| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 279, Issue 28, 29718-29727, July 9, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
¶
From the
Section of Biology, University of Konstanz, P. O. Box 5560-M656 and the Department of Structural and Functional Biology, University of Insubria, via J. H. Dunant 3, 21100 Varese, Italy
Received for publication, February 4, 2004 , and in revised form, April 20, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-imino acids and, after hydrolysis, -keto acids, ammonia (or primary amines), and hydrogen peroxide. Both enzymes show a high pKa for flavin N-3H ionization, do not bind covalently the FAD cofactor, and react readily with sulfite (1-3), but they differ in substrate specificity. In addition to neutral D-amino acids (e.g. D-alanine, D-proline, etc., which are also good substrates of DAAO), GO catalyzes the oxidation of primary and secondary amines (e.g. glycine, sarcosine, etc.) partially sharing the substrate specificity with monomeric sarcosine oxidase (MSOX, EC 1.5.3.1
[EC]
), an enzyme that catalyzes the oxidative demethylation of sarcosine to yield glycine, formaldehyde, and hydrogen peroxide (4). According to investigations of the substrate specificity and of the binding properties, the GO active site seems to preferentially accommodate amines of a small size, such as glycine and sarcosine (1, 2). GO follows a ternary complex sequential mechanism with glycine, sarcosine, and D-proline as substrates in which the rate of product dissociation from the re-oxidized enzyme form represents the rate-limiting step (5). Such a kinetic mechanism is similar to that determined for mammalian DAAO on neutral D-amino acids and for the MSOX on L-proline (6, 7); the main difference is represented by the observed reversibility of the GO reductive half-reaction.
Taken together, however, these results do not really clarify the function of GO in Bacillus subtilis; they only outline a general catabolic role for GO (1, 2). Recent work proposed GO as the first enzyme in the biosynthesis of the thiazole moiety of thiamine pyrophosphate cofactor in B. subtilis (8). According to this hypothesis, GO catalyzes the oxidation of glycine to give the imine product that would be trapped with the thiocarboxylate intermediate bound to the following enzyme of the pathway (ThiS). In such a mechanism, the nucleophilic addition might occur at the active site of GO to avoid hydrolysis of the imino product. The known 2.3-Å resolution structure of GO (8) does not provide direct evidence of such a reactivity. However, this anabolic function of GO is particularly intriguing because amino acid oxidases are usually involved in the catabolic utilization of their substrates. From this point of view, GO resembles L-aspartate oxidase that converts L-aspartate to iminoaspartate using molecular oxygen or fumarate as electron acceptors, the first reaction in the NAD+ biosynthesis pathway in bacteria (9). Furthermore, GO is also the object of particular attention because it can be used in an in vitro assay, in parallel to DAAO, to detect and modulate the level of glycine or D-serine in the proximity of N-methyl-D-aspartic acid receptors in human brain.
Here we report the crystal structure of B. subtilis GO in complex with the inhibitor glycolate at 1.8 Å resolution. Although the inhibitor was found in an unexpected orientation, active site residues that are likely to bind the substrate or to assist in its oxidation have been tentatively identified on the basis of similarities with other related flavoprotein amine oxidoreductases. In fact, the structures of DAAO, MSOX, and L-amino acid oxidase have also been resolved (10-13); it can thus be expected that comparison of their active sites as well as the mode of interaction with the substrate/ligand would provide insights into the similarities and differences in the structure-function relationships of flavoenzymes active on similar compounds. In addition, and with the aim of clarifying the role of GO in B. subtilis, the effect of different carbon and nitrogen sources on cell growth and on the level of GO expression has been investigated.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
0.08 in 500 ml of minimal medium (2-liter flasks) and grown for 16 h as reported above. The cells were collected by centrifugation, suspended in 2 ml of 1x minimal salt solution, and finally used to inoculate 600 ml of minimal medium supplemented with the appropriate nutrients (initial A600 0.08). Alternatively, a classic M9 medium (containing 1x minimal salt solution, 22 mM glucose, 2 mM MgSO4, and 0.1 mM CaCl2) supplemented with the appropriate nutrient was also used. The cells were grown in flasks as reported above and collected at different growth phases by centrifugation (4000 rpm for 10 min at room temperature) from 100 ml of fermentation broth. Cell growth was followed by absorbance measurements at 600 nm. The crude extracts were prepared according to Ref. 14. In detail, 1 g of cell paste was added to 5 ml of a 2 mg/ml lysozyme solution in TE buffer, pH 8.0, and incubated for 30 min at 37 °C; this sample was then centrifuged at 5000 rpm for 20-30 min at room temperature and 10 ml of lysis buffer added to the pellet (50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 100 mM NaCl, 1.1 mM phenylmethylsulfonyl fluoride, 5 µg/ml DNase I). After 15-30 min of incubation, the crude extract was recovered by centrifugation at 14,500 rpm for 15 min at 4 °C.
