J. Mol. BioZ. (1992) 225, 791-809

Crystal Structure of Yeast Cu,Zn Superoxide Dismutase Crystallographic

Refinement at 2.5 A Resolution

K. Djinovic l, G. Gattil, A. Codal, L. Antolini2, G. Pelosi3 A. Desideri4, M. Falconis, F. MarmocchV G. Rotilio5 and M. Bo1ognesil-f ‘Dipartimento di Genetica e Microbiologia and Centro Interuniversitario Studio Macromolecole Informazionali, Universitd di Pavia Via Abbiategrasso 207, I-27100 Pavia, Italy 2Dipartimento di Chimica, Universitd di Modena Via Giuseppe Campi 183, I-41100 Modena, Italy 31stituto di Chimica Generale ed Inorganica, Universiki Viale delle Scienze, I-43100 Parma, Italy

di Parma

4Dipartimento di Chimica Organica e Biologica Universiki di Mess&a, 98166 Mess&a, Italy ‘Dipartimento di Biologia, Universiki di Romu “Tor Vergtata” Via Orazio Raimondo, 00173 Roma, Italy ‘Dipartimento di Biologia Cellulare, Universitci di Camerino Via Scalzino 5, I-62032 Camerino, Italy (Received 2 October 1991; accepted 22 January

1992)

The structure of Cu,Zn yeast superoxide dismutase has been determined to 2,5 A resolution. The enzyme crystallizes in the P2,2,2 space group with two dimeric enzyme molecules per asymmetric unit. The structure has been solved by molecular replacement techniques using the dimer of the bovine enzyme as the search model, and refined by molecular dynamics crystallographic with crystallographic pseudo-energy terms, followed by conventional restrained refinement. The R-factor for 32,088 unique reflections in the 160 to 2.5 A resolution range (g&2% of all possible reflections) is O-158 for a model comprising two protein dimers and 516 bound solvent molecules, with a root-mean-square deviation of 0.016 A from the ideal bond lengths, and an average B-factor value of 29.9 A’. A dimeric molecule of the enzyme is composed of two identical subunits related by a noncrystallographic 2-fold axis. Each subunit (153 amino acid residues) has as its structural scaffolding a flattened antiparallel eight-stranded p-barrel, plus three external loops. The overall three-dimensional structure is quite similar to the phylogenetically distant bovine superoxide dismutase (55 O/camino acid homology), the largest deviations can be observed in the regions of amino acid insertions. The major insertion site hosting residues Ser25A and Gly25B, occurs in the 2,3 B-turn between strands 2b and 3c, resulting in the structural perturbations of the two neighbouring strands. The second insertion site, at the end of the 3c P-strand in the wide Greek-key loop, hosts the Asn35A residue, having an evident effect on the structure of the loop and possibly on the neighbouring 5,4 p-turn. The salt bridge Arg77-Asp99 and the disulphide bridge Cys55-Cys144 stabilize the loop regions containing the metal ligands. The stereochemistry of the two metal centres is conserved, with respect to the bovine enzyme. The Cu 2+ ligands show an uneven distortion from a square plane, while Zn2+ co-ordination geometry is distorted tetrahedral. The imidazole ring of the His61 residue forms a bridge between Cu and Zn ions. A solvent peak compatible with a fifth ligand is observed 2.0 A away from the copper in the active site channel, which is filled by ordered water molecules that possibly contribute to the stability and function of the enzyme. 7 Author to whom all correspondence should be addressed. 791

0022%2836/92/l

10791-19

$03.00/O

0

1992 Academic

Press Limited

792

K. Djinovic

et al.

The charged residues responsible for the electrostatic guidance of the substrate to the active site (Glu130, Glu131, Lys134 and Arg141) are fairly conserved in their positions, some of them showing different interactions in the four chains due to the intermolecular contacts between the dimers. The B-barrel of the bovine enzyme is closed at opposite ends by apolar interactions involving Leu104 and Leu36; the latter is substituted by Ser36 in the yeast enzyme. Three concerted substitutions Ala87+Thr, His41 +Arg and Glullg-*Ala allow for the formation of a “plug” for the b-barrel by means of a hydrogen bonding scheme. Keywords: superoxide dismutase; copper; zinc; X-ray; enzyme structure

1. Introduction Superoxide dismutase enzymes are metalloproteins containing either copper and zinc, or manganese, or iron as natural prosthetic groups, which catalyse the disproportionation of the superoxide radical anion into molecular oxygen and hydrogen peroxide. This molecular function is apparently of primary importance for the evolution of aerobic risk organisms, in view of the high oxidative produced by radical chain reactions initiated by 0;. The distribution of the three types of superoxide dismutases (SOD?) in organisms is considered to be characteristic of the evolutionary state of the organism and of the associated cell organelles. Structural studies have shown that the iron- and manganeseSOD types are closely related but bear little resemblance to the Cu,Zn SOD (Stallings et al., 1984). The iron-SOD has usually been isolated as a dimer of identical J4, = 20,000 subunits from prokaryotes and anaerobic bacteria. The manganese type has been found as an oligomer of identical &Z, = 20,660 subunits in prokaryotes and mitochondria of eukaryotic cells (Ludwig et al., 1991), while the Cu,Zn enzyme is predominantly found in the cytoplasm of all higher eukaryotic cells (Stallings et al., 1987; Parker & Blake, 1988). The structurefunction relationships in the Cu,Zn enzymes have been thoroughly investigated in the bovine Cu,Zn SOD (Bannister et al., 1987) by high resolution crystallographic analysis (Tainer et al., 1982) and by extensive spectroscopic studies (Bannister et al., 1987). The isolated protein contains Cu2+ and Zn2+ at the active site, where the catalytically active Cu2+ is cyclically reduced and oxidized during the superoxide encounters with successive t Abbreviations used: SOD, superoxide dismutase; y-SOD, yeast superoxide dismutase; b-SOD, bovine erythrocyte superoxide dismutase; s-SOD, spinach superoxide dismutase; PEG, polyethylene glycol; r.m.s., root-mean-square; m.d.x., molecular dynamics refinement with crystallographic pseudo-energy term; e.m.x., restrained crystallographic refinement with energy minimization. Residues of y-SOD have been numbered according to the numbering scheme of b-SOD (Tainer et aZ., 1982). The 4 y-SOD chains present in the asymmetric unit have been identified with the letters A, B, C and D. Their sequence numbers range from 2 to 151, 202 to 351, 402 to 551 and 602 to 751, respectively.

substrate. In the first step, an electron from one superoxide radical is donated to the catalytic centre with the formation of molecular oxygen and of a Cu+, which in turn donates one electron to a second superoxide radical to produce, together with two protons, hydrogen peroxide (Fielden et al., 1974). The kinetics of the process are dominated by the diffusion of the negatively charged superoxide radical anion into the active site channel. The process is controlled by an evolutionary constant distribution of electrostatic charges around the active site (Desideri et al., 1989. 1992). The amino acid sequences of Cu,Zn SODS (Bannister rt al., 1987; Getzoff et al., 1989) show that the enzyme has been highly preserved during evolution. Among the eukaryotic systems, yeast SOD (y-SOD) has the highest sequence divergence from the bovine enzyme (55% amino acid identit,y: Bannister et al., 1987). Moreover, y-SOD shows lower resistance to urea (Barra et al.. 1979), and a t’ransition temperature for reversible thermal unfolding that is 16 deg. C lower than that reported for the irreversible thermal denaturation of the bovine enzyme (Arnold & Lepock, 1982). The functional y-SOD molecule consists of a dimer of identical subunits, 153 amino acid residues each (M, = 16,000): one Cu2+ Zn2+ pair is associated to each subunit. Tw’o three-dimensional structures of eukaryotic Cu,Zn SOD have so far been reported, for the bovine erythrocyte Cu,Zn enzyme (b-SOD: Tainer rf al.. 1982) and for the spinach (Spinacia oleracea) superoxide dismutase (s-SOD: Kitagawa ef al., 1991). These structural investigations have shown that each subunit has as its structural scaffolding a flattened. antiparallel, eightstranded b-barrel. plus three external loops. In this context we undert,ook the crystallographic study of a new and phylogenetically distant SOD in order to elucidate the development of the enzyme tertiary structure and its relation with the electrostatic and functional properties of the molecule during evolution. 2. Experimental (a) CrystalZization y-SOD was purified as described by Barra et al. (1979). Crystals of the enzyme were obtained by the vapour diffusion technique, adopting the sitting drop experimental setup (MacPherson, 1982). A solution of yeast protein at a concentration of 15 mg/ml in 25 mM-citrate.

793

Crystal Structure of Yeast Cu,Zn Superoxide Dismutase

Table 1 Summary of the data collection and processing data set

High resolution data set

1.009

1909

370 3.0, 1.5 3.5

20, 1.0

Low resolution

Parameter Wavelength (A) Crystal-to-film distance (mm) Oscillation range ( ’ ) Resolution

limit (A)

Total reflections measured Total reflections unique

2505 2.5 131,852 32,538 6.0 982

R merg(?a

Completeness (%)$

t Lr, = xZIZi- (Z)l/zx:lZil, where (I) is the mean value of i intensity measurements. $ Of the theoretical reflections in the l@O to 2.5 A resolution range.

