.I. Mol. Bid. (1992) 224, 179-205

Refined Crystal Stru!ture of Ascorbate Oxidase at l-9 A Resolution Albrecht Messerschmidt, Rudolf Ladenstein, Robert Huber Max-Planklnstitut fiir W-8033 Martinsried,

Biochemie F.R.G.

Martin0 Bolognesi Dipartimento Universita

di Genetica e Microbiologia di Pavia, I-27100 Pavia, Italy

Luciana Avigliano Dipartimento di Scienze e Tecnologie Biomediche Universita dell’Aquila, I-67100 L’Aquila,

e di Biometrk Italy

Raffaele Petruzzelli, Antonello Rossi and Alessandro Finazzi-Agr6 Dipartimenti Universita

di Biologia e Medicina “Tor Vergata”. I-00173

(Received 15 August

h’perimentale

Roma, Italy

1991; accepted 26 November

1991)

The crystal structure of the fully oxidized form of ascorbate oxidase (EC 1.10.3.3) from Zucchini has been refined at 1.90 A (1 a = 91 nm) resolution, using an energy-restrained least-squares refinement procedure. The refined model, which includes 8764 protein atoms, 9 copper atoms and 970 solvent molecules, has a crysta,llographic R-factor of 293% for 85,252 reflections between 8 and 1.90 A resolution. The root-mean-square deviation in bond lengths and bond angles from ideal values is 0.011 A and 299”, respectively. The subunits of 552 residues (70,000 M,) are arranged as tetramers with 02 symmetry. One of the dyads is realized by the crystallographic axis parallel to the c-axis giving one dimer in the asymmetric unit. The dimer related about this crystallographic axis is suggested as the dimer present in solution. Asn92 is the attachment site for one of the t’wo N-linked sugar moieties, which has defined electron density for the N-linked K-acetylglucosamine ring. Each subunit is built up by three domains arranged sequentially on the polypeptide chain and tightly associated in space. The folding of all three domains is of a similar /?-barrel type and related to plastocyanin and azurin. An analysis of intra- and intertetramer hydrogen bond and van der Waals interactions is presented. Each subunit has four copper atoms bound as mononuclear and trinuclear species. The mononuclear copper has two histidine, a cysteine and a methionine ligand and represents the type-l copper. It is located in domain 3. The bond lengths of the type-l copper centrr are comparable to the values for oxidized plastocyanin. The trinuclear cluster has eight histidine ligands symmetrically supplied from domain 1 and 3. It may be subdivided into a pair of copper atoms with histidine ligands whose ligating N-atoms (5 NE2 atoms and one ND1 atom) are arranged trigonal prismatic. The pair is the putative type-3 copper. The remaining copper has two histidine ligands and is the putative spectroscopic hype-2 copper. Two oxygen atoms are bound to the trinuclear species as OH- or 02- and bridging the putative type-3 copper pair and as OH- or H,O bound to the putative type-2 copper trans to the copper pair. The bond lengths within the trinuclear copper site are similar to comparable binuclear model compounds. The putative binding site for the reducing substrate is close to the type-l copper. Two channels providing access from the solvent to the trinuclear copper site, the putative 179 (N)22~2836/92/05017!)~%7

$03.00/O

IQ 1992 Arademk

Press Limited

binding sit P of the dioxygen. c~mld br idrnt.ified. A catalytic, mt:c*hanism ol’as~rbatt~ oriclas;c~ is proposed based on the availablr kinetic data. and thr t hre~,-dimerrsion~rl struc+urcL and t h(~ associated electron transfer pr0cessr.s arc’ discussed. ascorbat,e oxidase; refntd crystal struct,ure: blue copper oxidasta: multi-copper enzyme: trinuclear copper site; rlecW)n transftlr

Krywords:

1. Introduction Ascorbate oxidase (AOaset) is a blue multicopper oxidase that catalyses t,he four-electron reduction of dioxygen to water with concomitant one-electron oxidation of the reducing organic (Malkin & substrate Malmstriim, 1970). (:opper-dependent, Abase is found only in higher plants (Chichiricco rt al.. 1989). The enzyme from (Yucurhita pepo m,edullosa (green Zucchini) is a dimer of 140,000 M, containing eight copper ions of three different spectroscopic forms classified as type- I type-2 and type-3 according to Vgnngard’s proposal (see Fee, 1975). The immunohistochemical localizat ion of AOase in green Zucchini reveals that, Abase is distributed in all specimens examined ubiquitously over vegetative and reproductive organs. At the cellular level the enzyme is linked with the cell wall and cytoplasm (Chichiricco et al.. 1989). The in c:i~ role in plants of ascorbate and L40ase is still under debate. As catechols and polyphenols are also substrates in aitro (Marchesini et al.. 1977), AOase might be involved in biological processes likt fruit ripening. A role in a redox system, as an alt,ernativc to the mit,ochondrial chain, “in growth promotion 01‘ in susceptibility to disease has also been post’ulated (Butt’,

1980).

Primary structures of Abase from cucumber (Ohkawa et al., 1989) and pumpkin (Osaka rt rrl.. 1990) have recently been reported. The preliminar? three-dimensional * X-ray structure of the full! oxidized form of AOase from Zucchini has been published (Messerschmidt et al.. 1989) and the polypeptide fold and the (*o-ordination of the mononucalear blue copper site and the unprecedented trinuclear copper sitme have been described. Thr structural relationship to the other blue copper oxidases, lactase (EC’ 1.10.3.2) and ceruloplasmin (IN: 1.12.3.1), has been demonstrated hy amino acid sequence alignment based on the spatial structure of AOase from Zucchini (Messerschmidt & Huber. 1990). All canonical copper ligands are conserved with the exception of t,he methionine ligand of thr mononuclear blue type-l copper, which is leucinr in

domain 2 of human ceruloplasmin (Takahashi it nl.. 1984) and in laccasr from Nf~~rospora c’rassrx (Cirrmann et al.. 1988). It has recent’ly been shown hy a recombinant Met121 -+ Leu mutant of azurin from I’stwdomonas

aeruginosa

that

t,he met,hionine

is not

t Abbreviations used: XOasr, asc~orbate oxidase (EC’ I. 10.3.3): r.rn.8.. root-mean-square: e.p.r,. ele&ron paramagnetic resonance; c.d., circular dichroism: ET. electron transfer.

