Genetic and genomic analysis of oxygen consumption in mice

We estimated genetic parameters of oxygen consumption (OC), OC per metabolic body weight (OCMBW), and body weight at three through eight weeks of age in divergently selected mice populations, with an animal model considering maternal genetic, common litter environmental, and cytoplasmic inheritance effects. Cytoplasmic inheritance was considered based on maternal lineage information. For OC, estimated direct heritability was moderate (0.32) and estimated maternal heritability and proportion of the variance of cytoplasmic inheritance effects to the phenotypic variance were very low (both <0.03), implying that causal genes for OC could be located on autosomes. To assess this hypothesis, we attempted to identify possible candidate causal genes by performing pool-seq using pooled DNA samples from mice in high and low OC lines and selective signature detection. We made a list of possible candidate causal genes for OC, including those relating to electron transport chain and ATP-binging proteins (Ndufa12, Sdhc, Atp10b, etc.), Prr16 encoding Largen protein, Cry1 encoding a key component of the circadian core oscillator, and so on. The results could contribute to elucidate the genetic mechanism of OC, an indicator for maintenance energy requirement and therefore feed efficiency.

with their own phenotypic records belonged to one of the 17 maternal lineages. On the other hand, G17 11 mice with their own phenotypic records belonged to only 1 and 4 lineages in HOC and LOC lines, 12 respectively. 13

| Statistical analysis
14 Two mixed linear models, namely models 1 and 2, were exploited to estimate genetic parameters. where y is the vector of phenotypic records; b is the vector of fixed effects; a is the vector of direct 18 additive genetic effects; e is the vector of errors; X and Z is the design matrices. In this study, the data 19 from the two lines were merged and then simultaneously analyzed. For OC and OCMBW, fixed 20 effects were sex, generation × line, and days of age at measuring OC (linear and quadratic covariates). 21 For BW traits, fixed effects were sex and generation × line. The mean and variance-covariance of the 22 random effects were as follows: trait heritability, and two-trait analysis was performed to estimate genetic correlation between two traits. 5 Genetic trends for OC, OCMBW, and BW8 were estimated using EBVs from the single-trait analysis. 6 Model 2 was as follows:   The absolute values of the estimated genetic correlations of BW traits with OC and OCMBW 7 were larger when the age at measuring BW was closer to that at measuring OC (Table 3). Using an

23
Estimated maternal heritabilities were about 0.1 or lower for all traits and smaller than twice the standard errors (Table 4). Direct-maternal genetic correlations were estimated to be negative for OC 1 and OCMBW and positive for BW traits, but their standard errors were large. These results suggest that 2 there is little need to consider maternal genetic effects. Accurate genetic parameter estimation requires a 3 sufficient number of records, as well as an appropriate model (Thompson, 1976) and data structure 4 (Gerstmayr, 1992). It has been argued that negative estimates of direct-maternal genetic correlations caused by the fact that all same-sex littermates were kept in the same cage after weaning. In this 4 population, full-sibs of the same sex were kept in the same cage after weaning from G0 to G10, and 5 therefore, there might be possible confounding of the effects of rearing environments before and after 6 weaning on BW traits. In addition, the number of pups was adjusted to 6 in principle, but some 7 exceptions were observed. Data were also analyzed after excluding such exceptions, and the results 8 were similar with those obtained when the data were not excluded (results not shown). Therefore, it 9 was considered unlikely that differences in the number of pups were not the main factor for the 10 common litter environmental effects in this study.

11
Estimated values of the proportion of the variance of the cytoplasmic inheritance effects, 12 regarding as mitochondrial DNA, were very low for all traits (Table 4). Also, adding the effects of  According to the moderate direct heritability and negligible proportion of the variance of cytoplasmic 10 inheritance effects (Table 4), we speculated that the genetic causes of the difference in OC between 11 lines are likely to exist on the nuclear genome. In order to test this hypothesis as much as possible, we 12 conducted a selective signature analysis using the results of pool-seq. Accumulating insight into genes 13 relating to OC not only could contribute to develop easier approach for estimating MER and RFI (e.g., per autosome are shown in Figures S1 through S19. We adopted the Bonferroni-corrected value (0.05/4,147,085 = 1.2×10 −8 ) as a conservative threshold, but aiming to further reduce false positives, 1 genomic regions with ≥5 consecutive variants with p-values lower than the threshold were detected as 2 selective signatures. We finally extracted the 2,792 genes as possible candidate causal genes relating to 3 OC (Table S1).

17
Genes previously reported as candidate ones for feed efficiency and related traits were also included,  18 We estimated genetic parameters using an animal model including maternal genetic, common litter 19 environmental, and cytoplasmic inheritance effects for OC, OCMBW, and BW at 3 through 8 weeks of