The Crystal Structure of the Escherichia coliMobA Protein Provides Insight into Molybdopterin Guanine Dinucleotide Biosynthesis
2000; Elsevier BV; Volume: 275; Issue: 51 Linguagem: Inglês
10.1074/jbc.m007406200
ISSN1083-351X
AutoresM.W. Lake, Carrie A. Temple, K.V. Rajagopalan, Hermann Schindelin,
Tópico(s)Porphyrin Metabolism and Disorders
ResumoThe molybdenum cofactor (Moco) is found in a variety of enzymes present in all phyla and comprises a family of related molecules containing molybdopterin (MPT), a tricyclic pyranopterin with a cis-dithiolene group, as the invariant essential moiety. MPT biosynthesis involves a conserved pathway, but some organisms perform additional reactions that modify MPT. In eubacteria, the cofactor is often present in a dinucleotide form combining MPT and a purine or pyrimidine nucleotide via a pyrophosphate linkage. In Escherichia coli, the MobA protein links a guanosine 5′-phosphate to MPT forming molybdopterin guanine dinucleotide. This reaction requires GTP, MgCl2, and the MPT form of the cofactor and can efficiently reconstituteRhodobacter sphaeroides apo-DMSOR, an enzyme that requires molybdopterin guanine dinucleotide for activity. In this paper, we present the crystal structure of MobA, a protein containing 194 amino acids. The MobA monomer has an α/β architecture in which the N-terminal half of the molecule adopts a Rossman fold. The structure of MobA has striking similarity to Bacillus subtilis SpsA, a nucleotide-diphospho-sugar transferase involved in sporulation. The cocrystal structure of MobA and GTP reveals that the GTP-binding site is located in the N-terminal half of the molecule. Conserved residues located primarily in three signature sequence motifs form crucial interactions with the bound nucleotide. The binding site for MPT is located adjacent to the GTP-binding site in the C-terminal half of the molecule, which contains another set of conserved residues presumably involved in MPT binding. The molybdenum cofactor (Moco) is found in a variety of enzymes present in all phyla and comprises a family of related molecules containing molybdopterin (MPT), a tricyclic pyranopterin with a cis-dithiolene group, as the invariant essential moiety. MPT biosynthesis involves a conserved pathway, but some organisms perform additional reactions that modify MPT. In eubacteria, the cofactor is often present in a dinucleotide form combining MPT and a purine or pyrimidine nucleotide via a pyrophosphate linkage. In Escherichia coli, the MobA protein links a guanosine 5′-phosphate to MPT forming molybdopterin guanine dinucleotide. This reaction requires GTP, MgCl2, and the MPT form of the cofactor and can efficiently reconstituteRhodobacter sphaeroides apo-DMSOR, an enzyme that requires molybdopterin guanine dinucleotide for activity. In this paper, we present the crystal structure of MobA, a protein containing 194 amino acids. The MobA monomer has an α/β architecture in which the N-terminal half of the molecule adopts a Rossman fold. The structure of MobA has striking similarity to Bacillus subtilis SpsA, a nucleotide-diphospho-sugar transferase involved in sporulation. The cocrystal structure of MobA and GTP reveals that the GTP-binding site is located in the N-terminal half of the molecule. Conserved residues located primarily in three signature sequence motifs form crucial interactions with the bound nucleotide. The binding site for MPT is located adjacent to the GTP-binding site in the C-terminal half of the molecule, which contains another set of conserved residues presumably involved in MPT binding. molybdenum cofactor molybdopterin molybdopterin guanine dinucleotide dimethyl sulfoxide reductase The molybdenum cofactor (Moco)1 is an essential component of a diverse group of enzymes catalyzing important redox transformations in the global carbon, nitrogen, and sulfur cycles (1Hille R. Chem. Rev. 1996; 96: 2757-2816Crossref PubMed Scopus (1470) Google Scholar, 2Kisker C. Schindelin H. Rees D.C. Annu. Rev. Biochem. 