Microbiome variations induced by delta9-tetrahydrocannabinol predict weight reduction in obese mice

IntroductionObesity and high-fat diets induce consistent alterations in gut microbiota composition. Observations from epidemiological reviews and experiments also illustrate weight regulation effects of delta9-tetrahydrocannabinol (THC) with microbiome shifts. Therefore, we investigated the weight-loss potential of THC in obese mice models and to elucidate the contribution of specific gut microbiome changes in THC-induced weight loss.MethodsHigh-fat diet induced obese mice were treated with oral THC supplementation for two weeks and compared with controls. In addition to measuring weight, fecal samples were obtained at various timepoints, sequenced for bacterial 16s rRNA content and analyzed using QIIME2. Alpha and beta diversity were computed followed by linear mixed effects (LME) modeling of bacterial relative abundance relationship to THC treatment and weight change.ResultsIn both male and female mice, the THC group had significantly greater average weight loss than controls (−17.8% vs. −0.22%, p<0.001 and −13.8% vs. +2.9%, p<0.001 respectively). Male mice had 8 bacterial taxonomic features that were both significantly different in relative abundance change over time with THC and correlated with weight change. An LME model using three bacterial features explained 76% of the variance in weight change with 24% of variation explained by fixed effects of feature relative abundance alone. The model also accurately predicted weight change in a second male mouse cohort (R=0.64, R2=0.41, p=<0.001). Female mice had fewer significant predictive features and were difficult to model, but the male-produced 3-feature model still accurately predicted weight change in the females (R=0.66, R2=0.44, p<0.001).ConclusionUsing a stepwise feature selection approach, our results indicate that sex-specific gut microbiome composition changes play some role in THC-induced weight loss. Additionally, we illustrated the concept of microbiome feature-based modeling to predict weight changes.