Abstract

Oxidative phosphorylation and fatty acid oxidation are two major metabolic pathways in mitochondria. Acyl-CoA dehydrogenase 9 (ACAD9), an enzyme assumed to play a role in fatty acid oxidation, was recently identified as a factor involved in complex I biogenesis. Here we further investigated the role of ACAD9's enzymatic activity in fatty acid oxidation and complex I biogenesis. We provide evidence indicating that ACAD9 displays enzyme activity in vivo. Knockdown experiments in very-long-chain acyl-CoA dehydrogenase (VLCAD)-deficient fibroblasts revealed that ACAD9 is responsible for the production of C14:1-carnitine from oleate and C12-carnitine from palmitate. These results explain the origin of these obscure acylcarnitines that are used to diagnose VLCAD deficiency in humans. Knockdown of ACAD9 in control fibroblasts did not reveal changes in the acylcarnitine profiles upon fatty acid loading. Next, we investigated whether catalytic activity of ACAD9 was necessary for complex I biogenesis. Catalytically inactive ACAD9 gave partial-to-complete rescue of complex I biogenesis in ACAD9-deficient cells and was incorporated in high-molecular-weight assembly intermediates. Our results underscore the importance of the ACAD9 protein in complex I assembly and suggest that the enzymatic activity is a rudiment of the duplication event.

INTRODUCTION

Acyl-CoA dehydrogenase 9 (ACAD9), initially proposed as an enzyme of the mitochondrial fatty acid β-oxidation pathway (1), was recently shown to be crucial for oxidative phosphorylation complex I assembly (2). Indeed, mutations in ACAD9 result in complex I deficiency associated with symptoms very characteristic of oxidative phosphorylation disorders (2–7). Sequence, phylogenetic and protein modeling data, however, show that the protein is an acyl-CoA dehydrogenase. These enzymes catalyze the α-β-dehydrogenation of acyl-CoA esters and as such perform the first step of the fatty acid oxidation cycle as well as oxidation steps in amino acid metabolism. ACAD9 is most homologous to very-long-chain acyl-CoA dehydrogenase (VLCAD), suggesting it can catalyze the initial step in long-chain fatty acid oxidation (1,2,8). Moreover, seemingly conflicting with its role in oxidative phosphorylation, recombinant ACAD9 displays activity towards long-chain acyl-CoAs, very similar to VLCAD (9). The catalytic sites have been conserved throughout all the vertebrate organisms that express the protein. Both VLCAD and ACAD9 function as a dimer, which is a prerequisite to bind and convert long-chain fatty acids (9).

Previous work showed that ACAD9 is active in the same complex I assembly stages as NDUFAF1 and Ecsit (2). These proteins are peripherally bound to the inner mitochondrial membrane and physically associate with two rather late-stage assembly intermediates. More recently, TMEM126B has been identified as an interaction partner of these proteins and may provide the peripheral membrane anchoring, because it contains a transmembrane domain (10). The function of ACAD9 in complex I assembly is crucial as is illustrated by the identification of 16 patients with complex I deficiency owing to mutations in ACAD9 (2–7). Interestingly, some of these patients benefit from riboflavin administration (3,4). Riboflavin is a precursor of flavin adenine dinucleotide (FAD), the enzyme-bound prosthetic group of all acyl-CoA dehydrogenases. The role of the FAD cofactor in complex I assembly is unclear, but riboflavin administration can promote the stability of a specific mutant glutaryl-CoA dehydrogenase, another member of the acyl-CoA dehydrogenase family (11). In this paper, we set out to find whether ACAD9 indeed uses its catalytic core in vivo and explored whether this protein physically links the two major mitochondrial metabolic pathways: oxidative phosphorylation and fatty acid oxidation.

RESULTS

ACAD9 is responsible for the production of C14:1-carnitine in VLCAD-deficient patients

At the metabolite level, VLCAD deficiency is characterized by a pronounced accumulation of C14:1-carnitine. In fact, C14:1-carnitine is used in neonatal screening programs and regular diagnostics to identify VLCAD deficiency. Previous work has established that C14:1-carnitine is derived from oleic acid, suggesting that VLCAD patients still possess long-chain acyl-CoA dehydrogenase activity that is sufficient to catalyze at least two cycles of β-oxidation (12,13). We reasoned that this long-chain acyl-CoA dehydrogenase activity could be performed by ACAD9. To test this, we performed a shRNA knockdown of ACAD9 in two different VLCAD-deficient fibroblast lines using lentiviral transduction, which resulted in a pronounced decrease in ACAD9 protein levels and as a consequence a decrease in the complex I subunit NDUFB8 (Fig. 1A). First, we determined oleate oxidation rate. Knockdown of ACAD9 further decreased oleate oxidation in VLCAD-deficient cell lines, but increased oleate oxidation in control cell lines (Fig. 1B). To address this in more detail, we incubated the cell lines with oleate and palmitate and quantified the formation of specific acylcarnitines in the medium. Incubation of VLCAD-deficient fibroblasts with oleate led to the accumulation of C18:1- and C14:1-acylcarnitine (Fig. 1C), whereas upon incubation with palmitate, there was accumulation of C16- and C12-acylcarnitine (Fig. 1D). These results confirm that there is low residual capacity to perform the first two cycles of β-oxidation, despite the fact that both cell lines have a VLCAD null allele. ACAD9 knockdown decreased the accumulation of C14:1- and C12-carnitine and caused a parallel increase in the accumulation of C18:1- and C16-carnitine (Fig. 1C and D). ACAD9 knockdown in control fibroblasts did not change acylcarnitine formation. The same results were obtained when we used [U-13C]-oleate and [U-13C]-palmitate (Supplementary Material, Fig. S1). This result proves that in VLCAD-deficient cell lines, ACAD9 is responsible for the residual FAO capacity and the production of C14:1- and C12-carnitine, two very important diagnostic markers of VLCAD deficiency.

