Role of an unclassified lachnospiraceae in the pathogenesis of type 2 diabetes: a longitudinal study of the urine microbiome and metabolites

ABSTRACT Recent investigations have revealed that the human microbiome plays an essential role in the occurrence of type 2 diabetes (T2D). However, despite the importance of understanding

the involvement of the microbiota throughout the body in T2D, most studies have focused specifically on the intestinal microbiota. Extracellular vesicles (EVs) have been recently found to

provide important evidence regarding the mechanisms of T2D pathogenesis, as they act as key messengers between intestinal microorganisms and the host. Herein, we explored microorganisms

potentially associated with T2D by tracking changes in microbiota-derived EVs from patient urine samples collected three times over four years. Mendelian randomization analysis was conducted

to evaluate the causal relationships among microbial organisms, metabolites, and clinical measurements to provide a comprehensive view of how microbiota can influence T2D. We also analyzed

EV-derived metagenomic (_N_ = 393), clinical (_N_ = 5032), genomic (_N_ = 8842), and metabolite (_N_ = 574) data from a prospective longitudinal Korean community-based cohort. Our data

revealed that _GU174097_g_, an unclassified _Lachnospiraceae_, was associated with T2D (_β_ = −189.13; _p_ = 0.00006), and it was associated with the ketone bodies acetoacetate and

3-hydroxybutyrate (_r_ = −0.0938 and −0.0829, respectively; _p_ = 0.0022 and 0.0069, respectively). Furthermore, a causal relationship was identified between acetoacetate and HbA1c levels

(_β_ = 0.0002; _p_ = 0.0154). _GU174097_g_ reduced ketone body levels, thus decreasing HbA1c levels and the risk of T2D. Taken together, our findings indicate that _GU174097_g_ may lower the

risk of T2D by reducing ketone body levels. SIMILAR CONTENT BEING VIEWED BY OTHERS FUNCTIONAL ALTERATIONS AND PREDICTIVE CAPACITY OF GUT MICROBIOME IN TYPE 2 DIABETES Article Open access 16

December 2023 GUT MICROBIOME STUDIES IN CKD: OPPORTUNITIES, PITFALLS AND THERAPEUTIC POTENTIAL Article 10 November 2022 ALTERATIONS OF GUT MICROBIOTA IN BIOPSY-PROVEN DIABETIC NEPHROPATHY

AND A LONG HISTORY OF DIABETES WITHOUT KIDNEY DAMAGE Article Open access 27 July 2023 INTRODUCTION Recent studies have revealed that the intestinal microbiota plays essential roles in host

energy homeostasis, body adiposity, blood sugar control, insulin sensitivity, hormone secretion, and the pathogenesis of metabolic diseases, such as type 2 diabetes (T2D) and obesity1,2,3.

However, most of these studies analyzed stool samples and therefore obtained limited information relative to insights from direct sampling of the intestinal mucosa, which is not possible in

most cases. In addition, the composition of microbial communities in stool samples is greatly affected by the specific compartment in which they reside, such as the mucous membrane4.

Microbial communities also differ based on their source, ranging from the intestines, skin, and airways, which are frequently studied, to urine and blood, which are generally sterile

environments5. Therefore, it is important not only to understand the role of the intestinal microbiota but also to consider the function and combined contribution of the all microbiota

throughout the body. Extracellular vesicles (EVs) have been recently suggested to function as the main messengers between intestinal microorganisms and the host. EVs travel long distances

within and between body tissues6 and have been used as biomarkers of atopic dermatitis, alcoholic hepatitis, and asthma7,8,9,10. Microbiota-derived EVs can enter the circulatory system

through the intestinal barrier. They are suspected to play a key role in the development of insulin resistance, potentially providing important clues into the pathogenesis of T2D. For

example, EVs derived from _Pseudomonas panacis_ are present in the stool samples of high-fat diet-fed mice. They can infiltrated the gut barrier and block the insulin pathway in skeletal

muscle and adipose tissue, inducing the development of insulin resistance and glucose intolerance11. However, microbiota-derived EVs are highly variable, as they are modulated by different

factors, such as age and sex. Therefore, caution should be exercised when inferring causal relationships based on the statistical analysis of microbiota-derived data. Furthermore,

longitudinal microbiota studies may allow for stronger inferences than cross-sectional studies12 and may allow for the detection of microorganisms related to the progression of T2D in

healthy subjects. However, existing studies have been predominantly cross-sectional in nature and are based on correlation analyses. As a result, these studies are unable to comprehensively

provide an understanding of the exact roles of the intestinal microbiota and EVs in metabolic disease development. Therefore, in the present study, we investigated the prospective Korean

Association REsource project (KARE) cohort13. By tracking changes in microbiota-derived EVs in urine samples from Korean adults collected three times over four years, we explored the

potential associations between microorganisms and T2D progression. Furthermore, using genomic and metabolite data from the KARE cohort, we conducted a multiomics analysis to investigate the

specific role of microorganisms potentially involved in T2D pathogenesis. We expect our findings to provide information regarding how microbes, the substances they produce, and their

byproducts interact with the human body and affect metabolic disease development. In addition, we evaluated causal relationships among microbial organisms, ketone bodies, and clinical

measurements, with the aim of further elucidating the relationship between T2D and the microbiota. MATERIALS AND METHODS COHORT AND STUDY DESIGN The KARE cohort is a prospective study cohort

involving subjects from the rural community of Ansung and the urban community of Ansan in South Korea. The KARE project began in 2001 as part of the Korean Genome Epidemiology Study14. We

used data from urine samples taken from subjects in 2013, 2015, and 2017, which we refer to as phases 1, 2, and 3 in this study. After collection, the urine was stored at –80 °C. For the

1,891 subjects whose urine samples were available, age, sex, and body mass index (BMI) were matched via 2:1:1 propensity score matching. As a result, a healthy group (healthy in all phases,

_N_ = 328), a T2D-at-risk group (T2D-at-risk in all phases, _N_ = 164), and a T2D group (T2D in any of the three phases, _N_ = 164) were selected. From the remaining unmatched subjects, 35

T2D subjects were also included. Consequently, 691 subjects were finally included, and their 2,072 urine samples were subjected to microbiota analysis. Metagenomic, metabolite, clinical, and

genomic data were subjected to comprehensive analyses (Fig. 1). OPERATIONAL DEFINITION OF T2D AND RELATED PHENOTYPES Study participants were categorized into control individuals,

T2D-at-risk patients, and T2D patients. T2D and T2D-at-risk patients were diagnosed on the basis of the American Diabetes Association criteria, which are provided in Supplementary Table 1.

