Crystal structure of cis-aconitate decarboxylase reveals the impact of naturally occurring human mutations on itaconate synthesis

Abstract

cis-Aconitate decarboxylase (CAD, also known as ACOD1 or Irg1) converts cis-aconitate to itaconate and plays central roles in linking innate immunity with metabolism and in the biotechnological production of itaconic acid by Aspergillus terreus We have elucidated the crystal structures of human and murine CADs and compared their enzymological properties to CAD from A. terreus Recombinant CAD is fully active in vitro without a cofactor. Murine CAD has the highest catalytic activity, whereas Aspergillus CAD is best adapted to a more acidic pH. CAD is not homologous to any known decarboxylase and appears to have evolved from prokaryotic enzymes that bind negatively charged substrates. CADs are homodimers, the active center is located in the interface between 2 distinct subdomains, and structural modeling revealed conservation in zebrafish and Aspergillus We identified 8 active-site residues critical for CAD function and rare naturally occurring human mutations in the active site that abolished CAD activity, as well as a variant (Asn152Ser) that increased CAD activity and is common (allele frequency 20%) in African ethnicity. These results open the way for 1) assessing the potential impact of human CAD variants on disease risk at the population level, 2) developing therapeutic interventions to modify CAD activity, and 3) improving CAD efficiency for biotechnological production of itaconic acid.

Keywords: cis-aconitate; decarboxylase; enzymology; itaconic acid; macrophage.

Fig. 1.

The formation of itaconate catalyzed by CAD. Bentley and Thiessen (43) showed that CAD removes the C5 carboxyl group and that a proton from the solvent is added to position C2.

Fig. 2.

Enzyme kinetics of wild-type and mutant CADs. (A) Michaelis–Menten plots showing the differences between CAD from human, mouse, and Aspergillus. (B) Effect of pH on enzyme activity. Enzymes were incubated with 8 mM cis-aconitate for 10 min at 37 °C and formation of itaconic acid was measured by HPLC. Error bars indicate the SD of triplicate assays. (C) Michaelis–Menten and Lineweaver–Burk plots showing the effect of the Asn152Ser and Arg273His mutations on hCAD. The diagram on the right is a zoom-in of the complete diagram shown in the Inset. Error bars in A and C indicate SE of triplicate assays at pH 6.5. Data were fitted to the Michaelis–Menten equation using SigmaPlot, resulting in turnover number (k_cat), Michaelis constant (K_M), k_cat/K_M and R², indicating the goodness of fit (D); 95% CI, 95% confidence intervals.

Fig. 3.

Crystal structure of CAD. (A) Surface and cartoon representation of a dimer of hCAD (Left) with 1 polypeptide chain colored to highlight the 2 domains (blue/red) and 1 chain colored from the N terminus (blue) to the C terminus (red). (Right) The active site with residues potentially involved in substrate binding or catalysis shown as orange sticks. Positive difference electron density indicating the presence of an unidentified ligand is shown in green (contoured at +3 σ). (B) Superposition of hCAD (light gray/orange) with a homology model of A. terreus CAD generated by Phyre2 (dark gray/green). The 2 proteins share 25% sequence identity. The residues of the active site and at the positions of frequent polymorphisms are shown as sticks. Conservative substitutions are highlighted in bold and italics, nonconservative substitutions are bold and underlined. Of 11 residues, 4 are conserved between the enzymes, 3 are conservative, and 4 nonconservative substitutions. (C) hCAD active site residues and residues at positions of frequent polymorphisms shown as sticks. The effect of single-point mutations relative to activity of the wild-type enzyme is shown by colored labels ranging from red (no activity) via orange (0.3 to 0.6%) and yellow (0.6 to 6%) to green (6 to 25%). Mutations leading to unchanged/increased enzymatic activity are highlighted in blue.

Fig. 4.

