Crystal structure of human glyoxalase II and its complex with a glutathione thiolester substrate analogue

Research Article

1067

Crystal structure of human glyoxalase II and its complex with a glutathione thiolester substrate analogue Alexander D Cameron1*†, Marianne Ridderström2, Birgit Olin2 and Bengt Mannervik2 Addresses: 1Department of Molecular Biology, Uppsala University, Biomedical Center, Box 590, S-751 24, Uppsala, Sweden and 2Department of Biochemistry, Uppsala University, Biomedical Center, Box 576, S-751 23, Uppsala, Sweden.

Background: Glyoxalase II, the second of two enzymes in the glyoxalase system, is a thiolesterase that catalyses the hydrolysis of S-D-lactoylglutathione to form glutathione and D-lactic acid. Results: The structure of human glyoxalase II was solved initially by single isomorphous replacement with anomalous scattering and refined at a resolution of 1.9 Å. The enzyme consists of two domains. The first domain folds into a four-layered β sandwich, similar to that seen in the metallo-β-lactamases. The second domain is predominantly α-helical. The active site contains a binuclear zinc-binding site and a substrate-binding site extending over the domain interface. The model contains acetate and cacodylate in the active site. A second complex was derived from crystals soaked in a solution containing the slow substrate, S-(N-hydroxy-N-bromophenylcarbamoyl)glutathione. This complex was refined at a resolution of 1.45 Å. It contains the added ligand in one molecule of the asymmetric unit and glutathione in the other.

Present address: †Structural Biology Laboratory, Department of Chemistry, University of York, Heslington, York, UK, YO10 5DD. *Corresponding author. [email protected] Key words: binuclear zinc site, crystal structure, glyoxalase II, glutathione, thiolesterase Received: 24 March 1999 Revisions requested: 27 April 1999 Revisions received: 17 May 1999 Accepted: 25 May 1999

Conclusions: The arrangement of ligands around the zinc ions includes a water molecule, presumably in the form of a hydroxide ion, coordinated to both metal ions. This hydroxide ion is situated 2.9 Å from the carbonyl carbon of the substrate in such a position that it could act as the nucleophile during catalysis. The reaction mechanism may also have implications for the action of metallo-β-lactamases.

Introduction The glyoxalase system catalyses the conversion of toxic 2-oxoaldehydes to the corresponding 2-hydroxycarboxylic acids using glutathione (GSH) as a coenzyme. The system consists of two distinct enzymes, glyoxalase I and glyoxalase II, and occurs in many organisms from bacteria to man [1,2]. The substrate for glyoxalase I is the thiohemiacetal produced when GSH reacts nonenzymatically with a 2-oxoaldehyde (Figure 1). Glyoxalase I catalyses the isomerisation of this thiohemiacetal to produce the thiolester of GSH and the corresponding 2-hydroxycarboxylic acid. Glyoxalase II then catalyses the hydrolysis of the thiolester to produce GSH and a free 2-hydroxycarboxylic acid. The primary physiological function of the glyoxalase system appears to be the detoxication of methylglyoxal, which is produced as a byproduct of metabolism [3]. A recent report indicates that glyoxalase II may also be important in the regulation of spermatogenesis [4]. The glyoxalase system has been studied with regard to diabetes because the accumulation of methylglyoxal in diabetics can lead to pathophysiological complications (reviewed in [5]). It has also been targeted in the design of novel anti-protozoal [6] and anti-tumour drugs [7–9].

Published: 26 August 1999 Structure September 1999, 7:1067–1078 http://biomednet.com/elecref/0969212600701067 0969-2126/99/$ – see front matter © 1999 Elsevier Science Ltd. All rights reserved.

Previously we published the structure of human glyoxalase I [10]. We now report the structure of human glyoxalase II (EC 3.1.2.6). The enzyme is a monomer of molecular weight 29 kDa and has no sequence similarity to glyoxalase I. It has a broad specificity for thiolesters but a high specificity for the glutathione moiety [11]. It has been cloned and heterologously expressed in Escherichia coli Figure 1

O R

O

O

C2 C1 H

+

GSH

R

Glyoxalase I

OH

C2 C1 SG H

OH O R

Glyoxalase II

OH R

C2 C1 SG

C2 C1

OH + GSH O

H

H H2O

Structure

The glyoxalase system. GSH represents reduced glutathione (γ-L-GluL-Cys-Gly).

1068

Structure 1999, Vol 7 No 9

[12]. Recently, based on sequence homology, it has been shown that glyoxalase II is a member of a superfamily of hydrolases [13,14]. The family includes the metallo-β-lactamases for which the structures from three sources have been published [15–17]. These enzymes contain either a mononuclear [15] or a binuclear [16,17] zinc-binding site. Throughout the family, most of the residues involved in coordination of the zinc ions in the metallo-β-lactamases are conserved, suggesting that the glyoxalase II enzymes may contain at least one metal-binding site [13]. Previously metal analyses had not been done on human glyoxalase II but an examination of the enzyme from Arabidopsis thaliana showed that the enzyme binds two moles of zinc per mole of II and that zinc is essential for activity [18]. In the present investigation, the crystal structure of human glyoxalase II has been solved by single isomorphous replacement combined with anomalous scattering. Two complexes are described. The first contains acetate and cacodylate ions, from the crystallisation mixture, bound in the active site. The second is derived from crystals soaked in a solution containing the slow substrate S-(N-hydroxy-N-bromophenylcarbamoyl)glutathione (HBPC–GSH) [19]. In this structure, HBPC–GSH can be seen in one molecule of the asymmetric unit and GSH, the product of the reaction, can be seen in the other.

N-terminal domain, including residues 1–173, has the topology of a four-layered β sandwich with two mixed β sheets flanked by α helices. The first half of the sandwich has a βββαβαββ topology; the second folds as a ββββαβ unit and is aligned at an angle of ~20° to the first (Figure 2a). The βββαβ motif from the two halves (β1β2β3α1β4 and β8β9β10α3β11) can be superimposed by a twofold rotation. We can align 39 out of 57 pairs of Cα atoms with an rmsd of 1.4 Å (see the Materials and methods section). A structure-based sequence alignment shows that the sequence identity from the N- and C-terminal halves is ~14%. Residues 136–141, situated directly after β10 form a β hairpin at an angle of approximately 80° to the direction of β10 (Figure 2). This is part of an extended loop structure that is not present in the N-terminal half of the sandwich. The second domain is situated at one edge of the first and consists of residues 174–260 folded into five α helices. That this domain would be predominantly α-helical was suggested from circular dichroism studies [21]. Directly preceding helix α6, there is a short section of chain that extends the second sheet of domain 1 by hydrogen bonding to β11. As can be seen from Figure 2, the two domains interact very tightly. In particular, the hairpin loop of the N-terminal domain protrudes into the C-terminal domain and makes both hydrophobic and hydrogen-bonding interactions with residues on helices α4 and α6. Altogether, the C-terminal domain buries ~1200 Å2 of the surface area belonging to the N-terminal domain.

Results and discussion The structure of the enzyme containing acetate and cacodylate was refined against data extending to a dmin of 1.9 Å, to an R factor of 18.5% (15–1.9 Å) and a corresponding Rfree [20] of 23.9%. The final model contains all residues from the two protein molecules of the asymmetric unit, which are related by a local twofold axis. Associated with each protein molecule are two zinc ions, an acetate ion, a cacodylate ion and an arsenic atom covalently linked to Cys153. There are a total of 369 water molecules. Two Mn ions and one Cl ion bind at the interface between the two molecules. The two molecules are very similar, with a root mean square deviation (rmsd) in the positions of all corresponding pairs of Cα atoms of 0.37 Å. The structure of the enzyme in complex with HBPC–GSH was refined against data extending to a resolution of 1.45 Å. The final model has an R factor of 20.4% (20–1.45 Å), an Rfree of 23.2 and includes all protein residues, four zinc ions, one molecule of HBPC–GSH, one molecule of GSH and 357 water molecules. The rmsd in the positions of all corresponding pairs of Cα atoms between the two molecules of the asymmetric unit of this structure is 0.35 Å. Between the A molecules of the two different complexes the rmsd is 0.22 Å, whereas for the B molecules it is 0.25 Å.