T2D status was then stratified into _T2D-at-risk/T2D_ (0 for healthy; 1 for T2D-at-risk and T2D) and _binary_T2D_ (0 for healthy and T2D-at-risk; 1 for T2D). In addition, we considered other

T2D-related indicators, such as BMI, HbA1c levels, fasting glucose and insulin levels, 60- and 120-min plasma glucose levels, and insulin levels after a 75 g oral glucose tolerance test in

our analysis. Age, the levels of total cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, kidney- and liver-related disease indicators (blood urea nitrogen (BUN),

creatinine, aspartate aminotransferase (AST), and alanine aminotransferase (ALT) C-reactive protein (CRP), white blood cell (WBC) count, red blood cell (RBC) count, hemoglobin, hematocrit,

and platelet count) were also collected. The homeostatic model assessment for insulin resistance (HOMA-IR) was calculated using fasting glucose and fasting insulin levels15. Descriptive

statistics for all variables were generated using Rex software (RexSoft Inc., Seoul, Korea) (Supplementary Table 2)16. EV ISOLATION AND DNA EXTRACTION For EV isolation, urine samples were

subjected to differential centrifugation at 10,000 × g and 4 °C for 10 min using a microcentrifuge (Labogene 1730R; Bio-Medical Science, Seoul, Korea)17. To remove bacteria, foreign

particles, and waste, the supernatant was filtered through a 0.22-µm filter (Inchpor2 Syringe Filter; Inchemtec, Seoul, Korea). The isolated EVs were boiled at 100 °C for 40 min and

centrifuged at 18,214 × g and 4 °C for 30 min to eliminate floating particles and impurities. The supernatant was collected and subjected to DNA extraction using a PowerSoil® DNA Isolation

Kit (MO BIO Laboratories, Carlsbad, CA, USA) according to the manufacturer’s protocol. DNA was quantified using the QIAxpert system (Qiagen, Hilden, Germany). 16 S RRNA SEQUENCE DATA

PROCESSING Paired-end sequencing of the V3-V4 region of the bacterial 16 S rRNA gene was conducted at MD Health care (Seoul, Korea) with the MiSeq Reagent Kit v3 (600 cycles, Illumina, San

Diego, CA, USA) using the widely used primers 16S_V3_F (5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACA-GCCTACGGGNGGCWGCAG-3′) and 16S_V4_R

(5′-GTCTCGTGGGCTCGGAGATGTGTATA-AGAGACAGGACTACHVGGGTATCTAATCC-3′). Adaptor sequences were detected and removed using CUTADAPT software (https://cutadapt.readthedocs.io) with a minimum overlap

of 11, a maximum error rate of 10%, and a minimum length of 1018. Sequences were merged using CASPER (http://best.snu.ac.kr/casper) with a mismatch ratio of 0.27 and filtered based on the

Phred (Q) score, resulting in sequences of 350–550 bp in length19,20. After the merged sequences were dereplicated, chimeric sequences were detected and removed using VSEARCH

(https://github.com/torognes/vsearch) and the Silva Gold reference database for chimeras21. Open-reference operational taxonomic unit (OTU) picking was conducted based on the EzTaxon

database using UCLUST (http://www.drive5.com/usearch)22,23. For each OTU, we calculated its proportion among all OTUs and determined the mean value across all subjects. If the resulting

value was <0.001, the OTU was excluded24. Among the 691 subjects, those with a read count <3000 or whose genomic data were not available in any phase were excluded. As a result, 1179

samples from 393 subjects, including 70 genera, were used for subsequent analyses. PREDICTION OF FUNCTIONAL PROFILES FROM 16 S RRNA METAGENOMIC DATA The functional potential of microbial

communities can be predicted from their phylogeny. Tax4fun uses evolutionary modeling to predict metagenomes based on 16 S data from the SILVA reference genome database. The SILVA-based 16 S

rRNA profile was used to estimate a taxonomic profile of prokaryotic Kyoto Encyclopedia of Genes and Genomes (KEGG) organisms. The estimated abundances of KEGG organisms were normalized

using the 16 S rRNA copy number obtained from the National Center for Biotechnology Information (NCBI) genome annotations. Finally, the normalized taxonomic abundances were used to linearly

combine the precomputed functional profiles of KEGG organisms to predict the functional profile of the microbial community25. Similar to the analysis of OTUs, we calculated the mean of the

relative proportions across all subjects for each functional profile. If the resulting value was <0.001, the functional profile was excluded from the analysis. As a result, 238 functional

profiles were retained for analysis. METABOLITE ANALYSIS OF KETONE BODIES Serum metabolites were analyzed using the Agilent 1290 Infinity LC and Agilent 6490 Triple Quadrupole MS systems

(Agilent Technologies, Palo Alto, CA, USA). The levels of acetoacetate and 3-hydroxybutyrate from subjects included in the metagenomics dataset were determined in the multiple reaction

monitoring mode. A batch normalizer was used to correct for possible batch effects26. ANALYSIS OF BACTERIAL COMPOSITION AND MICROBIAL VARIANCE We calculated alpha- and beta-diversity indices

using R (v3.6.2) after read number normalization with the Rarefy function in the R package GUniFrac (v1.1). The R package Fossil (v0.4.0) was used to obtain Chao1 and ACE diversity indices.