1997; 66: 233-267Crossref PubMed Scopus (439) Google Scholar). The Moco consists of a mononuclear molybdenum or tungsten coordinated by the dithiolene moiety of a family of tricyclic pyranopterin structures, the simplest of which is commonly referred to as molybdopterin (MPT) (3Rajagopalan K.V. Adv. Enzymol. 1991; 64: 215-290PubMed Google Scholar, 4Rajagopalan K.V. Johnson J.L. J. Biol. Chem. 1992; 267: 10199-10202Abstract Full Text PDF PubMed Google Scholar). In addition to MPT, various dinucleotide forms of the cofactor exist such as molybdopterin guanine dinucleotide (MGD) and molybdopterin cytosine dinucleotide (5Meyer O. Frunzke K. Tachil J. Volk M. Stiefel E.I. Molybdenum Enzymes, Cofactors and Chemistry. American Chemical Society, Washington, D. C.1993Google Scholar, 6Johnson J.L. Bastian N.R. Rajagopalan K.V. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 3190-3194Crossref PubMed Scopus (155) Google Scholar). The available crystal structures of enzymes containing Moco variants (7Chan M.K. Mukund S. Kletzin A. Adams M.W.W. Rees D.C. Science. 1995; 267: 1463-1469Crossref PubMed Scopus (530) Google Scholar, 8Romão M.J. Archer M. Moura I. LeGall J. Engh R. Schneider M. Hof P. Huber R. Science. 1995; 270: 1170-1176Crossref PubMed Scopus (457) Google Scholar, 9Schindelin H. Kisker C. Hilton J. Rajagopalan K.V. Rees D.C. Science. 1996; 272: 1615-1621Crossref PubMed Scopus (440) Google Scholar, 10Schneider F. Löwe J. Huber R. Schindelin H. Kisker C. Knäblein J. J. Mol. Biol. 1996; 263: 53-69Crossref PubMed Scopus (254) Google Scholar, 11Kisker C. Schindelin H. Pacheco A. Wehbi W.A. Garrett R.M. Rajagopalan K.V. Enemark J.H. Rees D.C. Cell. 1997; 91: 973-983Abstract Full Text Full Text PDF PubMed Scopus (433) Google Scholar, 12Boyington J.C. Gladyshev V.N. Khangulov S.V. Stadtman T.C. Sun P.D. Science. 1997; 275: 1305-1308Crossref PubMed Scopus (484) Google Scholar, 13McAlpine A.S. McEwan A.G. Shaw A.L. Bailey S. J. Biol. Inorg. Chem. 1997; 2: 690-701Crossref Scopus (152) Google Scholar, 14Czjzek M. Santos J.-P.D. Pommier J. Giordano G. Méjean V. Haser R. J. Mol. Biol. 1998; 284: 435-447Crossref PubMed Scopus (158) Google Scholar, 15Dias J.M. Than M.E. Humm A.E. Huber R. Bourenkov G.P. Bartunik H.D. Bursakov S. Calvete J. Caldeira J. Carneiro C. Moura J.J.G. Moura I. Romao M.J. Structure. 1999; 7: 65-79Abstract Full Text Full Text PDF PubMed Scopus (283) Google Scholar, 16Dobbek H. Gremer L. Meyer O. Huber R. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 8884-8889Crossref PubMed Scopus (215) Google Scholar, 17Hu Y. Faham S. Roy R. Adams M.W.W. Rees D.C. J. Mol. Biol. 1999; 286: 899-914Crossref PubMed Scopus (117) Google Scholar) have revealed considerable structural diversity in the cofactor. The cofactor can exist in a mononucleotide as well as various dinucleotide forms, and either one or two copies of the different Moco variants ligate the metal. The dinucleotide forms of the cofactor appear to be present only in eubacteria. All known molybdoenzymes in Escherichia coli contain the bis-MGD form of the Moco that consists of two MGD molecules and a single molybdenum. Based on sequence similarities, all of these enzymes belong to the DMSOR family of Moco-containing enzymes (1Hille R. Chem. Rev. 1996; 96: 2757-2816Crossref PubMed Scopus (1470) Google Scholar, 2Kisker C. Schindelin H. Rees D.C. Annu. Rev. Biochem. 1997; 66: 233-267Crossref PubMed Scopus (439) Google Scholar). The crystal structure ofRhodobacter sphaeroides DMSOR (9Schindelin H. Kisker C. Hilton J. Rajagopalan K.V. Rees D.C. Science. 