10 mM-phosphate buffer (pH 65), 6% (w/v) polyethyleneglycol (average M, 4000) were equilibrated against a 12% PEG solution at the same pH, at 28°C. Single crystals with prismatic morphology, about @7 mm in each dimension, which were unstable at room temperature in their mother liquor, could be grown in 2 to 5 weeks. They were stabilized by transfer to 35% PEG solutions in the same buffer medium. The preliminary characterization of the crystals has been reported elsewhere (Frigerio et al., 1989). (b) Data collection

and reduction

The crystals belong to the orthorhombic

space group

P2,2,2 with unit cell dimensions of a = 105.2 A, b = 142% ip, c = 62.1 A (1 A = @l nm); the packing density

parameter V,,, (Matthews, 1968) indicates that a tetramer is present in the asymmetric unit, corresponding to 66% bv volume solvent content of the unit cell (or 62,000 A3 bking occupied by the buffer medium). This amounts to approxirnately 1300 water molecules per asymmetric unit. A single crystal was used to collect 3-dimensional diffraction data to 2.5 A using the EMBL beam-line X11 at’ the DORIS storage ring (DESY, Hamburg). The data were recorded at 4°C using a modified Arndt-Wonnacott oscillation camera (Nyborg & Wonnacott, 1977), employing a locally developed image plate system as detector. A total of 90” rotation data were measured, the crystal spinning around the [l lo] direction. The exposure times were from 14 to 121 s. The refinement of orientation and integration of the intensities was performed using the MOSFLM suite of programs (Leslie et aE., 1986), modified for processing of the image plate data. The processing is summarized in Table 1. Merging of the observed intensities into a unique data set of reflections was performed with the ROTAVATA/AGROVATA programs from the CCP4 suite, supplied by the SERC Daresbury Laboratory (UK). The intensities were converted to the structure factor amplitudes using the program TRUNCATE (French & Wilson, 1978). (c) Structure

determination

The full details of structure solution followed by partial refinement using molecular dynamics have been published (Djinovic et al., .1991). In brief, molecular replacement was used to solve the structure and get an initial set of phases. The search model was the “blue-green” dimer of the b-SOD (Tainer et al., 1982), deposited as data set

2SOD with the Protein Data Bank (Bernstein et al., 1977), having all the side-chains not common to the y-SOD amino acid sequence truncated to alanine. The MERLOT (Fitzgerald, 1988) package of programs, which includes the fast-rotation function of Crowther (1972), the rotation function of Lattman t Love (1970) and a translation function of Crowther & Blow (1967), was used. The calculated structure factors were determined from the model dimer placed in a cubic unit cell of Pl symmetry with cell edges of 100 A. The orientational parameters of the 2 dimers in the asymmetric unit were determined using the data in the resolution range between l@O and 45 A, with a Patterson integration radius of 269 A. The 2 peaks used, corresponding to the correct orientations of the 2 dimers in the asymmetric unit, had root-mean-square (r.m.s.) values of 633 and 506. Translation search calculations on a 1.0 A grid for the data between %O and 3.5 A for the Harker sections x = y = l/2, z = 0 gave a self-consistent set of vectors that positioned the 2 dimers of the asymmetric unit with respect to the crystallographic symmetry elements (Djinovic et al., 1991). (d) Molecular dynamics crystallographic

and conventional rejinement

Structure factors were calculated for the appropriately rotated and translated co-ordinates of the blue-green b-SOD dimers, with residues truncated as described above. The initial crystallographic R-factor (defined as: R = ~llF,l-lFJl/ZIFOl, where IF,,1and IF,1 are the observed and calculated structure factor amplitudes, respectively, and summations include the data in the specified range) in the 7.0 to 30 A resolution range was @449 for this model. In order to determine the molecular replacement solution more accurately, rigid body refinement cycles were run, starting with the data in the 60 to 10.0 A resolution range, using the program TNT (Tronrud et al., 1987), with an overall temperature factor of 2@0 A’. The 4 polypeptide chains in the asymmetric unit were refined as independent moieties, allowing 3 positional and 3 rotational degrees of freedom for each subunit. During the refinement, the resolution range was increased to 45 A and the R-factor dropped to 0.406 for the diffraction data between l@O and 45 A. The crystallographic refinement was then continued simulation package GROMOS using the molecular (van Gunsteren & Berendsen, 1987). modified by Fujinaga et al. (1989) to include the crystallographic pseudo-energy with crystallographic term. Energy minimization restraints and molecular dynamics, also including crystallographic terms, have been used in the course of the refinement. The starting model submitted to the m.d.x. refinement was constructed by proper replacement of the truncated amino acid side-chain with those indicated by the y-SOD sequence (Steinman, 1980). The starting model was not manually corrected for the poor geometry and short intra- and intermolecular contacts. A chain interruption was intentionally left at the insertion site at position 31, as indicated by the sequence alignment. Seven residues (32, 32A, 32B, 33, 34, 35 and 35A) were deleted from the model, and free C and N termini were left in the polypeptide chain at positions 31 and 36, respectively. The initial R-value for this model was 0.472 for the data between 60 and 3.0 A resolution. Energy minimization using X-ray restraints (e.m.x.) was therefore first performed (150 cycles in the resolution range from 6.0 to 3.0 A) to remove the initial strain in the structure due to the poor stereochemistry.

794

K. Djinovic et al.

m.d.x. refinement was typically performed in series of @l to @6 ps/cycle, consisting of steps of At = 2 fs. During the m.d.x. refinement, the system was coupled to a constant heat bath using the temperature relaxation time T = @l ps (Berendsen et al, 1984). The interaction parameters of GROMOS for covalent bond, angle, dihedral, improper dihedral, Lennard-Jones and electrostatic interactions for simulation in vacua were used without any alternations (van Gunsteren & Berendsen, 1987). No restraints on bond lengths were applied and B factor values of 2@0 A2 were assigned to each atom. The cut-off radius for calculating electrostatic interacations was set to &O A. The balance between crystallographic and energy terms was controlled by applying a weight that was approximately equal to the observed r.m.s. difference between IF,,/ and IF,1 during the refinement. The m.d.x. refinement was initialized at, T = 300 K with 100 cycles using the data between l@O and 50 8; the resolution was then gradually increased to 2.5 & limiting the data t’o 60 A in order to avoid the contribution of the solvent structure at lower scattering angles. Each extension of the resolution range during the m.d.x. refinement was preceded by e.m.x. energy minimization. After 67 ps of m.d.x. simulation time, followed by e.m.x. minimization (R = @299 for 60 to 25 a data) an electron density map showed the amino acid alignment suggested by Getzoff et

al. (1989) to be more consistent with the observed electron density in the insertion region 31 to 36. Therefore residues 32 to 35A were all built into the model as Ala, residues 26 to 31 of the adjacent b-strand were trimmed to their CB atoms and a break in the polypeptide chain was introduced to leave enough room for residues 24 and 25, and for the insertions 25A and 25B. The 2 alternative amino acid sequence alignments of the bovine and yeast SOD discussed above and the amino acid sequence alignment of the 2 proteins (Steinman, 1980; Getzoff et al., 1989) are presented in Table 2. Further refinement proceeded by e.m.x. minimization in the 6.0 to 3.0 a resolution range. followed by m.d.x. simulation, gradually increasing the resolution and the weight for the crystallographic terms. An electron density map calculated after an additional 36 ps of simulation time (R = 0.280 for the data between @Oand 2.5 8) clearly showed the side-chains of the remaining trimmed residues in all 4 chains as well as the backbone tracing for missing residues in 3 out of 4 monomers. The residues 24, 25, 25A and 25B of the chains A, B and D were individually built into the model, while for chain C the corresponding segment of the D subunit was adopted by rotation around the local 2-fold axis of the dimer. The refinement was continued by e.m.x./m.d.x. minimization. As the m.d.x. simulation seemed to reach the convergence after a total

Table 2 Amino A. Sequence alignments

acid sequence alignments

of bovim (B) and yeast ( Y) mpero.ridr

dismutcws

B Y

10 20 AC-ATKAVCVLKGDGPVQGTIHFEAKGD--TVVVTGSITGVQAVAVLKGDAGVSGVVKFEQASESEPTTVSYEIAGN

B B

LTEG:HGFHVHQFG&TQGCTSAG:HFNPLSKKH:GPKD SPNAERGFHIHEFGDATDGCVSAGPHFNPFKKTHGAPTD

B Y

EERHVGDLGNVTADKNGVAIVDIVDPLISL~GEYSIIGRT

13

MVVHEKPDDLGRGGNEESTKTGNAGSRLACGVIGIAK VVIHAGQDDLGKGDTEESLKTGNAGPRPACGVIGLTN

80

90

110

EVRHVGDMGNVKTDEDGVAKGSFKDSLIKLIGPTSVVGRS 120

Y

100

30

B. Two

alternative

130

140

150

of the bovine (B) and yeast ( Y) superoxide dismutases in the regim oj

alignments

th,e insertions (a) B Y

B Y

20 25 FEAKGDTVVVTGSFEQASESEPTTVST

33 -ITG-LTEGD

30

40

EIAGNSPNAE

(b) B Y B

T

20 25 26 FEAKGD--TVVVTG FEQASESEPTTVST 36 SITG-LTEGD EIAGNSPNAE

30

40

Residues are numbered according to the bovine sequence and those identical are printed in bold characters. Amino acid residues are displayed according to their l-letter code. (a) Corresponds to the alignment by Steinman (1980); (b) to the alignment by Getzoff et al. (1989).