The Shannon index and Simpson’s diversity index were calculated using the Vegan package in R (v2.5.6). Taxonomy-based ring charts were created using the Krona tool27. PERMANOVA is a

nonparametric multivariate analysis of variance test based on pairwise distances28. The R package pldist was used to obtain the microbial variance for individuals in repeated measurements of

microbial profiles. pldist summarizes within-individual shifts in the microbiome composition and compares these across individuals. pldist also calculates dissimilarities based on a novel

transformation of relative abundances, which are then extended to more than two time points. They are then incorporated into a chosen beta-diversity, which, in our case, was Bray–Curtis

dissimilarity. PERMANOVA was performed for biochemistry-related KARE phenotypes using the adonis function in R. PERMANOVA can be applied to the cross-sectional data, and thus, the phenotypes

were averaged for phases 1, 2, and 3. STATISTICAL ANALYSIS OF THE EFFECT OF THE MICROBIOME ON T2D AND DIABETES RISK INDICATORS For each taxon and functional profile, a generalized linear

mixed model (LMM) with the logit link function was used to find associations with _binary_T2D_ and _T2D-at-risk/T2D_, whereas an LMM was used for log-transformed diabetes risk indicators. A

random effect with a compound symmetry structure for each time point was incorporated to adjust the similarity of T2D status for the same subject at different time points, and the sandwich

estimator was used to find a robust estimate against the misspecified covariance matrix. To accommodate the multiple testing problem, _p_ values were adjusted for the false discovery rate

(FDR) using the Benjamini–Hochberg method29. NETWORK ANALYSIS OF A T2D-RELATED TAXON BASED ON MULTIOMICS DATA To assess overall associations using repeatedly measured multiomics data, we

first modeled an LMM using log-transformed diabetes risk indicators as response variables and age in phase 1 as well as sex as explanatory variables with a compound symmetry structure for

its covariance. We modeled an LMM with a T2D-related taxon as the response variable with the same covariates and covariance structure. For each combination of diabetes risk indicators and a

T2D-related taxon, two different sets of residuals were obtained, and Spearman correlations between the residuals were calculated. Similarly, the association between a chosen microbial

marker and the levels of ketone bodies was analyzed. Network analysis was conducted to calculate simple correlations among diabetes risk indicators, a chosen taxon, and ketone bodies. Edge

width was calculated as –log10 of the _p_ value. The network was visualized using the R package visNetwork (v2.0.8). GENOTYPING, IMPUTATION, AND QUALITY CONTROL Quality control and genotype

imputation were performed according to the standard quality control and imputation protocols for the genotypes of 8842 KARE cohort participants30. After quality control, 8216 subjects with

17,716,215 single-nucleotide polymorphisms (SNPs) were included in the analysis. In total, the data of 351 subjects with a read count <3,000 and nonmissing T2D status for all phases were

used for a genome-wide association study (GWAS) of metagenomic data. A total of 574 subjects who had no missing metabolite levels and T2D status for all three phases were selected for a GWAS

of metabolite levels. Among the subjects not included in the metabolite or metagenome GWAS, 3542 subjects had KARE phenotypes for the three phases and were thus included in a GWAS of KARE

phenotypes. We excluded subjects in the metabolite or metagenome GWAS for the purposes of a two-sample Mendelian randomization (MR) study. Details are provided in Supplementary Fig. 1, and

all the associated SNPs from each GWAS are listed in Supplementary Table 3. MR ANALYSIS MR uses genetic variants that are not associated with conventional confounders of observational

studies and is therefore considered analogous to randomized controlled trials31. Randomly selected alleles are transmitted from parents, and genotypes can be assumed to be independent, with

many potential confounders. This randomization produces unbiased estimates for the associations between the main exposures and outcomes. Thus, genetic variants associated with the main

exposure were used as instrumental variables. There are two types of MR, namely, two-sample MR and one-sample MR. The former uses two independent datasets with nonoverlapping samples for the

association of SNP exposure and SNP outcome (as opposed to one-sample MR) It is less likely to lead to inflated type 1 error rates and false-positive findings when compared to one-sample

MR. Two-sample MR was conducted to identify the effect of a microbial taxon or each ketone body on KARE phenotypes by using no overlapping samples. One-sample MR was conducted to estimate

the effect of a chosen taxon on each ketone body. For one-sample MR, we conducted two-stage least-squares regression. The first stage consisted of a regression for SNP exposure, and the

second stage consisted of a regression for the outcome of interest on the fitted values from the first-stage regression. The estimator of the coefficient for first-stage fitted values in the

second-stage model is the causal estimate32,33. F-statistics from the first-stage regression were examined to avoid weak instrument bias34. The Durbin-Wu-Hausman (DWH) test for

endogeneity35 was used to evaluate whether there is any evidence that the causal estimate differs from the ordinary least square estimate of exposure and outcome. For two-sample MR, the

average _F_-statistic was used to avoid weak instrument bias. The inverse-variance-weighted (IVW) method, Cochran’s _Q_ test, and MR-PRESSO global test were used to confirm the heterogeneity

assumption, and _I_2 was used for the no measurement error (NOME) assumption. To enhance the validity of MR analysis, we considered the extensive range of existing MR methods, including

IVW, MR-egger, MR-egger with SIMEX correction, median-weighted method, and MR-PRESSO, and selected the recommended MR method based on the violations of MR assumptions36. RESULTS LONGITUDINAL

CHANGES IN THE URINE MICROBIAL COMPOSITION OVER FOUR YEARS The alpha-diversity of the urine microbiome decreased during the follow-up period, which may have been an effect of aging