1996; 272: 1615-1621Crossref PubMed Scopus (440) Google Scholar, 18Li H.-K. Temple C.A. Rajagopalan K.V. Schindelin H. J. Am. Chem. Soc. 2000; 32: 7673-7870Crossref Scopus (170) Google Scholar) revealed that the bis-MGD Moco is deeply buried in the enzyme. The molybdenum is ligated by the sulfurs of both dithiolene groups with an approximate 2-fold axis of symmetry passing through the metal. Only the molybdenum and some of its coordinating ligands are accessible from the solvent through an active site funnel. Genes involved in Moco biosynthesis have been identified in eubacteria, archaea, and eukarya. Besides the evolutionarily conserved common aspects of Moco biosynthesis (19Rajagopalan K.V. Neidhardt F.C. Escherichia coli and Salmonella typhimurium. American Society for Microbiology, Washington, D. C.1996: 674-679Google Scholar, 20Rajagopalan K.V. Biochem. Soc. Trans. 1997; 25: 757-761Crossref PubMed Scopus (64) Google Scholar), an additional step is present in eubacteria in which the MPT form of the cofactor is converted into a dinucleotide form (21Johnson J.L. Indermaur L.W. Rajagopalan K.V. J. Biol. Chem. 1991; 266: 12140-12145Abstract Full Text PDF PubMed Google Scholar). In E. coli, the mobAB locus is responsible for the final step in Moco biosynthesis. E. coli MobA, also referred to as protein FA, consists of 194 amino acids and is responsible for the conversion of MPT and GTP to MGD (21Johnson J.L. Indermaur L.W. Rajagopalan K.V. J. Biol. Chem. 1991; 266: 12140-12145Abstract Full Text PDF PubMed Google Scholar, 22Palmer T. Vasishta A. Whitty P.W. Boxer D.H. Eur. J. Biochem. 1994; 222: 687-692Crossref PubMed Scopus (48) Google Scholar). Proteins with homology to MobA are present in a variety of eubacteria (Fig. 1). It is currently unclear whether Moco-containing enzymes in all of these organisms utilize MGD or other dinculeotide forms of the cofactor. MobB is a GTP-binding protein, with weak intrinsic GTPase activity (23Eaves, D. J., Palmer, T., and Boxer, D. H. (1997)246, 690–697.Google Scholar). Reportedly, MobB is not essential forin vitro reconstitution of apo-nitrate reductase, but it appears to increase the yield of active nitrate reductase (24Palmer T. Santini C.-L. Iobbi-Nivol C. Eaves D.J. Boxer D.H. Giordano G. Mol. Microbiol. 1996; 20: 875-884Crossref PubMed Scopus (145) Google Scholar).Figure 1Sequence alignment of six MobA proteins from different species. Strictly conserved and similar residues arehighlighted in black and enclosed in abox, respectively. The four signature sequences are labeledI–IV. These six sequences were chosen from a set of MobA sequences to include those organisms most likely to synthesize the MGD form of the Moco. This alignment was generated with the program ALSCRIPT (44Barton G.J. Protein Eng. 1993; 6: 37-40Crossref PubMed Scopus (1110) Google Scholar). Secondary structure elements as determined with PROMOTIF (45Hutchinson E.G. Thornton J.M. Protein Sci. 1994; 3: 2207-2216Crossref PubMed Scopus (841) Google Scholar) are indicated.View Large Image Figure ViewerDownload Hi-res image Download (PPT) The process of cofactor insertion into Moco-containing enzymes has been studied in some detail (24Palmer T. Santini C.-L. Iobbi-Nivol C. Eaves D.J. Boxer D.H. Giordano G. Mol. Microbiol. 1996; 20: 875-884Crossref PubMed Scopus (145) Google Scholar, 25Rothery R.A. Grant J.L.S. Johnson J.L. Rajagopalan K.V. Weiner J.H. J. Bacteriol. 