Structural basis for defective membrane targeting of mutant enzyme in human VLCAD deficiency

Abstract

Very long-chain acyl-CoA dehydrogenase (VLCAD) is an inner mitochondrial membrane enzyme that catalyzes the first and rate-limiting step of long-chain fatty acid oxidation. Point mutations in human VLCAD can produce an inborn error of metabolism called VLCAD deficiency that can lead to severe pathophysiologic consequences, including cardiomyopathy, hypoglycemia, and rhabdomyolysis. Discrete mutations in a structurally-uncharacterized C-terminal domain region of VLCAD cause enzymatic deficiency by an incompletely defined mechanism. Here, we conducted a structure-function study, incorporating X-ray crystallography, hydrogen-deuterium exchange mass spectrometry, computational modeling, and biochemical analyses, to characterize a specific membrane interaction defect of full-length, human VLCAD bearing the clinically-observed mutations, A450P or L462P. By disrupting a predicted α-helical hairpin, these mutations either partially or completely impair direct interaction with the membrane itself. Thus, our data support a structural basis for VLCAD deficiency in patients with discrete mutations in an α-helical membrane-binding motif, resulting in pathologic enzyme mislocalization.

Fig. 1. Structural and functional characterization of human VLCAD by X-ray crystallography, HDX MS analysis, and liposomal translocation assay.

a Dimeric structure of human VLCAD ΔEx3 (PDB ID 7S7G) demonstrating its subcomponents and expanded MBR residues (purple) that extend outward from the protein core. See Supplementary Table 1 for data collection and results of refinement. b Crystal contacts between neighboring VLCAD dimers involve reciprocal interactions between the MBR of one dimer and the surface groove of another. c Relative deuterium uptake (%) of full-length VLCAD in solution after 10 s, 1 m, 10 m, and 1 h of deuteration. Residues are numbered using the mature (cleaved leader sequence) form of human VLCAD. HDX MS experiments were performed twice using independent preparations of VLCAD protein. See Supplementary Data 1 for the HDX MS data used to create this figure. d Translocation of VLCAD, but not its ΔMBR deletion mutant, to liposomes bearing the lipid composition of the inner mitochondrial membrane, as monitored by VLCAD western analysis of SEC fractions. The experiment was repeated twice with independent preparations of VLCAD protein with similar results. e Quantitation of VLCAD observed in liposomal translocation assay fractions (d) by densitometry using ImageJ software.

Fig. 2. Comparative SEC, enzymatic activities, HDX MS profiles, and membrane interaction analyses of wild-type VLCAD protein and two proline-mutant variants observed in human VLCAD deficiency.

a SDS-PAGE and Coomassie stain of expressed and purified full-length VLCAD protein and its A450P and L462P mutants. The experiment was repeated twice using independent preparations of VLCAD proteins with similar results. b SEC elution profiles of wild-type, A450P, and L462P VLCAD proteins. c Comparative enzymatic activities of wild-type, A450P, and L462P VLCAD proteins, as assessed using a series of long-chain substrates and ferrocenium hexafluorophosphate as the electron acceptor. Data are mean ± s.e.m. for experiments performed in technical quadruplicate and repeated twice with independent preparations of assay reagents with similar results. d, e Deuterium difference plots showing the relative deuterium incorporation in solution of VLCAD A450P (d) and L462P (e) minus the relative deuterium incorporation of wild-type VLCAD, as measured after 10 s, 1 m, 10 m, and 1 h of deuteration. Regions of deprotection and protection above 0.5 Da (dotted line) are considered meaningful. HDX MS experiments were performed twice with independent preparations of VLCAD proteins. See Supplementary Data 1 for the HDX MS data used to create this figure. f Comparative liposomal translocation of wild-type, A450P, and L462P VLCAD proteins, as monitored by VLCAD western analysis of SEC fractions. The experiment was repeated twice with independent preparations of VLCAD proteins with similar results. g Quantitation of VLCAD observed in liposomal translocation assay fractions (f) by densitometry using ImageJ software.

