Complex I assembly function and fatty acid oxidation enzyme activity of ACAD9 both contribute to disease severity in ACAD9 deficiency

Abstract

Acyl-CoA dehydrogenase 9 (ACAD9) is an assembly factor for mitochondrial respiratory chain Complex I (CI), and ACAD9 mutations are recognized as a frequent cause of CI deficiency. ACAD9 also retains enzyme ACAD activity for long-chain fatty acids in vitro, but the biological relevance of this function remains controversial partly because of the tissue specificity of ACAD9 expression: high in liver and neurons and minimal in skin fibroblasts. In this study, we hypothesized that this enzymatic ACAD activity is required for full fatty acid oxidation capacity in cells expressing high levels of ACAD9 and that loss of this function is important in determining phenotype in ACAD9-deficient patients. First, we confirmed that HEK293 cells express ACAD9 abundantly. Then, we showed that ACAD9 knockout in HEK293 cells affected long-chain fatty acid oxidation along with Cl, both of which were rescued by wild type ACAD9. Further, we evaluated whether the loss of ACAD9 enzymatic fatty acid oxidation affects clinical severity in patients with ACAD9 mutations. The effects on ACAD activity of 16 ACAD9 mutations identified in 24 patients were evaluated using a prokaryotic expression system. We showed that there was a significant inverse correlation between residual enzyme ACAD activity and phenotypic severity of ACAD9-deficient patients. These results provide evidence that in cells where it is strongly expressed, ACAD9 plays a physiological role in fatty acid oxidation, which contributes to the severity of the phenotype in ACAD9-deficient patients. Accordingly, treatment of ACAD9 patients should aim at counteracting both CI and fatty acid oxidation dysfunctions.

Figure 1.

Targeted gene disruption of ACAD9 in HEK293 cells selectively affects long-chain FAO and not medium-chain FAO. (A) Cell lysates from HEK293 cells (293), mock-treated HEK293 cells (Mock) and two ACAD9 −/− cell clones derived from TALEN-transfected cells (clones A and B both with ACAD9-null mutations) were subjected to SDS–PAGE and western blotting with the indicated antibodies. β-actin was used as control. (B) Etomoxir-sensitive palmitate oxidation rate (top) and palmitoyl-CoA dehydrogenase activity (bottom) were determined in HEK293, mock-treated (Mock-Tx) and the two ACAD9 −/− clones (clones A and B). All values are presented as average ± SD (n = 4 for each condition); *P < 0.05. (C) Octanoate oxidation rates were determined in HEK293 cells and TALEN-mediated ACAD9 −/− clone A. All values are presented as average ± SD (n = 4). N.S., not significant.

Figure 2.

Targeted gene disruption of ACAD9 in HEK293 cells affects formation of supercomplexes (A) as well as CI activity and ECSIT binding (B). (A) Mitochondria isolated from HEK293, mock-treated HEK293 and the two ACAD9 −/− clones A and B were permeabilized with digitonin and resolved by BNGE (Blue Native Gel Electrophoresis) followed by Coomassie Blue staining to visualize individual respiratory chain complex bands (CI to IV) and supercomplexes (SC). (B) Top: BNGE analysis of CI in-gel activity was performed in isolated mitochondria from HEK293, mock-treated HEK293 and the two ACAD9 −/− clones A and B. Bottom: SDS–PAGE immunodetection of ECSIT and β-actin in cell lysates from the same cell lines.

Figure 3.

Long-chain FAO and CI defects are restored in ACAD9-deficient cells by transfection of an ACAD9 expression vector. (A) Re-expression of ACAD9 in ACAD9 −/− HEK293 clone A cells transfected with empty pcDNA3.1 vector (Vect) or pcDNA-ACAD9 (ACAD9). The ACAD9 antigen was visualized with green fluorescently tagged antibodies, and mitochondrial cytochrome c oxidase (COX) was visualized with red fluorescently tagged antibodies. The merged image (arrow) shows co-localization of ACAD9 and COX in mitochondria as yellow. Scale bar, 10.75 μm. (B) Etomoxir-sensitive palmitate oxidation rates were determined in pcDNA3.1 (Vect) and pcDNA-ACAD9 (ACAD9) transfected clone A. Values are presented as average ± SD (n = 4); *P < 0.05. (C) Left panel: mitochondria isolated from TALEN-mediated ACAD9-deficient HEK293 clone A transfected with pcDNA3.1 vector (Vect) or pcDNA-ACAD9 (ACAD9) were permeabilized with digitonin and resolved by BNGE followed by Coomassie Blue staining to visualize individual respiratory chain complex bands (CI to IV) and supercomplexes (SC). Right: BNGE analysis of CI in-gel activity was performed in isolated mitochondria from pcDNA3.1-transfected (Vect) and pcDNA-ACAD9-transfected (ACAD9) clone A (top). Cell lysates from these two cell lines were subjected to SDS–PAGE and western blotting with the indicated antibodies. β-actin was used as control.

