EPFL/LISP BXD Muscle Polar Metabolites HFD (Jun14) **

Download datasets and supplementary data files


The BXD genetic reference population is a recombinant inbred panel descended from crosses between the C57BL/6 (B6) and DBA/2 (D2) strains of mice, which segregate for about 5 million sequence variants. Recently, some these variants have been established with effects on general metabolic phenotypes such as glucose response and bone strength. In this study, we examined genetic variants across 40 strains of BXD and the two founder lines, in addition to a major environmental influence—long term feeding with chow diet (CD) or high fat diet (HFD)—to see how metabolic gene expression varies by genotype and environment, and gene-by-environment interactions. It is difficult to say whether the major muscle phenotypes were broadly affected by HFD directly, or whether the phenotypic variance was as a consequence of the induced weight gain and other deleterious effects of the HFD (e.g. see phenotype traits ID 17717 and 17718 for nighttime activity in CD and HFD cohorts, respectively).

About cases

40 strains of the BXD family (BXD43 – BXD103) and both parental strains (C57BL/6 and DBA/2) were born and raised at the EPFL in Switzerland prior to inclusion in this study. For each strain, roughly 10 male animals were born and weaned at 3 weeks of age. These cohorts were then separated evenly into two cohorts at 8 weeks of age: up to 5 animals per strain on a chow diet (6% kcal/fat, 20% protein, 74% carbohydrate) and up to 5 animals per strain on high fat diet (60% kcal/fat, 20% protein, 20% carbohydrate). A total of 195 CD animals and 189 HFD animals comprised these original 82 cohorts (42 on CD, 40 on HFD, all 40 overlap except for 2 strains, BXD60 and BXD92a (aka BXD65b), which are unique to CD). For the next 8 weeks, animals adjusted to the diet and housing situation. From 16 to 24 weeks of age, animals were phenotyped for respiration, oral glucose response, cold tolerance, basal activity, VO2max exercise, and voluntary exercise. Animals were communally housed until the last 5 weeks of the experiment, when the animals could rest, undisturbed except for a weekly cage change and weighing. Animals were fasted overnight prior to sacrifice, which occurred between 9am and 11am after isoflurane anesthesia and perfusion.

About tissue

The quadriceps were taken near the end of the pipeline, roughly 6–10 minutes after completion of perfusion. Quadriceps were collected by cutting evenly across the upper axis of the femoral bone. Quadriceps muscle lying below or beside this femoral axis was not taken. The quadriceps was then placed in liquid nitrogen. At a later date, quadriceps were retrieved from -80 storage and shattered in liquid nitrogen. Roughly ~100 mg fragments were taken at random for mRNA preparation for every individual. To account potential differences in the tissue taken (as the tissue was not pulverized into powder), all ~5 animals per cohort had their RNA prepared, and then were pooled evenly (by µg of RNA) into a single RNA sample for each cohort. These pooled RNA samples of approximately 30 µg RNA were then purified using RNEasy, then sent out for array analysis. All RIN values were > 8.0. 

Tissue preparation was done in two different steps: the first half of the cohorts were run in October 2011, while the second half was run in August 2012. This tissue was the first microarray run at this platform, and thus the quadriceps served as a pilot to ensure that data quality would be good. The first half of the cohorts are: CD: C57, DBA, 44, 45, 51, 55, 61, 62, 63, 66, 70, 73, 75, 80, 83, 87, 89, 90, 96. HFD: C57, DBA, 44, 45, 51, 55, 61, 62, 63, 66, 70, 73, 75, 80, 83, 87, 90, 96, 100. The second half of the cohorts were the remainder. The preparation of all samples was done by the same technician with the same technique, other than the year time differential

About platform

All arrays were Affymetrix Mouse Gene 1.0 ST, run in either October 2011 or August 2012 at the University of Tennessee Health Science Center. All .CEL files were normalized together after the second batch of data was collected.

About data processing

In general, the array data that we put in GeneNetwork has be logged and then z normalized, but instead of leaving the mean at 0 and the standard deviation of 1 unit, we shift up to a mean of 8 units and increase the spread by having an standard deviation of 2 units (what we call 2Z + 8 normalized data). This removes negative values from the tables.


Williams EG, Andreux P, Auwerx J


Pirinen et al., Cell Metabolism 2014. Pharmacological inhibition of poly(ADP-ribose) polymerases improves fitness and mitochondrial function in skeletal muscle. PMID 24814482.

Williams et al., PLoS Genetics 2014. An evolutionarily conserved role for the aryl hydrocarbon receptor in the regulation of movement. PMID 25255223.


We would like to thank Cristina Cartoni, Sébastien Lamy, and Charles Thomas at the Center of Phenogenomics (CPG, EPFL) for help in establishing and phenotyping the BXD mice. We thank the Molecular Resource Center of Excellence at The University of Tennessee Health Science Center processing all microarrays. The work in the Auwerx lab was supported by the Ecole Polytechnique Federale de Lausanne, the EU Ideas program (AdG-23138 and AdG-322424), the NIH (R01HL106511-01A and R01AG043930), and the Swiss National Science Foundation (31003A- 124713 and 31003A-125487 and CSRII3-1362).