Figure 1.

ACAD9 displays acyl-CoA dehydrogenase activity in vivo. (A) SDS western blot immunodetection of ACAD9, VLCAD, subunits of oxidative phosphorylation complex I, II, III and V and loading control β-actin on two control (C1 and C2) and two VLCAD-deficient (VLCADD1 and VLCADD2) patient fibroblasts, after stable knockdown of scrambled RNA (shControl) and stable knockdown of ACAD9 (shACAD9). (B) Oleate oxidation assay in these fibroblasts. All values are presented as average ± SD, n = 4 for each condition. (C and D) Levels of indicated acylcarnitines (in pmol/μg protein/72 h) after incubation with (C) 100 μM oleate and (D) 100 μM palmitate. Values are presented as average + SD, n = 2 for each condition.

Figure 1.

ACAD9 displays acyl-CoA dehydrogenase activity in vivo. (A) SDS western blot immunodetection of ACAD9, VLCAD, subunits of oxidative phosphorylation complex I, II, III and V and loading control β-actin on two control (C1 and C2) and two VLCAD-deficient (VLCADD1 and VLCADD2) patient fibroblasts, after stable knockdown of scrambled RNA (shControl) and stable knockdown of ACAD9 (shACAD9). (B) Oleate oxidation assay in these fibroblasts. All values are presented as average ± SD, n = 4 for each condition. (C and D) Levels of indicated acylcarnitines (in pmol/μg protein/72 h) after incubation with (C) 100 μM oleate and (D) 100 μM palmitate. Values are presented as average + SD, n = 2 for each condition.

Complex I deficiency does not lead to the production of C14:1- and C12-carnitine

To confirm that the changes in the accumulation of the long-chain acylcarnitines in VLCAD- and ACAD9-deficient cells were not due to decreased complex I activity caused by ACAD9 depletion but owing to the loss of the catalytic activity of the enzyme, we inhibited complex I activity in control and VLCAD-deficient fibroblasts with rotenone and determined the consequences on the acylcarnitine profile upon palmitate loading (Fig. 2A and B). Although inhibition of complex I with rotenone led to a decrease in the production C12-acylcarnitine in VLCAD-deficient fibroblasts, it did not produce a parallel increase in the production of C16-carnitine as was observed after ACAD9 knockdown (Fig. 2A and B).

Figure 2.

Complex I deficiency does not lead to the production of C14:1- and C12-carnitine. (A and B) Accumulation of indicated acylcarnitines in medium of fibroblasts treated with 100 µM [U-13C]-palmitate for 3 days in the absence or presence of 0.1 µM rotenone. (C) SDS western blot immunodetection of VLCAD, ACAD9, complex I subunits, NDUFS3 and NDUFB8, subunits of complex II, III and V and loading control β-actin on control and VLCAD-deficient patient fibroblasts (VLCADD), after shRNA-mediated knockdown of ACAD9 (A#1; A#2), or NDUFS3 (N#1; N#2). (D and E) Levels of indicated acylcarnitines in medium of fibroblasts treated with 100 µM [U-13C]-palmitate for 3 days. Values are presented as average + SD, n = 2 for each condition.

Figure 2.

Complex I deficiency does not lead to the production of C14:1- and C12-carnitine. (A and B) Accumulation of indicated acylcarnitines in medium of fibroblasts treated with 100 µM [U-13C]-palmitate for 3 days in the absence or presence of 0.1 µM rotenone. (C) SDS western blot immunodetection of VLCAD, ACAD9, complex I subunits, NDUFS3 and NDUFB8, subunits of complex II, III and V and loading control β-actin on control and VLCAD-deficient patient fibroblasts (VLCADD), after shRNA-mediated knockdown of ACAD9 (A#1; A#2), or NDUFS3 (N#1; N#2). (D and E) Levels of indicated acylcarnitines in medium of fibroblasts treated with 100 µM [U-13C]-palmitate for 3 days. Values are presented as average + SD, n = 2 for each condition.

To further demonstrate the role of ACAD9 in the formation of C14:1- and C12-carnitine and to verify that the changes in acylcarnitine production were not caused by a decrease in total amount of fully assembled complex I, we performed knockdown of NDUFS3 as well as ACAD9 using two different shRNAs for both genes. All shRNAs gave efficient knockdown of the targeted protein (Fig. 2C). In addition, NDUFS3 knockdown decreased NDUFB8, showing that the complex I content was decreased. The different fibroblast cell lines were loaded with palmitate. Again ACAD9 knockdown very effectively decreased C12-carnitine in VLCAD-deficient cell lines, which again coincided with an increase in C16-carnitine (Fig. 2D and E). NDUFS3 knockdown did not have these pronounced effects on acylcarnitine production. This unambiguously demonstrates that ACAD9 has enzyme activity and is responsible for the production of C14:1- and C12-carnitine in VLCAD deficiency. It also shows that decreased complex I activity is not able to produce similar effects on acylcarnitine production.