Crystal

Structure

of Yeast Cu,Zn

Superoxide

Dismutase

795

Table 3 Summary

rejinement with stereochemical restraints TNT program suite

of the crystallographic

using the

r.m.s. Resolution

Stage 1

(A)

&nit (%I

&in (%)

No. cycles

290 28.8 26.2 20.9 18.1 18.2

22.0

52 68 34 25 24 21

60-2.5 100-2.5 100-2.5 100-2.5 10.0-2.5 10.0-2.5

2 3 4 5 6

199 17.8 16.5

159 158

of 11.9 ps of simulation time, the temperature was increased to 600 K (t = 001 ps) and after @6 ps of molecular dynamics minimization the model was cooled back to 300 K (t = @05 to @1 ps) The resulting model was submitted to e.m.x. minimization in order to release the system’s kinetic energy. This process led to an R-factor of 0268 for the intensity data between GO and 2.5 A (Djinovic et al., 1991). The refinement of this molecular model continued by conventional restrained crystallographic refinement of the co-ordinates and temperature factors using the TNT suite of programs (Tronrud et al., 1987). Electron density maps with coefficients (21F,I -IF,I) and (IF,] -I1p,I) and calculated phases were computed at regular intervals, whenever convergence to a local minimum was shown by a contained decrease in the R-factor and small r.m.s. movements of shifts of the stereochemical parameters, indicating the need for manual intervention. Table 3 gives a summary of the crystallographic refinement in terms of R values and r.m.s. co-ordinate shifts. Six stages of restrained crystallographic refinement of co-ordinates and temperature factors, alternated by model-building sessions on a graphics station, using the FRODO software (Jones, 1978), led to a final R value of 0158 for 32,088 observed structure factors in the l@O to 25 A resolution range (98.2% of all the theoretical reflections in this shell). The solvent contribution to structure factor calculation was not considered until stage 2 of the conventional refinement (see Table 3), during which the crystallographic data were extended to the lower resolution shell (l@O A). Systematic searches on the difference electron density maps calculated with coefficients (IF,1 - IF,I) and

co-ord. shift 0070

0010 0.056 0.003 0006 0.007

calculated phases at various stages of the refinement led to the inclusion of 516 water molecules in the present structure. The positions of the solvent molecules with high temperature factors and of those apparently free from hydrogen bonds were carefully examined and, for these cases, only those that appeared with electron density higher than 4 electron density standard deviations were included in the model. A plot of R-factor versus resolution is shown in Fig. 1. The profile of the curve is typical: low angle data with significant scattering from non-modelled bulk solvent and omitted hydrogen atoms show poor agreement. High resolution data, on the other hand, tend to be weak and so have larger associated error. Table 4 presents the summary of the refinement results; the small r.m.s. shift in coordinates (@007 A) indicates that the refinement had reached convergence. During the refinement, the metal-to-ligand bonds were restrained to the average Cu’+-N (1.98 A), Zn*+-N (2.02 A), Zn*+ -0 (1.95 A) bond lengths recovered from the Tables of Bond Lengths determined by X-ray and neutron diffraction (Orpen et al., 1989). Possible H-bonds were identified by calculation of the distance separating all pairs of potentially hydrogen bonding atoms. Potential hydrogen bond pairs from the initial sample were checked for distance (between donor D and acceptor A, less than 3.5 A) and linearity (angle AHD = 180” f 40”) criteria. Main-chain dihedral angles were determined by the WHATIF program package (Vriend, 1990). The secondary structure of the protein was analysed by the DSSP program (Kabsch BE Sander, 1983). Secondary

Table 4 Rejinement

parameters

and results for the final

rejbement cycle Parameter No. of cycles R (%) No. of reflections

21 158 (resol. range in A)

No. of protein atoms No. of solvent atoms r.m.s. co-ordinate shift in the

32,088 (10.0-2.5) 4432 ( + 8 metal ions)

516 om7

final cycle (A)

0.121 ’ 0.10

’

0.15

’

’

0.20

’

’

0.25

’

’

0.30

’

’

0.35

’

’

040

2 Sin B/X

Figure 1. Plot of R-factor verm resolution for the data between l@O and 25 8. Each point is the higher resolution limit for a range of 2 sin e/n. The 1st point represents data from 10.0 to 63 A.

r.m.s. B shift in the final cycle (A’) r.m.s. deviations from ideal values: Bond length (A) Bond angle (“) Torsion angle (“) Trigonal plane (A) Planar groups (A) Non-bonded contacts (A)

218

0016 2912 27.16

0019 0.020

0098

K. Djinovic et al. to 63” and N-O distance up to 5.2 A for perfect alignment. Data processing and reduction were performed on the Ethernet-based MicroVAX cluster at EMBL, Hamburg (Germany), molecular replacement and molecular dynamics calculations on a Convex C220 S computer at CILEA, Milan0 (Italy), while conventional refinement and molecular graphics were carried out on a VAX8530 system and on a PS330 Evans & Sutherland graphics computer at the University of Pavia. The Figures were prepared using the molecular graphics package WHATIF (Vriend, 1990).

-20 "0 -30 ; 2 C

-40 -50 -60 -70

0

IO

20

30

(sin B/Xl2 x IO3

3. Results

Figure 2. Plot of (In oA) versus (sin @/A)’ based on the method of Read (1986). The y-intercept (=@5 In (E[&./E,]) has the value of 0@0216, so that Zp/CN = 1604. The slope of the line is equal to -2631892 (the mean square co-ordinate error). The slope of the linear regression line is - 1.3491, and so the r.m.s. co-ordinate error is 0226 A. Dividing this value by a factor of (37~/8)“~ gives a mean co-ordinate error of 621 A.

structure recognition is based on hydrogen bonding patterns, where absence or presence of a hydrogen bond can be characterized by a single decision parameter, a cutoff in bond energy. Hydrogen bond energies up to -95 kcal/mol were assumed with the misalignment of up

and Discussion

(a) Quality of the jind

As there is no straightforward method for evaluating the co-ordinate error of a macromolecular structure refined by a restrained least-squares algorithm, the method of Read (1986) was used to calculate an estimate for the co-ordinate r.m.s. error. Fi ure 2 shows a plot of In ((T*) versus ((sin #)/A) Fi, from which a positional r.m.s. error of 0.21 il can be estimated for the refined y-SOD model. A good indicator of the stereochemical correctness of the main-chain folding of a protein is provided by the Ramachandran plot. Figure 3 displays the

+

ma

-180

model

&ldcg.l

180 -I-

180

180

Figure 3. The 4, IJ?plot for the refined model of y-SOD. The inner lines define areas of fully allowed conformations with 7 (N-Ca-C) = 110” for the non-glycine residues. The outer lines show the regions obtained by relaxing the van der Waals’ contact constants as well as by allowing 7 to increase to 115” (Ramakrishnan & Ramachandran, 1965). All residues falling in the forbidden regions of the digram (lower right quadrant) are glycine.

797

Crystal Structure of Yeast Cu,Zn Superoxide Dismutase

0

IO

20

30

40

50

50

70 Residue

80

go

100

110

I20

130

140

I50

number

Figure 4. Plot of average B-factor values for the main-chain (0) and side-chain ( - ) atoms wersus sequence number for the D subunit of the y-SOD model.

angles for the two y-SOD dimers after the refinement. The allowed regions for the non-glycine residues are indicated according to Ramakrishnan & Ramachandran (1965). p-structure and “bridge” regions are highly populated, the latter mostly representing residues in /?-turns or loop regions. The largest deviations from ideality in 4, II/, for nonglycine residues, are found for Leu124 in all four subunits, and Asn137 in chains A and C. These residues, with left cl-helical conformation, also fall outside the normally allowed regions in the model of b-SOD (Tainer et al., 1982). Individual atomic isotropic temperature factors were refined along with the atomic co-ordinates, once the refinement of positional parameters reached convergence. The average temperature factor for the whole model, including 516 water molecules, is 299 AZ. The average temperature factor for 4361 non-hydrogen atoms considered in the structure factor calculations is 27.0 A2. The average main-chain and side-chain temperature factors per residue for the subunit D are depicted in Figure 4. The residues with the largest B-factor values correspond mainly to three regions in the protein molecule: (1) the 24 to 25B b-turn, accommodating two out of the three amino acid insertions (with respect to the bovine enzyme), (2) residues of the Zn-ligand region in the 6,5 loop (His61 to Met82: Getzoff et al., 1989), except for the histidine residues binding to the Zn ion, (3) residues 124 to 138 of the 7,8 loop that lie around the rim of the active site channel. This observation is in agreement with the normal mode calculations on b-SOD (Fisher et aZ., 1990), which show large atomic displacements from equilibrium position for the residues Glu130, Glu131 and Lys134. These atomic movements are attributed to an opening-andclosing movement around the active site region due to the swinging of the hanging loop containing the residues that are believed to be important for the electrostatic guidance of the substrate into the active site. Moreover, the overall temperature factor

profile observed for y-SOD, although on average displaced towards higher B values, follows closely that found in b-SOD; in particular, one of the loops displaying the highest B factor values is the 7,8 loop (residues 124 to 129) in both proteins, suggesting a functional role for this segmental flexibility. While the electron density is well-defined almost throughout the whole asymmetric unit, some of the larger side-chains on the surface of the model show weak electron density and cannot be fitted properly to it. These regions are listed in Table 5. In order to assess the local deviations of the y-SOD tertiary structure from a common scaffolding, the best fit molecular overlays of the two y-SOD dimers within the asymmetric unit have been calculated. When the C” backbone of the AB dimer is overlayed onto the CD dimer backbone, a r.m.s. deviation of @446 a is obtained. The residues showing largest deviations within either A or B chains can be grouped in three areas which include the 24 to 25B insertion loop (largest deviation 1.17 A at residue 24 of the B chain), the 6,5 loop (largest deviation 1.41 A at residue 53 of the A chain), the active site rim 128 to 130 (largest deviation 1.43 A at residue 128 of the B chain), and the C terminus (with a deviation of 4.10 A at residue 151 of the B chain). Table 6 lists the subunit-to-subunit pairwise C” backbone comparison for the chains A through D for the y-SOD structure. A detailed analysis of the residues that deviate most in these pairwise monomer comparisons confirms the location of the largest backbone structural deviations in the loops listed above. Figure 5 shows the difference in C’ positions after superposition of the fully refined CD dimer of the y-SOD model and the initial search model, the “blue-green” dimer of the b-SOD (r.m.s. deviation is @977 A). The largest atomic displacements in all four chains of y-SOD with respect to the starting model can be observed in the tight turn between the strands 2b and 3c, accommodating two out of the three insertions, and in the non-repetitive structure

798

K. Djirwvic

et al.

