Solution Structural Studies of GTP:Adenosylcobinamide-Phosphateguanylyl Transferase (CobY) from Methanocaldococcus jannaschii

GTP:adenosylcobinamide-phosphate (AdoCbi-P) guanylyl transferase (CobY) is an enzyme that transfers the GMP moiety of GTP to AdoCbi yielding AdoCbi-GDP in the late steps of the assembly of Ado-cobamides in archaea. The failure of repeated attempts to crystallize ligand-free (apo) CobY prompted us to explore its 3D structure by solution NMR spectroscopy. As reported here, the solution structure has a mixed α/β fold consisting of seven β-strands and five α-helices, which is very similar to a Rossmann fold. Titration of apo-CobY with GTP resulted in large changes in amide proton chemical shifts that indicated major structural perturbations upon complex formation. However, the CobY:GTP complex as followed by 1H-15N HSQC spectra was found to be unstable over time: GTP hydrolyzed and the protein converted slowly to a species with an NMR spectrum similar to that of apo-CobY. The variant CobYG153D, whose GTP complex was studied by X-ray crystallography, yielded NMR spectra similar to those of wild-type CobY in both its apo- state and in complex with GTP. The CobYG153D:GTP complex was also found to be unstable over time.


Introduction
Coenzyme B 12 (a.k.a. adenosylcobalamin or AdoCbi) is the largest, non-polymeric molecule with biological activity. AdoCbi belongs to the broadly distributed family of cyclic tetrapyrrole molecules known as 'The Pigments of Life', which includes hemes, factor F 430 , and chlorophylls [1]. The core ring structure of AdoCbi (a.k.a. the corrin ring) contains a cobalt ion chelated by pyrrolic nitrogens. On the upper (beta) face of the ring, a covalent bond links 5 0 -deoxyadenosine (Ado) and the Co ion. This unique organometallic bond is critical to the function of the coenzyme. The lower (alpha) face of the ring features a nucleotide loop tethered to a substituent of the ring via a phosphodiester bond. Two features unique to the nucleotide loop are the alpha-N-glycosidic bond between the base and ribosyl moiety, and the diversity in the base [2]. 'Cobamide' is the term used to refer to complete B 12 -like molecules, regardless of their base. The best known cobamide is cobalamin, which contains 5,6-dimethylbenzimidazole as its base.
The assembly of the nucleotide loop evolved differently in bacteria and archaea. In both domains, the pathway starts with the synthesis of AdoCbi-P, which is then converted to AdoCbi-GDP, the so-called activated corrin ring. The difference between the way archaea and bacteria synthesize AdoCbi-GDP lies in the guanylyl transferase that transfers the GMP moiety of GTP to AdoCbi-P. Bacteria use a bi-functional kinase/guanylyl transferase enzyme (CobU, EC 2.7.7.62) [3][4][5], whilst archaea evolved CobY (E.C. 2.7.7.62), a guanylyl transferase that lacks kinase activity [6]. Crystal structures of CobU in its apo form and in complex with GMP are available (PDB 1C9K [5] and 1CBU [7], respectively). The crystal structure of CobY G153D in complex with GTP is also available (PDB 3RSB) [8], but efforts to crystallize the apo-forms of CobY or CobY G153D were unsuccessful.
Results of biochemical experiments performed during the course of this work revealed that two subunits of apo-CobY bind one GTP molecule with a binding constant of K b = 2.0 × 10 −5 M -1 and a dissociation constant of K d = 5.0 × 10 −6 M, but apo-CobY failed to bind GTP analogues, such as GMP-PNP, GMP-PCP or even GDP [9]. CobY binds GTP first before binding AdoCbi-P 200 [9]. The Ado moiety of the corrinoid is required for binding, but the order of binding is clear. The G153D variant of CobY (CobY G153D ) crystallized in the presence of GTP and led to the determination of the 3D structure of the complex by X-ray crystallography at a resolution of 2.8 Å [8]. Repeated failed attempts to crystallize the apo-CobY protein prompted us to explore solution NMR spectroscopy as a means for determining the structure of apo-CobY and its complex with GTP. To aid in answering how CobY binds GTP and is involved in transferring the GMP moiety to AdoCbi-P, we conducted structural studies using nuclear magnetic resonance (NMR) spectroscopy. We report here the solution structure of apo-CobY, which has allowed comparison with the X-ray structure of CobY G153D . We also present NMR studies of CobY G153D and interactions of the proteins with GTP.

