Christopher D. Moyes

Cytochrome c oxidase (COX) subunit 4 has two paralogs in most vertebrates. The mammalian COX4-2 gene is hypoxia responsive, and the protein has a disrupted ATP-binding site that confers kinetic properties on COX that distinguish it from COX4-1. The structure-function of COX4-2 orthologs in other vertebrates remains uncertain. Phylogenetic analyses suggest the two paralogs arose in basal vertebrates, but COX4-2 orthologs diverged faster than COX4-1 orthologs. COX4-1/4-2 protein levels in tilapia tracked mRNA levels across tissues, and did not change in hypoxia, arguing against a role for differential post-translational regulation of paralogs. The heart, and to a lesser extent the brain, showed a size-dependent shift from COX4-1 to COX4-2 (transcript and protein). ATP allosterically inhibited both velocity and affinity for oxygen in COX assayed from both muscle (predominantly COX4-2) and gill (predominantly COX4-1). We saw some evidence of cellular and subcellular discrimination of COX4 paralogs in heart. In cardiac ventricle, some non-cardiomyocyte cells were COX positive but lacked detectible COX4-2. Within heart, the two proteins partitioned to different mitochondrial subpopulations. Cardiac subsarcolemmal mitochondria had mostly COX4-1 and intermyofibrillar mitochondria had mostly COX4-2. Collectively, these data argue that, despite common evolutionary origins, COX4-2 orthologs of fish show unique patterns of subfunctionalization with respect to transcriptional and posttranslation regulation relative to the rodents and primates that have been studied to date.

Cold acclimation in fish typically increases muscle mitochondrial enzymes. In mammals, stressors that increase mitochondrial content are mediated though transcriptional regulators, including nuclear respiratory factor-1 (NRF-1). Focusing on the goldfish gene for cytochrome c oxidase (COX) subunit 4-1, we analyzed the regulatory regions in various contexts to identify a mechanistic link between NRF-1 and cold-induced mitochondrial proliferation. Promoter analysis implicated two putative NRF-1 sites: one in the proximal promoter and a second in exon 1, which encodes the 5′ untranslated region (5′-UTR). Transfection into mouse myoblasts showed that deletion of a region that included the proximal NRF-1 site reduced promoter activity by 30%, however mutagenesis of the specific sequence had no effect. Thermal sensitivity analyses performed in rainbow trout gonadal fibroblasts (RTG-2) showed no effect of temperature (4 vs. 19°C) on reporter gene expression. Likewise, reporters injected into muscle of thermally-acclimated goldfish (4 vs. 26°C) showed no elevation in expression. There was no difference in thermal responses of COX4-1 promoter reporters constructed from homologous regions of eurythermal goldfish and stenothermal zebrafish genes. NRF-1 chromatin immunoprecipitation of thermally acclimated goldfish muscle showed no temperature effect on NRF-1 binding to either the proximal promoter or 5′-UTR. It remains possible that the cold-induced up-regulation of COX4-1 expression is a result of NRF-1 binding to distal regulatory regions or through indirect effects on other transcription factors. However, the proximal promoter does not appear to play a role in mediating the thermal response of the COX4-1 gene in fish.