Protein Analyses, Enzyme Assay, and Gel-permeation Chromatography—B. subtilis crude extracts were employed for the following assays: (a) determination of protein concentration (Biuret method); (b) determination of GO activity (polarographic assay, see below); and (c) determination of the total GO concentration (by Western blot analysis). A fixed amount of protein (
300 µg) from the crude extract (the soluble fraction obtained after cell disruption and centrifugation) was separated by SDS-PAGE and electroblotted to a nitrocellulose membrane. The same analysis was also performed on whole cell samples by separation of the proteins corresponding to 240 µl of fermentation broth. GO was detected by immunostaining using monospecific rabbit anti-GO antibodies and visualized using anti-rabbit IgG alkaline phosphatase conjugated with 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium chloride, as reported previously (2), or anti-rabbit IgG horseradish peroxidase conjugated with a chemiluminescent substrate (SuperSignal West Pico, Pierce). The amount of anti-GO immunoreactive protein was determined by densitometric analysis using a 50-1000-ng titration curve obtained using pure GO.
Glycine oxidase activity was assayed in a thermostated Hansatech oxygen electrode measuring the oxygen consumption at pH 8.5 and 25 °C with 10 mM sarcosine as substrate (1, 2). One unit of GO is defined as the amount of enzyme that converts 1 µmol of substrate (sarcosine or oxygen) per min at 25 °C.
The oligomerization state of GO was investigated by means of gel-permeation chromatography on a SuperdexTM 200 HR 10/30 column (Amersham Biosciences) using 50 mM sodium pyrophosphate, pH 8.5, 5% glycerol, and 250 mM NaCl as elution buffer (1). The pH effect on GO oligomerization state was determined by chromatographic separation using 50 mM potassium phosphate (pH 6.5 and 7.5), 50 mM sodium pyrophosphate (pH 8.5), or 25 mM sodium pyrophosphate and 25 mM sodium carbonate (pH 9.5), all containing 250 mM sodium chloride and 5% (v/v) glycerol.
Limited Proteolysis Experiments—GO (1 mg/ml) was incubated at 25 °C in 50 mM sodium pyrophosphate, pH 8.5, and 1% glycerol with 10% (w/w) trypsin, chymotrypsin, or SV8 protease. For electrophoretic analysis, 1.5 mM phenylmethylsulfonyl fluoride was added to protein samples (10 µg of GO) taken at different times after the addition of protease and immediately frozen for analysis by native PAGE on a 7.5% (w/v) polyacrylamide gel or diluted in the sample buffer for SDS-PAGE, boiled for 3 min, and then loaded on a 12% (w/v) polyacrylamide gel. Gels were stained with Coomassie Blue R-250 and, only for gels from native PAGE, stained for GO activity as reported previously (1). The oligomerization state of proteolyzed GO samples was determined by gel-permeation chromatography (see above), and their N-terminal sequence was determined by means of automated Edman degradation using a Procise protein sequencer (Applied Biosystems).
Preparation of the Protein and Crystallization—Wild-type GO was expressed in Escherichia coli using the pT7-HisGO expression system in BL21(DE3)pLysS E. coli cells (2). The pT7-HisGO encodes a fully active fusion protein with 13 additional residues at the N terminus of GO (MHHHHHHMARIRA). The purified protein was concentrated up to 15 mg/ml and equilibrated in 50 mM disodium pyrophosphate buffer, pH 8.5, 10% glycerol by gel-permeation chromatography on a Sephadex G-25 (PD10) column.
