The incidence of type 2 diabetes (T2D) has been increasing globally, and a growing body of evidence links type 2 diabetes with altered microbiota composition. Type 2 diabetes is preceded by a long prediabetic state characterized by changes in various metabolic parameters. We tested whether the gut microbiome could have predictive potential for T2D development during the healthy and prediabetic disease stages. We used prospective data of 608 well-phenotyped Finnish men collected from the population-based Metabolic Syndrome in Men (METSIM) study to build machine learning models for predicting continuous glucose and insulin measures in a shorter (1.5 year) and longer (4 year) period. Our results show that the inclusion of the gut microbiome improves prediction accuracy for modeling T2D-associated parameters such as glycosylated hemoglobin and insulin measures. We identified novel microbial biomarkers and described their effects on the predictions using interpretable machine learning techniques, which revealed complex linear and nonlinear associations. Additionally, the modeling strategy carried out allowed us to compare the stability of model performance and biomarker selection, also revealing differences in short-term and long-term predictions. The identified microbiome biomarkers provide a predictive measure for various metabolic traits related to T2D, thus providing an additional parameter for personal risk assessment. Our work also highlights the need for robust modeling strategies and the value of interpretable machine learning.IMPORTANCE Recent studies have shown a clear link between gut microbiota and type 2 diabetes. However, current results are based on cross-sectional studies that aim to determine the microbial dysbiosis when the disease is already prevalent. In order to consider the microbiome as a factor in disease risk assessment, prospective studies are needed. Our study is the first study that assesses the gut microbiome as a predictive measure for several type 2 diabetes-associated parameters in a longitudinal study setting. Our results revealed a number of novel microbial biomarkers that can improve the prediction accuracy for continuous insulin measures and glycosylated hemoglobin levels. These results make the prospect of using the microbiome in personalized medicine promising.

Interindividual variation in the composition of the human gut microbiome was examined in relation to demographic and anthropometric traits, and to changes in dietary saturated fat intake and protein source. One hundred nine healthy men and women aged 21 to 65, with BMIs of 18 to 36, were randomized, after a two-week baseline diet, to high (15% total energy [E])- or low (7%E)-saturated-fat groups and randomly received three diets (four weeks each) in which the protein source (25%E) was mainly red meat (beef, pork) (12%E), white meat (chicken, turkey) (12%E), and nonmeat sources (nuts, beans, soy) (16%E). Taxonomic characterization using 16S ribosomal DNA was performed on fecal samples collected at each diet completion. Interindividual differences in age, body fat (%), height, ethnicity, sex, and alpha diversity (Shannon) were all significant factors, and most samples clustered by participant in the PCoA ordination. The dietary interventions did not significantly alter the overall microbiome community in ordination space, but there was an effect on taxon abundance levels. Saturated fat had a greater effect than protein source on taxon differential abundance, but protein source had a significant effect once the fat influence was removed. Higher alpha diversity predicted lower beta diversity between the experimental and baseline diets, indicating greater resistance to change in people with higher microbiome diversity. Our results suggest that interindividual differences outweighed the influence of these specific dietary changes on the microbiome and that moderate changes in saturated fat level and protein source correspond to modest changes in the microbiome.IMPORTANCE The microbiome has proven to influence health and disease, but how combinations of external factors affect the microbiome is relatively unknown. Diet can cause changes, but this is usually achieved by altering macronutrient ratios and has not focused on dietary protein source or saturated fat intake levels. In addition, each individual's unique microbiome profile can be an important factor during studies, and it has even been shown to affect therapeutic outcomes. We show here that the effects of individual differences outweighed the effect of experimental diets and that protein source is less influential than saturated fat level. This suggests that fat and protein composition, separate from macronutrient ratio and carbohydrate composition, is an important consideration in dietary studies.

The metabolic syndrome is a cluster of conditions that increase an individual's risk of developing diseases. Being physically active throughout life is known to reduce the prevalence and onset of some aspects of the metabolic syndrome. Furthermore, previous studies have demonstrated that an individual's gut microbiome composition has a large influence on several aspects of the metabolic syndrome. However, the mechanism(s) by which physical activity may improve metabolic health are not well understood. We sought to determine if endurance exercise is sufficient to prevent or ameliorate the development of the metabolic syndrome and its associated diseases. We also analyzed the impact of physical activity under metabolic syndrome progression upon the gut microbiome composition. Utilizing whole-body low-density lipoprotein receptor (LDLR) knockout mice on a "Western Diet," we show that long-term exercise acts favorably upon glucose tolerance, adiposity, and liver lipids. Exercise increased mitochondrial abundance in skeletal muscle but did not reduce liver fibrosis, aortic lesion area, or plasma lipids. Lastly, we observed several changes in gut bacteria and their novel associations with metabolic parameters of clinical importance. Altogether, our results indicate that exercise can ameliorate some aspects of the metabolic syndrome progression and alter the gut microbiome composition.

