Crystal structure of a PduO‐type ATP:cobalamin adenosyltransferase from Burkholderia thailandensis

Abstract

ATP:cobalamin adenosyltransferase (EC 2.5.1.17), an enzyme in the vitamin B12 (cobalamin) metabolic pathway, converts cobalamin to adenosylcobalamin (coenzyme B12) by transferring a 5′-deoxyadenosyl moiety from ATP to the cobalt atom in cobalamin.1 Adenosylcobalamin functions as a radical reservoir, and enzymes utilize this organometallic cofactor to catalyze 1,2-rearrangement reactions that interchange a hydrogen atom and a variable group (OH, NH2, or a carbon-containing group) on adjacent carbons.2 Amino acid sequence analysis have revealed three types of adenosyltransferases: CobA, PduO, and EutT.3 The crystal structure of CobA from Salmonella typhimurium4 showed a homodimer with mixed α/β-folds, but the structure of PduO from Thermoplasma acidophilum5 revealed a homotrimer with mainly α-folds (a five-helix bundle); a three-dimensional structure of EutT has not yet been reported. They share little sequence identity (<20%) and have thus far yielded different three-dimensional structures, even though they catalyze the same reactions, suggesting that particular types of adenosyltransferases are specialized for particular B12-dependent enzymes or for the de novo cobalamin biosynthesis.6 Among the adenosyltransferases, PduO is the most widely distributed enzyme with homologs occurring from archaea to human. PduO has drawn considerable interest, because the human PduO homolog (MMAB gene) is defective in at least two cobalamin metabolic disorders: methylmalonic aciduria and metabolic ketoacidosis.7 Several crystal structures of PduO adenosyltransferases are available in the Protein Data Bank database, but to date only four crystal structures have been reported (from T. acidophilum,5 human,8 Lactobacillus reuteri,9 and Sulfolobus tokodaii10). Of these structures, only two (human and L. reuteri) described the substrate-binding site with ATP bound or detailed the role of the conserved residues around the N-terminus. The PduO protein from Burkholderia thailandensis possesses 35 and 40% sequence identity to the T. acidophilum and human PduO proteins, respectively. It was predicted to be a PduO-type adenosyltransferase, because all conserved residues comprising the putative PduO active site aligned completely with known PduO adenosyltransferases. In this report, we have determined crystal structures for this PduO adenosyltransferase from B. thailandensis at a 1.8-Å resolution, with and without its substrates, a Mg2+ ion and an ATP molecule (MgATP). Furthermore, we demonstrated concrete and specific functions for the conserved ATP-binding residues by comparing the native and MgATP-complexed structures. These structural studies provide valuable information for understanding the molecular mechanism of human methylmalonic aciduria, a cobalamin metabolic disorder disease. The full-length PduO gene was amplified by polymerase chain reaction (PCR) from the genomic DNA of B. thailandensis. The PCR product was digested with NcoI and XhoI restriction enzymes and ligated into the pET-28a expression vector (Novagen), which contained a hexahistidine tag at the C-terminus. The resulting plasmid was transformed into E. coli BL21 (DE3), and the cells were grown at 37°C in Luria-Bertani medium supplemented with kanamycin (50 μg/mL). Expression of the recombinant PduO protein was induced by 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) when the cells had reached an optical density at 600 nm of about 0.45. The cells were grown for an additional 12 h at 18°C and harvested by centrifugation at 6000 rpm for 30 min at 4°C. The pelleted cells were suspended in buffer A (20 mM Tris-HCl pH 7.9, 500 mM NaCl, 10 mM β-mercaptoethanol) and homogenized by sonication. The crude lysate was centrifuged at 15,000 rpm for 1 h at 4°C, and the supernatant was loaded onto a Ni2+-chelated HiTrap metal-chelating column (GE Healthcare) that had been equilibrated with buffer A. The protein was eluted with a linear gradient of buffer A containing 1M imidazole. The fractions containing PduO were pooled and concentrated before purification. Final purification of the PduO protein was achieved by gel filtration on a HiLoad 16/60 Superdex 200 (GE Healthcare) that had been equilibrated with buffer B (20 mM Tris-HCl pH 7.9, 200 mM NaCl, 4 mM DTT). The purified protein was concentrated to 15 mg/mL using an Amicon Ultra-15 ultrafiltration device (Millipore) for crystallization experiments. This procedure yielded ∼45 mg of PduO protein from a 1-L culture. The homogeneity of the purified protein was confirmed by SDS-polyacrylamide gel electrophoresis. The protein concentration was estimated by the Bradford method using a Bio-Rad protein assay reagent. Preliminary crystallization experiments were carried out by the sitting-drop vapor diffusion method at 22°C on 96-well plates, using a Hydra II Plus One crystallization robot (Matrix Technologies). Initial crystallization trials were set up using Crystal Screen I and II, MembFac Screen, Natrix Screen, Index (Hampton Research), and Macrosol (Molecular