1995; 177: 2057-2063Crossref PubMed Google Scholar, 26Rothery R.A. Magalon A. Giordano G. Guigliarelli B. Blasco F. Weiner J.H. J. Biol. Chem. 1998; 273: 7462-7469Abstract Full Text Full Text PDF PubMed Scopus (56) Google Scholar). In all of the available crystal structures, the Moco is buried deeply inside the protein, suggesting that cofactor insertion is coupled to the final steps of protein folding. In the case of R. sphaeroides DMSOR, two forms of the apoenzyme have been characterized (27Masui H. Satoh M. Satoh T. J. Bacteriol. 1994; 176: 1624-1629Crossref PubMed Google Scholar). One appears to be partially unfolded based on its altered mobility on native gels and gel permeation chromatography columns, while the other form is indistinguishable from active DMSOR using these techniques. In the absence of molybdopterin cytosine dinucleotide, CO dehydrogenase from Hydrogenophaga pseudoflava will incorporate a variety of cytidine nucleotides (28Hänzelmann P. Meyer O. Eur. J. Biochem. 1998; 255: 755-765Crossref PubMed Scopus (34) Google Scholar), and DMSOR from R. sphaeroideswill incorporate either GMP or GDP in the absence of MGD (29Temple C.A. Rajagopalan K.V. J. Biol. Chem. 2000; 275: 40202-40210Abstract Full Text Full Text PDF PubMed Scopus (57) Google Scholar). Whereas some Moco-containing enzymes such as E. coli nitrate reductase require a specific chaperone for Moco insertion (24Palmer T. Santini C.-L. Iobbi-Nivol C. Eaves D.J. Boxer D.H. Giordano G. Mol. Microbiol. 1996; 20: 875-884Crossref PubMed Scopus (145) Google Scholar), others such as DMSOR only require MobA as single protein component forin vitro cofactor assembly and insertion. This latter aspect is clearly illustrated in the accompanying paper (29Temple C.A. Rajagopalan K.V. J. Biol. Chem. 2000; 275: 40202-40210Abstract Full Text Full Text PDF PubMed Scopus (57) Google Scholar) on an in vitro reconstitution system in which MobA only requires GTP, MgCl2, and the MPT form of the cofactor to efficiently generate active bis-MGD-containing R. sphaeroides DMSOR from apoprotein. These studies also indicate that high concentrations of MobA inhibit the reconstitution reaction, presumably due to binding of MGD to MobA. In order to gain a better understanding of MGD biosynthesis and the subsequent incorporation of this cofactor into molybdoenzymes, we have determined the high resolution crystal structure of E. coliMobA, with and without bound Mn-GTP. The active site of MobA is located in a deep depression on the molecular surface, and the co-crystal structure of MobA with GTP identifies the substrate-binding site. MGD biosynthesis presumably occurs in an elongated groove on the surface of the enzyme that is well suited for binding MGD. MobA was overexpressed and purified as described previously (29Temple C.A. Rajagopalan K.V. J. Biol. Chem. 2000; 275: 40202-40210Abstract Full Text Full Text PDF PubMed Scopus (57) Google Scholar). Crystals of MobA were obtained by vapor diffusion of an enzyme solution at a concentration of 12 mg/ml against a reservoir containing 2.5–3m sodium acetate in 0.1 m cacodylic acid, pH 6.5. The crystals reached their final size of 0.1 mm in each dimension within 10–20 days. The crystals belong to the tetragonal space group I422 with a = 124.1 Å and c = 67.0 Å and contain one molecule in the asymmetric unit. The structure of MobA was solved by single isomorphous replacement and anomalous scattering using a platinum derivative. The heavy atom derivative was prepared by soaking a crystal for 24 h in mother liquor containing 2 mm K2[PtCl4]. After transfer into a cryoprotectant containing 25% glycerol, the crystals were cryocooled, and diffraction data were collected to 1.65-Å (native) and 2.7-Å (platinum derivative) resolution at beam lines X26C and X12B, respectively, at the National Synchrotron Light Source at Brookhaven National Laboratory. Both beam lines were equipped with an ADSC Quantum 4 CCD detector, and diffraction data were indexed, integrated, and scaled with the HKL software (30Otwinowski Z. Minor W. Methods Enzymol. 