Materials and Methods
Protein production and sample preparation [U-15 N]-CobY, [(U-13 C, 15 N]-CobY, and [U-13 C, 15 N]-CobY G153D protein samples containing 196 amino acids (residues 1-196) used NMR studies were produced in minimal medium according to the protocol described previously [10], except that E. coli BL21-CodonPlus 1 (DE3)-RIL (Stratagene)was used for protein production, and cultures were grown in Erlenmeyer flasks. The M.jannaschiicobY gene was expressed from plasmid pCobY14 [9]. Proteins were purified as previously reported [9] with the following modifications. Cell-free extract was applied to a 5 mL HiTrap phenyl (high-sub) FF column (GE Healthcare) equilibrated with tris (hydroxymethyl) aminomethane hydrochloride buffer (50 mM Tris-HCI, pH 8.0 at 4°C) containing 55 g/L (NH 4 ) 2 S0 4 . Protein was eluted at a flow rate of 5 mL / min with a linear gradient to 100% Tris-HCI buffer. CobY-containing fractions were concentrated and dialyzed against Tris-HCI buffer. Protein purity was assessed as previously reported [9] and found to be >95% homogeneous (data not shown). Ion exchange chromatography was therefore omitted.
[U-13 C, 15 N]-CobY protein used for structure determination was further dialyzed against 50 mM deuterated Tris buffer (pH 8.0 at 4°C) containing 50mM NaCl, 5 mM dithiothreitol (DTT) and 10 mM MgCl 2 . To prevent bacterial growth, 0.2% NaN 3 was added to all samples and proteins were stored at 4°C.

NMR Data Collection and Analysis
All NMR spectra were recorded at the National Magnetic Resonance Facility at Madison (NMRFAM) on Varian VNMRS (600 MHz, 800 MHz and 900 MHz) spectrometers equipped with triple-resonance cryogenic probes. The temperature of the sample was regulated at 40°C. Sequence specific backbone resonance assignments were conducted for CobY using a series of 2D and 3D heteronuclear NMR spectra. NMR data were collected for both CobY containing 2.0 mM [U-13 C, 15 N] protein dissolved in NMR buffer with 50 mM Tris, 5 mM DTT, 50 mM NaCl, 10 mM MgCl 2 ,95% H 2 O, 5% D 2 O. Raw NMR data were processed with NMRPipe [11] and analyzed using the programs XEASY [12] and NMRFAM-SPARKY [13]. 2D 1 H-15 N HSQC and 3D HNCO data sets were used to identify the number of spin systems, and these identifications plus 3D HNCACB and 3D CBCA(CO)NH data sets were used as input to the PINE server [14] to determine sequence specific backbone resonance assignments. In addition, backbone resonance assignments were confirmed on the basis of 3D 15 N-edited 1 H-1 H 3D-NOESY. 2D 1 H-13 C HSQC, 3D HBHA(CO)NH, 3D HC(CO)NH, 3D C(CO)NH experiments were used to assign the side chain and HB and HA resonances. 3D 15 N-edited 1 H-1 H NOESY (100 ms mixing time), and 3D 13 C-edited 1 H-1 H NOESY (120 ms mixing time) experiments were used to derive the distance constraints to determine the three dimensional structure of protein [15]. Standard pulse sequences were used to record steady state [ 1 H]-15 N NOE and 15 N relaxation (T 1 , T 2 ) data [16]. To determine the 15 N T 1 values, multiple interleaved NMR spectra were recorded with relaxation delays of 10, 100, 200, 400, 600, 800, 1000, 1200, and 1400 ms. To determine 15 N T 2 values, multiple interleaved NMR spectra were recorded with delays of 10, 30, 50, 70, 90, 110, and 150 ms. Relaxation rates were calculated by leastsquares fitting of peak heights versus relaxation delay to one single exponential decay by using NMRFAM-SPARKY. The reported error estimates are standard deviations derived from fitting the data. Steady-state [ 1 H]-15 N NOE values were calculated from the ratio of peak heights in a pair of NMR spectra acquired with and without proton saturation. The signal-to-noise ratio in each spectrum was used to estimate the experimental uncertainty.

Structure calculation and analysis
For the structure calculation, 15 N resolved 1 H-1 H 3D NOESY and 13 C resolved 1 H-1 H 3D NOESY spectra were used to derive the intra molecular distance restraints. TALOS+ software [17] was used to derive backbone dihedral angle restraints φ and ψ from 1 H, 15 N, 13 CA, 13 CB, 13 C 0 chemical shifts. CYANA (version 3.0) [18] was used for automated NOESY peaks assignments and structure calculation. NOESY peaks assigned automatically by CYANA were used as a guide to further refine the structure. Programs MOLMOL [19] and PYMOL [20] were used, respectively, to calculate the root mean square deviation (rmsd) and for graphical analysis. The PSVS server [21] was used to check the quality of the structure.