(Supplementary Fig. 2). A nonmetric multidimensional scaling plot based on beta diversity also revealed a gradual change in microbiota composition with age (Supplementary Fig. 3). The

overall microbiome composition at the phylum and genus levels is presented in Fig. 2 and Supplementary Fig. 4, respectively. _Verrucomicrobia_, _Bacteroidetes_, and _Firmicutes_ were the

predominant phyla, whereas _Akkermansia_ and _Bacteroides_ were the predominant genera. T2D AND OTHER CLINICAL TRAITS EXPLAINED BY MICROBIAL VARIANCE We investigated the associations between

various clinical phenotypes and microbial compositions using PERMANOVA (Supplementary Fig. 5). HbA1c, WBC, hematocrit, _binary_T2D_, and age in phase 1 significantly explained changes in

microbial composition during the follow-up period (_p_ = 0.0061, 0.0107, 0.0110, 0.0409, and 0.0290, respectively; FDR-adjusted _p_ = 0.1027, 0.1027, 0.1027, 0.2290, and 0.2030,

respectively). HbA1c and _binary_T2D_ partially explained the variance in microbial changes over the 4 years, indicating that the longitudinal change in microbiome composition may be more

closely associated with T2D-related phenotypes than with other clinical traits. TAXA AND FUNCTIONAL PROFILES ASSOCIATED WITH T2D AND DIABETES RISK INDICATORS In an association analysis of 70

genera with _binary_T2D_ and _T2D-at-risk/T2D_ phenotypes, _GU174097_g_, an unclassified _Lachnospiraceae_, was found to exhibit a significant association with these phenotypes and was more

abundant in healthy subjects than in diabetic or prediabetic patients (Table 1). We divided the samples into four groups. The _Healthy in Phases 1-3_ group included subjects who were

healthy in phases 1, 2, and 3. The _T2D in Phases 1-3_ group consisted of subjects who had T2D in phases 1, 2, and 3. The _Healthy to T2D-at-risk/T2D_ group included subjects who were

healthy in phase 1 and became T2D patients or T2D-at-risk in phase 3. The _T2D-at-risk/T2D to Healthy_ group included subjects who were T2D-at-risk/T2D in phase 1 and healthy in phase 3. The

relative abundance of _GU174097_g_ in subjects who were healthy at baseline but changed to the _T2D-at-risk/T2D_ group at phase 2 or 3 decreased with the development of T2D (_p_ = 0.0001).

Conversely, its relative abundance in the _T2D-at-risk/T2D to Healthy_ group exhibited no tendency to decrease (_p_ = 0.19) (Fig. 3). Supplementary Fig. 6 shows the profiles of _GU174097_g_

for randomly selected subjects. The relative abundance of _GU174097_g_ in subjects who were healthy at baseline but changed _to T2D-at-risk/T2D_ at phase 2 or 3 tended to decrease. Most T2D

patients had small relative abundances of _GU174097_g_ at baseline. In summary, _GU174097_g_ was clearly associated with the progression of diabetes over time, and this association was not

simply based on diabetic or nondiabetic status. To investigate the T2D-associated microbial functional profiles, 238 functional profiles were evaluated. The significant associations at an

FDR-adjusted significance of 0.1 are presented in Supplementary Table 4. The _T2D-at-risk/T2D_ phenotype was related to the cationic antimicrobial peptide. Furthermore, the biosynthesis of

fatty acids, coenzyme A (CoA), and secondary metabolites as well as oxidative phosphorylation were significantly associated with the _Binary_T2D_ phenotype at an FDR-adjusted significance of

0.1. Next, we investigated the associations between the log-transformed diabetes risk indicators and genera, and significant associations at an FDR-adjusted significance of 0.1 were

identified. Twelve, four, and 20 genera were significantly associated with HbA1c, glucose, and insulin levels, respectively. In particular, _Hafnia_ was associated with HbA1c and 60- and

120-min insulin levels, whereas _AB185816_g_ and _Akkermansia_ were associated with HbA1c, fasting glucose, and 60-min insulin levels (Supplementary Table 5). ASSOCIATIONS BETWEEN

T2D-RELATED UNCLASSIFIED _LACHNOSPIRACEAE_ AND DIABETES RISK INDICATORS AND KETONE BODIES To confirm the association between _GU174097_g_ and T2D, we performed extensive validation analysis

using clinical and metabolite data. We analyzed the association between _GU174097_g_ and diabetes risk indicators (Table 2). Among all glucose- and insulin-related variables, _GU174097_g_

was significantly and positively associated with the 60-min insulin level. Thereafter, we analyzed the potential associations between ketone bodies and the T2D-related taxon, since ketone

bodies have been suggested as markers of disrupted glucose metabolism in prediabetic patients37. The ketone bodies 3-hydroxybutyrate and acetoacetate exhibited significant negative

correlations with _GU174097_g_ (_r_ = –0.0829 and –0.0938, respectively; _p_ = 0.0069 and 0.0022, respectively) (Table 3). Supplementary Fig. 7 shows the tendency of high acetoacetate and

3-hydroxybutyrate concentrations coinciding with the low abundance of _GU174097_g_. The ion abundances of acetoacetate and 3-hydroxybutyrate did not rise beyond 2000 and 10000, respectively,

when the relative abundance of _GU174097_g_ was high. Finally, we established an association network for diabetes risk indicators and ketone bodies, as the same observed correlations can

imply completely different biological processes. For example, if high levels of glucose or HbA1c tend to appear in parallel to high levels of insulin, insulin resistance may be present.

However, if high levels of glucose or HbA1c are observed in parallel to low levels of insulin, insulin secretion may have suppressed glucose or HbA1c levels. Network analysis indicated

strong associations among the diabetes risk indicators (Supplementary Fig. 8). In particular, the 60-min insulin level exhibited a strong negative correlation with HbA1c levels, suggesting

that the former can decrease the latter. Ketone bodies exhibited negative correlations with fasting insulin and 60-min insulin levels and positive correlations with 60- and 120-min glucose

levels. CAUSAL RELATIONSHIP BETWEEN THE T2D-RELATED TAXON AND KETONE BODIES AND THE DIABETES RISK INDICATORS One-sample MR did not reveal any significant causal relationship between

_GU174097_g_ and ketone bodies and vice versa (Supplementary Table 6). To verify whether a causal relationship existed between the abundance of _GU174097_g_ or the levels of ketone bodies

and diabetes risk indicators, two-sample MR analysis was performed. Extensive assumption checks were conducted to enhance the validity of the two-sample MR analysis (Supplementary Table 7).

