What determines an individual's lifespan? The intuitive answers of "genetics" or "environment" do not tell the full story-even isogenic populations of C. elegans reared in a controlled environment show the same degree of inter-individual variation in lifespan as that seen in outbred human populations. One hypothesis is that individuals are driven toward different fates by stochastic differences in the expression of key regulatory genes. We recently showed that the expression levels of 10 out of the 22 microRNA promoter::GFP constructs we examined were sufficient to predict an individual's future lifespan. At least two of these GFPs predict lifespan independent of the activity of
daf-16, and at least three report on lifespan redundantly despite being expressed in distinct tissues. We wondered whether these results might be evidence of a global gene expression state that determines (or is determined by) an individual's remaining lifespan. To address this, we set out to understand the global transcriptional states underlying high vs. low expression of these lifespan-predictive GFPs. We performed RNA-seq on populations sorted by the GFP intensity of several of these predictive markers, as well as intestinal autofluorescence, a phenomenological biomarker of aging. We found that, regardless of biomarker, the differentially expressed genes between prospectively long- vs. short-lived populations were extremely similar. However, we also noted that these differences overlapped significantly with genes that change over time during chronological aging. We reasoned that this was due to the "apparent age" of the worms; that is, worms predicted to be long-lived had gene expression reflective of chronologically younger animals, while worms predicted to be short-lived showed premature genetic hallmarks of aging. We then asked whether this accelerated or retarded "apparent age" was the only difference between prospectively long- vs. short-lived populations. By first aligning each population to a high-resolution transcriptional time course, we account for and remove the effects of "apparent age," revealing a signature that specifically reflects long vs. short future lifespan. Surprisingly, we found that, for all the predictive markers we tested, the expression of germline-related genes is highly related to future lifespan, even after controlling for the "apparent age" of the transcriptome.