The first domain of glyoxalase II is structurally similar to the whole structure of the metallo-β-lactamases (Figure 2). Within this domain, the rmsd in the positions of 144 Cα atoms from the metallo-β-lactamase from Bacteroides fragilis [16] that can be superimposed to within 3.8 Å of the corresponding atoms of glyoxalase II is 1.7 Å (see the Materials and methods section). Twenty-five percent of the matched residues are identical in the two sequences. In comparison to glyoxalase II, the N-terminal sheet of the metallo-β-lactamases contains two extra strands at the N terminus. This means that in these enzymes the motif common to the two halves of the sandwich is slightly larger than that seen in glyoxalase II (ββββαβα) [15]. Although there is only one domain in the metallo-β-lactamases, there is a topologically equivalent strand in these enzymes to the strand that hydrogen bonds to β11 of glyoxalase II [22]. As in glyoxalase II, this strand immediately precedes an α helix (α6 in glyoxalase II, the C-terminal helix in the metallo-β-lactamase); however, it follows directly on from the β sandwich rather than via the α−α motif seen in glyoxalase II. As can be seen in Figure 2, the β hairpin in glyoxalase II is replaced by a gentler loop structure in the metallo-β-lactamases.

Overall structure

Active site

The protein consists of two structural domains (Figure 2a), as was postulated by limited-proteolysis studies [21]. The

Within each molecule there is clear electron density for a binuclear metal site (Figure 3). Metal analysis shows that

Research Article Structure of human glyoxalase II Cameron et al.

1069

Figure 2 Overall structure of glyoxalase II and a comparison with the metallo-β-lactamase. (a) Schematic representation of glyoxalase II. The molecule has been colour ramped according to residue number starting with red at the N terminus and finishing with blue at the C terminus. The β strands of the first domain and the α helices of the second domain have been labelled for clarity. The metal ions and the coordinating residues are represented by balls and sticks. The figure was prepared using MolScript version 2.1 [54]. (b) Similar view of the metallo-βlactamase from B. fragilis [16] after it had been superimposed on glyoxalase II. (c) Topology diagram of glyoxalase II. The approximate position of the twofold axis relating the two halves of the β sheet is indicated. The numbering shows the N and C termini, respectively, of each secondary structure element. The secondary structural elements were defined using the YASSPA algorithm implemented in O [45].

(a)

(b) C

α7 α8

C

α4

β6

Hairpin α6

β11

α5

β5 β7

β9 β8 β10 β2

β4

β3

β1

N

N

(c)

258 α8

β12 256 215

254

α6

α7

220

241

236

α3

230 171

206

α5

α4 191

β11

176

186 8 β1 2

149

157

133

114

β10

108

β9

β8

93 β7

167

127

121

100

97

12

28

53

77

90

β2 18

N

β3

β4 24

33

β5 49

63

β6 73

87

α2

α1 43

69 Structure

there are ~1.5 moles of zinc per mole of protein. Though the results also show the presence of 0.7 moles of iron per mole of protein, taken together with the metal-binding studies on the A. thaliana enzyme [18], it seems that human glyoxalase II is a zinc metalloenzyme. It is possible that the zinc content is slightly less than expected because some metal exchange occurred during purification. The metalbinding site is situated at a topologically equivalent position to that observed in the metallo-β-lactamases at one edge of the β sandwich (Figure 2). One of the metal ions is less than 1 Å from the rotation axis that would superimpose one β sheet onto the other. The active site extends from this metal-binding site across the domain interface. HBPC–GSH is clearly defined in the structure derived from the crystal soaked in this compound (Figure 3) binding in a groove on the surface of the molecule.

However, it is only seen in one molecule of the asymmetric unit. In the other molecule, glutathione, the product of the reaction, can be observed. In the other structure obtained from crystals grown in the presence of acetate and cacodylate these entities are clearly seen in the active site of both protein molecules. Arsenate ions have, in fact, been reported to inhibit the enzyme [11]. Metal binding

The coordination sphere of the two metal ions varies slightly depending on the ligand bound in the active site. In all cases, seven protein residues and one water molecule interact directly with the two zinc ions (Table 1, Figure 4). The two metals are separated by a distance of 3.3–3.5 Å and are bridged by both a water molecule and the Oδ1 atom of Asp134. The latter, however, appears to

1070

Structure 1999, Vol 7 No 9

Figure 3

Y175

Y175

H173

H173

D134

D134

H59

H59

Z2

Z2

Z1

Stereoview of the electron density in the active site of the HBPC–GSH complex. The density is calculated with coefficients 2mFobs–dFcalc using the final structure and contoured at 1σ. The ligand is shown with orange carbon atoms and protein residues are shown with khaki carbon atoms. Although the density for the phenyl group is poor, it is clear how the ligand binds. The γ-glutamate of the HBPC–GSH is not shown beyond the carbonyl oxygen. The figure was prepared in O.

Z1 H110

H110

H54

H54 H56

H56 Structure

two metal ions, respectively (Figure 4a). This gives each of the two zinc ions an octahedral coordination. When glutathione is present in the active site, another water molecule is situated 2.5 Å from zinc 1 and the sulphur of the glutathione is positioned 2.8 Å from zinc 2 in similar positions to the two oxygens of the cacodylate ion (Figure 4b). In the HBPC–GSH complex the carbonyl oxygen of the substrate analogue replaces the water molecule bound to zinc 1 when glutathione is present and the sulphur is a little further from zinc 2 (Figure 4c, Table 1). Despite the differences in the coordination there is very little change in the positions of the liganding residues and the coordination can be described as octahedral or square pyramidal. Though it is more common that zinc ions in protein structures have a tetrahedral coordination, they often have a

be in a less favourable position for the coordination of zinc 1 than for zinc 2. Zinc 1 is further from the carboxylate oxygen, is more out of the plane of the carboxylate group and interacts with the anti rather than the more favourable syn lone pair [23] of the aspartic acid. Zinc 1 is also coordinated by His54 Nε2, His56 Nδ1 and His110 Nε2, while zinc 2 interacts with Asp58 Oδ1, His59 Nε2 and His173 Nε2. These residues come from four different regions of the first domain. Superposition of the two β sheets upon one another using the twofold rotation mentioned above brings His54 onto His173. These residues are also matched by a structure-based sequence alignment of the two sheets. In the cacodylate-bound structure, the coordination sphere of each metal is completed by the two oxygen atoms of the cacodylate ion interacting with the Table 1

Distance from metals of coordinating residues in human glyoxalase II. Metal

Coordinating atom

Complex with cacodylate

Complex with HBPC–GSH

Residue

Atom

A

B

A

B

Nε2 Nδ1 Nε2 Oδ2 O/OZ1

2.31 2.32 2.25 2.21 2.23

2.23 2.24 2.25 2.31 2.01

2.15 2.21 2.13 2.58 2.50

2.14 2.23 2.14 2.57 3.28

Zn 1

His54 His56 His110 Asp134 CAC/Wat2 HBPC–GSH† Wat1

O

2.12

2.14

2.00

2.00

Zn 2 Zn 2 Zn 2 Zn 2 Zn 2 Zn 2

Asp58 His59 Asp134 His173 CAC/GSH† Wat1

Oδ2 Nε2 Oδ2 Nε2 O1/S O

2.34 2.16 2.07 2.12 2.37 2.11

2.22 2.05 2.14 2.08 2.37 1.92

2.29 2.10 2.09 2.13 2.86 2.16

2.26 2.08 2.08 2.15 3.32 2.03

Zn 1 Zn 1 Zn 1 Zn 1 Zn 1

*Water molecule in turn hydrogen bonded to Asp146. †See text.

Second sphere ligand

Thr53 Oγ1 Wat–Asp 146 Oε2* Lys143 O Asp134 O

Bridging water Asp29 Oδ2 Asp134 O Asp11 Oδ1

Research Article Structure of human glyoxalase II Cameron et al.

1071

Figure 4

Metal coordination in glyoxalase II. (a) The complex with cacodylate. (b) The molecule with GSH bound. (c) The molecule with HBPC–GSH bound. The protein residues are shown with brown carbon atoms and the ligands with red carbon atoms. Only the Cα, Cβ

and S atoms of the cysteine moiety of the GSH and HBPC–GSH are shown. The figure was generated in MolScript version 2.1 [54] and rendered in Raster3D [55].

higher coordination number when present in catalytic sites [24], particularly when part of a zinc cluster [25]; for example, both leucine aminopeptidase [26] and purple acid phosphatase [27] contain an octahedrally coordinated zinc ion as part of a binuclear metal cluster. Each of the residues liganding the zinc ions in glyoxalase II are stabilised by either direct or indirect hydrogen-bonding interactions with other residues (Table 1).

of the zinc ions than the other [28,29]. Recently, a model for the zinc-binding site in glyoxalase II was proposed based on the structure of the B. fragilis enzyme [18]. Although the model correctly predicted six out of seven of the residues binding the zinc ion as well as the shared water molecule, the modellers did not foresee either that His59 would be a zinc ligand or that Asp134 would interact with both metal ions. In the metallo-β-lactamases, the coordination of the metal ions has been described as tetrahedral and distorted trigonal bipyramidal for zincs 1 and 2, respectively [16,17]. The geometry around zinc 2, however, is similar to that of the equivalent ion in glyoxalase II if the active site ligands are excluded from the analysis. Whereas for the S. maltophili metallo-β-lactamase, angles of 146°, 102° and 166° are given for the Wat1–Zn2–His255, His255–Zn2–His89 and Asp88–Zn2–Wat2 angles, respectively [17]; the corresponding angles in glyoxalase II are ~150° (Wat1–Zn2–His173), ~105° (His173–Zn2–His59) and ~164° (Asp58–Zn2–Asp134).