Catalytically inactive ACAD9 rescues complex I activity in ACAD9 deficiency

To determine whether the catalytic activity of ACAD9 is required for its function in oxidative phosphorylation, we attempted to rescue the two previously described ACAD9 patient fibroblast cell lines (2) with a catalytically inactive ACAD9 mutant protein. For this, we generated an E426Q ACAD9 mutant enzyme. In VLCAD, this glutamate (E462) is the catalytic base and mutating it to a glutamine renders the enzyme inactive (14). A model of the ACAD9 protein showed that E426 is indeed located at the catalytic site of the protein (Fig. 3A). To prove that ACAD9 E426Q is catalytically inactive, we expressed wild-type (WT) or mutant ACAD9 in Escherichia coli and measured activity with palmitoyl-CoA. Purified ACAD9 catalyzed the formation of trans-2-palmitoyl-CoA, whereas no activity was observed for the purified E426Q mutant (Fig. 3B). Next, we complemented the ACAD9 patient cell lines by lentiviral transduction with the WT ACAD9 and E426Q ACAD9 mutant. As expected, WT ACAD9 rescued decreased Ecsit levels (Fig. 3C) and decreased complex I abundance (Fig. 3D), in gel activity (Fig. 3E) and in vitro activity (Fig. 3F). Remarkably, the mutant ACAD9 was also able to complement the ACAD9-deficient cell line (Fig. 3C–F). Complex I enzyme measurements revealed that WT and mutant ACAD9 were both able to rescue the complex I deficiency, although in patient 2, the mutant ACAD9 was slightly less effective (Fig. 3F). In control cells, mutant ACAD9 seemed to decrease complex I activity, which might suggest a dominant negative effect of the catalytically inactive mutant ACAD9 on complex I biogenesis (Fig. 3F). This decrease in activity was not reflected by parallel decreases in Ecsit levels (Fig. 3C), complex I abundance (Fig. 3D) and complex I in gel activity (Fig. 3E).

Figure 3.

Complementation of ACAD9-deficient fibroblasts with WT and catalytically inactive ACAD9 (E426Q). (A) Model of ACAD9 with the introduced E426Q mutation. The two ACAD9 monomers are shown in green and blue (surface view) by element; carbon is cyan, nitrogen is blue, oxygen is red and yellow dots indicate hydrogen bonds. The location of the mutation is indicated with an arrow. Glutamate 426 forms a hydrogen bond with arginine 304, the change to a glutamine does not affect its size or form however, the amino acid loses its negative charge and therefore the amino acid will prevent the formation of a charge transfer complex. (B) Enzyme activity of purified WT and mutant E426Q ACAD9. The chromatogram shows the substrate C16-CoA (palmitoyl-CoA, peak a) and the product of the reaction C16:1-CoA (palmitenoyl-CoA, peak b). The immunoblot of the eluates shows the presence of ACAD9; eluate 3 was used to measure the activity. (C–F) Complementation of ACAD9 patient fibroblasts. –, not infected; Cox8, infected with V5-tagged COX8 leader sequence; WT, infected with V5-tagged ACAD9 WT; mut, infected with V5-tagged ACAD9 E426Q. (C) SDS western blot immunodetection of V5, Ecsit, complex II subunit SDHA and loading control porin. (D) BN–PAGE immunodetection of complex I subunit NDUFA9, complex II-SDHA, complex III-core 2, complex IV-cox4 and complex V-ATPase-α. (E) BN–PAGE analysis of complex I in-gel activity (CI-IGA) and western blot immunodetection of loading control complex II-SDHA. (F) Measurement of complex I activity normalized on citrate synthase in mU CI activity/U CS. Values are presented as average + SD, n = 2 for each condition.

Figure 3.

Complementation of ACAD9-deficient fibroblasts with WT and catalytically inactive ACAD9 (E426Q). (A) Model of ACAD9 with the introduced E426Q mutation. The two ACAD9 monomers are shown in green and blue (surface view) by element; carbon is cyan, nitrogen is blue, oxygen is red and yellow dots indicate hydrogen bonds. The location of the mutation is indicated with an arrow. Glutamate 426 forms a hydrogen bond with arginine 304, the change to a glutamine does not affect its size or form however, the amino acid loses its negative charge and therefore the amino acid will prevent the formation of a charge transfer complex. (B) Enzyme activity of purified WT and mutant E426Q ACAD9. The chromatogram shows the substrate C16-CoA (palmitoyl-CoA, peak a) and the product of the reaction C16:1-CoA (palmitenoyl-CoA, peak b). The immunoblot of the eluates shows the presence of ACAD9; eluate 3 was used to measure the activity. (C–F) Complementation of ACAD9 patient fibroblasts. –, not infected; Cox8, infected with V5-tagged COX8 leader sequence; WT, infected with V5-tagged ACAD9 WT; mut, infected with V5-tagged ACAD9 E426Q. (C) SDS western blot immunodetection of V5, Ecsit, complex II subunit SDHA and loading control porin. (D) BN–PAGE immunodetection of complex I subunit NDUFA9, complex II-SDHA, complex III-core 2, complex IV-cox4 and complex V-ATPase-α. (E) BN–PAGE analysis of complex I in-gel activity (CI-IGA) and western blot immunodetection of loading control complex II-SDHA. (F) Measurement of complex I activity normalized on citrate synthase in mU CI activity/U CS. Values are presented as average + SD, n = 2 for each condition.

