Mitochondrial H2O2 Generation Via the Acyl-CoA Dehydrogenases

Yuxun Zhang, Megan E. Beck, Sivakama S. Bharathi, Xue-Jun Zhao, and Eric S. Goetzman*

Children’s Hospital of Pittsburgh of UPMC, Pittsburgh, PA, USA

*Corresponding author:  Eric Goetzman, PhD; Rangos 5117, Children’s Hospital of Pittsburgh, 4401

Penn Ave, Pittsburgh, PA, 15224;, 412-692-7952

Fatty acid oxidation (FAO)-driven H2O2 has been shown to be a major source of oxidative stress in liver and kidney of rodent models of obesity/diabetes. One mechanism appears to be the direct leakage of electrons from mitochondrial FAO flavoproteins to oxygen. Recently, very long-chain acyl-CoA dehydrogenase (VLCAD) was shown to have partial oxidase activity and to produce H2O2 in the livers of mice maintained on a high-fat diet. Rodents are also known to highly express long-chain acyl-CoA dehydrogenase (LCAD), another flavoprotein that functionally overlaps with VLCAD. In the present work, we sought to compare the dehydrogenase activities and H2O2-producing capacities of VLCAD and LCAD using isolated liver mitochondria, recombinant enzymes, and the human hepatic cell line HepG2. First, the rate of palmitoylcarnitine-stimulated H2O2 release was found to be significantly lower from LCAD-/- mouse liver mitochondria compared to wildtype mitochondria, while release of H2O2 from VLCAD-/- mitochondria was significantly higher. This suggests that LCAD, rather than VLCAD, is a source of H2O2 in mouse liver. At the same time, mitochondrial respiration was significantly reduced in VLCAD-/- liver but was unaltered in LCAD-/- liver, indicating that loss of VLCAD—but not LCAD—can induce secondary respiratory-chain defects. In vitro, recombinant mouse LCAD (mLCAD) exhibited stronger dehydrogenase activity than mouse VLCAD (mVLCAD) with palmitoyl-CoA as substrate, and produced H2O2 at a significantly higher rate. We then investigated the activities of human LCAD (hLCAD) and human VLCAD (hVLCAD). Recombinant hLCAD was significantly less active than both mLCAD and hVLCAD as a dehydrogenase, but produced more H2O2, particularly with stearoyl-CoA as substrate. The Vmax of recombinant hLCAD assayed as an oxidase was about 10-fold higher than the Vmax of hVLCAD assayed as an oxidase, but still 50-fold lower than the Vmax of the peroxisomal acyl-CoA oxidase-1 (ACOX1).

Recombinant hLCAD’s turnover number as an oxidase was approximately 100-fold lower than its turnover number as a dehydrogenase, indicating that the oxidase function is a minor side reaction. Yet, stable transfection of an hLCAD expression construct into HepG2 cells, which express very little endogenous LCAD, doubled the rate of fatty acid-stimulated H2O2 release into the media. Finally, by western-blotting human primary hepatocyte lysates against a standard curve of recombinant LCAD/VLCAD, we estimated that the amount of hLCAD antigen is about half the amount of hVLCAD antigen in normal human liver. Human kidney and pancreas also were observed to co-express LCAD and VLCAD. This suggests that several human tissues may have the potential for generating mitochondrial H2O2 production through hLCAD. Based on the observation of increased H2O2 release from hepatic mitochondria in the VLCAD-/- mouse model, we speculate that H2O2 generation through LCAD could become exacerbated in human tissues in the absence of VLCAD, such as in patients with VLCAD deficiency.