Optimization of NMR sample conditions
By optimizing the buffer composition and temperature, we discovered conditions that led to sharp and uniform signals in the 1 H-15 N HSQC spectrum ( Fig 1A) and good triple-resonance and NOESY data, as needed for a successful structure determination. The final conditions were:2 mM protein in 50 mM TRIS buffer pH 7.0 containing 50 mM NaCl and 10 mMMgCl 2 . Data were collected at 40°C.

Structure of apo-CobY
The solution NMR structure of apo-CobY was determined from 3246 distance constraints from NOESY spectra and 220 angle constraints derived from chemical shifts by using the TALOS+ program [17]. Two hundred refined structures were generated, and the best 20 conformers, those with lowest energy that showed the fewest constraint violations with CYANA doi:10.1371/journal.pone.0141297.g001 [18], were chosen for additional water bath refinement using PONDEROSA-C/S [22] assisted Xplor-NIH [23]. Statistics for the solution structure ( Table 1) are indicative of its high quality. The average number of constraints per residue was 17.6, and, of these, an average of 4.2 per residue were long-range constraints. The root mean square deviation (rmsd) for backbone heavy atoms was< 1.0 Å overall and~0.6 Å for backbone heavy atoms in regular secondary structure. Of the backbone torsion angles, 91% were in the most favored and 8% were in additionally allowed regions of the Ramachandran plot. The PROCHECK [24] Z-scores for backbone / all atoms were−0.04/ −0.71. The structure consists of a mixed α/β fold (Fig 2). The seven β-strands (A-G) consist of resi- V157), and β G (E167−V170). The five α-helices (I-V) consist of residues α I (L29−K39), α II (P54−Y64), α III (Y80−Y90), α IV (K108−K123), and α V (T180−K192). The orientations of the β-strands make up a twisted β-sheet (Fig 2A, 2B, and 2C); six of the seven β-strands are arranged in parallel fashion: β C (") β B (") β A (") β D (") β F (#) β E (") β G ("). In addition, a short and stable β-hairpin is located between residues I22 and L29, and a short anti-parallel β-sheetlike structure is formed by residues D101-N104 and I175-N177. Four α-helices (I, II, IV, and V) are arranged on one side of the β-sheet, whereas one α-helix (IV) is on the other side of the β-sheet. α-Helix I is in contact with β-strands A and B, whereas α-helix II is in contact with βstrands B and C. α-helices I and II also contact one other; α-helix III contacts half of the βsheet (β-strands C, B, A, D, and F); and αhelix IV contacts the other side of the β-sheet (βstrands A, D, F, E, and G). The loops connecting the secondary structural elements (α 1 −β A , α 4 −β D , β E −β F , and β G −α 5 ) are highly flexible and unstructured. Resonance assignments could not be obtained for residues in some of these loops because of exchange broadening, which led to the disappearance of the amide peaks. The backbone rmsd plotted against the amino acid sequence (Fig 2D) shows that the polypeptide chain is flexible between residues 8-20 and133-153. The C-terminal helix is also relatively dynamic as determined from heteronuclear NOE values. The coordinates were deposited in the Protein Data Bank (PDB) with accession code 2MZB, and the chemical shifts were deposited in Biological Magnetic Resonance Data Bank (BMRB) with accession code 25482.

Dynamics of ligand free CobY
One of the advantages of NMR spectroscopy over X-ray crystallography is its ability to provide information about the dynamic properties of proteins in solution. For CobY we accomplished this by measuring nitrogen spin-lattice (T 1 ) and spin-spin (T 2 ) relaxation times and heteronuclear NOEs ( 15 N NOEs) for backbone amide resonances. As shown in Fig 5, we found that the relaxation parameters are fairly similar throughout the polypeptide chain, indicating uniform overall protein dynamics. The only exceptions were inflexible loops connecting regular secondary structure elements, in particularα 1 -β A , α 4 -β D , β E -β F , and β G -α 5 ; these regions yielded weak electron density in the X-ray studies of CobY. The average T 1 (~750 ms) and T 2 (~70 ms)values are those expected for a monomeric protein of~23 kDa and are consistent with the 3D NMR solution structure. In addition, the rotational correlation times (τ c ) of the amide protons calculated from T 1 and T 2 revealed an average τ c = 9.9 ± 0.7 ns (Fig 5) consistent with monomeric protein in solution. By contrast, the X-ray structure of the CobY G153D :GTP complex was modeled as a weak dimer [8].
Comparison of the NMR structure of apo-CobY and the X-ray structure of the CobY G153D :GTP complex The solution structure of apo-CobY and the crystal structure of CobY G153D :GTP complex(PDB 3RSB) [8] exhibit similar 3D folds (structures superimposed in Fig 6A). The elements of regular secondary structure (consisting of 67 amino acid residues) superimposed with an average rmsd = 0.83Å. The X-ray structure has a relatively low resolution (2.8 Å), and electron density was not identified for residues 8-11, 74-81, 126-127, and 192-196. Perhaps the largest structural differences are in α-helix-III, which is structured in the NMR solution structure but unstructured in the X-ray structure (Fig 6B). Although isothermal titration calorimetry (ITC) studies suggested one GTP molecule per two units of CobY [9], the X-ray structure of the CobY G153D :GTP complex was modeled as a dimer with one GTP molecule bound to each subunit. The ITC results were obtained with active enzyme that may have been turning over during the experiment. Our NMR experiments with both CobY and CobY G153D are consistent with a 1:1 complex.