Catalytically mutant ACAD9 incorporates into known higher-molecular-weight complexes

To further investigate a potential dominant negative effect of catalytically inactive ACAD9 on complex I activity, we generated inducible HEK293 cell lines, which express WT or mutant ACAD9 with different tags. First, we confirmed that the tagged WT and mutant ACAD9 were expressed and properly built into higher-molecular-weight complexes. GFP-tagged WT, but also mutant ACAD9, was incorporated into higher-molecular-weight complexes (Fig. 4A). As it is difficult to distinguish subcomplexes on one-dimensional blue native gels and the GFP tag is rather long, we also analyzed MycHis- and TAP-tagged ACAD9. Two-dimensional PAGE analysis of the MycHis-tagged WT and mutant ACAD9 revealed that both proteins are assembled into the same subcomplexes (Fig. 4B). To further corroborate that mutant ACAD9 indeed does not affect complex I assembly nor function, we analyzed the expression level of ACAD9's interaction partner Ecsit and measured complex I activity. First, we checked the expression level of three different MycHis-tagged mutant ACAD9 clones compared with endogenous ACAD9. The expression level of the tagged ACAD9 was higher than endogenous ACAD9 (Fig. 4B and C, endogenous signal indicated with arrows). Despite the overexpression of mutant ACAD9 and its assembly into complex I subcomplexes, the expression of Ecsit was not affected (Fig. 4C). Blue native analysis of these clones showed that the levels of fully assembled complex I were also not affected (Fig. 4D). Moreover, complex I activity was not decreased after overexpression of mutant ACAD9 (Fig. 4E). The control HEK293 cells, tagged WT and mutant clones all showed a slight decrease in complex I activity probably caused by the addition of doxycycline, a known mitochondrial toxin, to the culture medium. Thus, despite high levels of overexpression, we were unable to find dominant negative effects of catalytically inactive ACAD9 on complex I biogenesis and activity.

Figure 4.

Catalytically inactive ACAD9 is incorporated in higher-molecular-weight complexes and does not disturb complex I assembly and function. (A) Fluorescence scanning of blue native gels of mitochondrial fractions of HEK293 lysates expressing GFP-tagged mutant ACAD9 (mut-GFP+dox) or GFP-tagged WT ACAD9 (WT-GFP+dox). (B) Two-dimensional analysis of HEK293 mitochondrial lysate of MycHis-tagged WT (WT-MycHis) and mutant (mut-MycHis) ACAD9 with ACAD9 antibody; the arrows indicate endogenous ACAD9. Complex I (CI) and subcomplex 3 (Sub 3) are indicated with NDUFS3 antibody. (C) SDS western blot analysis of three mutant MycHis clones (1, 2, 3) with ACAD9, Ecsit and loading control porin. The arrows indicate endogenous ACAD9. (D) BN–PAGE western blot analysis of three mutant MycHis clones (1, 2, 3) with CI-NDUFA9, CV-ATPase-α and loading control CII-SDHA. (E) Measurement of complex I activity normalized on citrate synthase in mU CI activity/U CS in control HEK293 cells, and HEK293 T-rex cells inducible for MycHis- and TAP-tagged WT and mutant ACAD9 without (−) and after incubation with doxycycline (+dox). Values are presented as average + D, n = 2 for each condition.

Figure 4.

Catalytically inactive ACAD9 is incorporated in higher-molecular-weight complexes and does not disturb complex I assembly and function. (A) Fluorescence scanning of blue native gels of mitochondrial fractions of HEK293 lysates expressing GFP-tagged mutant ACAD9 (mut-GFP+dox) or GFP-tagged WT ACAD9 (WT-GFP+dox). (B) Two-dimensional analysis of HEK293 mitochondrial lysate of MycHis-tagged WT (WT-MycHis) and mutant (mut-MycHis) ACAD9 with ACAD9 antibody; the arrows indicate endogenous ACAD9. Complex I (CI) and subcomplex 3 (Sub 3) are indicated with NDUFS3 antibody. (C) SDS western blot analysis of three mutant MycHis clones (1, 2, 3) with ACAD9, Ecsit and loading control porin. The arrows indicate endogenous ACAD9. (D) BN–PAGE western blot analysis of three mutant MycHis clones (1, 2, 3) with CI-NDUFA9, CV-ATPase-α and loading control CII-SDHA. (E) Measurement of complex I activity normalized on citrate synthase in mU CI activity/U CS in control HEK293 cells, and HEK293 T-rex cells inducible for MycHis- and TAP-tagged WT and mutant ACAD9 without (−) and after incubation with doxycycline (+dox). Values are presented as average + D, n = 2 for each condition.

DISCUSSION

Here, we show that ACAD9's catalytic core has conserved its ability to convert acyl-CoA into trans-2-enoyl-CoAs in vivo. In VLCAD-deficient fibroblasts, ACAD9 is responsible for the production of C14:1-carnitine from oleate, which is the marker used to diagnose VLCAD deficiency in metabolite diagnostics and neonatal screening programs. We show that ACAD9 is able to initiate the first and second cycle in the fatty acid oxidation of palmitate and oleate, the two main long-chain fatty acids used for the mitochondrial fatty acid oxidation. We were only able to demonstrate this activity in VLCAD-deficient fibroblasts and not in control cells, indicating that ACAD9 activity levels are low in healthy cells. This is corroborated by the fact that upon overexpression in E. coli, the Vmax of ACAD9 is relatively low compared with VLCAD (9,15). In addition, a high level of overexpression of ACAD9 in HEK293 cells did not produce a detectable increase in palmitoyl-CoA dehydrogenase activity (data not shown). Moreover, ACAD9-deficient patients do not display any fatty acid oxidation disturbances.