The recombinant form of GO was crystallized following dynamic light scattering analysis (DynaPro, Protein Solutions Ltd.) by the hanging drop vapor diffusion method, mixing 1 µl of reservoir and 1 µl of protein solution (13 mg/ml) at 18 °C. Crystals with two different space groups were obtained. GO crystals of the space group P6122 were obtained by using a reservoir solution containing 1 M sodium citrate, pH 6.2; the hexagonal crystals grew in 2-3 weeks. Crystals of the space group C2221 were obtained using a GO solution containing 30 mM sodium glycolate and a reservoir solution containing 100 mM imidazole (pH 8.2), 200 mM calcium acetate, and 10% (w/v) PEG 1000; the orthorhombic crystals grew in 1-3 days. Prior to flash-freezing using liquid nitrogen, the crystals were soaked with the corresponding reservoir buffer containing 13% (v/v) ethylene glycol. Heavy atom derivatives were prepared by addition in the reservoir buffer of KAu(CN)2 or K2Pt(CN)4 to the hexagonal crystal form of GO (1 mM final concentration) and incubating them for 12-16 h. For cryoprotection, the cryoprotectant buffer with an added 1 mM of the corresponding heavy atom salt was used.
Solution of the GO Crystal Structure—All data sets were collected under cryogenic conditions at 100 K. For collecting data of the heavy atom derivatives the MAR345 image plate system with a rotating anode x-ray source (Schneider, Offenburg, Germany) was used. A native data set from the hexagonal crystal was measured at the DESY (EMBL, Hamburg) using the MAR345 image plate system. The 1.8-Å GO-glycolate complex was measured from the orthorhombic crystal at the PSI/SLS (Villigen, Switzerland) using a Mar CCD detector 165 mm in diameter. Space group determination and data reduction were carried out with XDS (15). The programs SOLVE/RESOLVE (16, 17) were used to solve phases by multiple isomorphous replacement. With MOLREP (18), which is part of the CCP4 package (19), and monomeric sarcosine oxidase as a model (Protein Data Bank code 1el5 [PDB] ), the position of one of the two molecules per asymmetric unit was found in the map calculated by RESOLVE. The second molecule was positioned using O (20). Alternating refinement in CNS (21) and model building in O was carried out until an Rfree of 30% was reached. This model was used with MOLREP to find the four molecules in the asymmetric unit of the orthorhombic crystal. The GO-glycolate complex was refined with REFMAC5 (22), and the secondary structure was analyzed with DSSP (definition of secondary structure of proteins given a set of three-dimensional coordinates) (23). For detailed data collection statistics see Table I and for refinement statistics see Table II. Structure plots were produced with the programs MOLSCRIPT (24), RASTER3D (25), and DINO (26).
|
|
Accession Numbers—The coordinates and structure factors of glycine oxidase in complex with the inhibitor glycolate have been deposited in the Protein Data Bank under accession code 1RYI.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
-helices) and 18 
-strands (Fig. 2). The position of the residues that initiate and terminate the secondary structure elements is frequently different in our structure compared with the previously shown topology (8). The main difference is represented by the three newly identified 3/10 helices (depicted in yellow in Fig. 2); the overall topology is not changed. GO is a two-domain protein, which consists of a FAD-binding domain and a substrate-binding domain. The main structural elements are central antiparallel -sheets, as first observed in the flavoprotein p-hydroxybenzoate hydroxylase (29). The classic FAD-binding domain is common to the glutathione reductase (GR) class of flavoproteins (30). In GO, this motif consists of a six-stranded -sheet composed of five parallel -strands (strands 6, 2, 1, 10, and 18) and one additional antiparallel strand (strand 17) and flanked on one side by three -helices (helices 1, 7, and 10) and on the other side by a three-stranded antiparallel -sheet (strands 7-9) and a small -helix (helix 8). The polypeptide chain crosses between the two domains four times (after helix 4, strand 5, helix 8, and strand 16). The most significant differences between the GO structure and both MSOX and RgDAAO are represented by -helix 8, which is missing in the other two enzymes, a different -helix 3 and 4 topology, which is fused to a single continuous helix in RgDAAO and MSOX, and by the three stranded -sheet (strands 7-9 in GO) of the flavin-binding domain, which is conserved in all GR family members and is not conserved in RgDAAO (10, 11) (see Fig. 1). Another main topological difference with RgDAAO is the absence of the loop consisting of 21 amino acids connecting F5 and F6 in RgDAAO (Fig. 1B), which is involved in monomer-monomer interaction and is not present in other known DAAO sequences (11). Concerning the substrate-binding domain, an element corresponding to -helix 6 of GO is not conserved in RgDAAO. Analogously to the yeast DAAO, GO shares an active site loop (connecting strands 13 and 14) that is shorter by 5-8 residues as compared with pkDAAO and MSOX.
|
|
7 and 8 of the -meander (Fig. 1), the electron density for four amino acids (Arg180-Ala183) is weak, indicating that part of the loop is very flexible. The region Ala55-Asp60 after helix 2 also shows weak electron density. In MSOX this region corresponds to a flexible loop (Tyr55-Tyr61) that changes from a disordered to a weak electron density following the binding of an active site ligand, thus shielding the positive surface potential at the FAD site (12). This loop, and in particular the side chain of Arg52, was thus suggested to act in MSOX as a switch for active site accessibility (see below).