Although the zinc-binding site in glyoxalase II is similar to that observed in the metallo-β-lactamases, it is not identical. Indeed, the zinc-binding site is not completely conserved among the metallo-β-lactamases from the three species from which the structure has been elucidated. All of the structures contain a histidine at the equivalent of positions 54, 56, 110 and 173 in glyoxalase II and an aspartic acid at position 58. The variation comes at the other two residues. Asp134 is replaced by a cysteine in the metallo-β-lactamases from B. fragilis and Bacillus cereus and by a serine in the enzyme from Stenotrophomonas maltophili. When the residue is a cysteine, it only coordinates zinc 2 [16], and when a serine, it interacts with the zinc ion only through an intervening water molecule [17]. Zinc 2 is the zinc ion that is in a more favourable position for interacting with Asp134 in glyoxalase II. In the metallo-β-lactamases from B. fragilis and B. cereus, His59 is replaced by a cysteine and an arginine, respectively. Neither of the latter residues are in positions to coordinate the zinc ion. In both the metallo-β-lactamases from B. fragilis and S. maltophili there is a water molecule shared between the two metal ions as seen in glyoxalase II. In the enzyme from B. cereus, however, the situation is slightly different and the enzyme is able to bind either one or two metal ions. Where there are two zinc ions in the active site, there is also a shared water molecule but it is much closer to one

Substrate-binding site

The glutathione moiety of HBPC–GSH is tightly bound to the protein via its glycine and cysteine residues (Figure 5a,b). In contrast, the γ-glutamate does not make any specific interactions with the protein beyond its carbonyl oxygen atom and it is disordered in the electron density. This agrees with previous studies that indicate that the γ-glutamate is not necessary for binding [30]. It differs, however, from the situation observed in glyoxalase I, where with the same ligand both the glycine and γ-glutamate residues are involved in hydrogen-bonding interactions with the protein [31]. There are three basic residues in proximity to the carboxylate group of the glycine: Arg249, Lys252 and Lys143 (Figure 5). Arg249 and Lys252 are within hydrogen-bonding distance of the

1072

Structure 1999, Vol 7 No 9

carboxylate. These residues belong to the final α helix of the second domain. Arg249 is anchored in position by further hydrogen-bonding interactions with the carboxylate of Asp253 and the carbonyl oxygen of Cys141. The hydrophobic part of the Lys252 sidechain packs against the phenyl ring of Phe180. Lys143 is situated a little further from the carboxylate and instead hydrogen bonds to the carbonyl oxygen of the γ-glutamate. This residue is situated just after the β hairpin. The carbonyl oxygen of Cys141, which is situated on the hairpin, interacts with the glutathione via a water molecule. The glutathione makes further interactions with Tyr175 and Tyr145, the details of which can be seen in Figure 5b.

Figure 5 (a)

(b)

Asn 179 Lys 252 Cys 141 2.86 2.97 3.02

Tyr 175 2.812.86 3.11 2.64

2.97

Phe 137

3.03

Arg 249 3.14

2.93

Tyr 145

3.21 2.89

Lys 143

3.30 2.67

His 173

HBPC-GSH

3.32

Zn 2 3.09 3.26 2.03

His 56

The S-hydroxybromophenylcarbamoyl (HBPC) binds over the zinc site with its phenyl ring stacking against the imidazole ring of His56. Based on the lower KM of S-mandeloylglutathione than that of S-lactoylglutathione it had been proposed that glyoxalase II would have a hydrophobic pocket for substrate binding [11,30,32] such as occurs in glyoxalase I [10]; however, there is no such pocket and the bromophenyl group of the HBPC–GSH binds in a groove in the protein and is completely accessible to solvent. The hydroxycarbamoyl group adopts a trans conformation above the binuclear zinc site such that the zincbound water molecule is 2.9 Å from the C1 atom in a direction perpendicular to the plane of the hydroxycarbamoyl and the carbonyl oxygen is 3.3 Å from zinc 1, as described previously. The hydroxyl oxygen of the analogue is not involved in hydrogen bonding. It has, in fact, been shown that there is no requirement for substrates to have a hydroxyl group at this position [33]. These are the only close interactions made by the S-substituent of glutathione and it would appear that the molecular recognition is through the glutathione moiety and the interactions of the thiolester group with the zinc ions. The absence of close contacts with the protein may explain why the density is less well defined for the HBPC than for the glutathione moiety (Figure 3). It is also possible, however, that the HBPC–GSH has undergone some hydrolysis, as this molecule is a slow substrate for glyoxalase II [19].

3.28

2.00

Zn 1

Structure

View of the active site in glyoxalase II. (a) Residues coordinating the zinc ion or those within hydrogen-bonding distance of the HBPC–GSH are represented as balls and sticks with yellow carbon atoms. The HBPC–GSH is shown as a ball-and-stick representation with orange carbon atoms. The surrounding schematic view of the backbone is coloured as in Figure 2a. The HBPC–GSH is truncated as in Figure 3. (b) Schematic diagram showing the interactions between the HBPC–GSH and the protein. The ligand is shown with purple bonds and the protein residues with brown bonds. Atoms and residues involved in hydrophobic contacts are shown as fanned by red dashes. The figure was made using LIGPLOT [56].

In the molecule in which glutathione has bound in the active site, it adopts a very similar conformation to the equivalent moiety of the HBPC–GSH. The proximity of a crystallographically related molecule would appear to prevent the HBPC–GSH from binding in this molecule of the asymmetric unit. When acetate binds in the active site, it mimics the glycine carboxylate group of the glutathione. There are no significant differences in mainchain conformation among the structures with GSH bound, with HBPC–GSH and with cacodylate and acetate ions bound. Although we do not have a true nonliganded structure with which to compare, the absence of any conformational changes among the various molecules, crystallised under slightly different conditions, agrees with

Research Article Structure of human glyoxalase II Cameron et al.

1073

Figure 6 β1 Human Marmoset Rat A_thaliana S_mansoni C_elegans Yeast_glo2 Yeast_glo4 R_capsulatus R_blasticus A_thaliana_mit E_coli B_aphidicola Synechocystis S_pombe H_influenzae

. . . . . . . . . . K . . . . .

. . . . . . . . . . I . . . . .

. . . . . . . . . . T . . . . .

. . . . . . . . . . H . . . . .

. . . . . . . M . . F . . . . .

. . . . . . . K . . C . . . . .

. . . . . . . F . . S . . . . .

. . . . . . . L . . I . . . . .

. . . . . . . L . . S . . . . .

. . . . . . . Q . . N . . . . .

. . . . . . . Q . . M . . . . .

. . . . . . . I . . P . . . . .

. . . . . . . R M M S . . . M .

. . . . . M . N T A S . . . S .

M M M M . V M M L L L M M M F .

K K K K . Y Q H E N K N I D Q M

V V I I . V V V L L I L L I I L

E E E F . K K K T L E N T H L F

V V L H . S S P P T L S S R P A

L L L V . L I I I V V I I L I L

β2 P P P P . L K K P P P P S P P P

A A A C . R M M C C C A V A M A

L L L L . R R R L L S F L L W L

10 TD TD TD QD .. AD WE WL TD KD KD DD KD AD VG ND

α2 Human Marmoset Rat A_thaliana S_mansoni C_elegans Yeast_glo2 Yeast_glo4 R_capsulatus R_blasticus A_thaliana_mit E_coli B_aphidicola Synechocystis S_pombe H_influenzae

W W W W L Y Y Y A D D H I G A D

D D D D D D D D D D D D D D D D

H H H H H H H H H H H H H H H H

60 AG AG AG AG AG CG AD SG IA IQ IG VG VG VG SG TQ

G G G G G G G G G A G G G A G G

N N N N N N N N V V N V V N N V

E E E E L E A L E P A K E R L S

K K K K D G D A A E E E E E N A

L L L I L F I L L I L L I L L F

V V V K V R L Y A H K V I L K K

. . . . T . K S . . . . . . . .

. . . . K . Y I . . . . . . . .

. . . . C . L L . . . . . . . .

. . . . . . S T . . . . . . . .

. . . . . . G G . . . . . . T .

. . . . . . G G . . . . . . Q .

. . . . . . V V . . . . . . D .