These observations argue against a primary role for ACAD9 in β-oxidation of fatty acids. Therefore, we hypothesized that ACAD9 activity could be required for its function in complex I biogenesis. As no patients have been identified with a mutation in one of the amino acids required for catalytic activity, we created a catalytic mutant (E426Q) and expressed this mutant in ACAD9-deficient fibroblasts. This mutant ACAD9 protein rescued complex I deficiency in cells of two different patients. In one patient, catalytically inactive ACAD9 seemed less effective than the WT protein. Also, expression of catalytically inactive ACAD9 in WT cells suggested a dominant negative effect on complex I biogenesis. We suggest that these inconsistencies are explained by the inherent difficulties of complementation in primary fibroblasts. Indeed, follow-up experiments in HEK293 cells failed to demonstrate a dominant negative effect of the catalytically inactive mutant ACAD9 on complex I activity and confirmed that this mutant protein was assembled into higher-molecular-weight complexes illustrating its functionality. From these observations, we conclude that ACAD9's enzymatic activity is dispensable for its function in complex I assembly.

Riboflavin administration can alleviate symptoms in ACAD9-deficient patients and increases complex I activity in fibroblasts from these patients (4,5). This observation, however, does not prove that ACAD9 enzyme activity is necessary for complex I assembly. Riboflavin administration does not only promote FAD binding, it may also function as a chemical chaperone (16) and as such improve folding of specific ACAD9 mutant proteins. Indeed, riboflavin administration can promote the stability of a specific mutant glutaryl-CoA dehydrogenase (11). Similarly, riboflavin may also result in higher protein levels of mutant ACAD9 and as such rescue complex I assembly. The effects of riboflavin on ACAD9 protein levels should be investigated in these specific cell lines.

A crucial role for ACAD9 in complex I biogenesis has been unambiguously demonstrated. A role in fatty acid oxidation remains at best questionable. We now show for the first time ACAD9's enzymatic activity in a cell based assay. This activity could, however, only be unambiguously established in VLCAD-deficient cells. A remaining question is whether the dehydrogenase activity of ACAD9 in VLCAD null mutants really demonstrates a function for ACAD9 in fatty acid oxidation. Indeed, the low ACAD9 activity could be a mere rudiment of the duplication event. The fact that activity is preserved despite a long evolution time argues in favor of a role in fatty acid oxidation. However, there are only few mutations possible that will completely inactivate the protein, while leaving the protein structure intact. These specific mutations could be considered rare events in evolution. Mutations that decrease activity, but leave the protein structure intact, are probably more frequent. Of course, one alternative hypothesis might be that ACAD9 functions in mitochondrial fatty acid oxidation in very specific tissues or has some very specific acyl-CoA substrates. ACAD9, however, is ubiquitously expressed, arguing against a tissue-specific role. The reported substrate specificity is rather broad (9) arguing against a specific substrate.

To date, 16 patients with 12 different mutations in the coding region of ACAD9 gene have been identified, which mainly comprise missense mutations (3,7). Interestingly, so far no patients with two ACAD9 null mutations have been described, suggesting that the complete absence of the protein is lethal. The clinical symptoms range in severity and in organs affected (3,7). Remarkably, ACAD9 mutations result frequently, but not exclusively, in hypertrophic cardiomyopathy. However, the number of ACAD9-deficient patients described so far is still too low to draw conclusions on clinical features or to make reliable genotype–phenotype correlations.

In summary, we have shown that ACAD9 can display acyl-CoA dehydrogenase activity in vivo and is responsible for the formation of the C14:1- and C12-carnitine in VLCAD deficiency. There is currently no other evidence to support that ACAD9 is necessary for fatty acid oxidation. In contrast, the function of ACAD9 in complex I assembly has been firmly established by now. We now show that ACAD9 activity is dispensable for complex I assembly.

MATERIALS AND METHODS

Cell culture

The cells used for the acylcarnitine profiling were cultured in Dulbecco's modified Eagle's medium (DMEM, Lonza) containing 10% fetal bovine serum (FBS, PAA), 1% penicillin, streptomycin, fungizone (Life Technologies) and 22.5 mM Hepes. Human skin fibroblasts in the complementation experiment were cultured in medium 199 (Life Technologies) supplemented with 10% FBS and 1% penicillin/streptomycin. HEK293FT cells for virus production were cultured in DMEM supplemented with 10% FBS, 1% penicillin/streptomycin, MEM non-essential amino acids (Life Technologies), 1 mM sodium pyruvate (Life Technologies) and l-glutamine (Life Technologies). Flp-In T-REx HEK293 cells (Life Technologies) were cultured for selection in the presence of 200 μg/ml hygromycin (Millipore) in DMEM supplemented with 10% FBS, 1% penicillin/streptomycin and 50 μg/ml blasticin (Life Technologies). To induce expression of the fusion proteins, 1 μg/ml of doxycycline (Sigma) was added to the medium, followed by incubation for 72 h.

Lentiviral knockdown on fibroblasts

The Mission lentiviral shRNA system from Sigma was used to obtain ACAD9 and NDUFS3 knockdown. The virus was produced after cotransfection of pMD2G, pMDL/RRE, pRSV/REV and the pLKO.1 vector (shRNA against ACAD9, NDUFS3 or control) in HEK293 cells. For NDUFS3: TRCN0000036617, TRCN0000036615. For ACAD9: TRCN0000026490, TRCN0000026450.

Virus supernatant from HEK293 cells were used to infect control and VLCAD-deficient fibroblasts [VLCADD1; c.AGTTdel798-801 VLCADD2; c.GTTAdel799-802 (12,17)]. One day after transduction, medium was replaced by culture medium containing 0.5 or 5 μg/ml puromycin to select for virus containing cells.