Homology of GO with Other Amine Oxidoreductases—GO exhibits the highest sequence conservation with the -subunit of heterotetrameric sarcosine oxidase, sarcosine dehydrogenase, and dimethylglycine dehydrogenase (24-27% identity), less similarity with the sequences of MSOX and pipecolate oxidase (21% identity), and a modest similarity with DAAO and D-aspartate oxidase (18.4% identity) (2). Furthermore, the primary sequence of GO shows a high degree of conservation with the product of thiO gene of Rhizobium etli (23.0% of sequence identity). thiO is the second open reading frame of four genes (thiC, thiO, thiG, and thiE) located on plasmid pb, which are involved in the synthesis of thiamine in R. etli (31). R. etli ThiO protein (a 327-amino acid protein) contains at its N terminus a flavin adenine dinucleotide-binding motif and shares many of the residues involved in the catalytic site of DAAO; it has been also suggested that ThiO may have amino acid oxidase activity (31). In E. coli five genes (thiC, thiE, thiF, thiG, and thiH), proposed to be a single transcription unit, are involved in thiamine biosynthesis (32). The ThiO protein of E. coli shows a limited sequence identity (12.5%) with B. subtilis GO.
By structural overlay, the GO structure was compared with that of other flavoprotein oxidases (see Table III). Based on structural and sequence homologies, GO can be classified as a member of the large GR family (all the family members adopt the Rossmann fold) (33, 34) and further into the subgroup GR2, which was reported to show sequence similarity mainly within 30 residues in the N-terminal region (34). Our superposition procedure identified large, structurally homologous parts of GO with DAAO and MSOX (Table III). Although comparison of GO with dimethylglycine oxidase shows that 284 of 364 residues lie within the distance cut off of 3.5 Å, the r.m.s. deviation of those residues is high (1.84 Å), reflecting notable structural differences between GO and dimethylglycine oxidase. The enzymes pkDAAO, RgDAAO, and MSOX can be superimposed with a smaller r.m.s deviation (1.53-1.58 Å) with 222-256 of the 364 residues of the GO structure; this reflects the structural similarity of GO with both DAAOs and MSOX.
|
FAD Binding—Each GO monomer contains one noncovalently bound FAD molecule (1, 2). The FAD-binding patterns of GO, DAAO, and MSOX share an overall similarity; in all these enzymes the FAD-binding domain contains the conserved Rossmann fold motif (1, 1, and 2) (33), which serves as a dinucleotide-binding motif (35). The central part of this consensus motif is the sequence GXGXXG (Gly11, Gly13, and Gly16 of helix 1) with the N-terminal end of helix 1 pointing toward the FAD pyrophosphate moiety, as observed for other dinucleotide-binding proteins (35). The binding of FAD in GO is that typical of the GR2 subfamily (34); the prosthetic group adopts an extended conformation with the isoalloxazine ring located at the interface between the FAD-binding domain and the substrate-binding domain, facing with its re-side the inner part of the substrate-binding cavity. The cavity is located distant from the monomer-monomer interaction sites facing toward the bulk solvent. The whole cofactor is buried inside the protein (Fig. 1), and the isoalloxazine ring is not directly solvent-accessible. A similar situation was also observed in DAAO (10, 11), whereas in p-hydroxybenzoate hydroxylase the flavin benzene ring is exposed to the bulk solvent, allowing the flavin to adopt two different conformations (36). The large majority of the potential FAD hydrogen bonds are formed with the protein residues, thus resulting in a tight net as shown in Fig. 3, a Kd value for the apoprotein-FAD complex of 5 ± 2 x 10-8 M has been calculated.2 The isoalloxazine ring is held in place by a hydrogen bond between its N-3H-C-4=O and the backbone of Gly48 and Met49, whereas N-5 is within hydrogen bond distance to one of the oxygens of the inhibitor glycolate (this flavin position interacts with the backbone NH group of Gly52 and Ala49 in RgDAAO and pkDAAO, respectively, and with a water molecule in MSOX). The flavin N-1 is within hydrogen bond distance to the Arg329 backbone carbonyl oxygen in GO (such an interaction is absent in pkDAAO and corresponds to RgDAAO Ser335 and MSOX Lys348). The environment of the O-2 position is slightly different between GO and DAAOs. In GO the O-2 forms two H-bonds with the backbone NH group of Ile332 and Leu333 (Fig. 3). In pkDAAO O-2 interacts with a threonine and with the partial positive charge of a dipole from helix F5 (10), and in MSOX the flavin O-2 forms a hydrogen bond to the side chain nitrogen of Lys348 (12).
