Experimental Diets Dictate the Metabolic Benefits of Lactobacillus Strains in Obesity

Background: Growing evidence supports the use of probiotics to prevent or mitigate obesity-related dysmetabolism and non-alcoholic fatty liver disease (NAFLD). However, frequent reports of responders versus non-responders to probiotic treatment warrant a better understanding of key modifiers of host-microbe interactions. The influence of host diet on probiotic efficacy, in particular against metabolic diseases, remains elusive. Method: We fed C57BL6/J mice a low fat reference diet or one of two energy-matched high fat and high sucrose diets for 12 weeks; a classical high fat diet (HFD) and a customized fast food-mimicking diet (FFMD). During the studies, mice fed either obesogenic diet were gavaged daily with one of two probiotic Lactobacillus strains, namely Limosilactobacillus reuteri (L. reuteri) or Lacticaseibacillus paracasei subsp. paracasei (L. paracasei), or vehicle. Results: The tested probiotics exhibited a reproducible efficacy but dichotomous response according to the obesogenic diets used. Indeed, L. paracasei prevented weight gain, improved insulin sensitivity, and protected against NAFLD development in mice fed HFD, but not FFMD. Conversely, L. reuteri improved glucoregulatory capacity, reduced NAFLD development, and increased distal gut bile acid levels associated with changes in predicted functions of the gut microbiota exclusively in the context of FFMD-feeding. Conclusion: We found that the probiotic efficacy of two Lactobacillus strains is highly dependent on experimental obesogenic diets. These findings highlight the need to carefully consider the confounding impact of diet in order to improve both the reproducibility of preclinical studies and their clinical research translatability. ### Competing Interest Statement This study was partly funded by Novozymes A/S (NZ) which provided the bacterial strains and was granted a patent related to the findings (WO20084051). NZ contributed to study design and parts of the analyses but had no role in data integration, interpretation and conclusion.