an essential component of the blue type-l site (Karlsson rt al.. 1989). In this paper. the refined crystal structure

chopper of the

fully oxidized form of AOase from green Zucchini is described and its implications for the catalytic electron mechanism. transfer processes and activation of dioxygen are discussed. 2. Experimental The solution of the X-ray structure of AOasr from Zucchini has been described in detail (Mesxerxchmidt ef al.. 1989). Two different crystal forms with space groups P2,2,dI and 1’2,2,2 were used in the crystal structure determination. Multiple isomorphous replacrrnrnt wit,h subsequent electron density averaging about the noncrystallographic symmetries in both csrystal forms and shell-wise phase expansion from 3.5 A to 2.5 A (I X = 0.1 nm) resolution using c~c%c~rlec*troti density averaging has been applied. The subunits form trtrarnrrs with /I2 s,vmmrtry. Thug 1’%,2,2 cryst,al form has a dimer in the asymmetric. unit,. The ktramrr is genrrakd t)v the carystallographic dyatl axis. In the P8,2,2, crystal km a t’etramrr is the asynlmetric unit. Model building was performed wit,h FR,Ol)O ( l~OUi/) I~niyur reflections Sowrejected unique refrct. I)ata mllplrtrnrsh K-l.9 A 193-1~90 A (‘rystal to crystal R-rnrrgrt

4 043 mm x 0.4 mm x 0.3 mm 249,029 90.635 86,529 8ij.6”) 349 0;

observed of reflect

of paramet,ers Final difference Fourier map

I’ROTETIS (Steigemann, 1974) was used for scaling and loading. Individual measurements for a unique reflection were rejected when deviating more than 407, from the mean value rejinemmt

(‘rystallographic refinement was carried out with the 1990). Use was made program system X-PLOR (Briinger, only of the conventional restrained least-squares refinrment procedure on positional parameters and B-factors. Refinement proceeded in several cycles, inberrupted for manual rebuilding on the basis of ZF,- F, difference Fourier maps. The summary of the restrained leastsquares refinement is listed in Table 2. In the early rounds of refinement an overall B-factor of 15 A* was used. In later stages, individual B-factors were refined but restrained to a standard deviation of 2.25 A2 for the difference of H-factors (AB) of bonded atoms and 3.0 A2 for atoms related by bond angles. Solvent molecules were introduced after the R-value was 260/, when they were in stereochemically reasonable positions and had high difference electron densities. The copper sites were refined with weak energy restraints to target values. The geometry of the mononuclear copper site was averaged from poplar plastocyanin (Guss & Freeman. 1983) and azurin from A Zcaligenea denitrijkzns (Baker, 1988). The geometry for the trinuclear copper centre was derived from 2 binuclear model compounds with nitrogen and oxygen copper copper ligands (Kartin et al.. 1984; Chaudhuri et al.. 1985). Weak potential energies were applied to Cu-ligand bonds (80 kcal mol-’ .% ’ Cu-equatorial for ligand. 30 kcal molt’ A-’ for Cu-axial ligand: 1 cal = 4.184 eJ) and ligand&aopper --ligand angles (10 kcal mot - ’ deg- ‘). As a check on the Cu-site geometry, all restraints were removed and the model subjected to extensive rounds of positional refinement. The changes in the mean values of bond lengths and angles were similar in magnit,ude to the mean positional errors of the atoms (025 a) as estimated from a Luzzati plot (Luzzati. 1952). Therefore. the weak restraints on the copper-site geometry were maintained over the whole rounds of refinement. No non-crystallographic. symmetry restraints were imposed on the dimer. The c*o-ordinat,es have been deposited in the lirookhaven Protein I)ata Rank (Bernstein rl al.. 1977). (e) Solvent accessibility Solvent-accessibility areas were calculated using the algorithm of Lee & Richards (1971), with a probe radius of 1.4 A.

all protein copper solvent

x+14~ x xi 252 $30, 339 O(, I)06 4

atoms atoms atoms atoms

9743 8761 9 970

Sumber

Avrrxge

((1) Model

x~oo- I .99 A 1.9% I90 A

r.m.s. co-ordinate shift on last cycle (last manual intervention) Z;umber of atoms

940,

t K-merge = zXlZ(h), - (I(h))l/ZZZ(h),, with In, intmsit,y in the it,h SOUIYPand (Z(h)), mean int,rnsity tton h 0ver all measurements c,f Z(h).

Resolution rangr Number of reflert,ions R = zIF,-F,I/zF,

temperature

factor

40,692

highest peak standard dewation all atoms protein atoms solvent at,oms

deviation from standard geometries

r.m.8.

(f) Cor~nolly

dot surface

and dockin!/

045 e/A3 0.09 e/A3 22+3 A2 21.1 A2 3X.4 A2

hnds

0.01 1 ‘1

an&s

299

r.prinwnts

(Connolly dot surfaces (Connolly. 1983) were calculated and displayed by the program MAI5 of I). Turk. The docking experiments were carried out wit,h program MAIN as well using the same force tield as during the X-PLOR-refinement. Co-ordinates and ryuilibrium geometries for the ascorbate molecule were taktln from the crystal structure (Hvoslrff. 1969). The force field constants were set) to values comparable to the protein part.