1997; 276: 307-326Crossref Scopus (38440) Google Scholar). The platinum data set was supposed to be part of a multiwavelength anomalous diffraction experiment, but due to time constraints only two data sets, one at the inflection point and one at a remote wavelength, were collected. Since the derivative and native data were isomorphous, the remote wavelength data set collected at 1.033 Å was chosen as the heavy atom-derivative data set. One major and three minor platinum sites were detected by direct methods using SHELX (31Sheldrick G.M. Acta Crystallogr. A. 1990; 46: 467-473Crossref Scopus (19232) Google Scholar). Phase refinement was performed with SHARP (32DeLaFortelle E. Bricogne G. Methods Enzymol. 1997; 276: 472-494Crossref PubMed Scopus (1797) Google Scholar) at 2.7-Å resolution followed by solvent flattening and gradual phase extension to 1.7 Å with SOLOMON (33Abrahams J.P. Leslie A.G.W. Acta Crystallogr. D. 1996; 52: 30-42Crossref PubMed Scopus (1141) Google Scholar). The packing density in the crystals is fairly low, the Matthew's coefficient is 3 Å3/Da, corresponding to a solvent content of 59%, and this probably contributed to the fact that phases could be successfully extended by solvent flattening only. Subsequently, the resulting phases were used to autobuild the polypeptide chain using amplitudes to 1.65-Å resolution with ARP (34Perrakis A. Morris R. Lamzin V.S. Nat. Struct. Biol. 1999; 6: 458-463Crossref PubMed Scopus (2563) Google Scholar) after 5% of the data had been set aside to calculate the free R-factor. ARP was able to build 164 out of 194 residues in four chains. Most of the remaining residues and the side chains were fitted into the resulting electron density maps with the program O (35Jones T.A. Zou J.Y. Cowan S.W. Kjeldgaard M. Acta Crystallogr. A. 1991; 47: 110-119Crossref PubMed Scopus (13006) Google Scholar). The model was refined with REFMAC (36Murshudov G. Vagin A. Dodson E. Acta Crystallogr. D. 1997; 53: 240-255Crossref PubMed Scopus (13811) Google Scholar), and water molecules were added with ARP. All data between 20 Å and the respective high resolution limits were included. Additional calculations were done with the CCP4 suite (37Bailey S. Acta Crystallogr. D. 1994; 50: 760-763Crossref PubMed Scopus (41) Google Scholar). Initial attempts to soak nucleotides into MobA failed, and no density features reflecting a bound nucleotide could be detected. This was attributed to the fact that the high concentrations of sodium acetate necessary for crystallization interfered with nucleotide binding. Subsequently, the concentration of sodium acetate was reduced to 1 m, since the crystals were able to tolerate this decrease in salt concentration. Soaking experiments at a concentration of 20 mm GTP in the absence and presence of stoichiometric amounts of divalent cations (Mg2+ and Mn2+) were repeated under these low salt conditions. Diffraction data were collected from various soaked and cryocooled crystals at beam line X26C at the National Synchrotron Light Source. Data were processed with the HKL suite (30Otwinowski Z. Minor W. Methods Enzymol. 1997; 276: 307-326Crossref Scopus (38440) Google Scholar). Refinement against all of these data sets was performed using REFMAC, and the same reflections as in the native data set were used for calculating the free R-factor. The nucleotide was most clearly defined in crystals soaked with GTP and Mn2+, and the model described here is derived from a 1.75-Å data set collected at beam line X26C from a crystal that was soaked in 20 mm GTP and 20 mm MnCl2. Refinement was performed with REFMAC and ARP as described above for the native data set. The crystal structure of MobA was solved by single isomorphous replacement with anomalous scattering (TableI) using a tetragonal crystal form (space group I422). The structure has been refined at 1.65-Å resolution to a crystallographic R-factor of 0.206 (Rfree = 0.243) (TableII). The overall quality of the model is very good as judged by the low free R-factor, the low deviations from stereochemical ideality, and the appearance of the Ramachandran diagram. 