Vertebrates possess two paralogs of cytochrome c oxidase (COX) subunit 4: a ubiquitous COX4-1 and a hypoxia-linked COX4-2. Mammalian COX4-2 is thought to have a role in relation to fine-tuning metabolism in low oxygen levels, conferred through both structural differences in the subunit protein structure and regulatory differences in the gene. We sought to elucidate the pervasiveness of this feature across vertebrates. The ratio of COX4-2/4-1 mRNA is generally low in mammals, but this ratio was higher in fish and reptiles, particularly turtles. The COX4-2 gene appeared unresponsive to low oxygen in nonmammalian models (zebrafish, goldfish, tilapia, anoles, and turtles) and fish cell lines. Reporter genes constructed from the amphibian and reptile homologues of the mammalian oxygen-responsive elements and hypoxia-responsive elements did not respond to low oxygen. Unlike the rodent ortholog, the promoter of goldfish COX4-2 did not respond to hypoxia or anoxia. The protein sequences of the COX4-2 peptide showed that the disulfide bridge seen in human and rodent orthologs would be precluded in other mammalian lineages and lower vertebrates, all of which lack the requisite pair of cysteines. The coordinating ligands of the ATP-binding site are largely conserved across mammals and reptiles, but in Xenopus and fish, sequence variations may disrupt the ability of the protein to bind ATP at this site. Collectively, these results suggest that many of the genetic and structural features of COX4-2 that impart responsiveness and benefits in hypoxia may be restricted to the Euarchontoglires lineage that includes primates, lagomorphs, and rodents.

Do the transcriptional mechanisms that control an individual's mitochondrial content, PGC1α (peroxisome proliferator-activated receptor γ coactivator-1α) and NRF1 (nuclear respiratory factor-1), also cause differences between species? We explored the determinants of cytochrome c oxidase (COX) activities in muscles from 12 rodents differing 1,000-fold in mass. Hindlimb muscles differed in scaling patterns from isometric (soleus, gastrocnemius) to allometric (tibialis anterior, scaling coefficient = −0.16). Consideration of myonuclear domain reduced the differences within species, but interspecies differences remained. For tibialis anterior, there was no significant scaling relationship in mRNA/g for COX4-1, PGC1α, or NRF1, yet COX4-1 mRNA/g was a good predictor of COX activity (r² = 0.55), PGC1α and NRF1 mRNA correlated with each other (r² = 0.42), and both could predict COX4-1 mRNA (r² = 0.48 and 0.52) and COX activity (r² = 0.55 and 0.49). This paradox was resolved by multivariate analysis, which explained 90% of interspecies variation, about equally partitioned between mass effects and PGC1α (or NRF1) mRNA levels, independent of mass. To explore the determinants of PGC1α mRNA, we analyzed 52 mammalian PGC1α proximal promoters and found no size dependence in regulatory element distribution. Likewise, the activity of PGC1α promoter reporter genes from 30 mammals showed no significant relationship with body mass. Collectively, these studies suggest that not all muscles scale equivalently, but for those that show allometric scaling, transcriptional regulation of the master regulators, PGC1α and NRF1, does not account for scaling patterns, though it does contribute to interspecies differences in COX activities independent of mass.

SUMMARY: Mitochondrial content, central to aerobic metabolism, is thought to be controlled by a few transcriptional master regulators, including nuclear respiratory factor 1 (NRF-1), NRF-2 and peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α). Though well studied in mammals, the mechanisms by which these factors control mitochondrial content have been less studied in lower vertebrates. We evaluated the role of these transcriptional regulators in seasonal changes in white muscle cytochrome c oxidase (COX) activity in eight local fish species representing five families: Centrarchidae, Umbridae, Esocidae, Gasterosteidae and Cyprinidae. Amongst centrarchids, COX activity was significantly higher in winter for pumpkinseed (2-fold) and black crappie (1.3-fold) but not bluegill or largemouth bass. In esociforms, winter COX activity was significantly higher in central mudminnow (3.5-fold) but not northern pike. COX activity was significantly higher in winter-acclimatized brook stickleback (2-fold) and northern redbelly dace (3-fold). Though mudminnow COX activity increased in winter, lab acclimation to winter temperatures did not alter COX activity, suggesting a role for non-thermal cues. When mRNA was measured for putative master regulators of mitochondria, there was little evidence for a uniform relationship between COX activity and any of NRF-1, NRF-2α or PGC-1α mRNA levels Collectively, these studies argue against a simple temperature-dependent mitochondrial response ubiquitous in fish, and suggest that pathways which control mitochondrial content in fish may differ in important ways from those of the better studied mammals.