Table 5 Residues with poorly dejked side-chains Residue

Atoms

Residue

A. Chain A Gln3 Ala23 Ser25A Glu75 Asp128 Thr129 Glu130 Lys134

B. Chain B OEl, NE2 CB CB, CG CD, OEl, OE2 N, CB, CG, ODl, OD2 CA, CG2, CB, OGl ODl, OEl CD, CE, NZ

Lysl9 Glu21 Glu25 Lys66 Glu75 Glu89 Asp96 Lys98

NZ CD CE, NZ 0, CB, CG, NDl, N, CG2 CG

Lysl9 Ser24 Glu25B Glu75 Lys86 Glu130

C. Chain C Lys66 Lys86 Lys126 Asp128 Thr129 Glu130

Atoms

D. Chain

OD2

CD, CE, NZ CD, OEl, OE2 CG CE, NZ CG on the border of density OEl, OE2, CB, CG, CD CB NZ D CD OG OEl CB, CD, CD,

on the border of density CG, CD, OEl, OE2 CE, NZ OE2

Table 6

forming the Greek-key connection between strands 6d and 3c (residues 36 to 38), where the third amino acid insertion is placed. Other regions containing residues that moved more than 2 A during the refinement are: (1) residues 88 to 89 of the /?-turn connecting strands 5e and 4f; (2) residues 95 to 97 of the p-strand 4f; (3) residues 105 to 107 of the two tight turns (106 to 113) in the 4,7 loop; (4) residues 128 to 130 in the active site rim, before the sixresidue a-helix.

Pairwise deviations among the four subunits in an asymmetric unit (A)

A B C D

A

B

c

D

I 0.399 0356 0.310

0254 I 0336 0429

0.228 0225 / 6340

6198 0262 0231 I

(b) Subunit interactions

The upper right values of the Table show the distance deviations of C” atoms; the lower left those for all non-hydrogen atoms in protein.

The two y-SOD dimers in the asymmetric unit (Fig. 6) are contacting each other through only four polar interactions as listed in Table 7A. It can be seen that subunit D is hydrogen bonded to both molecules of the AB dimer, while subunit C does not contribute to the packing interaction within the asymmetric unit. The residues involved in the dimer-dimer hydrogen bonding network are placed in three regions of the molecule: (1) in the b-turn accommodating two out of the three insertions that extend the 24 to 25B B-turn and therefore facilitate the intermolecular interaction; (2) in the a-helix (130 to 135); (3) in the Znl p-turn (63 to 66)

of the Zn ligand region of the 6,5 loop, the latter contact being between main-chain atoms. The hydrogen bonds stabilizing the threedimensional structure of the crystal (i.e. between molecules related by crystallographic symmetry) involve main-chain-main-chain (2), side-chain-sidechain (19) and side-chain-main-chain (1) interactions with their symmetry-related partners in different asymmetric units (see Table 7B). From an overall inspection of the crystal packing, it appears that no contacts between different asymmetric units

Monomer

0

C

Monomer D

IO 20 30 40 50 60 70 so 90 loo 110 I20 130 140 1502002,0220

230

240

250

260

270

m290”o

310

320

330

340

350

Residue number

Figure 5. The difference of y-SOD.

(A) in C” positions

between

the initial

blue-green

b-SOD

model and the fully

refined

CD dimer

Crystal Structure of Yeast Cu,Zn Superoxide Dismutase

Figure 6. Stereoscopic view of the C” atom tracing of the asymmetric chains are labelled according to the convention adopted in this paper.

are established among the same functional dimers; so neither A nor B (C or D) chains are in contact with symmetry-related A and/or B (C and/or D) chains. The C” positions of the residues involved in main-chain-main-chain hydrogen bonded contacts (Aspll, Gly13, Lys66 and Glu130) are little affected by these interactions even if the comparison is made with the initial b-SOD model (maximal r.m.s. deviation is 1.68 A at residue Gly13 of the chain B). On the other hand, inspection of the regions containing residues contributing to the side-chain-side-chain contacts shows that some of the hydrogen bonding

Table 7 Contacts in y-SOD Residue

Atom

Chain

Residue

Atom

A. Dimer-dimer contacts in the asymmetric unit OEl A Lys134 NZ Glu25 Lys134 NZ OE2 A Glu25 OEl OE2 A Glu130 Glu25 Pro64 0 B Glu25 N B. Protein-protein Asp11 Lysl9 Glu21 Glu21 Glu32 Glu32 Glu89 Lys94 Lys94 Ser96 Asn151 car151 Asp50 Asp53 Ser57 Ser57 Glu130 Glu130 Glu131 Lys134 Arg141

0 NZ OEl OEl OEl OEl OEl NZ NZ OG ND2 OH ODl OD2 OG OG N OE2 OEl NZ NH2

crystal packing contacts N A Gly13 A Lysl9 NZ OEl Glu21 A Glu21 OE2 A OD2 A Asp53 Set-57 OG A NH2 Arg41 A A Asp50 ODl OGl A Thrl50 ND2 A Asn151 OG A Ser96 OG Ser96 A NZ B Lys94 OEl B Glu32 OE2 B Glu32 OEl B Glu32 0 B Lys66 B Lys67 NZ ND2 Asn38 B ODl B Asn38 OD2 B Glu89

Chain

Disk (4

D D D D

304 241 280 300

C D D D C C C C D D D D D D D D C C D D D

347 317 342 2.91 2.89 259 333 250 328 308 352 303 287 285 344 2.50 3.19 281 329 349 349

unit of the y-SOD

799

containing

2 dimers. The 4

regions have been displaced considerably ( > 2 A) during the refinement, as compared to the bovine enzyme, and differ also within the same asymmetric unit. The superposition of the C” backbones of the bovine “blue” subunit and yeast D subunit (Fig. 7) shows that the B-sheet containing strands la, 2b, 3c and 4f has been shifted towards the C terminus; such a movement might be due to the two-residue insertion in the hairpin turn between strands 2b and 3c, resulting in a different orientation of the turn, stabilized by intermolecular hydrogen bonds with, the residues of this b-sheet. Sixteen amino acid residues contribute to the interactions between the two subunits of the functional dimer (Fig. 8), the majority of them being highly conserved through the evolutionary phyla (Getzoff et al., 1989). The intersubunit apolar interactions involve residues Va15, Va17, Ile147, Gly148 and Leu149. These residues, together with those provided by the other subunit, build up a hydraphobic

field around

the local 2-fold axis relating

the

two subunits of the dimer. The polar interactions, on the other hand, involve the backbone of evolutionary invariant residues Gly49 and Gly112, and the variable residue Leu149 (Ile or Thr in other known SOD amino acid sequences), that form four tight main-chain hydrogen bonds across the dimer interface (Table 8). Two additional side-chain-sidechain hydrogen bonds of %73 and 2.59 A stabilizing the subunit-to-subunit interactions are formed,

Table 8 Subunit-to-subunit

contacts as observed in the CD dimer

Residue

Atom

Chain

Residue

Atom

Chain

Dist. (4

Glu49 Glu112 Leu149 Leul49

N 0 0 N

D D D D

Leu149 Leu149 Gly49 Gly112

0 N N 0

C C C C

271 F54 2.60 282

800

K. Djinovic

Figure 7. Stereoscopic view of the c” atoms of y-SOD subunit

(broken

line).

Labels

Figure 8. Stereoscopic view crystallographic dyad axis.

have been placed

of the C” atom

D subunit (continuous 15th C” atom.

on every

tracing

et al.

of the y-SOD

between Asp50 ODl and Thr150 OGl atoms; ThrlBO substitutes for Ala150 of the bovine enzyme. This pattern of interactions between the two polypeptide chains of the functional SOD molecule is also present in the s-SOD enzyme (Kitagawa et al., 1991).

line) superimposed

CD dimer

on the blue b-SOD

as seen perpendicular

to the pseudo

b-strands and in p-turns or loop regions. Some of the main-chain to side-chain and side-chain to sidechain hydrogen bonds play an important role in the Table 9 structure assignment for the D subunit of y-SOD as dejked by the DSSP program packoqe

Secondary

(c) Monomer fold and hydrogen bonds Table 9t shows the results of the secondary structure assignments for the D monomer as provided by the DSSP program (Kabsch & Sander, 1983). The overall agreement for the structural assignments of the yeast and bovine SODS is quite satisfactory. Nevertheless, minor differences in the actual lengths of B-strands or in the type of turns are scattered throughout the four y-SOD monomers. Similar considerations apply to the definition of the short helical region around residues 130 to 135 whose regularity varies in the four monomers. Table 10 reports the main-chain to main-chain, side-chain to main-chain, and side-chain to sidechain hydrogen bonds as observed in subunit D of y-SOD. Intermolecular hydrogen bonds between polypeptide chains within the asymmetric unit are listed in Tables 7 and 8. The intramolecular mainchain to main-chain hydrogen bonds are mostly involved in connecting residues of adjacent t The numbering according to Tainer

of the individual et al. (1982).

b-strands

is

Structural element

Name

cr-Helix

/I-Strand /i-Strand /?-Strand fiStrand $-Strand fi-Strand /?-Strand /kStrand D-Turn B-Turn B-Turn /l-Turn B-Turn B-Turn B-Turn P-Turn Loop Loop Loop Loop

Residues 139135

la 2b 3c 4f 5e 6d 7g 8h 2-3 Znl Zn2 Zn3 5-4 Gl G2 HeI 4,7 635 6,5 7,8

3-8 15-21 27-35 93-99 81-86 4046 114-117 143-148 24-25B 63-66 71-74 78-81 88-91 106109 110-113 123-126 100-l 12 47-60 61-82 119-142

Comment Extends to residues 129 and 136 to 138 with B-turns Big

Small

Disulphide region Zn-ligand region

Crystal

Structure

of

Table 10 (continued)