Figure 4.

Mouse double-knockout Acadl −/−, Acadvl −/− (2KO) neonatal fibroblasts exhibit residual long-chain FAO capacity. (A) Cell lysates from wild type (WT) and double-knockout mouse fibroblasts (2KO) were subjected to SDS–PAGE and western blotting with the indicated antibodies. β-actin was used as control. Etomoxir-sensitive palmitate oxidation rates (B) and palmitoyl-CoA dehydrogenase activity (C) were determined in WT and 2KO double-knockout mouse fibroblasts. Values are presented as average ± SD (n = 4); *P < 0.05.

Figure 5.

ACAD9 missense mutations affect FAO independent from its CI assembly function. (A) Prokaryotic mutagenesis and expression studies of 16 ACAD9 alleles containing missense variants. Each allele with the predicted amino acid substitution shown at the top of the figure was expressed in E. coli, and the cell-free extract was analyzed by SDS–PAGE followed by western blotting with anti-ACAD9 antibodies (middle). Palmitoyl-CoA dehydrogenase activity (ACAD9 activity) in cell-free extracts following prokaryotic expression is given on the bottom line. ND: non-detectable. ACAD9 activity of each mutant protein is expressed as % of the activity of wild-type (WT) ACAD9 expressed in E. coli. (B) SDS–PAGE analysis of recombinant wild type-His-tagged ACAD9 (WT) and two mutants ACAD9 proteins chosen as an example (E413K, unstable and R518H, stable) during limited trypsin digestion. The molar ratio of protein to trypsin was 1000:1. The reactions were initiated by adding trypsin into 1 mg/ml ACAD9 proteins in 25 mm sodium phosphate buffer, pH 7.6, containing 100 mm NaCl and 10% glycerol. Aliquots were taken at indicated times for SDS–PAGE analysis. While band a represents the entire mature ACAD9 peptide (65 kDa), the molecular weight of bands b and c correspond to the N-terminal domain (45 kDa) and C-terminal domain (17 kDa) of the mature ACAD9 protein, respectively. Bands b1 and b2 correspond to an additional cleavage site in the loop region connecting N-terminal and C-terminal domain. Upon addition of trypsin, both WT, E413K and R518H mutants are quickly digested into two pieces: the N-terminal domain (45 kDa) and the C-terminal domain (17 kDa). However, as the trypsin digestion continues, the N-terminal and C-terminal domains of E413K were further cleaved into smaller pieces (instability), whereas the two domains of WT and R518H are very stable. (C) Comparison between crude extracts (white) and purified protein data (gray) for each mutants. S: stability of the recombinant ACAD9 proteins from crude extracts evaluated on western blots (+++: very stable; −: very unstable). T: stability of the recombinant purified ACAD9 proteins after trypsin proteolysis (+++: very stable; −: very unstable). C: ACAD activity of crude extract. %: percentage of the activity of wild type recombinant ACAD9. P: specific activity of the recombinant purified ACAD9 expressed as μmole of C16-CoA oxidized/min/mg of ACAD9 protein in the presence of added FAD. NA: not available. ND: non-detectable. MW: molecular weight.

Figure 6.

Molecular modeling of the ACAD9 mutations. Two perpendicular views of ribbon representations of a human ACAD9 monomer model (11) showing the active site with bound FAD and the acyl moiety of myristoyl-CoA and the location of missense mutations found in ACAD9-deficient patients. The second monomer, hidden, lies between the C-terminus (blue) α-helix and the rest of the monomer (gray). The model was generated using the human VLCAD crystal structure coordinates (PDB: 3B96) (18). The peptide stretch 448–494 is not represented as its equivalent in the template molecule, the VLCAD crystal, is disordered.