N N N N . N N N N N N N N N N N

Y Y Y Y . F Y Y Y F Y Y Y Y Y Y

M M M S . M C S A A A I V I A I

Y Y Y Y . Y Y Y W F Y W W F Y W

L L L L . I L L L L L V I L L L

β3 V V I I M V L L W L L L L L L Y

I I I I S K S S H H H N Y H L Q

D D D D S K D T D D D D N D C R

. . . . . S . . T A E E D R . E

20 DE DE ED ES EQ SE SK ED AT AS DT A. N. QR EE N.

β5 . . . . . . K C . . . . . . . .

. . . . R . E Q . . . . . . . .

. . . . K . K E . . . . . . . .

K K K Q Q R N N Q A E E R A K K

L L L L G Q P S A A R K R K E R

70 ES EP EP VP LE FP TS GH TG TG YG FP YP YP FP YP

G G G D G N K D . . . Q N N H T

L L L I L V V I A A A I V L V V

K K K K K M E K K R K V T E T P

V V V V V I V I V V V V V V I I

Y Y Y Y Y L I I A H I Y F Y Y Y

T T T T . A N R . . . . . . T .

K K Q G . R K R D G G G C N R L

E E E D . A K N T R T R S Q Q P

A A A A . A S S V T V C C A A L

A A A A . L W W V A G L I A A I

I I V V L L L L V L V I I V I I

V V V V Q V I I V V V V I V V V

β6 G G G G G G G G G G G G G G G G

G G G G G G G G A A S P P G G P

. . . S . . . . A A A Q E V . Q

. . . . . . . . A A V E E E . E

. . . . . . . . D D D T T D . C

D D D L D D S S A A K Q K Q S E

80 DR DR DR DK DR SR KD KS HR HR HR DK TR GR DQ KK

I I I V I I C S L L I . . I . .

G G G K G P P P P P P . . P N .

A G A G G A K G P P G G N G G G

L L L C L M V V L L I T V Q V A

T T T T T D T T D D D T N T S T

H H H D D R I E H H I Q K V H Q

K K K A T H I V P A L V I F V I

I G V V V V P P V L L V V L L V

D D D D D D D D D D D D D D D D

30 PV PV PV PV PV LV PA PA VP AP PS PG PG PA PA LP

Q Q Q D E N E E E E E D V E E E

P P P P P E P P A A A A S A V T

Q Q Q E D E P L A A A E E K N D

K K K K K D E E P P P P D P V K

V V V V I Y V V V V V V I V V L

V V I I L K L S L L I L I L M F

β7 90 TH TH TH DN SH KH EN DN AP RE KD KD KQ RD QD NE

L L L G G G L L G G S G G G K G

S S S D Y D K Q D D D E D D E K

T T T K K M K Q V R K T V R T I

L L L L X A L Y L V W A I L L L

Q Q E T R E H H H A M F K S R T

D E E A S E P P K S E N K D P A

A E T S A L E K V A A A K C I W

α1

β4

40 AR AK VK AE VT AD LT LS LT LA LS IA IE LE LK LE

50 VL VL VL VL IL VL IV IV IL IL IL IF IL IY IL VL

. . . . . . E A . . . . . . K .

. . . . . . . . . . . . . . K .

. . . . . . . . . . . . . . L .

S S S . T T T T . . . . . E S .

E Q E E K S D G . . . . . T N .

P P P N E P E E A S S . . I E .

V V V L L F L L L I F V L F I A

P P S P G G R Q G A A P P G H N

A A A A A V C C L L T Y Y D A H

130 VFT VFT VFT VFT VFT VFT IFT IFT LFS AFT IFT LFC IFC LFC VFT LFC

α3

hairpin G G G G G G G G G G G G G G G G

D D D D D D D D D D D D D D D D

T T T T T T T T S S L T T T T A

L L L L L L L L L L I L L I L L

F F F F F F F F M M Y F F F F F

V V V V L I T I S A S S S A N S

A A A A G A A A W A L G G G A A

140 GCG GCG GCG GCG GCG GCG GCG GCG GCG GCG SCG GCG GCG GCG GCG GCG

ZB

L L L L L L L F M Q A Y Y F . V

A A A A E E D D D G T R K Q . E

W W W W A W E A R R R K K T . K

. . . . . . . . . . . . . . . Q

G G R Q G N K K R G N N N G E N

V V V A L I I K W W W W W A I A

K M K K R D S S Q R N Q K D D T

L L L I L I V I L L L P L L L I

T T T K E T E D H T T E I V Q E

G G G G V A G G G G A L L A G N

. . . Q A . . . . . . . . . . .

S S S D T G D N M A G G K D N .

100 LNV LNV LSV INI STF LQI LEI LRV EAA ESA HEV HEF SEI REA VQI YQI

K K K L I K T T Q V R S H T E D

C C C A A C C C V V I V V V A V

T T T F S A A A H D Y A A T A A

T T T T T T N N I L N L L N T L

T T T T T T T T T T T T T T T T

H H H H H H H H H H H H H H H H

H H H H H H H H H H H H N H H E

H H H H H H H H H H H H H H H H

Z2

Z2

K K K K E A . . . . . . . . . .

G G G G N D P L . . . . . S . .

β9 L L L L W L I I I I L I F Y L I

A S S H L S R R A D D A L F H P

T T T T H T T T A V T T T V T T

P P P P H P P P D P P P P P P G

K K K K R R R R R R T R R R R R

F F F F F F F F L L L L V L F V

Y Y Y F F F F F F F S F F F F F

E E E E E E E E E E E E K E E T

G G G G G G G G G G G G N G G G

T T T T T T T T T T T T K T T N

A A A A A A G G P P P A F P A Y

150 DEM DEM DEM EQM EQM PQM EEM RDM AQM AEM EQM SQM FDM AQM AEM ALM

C C Y Y F D D D F W L Y Y V H F

β11 K K K Q K V I M D A S Q Q Q I E

A A A S A A A A T S S S S S A G

L L L L L L L L L L L L I I L L

. . . . . N N N . . . . . . N .

. . . . . . N Q . . . . . . . .

. . . . . . S I . . . . . . . .

. . . . . . I M . . . . . . . .

L L L C I . L L . . . . . . . .

E E E V E E E R T S Q K N G A Q

V V V T V I T A R K K K F K V R

160 LGR LGR LGR LAA LSK LKN VGR VGE LAA LAA IVS LSA IKS LRQ LSS LNT

. . . . . . . . . . . . . . . R

. . . . . . . . . . . . . . . I

. . . . . . . . . . . . . . . F

200 AKE AKE AKE ARQ VKR AQR LEQ LEQ VTA LSE VAH VKE IKK VQG LEG VET

K K K Q L Q F Y L M L L H D L Q

Y Y N R R I C C R R R R I R C R

L F L L L L Q T L L L L L L L L

P P P P P P N N P P P P P P P P

. . . . . . W W . . . . . . . .

P P P K T V S N P P D D K D N D

D D D P T E K K E E D D K Q N E

T T T T T T T V T T T T T T T T

R R K Q K Q R K R L N L I R V I

V V V V V I V I V I I V L L I V

C C C C R C C C G G G G G G C G

110 HTS HTS HTS HTK HTT HTS HTR HTK HTL HTR HTQ HTL HTL HTR HTR HTK

120 S.. T.. S.. NG. T.. TNP K.. K.. PG. PG. PG. K.. K.. AP. HS. D..

G G G G G G D D G G G G G G D Q

H H H H H H S S H H H H H H S H

I I I I I I I I I I I I V I I V

C C C S C C C C A A S C S A C S

Y Y Y Y Y Y Y Y W F F Y Y Y F F

F F F Y L H Y Y Y H Y F Y Y Y L

V V V V V I V I L F F S L F A V

R R R L L C R R L L L L M L S E

. . . . . . . A . . . . . . K T

H H H T T H K K S A S S S T H V

V V V V V V I I L L V I I V L L

E E E E E E Y Y E E E L L D Q V

P S P P P P P S P P P P P P S D

G G G N K G Q D G G K H F S . K

190 NAA NAA NTA NGK NEA NEK VGE IGQ NGR NPA NET DLS DKK NKD .EA SAV

S S A Q A E S K L R S A S Q N A

I I I A S Q K S A E Q K Q R H E

G G G D N G H N G G G N N G N N

E E E L Q G E E E R L Q K Q Q K

P P P P A F V C P L P I T A F P

210 TVP AVP TVP TIP SVP TVP TAG TTG SLP PMP SIP TLP SLP TIP IAG SLP

. . . . . . R H . . . . . . H .

S S S S G S F F V W T V V S I T

T T T T T T T T T T T I S W T T

L L L L I V L L L L V L L L M L

A A A E G A K R G A K K E G G K

E E E E E E D D E E V N T T Q L

E E E E E E E E E E E E E E E E

R R R R R R R R R R Q R R R R R

250 REK REK REK NKK EEK EMK AMK KLK ARK ARK RAR SKK KEK GMK TLK KAK

D D D D D N N N D D D D D D N

Q Q Q Q R K R A R L R R D N Q

F F F W F F M M F F F F F F G

K K K R . . . . . . . . . . .