No weak instrument bias was observed (_F_-statistic >10). However, NOME assumptions were violated for all tests because _GU174097_g_, 3-hydroxybutyrate, and acetoacetate had seven, eight,

and five SNPs as their instrument variables, respectively, and these values were not sufficiently large for _I_2 > 90. In this case, if heterogeneity exists, MR–Egger (SIMEX) is

recommended; otherwise, IVW is recommended. As the InSIDE assumption cannot be statistically tested38, the weighted median method—a robust approach used in cases of InSIDE assumption

violation—has to be considered with each recommended method36. Therefore, MR–Egger (SIMEX) was used to estimate the causal effect of 3-hydroxybutyrate on 60-min insulin as well as that of

acetoacetate on HbA1c levels. The IVW method was used to estimate all other causal effects. To determine the causal effect of 3-hydroxybutyrate on 60-min insulin, rs2259835 was detected as

an outlier via MR-PRESSO at a significance level of 0.05 (Supplementary Table 8). Thus, rs2259835 had to be removed to prevent potential horizontal pleiotropy. The result of MR-PRESSO is

shown in Supplementary Table 9 and shows the estimates without outliers. The effect of acetoacetate on the HbA1c level was the only significant effect at an FDR-adjusted significance of

0.05, indicating that acetoacetate increases HbA1c levels (Supplementary Table 9). The results obtained using the weighted median method corroborated this significant association (_p_ = 

0.0475). DISCUSSION Recent microbiome studies have shown that T2D is associated with gut dysbiosis39,40,41 that can result in altered intestinal barrier function and a dysregulation of host

metabolic and signaling pathways42. Intestinal bacteria can promote insulin resistance by triggering inflammation via polysaccharides, which are components of the gram-negative bacterial

cell wall43. Furthermore, microbiota-derived EVs are expected to affect insulin resistance and provide a more in-depth understanding of T2D pathogenesis11. Various bacterial metabolites,

such as short-chain fatty acids (SCFAs), can modulate the function of various signaling pathways implicated in human health and can protect against insulin resistance43,44. The human

microbiota is highly variable, and this variability is determined by various external factors, such as diet, exercise, mobility, medication, and microbial cooccurrence patterns. Many of

these external factors also determine the risk of metabolic disease and are age-related45; that is, the intestinal microbiota and host phenotype are substantially altered with aging.

Furthermore, the effect of the intestinal microbiota on the host phenotype is also dependent on the age of the host. The estimation of within-subject covariate effects represents a robust

approach against between-subject confounders, and longitudinally measured microbiome data enable characterization of the effects of the microbiota on host disease risk. As most existing

studies have been cross-sectional in nature, the validity and interpretation of their results are limited. In turn, longitudinal studies are needed to comprehensively investigate the

association between the human microbiome and diseases, including T2D. Our longitudinal study revealed that a low abundance of _GU174097_g_ is a risk factor for T2D development. _GU174097_g_

has not been cultured to date. Multiomics data, including host genomic data, T2D-related metabolites, clinical information, and predicted functional metagenomic profiles, were utilized to

extensively validate our results via causality analysis. _GU174097__g is a member of the family _Lachnospiraceae_, and an association between _Lachnospiraceae_ and T2D risk has been reported

in several previous studies46,47. SCFA pathways, including the propanediol and acrylate signaling pathways, play important roles in mediating the effects of _Lachnospiraceae_ on T2D45.

Additionally, SCFA-producing bacteria affect epigenetic regulation in T2D patients and reduce the risk of developing T2D44,48. We found that _GU174097_g_ is positively correlated with the

60-min insulin level, and in turn, it is negatively correlated with HbA1c levels. This indicates that _GU174097_g_ reduces HbA1c levels and, thus, the risk of developing T2D by stimulating

insulin secretion. Next, we aimed to elucidate how _GU174097_g_ affects T2D through the regulation of 60-min insulin and HbA1c. Multiple mechanisms may underlie these associations, including

the effects of various microbiota-derived metabolites, including SCFAs, as previously suggested. In addition, ketone bodies have been reported not only as indicators of diabetic

hyperglycemia but also as markers of disturbed glucose metabolism in the prediabetic state37,48. Furthermore, fatty acid metabolism, CoA synthesis, and oxidative phosphorylation, all of

which are involved in ketogenesis or ketolysis, have been associated with T2D49. In our study, the ketone bodies 3-hydroxybutyrate and acetoacetate were negatively correlated with

_GU174097_g_ but positively correlated with the 60- and 120-min glucose levels. MR analysis was employed to investigate the effects of _GU174097_g_ and ketone bodies on diabetes risk

indicators. Although no causal relationship was observed between _GU174097_g_ and ketone bodies or other clinical variables, acetoacetate was found to be causally related to an increased

HbA1c level. HbA1c level is a major biomarker of T2D and explains the microbial beta-diversity. Furthermore, _GU174097_g_ was negatively correlated with acetoacetate. Therefore, our study

not only confirmed the importance of ketone bodies in T2D pathogenesis but also suggests an underlying mechanism for the association between _GU174097_g_ and T2D development. Previous

studies have reported that gut microbe–derived EVs can infiltrate the circulatory system through the gut barrier11,50. Furthermore, microbe-derived EVs in urine can reflect the lung and gut

microbiota of children with asthma51. Interestingly, T2D increases the co-occurrence of the same OTUs within the gut microbiome and microbe-derived EVs in urine samples52, which indicates

that these EVs may reflect the gut microbiota composition. _Coprococcus_, a member of the _Lachnospiraceae_ family, is one of the major butyrate-producing bacteria. It is known to utilize

metabolic intermediates essential for the synthesis of ketone bodies, such as acetoacetyl-CoA, 3-hydroxybutyryl-CoA, and crotonyl-CoA26, as energy sources to produce the SCFA butyrate. SCFAs

are considered beneficial for health and are considered to protect against T2D53. Thus, we hypothesize that _GU174097_g_ consumes acetoacetate to produce SCFAs. These SCFAS can promote

insulin secretion and decrease HbA1c levels, leading to a decreased risk of T2D. Our study had several limitations. First, as it was based on the metagenomic profiles of EVs, the microbial

compositions observed can differ from, and need to be further compared with, those of the intestinal microbiota. Second, as the genus-level taxonomy of _GU174097_g_ is unknown, ecological

and biological information on this species is limited. Third, published summary statistics of microbial GWAS are limited, and the sample size in the current microbial GWAS was small.