Table 10 Intramolecular

Residue

Atom

A. Main-chain Ve12 Gln3 Ala4 Ala4 va15 Va15 Ala6 Ala6 Ieu8 Leu8 Lys9 Lys9 Serl5 Serl5 V&117 V&117 Lysl9 Lysl9 Glu21 Glu21 Thr27 Thr27 V&b129 va129 Tyrfl Tyr31 Ile33 Ile33 Asn35A Asn38 Ala39 Argll Arg41 Gly42 Gly42 Phe43 His44 His44 Ile45 Ile45 His46 His46 Glu47 Phe48 Phe48 Asp53 cys55 Va156 Phe62 Asn63 His69 Ala71 His78 Asp88 Asp88 LyslO3 Gly106 V&l1 10 Gly112 va1115 Be117 Ala119 Asp123 Asp123 Thr129 Glu131 Gly136

hydrogen bonds observed in the D subunit of y-SOD Residue

Atom

801

Yeast Cu,Zn Superoxide Dismutase

Dist. (A)

to m&n-chain hydrogen bmda 316 Gln22 0 N 308 Thrl59 0 N 286 N Phe20 301 0 ii Phe20 257 Gly148 N 0 2.68 N Gly148 0 294 N 0 Va118 290 Va118 0 N 276 N 0 Gly16 328 0 N Glyl6 N 0 308 cys144 316 0 N cys144 288 N 0 Ala34 259 0 N Ala34 2.86 N 0 Glu32 328 0 N Glu32 N 273 0 Ser30 299 0 N 2.89 Thr28 N 0 285 0 N Thr28 N 0 305 Asp99 2.92 0 N Asp99 273 Phe97 N 0 279 0 N Phe97 283 N 0 Gly95 0 300 Gly95 N 274 N 0 Ala93 2.81 0 N Ala93 N 327 0 Gly91 N 0 2.74 Asp88 0 N 251 Thr87 0 307 V&185 N 293 0 N Va185 265 0 Gly120 N 0 255 His118 N 285 N 0 Gly83 310 Valll6 N 0 0 278 Va1116 N N 0 2.85 Gly80 0 N Phe62 296 278 Serll4 0 N Ser114 0 N 342 pro60 287 N 0 Ala58 N 0 306 Asp56 0 N 335 Ser57 0 N 310 Ala58 0 N 324 Gly59 0 N 323 Gly86 0 N 316 Lys66 0 N 305 0 276 Leu133 N Asp74 272 0 N Asp81 N 315 0 Va192 0 N 295 Gly91 N 0 290 ValllO N 0 2.77 Serl69 0 N 317 294 Argl13 N Ile147 0 286 ii Gly145 0 292 N Ala143 0 304 N Pro140 0 N 310 0 2.76 Gly136 N Lys126 N 0 340 Leu133 N 348 0 Thr135 N 0 290 Ser132 0 276 N

Angle (“)

95 21.0 186 187 322 84 264 88 181 37.1 17.7 294 184 203 294 186 147 lo-4 388 198 17.9 368 14.3 349 251 222 21.4 2w9 286 22.5 11.2 173 221 231 224 27.3 48 87 154 45 34.2 167 251 87 359 330 175 161 165 83 129 54 17.5 155 159 31.8 180 333 285 11.8 85 22.8 32 336 27.3 91 468

Residue

Atom

Residue

Atom

B. Side-chain to main-chain hydrogen bade Leul94 N OEl Ile105 N OEl LyslO3 NZ 0 Ile102 0 OGl 0 Gly91 ND2 Thr87 OGl 0 OGl Thr87 N 0 ND1 Gly59 Argll3 NH1 0 OG Ser114 0 Ser57 OG 0 Thr52 N OD2 OG Ser57 N Argl41 NH1 0 0 OG Gly54 0 Lys67 ODl N Lys67 ND2 Asn63 N Lys67 ND2 Asn63 His78 NE2 0 Lys67 As~81 OD2 N Gly70 ODl N Ala71 His78 0 NE Arg77 Va179 0 NH2 Arg77 Asp81 OD2 N His78 Asp122 N ODl Asn84 0 Asp122 ND2 Asn84 N Asp96 ODl Asp88 N ODl Va192 Asp88 OEl Cm89 N Glu89 0 ODl Asp96 Glu89 OG 0 Ser96 Gly95 N LeulOl ODl Asp99 N Ile102 ODl Asp99 0 NE Argl13 SerlO9 0 Leul64 OG &xl69 N OG Vallll SerlO9 0 Pro140 ND1 His118 N Leu124 ODl Asp122 N Gly125 ODl Asp123 NZ Lys126 0 Leul24 N OG Ser132 Gly127 N OGl Glu131 Thr129 0 Thr135 OGl Glu131 OEl OGl Thr135 Glu131

Gln22 Gln22 Glu25 Thr27 Asn35A Ser36 Ala39 His46 Glu47 Phe48 Asp56 Asp56 Asp53 cys55 Ser57

Did. (A)

Angle (“)

2,51 2.96 343 245 2.93 249 2.78 2.54 2.79 2.76 253 274 2.80 2.72 340 281 285 285 316 283 2.87 302 272 2.99 279 302 2.78 278 304 347 301 268 282 315 2.73 320 2.70 2.94 258 279 289 305 328 2.48

21.4 356 287 191 258 261 37.1 264 292 294 31.6 140 242 391 11.3 167 39.2 199 18.2 17.0 20.1 21.1 385 17.9 4.5 22.7 38.2 25.0 222 16.8 1.9 33.0 256 152 390 26.1 405 21.7 327 390 369 142 182 122

329 275 2.97 313 336 243 331 326 270 2.96 2.81 336 2.77 301

266 22.1 97 190 21.4 8.9 398 37.2 164 33 285 299 4.0 38

C. Side-chain to side-chain hydrogen bonds Gln3 Glu32 His44 His46 Asp50 Thr52 His61 His61 His69 His69 Arg77 Arg77 Asp123 Asp123

OEl OE2 NE2 NE2 OD2 OGl ND1 ND1 NE2 NE2 NH1 NH2 ODl OD2

Glu21 Lys94 Asp122 His61 Ser57 Asp50 His69 His78 Asp122 Asp122 Asp99 Asp99 Ser132 Ser132

OEl NZ OD2 NE2 OG OD2 ND1 ND1 ODl OD2 OD2 ODl OG OG

structural stabilization of the active site region (H&44-Asp1 22, GlyB%His46 and His 11%Pro140) as well as in the overall rigidity of the tertiary structure of the subunit (Gln22-Leu104, Thr27Asn35A-Gly91, Ala39-Thr87, Glu47Ile102, Argl13, Phe48-Ser114, Cys55-Arg144 and Arg77Asp99) and will be discussed elsewhere.

802

K. Djinovic

(d) Salt and disulphide

bridges

The intramolecular salt bridges present in the refined structure of b-SOD involve three residues that are mostly conserved in the SOD sequences. Thus, the ion pair Arg77-Asp99 is present in the y-SOD enzyme and, as in b-SOD, connects the /?-barrel to the 65 loop containing the zinc ligands. Residue Arg126 of b-SOD is Lys in y-SOD, allowing the establishment of the Asp74-Lys126 ion pair between loops 65 and 7,8, respectively. The presence (and the stereochemistry) of the additional salt link connecting Glu130 to Lys134 is affected by crystal contacts involving both side-chains (see Table 7B). At the monomer-to-monomer interface of the functional dimer, formation of the ArgllSGlu107 ion pair is prevented by the GlulO7-+Pro substitution. The side-chain of Argl13 can, however, reach the carbonyl oxygen atom of Glu47, in the same subunit, yielding a strong intramolecular hydrogen bond (2.79 A), and can hydrogen bond through ordered water molecules to the peptide backbone at residues Pro107 and SerlO9. Two amino acid substitutions in y-SOD as compared to the b-SOD (Ser32+Glu and Ile94+Lys) allow the establishment of an ion pair between p-strands 3c and 4f. This occurs properly only in subunits A and D of y-SOD, because of favourable crystal contacts that set the mutual orientation of the two charged side-chains. The b-SOD dimer is stabilized by an additional salt link between Asp50 and Lysl51. In y-SOD residue 151 is Asn, preventing the formation of the same ion pair as in b-SOD; however, a productive contact is formed, in a structurally equivalent manner, taking advantage of the Alal50+Thr substitution. Thus the C-terminal segment of the y-SOD subunit is stabilized by strong hydrogen bonds between Asp50 ODl and Thrl50 OGl atoms and between two polypeptide chains, as discussed above. The two Cys residues in the y-SOD structure are at positions 55 (6,5 loop) and 144 (Sh B-strand), respectively, and form a covalent bond of 2.04 A. The disulphide bridge shows a left-handed chirality, with torsion angles within the expected ranges for the left-handed spiral as defined by Thornton (1981). The side-chain dihedral angles, starting from Cys55, are: x1 = -58”. xZ = -54”. x3 = -87”, x; = -94”, x; = -63”, with the c” to C” distance of 5.83 A. The x1 angles do not deviate much from the ideal gauche+ conformation, while xZ values differ more from the predominant conformation of -80”. The solvent accessibility of the Cys side-chains range from 12 to 18 A2 in the four subunits. (e) Structural

details at the substituted acid sites

amino

In y-SOD, 22 amino acid substitutions alter the nature of the exposed electrostatic charges when compared to b-SOD; however, because of the compensation of positively and negatively charged

et al.

residues, the overall change in the electrostatic charge of the protein corresponds to the loss of one positive charge, as compared with the bovine enzyme. This is in qualitative agreement with the lower isoelectric point of y-SOD (46: O’Neill et al., 1988), as compared to the b-SOD (5.2: Civalleri et al., 1982). In this framework, residues De94 and Va198 are substitued by Lys in y-SOD, probably affecting the chain conformation of strand 4f, which shows a substantial deviation from the conformation observed for the same strand of b-SOD. Similarly, two substitutions occur in the neighand boring strand 5e, which hosts Thr86-+Lys Lys89+Glu substitutions. The Ile94+Lys substitution mentioned above finds cornpension in the Ser32+Glu substitution in strand 3c; a hydrogen bond connects atoms Lys94 NZ and Glu32 OE2 (2.75 A). A sterically compensating substitution is observed at residues 31 and 95, which face each other in strands 3c and 4f, respectively. The substiand Va195+Gly allow Gly31 -+Tyr tutions accommodation of the bulky side-chain of Tyr31 in the yeast enzyme. This substitution may be at the basis of the large deviation’of residue Gly95 in the C” backbone comparison with the bovine SOD. Position 31 is usually occupied by Gly in superoxide dismutases. It contains Tyr only in the Neurospora crassa enzyme (Getzoff et al., 1989) which, accordingly, as observed in y-SOD, shows Gly at position 95. The SOD B-barrel is closed at two opposite ends by polar and apolar interactions that prevent the access of solvent molecules to the inner core of the protein. At the lower end (see Fig. 9), residue LeulO4, which is highly conserved throughout, the SOD sequences, provides a bulky hydrophobic “plug” to the B-barrel. Because of the Ala22+Gln substitution and the insertion of residue Glu25B. two strong hydrogen bonds (Gln22 OEI-Leu104 N. 2.51 A; Gln22 NE2-Glu25B 0, 2.51 A) additionally contribute to the stabilization of this lower-end molecular lid. At the other “pole” of the b-barrel. the non-polar amino acid residue Leu36 is substituted with Ser in y-SOD. The three concerted substitutions at positions Ala87-+Thr, His41 -+Arg and Glu 119-+Ala allow t)he formation of a hydrogen bonding network (Fig. 9) which, in the absence of the bulky Leu36 side-chain, efficiently closes this end of the b-barrel (Thr87 OGSer36 0. 2.51 A; Ser36 OG--Arg41 NHl, 2.96 A; Arg41 NEAla1 19 0, 3.34 A). Similar amino acid substitutions at these positions (Ser36, Thr87, TAys41 and AlallY) present in the N. crassa enzyme (Getzoff et al.. 1989) imply an analogous scheme of contacts aching as a physical “lid” to this end of the P-barrel. As anticipated by Getzoff et al. (1989) on the basis of the sequence alignments, the major amino acid insertion site occurs between strands 2b and 3c of the B-barrel. As a result of the insertion of Ser25A and Glu25B in y-SOD, the struct#ure of the 2,3 B-turn hosting the inserted residues is altered substantially and perturbations are particularly evident in both the neighboring /?-strands (residues