3. Results and Discussion (a) The jkal

model

The final model of Abase consists of 8764 protein atoms, nine copper atoms and 970 water molecules. The model was kept close to standard geometry throughout refinement. The final r.m.s. deviations from ideal bond lengths and angles are O+ll A and 2.99”. The final R-factor was 20.3’?,, for all data between 8.0 A and 1.9 i\ resolution. The average temperature factor for all atoms is 22.8 x2. The corresponding numbers of protein and solvent atoms are %I.1 A* and 384 A2. respecti\FelT. The mean positional error of the atoms a,s rstlmat,ed from a, Luzzati plot (Luzzati, 1952) is 055 B. As denidiscussed in detail for azurin from Alca.ligrnes tri$cans (Baker. 19&S), this estimate Indicates a maximum co-ordinate error. For well-defined parts of the structure, especially the b-strands. internal side-chains and regions around the copper sites, the error is likely to be lower. The largest peaks in the final difference map (@3 to 0.5 e/A3) were all associated with disordered residues. In sotnr cases two alternative conformations were clearly visible, but no attempt was made to model more than one conformation for a side-chain. The SF,-- It’, maps contoured at la show continuous density for all main-chain atoms with the exception of the two C-terminal residues. Several side-chains, all located on t,he protein surface, are poorly defined a,nd have high thermal parameters (H > 60 A2).

Figure 1. Stereo plot of t.he iv-acet,yl-~lucosaminr density map contoured at I.0 (r.

sugar ring bound to :\snW. :4tomk

During the late stage of refinement, the site of attachment of an N-glycosidic linked carbohydrate chain could be identified. Protein amino acid sequencing (Petruzzelli & Rossi, unpublished results) had indicated a histidine 92. An electron density map obtained by omitting this residue during phase calculation clearly showed density side-chain linked to an for an asparagine N-glucosamine sugar ring, however. Figure 1 depicts t.he final 2p0-fJC map at this region of subunit A. The Asn92-Phe93-Thr94 sequence also present in pumpkin AOase (Esaka et al., 1990) is characteristic for an N-glycosidic carbohydrate attachment site. D’Andrea et al. (1988) determined number and primary structure of the carbohydrate moieties attached to AOase from Zucchini. They found two identical carbohydrate moieties per subunit linked with ilr-acetyl-glucosamine to asparagine.

(b) O~~erall description

rrlotlrl

plus APO- PC elrc.troil

of the subunit

structzcrr

The subunits are arranged as homo-tetramers in the crystals. The structure of a subunit is shown schematically in Figure 2. Each subunit of 552 amino acid residues has a globular shape with dimensions of 49 A x 53 A x 65 A and is huilt up by three domains arranged sequentially on the polypeptide chain tightly associated in space. The folding of all three domains is of a similar P-barrel type. It, is distantly related to the small blue copper proteins like plastocyanin or azurin. Domain 1 is made up of two four-stranded B-sheets (Fig. 2(b)), which form a P-sandwich structure. Domain 2 consists of a six-stranded and a five-stranded p-sheet. Finally. domain 3 is built. up of two fivestranded a-sheets, which form the P-barrel structure and a four-stranded b-sheet, which is an extension at the N-terminal part of this domain. A topology

(bl

Figure 2. A schematic representation of the monomer structure AOase. (a) Monomer plus copper ions, (b) assignment of the secondary structure elements. B-Sheets are represented by arrows. sc-helices by helical ribbons. The Figures werr produced by the program RIBBON (Priestle. 1988).

Rejined Structure

in..,

of Ascorbate

Oxida.se

183

/

C”.., ‘

Figure 3. A topology/packing diagram of the domains of the Abase monomer compared with plastocyanin and azurin. Each p-strand is represented by a triangle whose apex points up or down whether the strand is viewed from (1 or N t~emminus. a-Helices are represented by circles.

diagram of Abase for all three domains and of the related structures of plastocyanin and azurin is shown in Figure 3, AOase contains seven helices. Domain 2 has a short a-helix (al) between strands A2 and B2. Domain 3 exhibits five short helices, which are located between strands D3 and E3 (as). 13 and 53 (ah). M3 and N3 (as) as well as at the C terminus (a,, M,). Helix a2 connects domain 2 and domain 3.t A comparison of the different variants of the P-barrel domain structure in Figure 2 shows that domain 1 of Abase has the simplest b-barrel with only two four-stranded a-sheets. Plastocyanin and azurin are quite similar but between strands 4 (El) and 6 (Fl) they have insertions of one strand (plastocyanin) or one strand and an a-helix (azurin). Domain 2 has one additional strand H2 in sheet D next to strand E2 (sheet B and strand El in domain 1) and t)wo additional strands, F2 and G2, in sheet C next to strand I2 (sheet A and strand Fl in domain 1). Domain 3 resembles domain 2 except for the insertion of the short a-helices and the addition of t,he four-stranded b-sheet at its N terminus. The mononuclear copper site is located in domain 3 and the trinuclear copper species is bound between domain 1 and domain 3 (see Fig. 2(a)). The copper site geometries will be discussed later. Each monomer exhibits three disulphide bridges. These are between domain 1 and domain 2 (CyslS(~~~201); domain 1 and domain 3 (Cys81-Cys538) and within domain 2 (Cys180-Cys193). Three putative attachment sites for A’-glycosidic linked carbohydrate moieties are present in the amino acid sequence of AOase from Zucchini (Asn92-Phe93-Thr94; Asn325-Phe326-Thr327: Asn440-Leu44 1-Ser442), but only Asn92 shows density for an X-acetyl-glucosamine group. There is 7 The 3 unlabelled 3,, helices in Table 3 are not taken into account because they consist of 3 residues only.

Figure 4. Ramachandran plots of main-chain dihedral angles for AOase. (a) Subunit A. (b) Subunit H. Glycine residues are indicated as squares.

no significant difference in the B-values put.ative carbohydrate binding sites. (~a)Secondary

for the three

structure

Secondary st,ructure was assigned on the basis of main-chain hydrogen bonding using the algorithm of Kabsch & Sander (1983) with a cut-off energy of 0.5 kcal/mol for the definition of a hydrogen bond interaction. Ramachandran plots of the refined models for subunit A and B (Fig. 4(a) and (b)) show that all non-glycine residues he in or close to energetically allowed regions. All residues with positive O-values are located in turns of the polypeptide chain.