Background

It is unclear how high fructose consumption induces disparate metabolic responses in genetically diverse mouse strains.

Objective

We aimed to investigate whether the gut microbiota contributes to differential metabolic responses to fructose.

Methods

Eight-week-old male C57BL/6J (B6), DBA/2J (DBA), and FVB/NJ (FVB) mice were given 8% fructose solution or regular water (control) for 12 wk. The gut microbiota composition in cecum and feces was analyzed using 16S ribosomal DNA sequencing, and permutational multivariate ANOVA (PERMANOVA) was used to compare community across mouse strains, treatments, and time points. Microbiota abundance was correlated with metabolic phenotypes and host gene expression in hypothalamus, liver, and adipose tissues using Biweight midcorrelation. To test the causal role of the gut microbiota in determining fructose response, we conducted fecal transplants from B6 to DBA mice and vice versa for 4 wk, as well as gavaged antibiotic-treated DBA mice with Akkermansia for 9 wk, accompanied with or without fructose treatment.

Results

Compared with B6 and FVB, DBA mice had significantly higher Firmicutes to Bacteroidetes ratio and lower baseline abundance of Akkermansia and S24-7 (P < 0.05), accompanied by metabolic dysregulation after fructose consumption. Fructose altered specific microbial taxa in individual mouse strains, such as a 7.27-fold increase in Akkermansia in B6 and 0.374-fold change in Rikenellaceae in DBA (false discovery rate <5%), which demonstrated strain-specific correlations with host metabolic and transcriptomic phenotypes. Fecal transplant experiments indicated that B6 microbes conferred resistance to fructose-induced weight gain in DBA mice (F = 43.1, P < 0.001), and Akkermansia colonization abrogated the fructose-induced weight gain (F = 17.8, P < 0.001) and glycemic dysfunctions (F = 11.8, P = 0.004) in DBA mice.

Conclusions

Our findings support that differential microbiota composition between mouse strains is partially responsible for host metabolic sensitivity to fructose, and that Akkermansia is a key bacterium that confers resistance to fructose-induced metabolic dysregulation.

Objective

Lipocalin-2 (LCN2) is a secreted protein involved in innate immunity and has also been associated with several cardiometabolic traits in both mouse and human studies. However, the causal relationship of LCN2 to these traits is unclear, and most studies have examined only males.

Methods

Using adeno-associated viral vectors we expressed LCN2 in either adipose or liver in a tissue specific manner on the background of a whole-body Lcn2 knockout or wildtype mice. Metabolic phenotypes including body weight, body composition, plasma and liver lipids, glucose homeostasis, insulin resistance, mitochondrial phenotyping, and metabolic cage studies were monitored.

Results

We studied the genetics of LCN2 expression and associated clinical traits in both males and females in a panel of 100 inbred strains of mice (HMDP). The natural variation in Lcn2 expression across the HMDP exhibits high heritability, and genetic mapping suggests that it is regulated in part by Lipin1 gene variation. The correlation analyses revealed striking tissue dependent sex differences in obesity, insulin resistance, hepatic steatosis, and dyslipidemia. To understand the causal relationships, we examined the effects of expression of LCN2 selectively in liver or adipose. On a Lcn2-null background, LCN2 expression in white adipose promoted metabolic disturbances in females but not males. It acted in an autocrine/paracrine manner, resulting in mitochondrial dysfunction and an upregulation of inflammatory and fibrotic genes. On the other hand, on a null background, expression of LCN2 in liver had no discernible impact on the traits examined despite increasing the levels of circulating LCN2 more than adipose LCN2 expression. The mechanisms underlying the sex-specific action of LCN2 are unclear, but our results indicate that adipose LCN2 negatively regulates its receptor, LRP2 (or megalin), and its repressor, ERα, in a female-specific manner and that the effects of LCN2 on metabolic traits are mediated in part by LRP2.

Conclusions

Following up on our population-based studies, we demonstrate that LCN2 acts in a highly sex- and tissue-specific manner in mice. Our results have important implications for human studies, emphasizing the importance of sex and the tissue source of LCN2.