Therefore, the number of SNPs used as instrumental variables in our MR analysis was also suboptimal. Future studies should include a large sample size to identify more associated SNPs and

increase the power of MR analysis. Therefore, the mechanisms underlying T2D pathogenesis could be further identified and characterized. Fourth, even though extensive methods were used to

validate assumptions in our MR analysis and enhance the validity of causal analysis, the MR results were not easy to interpret. Ketone bodies and diabetes risk indicators were highly

correlated and interacted with each other. Additional in vivo and in vitro experiments may clarify the associations identified herein. Our study revealed that _GU174097_g_, an unclassified

_Lachnospiraceae_, is associated with T2D and ketone bodies. Furthermore, we found a potential causal relationship between ketone body acetoacetate and HbA1c levels. Our findings indicate

that _GU174097_g_ may lower the risk of developing T2D via the reduction in ketone body levels. Although the mechanisms by which _GU174097_g_ and ketone bodies affect T2D development have

not been elucidated, further large-scale longitudinal studies as well as in vivo and in vitro experiments could contribute to unraveling these mechanisms. DATA AVAILABILITY Raw datasets

generated during the current study are available in the NCBI Sequence Read Archive (BioProject id PRJNA716550; SRA accession id SAMN18437890-SAMN18438579, SAMN18443936-SAMN18444626,

SAMN18446963-SAMN18447653). REFERENCES * Fan, Y. & Pedersen, O. Gut microbiota in human metabolic health and disease. _Nat. Rev. Microbiol._ 19, 55–71 (2021). Article  CAS  PubMed 

Google Scholar  * Johnson, E. L., Heaver, S. L., Walters, W. A. & Ley, R. E. Microbiome and metabolic disease: revisiting the bacterial phylum Bacteroidetes. _J. Mol. Med._ 95, 1–8

(2017). Article  CAS  PubMed  Google Scholar  * Sanz, Y., Olivares, M., Moya-Pérez, Á. & Agostoni, C. Understanding the role of gut microbiome in metabolic disease risk. _Pediatr. Res._

77, 236–244 (2015). Article  PubMed  Google Scholar  * Zmora, N. et al. Personalized gut mucosal colonization resistance to empiric probiotics is associated with unique host and microbiome

features. _Cell_ 174, 1388–1405. e21 (2018). Article  CAS  PubMed  Google Scholar  * Proal, A. D., Albert, P. J. & Marshall, T. G. Inflammatory disease and the human microbiome. _Discov.

Med._ 17, 257–265 (2014). PubMed  Google Scholar  * Ahmadi Badi, S. et al. Microbiota-derived extracellular vesicles as new systemic regulators. _Front. Microbiol._ 8, 1610 (2017). Article

  PubMed  PubMed Central  Google Scholar  * Yang, J. et al. Diagnostic models for atopic dermatitis based on serum microbial extracellular vesicle metagenomic analysis: a pilot study.

_Allergy Asthma Immunol. Res._ 12, 792–805 (2020). Article  CAS  PubMed  PubMed Central  Google Scholar  * An, J., McDowell, A., Kim, Y. K. & Kim, T. B. Extracellular vesicle-derived

microbiome obtained from exhaled breath condensate in patients with asthma. _Ann. Allergy Asthma Immunol._ 126, 729–731 (2021). Article  PubMed  Google Scholar  * Lee, J. H. et al.

Metagenome analysis using serum extracellular vesicles identified distinct microbiota in asthmatics. _Sci. Rep._ 10, 15125 (2020). Article  CAS  PubMed  PubMed Central  Google Scholar  *

Kim, S. S. et al. Microbiome as a potential diagnostic and predictive biomarker in severe alcoholic hepatitis. _Aliment. Pharmacol. Ther._ 53, 540–551 (2021). CAS  PubMed  Google Scholar  *

Choi, Y. et al. Gut microbe-derived extracellular vesicles induce insulin resistance, thereby impairing glucose metabolism in skeletal muscle. _Sci. Rep._ 5, 1–11 (2015). Article  Google

Scholar  * VanderWeele, T. J., Jackson, J. W. & Li, S. Causal inference and longitudinal data: a case study of religion and mental health. _Soc. Psych. Psych. Epid._ 51, 1457–1466

(2016). Article  Google Scholar  * Cho, Y. S. et al. A large-scale genome-wide association study of Asian populations uncovers genetic factors influencing eight quantitative traits. _Nat.

Genet._ 41, 527–534 (2009). Article  CAS  PubMed  Google Scholar  * Kim, Y., Han, B.-G. & Group, K. Cohort profile: the Korean genome and epidemiology study (KoGES) consortium. _Int. J.

Epidemiol._ 46, e20–e20 (2017). Article  PubMed  Google Scholar  * Pisprasert, V., Ingram, K. H., Lopez-Davila, M. F., Munoz, A. J. & Garvey, W. T. Limitations in the use of indices

using glucose and insulin levels to predict insulin sensitivity: impact of race and gender and superiority of the indices derived from oral glucose tolerance test in African Americans.

_Diabetes care_ 36, 845–853 (2013). Article  CAS  PubMed  PubMed Central  Google Scholar  * RexSoft, Rex: Excel‐based statistical software. http://rexsoft.org/ (2018). * Lee, E. Y. et al.