1x4

-__.-

_..-..

Table 3 Pamrnrfrrs

oj helices Subunit

Suhunit A Helix namr

Residues

(@)t

(@>I.

wt

Ii wt

1a!& 156 3POb327 376-381 slob514 533m15‘xi e 540~ 545

-6X --et5 -02 -til -- i:! -70 ~ (2

(8) (6) (9) (6) (20) (X)

-34 - 39 -41 -40 -2i -37 -II

(1 I) (9) (4) (5) (9) (I 1)

-il -63 -63 -60 -7% --7x -62

(H) (7) (IO) (6) (16) (23)

- 33 ~-41 -11 --xl -28 -30 - -4I

(I I ) (IO) (S) (6) (7) (17)

466~47 I 7Ok7”M 1X4-186 525~527 1 .

--6X -x0 -x2 --7” - il

(15) (21) (13) (IX)

-21 -ti - ti -14 -IX

(15) (18) (3) (14)

-6X -Xl -67 -79 - 71

(12) (23) (13) (23)

-20 --6 -20 -x -IX

(12) (13) (16) (17)

t The r.m.8. deviation from the mem is given in parentheses. $ Has one I -+ i+5 hydrogen bond at its (‘-terminal end to give it short length n-helix. 5 Values from liarlow & Thornton (1988)

(i) Ile1ice.s and turns

There are six a-helices in the monomer of AOase involving a total of 37 residues or 6.7 ?:, of the polypeptide. In addition there are four short 3,, helices. The average conformation angles of t’hese secondary structures for subunit A and K are summarized in Table 3. The helical paramet,ers for both cr-helices and 3,, helices are close to those reported by Barlow & Thornton (1988). Except’ions are tlj and cx5, which have one i --t i + .5 bond at t’heir (‘-terminal end to give a short length n-helix. The turn parameters of all reverse t.urns are listed in Table 4 with their classification according to (Crawford et ab. (1973) and Richardson (1981). The preference for glycine at position 3 in the turn is very pronounced. There are three cis-proline turns (Huber bz Steigemann, 1974) and one Asx turn (Raker & Hubbard, 1984), which displays conventional geometry. The cis-proline residues have no unusual function in the structure. (ii) B-Sheet structure The residues involved in the seven p-sheet strut:t,ures and t’he hydrogen bonding patterns are shown in Figure 5. All sheets exhibit the characteristic right,-handed twist when viewed along the strand direction. There is a “wide” a-bulge (Richardson, 1981) in /?-sheet C of domain 2 (Fig. 5(b)) at residues Va1238 and Va1239. /II-Sheet E of domain 3 (Fig. 5(c)) contains a classic j-bulge (Richardson, 1981) at residues Leu457 and Gly458. (iii) Main-chain to side-chain hydrogen bonds There are numerous main-chain to side-chain bonds in AOase structure. hydrogen the Interactions involving Asn or Gln side-chains are of special interest because according to Baker & Hubbard (1984) they make a relatively large proportion of “long-range” interactions. providing

important cross-links in protein structures. Table 5 summarizes these t’ypes of hydrogen bonds in the subunit A and R of the AOase struct’ure. All interactions are intradomain with t’he exception of ND2 Asn22 to 0 His146. which is between domain 1 and domain 2. (d) Contact surface areaS within

thP tetramer

Contact surface areas for the monomer-monomer interactions within the tetramer were calculated using the algorithm of Lee & Richards (1971). These results are displayed in Table 6. It is evident that the contact surface areas between monomers A and C, and B and D are by far largest. These interfaces are mediated by the crystallographic dyad. Figure 6 displays t’he atomic structures of the tetramer and the three principal dimer contacts. AOasr from Zucchini occurs in solution as a dimer, which is likely the A, C and R, D dimer. Dimers A. (’ and 13. D are related by the local dyad and nearly identical. The tet)ramer is a dimer of dimers. which is related by a local dyad. (e) TPmperaturr

,fafac’tors

The variation of the average main-chain and sidechain atomic temperature factors along the polxpeptide chain in subunit A is shown in Figure 1. Subunit K exhibits a similar distribution. As expected, the temperature factors are greatest for loop regions connecting elements of secondary strucature or for segments between the indrvidual domains. Except for t,he N and (J t’erminus, there are eight loop regions with average main-chain R-values greater than 30 A*. The loop around residue 130 is part of the connecting segment between domain 1 and domain 2, loops around residues 199 and 274 are within domain 2, loops

Refined Structure

of Ascorbate

Oxidase

1x5

Table 4 Parameters

of turns in AOase N...O

y’,

@3

N-H

distance (8)

y3

0

angle (“) 13

A. Ruxrw fums (0; -+ Xi, 3 hydrogen bond) 16-19 ahi~ -58 -44 -35 -46 -120 27-30 ISGQ 5% 44 48 39 75 39-4% - 65 NAGD -59 127 175 1. 117 - 74 61-M RHGI I%0 108 -71 115 -61 107 91 66-69 &RUT -67 109 84--87 NPGE - 50 133 -40 133 81 - 137 -102 107-110 LGMQ 59 56 -146 - 55 10.5 -55 121 124-127 PQGK 103 4% I ‘71-I i5 LSGR 49 57 12 79 IXl-1X4 RIAA -75 -14 -74 -7 -114 -%H 187-190 I)SSI, -56 -83 -57 - 10 -2; 199-N% ESCA - 64 - 100 -64 -26 121 231-234 I( :NH -60 57 - 58 137 .- I 241-“44 ‘WGN 43 5‘3 42 81 -54 -6% 127 257 -%60 VSG6 I& 107 - 76 271-274 NPSE -80 -6 - 3 -94 302-305 S\‘SK . -Iii -84 --6X - 70 - I8 333- 336 AMGS 137 106 - 53 -66 119 .?I0 97 43 54 44 366~-359 ING\ 49 364-367 INDI 65 4” 5% 44 - 7x - .i -120 384 -387 LLHA -71 I6 - 1:s -94 - 75 398 -40 I PED\ - 56 -3 - 74 2 -94 402 405 I)IDl --78 r -54 -121 -47 -33 -iY 409-41” NEK7 99 l(k? -5x 12% 4”2 KIGE - 63 I 4% 127 -46 126 486 489 Fl’Yi(: -37 48