89.2% of the residues were found in the most favored regions of the Ramachandran diagram as defined in PROCHECK (38Laskowski R.A. McArthur M.W. Moss D.S. Thornton J.M. J. Appl. Crystallogr. 1993; 26: 283-291Crossref Google Scholar). Ser72 and Leu73 located in a flexible loop are the only residues in disallowed regions of Phi-Psi space. The final model of the apoenzyme contains residues 2–16 and 23–192, a zinc ion with 0.5 occupancy, an acetate ion, and 177 water molecules. Residues 17–22 and 193–195 are disordered, and the N-terminal Met is not present, presumably due to posttranslational removal.Table IData collection and structure solutionParametersNativePlatinumMn-GTPResolution limits (Å)50–1.6550–2.750–1.75Wavelength (Å)1.251.0331.1Rsym0.086 (0.411)0.086 (0.327)0.084 (0.441)Completeness0.978 (1.0)0.949 (0.959)0.996 (1.0)〈I〉/〈sigI〉9.4 (1.2)11.5 (2.3)14.0 (1.9)Phasing power1.28/1.46RCullis0.796FOM0.336/0.267Rsym = ∑hkl∑i‖Ii − 〈I〉‖/∑hkl∑i Ii , where Ii is the ith measurement and 〈I〉 is the weighted mean of all measurements ofI. 〈I/sigI〉 indicates the average of the intensity divided by its S.D. Numbers in parentheses refer to the respective highest resolution data shell in each dataset. SIRAS phasing was performed to 2.7-Å resolution. Phasing power is the mean value of the heavy atom structure factor amplitude divided by the lack of closure for isomorphous/anomalous differences.RCullis is the lack of closure divided by the absolute of the difference between FPH andFP for isomorphous differences of centric data. FOM is the figure of merit given for acentric/centric data. Open table in a new tab Table IIRefinement statisticsParametersNativeMn-GTP complexResolution limits (Å)20–1.6520–1.75Number of reflections29,82024,531Number of protein/cofactor/solvent atoms1443/0/1821455/33/173Data/parameter ratio4.63.7R (Rfree)0.204 (0.243)0.183 (0.226)Deviations from ideal values Bond distances (Å)0.0190.022 Bond angles (°)1.962.24 Chiral volumes (Å3)0.1240.123 Planar groups (Å)0.0080.009 Torsion angles (°)5.8/18.75.4/20.1 B-factors of bonded atoms (Å2)2.1/3.4/4.7/7.62.3/3.7/5.3/8.1Ramachandran statistics0.892/0.096/0/0.0130.885/0.115/0/0Overall average B-factor (Å2)24.627.5Rcryst = Σhkl∥Fo‖ − ‖Fc∥/ Σhkl‖Fo‖, whereFo and Fc are the observed and calculated structure factor amplitudes. Rfree is defined as Rcryst for 5% of the data randomly omitted from refinement. Ramachandran statistics indicate the fraction of residues in the most favored, additionally allowed, generously allowed, and disallowed regions of the Ramachandran diagram as defined by PROCHECK. Open table in a new tab Rsym = ∑hkl∑i‖Ii − 〈I〉‖/∑hkl∑i Ii , where Ii is the ith measurement and 〈I〉 is the weighted mean of all measurements ofI. 〈I/sigI〉 indicates the average of the intensity divided by its S.D. Numbers in parentheses refer to the respective highest resolution data shell in each dataset. SIRAS phasing was performed to 2.7-Å resolution. Phasing power is the mean value of the heavy atom structure factor amplitude divided by the lack of closure for isomorphous/anomalous differences.RCullis is the lack of closure divided by the absolute of the difference between FPH andFP for isomorphous differences of centric data. FOM is the figure of merit given for acentric/centric data. Rcryst = Σhkl∥Fo‖ − ‖Fc∥/ Σhkl‖Fo‖, whereFo