Global proteomic profiling of native outer membrane vesicles derived from Escherichia coli. _Proteomics_ 7, 3143–3153 (2007). Article  CAS  PubMed  Google Scholar  * Martin, M. Cutadapt

Removes Adapter Sequences from High-Throughput Sequencing Reads. _EMBnet J._ 17, 10–12 (2011). Article  Google Scholar  * Kwon, S., Lee, B. & Yoon, S. CASPER: context-aware scheme for

paired-end reads from high-throughput amplicon sequencing. _BMC Bioinforma._ 15, S10 (2014). Suppl 9. Article  CAS  Google Scholar  * Bokulich, N. A. et al. Quality-filtering vastly improves

diversity estimates from Illumina amplicon sequencing. _Nat. Methods_ 10, 57–59 (2013). Article  CAS  PubMed  Google Scholar  * Rognes, T., Flouri, T., Nichols, B., Quince, C. & Mahé,

F. VSEARCH: a versatile open source tool for metagenomics. _PeerJ_ 4, e2584 (2016). Article  PubMed  PubMed Central  Google Scholar  * Yoon, S.-H. et al. Introducing EzBioCloud: a

taxonomically united database of 16S rRNA gene sequences and whole-genome assemblies. _Int. J. Syst. Evol. Micr._ 67, 1613 (2017). Article  CAS  Google Scholar  * Edgar, R. C. Search and

clustering orders of magnitude faster than BLAST. _Bioinformatics_ 26, 2460–2461 (2010). Article  CAS  PubMed  Google Scholar  * Li, K., Bihan, M. & Methé, B. A. Analyses of the

stability and core taxonomic memberships of the human microbiome. _PLoS one_ 8, e63139 (2013). Article  CAS  PubMed  PubMed Central  Google Scholar  * Aßhauer, K. P., Wemheuer, B., Daniel,

R. & Meinicke, P. Tax4Fun: predicting functional profiles from metagenomic 16S rRNA data. _Bioinformatics_ 31, 2882–2884 (2015). Article  PubMed  PubMed Central  CAS  Google Scholar  *

Wang, S.-Y., Kuo, C.-H. & Tseng, Y. J. Batch Normalizer: a fast total abundance regression calibration method to simultaneously adjust batch and injection order effects in liquid

chromatography/time-of-flight mass spectrometry-based metabolomics data and comparison with current calibration methods. _Anal. Chem._ 85, 1037–1046 (2013). Article  CAS  PubMed  Google

Scholar  * Ondov, B. D., Bergman, N. H. & Phillippy, A. M. Interactive metagenomic visualization in a Web browser. _BMC Bioinforma._ 12, 1–10 (2011). Article  Google Scholar  * Anderson,

M. J. A new method for non‐parametric multivariate analysis of variance. _Austral Ecol._ 26, 32–46 (2001). Google Scholar  * Benjamini, Y. & Hochberg, Y. Controlling the false discovery

rate: a practical and powerful approach to multiple testing. _J. Roy. Stat. Soc. B Met._ 57, 289–300 (1995). Google Scholar  * Gim, J. et al. A between ethnicities comparison of chronic

obstructive pulmonary disease genetic risk. _Front. Genet._ 11, 329 (2020). Article  PubMed  PubMed Central  Google Scholar  * Greco M, F. D., Minelli, C., Sheehan, N. A. & Thompson, J.

R. Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome. _Stat. Med._ 34, 2926–2940 (2015). Article  PubMed  Google Scholar  * Richard, M. A. et

al. Altered mechanisms of genital development identified through integration of DNA methylation and genomic measures in hypospadias. _Sci. Rep._ 10, 1–12 (2020). Article  CAS  Google

Scholar  * Bell, J. A. et al. Influence of puberty timing on adiposity and cardiometabolic traits: a Mendelian randomisation study. _PLoS Med._ 15, e1002641 (2018). Article  PubMed  PubMed

Central  CAS  Google Scholar  * Staiger, D. & Stock, J. H. Instrumental variables regression with weak instruments. _Econometrica_ 557–586 (1997). * Hausman, J. A. Specification tests in

econometrics. _Econometrica_ 1251–1271 (1978). * Jin, H., Lee, S. & Won, S. Causal evaluation of laboratory markers in Type 2 Diabetes on cancer and vascular diseases using various

mendelian randomization tools. _Front. Genet._ 11, 597420 (2020). Article  CAS  PubMed  PubMed Central  Google Scholar  * Mahendran, Y. et al. Association of ketone body levels with

hyperglycemia and type 2 diabetes in 9,398 Finnish men. _Diabetes_ 62, 3618–3626 (2013). Article  CAS  PubMed  PubMed Central  Google Scholar  * Bowden, J. Misconceptions on the use of

MR-Egger regression and the evaluation of the InSIDE assumption. _Int. J. Epidemiol._ 46, 2097–2099 (2017). Article  PubMed  PubMed Central  Google Scholar  * Wen, L. & Duffy, A. Factors

influencing the gut microbiota, inflammation, and type 2 diabetes. _J. Nutr._ 147, 1468S–1475S (2017). Article  PubMed  PubMed Central  CAS  Google Scholar  * Cani, P. D., Osto, M., Geurts,

L. & Everard, A. Involvement of gut microbiota in the development of low-grade inflammation and type 2 diabetes associated with obesity. _Gut microbes_ 3, 279–288 (2012). Article 

PubMed  PubMed Central  Google Scholar  * Upadhyaya, S. & Banerjee, G. Type 2 diabetes and gut microbiome: at the intersection of known and unknown. _Gut microbes_ 6, 85–92 (2015).

Article  CAS  PubMed  PubMed Central  Google Scholar  * Sharma, S. & Tripathi, P. Gut microbiome and type 2 diabetes: where we are and where to go? _J. Nutr. Biochem._ 63, 101–108

(2019). Article  CAS  PubMed  Google Scholar  * Vallianou, N. G., Stratigou, T. & Tsagarakis, S. Microbiome and diabetes: where are we now? _Diabetes Res. Clin. Pract._ 146, 111–118

(2018). Article  CAS  PubMed  Google Scholar  * Puddu, A., Sanguineti, R., Montecucco, F. & Viviani, G. L. Evidence for the gut microbiota short-chain fatty acids as key

pathophysiological molecules improving diabetes. _Mediat. Inflamm._ 2014, 162021 (2014). Article  Google Scholar  * O’Toole, P. W. & Jeffery, I. B. Gut microbiota and aging. _Science_

350, 1214–1215 (2015). Article  PubMed  CAS  Google Scholar  * Vacca, M. et al. The controversial role of human gut lachnospiraceae. _Microorganisms_ 8, 573 (2020). Article  CAS  PubMed

Central  Google Scholar  * Ibrahim, K. S. et al. Characterisation of gut microbiota of obesity and type 2 diabetes in a rodent model. _Biosci. Microbiota Food Health_ 40, 2019–2031 (2020).