G

G

α4

170 YCG YCG ICG YCG YCG FPG YPG YPG CSG CSG YCG CCA CCS WCA YPG CPA

GM G

α6 Human Marmoset Rat A_thaliana S_mansoni C_elegans Yeast_glo2 Yeast_glo4 R_capsulatus R_blasticus A_thaliana_mit E_coli B_aphidicola Synechocystis S_pombe H_influenzae

H H H H R E E E R Q K N K L K Q

P P P E E A D D . . . . . G . .

Z2

β10 G G S . S S T E . . . . . G . .

K K K K Q K D E D A R A K T N K

β8

Z1Z1

Human Marmoset Rat A_thaliana S_mansoni C_elegans Yeast_glo2 Yeast_glo4 R_capsulatus R_blasticus A_thaliana_mit E_coli B_aphidicola Synechocystis S_pombe H_influenzae

. . . . . . . . . . . . . . K .

H H H H H H H H H H R H H H H H

E E E E E E E E E D E E E E E E

Y Y Y Y Y Y Y Y Y Y N Y Y Y Y Y

Z1

G

240 DPV DPV DPV SPI DPI DAV DRA PRS P.L D.T E.A QPE S.F EPA DPI ..L

T P T D K V Q V A V E E E R K E

T T T T T T T T T L T T T T T T

I I V V V V S K A D A L L L K L

N N N K K A D G A G G S S G S G

N N N N N N N N N N N N N N N N

T T T T A G I V V V A R V V V E

M M M M M M M M F F L F F F M F

R R R R K A D Q T A R A I G D T

A A A E T K K E A E R W L K E A

α7 F F F L L K V L R L K R K K K R

220 TYN TYN TYN ETN ATN ATN EFN GYN ATN ATN ACN QIN KIN RTN QFN EIN

P P P P P P P P P P P V I P P P

F F F F L F F F F F F F F F F F

M M M M M M M M L L L L L L M L

R R R R R R R R R R R R R R R Q

V V V V V V L L A A I T T W V A

R R K D S R E D D P S E N D T K

E E E K E E D D D L S D E N D T

. . . . . S . . . . . . . . . .

230 KTV KTV KTV PEI PDV EEI PKV RAV AAL PQM KDI IDL KTI PAI PEL ...

Q Q Q Q L Q Q R R K R I K Q Q .

Q Q Q E A K K L A A K N K A K .

180 LKF LKF LKF LEF LEF LKF VKF VSF GRF LRF LKF MKF LNF LKF VKF LAF

α5 A A A A G A V I A A A A A A A A

α8 H H H K H S A A A A S V A R H .

A A A L A I A V L V L I M V L .

G G G G K G G G G G S N G G G .

E E E C T T D D L L I E L M L .

T T T K T C T T P P P E K T N .

. . . . . . N A G . D T K . . .

. . . . . . N G D Q S L D . . .

. . . . . . S T A A A L T . . .

. . . . . . W Y A S T Q S . . .

V V I V I L L L A I I L L L L L

I V V I L A N N L L L I I L L E

R Q Q Q K A K K H I Q N K Q N K

E E E Q H Q A E D S S D K E H S

K K K K R K . . R R Y Y Y R . A

β12 M V V G . . . . . . . . . . .

P P P . . . . . . . . . . . .

R R R . . . . . . . . . . . .

D D D . . . . . . . . . . . .

Structure

Sequence conservation among the glyoxalase II enzymes. The sequences were aligned using ClustalW [37]. Human, Homo sapiens, SWISS-PROT accession no. Q16775; Marmoset, Callithrix jacchus, SWISS-PROT accession no. Q28333; Rat, Rattus norvegicus GENBANK accession no. U97667; A. thaliana, Arabidopsis thaliana, EMBL accession no. Y08357; S_mansoni, Schistosoma mansoni, SWISS-PROT accession no. Q26547; C_elegans, Caenorhabditis elegans, EMBL accession no. AL023828; Yeast_glo2, Saccharomyces cerevisiae, SWISS-PROT accession no. Q05584; Yeast_glo4, Saccharomyces cerevisiae, SWISS-PROT accession no. Q12320; R_capsulatus, Rhodobacter capsulatus, EMBL accession no. X99599; R_blasticus, Rhodobacter blasticus, SWISS-PROT accession no. P05446; A_thaliana_mit, Arabidopsis thaliana, GENBANK accession no. U90927 (mitochondrial); E_coli, Escherichia coli, SWISS-PROT accession no. Q47677; B_aphidicola, Buchnera aphidicola,

SWISS-PROT accession no. Q08889; Synechocystis, Synechocystis PCC6803, SWISS-PROT accession no. P72933; S_pombe, Schizosaccharomyces pombe, EMBL accession no. AL032681; H_influenzae, Haemophilus influenzae, SWISS-PROT accession no. P71374. Numbering corresponds to the human enzyme. The secondary structure is as shown in Figure 2. Vertical lines indicate the ends of the two β sheets of the sandwich. Residues that are identical to at least 14 out of the other 15 matched residues are coloured red and those identical to more than 11, blue. Residues that are conserved but not identical are coloured green. Conservation was calculated in ALSCRIPT [57] with a value of 6. The residues that coordinate zincs 1, 2 or both are denoted by Z1, Z2 and ZB, respectively. Residues within hydrogen-bonding distance of the GSH are shown by G. GM denotes that the glutathione interacts with the mainchain of the residue.

the results from limited proteolysis [21]. There is no equivalent in glyoxalase II of the flexible loop in the metallo-β-lactamases that has been proposed to fold over the active site when substrate binds [34].

Conservation

To date there are a total of seven sequences that are known to encode glyoxalase II enzymes. A BLAST search [35] located other sequences with similarity to glyoxalase II.

1074

Structure 1999, Vol 7 No 9

Figure 7

238 185 259

185

255

259

255

227

217 206

165

227

146

175 137

217 206 113 55

9

165

113 55

9

30

1

86

125 1

86

97 46

146

175 137

30 125

Stereoview of glyoxalase II coloured according to conservation as shown for the glyoxalase sequence in Figure 6. The metal ions and the coordinating residues are shown. The HBPC–GSH is shown in orange. The figure was prepared in ALSCRIPT.

238

97 72

46

72 Structure

The sequences that we believe are most likely to encode glyoxalase II enzymes have been aligned in Figure 6. This is fewer than in a sequence alignment reported by Bito et al. [36] since many of these sequences, when aligned using ClustalW [37], appear to lack the residues at the C terminus that interact with glutathione. The conserved residues are mainly involved in formation of either the metalbinding or the substrate-binding sites (Figure 7). Apart from the A. thaliana mitochondrial sequence, there is complete conservation of the seven residues directly involved in metal coordination as well as most of the key residues that are involved in glutathione binding (Arg249, Lys252, Tyr175 and Asn179). Lys43, however, is replaced in the majority of sequences by an arginine. In the sequence from A. thaliana mitochondria, His173, coordinated to zinc 2, is replaced by an arginine, Tyr175 by an asparagine, Arg49 by a glutamine, Lys252 by an arginine and Lys143 by a serine. Despite these changes among the important residues, Maiti et al. reported glyoxalase II activity [38]. It seems, therefore, that these concerted changes allow the enzyme to bind glutathione derivatives. These workers did report, however, that the activity of the mitochondrial enzyme was much lower than that of the cytoplasmic enzyme. They postulated that this was because of problems in expression; however, it could be that this enzyme does not bind glutathione as well as the others.

motif seems critical in maintaining the active site. A Cβ atom at residue 140 would clash with Asn179, resulting in a disruption of the glutathione-binding site. Similarly, any residue other than a glycine at position 142 would clash with Asp134, one of the metal ligands. The sulphur of Cys141 in this motif is in van der Waals contact distance of the sidechains of Arg249, Met152 and Phe224, which are all conserved. Residues Glu218 and Asn222, which belong to the second domain, hydrogen bond to the β hairpin and presumably stabilise its structure. Again, the sequence alignment suggests that mitochondrial A. thaliana glyoxalase II must have a slightly different active site. It is possible that the hydroxyl of the serine corresponding to Gly140 of the human enzyme will occupy the space occupied by the water molecule binding to glutathione that is seen in each of the molecules of our structures. The phenyl rings of Phe182 and Tyr145 seem to be important in stabilising Lys252 and Lys143, respectively, by stacking interactions with the hydrophobic sidechains of these residues. In the structure, the hydroxyl of Tyr145 is 3.3 Å from the carbonyl oxygen of the glycine moiety of the glutathione (Figure 5b) but in most sequences the tyrosine is replaced by a phenylalanine. Other conserved residues are involved in the formation of the hydrophobic core or are found in sharp turns. Reaction mechanism

Most of the other conserved residues also contribute to the formation of the active site. The sidechains of Asn12, Asp29 and Thr53 are all involved in hydrogen bonding with the metal-coordinating residues. Thr111, Gly133 and Glu174 are important in positioning the mainchain of the neigbouring zinc ligands. The GCG motif including residues 140–142 belongs to the β hairpin (Figure 7). This

Like glyoxalase I, the first enzyme in the glyoxalase pathway, glyoxalase II is a very efficient enzyme with a kcat/KM close to the diffusion limit using S-D-lactoylglutathione as substrate [39]. It has a broad pH optimum, with the enzyme isolated from human liver showing no variation in activity between pH 6.8 and 7.5. The reaction is thought to involve a nucleophilic attack on the C1 atom

Research Article Structure of human glyoxalase II Cameron et al.