-115 73 94 103 98 75 -113 110 82 -103 -80 - 101 64 70 93 -90 -84 107 7% 62 - 104 -88 - 104 -II4 128 58

34 i -16 -8 -II 4 36 - 33 7 31 -II 4 I6 I6 - 18 9

32 11 - 10 -18 - %5 -8 36 - I3 -4 30 I 9 3I -” - 14 I3

:9 I “4 3.1 33 %+I 3.1 32 2.9 :v I 3.3 2.9 3.1 34 2x 3. I 34

;I.- 33 I5 -I 6 4 IO -7 33

-1X - 3 -II IO .i 5 7 II - I9 38

I6 - 3 I5

17 -40 II

il. Rww.w turns nsmcicrtrd with r-hrlices 316 -319 Au-nn -46 m-51 327 330 TYRI -75 -74 513 516 LHMC: -5x ~- 60

-%9 -%O - 35

- 33 -16 -2x

- 110 - 75 -x7

- 105

(’ Lvrtrr I?,“ lW ttrrns .?A .5;. LHTE -209-2 I-’ SPKK 449 -452 LH(:H

-4x 162 151

-123 51 68

- I %O

52 :iH

-5 30 I4

- 3 29 2I

64 -134 59

65 -134 65

IO ~ 52 4.5

-I -44 40

-94

-99 -58

I3 167 14;

32 I77 I40

I). oprn t/rr,/s I77 1x0 QIVKq %%+2ytj TA LA 4:1x- 44 I KENI, IX, cis-fJrr,/iw %08--301 30- 33 1.w

161

F. As.c tH)./lh As11473 OI>l

turn‘s YLPS QI’PC: SKPI

-73 -49

-- 84 -_ 69

- 86

~- 84

-43 151 149

-110 -76 -71

-I26 --71 -- 59

136 -3 1 139

134 - 33 133

-68 -13x -11.5

~_7% - 134 -I26

138 124 128

140 I %I) 112

-68 - 95

- l SD2 275 Am ODl ND2 ND2 365 Asn ODI ND2 SD2 432 Gin OF,1 NE:2 433 Asn 01)1 ND2 435 Asn ODl SD2

Main-chain atoms

23 146 9 7 140 138 298

lie N His 0 Glu N LysO Leu N Asn 0 Tyr N

302 ser 0 305 350 348 394 464 462 434 443 441 436

Lys 0 Leu N Phe 0 Pro 0 Phe F Glv 0 Ai; S Glu 0 Ltw N McfO

(S

Distance 0) (A)

A 2.9 3.8 3.1 3. L 3.0 3.1 3.1 2.7 2.8 3.0 2.8 2.9 2.9 2.1) 2.X 2.x 3, I 2.7

of Ascorbate

Angle

B 2.9 3.0 2.9 30 2.8 3.0 3.3 2.8 2.8 27 2.8 29 3.1 2.9 3.1 2.8 3.1 27

0 (“)

A 160 148 179 180 180 180 163 149 180 180 I x0 167 178 180 154 170 157 174

B 166 138 173 180 170 180 165 155 180 180 180 178 176 180 156 179 I60 165

187

The intertetramer hydrogen bond interactions include only crystal packing contacts. Mainly residues at the N terminus (Gln2, Arg4, Tyr6, Lys7, Trp8, Asn28, Arg37), around residue 400 (His386, Glu399, Asp400, Asp402, Asp404), around residue 467 (Glu467. Ser470) and at the C terminus (Ser544, Ile546, Asn547, Asn551) are involved in these contacts. These residues have different structural environments in the various subunits. This is documented in an r.m.s. deviation analysis between subunit A and subunit B. The r.m.s. deviation for all main-chain atoms between subunit A and subunit B is about 0.22 ir versus 1.0 to 2.0 A for t’he regions involved in the crystal packing contacts. The van der Waals’ interactions with distances less than 4 A are additionally listed in Table 9. Most of them are identical or located close to the residues taking part in the hydrogen bond intera&ions.

of

N--H

Oxidase

(i) Copper

site geometriex

The mononuclear copper site is located in domain 3 and has the four canonical type-l copper ligands (His, Cys, His, Met) also found in plastocyanin and azurin. It is co-ordinated to the ND1 atoms of His445 and 512, the SG atom of the Cys507 and the SD atom of Met517 in a distorted trigonal pyramidal geometry. The SD atom is at the long apex (see Fig. 8). Bond lengths of the type-l copper for both subunits are displayed in Table 10. They are compared to oxidized poplar plastocyanin (Guss B Freeman, 1983) and azurin from Pseudomonas aeruginosa (Nar et al.: 1991). Figure 9 shows an overlay of the type-l copper site in azurin, plastocyanin and AOase The copper is penta co-ordinated in azurin (glycine) with the main-chain carbonyl oxygen of the residue preceding the first histidine copper ligand on the polypeptide chain. This carbonyl oxygen forms the fifth ligand. In poplar plastocyanin, the homologous residue is a proline, which is the beginning of a turn causing the carbonyl oxygen to be removed to a distance of 3.4 w from the type-l copper ion. In AOase. the (-orresponding

subunit (’ and D and no contacts between subunits A and D or subunits B and C reflecting the size of the contact surface areas and confirming that A, C and B, D are probably the dimers in solution. There is only one direct hydrogen bond interaction between subunits A, B or subunits C, D but 12 with a single bridging water molecule versus 10 and 8 between subunits A, C, or 11 and 8 between subunits B. D, respectively. The contacts between subunits A, B or C, D are obviously highly hydrated. An additional structural metal ion could be detected by calculating a difference Fourier map of a native data, set and a data set collected on crystals soaked in buffer solution containing @5 mM-diethylet al., 1989). The dithiocarbamate (Messerschmidt nature of this metal ion is not clear. The metal has HisA and HisB286 as ligands and lies on the local 2-fold axis. We assume it to be a copper ion but it could be zinc. The metal ion may stabilize the tetramer in the crystal.