Crystal Structure of Yeast Cu,Zn Superoxide Dismutase

803

Figure 9. Stereoscopic view of the C” atom trace of the D subunit, along with the positions of the apolar residue Leu104, closing the b-barrel at the lower end and the amino acid residues Ser36, Arg41, Thr87 and Ala119 forming a polar lid to the upper end of the p-barrel.

15 to 21 of b-strand 2b and residues 27 to 35 of b-strand 3~). The acceptance of different kinds of amino acid residues at this site is probably justified by both the solvent exposure of the loop and by its relative distance from the protein structure defining the active site. In this respect, it is remarkable that sea turtle (Caretta caretta) SOD shows a 12 amino acid insertion at the five residue turn linking b-strands la and 2b (Schinini et al., 1991). In y-SOD, part of the structural perturbation observed at /?-turn 2,3 may be ascribed to an intramolecular contact (dimer-to-dimer) occurring within the asymmetric unit. Residue Glu25 of chain A participates in an ion pair with Lys134 of the D chain, stabilizing the crystal lattice in this crystalline form (see Table 7). The second insertion site occurs at the end of strand 3c in the wide Greek-key loop. The insertion of Asn35A, which occurs only in yeast and N. crassa SODS, has an evident effect on the structure of the loop and can possibly be considered as the origin of structural perturbations observed in the neighboring 5,4 b-turn involving residues 88 to 91, which is of type I (Wilmot & Thornton, 1990) in y-SOD, while in b-SOD, b-strands 5e and 4f are linked by a five residue turn. Gly91 0 from this turn is connected by a strong hydrogen bond (2.87 A) to Asn35A ND2. (f) Active site The overall structural organization of the y-SOD active site is displayed in Figure 10(a). As expected, and anticipated by us (Djinovic et al., 1991), the stereochemistry of the two metal centres is substantially conserved, with respect to the b-SOD refined structure. The Cu2+-Zn2+ distance is 61 At, similar to that reported by Kitagawa et al. (1991) for S-SOD, but 92 A shorter than the distance 7 All bond distances hereafter listed are compared with the y-SOD monomer D.

observed in the 2-O A refined structure of b-SOD. Crystallographic refinement of the occupancies of CU2+ and Zn2+ indicates that both sites are fully occupied as opposed to partial occupancy of the zinc site proposed by Dunbar et al. (1984a,b) for the b-SOD dimer in solution. The copper co-ordination bonds to the ligands (ND1 His44, NE2 His46 NE2 His61 and NE2 Hisll8) in the four polypeptide chains average at 2.1 A, whereas the zinc-to-ligand distances average at 2.0 A for the three histidine residues (61, 69 and 78) co-ordinating the metal ion by their ND1 atoms, and 1.9 A for the aspartate residue (Asp81 ), and can be considered to be identical within the experimental error with the restrained metal-to-ligand bond distances used during the refinement. As observed in b-SOD (Tainer et aZ., 1982) and in s-SOD (Kitagawa et al., 1991), in this oxidation state of the copper ion, the bridging ligand His61 defines a plane that almost contains both the metal ions. The deviations of zinc and copper from the least-squares plane defined by the imidazole ring of His61 occur in the same direction towards the inner core of the molecule, and are 642 and 950 A, respectively. The zinc co-ordination geometry is distorted tetrahedral, whereas copper is in dist)orted square planar co-ordination. The copper ion is almost exactly contained in the least-squares plane defined by its four ligand atoms, deviating 006 A from it in the direction of the fifth ligand. A solvent peak compatible with a fifth isolated ligand is observed 2-O A away from the copper in the active site channel, where it is almost completely buried by the side-chains of Thr135 and copper ligands His44, His46 and His1 18. This fifth ligand is thus at the vertex of the distorted square pyramidal co-ordination shell, whose base is centred on the copper ion. The proposed entrance to the active site channel is defined by two rims of structure (polypeptide loops Va156 to Asn63 and Lys134 to Arg141). These deviations as loops, which show significant compared with the b-SOD structure (see Fig. 10(b)

804

K. Djinovic

et al.

(b)

Figure 10. (a) Stereoscopic drawing showing the geometry of the active site metals and their ligands as seen from the solvent. (b) A comparison of the rims of the substrate channels in b-SOD (continuous line) and y-SOD (broken line). The positions of the metal ions are indicated by crosses on the left (Cu’+) and right (Zn”) of His61.

and below), host residues that are relevant for the electrostatic guidance of the substrate, and accordingly show only limited amino acid substitutions. In particular, the Serl40+Pro and Leul42+Pro replacements occur around the highly . conserved Arg141 residue, without affecting in a significant manner the positioning of this catalytic residue with respect to the superoxide anion binding site near the copper. The distances between the Cu2+ and the NH2 and NE Arg141 atoms, believed to stabilize the bound superoxide ion, are 5.77 and 571 A, respectively. Moreover, residue Thr135 is also conserved in its position with a OGl-Cu2+ distance of 6*75 A. In the active site region the side-chain of residue Asp122, as observed in b-SOD, acts as a bridge between His44 (copper ligand) and His69 (zinc ligand). Of the two carboxylate oxygen atoms, ODl 122 finds a remarkable location among the three hydrogen bond donors His69 NE2 (2.70 A), and peptide nitrogen atoms of residues Asp123 (315 A) and Leu124 (2.948). On the other hand, the ODl 122 atom is at 2.97 A from the NE2 atom of copper ligand His44, but also at 296 A from NE2 of zinc ligand His69 (see Fig. 11). A totally buried water molecule, OH 1231, is at 2.69 A from OD2 122 and connects this residue, through two additional strong hydrogen bonds, to the peptide nitrogen atom of residue Gly83 and to the carbonyl oxygen atom of Gly70.

The active site cleft is filled by ordered water molecules that possibly participate in the stability and conformation of the protein structure. In this region, an almost ideal network of hydrogen bonded water molecules stretches from the copper ion fifth ligand (OH 1964 in the subunit D) along the molecular space comprised between the disulphide region of the 6,5 loop, to the a-helix and the 138 to 141 region. Water molecules from this network show well-defined electron density and, besides being mutually connected, are hydrogen bonded to mainchain and side-chain atoms from these regions. Among these, water molecules 1514 and 1515 are mostly responsible for the inaccessibility of the active site centre, together with the side-chain of the Arg141. The orientation of the side-chain of Arg141 is defined by strong hydrogen bonds between atoms Arg141 NH1 and Cys550, and Arg141 NH2 and Gly59 0. As shown in Figure 12, this conformation of the side-chain allows it to hydrogen bond to OH 1515 and, through this, a polar loop is completed, hydrogen bonding OH 1514 to the Gly139 0 atom; on the other hand, the copper fifth ligand (OH 1064) is hydrogen bonded to OH 1514 (see Fig. 12). Another water molecule (1359 OH) that plays a structural role is located as a hydrogen bond bridge between atoms His61 ND1 (the Cu2+-Zn2+ bridging residue in the 6,5 loop), and Lys134 0, which is located in the a-helix. Similarly, in the proximity of the zinc site water

805

Crystal Structure of Yeast Cu,Zn Superoxide Dismutase

Figure 11. Stereoscopic view of Asp122, hydrogen bonding the His44 (copper ligand) and His69 (zinc ligand) as well as OH 1231 that is connected through hydrogen bonds to Gly83 N and Gly70 0.

Figure 12. Stereoscopic drawing of Arg141, forming hydrogen bonds to Cys55 0 and Gly59 0. The positions of OH 1515, hydrogen bonding Arg141 NH2, and of OH 1514, making a polar interaction with the Gly139 0 and the copper fifth ligand OH 1004 are also indicated. .

molecule OH 1208 allows connection of the NE2 atom of the zinc ligand His78 to atoms Lys67 0 (of the zinc ligand region in the 6,5 loop) and Leu133 0 from the helical region.

(g) Electrostatics The diffusion of the charged substrate molecule to the active site of an enzyme is affected by electrostatic interactions between substrate and charged amino acid residues. Two types of effects can be observed: enzyme charge distribution can influence the non-specific collision rate for substrate encounters with any part of the enzyme surface, or it may steer the substrate to a particular area of the that an enzyme surface, altering the probability enzyme-substrate complex will result from collision (Sines et al., 1990). The influence of individual charged amino acid residues on collision rate and electrostatic steering of the superoxide radical towards

the active

site of SOD has been studied

in

order to understand the role of electrostatics in the overall catalytical process of the enzyme (Sines et al., 1996; Desideri et al., 1989, 1992).

