamini-Hochberg approach were used to identify the significantly altered metabolites in the study (see Table S9 for VIP scores, correlation coefficients and adjusted p-values). Metabolite enrichment for structural classes, tissues, and dietary factors was performed using MetaboAnalyst and FDR-adjusted p-values<0.05 were considered significant.141 Procrustes analyses To assess the relationship between microbial functional pathways and metabolite profiles, we performed Procrustes analysis, which evaluates the correlation between the distance matrices of two multivariate datasets by aligning their principal coordinate spaces and measuring structural similarity. We first constructed Bray-Curtis dissimilarity matrices for microbial functional pathways and metab- olite profiles. Each dataset underwent Principal Coordinate Analysis (PCoA), and all available axes were used in Procrustes analysis, implemented in the vegan package in R. Procrustes analysis optimally rotates, scales, and translates one ordination space onto the other to maximize alignment, which was quantified using the Procrustes correlation coefficient (rho). To assess statistical signifi- cance, we conducted a permutation test with 1,000 permutations, randomly shuffling microbial and metabolite data points and re- calculating the correlation each time (Figure S6A). This was repeated to assess the relationship between microbial enzymes and metabolite profiles (Figure S6C). Microbiome blood marker associations We computed pairwise Spearman’s correlations between 366 bacterial species, 321 pathways, 285 metabolites, and 51 clinical markers. Multiple testing correction was performed by computing FDRs using the Benjamini-Hochberg method and FDR-adjusted p < 0.05 were considered significant (see Table S14 for correlation coefficients, p-values and adjusted p-values). Hierarchical clus- tering was applied to the correlation matrix to identify five feature clusters. Enrichment of obesity related microbiome and clinical features in the clusters was evaluated with hypergeometric tests. An igraph122 object was created from the correlation matrix and global (degree centrality and betweenness) and cluster-specific (edge density and transitivity) network metrics were calculated (Figure S8C). Central nodes in the network were inferred from global and cluster-specific metrics. 24 Cell Reports 45, 116819, February 24, 2026 Article ll OPEN ACCESS Integrative multi-omics analysis reveals oral microbiome-metabolome signatures of obesity Introduction Results Study design and cohort description Effect of lifestyle and systemic health on the oral microbiome in the Emirati population Matched analysis highlights biomarkers of metabolic dysregulation in obesity Obesity-associated alterations in oral microbiome composition Obesity-associated functional changes correspond to metabolic alterations Oral bacterial species contributing to obesity-associated functional differences Integrative multi-omics analysis reveals microbe-metabolomic signatures linked to obesity-associated clinical markers Discussion Limitations of the study Resource availability Lead contact Materials availability Data and code availability Consortia Acknowledgments Author contributions Declaration of interests Supplemental information References STAR★Methods Key resources table Experimental model and study participant details Study design and mouthwash sample collection Ethics statement Method details Blood and urine biomarker measurements DNA extraction 16S rRNA gene sequencing 16S rRNA amplicon processing Identification of matched cohort and lifestyle and health-associated variable selection Blood and urine marker analysis Metagenomics library preparation and sequencing Metagenomic processing, taxonomic profiling, and functional characterization Metagenomic assembly and functional analyses Untargeted metabolomics using LC-MS and LC-MS/MS Quantification and statistical analysis Oral microbial diversity analyses using 16S rRNA amplicons Random forests classifier Oral microbial composition Oral microbiome-obesity associations Differential abundance Metabolite abundances Procrustes analyses Microbiome blood marker associations