Table 6 Contact

surfa,ce

area8 (A2) for the six different within B 6491

t Gweric~

transformations

relating

the tetramer c 1828.5 36.9

the monomers

Operation

monomer-monomer

D 37.5 1797.8 6491 within

A w (‘

t,he tetramer

T (4

M

Monomer

K -+ monomer

A

Monomer

A + monomer

C

-0.30762 -0.95151 -090141 - 190000 oaoo9o 0~00000

interactions?

of AOase

-@95151 0.30762 -000200 090000 - 1aoooo oaoooo

090234 090072 - l@oOoo oaoooO 090000 1aoooo

6963 50.63 52.37 106.70 OQO 090

-a \

.--

(b)

(c) Fig. 6.

Rejined Structure

of Ascorbate

Oxidase

189

Cd) Figure 6. Stereo drawings of the tetramer structure and the principal dimer structures, z-axis upwards. (a) Whole t.etramer. (b) Dirner A. C. (c) Dimer A, B. (d) I)imer A. U.

strand is extended moving the carbonyl oxygen to a distance of 4.6 .h away from the type-l copper ion. This extended strand contributes Glu443 t’o the formation of the binding site in the reducing substrate (see Fig. 14). The trinuclear copper site (see Fig. IO) has eight histidine ligands symmetrically supplied by domain 1 and domain 3 and two oxygen ligands. Seven histidine residues are ligated by their NE2 atoms to the copper ions whereas His62 is ligated to CU3 K3 by its ND1 atom. In the preliminary structural report on AOase (Messerschmidt et al., 1989) based on lower resolution data all eight histidine residues were modelled with their NE2 atoms as ligands to the copper atoms of the trinuclear site. The high resolution data allow an unequivocal interpretation shown in Figure 11 for subunit B. His62 is part of a j-sheet while His450 is not. An overlay of the relevant parts of domain 3 onto domain 1 shows t,hat, His62 comes closer with its main-chain atoms

C

50

100

150

200

250

300

350

400

450

500

550

Residue

Figure 7. Variation

of the average main-chain (continuous line) and side-chain temperature factors for subunit A. subunit H displays a similar distribution.

to the copper ion CU3 than the corresponding atoms of His450. As a consequence, the side-chain of His62 has to adopt a conformation where ND1 of the imidazole ring is ligated to the copper ion CU3. The trinuclear copper site may be subdivided into a pair of copper ions (CU2 K3, CU3 K3) with six histidine ligands arranged tri onal prismatic. The pair is bridged by OH- or 0 5 -, which leads to a strong antiferromagnetic coupling and makes this copper pair e.p.r. silent. The pair very likely reprethe spectroscopic type-3 copper. The sents remaining copper ion (CU4 K3) has two histidine ligands and an OH- or water ligand and probably represents the spectroscopic type-2 copper. Both oxygen ligands could be clearly detected from difference Fourier maps and the final electron density of the trinuclear copper site for subunit A is shown in Figure 12. An oxygen ligand in the centre of the three copper ions is not seen. Also the difference Fourier map did not show significant densit,y at this position. Cole et al. (1990) studied the electronic structure of the lactase trinuclear copper active site by the use of absorption, circular dichroism and lowtemperature magnetic circular dichroism spectroscopies. The assigned ligand field transition energies indicated that all three copper ions have tetragonal geometries and that the two type-3 copper ions are inequivalent. The latter interpretation fits well to our structural model as CU3 K3 of the copper pair is ligated to one ND1 and two NE2 atoms, but the former interpretation of tetragonal co-ordination geometries for all three copper ions is not consistent with the structure. The copper ions of the pair are both tetrahedrally co-ordinated whereas the type-2 copper has three ligands. The existence of a central oxygen ligand would give rise to a penta-co-ordination of both copper pair atoms (but not a tetragonal-pyramidal co-ordination) and a square planar co-ordination for the spectroscopic type-2 copper.

Table 7 interactions

involving

ion pa%rs

H-bonding distances (A)

Residues

Solvent accessibility area (Aqt

Intrasubunit

Intersubunit

u 3.0 2.9 3.2 2.9 3.0 3.2 3.2 3.1 2.9 3I 2.9 > 44 3.0 2.x 2.8 3.1 2.9 3.5 36 31 3.0 3I 3.0 3” 33 3, I 38 3.0 3.2 2.x 32 2.9 29 32 3.8 2.9 3.2 3.2 2.9 >44 > 4.0 2.9

A 3.0 2.9 34 3.9 2.9 2.8 3.2 :+0 29 3.3 2.9 3.4 2.9 3.0 2.X 3.4 3.0 3.2 3,,5 3.1 3, I 3.1 3.0 3.0 3.3 3.0 3.1 3.1 3.2 2.8 3.5 3. I 2.7 29 3.0 3.0 2.9 3.6 2.9 2.9 3.3 >4.0