It

has recently

been

observed

that

SODS with

fairly different protein net charge show the same catalytic efficiency and the same pattern of pH and ionic strength dependence (O’Neill et al., 1988; Capo et al., 1990). An indication of electrostatic steering of the substrate is seen in the unusual dependence of the reaction rate on the ionic strength. The rate of superoxide dismutation decreases with increasing salt concentrations, while a simple diffusion model predicts the opposite behaviour for the encounter of like-charged ions, since increased ion screening should decrease electrostatic repulsion. In contrast to reaction rates, collision rates exhibit a positive ionic strength dependence, characteristic of simple encounter of like-charged species. The electrostatic steering of the superoxide towards the catalytic copper cofactor has been attributed to charged residues placed around the active site region. Arg141 is invariant in naturally occurring SOD enzymes (Bannister et al., 1987) and is believed to be mostly responsible for local orienting effects, while the highly conserved residues Lys134 and Glu131 seem to have important roles in directing the long-range approach of the superoxide radical

K. Djinovic which, travelling in the direction of the active site, passes near the positively charged side-chain of Lys134; the non-productive association of 0; with Lys134 being reduced by the adjacent negatively charged carboxylate group of Glu131. This position of the opposite charged residues helps to direct the field vectors into the active site channel. The charged residues at positions 119 and 120 (Glu and Lys in b-SOD), have been conserved less strictly during evolution. The changes always occur in conjunction and the residue pair is therefore considered to have a small joint effect (Getzoff et al.; 1983; Sines et aE., 1996). Concerning the conservation of the electrostatic charges in the active site region, significant substitutions are observed in the y-SOD 7,8 loop. Close to the end of the 7g b-strand, residue Glull9 is subs% tuted by Ala. Next, Lysl20 (b-SOD) is Gly in y-SOD, starting the non-repetitive structure of the 7,s loop. The other charged residues, namely Glu130, Glu131, Lys134 and Arg141, are fairly conserved in their positions. Some of the charged residues participating in the active site region show different conformations and interactions in the four chains present in the asymmetric unit of y-SOD, mainly due to the crystal contacts between the dimers. This is particularly true for the residues Glu130 and Lys134, which are involved in intermolecular salt bridges, polar interactions mediated by water molecules, or are disordered according to the y-SOD subunit considered.

(h) Solvent structure A number of partially buried water molecules stabilize the tertiary structure of y-SOD through internal hydrogen bonds. Among these water molecules OH 1038 and OH 1039 (being connected to each other by a hydrogen bond of 3.04 A) connect the /?-strands la and 2b in the region of the loop between residues 10 and 14 (see Table 11). Another structural water molecule (OH 1089) occurs at the second insertion site, connecting the backbone of residue Gly35 with the 5,4 p-turn. The conformation of both these y-SOD regions deviates significantly from the reference structure of b-SOD (Fig. 5). Water molecule 1216 is a structural feature that is conserved in y-SOD as well as in b- and s-SOD’s (McRee et al., 1990; Kitagawa et al., 1991). It is almost completely buried and connects directly the first and the last b-strands of the polypeptide chain through direct hydrogen bonds to backbone atoms. In the case of y-SOD, it is additionally connected to other water molecules (OH 1258 and OH 1262), forming an almost ideal tetrahedron (see Table 11 and Fig. 13). Finally, a relevant structural role may be played by water molecule 1359 OH, which is bridging the polypeptide backbones of residues His61 (Cu 2i-Zn2+ bridging lig and), and Lys134 of the six-membered a-helix, which is also one of the key residues for the electrostatic attraction of the substrate towards the active site.

et al. Table 11 Water molecules stabilizing the structure of the D subunit in the y-SOD structure Water molecule 1038

1039

1089

1216

1359

Residue

Atom

Distance (A)

GlylO Gly13 8011039 GlylO GlylO Va114 so1103t3 Gly35 Asn90 so11090 Va17 Va1146 5011258 So11262 His61 Lys134 so11150 Sol1 358

0 N OH N 0 0 OH N 0 OH N 0 OH OH N 0 OH OH

3.15 2.48 3.04 308 3.04 :+13 3.04 3.0 1 2.52 2.93 %90 2.74 3.05 2.48 2.93 3.0 I :P‘?1 2.81

(i) Thermal stability of the protein Concerning the protein thermal stability, one has to distinguish between the rate of thermal inactivation and the intrinsic stability of the protein folded form, the latter being reflected by different values of the melting temperature (t,). It has been shown for a number of model proteins that the inactivation processes are primarily related to the presence of free thiol groups, capable of affecting the incorrect fold and/or aggregation of protein molecules (Matthews et al., 1987; Perry & Wetzel, 1987). In this respect, a recent study on a mutant of b-SOD has shown that replacement of residue Cys6 with Ala decreases the rate of thermal inactivation by a factor of 2 (McRee et al., 1990). On the other hand, the establishment of inner cavities in the protein core, resulting from sterically non-compensating amino acid substitutions can decrease the stability of folded protein. as shown by the lower t, of the b-SOD mutant mentioned above (t, wildtype = 892”C, t, mutant = 85.5”C: McRee et al.. 1990). The same kind of amino acid substitution occurs naturally in y-SOD, which shows a totally buried Ala residue at position 6. Cys6 in the b-SOD is in tight contact with the apolar residues Leu8> Phe20, Met115 and Ilel47. Of these, Met115 is substituted by Val in y-SOD, resulting in a packing cavity due to the absence of the SD atom of residue Metll5, and of the SG atom of residue Cys6. The Metll5-+Val substitution appears to be partially compensated, in terms of the sterical packing, by the concerted Leu82+Met substitution in y-SOD. Despite these compensating substitutions, occurring within the small B-sheet at strands 7g and 5e, no evident structural detail appears to compensate for the absence of the two bulky sulphur atoms of the residues Cys(Ala)6 and Met(Val)ll5 (in y-SOD). which face each other across the p-barrel. In the CysS--+Ala b-SOD mutant, the removal of the Cys6

Crystal Structure of Yeast Cu,Zn Superoxide Dismutase

807

Figure 13. Stereoscopic drawing of the OH 1216, having a structural role, connecting the 1st and last B-strand through direct hydrogen bonds to main-chain atoms Va17 N and Va1146 0. The water molecule is additionally co-ordinated by 2 other water molecules, OH 1258 and OH 1262, forming an almost ideal tetrahedral.

SG atom results in a tighter packing of the N and C terminus /?-strands (with decrease in their mutual

distance of 0.4 A). In spite of the absence of the two above-mentioned sulphur atoms in y-SOD, no compensating strand shifts seem to take place, the two strands being at the same distance as in the wild-type b-SOD. The inner cavity left can be one of the structural bases of the lower t, observed in y-SOD as compared to b-SOD. Among the physico-chemical properties that confer thermal stability to a protein molecule, a decrease of the configurational entropy of the unfolded chain has been shown to raise the t, in a manner (Matthews et al., 1987). consistent Systematic structural and physico-chemical investigations on phage T4 lysozyme have shown that substitution of Gly residues with CB branched sidechains or with proline residues, decreasing the number of configurational states available to the unfolded protein chain, decrease the overall AC for the folding process (Matthews, 1987). In y-SOD, which is less thermostable than b-SOD, a destabilizing substitution is observed at position 13 in the loop connecting fi-strands la and 2b. Residue Pro13 of b-SOD is substituted by Gly in y-SOD, in a position

that

could play

a key role in the assembly

of b-strands during the folding. According to Matthews et al. (19871, a similar Ala+Pro substitution in phage T4 lysozyme accounts for a free energy stabilization A(AG) of 2 to 4 kJ/mol, with an increase in t, of 2 deg. C. It has been shown that the formation of a disulphide bridge reduces the entropy of the unfolded polypeptide chain; the bigger the number of residues between the covalent cross-link, the larger the entropy

contribution

folded protein structure The

two

evolutionary

to the stabilization

of the

(Poland & Sheraga, 1965). invariant

cysteine

residues

(55 and 144) that form the disulphide bridge are separated by 88 amino acid residues, which is well above the most frequent separation of 10 to 14 residues (Thornton, 1981), giving, therefore, a

substantial contribution to the stability of the tertiary structure of the superoxide dismutases, particularly to the 6,5 loop. Area per la Ricerca di Trieste (Italy) is acknowledged for the financial support of K.D. This project was supported by grants from the Italian National Research Council target oriented Project “Biotecnologie e Biostrumentazione”, by the special project “Peptidi Bioattivi”, and from the Ministry of the University and Scientific Research. References Arnold, I. D. & Lepock, I. R. (1982). Reversibility of the thermal denaturation of yeast superoxide dismutase. FEBS

Letters,

146, 302-306.

Bannister,

J. V., Bannister, W. H. & Rotilio, G. (1987). Aspects of the structure, function, and applications of superoxide dismutase. CRC B&hem. 22, 111-180. Barra, D., Bossa, F., Marmocchi, F., Martini, F., Rigo, A. 6 Rotilio, G. (1979). Differential effects of urea on yeast and bovine copper, zinc superoxide dismutases, in relation to the extent of analogy of primary structures. Biochem. Biophys. Res. Commun. 86, 11991205. Berendsen, H. J. C., Postma, J. P. M., van Gunsteren, W. F., DiNola, A. & Haak, J. R. (1984). Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81, 3684-3690. Bernstein, F. C., Koetzle, T. F., Williams, G. J. B., Meyer, E. F., Jr, Bria, M. D., Rodgers, J. R., Kennard, O., Shimanouchi, T. & Tasumi, M. (1977). The protein data bank: a computer-based archival file for macromolecular studies. J. Mol. Biol. 112, 535-542. Capo, C. R., Polticelli, F., Calabrese, L., SchininB, M. E., Carri, M. T. & Rotilio, G. (1990). The copper, zinc superoxide dismutase isoenzymes of Xenopus laevis: purification, identification of a heterodimer and differential heat sensitivity. Biochem. Biophys. Res. Commun. 173, 1186-1193. Civalleri, L., Pini, C., Rigo, A., Federici, R., Calabrese, L. & Rotilio, G. (1982). Isolation and preliminary characterization of electrophoretic variants of dismutase. Mol. Cell. copper, zinc superoxide Biochem.