|
These interactions serve to stabilize the electrophilic character of the flavin ring, and the negative charge of the anionic form of the semiquinone (1) and, probably, of the fully reduced flavin of GO. The benzene ring of the isoalloxazine moiety makes van der Waals contacts with a pocket formed by the antiparallel -strands 11 and 15; the contacting residues are Ala45, Ala46, Ala47, Gly224, Cys226, Tyr246, Ala259, Gly300, and Arg302. In particular, Arg302 is 4 Å from the C-7 and C-8 position of the flavin. Most interesting, GO Gly300 occupies the same position as Cys315 in MSOX, the site of covalent flavinylation. The covalent attachment of the flavin cofactor to the apoprotein moiety requires the contribution of a base to act as a proton acceptor from the 8-CH3 group of FAD during tautomerization and possibly to generate the reaction thiolate from Cys315 (12). The proposed candidate is His45 in MSOX (in GO an alanine is instead present). Analogously to DAAO and MSOX, there are no acidic residues in the region surrounding the isoalloxazine ring.
The FAD diphosphate group forms one H-bond with Thr43 in GO. Five H2O molecules are found at optimal hydrogen bonding distance with four of the phosphate oxygen atoms (Fig. 3). The side chain of the highly conserved GO Asp34 of the FAD-binding domain interacts via three strong hydrogen bonds with the two OH groups of the AMP ribosyl moiety. The adenine moiety forms hydrogen bonds to the backbone of Val174 and a water molecule (Fig. 3). Although different amino acids interact with AMP in GO, MSOX, and DAAO, the overall picture is similar.
Mode of Oligomerization—Native GO from B. subtilis is a stable 169-kDa homotetramer whose oligomerization state is not dependent on protein concentration in the 0.01-13 mg/ml concentration range (2). The molecular mass of GO in solution estimated from dynamic light scattering and blue native PAGE (37) is 160 kDa and corresponds to the theoretical value of 173 kDa for the recombinant GO tetramer. Crystals of GO have the space group P6122 (hexagonal) and C2221 (orthorhombic). In both space groups GO crystallized as a tetramer. The tetramers of both space groups are identical and have a 222 point symmetry. The tetramer of the hexagonal crystal is also identical to the one described previously (8). In the orthorhombic crystal, there are four molecules (monomers) in the asymmetric unit, which build two identical tetramers; the first tetramer consists of the monomers A and B and their symmetry mates A' and B' after applying the symmetry operator (x, -y + 2, -z + 2) on chain A and B; the second tetramer consists of chains C and D and their symmetry mates C' and D' after applying the symmetry operator (-x, y, -z + 3/2). The difference between the two tetramers is the orientation of the crystallographic axis with respect to the tetramer (see Fig. 4).
|
9 and 16 of each subunit and buries a total surface of about 1770 Å2. There are eight hydrogen bonds between the two monomers, formed by the following residues: Pro270-Val204, Leu272-Met292, Gly273-Gln290, and Lys283-Glu276. Apart from the last pair of residues, the hydrogen bonds are between main chain N and C atoms. Moreover, the Leu275 from 9 forms a hydrophobic pocket with the 16 residues Val294 and Phe297. Most interesting, the side chain of Phe297 adopts two conformations. Interaction between monomers A and A' (and C and D') is due to residues belonging to loop 11-12, loop 4-3, and loop 13-14 and buries a total surface area of about 1870 Å2; no hydrogen bonds are present, but the loops fit together very tightly. The size of this contact area is about 250 Å2 larger than that reported previously (8). The third interaction between monomers A and B (and C and C') is due to residues from loops 15-9, 2-2, 6-7, and Met208 from 8, with a total buried surface area of only about 975 Å2. The accessibility from outside to the funnel leading to the active site of each monomer is not restricted by the quaternary structure of GO; the openings face the bulk solvent and are far away from each other. Moreover, the GO quaternary structure shows that interaction between the flavin cofactor of different subunits is not possible. Superposition of the four chains with LSQMAN shows that the r.m.s. deviation between the monomers (A-D) is in the 0.22-0.44 Å range. The r.m.s. deviation between the two tetramers (ABA'B' and CDC'D') is 0.35 Å. This result demonstrates that the two crystallographic tetramers are identical and that the r.m.s. deviation between tetramers is from the observed difference between the intrinsic structure of the monomers.