Dimensions). After obtaining the initial crystallization conditions that yielded microcrystals, crystal growth was scaled up and further optimized using hanging-drop vapor diffusion in 24-well plates. Each hanging drop was prepared by mixing 1 μL of protein solution and 1 μL of reservoir solution over 500 μL of reservoir solution. Optimal crystallization conditions were as follows: 100 mM sodium citrate, pH 5.7, 22% (w/v) isopropanol, and 12% (v/v) PEG 4000. To obtain crystals of the PduO-MgATP complex, native PduO crystals were soaked in reservoir solution containing 2 mM MgATP. For both the native and MgATP-complexed crystals, X-ray diffraction data were collected at beam lines 4A and 6C of the Pohang Light Source in Korea. For cryogenic experiments, crystals were flash-frozen in a liquid nitrogen stream, with 20% (v/v) glycerol as a cryoprotectant. Data sets for both the native and MgATP-complexed crystals were collected to a resolution of 1.8 Å, and they were processed and scaled using the HKL2000 program.11 Crystals for both native and MgATP-complexed PduO belonged to a C-centered orthorhombic space group C2221. The asymmetric units contain three molecules of PduO protein each, resulting in a crystal volume per protein mass (Vm) of 2.64 Å3/Da and a solvent content of 52% for the native crystals. The crystal structure of the PduO protein was solved by molecular replacement (MR) method using the CNS program,12 with the Mycobacterium tuberculosis PduO homolog (Protein Data Bank ID: 2G2D and 51% sequence identity) as a search model. After obtaining a good MR solution, we built the remainder of the protein model manually using the COOT program.13 Subsequent crystallographic refinements including bulk solvent correction, simulated annealing, positional refinement, and individual thermal factor refinement were performed with the CNS program. All nonglycine residues were located within the most favored or within additional allowed regions of the Ramachandran plot, as determined by analysis with the PROCHECK program.14 Data collection and refinement statistics are summarized in Table I. The final coordinates and structure factors for native and MgATP-complexed PduO have been deposited in the Protein Data Bank under accession codes 2ZHY and 2ZHZ, respectively. The PduO-type ATP:cobalamin adenosyltransferase from B. thailandensis consisted of five α-helices, α1 (residues 31–50), α2 (56–77), α3 (86–101), α4 (116–141), and α5 (147–169), which form a five-helix bundle; two antiparallel short β-strands at the N-terminus, β1 (residues 17–19) and β2 (25–27), are important for ATP binding. Similar to the currently available structures of other PduO adenosyltransferases, the PduO protein from B. thailandensis was a homotrimer. Within the overall native and MgATP-complexed structures, the monomeric subunits showed almost identical conformations. Over 172 Cα atoms, the root-mean-square deviations between each subunit ranged from 0.47 to 0.72 Å. The trimer assembly was mediated by ionic interactions (salt bridges) and hydrogen bonds between the subunits. The following residues comprised the interacting residues on the trimer interface in the B. thailandensis PduO protein (they are completely conserved within all known PduO proteins): Asp14 and Arg154, Asp40 and Arg125, and Glu41 and Arg130 (salt bridges); Gln66 and Pro112 and His122 and Arg125 (hydrogen bonds). All of the critical residues for enzyme activity and substrate binding among the PduO-type adenosyltransferases were also conserved in the B. thailandensis PduO protein [Fig. 1(a)]. Structure of B. thailandensis PduO adenosyltransferase. (a) Multiple sequence alignment of PduO-type ATP:cobalamin adenosyltransferases. The residue numbers refers to those in B. thailandensis. Completely conserved residues are shown in red boxes with white characters. Secondary structural elements are indicated above the sequences for B. thailandensis and below for human. Residues that participate in trimer interactions and ATP binding are indicated by blue and red triangles below the sequences, respectively. Bta, B. thailandensis; Lre, Lactobacillus reuteri; Bsu, Bacillus subtilis; Bha, Bacillus halodurans; Sto, Sulfolobus tokodaii; Tac, Thermoplasma acidophilum; Mtu, Mycobacterium tuberculosis; Hum, Homo sapiens. The panel (a) was produced with ESPript program.15 (b) Superposition of the PduO-type ATP:cobalamin adenosyltransferase structures from B. thailandensis, L. reuteri (shown in pink), and human (shown in blue). The native and MgATP-complexed structures from B. thailandensis are shown in light green and green, respectively. The ATP molecule is displayed as a stick model and the Mg2+ ion as a CPK model. (c) A stereoview of the ATP-binding site; the ATP-interacting residues from B. thailandensis, L.reuteri, and human are shown in green, pink, and blue, respectively. (d) Comparison of the active sites in the native and MgATP-bound structures from B. thailandensis. For clarity, the Mg2+ ion was omitted, and the same color scheme used for (b) was applied. All figures except (a) were drawn with PyMOL program.16 The overall structure of our MgATP-bound