Core Leader: Gary Churchill
The Statistical Core provides data analysis support for ongoing studies conducted within the JAX NSC including the development of new analysis methods and software necessary to support these projects. This core also disseminates JAX NSC data in conjunction with the Mouse Phenome Database and through the development of web services and interfaces to provide access to large-scale data resources. In addition the core also provide experimental design and analysis support to the aging research community.
In collaboration with Dr. Steven Gygi (Harvard University), we carried out proteomic profiling of 188 heart and kidney samples from our DO cross-sectional study tissue bank. We identified functional categories of genes that change with age and mapped genetic associations that are age-dependent. We demonstrated that effects of age on the proteome are not mediated through corresponding changes in mRNA. In contrast, proteins that differ between the sexes are primarily mediated through RNA. Our results suggest that the effects of age on the proteome are not genetically programmed but rather reflect accumulation of post-translational modifications over time. The data have been released with interactive analysis tools on the JAX NSC website. This study is in preparation for publication.
We have compiled lifespan data on more than 2200 DO mice from our own JAX NSC studies and from two independent studies carried out at JAX (by Drs. Karen Svenson and David Harrison) to investigate the responses to dietary restriction and rapamycin in DO mice (Figure 1). We carried out genetic mapping of lifespan as a meta-analysis across these studies and established that lifespan is heritable in a range similar to human studies (15-25%). Our findings suggest that many, perhaps hundreds, of individual loci influence lifespan. However, unlike human studies, the combined statistically significant loci in DO mice explain up to 50% of the genetic variation in lifespan in DO mice.
We have established sample size guidelines for intervention studies using DO mice. We computed sample sizes required to detect a change in lifespan between groups of treated and control animals using data from our DO longitudinal study (Figure 2). Due to the added genetic variability in lifespan, DO studies require approximately twice as many animals compared to a C57BL/6J study to detect the same magnitude of change. However, if we consider the possibility that the intervention effect may vary across strains, it makes economic sense to carry out initial studies on the outbred population rather than testing multiple inbred strains.
We are evaluating phenotypic (healthspan) data from our longitudinal study to identify potential predictors of lifespan. To date, our findings suggest that long-term predictions (beyond six months) are not better than chance expectation, despite many parameters that show trends with age. However, we can accurately predict death within the next six months (Fig. 3). Findings from the ongoing project suggest that while healthspan para-meters provide reliable indicators of “death’s door,” we have yet to discover the key to long-range prediction of lifespan.