In order to modify the stable tetrameric structure of GO, different approaches have been used. The tetrameric state of GO is not affected by pH in the pH range 6.5-9.5, as determined by means of gel-permeation chromatography. Treatment with increasing concentrations of thiocyanate (up to 1 M) results in a decrease of the peak corresponding to the tetrameric GO, but a corresponding increase in a peak at 46 kDa corresponding to the monomer was not observed. This result indicates that the lipophilic ion affects the stability of GO in solution but does not alter its monomer-monomer interaction. Tetrameric GO is also strongly resistant to proteolysis; treatment of GO (1 mg of protein/ml) with 10% (w/w) trypsin, chymotrypsin, or SV8 protease (at 25 °C and pH 8.5) for up to 4 h does not change its elution volume in gel-permeation chromatography. In contrast to GO, the dimeric oligomerization state of RgDAAO is affected by proteolysis (38). Only the N terminus of GO is sensitive to proteases. Western blot analysis and protein sequencing demonstrated that trypsin and chymotrypsin treatment, respectively, converted the 382 amino acids of recombinant GO monomer into a protein form that was shorter by 8-10 and 12-14 residues (lacking the His tag at the N terminus).
In contrast to this, the apoprotein form of GO is monomeric and rapidly converts to tetrameric holoenzyme upon addition of FAD.2 Considering that the FAD cofactor is involved in many protein core contacts (see above) and that the conversion to the apoprotein form abolishes the CD signal in the near-UV region due to the tertiary structure of the holoenzyme, it was deduced that the organization of the quaternary structure follows holoenzyme reconstitution.
The Active Site—The active site of GO is a cavity delimited by the two long -strands, 11 and 16, bent around the isoalloxazine ring of the flavin and the two short -strands, 13 and 14, located close to the substrate binding site; the flavin forms the bottom of the cavity (Fig. 5). The electron density is clearly seen for all atoms of the ligand glycolate in all four monomers of the asymmetric unit. The interactions can be described as follows: (a) Arg302 and Tyr246 form hydrogen bonds with one oxygen of the -COO- of the inhibitor glycolate (2.91 and 2.67 Å); (b) the second oxygen of the inhibitor -COO- group is at hydrogen bond distance to the N-5 of the FAD; (c) one terminal nitrogen of Arg329 and the main chain carbonyl of His244 form a hydrogen bond (3.02 Å) and are in close proximity to the -COO- of the bound ligand, probably preventing its rotation. Comparison of the structure of glycolate bound to GO with those of D-alanine in DAAO (11), dimethylglycine in MSOX (12), and acetylglycine in GO (8) shows that the -COO- of glycolate does not make two hydrogen bonds to the guanidyl group of GO Arg302 but instead that the ligand is unexpectedly rotated by 120° and binds in a different orientation, which is observed in all four molecules of the ASU. This binding mode may be favored by the small size of glycolate, and probably could resemble the binding mode of glycine, which has approximately the same size as the inhibitor we used.
|
-COO- oxygens of glycolate, is found at a position resembling that of RgDAAO Tyr223 (a residue involved, mainly by steric effects, in substrate binding) and of MSOX Tyr254 (whose role in substrate binding is considered marginal). GO Arg302 corresponds to RgDAAO Arg285 (and pkDAAO Arg283) and to MSOX Arg52; in all these enzymes, it is the residue that interacts with the -COO- of the substrate. In MSOX the movement of Arg52 following inhibitor binding induces a large replacement of a loop region that directs MSOX Glu57 into the active site (to bind MSOX Arg52) (12). Concerning the modulation of the substrate specificity, GO Tyr241 is located at a position resembling that of MSOX Met245 and RgDAAO Met213 (Fig. 5). This latter residue was recently demonstrated to modulate the substrate specificity of yeast DAAO (39); the introduction of a positive charge at this position by site-directed mutagenesis provided a DAAO variant active on both neutral and acidic D-amino acids. Most interesting, the loop found in pkDAAO and which was proposed by Mattevi et al. (10) to act as lid-controlling access to the active site, is absent in GO (as well as in RgDAAO and MSOX). The pkDAAO loop contains an important residue, Tyr224, which is probably involved in substrate/product fixation (10). The RgDAAO Tyr238 side chain, which was also proposed to change its position in order to allow substrate/product exchange, is placed at a position similar to that in pkDAAO Tyr223, GO Arg329, and MSOX His269 (Fig. 5). The latter residue was primarily considered the active site base, although now a different role has been proposed (40).