47, 3-9.

K. Djinovic Crowther, R. A. (1972). The fast rotation function. In The Molecular Replacement Method (Rossmann, M. G., ed.), pp. 173-178, Gordon 6 Breach, New York. Crowther, R. A. & Blow, D. M. (1967). Positioning a known molecule in an unknown crystal structure. Actu Crystallogr. 23, 544-548. Desideri, A., Falconi, M., Parisi, V. & Rotilio, G. (1989). Conservation of local electric fields in the evolution of Cu,Zn superoxide dismutase. FEBS Letters. 250, 45-48. Desideri, A., Falconi, M., Polticelli, F., Bolognesi, M., Djinovic, K. & Rotilio, G. (1992). Evolution of the catalytically important electric field in Cu,Zn superoxide dismut,ase: a study of six proteins with different electric charge and various degrees of sequence homology. J. Mol. Biol. 223, 337-342. Djinovic, K., Gatti, G.. Coda, A., Antolini, L., Pelosi, G.. Desideri, A., Falconi, M., Marmocchi, F., Rotilio, G. & Bolognesi, M. (1991). Crystal structure solution and molecular dynamics refinement of the yeast Cu,Zn superoxide dismutase. Acta Crystallogr. sect. B, 47, 918-927. Dunbar, J. C., Holmquist, B. & Johansen, J. T. (1984a). Asymmetric active site structures in yeast dicopper, dizinc superoxide dismutase. 1. Reconstitution of apo-superoxide dismutase. Biochemistry, 23, 43244329. Dunbar, J. C.: Holmquist, B. & Johansen, J. T. (19846). Asymmetric active site structures in yeast dicopper, dizinc superoxide dismutase. 2. pH-dependent incorporations of cobalt into the metal binding sites. Biochemistry, 23, 4330-4335. Fielden, E. M., Roberts, P. B., Bray, R. C., Lowe, D. J., Mautner, G. N., Rotilio, G. & Calabrese, L. (1974). The mechanism of action of superoxide dismutase from pulse radiolysis and electron paramagnetic resonance. B&hem. J. 139, 49-60. Fisher, C. L., Tainer, J. A., Pique, M. E. & Getzoff, E. D. (1990). Visualization of molecular flexibility and its effects on electrostatic recognition. J. Mol. Graph. 8, 125-132. Fitzgerald, P. M. D. (1988). MERLOT, an integrated package of computer programs for the determination of crystal structures by molecular replacement. J. Appl. Cystallogr. 21, 273-278. French, S. & Wilson, K. S. (1978). On treatment of negative intensity observations. A&z Cry8tallogr. sect. A, 34, 517-525. Frigerio, F., Falconi, M., Gatti, G., Bolognesi, M., Desideri, A., Marmocchi, F. BE Rotilio, G. (1989). Crystallographic characterization and the threedimensional model of yeast Cu,Zn superoxide dismutase. Biochem. Biophys. Res. Conzmun. 160, 677-681. Fujinaga, M., Gross, P. & van Gunsteren, W. F. (1989). Testing the method of crystallographic refinement using molecular dynamics. J. Appl. Crystallogr. 22, l-8. Getzoff, E. D., Tainer, J. A., Weiner, P. K., Kollman, P. A., Richardson, J. & Richardson, D. C. (1983). Electrostatic recognition between superoxide and copper, zinc superoxide dismutase. Nature (London), 306, 287-290. Getzoff, E. D., Tainer, J. A., Stempien, M. M., Bell, G. I. & Hallewell, R. A. (1989). Evolution of Cu,Zn superoxide dismutase and the greek-key B-barrel structural motif. Proteins, 5, 322-336. Jones, T. A. (1978). A graphics model building and refineJ. Appl. system for macromolecules. ment Crystallogr. 11, 268-272.

et al. Kabsch, W. & Sander, C. (1983). Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical structures. Biopolymers, 22, 2577-2637. Kitagawa, Y., Tanaka, N., Hata, Y., Kusonoki, M.. Lee, G., Katsube, Y., Asada, K., Aibara, S. & Morita, Y. (1991). Three-dimensional structure of Cu,Zn superoxide dismutase from spinach at 2.0 A resolution. J. B&hem. 109, 447-485. Lattman, E. E. & Love, W. E. (1970). Rotation search procedures for detecting a known molecule in a crystal. Actu Crystullogr. sect. B, 26, 1854-1857. Lepock, J. R.. Arnold, L. D., Torrie, B. H.; Andrews, B. & Kruuv. J. (1985). Structural analyses of various Cu++. Zn++ superoxide dismutases by differential scanning calorimetry and Raman spectroscopy. Arch. Bioehem. Biophys. 241, 243-251. Leslie, A. G. W., Brick, P. & Wonacott, A. J. (1986). An improved package for the measurement of oscillation photographs. CCP4 News, 18, 33-39. Ludwig, M. L., Metzger. A. L.. Pattridge, K. A. & Stallings, W. C. (1991). Manganese superoxide dismutase from Thermus thermophilus. A structural model refined at 1.8 d resolution. J. Mol. Biol. 219,335-358. MacPherson, A. (1982). Preparation and Analysis of Protein Crystals, pp. 82-159, Wiley and Sons. New York. Matthews, B. W. (1968). Solvent content of protein crystals. J. Mol. Biol. 33, 491-497. Matthews, B. W. (1987). Genetic and structural analysis of the protein stability problem. Biochemi&y, 26, 6885-6888. Matthews, B. W., Nicholson, H. & Becktel? W. J. (1987). Enhanced protein thermostability from site-directed mutations that decrease the entropy of unfolding. Proc. Nat. Acad. Sci., IJ.S.A. 84, 6663-6667. McRee, D. E., Redford, S., Getzoff, E., Lepock, J. R.. Hallewell, R. A. & Tainer, J. (1990). Changes in crystallographic structure and thermostability of Cu,Zn superoxide dismutitse mutant resulting from the removal of a buried cysteine. J. Biol. Chem. 244. 14234-14241. Nyborg, J. & Wonacott, A. J. (1977). The RotutiMz Method inCrystallography(Arndt,U.V.&Wonacott,A.J..eds). pp. 139-151, Elsevier, North-Holland, Amsterdam. O’Neill, P.. Davies, S., Calabrese, L., Capo, C., Marmocchi, F., Natoli, G. & Rotilio, G. (1988). The effects of pH and various salts upon the activity of a series of superoxide dismutases. Biochem. J. 251. 41-46. Orpen, A. G., Brammer, L., Allen, F. H., Kennard, 0.. Watson? D. & Taylor, R. (1989). Tables of bond lengths determined by X-ray and neutron diffraction. 2. Organometallic compounds and co-ordination complexes of the d- and f-block metals. J. Chem. Sot. Dalton, 51-583. Parker, M. W. & Blake, C. C. F. (1988). Crystal structure of manganese superoxide dismutase from Bacillus stearothemphilw at 2.4 A resolution. J. Mol. Biol. 199, 649-661. Perry, L. J. & Wetzel, R. (1987). The role of cysteine oxidation in the thermal inactivation of T4 lysozyme. Protein Eng., 1, 101-105. Poland, D. C. & Sheraga, H. A. (1965). Statistical mechanits of noncovalent bonds in polyamino acids. Biwpolymers, 3, 379-399. Ramakrishnan, C. & Ramachandran. C;. N. (1965). Stereochemical criteria for polypeptide and protein chain conformation. Biophys. J. 5, 908-933.

Crystal Structure of Yeast Cu,.!h &&peroxide Dismutase Read, R. J. (1986). Improved Fourier coefficients for maps using phases from partial structures with errors. Acta Crystullogr. sect. A, 42, 140-149. Schinina, M. E., D’Altri, L., Lania, A., Galtieri, A. & Capo, C. (1991). The copper, zinc superoxide dismutase from marine turtle Caretta caretta. In VI Convegno Nazionale, Proteine ‘91, 22-24 May 1991, p. 70, Trieste, Italy. Sines, J. J., Allison, S. A. & McCammon, J. A. (1990). Point charge distributions and electrostatic steering in enzyme/substrate encounter: Brownian dynamics of modified copper/zinc superoxide dismutases. Biochemistry, 29, 9403-9412. Stallinga, W. C., Pattridge, K. A., Strong, R. K. 6 Ludwig, M. L. (1984). Manganese and iron superoxide dismutases are structural homologs. J. Biol. Chem. 259, 10695-10699. Stallings, W. C., Pattridge, K. A., Strong, R. K., Ludwig, M. L., Yamakura, F., Isobe, T. & Steinman, H. M. (1987). Active site homology in iron and manganese superoxide dismutase. In Patterson and Pattersons (Glusker, J. P., Patterson, B. K. & Rossi, M., eds), pp. 505-513, Oxford University Press, Oxford.

Edited

809

Steinman, H. M. (1980). The amino acid sequence of copper-zinc superoxide dismutase from baker’s yeast. J. Biol. Chem. 225, 6758-6765. Tainer, J. A., Getzoff, E. D., Beem, K. M., Richardson, J. S. & Richardson, D. C. (1982). Determination and analysis of the 2 A structure of copper, zinc superoxide dismutase. J. Mol. BioZ. 160, 181-217. Thornton, J. M. (1981). Disulphide bridges in globular proteins. J. Mol. Biol. 151, 261-287. Tronrud, D. E., TenEyck, L. F. & Matthews, B. W. (1987). An efficient general-purpose least-squares refinement program for macromolecular structures. Acta Cryetallogr. sect. A, 43, 489-501. van Gunsteren, W. F. I!%Berendsen, H. J. C. (1987). BIOMOS. B&molecular Software, Laboratory of Physical Chemistry, University of Groningen, The Netherlands. Vriend, G. (1990). WHAT IF: a molecular modelling and design program. J. Mol. Graph. 8, 52-56. Wilmot, C. M. 6 Thornton, J. M. (1990). fi-Turns and their distortions: a proposed new nomenclature. Protein Eng. 3, 479-493.

by R. Huber