Controlling the uncontrolled variation in the diet induced obese mouse by microbiomic characterization.

Group sizes in an animal study are calculated from estimates on variation, effect, power and significance level. Much of the variation in glucose related parameters of the diet-induced obese (DIO) mouse model is due to inter-individual variation in gut microbiota composition. In addition, standard tandem repeats (STRs) in the non-coding DNA shows that inbred mice are not always homogenic. C57BL/6NTac (B6NTac) mice from Taconic and C57BL/6NRj (B6NRj) mice from Janvier Labs were fed a high calorie diet and treated with liraglutide. The fecal microbiota was sequenced before high-calorie feeding (time 1) and after diet-induced obesity instantly before liraglutide treatment (time 2) and mice were divided into clusters on the basis of their microbiota. Although liraglutide in both sub-strains alleviated glucose intolerance and reduced body weight, in a one-way ANOVA a borderline reduction in glycosylated hemoglobin (HbA1c) could only be shown in B6NTac mice. However, if the microbiota clusters from time 1 or time 2 were incorporated in a two-way ANOVA, the HbA1c effect was significant in B6NTac mice in both analyses, while this did not change anything in B6NRj mice. In a one-way ANOVA the estimated group size needed for a significant HbA1c effect in B6NTac mice was 42, but in two-way ANOVAs based upon microbiota clusters of time 1 or time 2 it was reduced to 21 or 12, respectively. The lowering impact on glucose tolerance was also powered by incorporation of microbiota clusters of both times in both sub-strains. B6NRj had up to six, while B6NTac had maximum three alleles in some of their STRs. In B6NRj mice in 28.8% of the STRs the most prevalent allele had a gene frequency less than 90%, while this was only 6.6% in the B6NTac mice. However, incorporation of the STRs with the highest number of alleles or the most even distribution of frequencies in two-way ANOVAs only had little impact on the outcome of data evaluation. It is concluded that the inclusion of microbiota clusters in a two-way ANOVA in the evaluation of the glucose related effects of an intervention in the DIO mouse model might be an efficient tool for increasing power and reducing group sizes in mouse sub-strains, if these have a microbiota, which influences these parameters.