No lid is evident in GO, but some side chain rearrangements occur upon inhibitor binding, in particular in Arg329 and Met261. We proposed that His244 should be involved in a system to drive the protons outside the active site; a proton should be taken up from the substrate by the latter residue and transferred to Arg329 and to Met261 that is outside the active site. In GO, the following residues form the entrance to the active site: Ala259, Thr260, Met261, Arg329, His244, Asp243, and Cys245 (Fig. 6). Site-directed mutagenesis of Met261 (to Tyr and His) supports such a conclusion. The mutants have properties quite similar to the wild-type (similar Vmax); the main change is a 10-fold increase in Km for the substrates and in Kd for the inhibitors and a change of the flavin redox properties in the free enzyme form.2 There is a striking difference between our structure and the conclusions derived by the unliganded structure (8) regarding the seal of the active site. Comparison of free GO with GO in complex with glyoxylate (Fig. 6) shows that the movements of residues Met261 and Arg329 do not fully protect the bound ligand from external solvent and thus the imine product from hydrolysis.
|
|
|
Effect of Phosphate and Thiamine Pyrophosphate on GO Activity—A previously detailed kinetic characterization of GO showed that the enzyme follows a classic Michaelis-Menten dependence of the reaction rate as a function of substrate concentration, without any indication of sigmoidal behavior (5). In the structure of GO reported previously (8), a phosphate ion was modeled into a buried, positively charged pocket formed by interaction of two monomers in proximity of the residues Arg89, Arg254, and Arg296. In the orthorhombic crystal, all three corresponding sites, which lie between the monomer pairs AA', BB', and CD, were investigated; neither the sites that are located on a crystallographic axis (AA' and BB') nor those on a noncrystallographic axis (CD) show any indication of bound phosphate ions at these sites. In agreement with our crystallographic data, the addition of phosphate (in the 1-130 mM concentration range) does not affect the activity of GO.
Because of the proposed role in B. subtilis of GO in the biosynthesis of the thiazole moiety of thiamine (8) and of its tetrameric oligomerization state, the effect of the proposed final product (thiamine and thiamine pyrophosphate) of this synthetic pathway on the enzymatic activity of GO was also assessed. As reported previously for phosphate, even the addition of 0.1, 1, and 50 mM thiamine and thiamine pyrophosphate does not affect the activity of GO.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
A further obscure point that hardly correlates with the proposed role in the anabolic production of thiamine is represented by the (wide) substrate specificity of GO. Is it only an evolutionary memory? GO is active on many of the compounds specifically oxidized by DAAO, MSOX, and monoamine oxidase (1, 2). In addition, the Vmax is similar to that of many of the substrates tested, because it is essentially due to the rate of product release from the reoxidized enzyme (5). The rate of flavin reduction by the substrate is similar among glycine and sarcosine (the best substrates of GO), and indeed the Km value for sarcosine is slightly lower than for glycine (5).
An additional intriguing point pertains to the oligomeric state of GO. Why is GO tetrameric, whereas all the other members of the amine flavooxidase family are monomeric or dimeric (no other members of the GR2 subfamily are tetrameric)? We demonstrated that GO does not show any allosteric behavior; the enzyme possesses classic Michaelis-Menten kinetics as a function of substrate concentration (5), and no effect on the enzymatic activity by phosphate, thiamine, and thiamine pyrophosphate is observed.