1075

Figure 8 A reaction mechanism for glyoxalase II proposed on the basis of the position of HBPC–GSH in the active site. GSH represents reduced glutathione (γ-L-Glu-L-Cys-Gly). The hydroxide ion is next to the carbonyl carbon, zinc 1 close to the carbonyl oxygen and zinc 2 near the sulphur of the HBPC–GSH. The hydroxide attacks the carbonyl carbon to form a negatively tetrahedral intermediate that may be stabilised by coordination to zinc 1. The C–S bond then breaks to yield the product. Presumably, in the apo enzyme, the sixth coordination positions of the zinc ions will be taken up by water molecules but these are not shown in the diagram.

of the substrate [32]. The structure shows a clear candidate for the nucleophile in the reaction. The water molecule that is coordinated to the two zinc ions is situated below the plane of the carbonyl bond of the HBPC–GSH, 2.9 Å from the C1 atom, in a stereochemically ideal position for attack. Since the HBPC–GSH is a slow substrate for glyoxalase II and the true substrate of the reaction is little different [19], it seems likely that this water molecule is indeed the nucleophile. Presumably, the interaction of the water molecule with the two zinc ions, would lower its pKa sufficiently for it to exist in the form of a hydroxide and alleviate the need for a base to abstract a proton. In addition to coordinating the zinc ions, the water molecule is within hydrogen-bonding distance of the Oδ1 atom of Asp58 (Table 1 and Figure 4). Asp58 is unlikely to play any role in proton transfer because of the interaction between its Oδ2 atom and zinc 2, but will help orient the hydroxide for attack and modify its pKa. By analogy with other metallo-hydrolases, nucleophilic attack would presumably result in a negatively charged tetrahedral intermediate. There are no protein residues near to the carbonyl oxygen of the HBPC–GSH that can help stabilise this intermediate. The hydroxyl of Tyr145 is 4.1 Å from the oxygen, but it is not conserved among the glyoxalase II enzymes and is not in a good position for hydrogen bonding. The carbonyl oxygen is, however, only 3.3 Å

from zinc 1. For leucine aminopeptidase, another binuclear zinc enzyme, a reaction has been suggested in which a hydroxide shared between two zinc ions attacks the substrate to form an intermediate in which the two oxygen atoms are coordinated to the zinc ion [26]. Although there is no direct evidence that this occurs in glyoxalase II, it is possible to model the putative transition-state structure with both oxygen atoms interacting closely with zinc 1. This interaction will provide the necessary stabilisation of the negative charge. To complete the reaction, the C1–S bond must be broken, the glutathione protonated and the products must diffuse from the active site to be replaced by water molecules. Without further information it is difficult to say how this occurs. In the structure, however, the sulphur atoms of the HBPC–GSH and GSH are 3.3 Å and 2.9 Å, respectively, from zinc 2. One possibility, therefore, is that the zinc ion will stabilise a thiolate ion formed upon bond breakage. Tyr175 could also play a part in the reaction mechanism. Its hydroxyl oxygen is positioned 3.6 Å from the sulphur; however, chemical modification studies suggest that a tyrosine is not important in the mechanism [32]. A possible reaction scheme is given in Figure 8. This reaction scheme supports the mechanism proposed for the binuclear metallo-β-lactamases in the hydrolysis of the amide bond of β-lactams. This scheme involves the

1076

Structure 1999, Vol 7 No 9

nucleophilic attack of the shared hydroxide on the carbonyl carbon of the β-lactam [16,17]. The interaction of the hydroxide with an aspartic acid seen in glyoxalase II is conserved in the metallo-β-lactamases. For the metallo-βlactamases it has proved difficult to obtain a crystal structure in complex with a substrate analogue [17]. Computer simulations of the binding of β-lactams to the metallo-βlactamases, however, result in models in which the carbonyl oxygen and nitrogen of the β-lactam interact with zincs 1 and 2, respectively, and the carbonyl carbon is situated above the shared hydroxide [16,17]. This mimics the situation observed in glyoxalase II but the nitrogen of the amide bond takes the position of the sulphur of the thiolester bond. Our structure, therefore, provides experimental evidence to corroborate the results of the computer-simulated binding studies of the metallo-β-lactamases and consequently supports the reaction mechanism put forward for these enzymes.

Biological implications The glyoxalase system catalyses the detoxication of methylglyoxal, which is produced as an unavoidable byproduct of glycolysis. The system consists of two enzymes, glyoxalase I and glyoxalase II, which work in concert using glutathione as a coenzyme to convert methylglyoxal or other 2-oxoaldehydes to 2-hydroxycarboxylic acids. Although the enzymes interact with glutathione derivatives, they are not structurally similar to other glutathione-linked enzymes of known structure. Instead, the two enzymes have been shown to belong to two distinct evolutionary classes. Glyoxalase II catalyses the hydrolysis of the thiolester bond of S-D-lactoylglutathione to form D-lactate and glutathione. It has recently been shown to have sequence similarity to other hydrolases, including the metallo-β-lactamases. Here, we describe the structure of human glyoxalase II in complex with a substrate analogue. The structure not only allows us to postulate a mechanism for the catalysed reaction, which may be relevant to many members of the family, but provides another example of how glutathione interacts with proteins. A detailed knowledge of the active site may also be useful in the design of inhibitors of the enzyme that can be used in further probing the function of the enzyme. The structure consists of two domains. The N-terminal domain shows the expected similarity with the metallo-βlactamases, a four-layered β sandwich containing a binuclear zinc-binding site at one edge. The C-terminal domain, on the other hand, is unique to glyoxalase II and consists mainly of α helices. The substrate analogue binds at the domain interface with the glycine and cysteine residues of the glutathione moiety involved in hydrogen-bonding interactions with the protein. In contrast to other glutathione-linked enzymes, the carboxylate group of the γ-glutamate does not make any specific

contacts with the protein. The position of the reactive centre of the ligand 2.9 Å from a water molecule coordinated to two zinc ions suggests that this water molecule, presumably in the form of a hydroxide ion, will act as the nucleophile in the reaction.

Materials and methods Complex with acetate and cacodylate Recombinant protein was expressed and purified as described previously [12]. Crystals were grown in hanging drops. To obtain the crystals for the native and derivative data sets, well liquor including 15–30% w/v PEG 2000 monomethyl ether, 0.2 M Mg-acetate, 0.1 M Na-cacodylate pH 6.5, and 2 mM DTT was mixed with an equal volume (5 µl) of 10 mg/ml glyoxalase II in 10 mM MOPS pH 7.1, containing 10 mM MgCl2 and 25% PEG 400. X-ray data collection statistics can be seen in Table 2. All data were collected from crystals flash frozen in a nitrogen gas stream at 100K. The low-resolution native and derivative data sets were collected on a Rigaku R-AXIS mounted on a rotating-anode X-ray source. The highresolution native data set was collected on a MAR research imaging plate on EMBL beamline X11 at DESY, Hamburg. Data were processed and scaled with the HKL suite of programs [40]. Further processing was carried out using programs from the CCP4 package [41]. The space group was determined to be P212121, with two molecules in the asymmetric unit. One mercury derivative, utilising the anomalous signal, was sufficient to solve the structure. The low-resolution native data set was used during structure solution. Initial heavy-atom sites were located using difference Patterson methods. Following refinement of the heavy-atom parameters in SHARP [42], the calculated phases were used to find minor sites. Phases were improved using DM [43] including solvent flattening, histogram matching and twofold averaging. The resulting map was of high quality so that a complete model could be built using skeletonised density, mainchain and sidechain databases and baton building methods [44,45]. Refinement was carried out against the high-resolution native data set initially in CNS [46] using the slow-cool procedure [47] and torsionangle dynamics [48], and later in REFMAC [49]. All data between 15 and 1.9 Å were used in the refinement. During the initial cycles strict noncrystallographic-symmetry constraints were applied between the two molecules of the asymmetric unit and in later cycles these were replaced with restraints. The progress of refinement was monitored with 5% of the data set aside to calculate Rfree [20]. Water molecules were located with ARP [50]. During the rebuilding stages it was apparent that there was a binuclear metal-binding site and zinc ions were inserted into the corresponding density. Other density observed in the active sites was modelled as acetate and cacodylate ions, respectively. A large peak was also observed in difference maps next to Cys153. This was presumed to be cacodylate covalently linked via an As–S bond but only the arsenic atom of the cacodylate was modelled. Density seen at the interface between the two molecules was assigned to be Mn and Cl ions. For the final structure, the rmsd from ideal bond lengths and angle distances [51] are 0.008 Å and 0.024 Å, respectively, and the rms B value between bonded atoms is 2.1 Å2. Only 2.3% of residues are outliers in a stringent boundary Ramachandran plot [52]. The average temperature factor of the protein is 19 Å2, whereas those for the four zinc ions are 20, 19, 19 and 18 Å2 for zincs 1 and 2 of the A and B molecules, respectively. The cacodylate ions have average temperature factors of 40 Å2 and the acetate ions of 20 Å2.

Complex with HBPC–GSH Crystals were again grown in hanging drops. The protein solution, containing 15 mg/ml enzyme in 10 mM MOPS pH 7.1 and 25% PEG 400, was equilibrated against the well liquor, which contained 4–20% PEG 2000 monomethyl ether, 0.1 M Na imidazole pH 6.5 and 2 mM DTT. The crystals were then soaked for a period of 1 hr 30 min in a solution

Research Article Structure of human glyoxalase II Cameron et al.

1077

Table 2 Data collection, processing and phasing statistics. Data set

Native

Cell dimensions (Å) (a, b, c) 39.3, 72.9, 162.3 Resolution range (Å) 30–2.8 Measured reflections 30,688 Unique reflections 11,532 Completeness (%) 93 (74)† Rmerge§ 5.3 (11.8) < I / sigI > 14 (7) Riso# Phasing power¶ (centric; acentric; anomalous) Rcullis¥ (centric; acentric; anomalous)

Thiomersal*

Native high

HBPC–GSH

39.3, 73.3, 163.1 30–3.0 25,177 9,048 89 (74) 6.0 (9.9) 17 (9) 31.1 (36.0)

39.1, 72.4, 162.1 15–1.9 126,060 36,838 99 (99) 6.3 (31.2) 14 (4)

39.4, 73.3, 164.0 43–1.45 437,671 84,815 99 (100) 11.5 (34.0)‡ 13 (5)

1.6; 2.0; 1.3 0.67; 0.66; 0.88

*Crystals soaked in the presence of 2 mM thiomersal for 4 hours. †Numbers in parentheses are for the highest resolution bins. ‡The high Rmerge is due to the collection of a low-resolution sweep after the highresolution data had been collected and the crystal had undergone some radiation damage. §

Rmerge =

∑∑ 〈I(hkl)〉 - I(hkl) /∑∑ I(hkl) i

hkl

i

hkl

ΣFPH–FP/ΣFP, where FPH is the observed structure-factor amplitude of the derivative and FP is the observed structure-factor amplitude of the native data set. ¶Phasing Power = rms (F /F H PH – FP+FH). ¥R cullis = ΣFPH – FP+FH/ΣFPH – FP.

#R = iso

i

i

containing 20 mM HBPC–GSH, 30% PEG 400 and 2 mM DTT buffered with MOPS at pH 7.1. Data were collected from frozen crystals on a Quantum-4 CCD detector at beamline ID14-EH4 at the ESRF, Grenoble and processed as above. Refinement was carried out in REFMAC using all data between 20 and 1.45 Å. The starting model for refinement included all protein atoms of the refined native structure. Water molecules were located with ARP. No noncrystallographic-symmetry restraints were imposed during refinement. HBPC–GSH was modelled in one molecule of the asymmetric unit and GSH in the other. For the final structure, the rmsd from ideal bond lengths and angle distances [51] are 0.011 Å and 0.026 Å, respectively, and the rms B value between bonded atoms is 2.1 Å2. Only 1.7% of residues are outliers in a stringent boundary Ramachandran plot [52]. The average temperature factor of the protein is 21 Å2, whereas those for the four zinc ions are 18, 17, 15 and 15 Å2 for zincs 1 and 2 of the A and B molecules, respectively. The HBPC–GSH has an average temperature factor of 31 Å2 and the GSH of 33 Å2.

Analysis All superpositions were carried out in O such that all matching Cα pairs were less than 3.8 Å apart after the superposition. Solvent-accessibility calculations were performed in X-PLOR [53].

Metal analysis Metal analyses were carried out by inductively coupled plasma atomic emission spectrometry and covered 11 elements (Al, Ca, Co, Cr, Cu, Fe, Mg, Mn, Mo, Ni and Zn). The protein used in the assay was purified as above but with the addition that all buffers were treated with Chelex100 (Bio-Rad Laboratories Inc., Hercules, Ca, USA).

Accession numbers Coordinates and structure factor amplitudes have been deposited in the PDB with entry codes 1qh3 and 1qh5 for the complexes with cacodylate and HBPC–GSH, respectively.

Acknowledgements The authors wish to acknowledge TA Jones, Uppsala University and GR Davies, University of York, for critical reading of the manuscript; D Creighton, University of Maryland for supplying HBPC-GSH; Eric Forsman, Department of Analytical Chemistry, Uppsala University for performing the metal analyses and Sean McSweeney at the ESRF and the staff at the

EMBL outstation at Hamburg for help with data collection. Support was provided by a grant to T Alwyn Jones from the Göran Gustafsson Stiftelse and a grant to BM from the Swedish Natural Science Research Council.

References 1. Mannervik, B. (1980). Glyoxalase I. In Enzymatic Basis of Detoxication (Jakoby, W. B., ed.), Vol. 2, pp. 263-273. Academic Press, New York. 2. Thornalley, P.J. (1990). The glyoxalase system: new developments towards functional characterization of a metabolic pathway fundamental to biological life. Biochem. J. 269, 1-11. 3. Richard, J.P. (1991). Kinetic parameters for the elimination reaction catalysed by triosephosphate isomerase and an estimation of the reaction’s physiological significance. Biochemistry 30, 4581-4585. 4. Ji, X., Moore, H.D., Russell, R.G. & Watts, D.J. (1997). cDNA cloning and characterization of a rat spermatogenesis-associated protein RSP29. Biochem. Biophys. Res. Commun. 241, 714-719. 5. Thornalley, P. J. (1993). The glyoxalase system in health and disease. Mol. Aspects. Med. 14, 287-371. 6. Thornalley, P.J., Strath, M. & Wilson, R.J. (1994). Antimalarial activity in vitro of the glyoxalase I inhibitor diester, S-p-bromobenzylglutathione diethyl ester. Biochem. Pharmacol. 47, 418-420. 7. Vince, R. & Wadd, W.B. (1969). Glyoxalase inhibitors as potential anticancer agents. Biochem. Biophys. Res. Commun. 34, 593-598. 8. Hamilton, D.S. & Creighton, D.J. (1992). Inhibition of glyoxalase I by the enediol mimic S-(N-hydroxy-N-methylcarbamoyl)glutathione. The possible basis of a tumor-selective anticancer strategy. J. Biol. Chem. 267, 24933-24936. 9. Thornalley, P.J. (1998). Glutathione-dependent detoxification of alphaoxoaldehydes by the glyoxalase system: involvement in disease mechanisms and antiproliferative activity of glyoxalase I inhibitors. Chem. Biol. Interact. 111-112, 137-151. 10. Cameron, A.D., Olin, B., Ridderström, M., Mannervik, B. & Jones, T.A. (1997). Crystal structure of human glyoxalase I — evidence for gene duplication and 3D domain swapping. EMBO J. 16, 3386-3395. 11. Uotila, L. (1973). Purification and characterization of S-2hydroxyacylglutathione hydrolase (glyoxalase II) from human liver. Biochemistry 12, 3944-3951. 12. Ridderström, M., Saccucci, F., Hellman, U., Bergman, T., Principato, G. & Mannervik, B. (1996). Molecular cloning, heterologous expression, and characterization of human glyoxalase II. J. Biol. Chem. 271, 319-323. 13. Neuwald, A.F., Liu, J.S., Lipman, D.J. & Lawrence, C.E. (1997). Extracting protein alignment models from the sequence database. Nucleic Acids Res. 25, 1665-1677.