NH1 Arg4-ODl Asp42 NE Arg4-ODI Asp42 OE2 GluS-NZ Lys51 OEI GluS-NZ Lys51 NH1 Arg37-OEl Glul36 NH2 Arg37-OE2 Glu136 NH1 Arg37-OE2 Glu136 NH1 Arg37-OD2 Asp122 NE Arg37-ODl Asp122 NH2 Arg67-ODl Asp498 NH2 Arg67pOD2 Asp498 OEl Glu87pNZ Lys534 NE Arglll-OEl Glu510 OE2 Glu129-NH1 Arg215 OEl Glu136-NH2 Arg217 OEl Glul66-NH1 Arg285 OEl Glu166-NE Arg285 NH2 Arg215-OE2 Glu240 ODl Asp242-NE Arg48O ODl Asp242-NH1 Arg480 OD2 Asp242SNE Arg480 ODl Asp253-NH1 Arg322 OD2 Asp253-NH2 Arg322 ODl Asp255NHl Arg283 OD2 Asp255-NE Arg283 OD2 Asp255kNHl Arg283 OE2 Glu260-NH1 Arg480 NH2 Arg285-OEl Glu443 NH2 Arg285kOE2 Glu443 NZ Lys324kODl Asp461 NE Arg345-OE2 Glu425 NH1 Arg345-OEl Glu425 NH2 Arg345-OE2 Glu395 OE2 Glu410-NZ Lys411 NH1 Arg346-ODl Asp428 NH 1 Arg346-OE2 Glu469 NH2 Arg346-OEl Glu469 NH2 Arg346-OE2 Glu469 ODl Asp428&NH 1 Arg494 OD2 Asp428-NE Arg494 OD2 Asp42SNHl Arg494 ODl Asp42SNE Arg494

A 11 0 51 *54 14 26 15 I6 II 3X 20 50 5 4 4 18 8 1 0 0 0 34 12 2 0 I 6 I3 6 32 8 24 I3 5I 0 12 I3 17 5 0 5 0

A --t B or C --t D. related about the local dyad

no ion pairs Intersubunit.

related about th le dgad parallel to the c-axis

OEl Glu21-NH2 OE2 Glu21-NH1

A -+(! 3.0 3.0

Arg162 Arg162

13-+ I) 3.0 3.2

Intertetramer i$ ymmetr\ operation NH1 ArgA4-ODl NH2 ArgA4-OD2 NH2 ArgA4-OD2

AspA AspA AspA

32 2.8 3.4

”

x+1/2,-Y+1/2,-2 x+1/2,Y+1/2,-2 x+1/2,-Y+1/2,-2

t The value given is the sum of the solvent-accessible interaction.

TX 0 0 0

area of the 2 atoms directly

Translation component TY T% 0 0 0 0 0 0 involved

in the

qf Ascorbate

Refined Structure

Oxidase

191

Table 8 Analysis

of hydrogen

bonds involving

Number To To To To To

main-chain 0 main-chain N side-chain 0 side-chain S water

506 297 359 207 579

Mean distance (A) 3.04 3.04 3.02 3.03 3.00

(024) (0.22) (0.25) (0.25) (923)

water Mean angle (0) 130 159 108 154

(20) (34) (34) (33)

Table 9 Intersubunit

and crystal packing

Hydrogen bond distances (A)

Residues involved A. Protein-protein

interactions Symmetry operation code?

interactions

Intersubunit OEl GluA21lNH2 ArgA162 OE2 GluA21lNHl ArgA162 GGl ThrA544OEl GluA55 0 ThrA54-N GlyA56 OE2 GluA87-N GlyA539 ND2 AsnA288-ODl AsnB288 NE ArgA530-0 AsnA548 NH1 ArgA530--0 AsnA548 NE ArgA530-0 ProA 0 IleA531-ND4 AsnA548 OG 1 ThrA533-ND2 AsnA548 OEl GluB21-NH2 ArgB162 OE2 GluB21-NH1 ArgB162 NE2 HisB53-OG SerB157 OGl ThrB54-OEl GluB55 0 ThrB544N GlyB56 OE2 GluB87-N GlyB539 NE ArgB530-0 AsnB548 NH1 ArgB530-0 AsnB548 NE ArgB530-0 ProB549 0 lleB531-ND2 AsnB548 OGl ThrB5333ND2 AsnB548 NE2 GlnA220 AspA NH 1 ArgA4-ODl AspA NH2 ArgA4-OD2 AspA NH2 ArgA4-OD2 AspA ND2 AsnA28-OE2 GluA467 NE2 GlnA125-OEl GluA439 NE2 HisA386-ODl AsnB551 NE2 HisA386-0 AsnB551 0 SerA544ND2 AsnB547 OH TyrBB-N GluB47 N ArgB37-OEl GluB467 NZ LysB338-ODl AsnB357 NE2 HisB386-ODl AsnA551 NE2 HisB38660 AsnA551 0 SerB544-ND2 AsnA547 ND2 AsnB547-0 SerA544

34 3.0 30 3.3 32 3.0 3.2 3.2 3.2 2.8 30 3.0 3.2 3.5 3.1 3.3 3.3 3.0 3.0 32 3.0 2.9 3.4 3.2 2.8 3..5 2.9 29 26 3.0 2.9 3.0 3.3 3.2 2.7 2.8 3.2 2.9

2 2 2 2 2 1 2 2 I 2 2 2 2 2 2 2 2 1 2 2 2 2 3 3 3 3 3 3 4 4 4 5 5 6 7 7 7 7

3.2, 2.8 3.3, 29 29, 2.7 2.9, 2.7 33, 2.8 3.0, 3.4 26,27 32, 2.6 3.1, 2.8 3.2, 2.9 28, 2.8 3.1, 35

1 2 2 2 2 2 2 1 2 1 1

13. Hydrogen bonds via a single bridging water molecule Intersubunit GE1 GlnA353-SOL82-ND2 AsnB189 0 TyrA186-SOL109N GluB439 OH TyrA382-SOL144-ODl AsnA548 N IleA531-SOL144-ODl AsnA548 OE2 GluA153-SOL195-ND1 HisA GE I GluA87-SOL212-N GlyA539 OE2 GluA87-SOL213-0 ProA OG SerA148-SOL222-ND1 HisA OE2 GluAI91-SOL519-0 TrpB163 OG SerA158-SOL538-N PheA14 GE I GluA443-SOL549ODl AsnBl89 GDl AspASOL542-NZ LysB438

192

A. Messerschmidt

et al.