Google Scholar  * Sharon, G. et al. Specialized metabolites from the microbiome in health and disease. _Cell Metab._ 20, 719–730 (2014). Article  CAS  PubMed  PubMed Central  Google Scholar

  * Kohlmeier, M. Nutrient Metabolism: Structures, Functions, and Genetics, in Food Science and Technology International Series 1st edn, 147–148 (Academic Press, London, 2003) * Kim, M. R.

et al. S taphylococcus aureus‐derived extracellular vesicles induce neutrophilic pulmonary inflammation via both T h1 and T h17 cell responses. _Allergy_ 67, 1271–1281 (2012). Article  CAS 

PubMed  Google Scholar  * Lee, Y. S., Kim, J. H. & Lim, D. H. Urine microbe-derived extracellular vesicles in children with asthma. _Allergy Asthma Immunol. Res._ 13, 75 (2021). Article

  CAS  PubMed  Google Scholar  * Nah, G. et al. Type-2 diabetics reduces spatial variation of microbiome based on extracellular vesicles from gut microbes across human body. _Sci. Rep._ 9,

1–10 (2019). Article  CAS  Google Scholar  * Sasaki, K., Sasaki, D., Hannya, A., Tsubota, J. & Kondo, A. In vitro human colonic microbiota utilises D-β-hydroxybutyrate to increase

butyrogenesis. _Sci. Rep._ 10, 1–8 (2020). Article  CAS  Google Scholar  Download references ACKNOWLEDGEMENTS This work was supported by the Industrial Core Technology Development Program

(20000134) funded by the Ministry of Trade, Industry and Energy (MOTIE, Korea). This study was conducted with bioresources from National Biobank of Korea, the Korea Disease Control and

Prevention Agency, Republic of Korea (NBK-2020-101). AUTHOR INFORMATION AUTHORS AND AFFILIATIONS * Department of Public Health Science, Graduate School of Public Health, Seoul National

University, Seoul, South Korea Kangjin Kim, Nam-Eun Kim & Sungho Won * Department of Medical Consilience, Graduate School, Dankook University, Seoul, South Korea Sanghun Lee * Brigham

and Women’s Hospital, Harvard Medical School, Boston, MA, USA Sang-Chul Park * Division of Pulmonary Sleep and Critical Care Medicine, Department of Internal Medicine, Korea University Ansan

Hospital, Ansan, South Korea Chol Shin * Institute of Human Genomic Study, College of Medicine, Korea University Ansan Hospital, Ansan, South Korea Chol Shin & Seung Ku Lee * Integrated

Metabolomics Research Group, Western Seoul Center, Korea Basic Science Institute, Seoul, South Korea Youngae Jung & Geum-Sook Hwang * Division of Allergy and Chronic Respiratory

Diseases, Center for Biomedical Sciences, National Institute of Health, Korea Center for Diseases Control and Prevention, Osong, Cheongju, South Korea Dankyu Yoon * Korea Medical Institute,

Seoul, South Korea Hyeonjeong Kim & Sanghyun Kim * Interdisciplinary Program for Bioinformatics, College of Natural Science, Seoul National University, Seoul, South Korea Sungho Won *

Institute of Health and Environment, Seoul National University, Seoul, South Korea Sungho Won Authors * Kangjin Kim View author publications You can also search for this author inPubMed 

Google Scholar * Sanghun Lee View author publications You can also search for this author inPubMed Google Scholar * Sang-Chul Park View author publications You can also search for this

author inPubMed Google Scholar * Nam-Eun Kim View author publications You can also search for this author inPubMed Google Scholar * Chol Shin View author publications You can also search for

this author inPubMed Google Scholar * Seung Ku Lee View author publications You can also search for this author inPubMed Google Scholar * Youngae Jung View author publications You can also

search for this author inPubMed Google Scholar * Dankyu Yoon View author publications You can also search for this author inPubMed Google Scholar * Hyeonjeong Kim View author publications

You can also search for this author inPubMed Google Scholar * Sanghyun Kim View author publications You can also search for this author inPubMed Google Scholar * Geum-Sook Hwang View author

publications You can also search for this author inPubMed Google Scholar * Sungho Won View author publications You can also search for this author inPubMed Google Scholar CORRESPONDING

AUTHORS Correspondence to Geum-Sook Hwang or Sungho Won. ETHICS DECLARATIONS COMPETING INTERESTS The authors declare no competing interests. ETHICS APPROVAL The protocol used in this study

was approved by the Institutional Review Board (IRB No. E1801/001-004) of Seoul National University. ADDITIONAL INFORMATION PUBLISHER’S NOTE Springer Nature remains neutral with regard to

jurisdictional claims in published maps and institutional affiliations. SUPPLEMENTARY INFORMATION SUPPLEMENTARY INFORMATION RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed under

a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate

credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article

are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and

your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this

license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Kim, K., Lee, S., Park, SC. _et al._ Role of an unclassified

_Lachnospiraceae_ in the pathogenesis of type 2 diabetes: a longitudinal study of the urine microbiome and metabolites. _Exp Mol Med_ 54, 1125–1132 (2022).

https://doi.org/10.1038/s12276-022-00816-x Download citation * Received: 14 July 2021 * Revised: 24 December 2021 * Accepted: 23 March 2022 * Published: 05 August 2022 * Issue Date: August

2022 * DOI: https://doi.org/10.1038/s12276-022-00816-x SHARE THIS ARTICLE Anyone you share the following link with will be able to read this content: Get shareable link Sorry, a shareable

link is not currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative