Structure of a membrane-embedded prenyltransferase homologous to UBIAD1

Abstract

Membrane-embedded prenyltransferases from the UbiA family catalyze the Mg2+-dependent transfer of a hydrophobic polyprenyl chain onto a variety of acceptor molecules and are involved in the synthesis of molecules that mediate electron transport, including Vitamin K and Coenzyme Q. In humans, missense mutations to the protein UbiA prenyltransferase domain-containing 1 (UBIAD1) are responsible for Schnyder crystalline corneal dystrophy, which is a genetic disease that causes blindness. Mechanistic understanding of this family of enzymes has been hampered by a lack of three-dimensional structures. We have solved structures of a UBIAD1 homolog from Archaeoglobus fulgidus, AfUbiA, in an unliganded form and bound to Mg2+ and two different isoprenyl diphosphates. Functional assays on MenA, a UbiA family member from E. coli, verified the importance of residues involved in Mg2+ and substrate binding. The structural and functional studies led us to propose a mechanism for the prenyl transfer reaction. Disease-causing mutations in UBIAD1 are clustered around the active site in AfUbiA, suggesting the mechanism of catalysis is conserved between the two homologs.

Figure 1. The UbiA fold.

(A) A representative reaction catalyzed by UbiA family members. Polyprenyl is transferred from diphosphate to 4HB to form 3-polyprenyl-4-hydroxybenzoate, a precursor to ubiquinone. The square brackets denote a single five-carbon prenyl unit. (B) The two conserved aspartate-rich motifs characteristic of the UbiA family. Residue numbers correspond to the Archaeoglobus fulgidus UbiA sequence. Ec, Escherichia coli; Sc, Saccharomyces cerevisiae; Hs, Homo sapiens; Af, Archaeoglobus fulgidus. (C) Topology diagram of AfUbiA, with the transmembrane helices colored in pairs of equivalent helices in the four-helix bundles. Dashed lines indicate the region of L2–3 that is disordered in the SeMet crystal structure. (D) Cartoon representation of the AfUbiA structure viewed from within the plane of the membrane (left) and from the extracellular side of the membrane (right). The transmembrane helices are colored according to the same scheme as in panel (C). Orange arrows indicate the two pseudosymmetric bundles. (E) Magnified stereo view of the boxed area in panel (D), showing residues from aspartate-rich Motif I and II as sticks.

Figure 2. The substrate-binding cavity and possible substrate tunnel.

(A) The solvent-accessible surface of the central cavity in the GPP-bound AfUbiA structure is shown as a blue mesh. Helix TM9 is highlighted in pink. (B) A ribbon representation of the AfUbiA structure in which the thickness and color of the ribbon indicate the degree of conservation. Highly conserved residues are thicker and colored dark purple; poorly conserved residues are thinner and colored teal. Residues colored yellow were not included in the multiple sequence alignment used to calculate the conservation scores. GPP and Mg²⁺ are shown in green. (C) Cutaway view of the central cavity and putative substrate tunnel from two perpendicular directions in the plane of the membrane, with the extracellular side on top. The molecule of GPP is shown as sticks. The pink arrow indicates a hydrophobic tunnel that opens into the membrane bilayer.

Figure 3. Substrate binding and active site.

(A) Stereo view of the GPP binding site, viewed from the cytoplasmic side of the membrane. Two Mg²⁺ atoms (green spheres) and a GPP molecule are shown in the binding site. Residues that potentially bind to Mg²⁺ and the diphosphate are labeled. (B) Stereo view of the active site from within the plane of the membrane. Conserved residues proposed to stabilize the intermediate state are labeled. The green mesh in both figures corresponds to F_o-F_c density contoured at 3.0 σ. (C and D) Binding of GPP to detergent-solubilized AfUbiA measured by ITC. Heats from successive injections of GPP were measured in the presence of 2 mM MgCl₂ (C) or 1 mM EDTA (D). Right panel in (C) shows the fit to a one-site model. (E) Table of thermodynamic values for GPP binding to WT and mutant AfUbiA measured by ITC. K_D, ΔH, and n were obtained by fitting a binding isotherm described in the Methods section. The thermodynamic relation ΔG = ΔH−TΔS was used to calculate −TΔS at 25°C with errors propagated. “N.D.” indicates no binding detected. Each value is the mean and s.e.m. of three ITC experiments.

Figure 4. Comparison of the AfUbiA and ApUbiA active sites.

(A and B) Interactions between Mg²⁺, GPP/GSPP, and residues in the first and second aspartate-rich motifs are shown for (A) AfUbiA and (B) ApUbiA (PDB accession 4OD5). All interaction distances of 3.0 Å or less are marked.

Figure 5. Conservation of active site residues across the UbiA family.

(A) A sequence alignment of AfUbiA, human UBIAD1, and EcMenA. Residues currently known to be mutated in SCD patients are highlighted in pink, red, or blue in the AfUbiA and HsUBIAD1 sequences. Red indicates mutated residues that are identical between the two proteins; blue indicates residues that align with gaps in the AfUbiA sequence. Green boxes indicate locations of residues that were mutated in the EcMenA functional assays. The colored bars above the alignment indicate locations of transmembrane helices in the AfUbiA crystal structure. (B) The locations of the active site residues marked with green boxes in panel (A) are shown relative to bound Mg²⁺ and GPP in the AfUbiA structure. Black labels correspond to residue numbers in the AfUbiA structure, green labels to the equivalent residues in EcMenA. (C) A menA ⁻ E. coli strain was transformed with plasmids containing either WT or mutant EcMenA, and grown in suspension cultures in an anaerobic chamber. The optical densities at 600 nm were measured after 24 h. Cells transformed with an unrelated protein (TrkH) were used as the negative control. Error bars are standard deviations of three experiments. Data used to calculate the bar graphs are shown in Table S2. (D) Membranes were purified from E. coli overexpressing WT or mutant EcMenA and incubated at 37°C for 10 min with 2 mM DHNA, 1 mM GPP, and 5 mM MgCl₂. Product formation was measured by HPLC and is shown as a percentage of the activity for WT EcMenA. Membranes from cells overexpressing EcUbiA, which is selective for 4HB as the prenyl acceptor, were used as a negative control. Error bars are standard deviations of three experiments. Data used to calculate the bar graphs are shown in Table S3. (E) Residues currently known to be mutated in SCD patients are shown as spheres on the structure of AfUbiA. (F) The same residues are shown as sticks in a closer view of the substrate-binding cavity from within the plane of the membrane (left) and from the intracellular side (right).

Figure 6. The proposed UbiA prenyltransferase reaction mechanism.

Schematic showing a potential three-stage ionization (i), condensation (ii), and elimination (iii) reaction mechanism for prenyl transfer, with key conserved residues on the enzyme highlighted. The prenyl acceptor shown is the proposed substrate for UBIAD1, the reduced form of menadione (2-methyl-1,4-dihydroxynaphthoquinol); residue numbers are for AfUbiA and human UBIAD1 (in parentheses).