Expression and characterization of mutations in human very long-chain acyl-CoA dehydrogenase using a prokaryotic system

Abstract

Very long-chain acyl-CoA dehydrogenase (VLCAD) catalyzes the first enzymatic step in the mitochondrial beta-oxidation of fatty acids 14-20 carbons in length. More than 100 cases of VLCAD deficiency have been reported with the disease varying from a severe, often fatal neonatal form to a mild adult-onset form. VLCAD is distinguished from matrix-soluble acyl-CoA dehydrogenases by its unique C-terminal domain, homodimeric structure, and localization to the inner mitochondrial membrane. We have for the first time expressed and purified VLCAD using a bacterial system. Recombinant VLCAD had similar biochemical properties to those reported for native VLCAD and the bacterial system was used to study six previously described disease-causing missense mutations including the two most common mild mutations (T220M, V243A), a mutation leading to the severe disease phenotype (R429W), and three mutations in the C-terminal domain (A450P, L462P, and R573W). Of particular interest was the finding that the A450P and L462P bacterial extracts had normal or increased amounts of VLCAD antigen and activity. In the pure form L462P had roughly 30% of wild-type activity while A450P was normal. Using computer modeling both mutations were mapped to a predicted charged surface of VLCAD that we postulate interacts with the mitochondrial membrane. In a membrane pull down assay both mutants showed greatly reduced mitochondrial membrane association, suggesting a mechanism for the disease in these patients. In summary, the bacterial expression system developed here will significantly advance our understanding of both the clinical aspects of VLCAD deficiency and the basic biochemistry of the enzyme.

Figure 1

The ΔEx3 VLCAD variant imports correctly into mitochondria. (A) Alignment of the full-length VLCAD, ΔEx3 VLCAD, and ACAD9 mature amino terminal sequences. (B) In vitro transcribed/translated precursor forms of ACAD9, ΔEx3 VLCAD, and VLCAD (loaded as controls into lanes 1, 4, and 7, respectively) were incubated with intact rat liver mitochondria for 45 min and then digested with proteinase K for 30 min to remove any remaining unimported precursor. Any imported protein will be protected from the protease and will be smaller in size due to processing of the mitochondrial leader sequence. Lanes 2, 5, and 8 show a mixture of remnant precursor (p) and protected mature (m) proteins following partial protease digestion (50 μg/ml proteinase K for 30 minutes) while in lanes 3, 6, and 9 only the protected, imported mature proteins remain following a full protease digestion (300 μg/ml proteinase K for 30 minutes).

Figure 2

Evaluation of enzyme activity and antigen in E coli crude lysates containing the recombinant VLCAD and various mutant VLCAD enzymes. (A) Activity was measured using C16:0-CoA as substrate. Bars represent means and standard deviations of triplicate assays. Note that mutant R429W had minimal detectable activity (0.4 ± 0.01 mU/mg) while R573W had no detectable activity. (B) VLCAD antigen was detected using a polyclonal antibody raised against the purified recombinant ΔEx3 VLCAD shown in lane 1. The top band represents VLCAD while the bottom band is due to non-specific immunoreactivity with an E coli protein. Lanes 2 through 8 contain 25 μg of E coli crude lysate from induced overnight C43 (DE3) cultures expressing the indicated VLCAD proteins in pET-21a.

Figure 3

Molecular sizing of wild type and mutant VLCADs. 50 μg of each purified VLCAD protein (as shown on a Coomassie-stained 10% SDS-PAGE gel in the lefthand inset: Lane 1, molecular weight markers; Lane 2, 5 μg purified ΔEx3 VLCAD; Lane 3, 5 μg purified A450P VLCAD; and Lane 4, 5 μg purified L462P VLCAD) was subjected to gel filtration chromatography and molecular mass was determined using the calibration curve shown in the righthand inset (standards used were: ribonuclease A, 13.7 kDa; carbonic anhydrase, 29 kDa; ovalbumin, 43 kDa; albumin, 67 kDa; aldolase, 158 kDa; and catalase, 232 kDa). The calculated molecular masses were 118 kDa, 119 kDa, and 122 kDa for L462P, A450P, and wild-type ΔEx3 VLCAD, respectively.

Figure 4

Comparison of the enzyme activity of purified wild-type VLCAD versus mutants A450P and L462P with various acyl-CoA substrates. Panel (A) shows specific activity assayed with myristoyl-CoA (C14:0), palmitoyl-CoA (C16:0), palmitoleoyl-CoA (C16:1), stearoyl-CoA (C18:0), oleoyl-CoA (C18:1), linoleoyl-CoA (C18:2), and arachidonyl-CoA (C20:0). Panel (B) depicts these same data but the specific activity with C16:0-CoA has been set to 100% and the relative activities with other substrates are shown as a percentage of C16:0 activity. Bars represent means and standard deviations of triplicate assays. Black bars show wild type activity, light gray bars show A450P activity, and dark gray bars show L462P activity.

Figure 5

Ribbon representation of a VLCAD homodimer model. The computer generated model was based on the known structures of several ACDs and straight chain acyl-CoA oxidase C-terminus (AOX). The six amino acids studied in mutagenesis experiments are shown in purple on either monomer A or B. Basic residues of the C-terminus are rendered in blue and acidic ones in red. The positively charged area at the top of the model is hypothesized to interact with the inner mitochondrial membrane.

Figure 6

Mitochondrial membrane binding is impaired in VLCAD mutants A450P and L462P. (A) Wild-type ΔEx3 VLCAD shows dose-dependent binding to mitochondrial membranes in vitro. A constant amount of pure VLCAD (3 μg) was incubated with increasing amounts of mitochondrial membranes and the percentage of enzyme activity appearing in the pellet fraction following centrifugation was determined. (B) Binding of mutant enzymes A450P and L462P to mitochondrial membranes as compared to wild-type VLCAD. 3 μg of purified wild type or mutant enzyme was incubated with 400 μg of mitochondrial membranes and the percentage of enzyme activity appearing in the pellet fraction following centrifugation was determined. SCAD, a mitochondrial matrix ACD enzyme, was used as a control and showed no membrane binding.