PduO from B. thailandensis was quite similar to those from human and L. reuteri [Fig. 1(b)]. The root-mean-square deviations were 1.12 Å over 162 Cα atom pairs and 1.10 Å over 150 Cα atom pairs, respectively. The ATP-binding residues in the MgATP-bound structures from human and L. reuteri aligned structurally with the corresponding residues in our B. thailandensis MgATP-bound structure (Gly13, Lys28, Arg129, Glu132, Arg133, and Gln153) and are also completely conserved among PduO-type adenosyltransferase proteins. Our MgATP-bound structure allowed us to describe the interaction between PduO and ATP in great detail. The NE2 nitrogen atom of the Arg129 residue in PduO hydrogen bonded with oxygen atoms within the ribose ring and phosphate bridge of the ATP. We observed additional hydrogen bonding between the ATP N1 adenosine nitrogen atom and the NH1 nitrogen atom of Arg133, as well as the N6 nitrogen atom from the adenosine amide group within ATP and the OE1 oxygen atom of Glu132. A hydrogen bond also formed between the NH1 nitrogen atom of Lys28 and β-phosphate oxygen of ATP. Oxygen atoms of γ-phosphate of ATP also hydrogen bonded with the main chain nitrogen atom of Gly13 and the OD1 oxygen atom of Asn153. Interestingly, the side-chain oxygen of Thr18 and the main chain carbonyl oxygen of Gly19 (corresponding to Ser68-Ser69 in the human and Thr13-Arg14 in the L. reuteri homologs) contributed to ATP binding by interacting with the β-phosphate and ribose oxygens of ATP, even though they are not conserved residues. One difference between the ATP binding in our B. thailandensis structure and those depicted in other previously solved MgATP-bound structures is the presence of an additional interaction with the ATP molecule. In our B. thailandensis structure, the NH2 nitrogen atom of Arg11 interacted with the oxygen atom of the γ-phosphate of ATP, but the side chains of the corresponding residues in L. reuteri (Lys6) and human (Lys61) point in a totally different direction, being incapable of interacting with ATP [Fig. 1(c)]. When comparing the native and MgATP-complexed structures of B. thailandensis, we observed no significant structural differences within their overall conformations (root-mean-square deviation value of 0.27 Å over 172 Cα atoms). Binding of the ATP molecule widened the cleft forming the substrate-binding site by swinging out the N-terminal loop (residues from 5 to 16). ATP binding also changed the side-chain conformation of Arg129, a residue that is completely conserved among PduO-type adenosyltransferases. In the native structure, the Arg129 residue could form hydrogen bonds with Glu132 and Asn153 and hold them in the positions necessary for them to interact with the ATP molecule. When the ATP molecule bounded to the protein, the direction of the Arg129 side chain moved toward the ATP molecule. This conformational shift allowed this arginine residue, as well as Glu132 and Asn153, to interact with ATP. Within our native structure, a hydrogen bond network allowed three ATP-interacting residues, Gly13, Lys28, and Glu132, to place their side chains in a conformation ready to bind ATP. Additionally, Asp14 kept the side-chain conformation of Asn153 in the exact position necessary for ATP binding, but Asp14 also played an important role in trimer formation through a salt bridge with Arg154 [Fig. 1(d)]. The most striking structural feature between the native and MgATP-bound structures of B. thailandensis PduO was the minimal structural alteration as a result of ATP binding, contrary to our expectations. These observations surprised us because previous native PduO structures determined by X-ray crystallography did not reveal an ordered N-terminal region that participated in ATP binding; only the MgATP-bound structures from human and L. reuteri showed that region clearly. We therefore assumed that ATP binding triggered a significant conformational change that increased the order within the N-terminal region. However, in our native structure of B. thailandensis PduO, we were able to definitively build an ordered N-terminal region without the presence of the bound MgATP. In addition, the ordered N-terminal region in the native structure had almost the same side-chain conformations within the substrate binding site as did the MgATP-bound structure. This surprising observation implied that formation of the active site in a PduO-type adenosyltransferase does not require significant conformational changes in N-terminal loop upon ATP binding. In other words, there is no need to first bind ATP in order to form the active site. As structures of PduO complexed with both ATP and cobalamin will more clearly define the roles of the conserved active site residues and allow us to propose a molecular mechanism of adenosyl transfer by this enzyme, cocrystallization experiments with B. thailandensis PduO, cobalamin, and ATP are in progress. We thank the staff for assistance during X-ray data collection experiments at beamlines 4A and 6C of Pohang Light Source, Korea. We also thank Dr. H. K. Song for his kind help in allowing us to use the Hydra II crystallization robot.