renal cancer (Supplementary Data 16). Interestingly, four mitochondrial genes, Hspd1, Hspa9, Clpx and Lonp1 that all encode components of the mitochondrial unfolded protein response (UPRmt) (ref. 19)—a still poorly characterized mitochondrial stress response pathway in mammals—show strong association with Fh1 (Fig. 3d and Supplementary Data 16). There is, furthermore, a significant correlation between Fh1 transcript levels and principal component scores of a group of UPRmt genes in mouse (Fig. 3e). In contrast, no genes involved in the cytoplasmic heat shock response (HSR) or the ER unfolded protein response (UPRer) are associated with Fh1, indicating a selective association between Fh1 and UPRmt in mammals (Fig. 3d). To validate this association, we examined the phenotypic impact of the highly conserved Caenorhabditis elegans Fh1 ortholog, fum-1 (86% sequence similarity) on unfolded protein responses. RNAi against fum-1 causes robust activation of the mitochondrial chaperone hsp-6 induces green fluorescent protein (hsp-6::gfp) reporter, indicative of the activation of the UPRmt (Fig. 3f). The response was organelle-specific, and fum-1 RNAi does not induce either hsp-4::gfp or hsp-16.2::gfp, reporters related to the UPRer or HSR, respectively (Fig. 3f). Thus, in the BXD family, a decrease of fumarate hydratase leads to a specific mitochondrial phenotype, characterized by an UPRmt.