Fig. 3. HDX MS analysis of wild-type and proline-mutant VLCAD proteins in the presence and absence of liposomal membranes.

a-c Deuterium difference plots showing the relative deuterium incorporation of wild-type (a), A450P (b), and L462P (c) VLCAD proteins in the presence of liposomes minus the relative deuterium incorporation in solution, as measured after 10 s, 1 m, 10 m, and 1 h of deuteration. Regions of deprotection and protection above 0.5 Da (dotted line) are considered meaningful. Peptides of the VLCAD C-terminal region that show protection in the presence of liposomal membrane, an effect notably impacted by proline mutagenesis, are shaded. HDX MS experiments were performed twice with independent preparations of VLCAD proteins. See Supplementary Data 1 for the HDX MS data used to create this figure.

Fig. 4. Computational and circular dichroism analyses support a model for proline-mediated disruption of a membrane-interacting hairpin located within VLCAD’s C-terminal domain.

a AlphaFold model structure of dimeric VLCAD demonstrating residues 440–473 as a helix-turn-helix hairpin. Residues A450 and L462, two sites of proline mutagenesis in human VLCAD deficiency, are colored in blue and red, respectively. b A surface view of the predicted α-helical hairpin demonstrates a hydrophobic interface surrounded by a perimeter of positively-charged residues, including lysines 440, 442, 452, 467 and arginines 470, 471, 472. c-e Model structures (AlphaFold) for wild-type (c), A450P (d), and L462P (e) VLCAD proteins demonstrate how proline mutagenesis can disrupt the structure of the α-helical hairpin. f-h Circular dichroism (CD) of wild-type (f), A450P (g), and L462P (h) peptides corresponding to residues 440–473 of the predicted α-helical hairpin in solution and in the presence of liposomes (relative α-helicity in the presence of liposomes of 2.5:1.7:1 for wild-type:A450P:L462P). CD experiments were performed in duplicate using independent preparations of peptides with similar results.

Fig. 5. Molecular dynamics simulations of VLCAD and its proline mutants in the presence of membrane.

a Simulated structure of a wild-type VLCAD monomer in contact with a cardiolipin-containing membrane bilayer (4:1 ratio of phosphatidylcholine [POPC]: tetraoleoyl cardiolipin [TOCL]), as demonstrated by molecular dynamics simulation after 125 ns of equilibration. Residues 446–474 of the α-helical hairpin insert deeply and stably into the lipid membrane. POPC, tan; TOCL, orange; FAD, yellow; substrate, green. b Independently computed electrostatic maps for wild-type VLCAD and lipid membrane, illustrating a complementary potential V between adjacent regions (blue: V > 0; red: V < 0; white, V = 0). c Representative electrostatic interactions between cationic (K, R) residues of VLCAD and cardiolipin headgroups of the lipid membrane or anionic (D, E) amino acids of the protein. To maintain clarity, only the charged phosphates (PO₄²⁻) of TOCL (orange) are shown. d, e Proline mutagenesis disrupts the local structure, resulting in more superficial membrane interaction compared to wild-type VLCAD, as demonstrated by molecular dynamics simulation of the A450P and L462P variants after 125 ns of equilibration. f Comparative geometric quantification of the membrane interactions of wild-type and proline-mutant VLCAD proteins, including, from top to bottom, (1) helical content hlx of the C-terminal region comprised of amino acids 436–488, (2) instantaneous tilt angle θ between a membrane-embedded α-helical segment (residues 460–474) and the membrane surface, (3) height d of the helix center-of-mass with respect to the membrane surface, as defined by a plane through phosphate atoms in the proximal membrane leaflet, whereby values with d < 0 lie below the lipid headgroups, and (4) solvent-accessible surface area A_SASA buried at the VLCAD-lipid interface. Trajectories were smoothed by averaging over 0.5 ns bins, and bands reflect the standard deviation within these bins. Calculations were facilitated by the collective variables module of VMD where applicable. Adjacent bar graphs depict mean values ± s.d. of the geometric parameters, calculated over the last 80 ns equilibration trajectory (n = 320).