Lentiviral complementation of patient fibroblasts

ACAD9 WT and the Cox8 leader sequence were cloned into pDONR201 as described before (2,18). The QuikChange®Site-Directed Mutagenesis kit (Stratagene) was used to generate the E426Q mutant (c.1276G>C). To obtain the lentiviral compatible vector containing C-terminally V5-tagged ACAD9 WT or ACAD9 E426Q mutant or Cox8 leader sequence, the pDONR201 vector with either of these sequences was recombined with pLenti6.2/V5-DEST™ Gateway® Vector using the Gateway LR Clonase II enzyme mix (Life Technologies).

HEK293FT cells were transfected to produce lentivirus with pLP1, pLP2, pVSV/G and one of the three expression vectors according to manufacturer's protocol (Life Technologies). The virus was harvested after 72 h and added to the human fibroblasts overnight [patient 1: homozygous for R518H; patient 2: compound heterozygous for E63X, E413 K (2)]. The next day the virus was removed and the medium replaced, and the subsequent day 2.5 μg/ml blasticidin was added to the medium to select the cells. The cells were analyzed several passages after virus infection.

Generation of inducible cell lines

To obtain the inducible C-terminally tagged cell lines, pDONR201-ACAD9 and pDONR201-ACAD9 E426Q were recombined with the GFP and MycHis-Destination vector by use of the LR Clonase II Enzyme Mix. The obtained constructs were transfected into Flp-In T-REx HEK293 cells by using SuperFect Transfection Reagent (QIAGEN) according to the manufacturer's manual.

Blue native, SDS–PAGE and in gel activity assays

Equal amounts of total fibroblast cell protein was separated on Nu-PAGE (4–12%) Bis–Tris gels and transferred onto nitrocellulose by semi-dry blotting (Figures 1 and 2). Forty or eighty micrograms of solubilized mitochondrial proteins was loaded on the lanes of 5–15% blue native gradient gels or one-dimensional 10% sodium dodecyl sulfate-polyacrylamide gel electrophoreses (SDS–PAGE) as described previously (Figures 3 and 4) (19,20). After electrophoresis, in-gel fluorescence was determined using an imaging analyzer FLA-5000 (Fujifilm) (473-nm excitation laser, FITC filter >510 nm). After electrophoresis or in gel fluorescence scanning, gels were processed for either complex I in gel activity (CI-IGA) analysis or immunoblotting. For blotting, proteins were transferred to a PROTRAN nitrocellulose membrane (Schleicher & Schuell).

Antibodies and ECL detection

Immunodetection was performed by the use of the following primary antibodies: CI-NDUFS3, CI-NDUFA9 (Life Technologies), CII-SDHA, CIII-core 2, CIV-COX4, CV-ATPase-α, total OXPHOS Rodent WB antibody cocktail (MitoSciences), ACAD9 (a gift from G. Vockley, University of Pittsburgh School of Medicine, PA), Ecsit (a gift from M. Ryan, La Trobe University, Melbourne, Australia), VLCAD (M01, clone 5D3, Abnova), β-actin (Sigma), Porin (VWR international) and V5 (Invitrogen). Goat-anti-rabbit and goat-anti-mouse IRDye CW 680 or IRDye CW 800 were used as secondary antibodies, to detect the proteins using the Odyssey system from LI-COR Biosciences (Figures 1 and 2). Secondary detection was performed using peroxidase-conjugated anti-mouse or anti-rabbit IgGs (Life Technologies) (Figures 3 and 4). Immunoreactive bands were visualized using the enhanced chemiluminescence kit (Thermo Scientific) and detected using the Chemidoc XRS+ system (Biorad).

Analysis of acylcarnitines

Fibroblasts cultured in 12-well plates were incubated in MEM medium (Gibco) with 1% penicillin, streptomycin, fungizone, 0.4 mM l-carnitine, 0.1% (w/v) BSA and 100 µM of the indicated fatty acids at 37°C, 5% CO2. After 72 h, the incubation was stopped by removing the medium from the cells. The medium was processed as described (12) using internal standards (50 pmol d3-C3-, 20 pmol d3-C6-, 20 pmol d3-C8-, 20 pmol d3-C10- and 20 pmol d3-C16-acylcarnitine). Semi-quantitative determination of the formed acylcarnitines in the medium was performed using tandem mass spectrometry.

Enzyme measurements

The activities of the mitochondrial respiratory chain complexes were determined in cultured fibroblasts according to established procedures (21,22). Values are expressed relative to the mitochondrial reference enzyme citrate synthase and complex IV (22).

Cloning and expression of ACAD9 in E. coli

The sequence of mature human ACAD9 was inserted into the prokaryotic expression vector pET-22b(+) (Novagen, USA). The 5′ primer consisted of 94 nucleotides and included an NdeI restriction site (shown in boldface), followed by the first 81 nucleotides of the mature ACAD9-coding sequence; the first 60 nucleotides of the coding sequence were altered to reflect E. coli codon usage bias: 5′ A GTA CGT CAT ATG GCT TTC GCT AAA GAA CTG TTC CTG GGC AAA ATC AAA AAA AAA GAA GTT TTC CCG TTC CCG GAA GTT AGC CAA GAT GAA CTT 3′. The 3′-primer consisted of the last 23 nucleotides of the coding region, followed by a SacI restriction site: 5′ ATT CGA GAG CTC GCA GGT GCG GTC CAG AGG GTG GG 3′ (restriction site is shown in bold with nucleotides altered to match to E. coli codon usage bias underlined). The amplified fragment was subcloned into the NdeI and SacI sites of the pET-22b(+) vector. The QuikChange®Site-Directed Mutagenesis kit (Stratagene) was used to generate the E426Q mutant (c.1276G>C). WT and mutant E426Q ACAD9 were expressed as His-tagged fusion proteins in E. coli (BL21 AI pGro7, Life Technologies) in Terrific Broth medium with 8 g/l glycerol at 30°C for 3 h. The proteins were purified on HisLink Protein Purification Resin (Promega) according to the manufacturer's protocol. The protein was eluted in a buffer containing 25 mM Tris–HCl, 500 mM NaCl and 500 mM Immidazole.