1078

Structure 1999, Vol 7 No 9

14. Melino, S., Capo, C., Dragani, B., Aceto, A. & Petruzzelli, R. (1998). A zinc-binding motif conserved in glyoxalase II, beta-lactamase and arylsulfatases. Trends Biochem. Sci. 23, 381-382. 15. Carfi, A., et al. & Dideberg, O. (1995). The 3-D structure of a zinc metallo-beta-lactamase from Bacillus cereus reveals a new type of protein fold. EMBO J. 14, 4914-4921. 16. Concha, N.O., Rasmussen, B.A., Bush, K. & Herzberg, O. (1996). Crystal structure of the wide-spectrum binuclear zinc beta-lactamase from Bacteroides fragilis. Structure 4, 823-836. 17. Ullah, J.H., et al. & Spencer, J. (1998). The crystal structure of the L1 metallo-beta-lactamase from Stenotrophomonas maltophilia at 1.7 Å resolution. J. Mol. Biol. 284, 125-136. 18. Crowder, M.W., Maiti, M.K., Banovic, L. & Makaroff, C.A. (1997). Glyoxalase II from A. thaliana requires Zn(II) for catalytic activity. FEBS Lett. 418, 351-354. 19. Murthy, N.S., Bakeris, T., Kavarana, M.J., Hamilton, D.S., Lan, Y. & Creighton, D.J. (1994). S-(N-aryl-N-hydroxycarbamoyl)glutathione derivatives are tight-binding inhibitors of glyoxalase I and slow substrates for glyoxalase II. J. Med. Chem. 37, 2161-2166. 20. Brünger, A.T. (1992). Free R value: a novel statistical quantity for assessing the accuracy of crystal structures. Nature 355, 472-475. 21. Aceto, A., et al. & Petruzzelli, R. (1998). Structural characterization of human glyoxalase II as probed by limited proteolysis. Biochem. Mol. Biol. Int. 44, 761-769. 22. Carfi, A., Duée, E., Paul-Soto, R., Galleni, M., Frère, J.M. & Dideberg, O. (1998). X-ray structure of the ZnII beta-lactamase from Bacteroides fragilis in an orthorhombic crystal form. Acta Crystallogr. D 54, 45-57. 23. Christianson, D.W. (1991). Structural biology of zinc. Adv. Prot. Chem. 42, 281-355. 24. Alberts, I.L., Nadassy, K. & Wodak, S.J. (1998). Analysis of zinc binding sites in protein crystal structures. Prot. Sci. 7, 1700-1716. 25. Vallee, B.L. & Auld, D.S. (1993). New perspectives on zinc biochemistry: cocatalytic sites in multi-zinc enzymes. Biochemistry 32, 6493-6500. 26. Sträter, N. & Lipscomb, W.N. (1995). Transition state analogue Lleucinephosphonic acid bound to bovine lens aminopeptidase: X-ray structure at 1.65 Å resolution in a new crystal form. Biochemistry 34, 9200-9210. 27. Sträter, N., Klabunde, T., Tucker, P., Witzel, H. & Krebs, B. (1995). Crystal structure of a purple acid-phosphatase containing a dinuclear Fe(III)-Zn(II) active-site. Science 268, 1489-1492. 28. Carfi, A., Duée, E., Galleni, M., Frère, J.M. & Dideberg, O. (1998). 1.85 Å resolution structure of the zinc (II) beta-lactamase from Bacillus cereus. Acta Crystallogr. D 54, 313-323. 29. Fabiane, S.M., Sohi, M.K., Wan, T., Payne, D.J., Bateson, J.H., Mitchell, T. & Sutton, B.J. (1998). Crystal structure of the zinc-dependent betalactamase from Bacillus cereus at 1.9 Å resolution: binuclear active site with features of a mononuclear enzyme. Biochemistry 37, 12404-12411. 30. Al-Timari, A. & Douglas, K.T. (1986). Inhibition by glutathione derivatives of bovine liver glyoxalase II (hydroxyacylglutathione hydrolase) as a probe of the N- and S-sites for substrate binding. Biochim. Biophys. Acta 870, 219-225. 31. Cameron, A.D., Ridderström, M., Olin, B., Kavarana, M.J., Creighton, D.J. & Mannervik, B. (1999). The reaction mechanism of glyoxalase I explored by an X-ray crystallographic analysis of the human enzyme in complex with a transition state analogue. Biochemistry 38, in press. 32. Ball, J.C. & Vander Jagt, D.L. (1981). S-2-hydroxyacylglutathione hydrolase (glyoxalase II): active-site mapping of a nonserine thiolesterase. Biochemistry 20, 899-905. 33. Vander Jagt, D.L. (1993). Glyoxalase II: molecular characteristics, kinetics and mechanism. Biochem. Soc. Trans. 21, 522-527. 34. Fitzgerald, P.M., Wu, J.K. & Toney, J.H. (1998). Unanticipated inhibition of the metallo-beta-lactamase from Bacteroides fragilis by 4-morpholineethanesulfonic acid (MES): a crystallographic study at 1.85 Å resolution. Biochemistry 37, 6791-6800. 35. Altschul, S.F., Gish, W., Miller, W., Myers, E.W. & Lipman, D.J. (1990). Basic local alignment search tool. J. Mol. Biol. 215, 403-410. 36. Bito, A., Haider, M., Hadler, I. & Breitenbach, M. (1997). Identification and phenotypic analysis of two glyoxalase II encoding genes from Saccharomyces cerevisiae, GLO2 and GLO4, and intracellular localization of the corresponding proteins. J. Biol. Chem. 272, 21509-21519. 37. Thompson, J.D., Higgins, D.G. & Gibson, T.J. (1994). ClustalW: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22, 4673-4680.

38. Maiti, M.K., Krishnasamy, S., Owen, H.A. & Makaroff, C.A. (1997). Molecular characterization of glyoxalase II from Arabidopsis thaliana. Plant Mol. Biol. 35, 471-481. 39. Guha, M.K., Vander Jagt, D.L. & Creighton, D.J. (1988). Diffusiondependent rates for the hydrolysis reaction catalyzed by glyoxalase II from rat erythrocytes. Biochemistry 27, 8818-8822. 40. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276A, 307-326. 41. Collaborative Computation Project Number 4. (1994). The CCP4 Suite: programs for protein crystallography. Acta Crystallogr. D 50, 760-763. 42. de La Fortelle, E. & Bricogne, G. (1997). Maximum-likelihood heavyatom parameter refinement for multiple isomorphous replacement and multiwavelength anomalous diffraction methods. Methods Enzymol. 276A, 472-494. 43. Cowtan, K. (1994). An automated procedure for phase improvement by density modification. Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography 31, 34-38. 44. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron-density maps and the location of errors in these models. Acta Crystallogr. A 47, 110-119. 45. Jones, T.A. & Kjeldgaard, M.O. (1997). Electron density map interpretation. Methods Enzymol. 277, 173-208. 46. Brünger, A.T., et al. & Warren, G.L. (1998). Crystallography and NMR system: a new software system for macromolecular structure determination. Acta Cryst. D 54, 905-921. 47. Brünger, A.T. & Krukowski, A. (1990). Slow-cooling protocols for crystallographic refinement by simulated annealing. Acta Crystallogr. A 46, 585-593. 48. Rice, L.M. & Brünger, A.T. (1994). Torsion angle dynamics: reduced variable conformational sampling enhances crystallographic structure refinement. Proteins 19, 277-290. 49. Murshodov, G.N., Vagin, A.A. & Dodson, E.J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta. Crystallogr. D 53, 240-255. 50. Lamzin, V.S. & Wilson, K.S. (1993). Automated refinement of protein models. Acta Crystallogr. D 49, 129-147. 51. Engh, R.A. & Huber, R. (1992). Accurate bond and angle parameters for X-ray protein structure refinement. Acta Crystallogr. A 47, 392-400. 52. Kleywegt, G.J. & Jones, T.A. (1996). Phi/Psi-chology: Ramachandran revisited. Structure 4, 1395-1400. 53. Brünger, A.T. (1992). X-PLOR version 3.1: A system for X-ray Crystallography and NMR., Yale University Press, New Haven, CT. 54. Kraulis, P.J. (1991). MolScript: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946-950. 55. Merrit, E.A. & Murphy, M.E.P. (1994). Raster3D version 2.0 — a program for photorealistic molecular graphics. Acta. Crystallogr. D 50, 869-873. 56. Wallace, A., Laskowski, R. & Thornton, J. (1995). LIGPLOT: a program to generate schematic diagrams of protein–ligand interactions. Protein Eng. 8, 127-134. 57. Barton, G. J. (1993). ALSCRIPT a tool to format multiple sequence alignments. Protein Eng. 6, 37-40.

Because Structure with Folding & Design operates a ‘Continuous Publication System’ for Research Papers, this paper has been published on the internet before being printed (accessed from http://biomednet.com/cbiology/str). For further information, see the explanation on the contents page.