Sequence alignment of CAD and related enzyme sequences. The Swiss-Prot/UniProt sequences used are: hCAD (IRG1_HUMAN), mCAD (IRG1_MOUSE), aCAD (CAD_ASPTE), IDS epimerase (A. tumefaciens, Q1L4E3_RHIRD), MmgE_Bacillus (Bacillus subtilis 2-methylcitrate dehydratase, MMGE_BACSU), and zCAD (zebrafish CAD, B0UYM1_DANRE). Active-site residues that were mutated to alanine in hCAD are labeled with blue triangles. Orange triangles indicate hCAD polymorphisms that reduced enzyme activity, while pink upward triangles indicate those with neutral/increasing effect. Secondary structure elements are presented on top: helices with squiggles, β-strands with arrows, turns with TT letters. Blue and brown indicate the large and the small domains, respectively. The alignment was prepared with T-Coffee (69) and ESPript (70). Red boxes and red text highlight identical and similar residues, respectively.

Fig. 5.

Active sites of hCAD and related enzymes. (A) Superposition of hCAD (1.7 Å) with mCAD (2.5 Å). The proteins have 84% sequence identity. The rmsd of the C_ɑ positions was 0.85 Å. Active-site residues and positions of frequent polymorphisms are shown as sticks. hCAD was rendered in dark red, dark blue, and orange, whereas light red, blue, and lime were used for mCAD. Differences between both proteins are underlined. The mCAD residue Lys152 was not modeled. (B) Conserved residues in the active site of hCAD, IDS epimerase from A. tumefaciens (PDB ID code 2HP3) and 2-methylcitrate dehydratase (MCDH) from B. subtilis (PDB ID code 5MUX). Asp93, His103, Lys207, and Lys272 are strictly conserved among all 3 enzymes. His163 of MCDH is shifted by 1 residue compared to His159 of hCAD or His150 of IDS epimerase, but the flexibility of the side-chain could compensate for this. Tyr145 of IDS epimerase is a potentially important residue in the enzyme’s reaction mechanism (27). The hydroxyl groups of this residue and Tyr318 of hCAD are positioned very closely in the superposition of the crystal structures.

Fig. 6.

Enzymatic activity of human CAD mutants. (A) The activities of wild-type CAD from human (hCAD) and mouse (mCAD) and of hCAD mutants were tested. Activities of purified enzymes (Left y axis) and itaconic acid production by transfected A549 cells (Right y axis) are shown. The activities of human CAD mutants were determined by incubation of purified proteins with 8 mM cis-aconitate for 10 min at pH 6.5 and 37 °C, followed by itaconic acid measurement by HPLC. Wild-type hCAD was compared to 8 alanine mutants of active site residues and to 8 mutants corresponding to natural polymorphisms of the human ACOD1 gene. Error bars indicate the SD of triplicate assays. (B) Western blot analysis of mCAD and wild-type and mutant mCAD overexpression in transfected A549 cells. β-Actin was used as loading control. Mock, transfection without DNA; vector, transfection with the empty pCMV6-Entry vector. (C) Melting temperatures (T_m) at which 50% of the protein is unfolded were determined by a thermal-shift assay. T_m differences (ΔT_m) between mutants of hCAD and the wild-type protein (T_m = 63 °C) are shown. Error bars indicate SDs from triplicate measurements.

Fig. 7.

Modeling of cis-aconitate binding and implications on the potential reaction mechanism of hCAD. (A) Fitting of cis-aconitate (gray) into the positive F_obs−F_calc difference electron density in the active site of hCAD (green mesh, contoured at σ = +3). The atoms used for the fit are also shown as spheres. The C1 and C6 carboxylates can be modeled well into the electron density, while the leaving C5 carboxylate cannot be attributed to the map. (B) Ligand interaction diagram and (C) corresponding 3D representation of cis-aconitate fitted into the active site as in A. The C1 and C6 carboxylates of the substrate are tightly coordinated by several hydrogen bonds to the enzyme, but the leaving C5 carboxylate is not involved in any interactions. Any possible conformation of this carboxylate moiety would be positioned it in a hydrophobic pocket (yellow), making it likely that this carboxylate would not be charged but protonated. This hydrophobic environment together with the polarizing hydrogen bonds at the C1 and C6 carboxylates would make the formation of an uncharged carbon dioxide favorable. The protonation that has been shown to occur at C2 (43) could be facilitated by His103, which was shown to be crucial for the reaction of hCAD. In this model, the side chain of this residue seems to be oriented appropriately for transferring a proton to C2 of cis-aconitate (dashed orange arrow/line).