Determination of ACAD9 activity

For the determination of ACAD9 activity, eluates from WT and mutant E426Q ACAD9 were incubated with palmitoyl-CoA. The standard mix contained 125 mM Tris–HCL (pH 8.0) and 0.4 mM ferrocenium hexafluorophosphate (23) plus 0.25 mM palmitoyl-CoA, in a final volume of 100 µl. The reaction was initiated by the addition of 80 µl reaction mix to 20 µl of five-times-diluted eluate. The reaction was performed at 37°C. After a 10-min incubation, the reaction was stopped using 10 µl of 2 M HCl, and samples were neutralized with a solution containing 10 µl 2 M KOH and 1 M MES (pH 6.0). After neutralization, 10 µl of 10 mM l-cysteine and 30 µl acetonitrile were added to the mixture. The samples were centrifuged at 20,000 g for 5 min, and the metabolites in the supernatant were analyzed on a C18 reversed-phase column by U(H)PLC, using an elution system of acetonitrile and 16.9 mM sodium phosphate buffer of pH 6.9 (24).

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported by the Netherlands Organization for Scientific Research (VIDI-grant No. 016.086.336 to S.M.H.).

ACKNOWLEDGEMENTS

The authors thank Hanka Venselaar (Centre for Molecular and Biomolecular Informatics, Nijmegen, the Netherlands) for modeling of the E426Q mutation in ACAD9, Lodewijk IJlst for expert advice on heterologous expression of ACAD9 in E. coli and Ronald Wanders for continuous support.

Conflict of Interest statement. None declared.

