Crystal structure of ADP/AMP complex of Escherichia coli adenylate kinase.

Adenylate kinases (AKs, ATP:AMP phosphotransferase, EC 2.7.4.3) are a family of enzymes which catalyze the following reaction: Mg2+ATP + AMP ↔ Mg2+ADP + ADP. Several structural works have revealed that AKs have two highly flexible domains which close over bound substrates.1 The LID domain covers ATP and the site of phosphoryl transfer and the AMP-binding domain closes on AMP when it is bound. Key residues to the domain closure and substrate binding are conserved five arginines in the active site. In Escherichia coli AK, they are Arg36, Arg88, Arg123, Arg156, and Arg167. A previous mutational study suggested an important role of Arg156 in catalysis2 but its function is not well defined although two E. coli AK structures were already solved with P1,P5-di(adenosine 5′)-pentaphosphate (Ap5A) and β-γ-imidoadenosine 5′-triphosphate (AMPPNP)/AMP, respectively.3, 4 Here, we report the crystal structure of E. coli AK with bound ADP and AMP determined at a resolution of 2.8 Å and propose possible roles of Arg156 in substrate binding and catalysis. E. coli AK was expressed and purified as described previously.5 Two crystals were grown by hanging drop vapor diffusion method to dimensions of 0.3 to 0.4 mm × 0.1 mm × 0.05 mm on edge using 2.2M ammonium sulfate, 1% PEG 2000 monomethyl ether, 50 mM MES pH 6.7, 10 mM EDTA, 0.1% sodium azide, and 3 mM ADP and AMP. Data sets were collected on an R-AXIS IIc imaging plate detector. The crystals were found to belong to the space group P21212 with two molecules in the asymmetric unit. The unit cell constants were determined to be a = 73.5, b = 84.6, c = 80.3. The two data sets were merged producing an overall completeness to 2.8 Å of greater than 99.0% with an Rmerge of 14.3%, and consisting of 12852 unique reflections. A molecular replacement solution was found using the program X-PLOR6 with the molecular model of a complex of E. coli AK and Ap5A (Protein Data Bank Entry 1AKE). ADP and AMP molecules were placed manually in their respective binding sites in both protein molecules in the asymmetric unit. The model was refined using noncrystallographic symmetry restraints by cycles of conjugate gradient minimization and temperature factor fitting in the program X-PLOR.6 The solvent addition cycles were conducted until no improvement observed in Rfree. Water molecules with high B-factors and low occupancies were rejected during refinement. The final model gives an R = 19.3% with Rfree = 27.6%. The coordinates were deposited with the Protein Data Bank (entry 2ECK). The ADP/AMP complex of the enzyme is in the fully closed conformation with the LID and AMP-binding domains closed over the bound substrates. The ADP and AMP occupy the ATP and AMP binding sites, respectively. They are bound in a roughly linear arrangement spanning the protein beneath the closed lid domain (Fig. 1). The terminal phosphates of each nucleotide face each other and the distance between the two phosphorus atoms is 6.8 Å, indicating that the reason the ADP/AMP complex is catalytically inactive is that there is a large separation between the terminal phosphates. The present structure shows that the interactions between AK and the bound nucleotides are almost identical to those of AMPPNP/AMP and Ap5A bound forms.3, 4 The α-phosphate of AMP is in the same position as seen in the AMPPNP/AMP complex. Likewise, the phosphate chain of ADP is in the same position and orientation, being bound by the protein's P-loop, as the α- and β-phosphates of AMPPNP and Ap5A in the other structures. The one exception is Arg156 in the LID domain. In our structure, Arg156 has moved slightly from its position in the Ap5A-bound structure, wherein it hydrogen bonds to the nonphysiological δ-phosphate of Ap5A, to form a hydrogen bond/salt bridge with the α-phosphate of AMP. When the present structure is compared to the AMP/AMPPNP complex, in which Arg156 interacts with the γ-phosphate of AMPPNP, we see that the guanidinium head group of Arg156 rotates about 180° around the CγCδ bond. Close-up view of the active site showing key residues involved in catalysis. Carbon, nitrogen, oxygen, and phosphorus atoms are shown in black, blue, red, and purple, respectively. Potential interactions between the residues and the substrates are displayed as broken lines. The absence of interaction between Arg156 and the bound ADP indicates that Arg156 may not be required for a closure of the LID domain over the ATP site. This is supported by the structure of a yeast AK mutant where the equivalent of Arg156 is mutated to Ile and an ATP analogue is bound in the ATP site.7 In this structure, the LID domain is in the fully closed conformation. Rather, Arg156 may induce movement of the LID domain depending on the state of the AMP-binding domain by its interacting with AMP when the substrate binds to the AMP-binding domain. This result suggests that the enzyme reaction may not be modeled accurately by a random bi–bi mechanism that most AKs are believed to follow because the mechanism requires independent motions of the LID and AMP-binding domains.8 In fact, the interaction between the equivalent of Arg156 in another AK and its bound AMP in the AMP-binding domain was found and the role of the Arg residue was proposed as a trigger in the possible cooperative motions of the two domains.9 The variation of the Arg156 position in different structures also displays mobility of the residue, supporting a role in stabilizing the enzyme's phosphoryl transfer reaction. Phosphate transfer in AKs needs to be stabilized and protected from water to prevent phosphate hydrolysis. The mobility of the Arg156 side chain between binding states suggests a role in stabilizing the top (the LID domain side) of the intermediate, while Lys13 in the P-loop and the Mg2+ ion, which is required for the reaction, stabilize the bottom and front.