and Fc are the observed and calculated structure factor amplitudes. Rfree is defined as Rcryst for 5% of the data randomly omitted from refinement. Ramachandran statistics indicate the fraction of residues in the most favored, additionally allowed, generously allowed, and disallowed regions of the Ramachandran diagram as defined by PROCHECK. MobA has an α/β structure and is composed of a central β-sheet containing seven strands (β1, β4–β9). With the exception of β8, which is located near the center of the sheet, all strands are parallel. The β-sheet is surrounded by three α-helices on either side with the C-terminal helix α7 located above the N-terminal end of the sheet (Fig. 2 A). β-Strands 1, 4, 5, and 6 and the surrounding α-helices 1–3 form the N-terminal half of MobA (residues 2–101) and have a βαβαβαβ topology typically observed in proteins containing the Rossman fold, but beyond β-strand 6, the structure of MobA is unrelated to the classical Rossman fold. An additional short β-hairpin is inserted into the first helix, which starts with one turn in 310 conformation followed by the hairpin, leading back into the remainder of the helix. The MobA monomer has a conical shape with a groove running across the base of the cone and the N terminus at the tip of the cone. The groove is about 25 Å long and has a maximum depth of ∼10 Å. It is very narrow (∼7 Å) and deep (∼10 Å) at its N-terminal end, but is considerably wider (∼15 Å) and shallow in the center. At the C-terminal end, the groove narrows down again and has a width of ∼10 Å. A view into the groove reveals a surprisingly large number of Pro residues, including the conserved prolines at positions 79, 103, and 133. Pro79 and Pro103 are located in two of the four signature sequence motifs described below. Pro133 is at the i + 1 position of a β-turn (residues 132–135), which interrupts β-strand 8. The presence of this turn allows the N-terminal end of this strand (β8A) to contact the adjacent strand β7 in a perpendicular fashion from outside the central β-sheet of MobA. After completion of the β-turn, the remainder of this strand (β8B) is inserted in an antiparallel fashion into the β-sheet. In all data sets examined, a prominent density feature linked to the side chain of Cys100 was observed in the 2Fo −Fc and Fo −Fc electron density maps, suggesting that this residue has been covalently modified. The exact type of modification could not be established based on the crystallographic data, but the sulfur of Cys is still present, since its position is the highest density feature of this residue. Sequence alignments of MobA proteins from different organisms reveal four highly conserved regions (Fig. 1) as follows: (i) a glycine-rich loop (residues 12–25) bearing a distant resemblance to the P-loop typically found in a variety of nucleoside triphosphate-hydrolyzing proteins (in the crystal structure of MobA, these residues form a loop that is partially disordered and connects the first β-strand and α-helix); (ii) residues 78–82, including the highly conserved Pro79 at the N-terminal end of helix α3; (iii) residues 100–103 in the loop following β6 (the covalently modified Cys100, the strictly conserved Asp101, and Pro103 are located in this motif); (iv) residues 176–182, including two conserved Asn residues at positions 180 and 182 located near the C-terminal end of a long loop, which leads into the C-terminal helix α7. In the three-dimensional structure of MobA, all four stretches form exposed loop regions located in the large groove on the molecular surface (Fig. 2 B). This region presumably represents the active site of MobA. The program DALI (39Holm L. Sander C. Trends Biochem. Sci. 1995; 20: 478-480Abstract Full Text PDF PubMed Scopus (1278) Google Scholar) was used to search the Protein Data Bank for structural homologues of MobA. The highest match with a Z-score of 9.5 was the nucleotide-diphospho-sugar transferase SpsA from Bacillus subtilis (40Charnock S.J. Davies G.J. Biochemistry. 