In order to clarify the role of GO in B. subtilis, the effect of different nitrogen and carbon sources on the growth and the induction of GO activity were also investigated. A wild-type strain of B. subtilis does not require thiazole to grow, thus confirming that it is able to synthesize thiamine. We also demonstrated that glycine and sarcosine, two putative in vivo GO substrates, do not support B. subtilis growth when used as the sole carbon or nitrogen source, whereas D-alanine can be used as the main carbon source (Table IV). This latter compound can be used only if it is catabolized by GO, because no other enzymatic activities have been reported in B. subtilis to deaminate D-amino acids. Otherwise, D-alanine could be converted to the corresponding L-isomer by an amino acid racemase (various racemases have been identified in B. subtilis genome, in particular a D-alanine racemase) and thus L-alanine should be used as a carbon source. This latter possibility is in good agreement with our experimental evidence showing that the amount of GO (activity and protein) is not increased in the absence of thiazole and using D-alanine as the sole carbon source, i.e. GO is not inducible. The evolution of the regulation of amino acid degradative enzymes in B. subtilis resulted in enzymes present at high levels in sporulating cells and spores, rather than preferentially expressed during nitrogen-limited growth (for a review see Ref. 41). Catabolite-repressed genes in B. subtilis are controlled by more than one global regulatory mechanism. Although none of the Bacillus catabolite repression mechanisms is understood at the molecular level, it is clear that they are not analogous to the cAMP-catabolite-responsive element-dependent mechanism that is operative in E. coli (for a review on catabolite repression and inducer control in Gram-positive bacteria, see Refs. 42 and 43). We performed a search for a 14-bp palindromic sequence element corresponding to the catabolite-responsive element (consensus sequence, TG(T/A)NANCGNTN(A/T)CA), which mediates catabolite repression in a number of B. subtilis genes (44). Catabolite-responsive elements show a striking variability in their sequence and position with respect to the transcriptional start sites of regulated genes. Two sequences similar to the consensus sequence for catabolite-responsive elements are found within the reading frame of the GO-coding gene (close to the 5' start site); with respect to the observed deviations of such elements from the canonical sequence (at least at three positions), GO cannot be considered as a protein active in carbon catabolite pathways.
We also showed that glycine and sarcosine cannot be used as a nitrogen source, whereas D-alanine supports B. subtilis growth even in the absence of L-glutamine (Table IV). Despite the different ecological habitats and life cycles of B. subtilis and enteric bacteria, such as E. coli, only minor differences in the physiology of ammonium assimilation have been reported between these bacteria (45). B. subtilis has no assimilatory glutamate dehydrogenase activity; therefore, ammonium assimilation occurs solely by the glutamate synthase pathway (its synthesis is raised during nitrogen-limited growth). Glutamine is the best nitrogen source for B. subtilis, followed by arginine, and even ammonium is a good nitrogen source. Expression of the arginine, proline, and histidine degradative enzymes has been reported to be substrate-inducible in both B. subtilis and enteric bacteria (46). However, and in contrast to the case for E. coli, their expression is not subjected to nitrogen regulation in B. subtilis, raising the possibility that this bacterium does not contain any global nitrogen-regulatory system. Our results show that the expression of GO is not modified by the different nitrogen sources used and in particular not by glycine, sarcosine, and D-alanine even if this latter compound can be used as the main nitrogen source and allows B. subtilis growth. Taken together, our microbiology experiments exclude regulation of GO synthesis according to the composition of the medium in B. subtilis and therefore do not support a main role of GO in the catabolic utilization of primary and secondary amines.
In conclusion, the structural and functional properties determined on GO demonstrate that it belongs to the amino acid oxidase class of flavoproteins (in particular to the GR2 subfamily) and that it is characterized by a broad substrate specificity, low kinetic efficiency, and a unique and stable tetrameric oligomerization state (depending on FAD binding). The combination of these investigations does not clarify how GO is involved in thiamine biosynthesis (mechanistic limitations of this role have been also reported in Ref. 8), nor do the microbiology experiments support a main catabolic role for this flavooxidase in B. subtilis.
|
FOOTNOTES |
|---|
* This work was supported by Italian MURST Grant Prot 2002057751 (to M. S. P.), by Fondo di Eccellenza 2001, University of Insubria (to M. S. P.), by FAR 2001 (to L. P.), and by FAR 2002 and 2003 (to M. S. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 
¶ To whom correspondence should be addressed. Tel.: 390-332-421506; Fax: 390-332-421500; E-mail: loredano.pollegioni{at}uninsubria.it.
1 The abbreviations used are: GO, glycine oxidase; DAAO, D-amino acid oxidase; RgDAAO, Rhodotorula gracilis D-amino acid oxidase; pk-DAAO, pig kidney D-amino acid oxidase; MSOX, monomeric sarcosine oxidase; GR, glutathione reductase; r.m.s., root mean square; ASU, asymmetric unit.
2 L. Pollegioni, personal communication.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
) and in the presence (
) of 2 µM thiazole alcohol.