REFERENCES

1
Zhang
J.
Zhang
W.
Zou
D.
Chen
G.
Wan
T.
Zhang
M.
Cao
X.
Cloning and functional characterization of ACAD-9, a novel member of human acyl-CoA dehydrogenase family
Biochem. Biophys. Res. Commun.
 , 
2002
, vol. 
297
 (pg. 
1033
-
1042
)
2
Nouws
J.
Nijtmans
L.
Houten
S.M.
van den Brand
M.
Huynen
M.
Venselaar
H.
Hoefs
S.
Gloerich
J.
Kronick
J.
Hutchin
T.
, et al.  . 
Acyl-CoA dehydrogenase 9 is required for the biogenesis of oxidative phosphorylation complex I
Cell Metab.
 , 
2010
, vol. 
12
 (pg. 
283
-
294
)
3
Garone
C.
Donati
M.A.
Sacchini
M.
Garcia-Diaz
B.
Bruno
C.
Calvo
S.
Mootha
V.K.
Dimauro
S.
Mitochondrial encephalomyopathy due to a novel mutation in ACAD9
JAMA Neurol.
 , 
2013
, vol. 
70
 (pg. 
1177
-
1179
)
4
Gerards
M.
van den Bosch
B.J.
Danhauser
K.
Serre
V.
van Weeghel
M.
Wanders
R.J.
Nicolaes
G.A.
Sluiter
W.
Schoonderwoerd
K.
Scholte
H.R.
, et al.  . 
Riboflavin-responsive oxidative phosphorylation complex I deficiency caused by defective ACAD9: new function for an old gene
Brain
 , 
2011
, vol. 
134
 (pg. 
210
-
219
)
5
Haack
T.B.
Danhauser
K.
Haberberger
B.
Hoser
J.
Strecker
V.
Boehm
D.
Uziel
G.
Lamantea
E.
Invernizzi
F.
Poulton
J.
, et al.  . 
Exome sequencing identifies ACAD9 mutations as a cause of complex I deficiency
Nat. Genet.
 , 
2010
, vol. 
42
 (pg. 
1131
-
1134
)
6
Haack
T.B.
Haberberger
B.
Frisch
E.M.
Wieland
T.
Iuso
A.
Gorza
M.
Strecker
V.
Graf
E.
Mayr
J.A.
Herberg
U.
, et al.  . 
Molecular diagnosis in mitochondrial complex I deficiency using exome sequencing
J. Med. Genet.
 , 
2012
, vol. 
49
 (pg. 
277
-
283
)
7
Nouws
J.
Wibrand
F.
van den Brand
M.
Venselaar
H.
Duno
M.
Lund
A.M.
Trautner
S.
Nijtmans
L.
Ostergard
E.
A patient with complex I deficiency caused by a novel ACAD9 mutation not responding to riboflavin treatment
JIMD Rep.
 , 
2013
8
Swigonova
Z.
Mohsen
A.W.
Vockley
J.
Acyl-CoA dehydrogenases: dynamic history of protein family evolution
J. Mol. Evol.
 , 
2009
, vol. 
69
 (pg. 
176
-
193
)
9
Ensenauer
R.
He
M.
Willard
J.M.
Goetzman
E.S.
Corydon
T.J.
Vandahl
B.B.
Mohsen
A.W.
Isaya
G.
Vockley
J.
Human acyl-CoA dehydrogenase-9 plays a novel role in the mitochondrial beta-oxidation of unsaturated fatty acids
J. Biol. Chem.
 , 
2005
, vol. 
280
 (pg. 
32309
-
32316
)
10
Heide
H.
Bleier
L.
Steger
M.
Ackermann
J.
Drose
S.
Schwamb
B.
Zornig
M.
Reichert
A.S.
Koch
I.
Wittig
I.
, et al.  . 
Complexome profiling identifies TMEM126B as a component of the mitochondrial complex I assembly complex
Cell Metab.
 , 
2012
, vol. 
16
 (pg. 
538
-
549
)
11
Chalmers
R.A.
Bain
M.D.
Zschocke
J.
Riboflavin-responsive glutaryl CoA dehydrogenase deficiency
Mol. Genet. Metab.
 , 
2006
, vol. 
88
 (pg. 
29
-
37
)
12
Chegary
M.
te Brinke
H.
Ruiter
J.P.
Wijburg
F.A.
Stoll
M.S.
Minkler
P.E.
van Weeghel
M.
Schulz
H.
Hoppel
C.L.
Wanders
R.J.
, et al.  . 
Mitochondrial long chain fatty acid beta-oxidation in man and mouse
Biochim. Biophys. Acta
 , 
2009
, vol. 
1791
 (pg. 
806
-
815
)
13
Roe
D.S.
Vianey-Saban
C.
Sharma
S.
Zabot
M.T.
Roe
C.R.
Oxidation of unsaturated fatty acids by human fibroblasts with very-long-chain acyl-CoA dehydrogenase deficiency: aspects of substrate specificity and correlation with clinical phenotype
Clin. Chim. Acta
 , 
2001
, vol. 
312
 (pg. 
55
-
67
)
14
Souri
M.
Aoyama
T.
Cox
G.F.
Hashimoto
T.
Catalytic and FAD-binding residues of mitochondrial very long chain acyl-coenzyme A dehydrogenase
J. Biol. Chem.
 , 
1998
, vol. 
273
 (pg. 
4227
-
4231
)
15
He
M.
Rutledge
S.L.
Kelly
D.R.
Palmer
C.A.
Murdoch
G.
Majumder
N.
Nicholls
R.D.
Pei
Z.
Watkins
P.A.
Vockley
J.
A new genetic disorder in mitochondrial fatty acid beta-oxidation: ACAD9 deficiency
Am. J. Hum. Genet.
 , 
2007
, vol. 
81
 (pg. 
87
-
103
)
16
Lucas
T.G.
Henriques
B.J.
Rodrigues
J.V.
Bross
P.
Gregersen
N.
Gomes
C.M.
Cofactors and metabolites as potential stabilizers of mitochondrial acyl-CoA dehydrogenases
Biochim. Biophys. Acta
 , 
2011
, vol. 
1812
 (pg. 
1658
-
1663
)
17
Andresen
B.S.
Olpin
S.
Poorthuis
B.J.
Scholte
H.R.
Vianey-Saban
C.
Wanders
R.
IJlst
L.
Morris
A.
Pourfarzam
M.
Bartlett
K.
, et al.  . 
Clear correlation of genotype with disease phenotype in very-long-chain acyl-CoA dehydrogenase deficiency
Am. J. Hum. Genet.
 , 
1999
, vol. 
64
 (pg. 
479
-
494
)
18
Hoefs
S.J.
Dieteren
C.E.
Distelmaier
F.
Janssen
R.J.
Epplen
A.
Swarts
H.G.
Forkink
M.
Rodenburg
R.J.
Nijtmans
L.G.
Willems
P.H.
, et al.  . 
NDUFA2 complex I mutation leads to Leigh disease
Am. J. Hum. Genet.
 , 
2008
, vol. 
82
 (pg. 
1306
-
1315
)
19
Calvaruso
M.A.
Smeitink
J.
Nijtmans
L.
Electrophoresis techniques to investigate defects in oxidative phosphorylation
Methods
 , 
2008
, vol. 
46
 (pg. 
281
-
287
)
20
Ugalde
C.
Janssen
R.J.
van den Heuvel
L.P.
Smeitink
J.A.
Nijtmans
L.G.
Differences in assembly or stability of complex I and other mitochondrial OXPHOS complexes in inherited complex I deficiency
Hum. Mol. Genet.
 , 
2004
, vol. 
13
 (pg. 
659
-
667
)
21
Janssen
A.J.
Trijbels
F.J.
Sengers
R.C.
Smeitink
J.A.
van den Heuvel
L.P.
Wintjes
L.T.
Stoltenborg-Hogenkamp
B.J.
Rodenburg
R.J.
Spectrophotometric assay for complex I of the respiratory chain in tissue samples and cultured fibroblasts
Clin. Chem.
 , 
2007
, vol. 
53
 (pg. 
729
-
734
)
22
Srere
P.A.
John
M.L.
Citrate synthase. In methods in enzymology citric acid cycle
1969
Academic press
(pg. 
3
-
11
)
23
Lehman
T.C.
Hale
D.E.
Bhala
A.
Thorpe
C.
An acyl-coenzyme A dehydrogenase assay utilizing the ferricenium ion
Anal. Biochem.
 , 
1990
, vol. 
186
 (pg. 
280
-
284
)
24
Wanders
R.J.
Ruiter
J.P.
IJlst
L.
Waterham
H.R.
Houten
S.M.
The enzymology of mitochondrial fatty acid beta-oxidation and its application to follow-up analysis of positive neonatal screening results
J. Inherit. Metab. Dis.
 , 
2010
, vol. 
33
 (pg. 
479
-
494
)

Author notes

Present address: Department of Pathology, Yale University School of Medicine, New Haven, CT 06520, USA.

Supplementary data