1999; 38: 6380-6385Crossref PubMed Scopus (309) Google Scholar). For comparison, two dissimilar proteins will have aZ-score below 2.0, and MobA matched against itself has aZ-score of 38.8. SpsA is involved in the production of the mature spore coat during sporulation, but its detailed substrate specificity is undefined. The structural match between MobA and SpsA is stunning, 150 Cα atoms of MobA can be superimposed with corresponding residues in SpsA with a root mean square deviation of 3.3 Å (Fig. 3). SpsA has been co-crystallized with either Mg-UDP or Mn-UDP, and the nucleotide binds near the N-terminal ends of β-strands 1–4. The binding site for the sugar is located in the C-terminal half of the molecule and might be mimicked by a glycerol molecule present in the crystallization solution. The structural similarity between SpsA and MobA is lower in this C-terminal region, reflecting different binding partners, either a sugar in SpsA or MPT in MobA. Other noteworthy matches of DALI include several dehydrogenases containing the Rossman fold, but these proteins are more distantly related to MobA as reflected by their lower Z-scores (Z < 4.2). Dynamic light scattering data of the purified protein 2M. W. Lake and H. Schindelin, unpublished results. at concentrations between 1 and 2 mg/ml indicate an equilibrium between a monomeric and oligomeric form of the enzyme. The molecular weight of the oligomeric form is most consistent with a decamer, but given the presence of an octamer in the crystal and its structure (see below) the oligomeric form observed by dynamic light scattering presumably corresponds to an octamer. In the tetragonal crystals, the MobA protein is present as an octamer (Fig. 4 A), which has D4 symmetry and is entirely generated by crystallographic symmetry operations in the tetragonal space group. The octamer has overall main chain dimensions of 75 × 75 × 65 Å and contains a central channel that runs along its 4-fold axis. The width of the channel is nonuniform, since it is restricted at the entrances to a diameter of ∼10 Å, whereas it is twice as wide near the center of the octamer. The restrictions on opposite ends of the octamer are separated by ∼25 Å, thus creating a roughly spherical central cavity. The restrictions at the entrances are imposed by residues 115–119. His115 is the second to last residue in helix α4, and residues 117–120 form a β-turn with Asp119 at the i + 2 position. The interior of the channel is lined with several hydrophilic side chains as well as polar main chain atoms. The N terminus is pointing into the channel near its center. Each monomer interacts with four other monomers, two along the 4-fold axis and two additional monomers along the 2-fold axes of symmetry. Although each of these contact areas only covers about 4% of the total solvent-accessible surface of the monomer, a total of 16.6% of the monomer surface is buried upon formation of the octamer. Unexpectedly, a zinc ion could be located in a monomer-monomer interface formed along one of the 2-fold symmetry axis (Fig. 4 B). The metal was identified as zinc by an anomalous Fourier calculation using amplitudes from the native data set collected at a wavelength of 1.25 Å, close to the K-absorption edge of zinc at 1.2834 Å, and the experimental phases after phase refinement and solvent flattening. The resulting map revealed a peak with a height of ∼32 times the root mean square deviation at the position of the metal. The metal is coordinated
Referência(s)