Complex formation of CobY with GTP
To probe the effect of added GTP on apo-CobY, we titrated a sample of 15 N-labeled apo-CobY with GTP and followed the chemical shifts in a series of 1 H-15 NHSQC spectra. Several amide cross peaks exhibited large perturbations (Fig 6D-6F), indicating that significant conformational changes accompanied the formation of the GTP complex. 1 H-15 N HSQC spectra acquired following the addition sub-stoichiometric amounts of GTP (not shown), exhibited two sets of peaks, one corresponding to free CobY and one to the CobY:GTP complex; this indicates a slow off rate for GTP dissociation.
Two factors appear to be responsible for these chemical shift changes: (i) GTP-induced ordering of the binding domain and (ii) electrostatic interactions between the ligand and protein backbone. In the X-ray structure of the CobY G153D : GTP complex, the tri-phosphate group of GTP is in the proximity of the region of the protein (residues 10-18) that appears disordered in the apo-enzyme both in X-ray crystal data, which lacked electron density for these residues, and in the solution spectra, which lacked signals from these residues as attributed to exchange broadening. The overlaid expansions of three regions of the 15 N HSQC spectra of CobY and CobY:GTP complex (Fig 6D-6F) indicate that the amide protons of A7, G8, K55, G79, G153 shift significantly upon GTP binding. The weighted chemical shift perturbations (CSPs) mapped on 3D structure of CobY G153D (Fig 6C) show that some are close to the GTP binding site and others are distant.  (Fig 7A) indicates that the chemical shift differences are small. The sample of 15 N-labeled CobY G153D was saturated with GTP, and the resulting spectrum was compared to that of apo-CobY G153D (Fig 7B). The chemical shift perturbations upon GTP binding are very similar to those observed with wild-type CobY (Fig 1B). The differences in the chemical shifts of CobY and CobY G153D are plotted as a function of residue number in Fig 8A and are mapped onto the 3D structure of the protein in Fig 8B. As expected, atoms in residues near residue 153 (the substitution site) exhibit the largest chemical shift differences.

Comparison of CobT and CobY G153D
The GTP complexes with wild-type CobY and CobY G153D proved to be unstable in solution. NMR spectra taken over time (not shown) indicated each complex converted to an unknown intermediate state over a period of about 24 hours (most probably the "switch-off" GDP  complex) and then converted over a period of days to species with spectra resembling those of the apo-proteins. Repeated attempts to make a stable GTP-CobY complex by reducing the temperature and saturating with GTP were unsuccessful. The instability of the complexes prevented us from determining their solution structures.

Conclusions
Solution NMR studies of apo-CobY yielded a 3Dstructure of high quality with a fold is similar to that of the low resolution X-ray structure of the CobY G153D :GTP complex [8]. We found CobY to be monomeric in solution in both its apo-and GTP-bound forms, whereas the X-ray structure of the CobY G153D :GTP complex was modeled as a homodimer. Other differences may reflect problems in tracing the chain in the X-ray map.
It is known that complexes of GTPases with GTP are conformationally flexible to allow for the conversion of GTP to GDP and transfer of the phosphate group. The proposed two-state mechanism has been extensively studied for small GTPases, such as Ras, RhoA, and Sec4 [34]. The active "switch-on" state has GTP bound, whereas the inactive "switch-off" state has GDP bound. Titration studies of CobY followed by NMR spectroscopy revealed that CobY forms a tight 1:1 complex with GTP. However, the complex was found to degrade over time, which prevented the determination of its solution structure.