Diet and exercise orthogonally alter the gut microbiome and reveal independent associations with anxiety and cognition

Abstract

Background: The ingestion of a high-fat diet (HFD) and the resulting obese state can exert a multitude of stressors on the individual including anxiety and cognitive dysfunction. Though many studies have shown that exercise can alleviate the negative consequences of a HFD using metabolic readouts such as insulin and glucose, a paucity of well-controlled rodent studies have been published on HFD and exercise interactions with regard to behavioral outcomes. This is a critical issue since some individuals assume that HFD-induced behavioral problems such as anxiety and cognitive dysfunction can simply be exercised away. To investigate this, we analyzed mice fed a normal diet (ND), ND with exercise, HFD diet, or HFD with exercise.

Results: We found that mice on a HFD had robust anxiety phenotypes but this was not rescued by exercise. Conversely, exercise increased cognitive abilities but this was not impacted by the HFD. Given the importance of the gut microbiome in shaping the host state, we used 16S rRNA hypervariable tag sequencing to profile our cohorts and found that HFD massively reshaped the gut microbial community in agreement with numerous published studies. However, exercise alone also caused massive shifts in the gut microbiome at nearly the same magnitude as diet but these changes were surprisingly orthogonal. Additionally, specific bacterial abundances were directly proportional to measures of anxiety or cognition.

Conclusions: Thus, behavioral domains and the gut microbiome are both impacted by diet and exercise but in unrelated ways. These data have important implications for obesity research aimed at modifications of the gut microbiome and suggest that specific gut microbes could be used as a biomarker for anxiety or cognition or perhaps even targeted for therapy.

Figure 1

Effects of diet and exercise on body weight, anxiety, and cognitive behavior. (A) HFD-fed mice were significantly heavier starting at 4 weeks of treatment (ND vs HFD) while exercised mice weighed significantly less at 5 weeks of treatment for ND groups (ND vs ND + exercise, *p < 0.05) and at 8 weeks of treatment for HFD groups (HFD vs HFD + exercise, * p < 0.05). HFD-fed mice were more anxious as measured in the Light/Dark exploration assay for both percentage of time spent in the lit compartment (B) and distance traveled while in the lit compartment (C). Exercised mice had enhanced learning and memory measured in the contextual fear conditioning assay (D). Body weight analyzed by repeated measures ANOVA with post-hoc t-test and significance defined as p < 0.05. Behavioral data was analyzed by two-way ANOVA with post-hoc Fisher’s LSD with significance indicated as *p < 0.05, **p < 0.01, and ***p < 0.001. All data presented as mean +/- SEM from n=10/group.

Figure 2

Changes in the relative abundances of gut bacteria by HFD or exercise. (A) HFD caused a bloom of OTU115 from the genus Streptococcus that returned to nearly undetectable levels in HFD + exercise mice. Bacterial of phyla Firmicutes (B), Bacteroidetes (C), and Tenericutes (D) of the gut microbiome were significantly altered by both HFD and exercise. Data was analyzed by two-way ANOVA with post-hoc Holm-Sidak t-tests with significance indicated as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. All data presented as mean +/- SEM from n=10/group.

Figure 3

Multidimensional analysis of diet and exercise reveals orthogonal changes in the gut microbiome. This analysis in multidimensional space demonstrates clear segregation of each of the four groups of mice with no overlap between groups.

Figure 4

Heat map of global analysis of diet and exercise reveals multiple distinct signatures with clear dietary effects and secondary exercise effects.

Figure 5

Several bacteria significantly associate with anxiety measures. Relative abundances of OTU69, 90, and 97 positively correlated with % time in light in the Light/Dark test (A-C) while OTU17, 30, and 72 negatively correlated with % time in light (D-F). Most likely family or genus shown on graph. Color scheme of data points follows the format as in Figure 1 (ND = blue, ND + exercise = green, HFD = red, and HFD + exercise in orange). Linear regression analysis for individual mice was performed with R² values indicating goodness of fit and p values (indicated on graph) for slope calculated by F test with Benjamini-Hochberg correction for false discovery rate.

Figure 6

Several bacteria specifically associate with cognitive measures. (A-D) The relative abundance of OTU’s 39 and 82 negatively correlated with % time freezing in the contextual fear conditioning test (A, B) while OTU79 and 57 positively correlated with % time freezing (C, D). Most likely family or genus shown on graph. Color scheme of data points follows the format as in Figure 1 (ND = blue, ND + exercise = green, HFD = red, and HFD + exercise in orange). Linear regression analysis for individual mice was performed with R² values indicating goodness of fit and uncorrected p values (indicated on graph) for slope calculated by F test.