Table 9 (continued) Hydrogen bond distances (A)

Residues involved

Symmetry operation codrt

ODl AsnA189-SOL644OEl GluB443 0 GluAZl-SOL690-NZ LysA151 0 SerB188SOL714-N TrpA163 0 AsnA189-SOL715-N TrpB163 N LysA438SOL1044-OG SerB188 OEl GluB555SOL1102OGl SerB54 N GluA439-SOL1109-0 TyrB186 0 GlyB539-SOL1130-NZ LysB534 OH TyrB382-SOL11444ODl AsnB548 OE2 GluB153-SOLl1955NDl HisB53 OEl GluB87-SOL1212-N GlyB539 OE2 GluB87-SOL1213-0 ProB373 OEl GluBll-SOL15280 SerB157 NZ LysA438-SOL15522NE2 GlnB313 0 IleA161-SOL16222OEl GluB191 ND2 AsnB84-SOL1685-0 AlaB

2.8, :+O 3.0. 3.2 29, 2.9 32. 2.8 34, 2.5 2.x. 3.1 2.9, 3.3 2.9. 3.3 2+3. 2% 3.0, 2.8 2.9, 2.9 2.7, 3.2 3.0. 3.1 2.9, 2.8 2.8, 3.5 3.4. 2.5

I 2 I I 1 2 1 2 2 2 2 2 I I 1 2

Intertetramer N GluA136-SOL158-ODl AspA 0 AspASOL571-ODl AspA OE2 GluA467-SOL617-0 LysA7 0 IleA546-SOL6699ND2 AsnB547 OE2 GluA399-SOL682-OH TyrA204 NH2 ArgA37-SOL746-0 GluA399 NE1 TrpBS-SOL1132-OG SerB470 0 SerB544-SOL17033ODl AsnA.547

2.9. 32. 3.0, 3.4. 30, 3.1, 28, 3.0.

3 8 8 4 8 3 5 7

C. van der Waals’ interactions;

atoms

involved in wnkwts less than 4 A

lntersubunit TyrA12 CB PheA14, CB, CG, CDl, CD2, CEl, CE2, CZ IleA23, CG2 HisA CB, CEl, NE2 ThrA54 CB, CG2, OGl

GluA55 CG ProA 0 GlyA86 0 GluA87 CG HisA150, CG, CD2, NDl, CEl, NE2 TrpA163 CB, CG, CDl, NE1 TyrA186 0 AspA ODl SerA188 0, OGl AsnA189 CA, CG, ODl, ND2 GluA191 OEl, OE2 LysA196 AsnA288 TyrA382 ArgA530

CE, NZ ND2 OH CG, CZ, NE, NH1

IleA 0 LeuA536 CD2 AlaA C, 0 CysA538 CB LeuA545 0 IleA CGl TyrB12 CB PheBl4 CDl, CEl, CE2, CZ

3.1 3.2 3.2 3.5 2.7 3.4 3.2 2.8

ValA154 CG2 ValA154 CG2 GlyAl55 N, CA ArgA162 CZ, NE, NHl, NH2 LysA151 CG GluA153 OE2 SerA157 OG GluA55 CA, CD, CG, OEl GlyA56 N LeuA107 CG GIyA108 CA GluA55 CG ProA CB ProA CB ProA CB HisA156, CG, CD2, NDl, CEl, NE2 AsnB189 CG, ODl, ND2 LysB438 CG LysB438 NZ IleBl61 CG2, 0 ValB366 CG2 TrpB163 CB, CG, CDl, CD2 IleBl61 0 ArgB162 CA, CB, CG LysB196 CE, NZ AsnB288, ODl AsnA548 ODl AsnA548 C, 0, CG, ODl, ND2 ProA C, 0 LysA556 CB, NZ AsnA548 CG, ODl, ND2 AsnA548 ND2 AlaA C. 0 AlaA CB IleA CD1 AsnA84 ND2 LeuA545 0 IleA CA AsnA547 N IleA CG 1 ValB154 CG2 LysB151 CG, 0 ValB154 CGl GlyB155 N, CA

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 I 2 I 2 2 3 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

Rejined Structure of Ascorbate Oxidase

193

Table 9 (continued) Symmetry operation code?

Residues involved

AsnB22 ODI HisB53, CB, CG, NDI,

SerB158 ArgBl62 LysBl51 GluB153 SerBl57

CEl, NE2

ThrB54 CB, 0, OGl

GluB55 CG AsnB84 ND2 ProB85 0 GluB87 CG, CD, OEl, OE2 LysBl96 CE, NZ TyrB382 OH ArgB530 CD, CG, NE, CZ, NH1

IleB531 0 ThrB533, N, CIA, CB, OGl, CG2 AlaB

0

LeuB545 0, CB, CD2

IleB546 CGI Intertetramer GlnA30 OEl, CG. NE2 ProA CB, CG ArgA37 CZ, NHl, NH2 AsnA39 ODl AspAl 0 LeuA384 CD2 HisA CD2, CEl, NE2 SerA544 C, 0 LeuA.545 CA. (I, 0 ArgB4 NH2 TyrB6 CA, CD1 LysB7 N, 0, CB ThrB35 0 ArgB37 N, 0, CB M&B334 CE LysB338 AsnB357 LeuB384 HisB386 SerB544

NZ ODl CD2 CD2, CEl, NE2 CB, 0, CG

AsnB547 CA, (‘B. ND2 t Legend for symmetry Symmetr,y operation code

operation

OG CD. NE, CZ NZ OE2 OG GluB55 C, CA, CB, CG, CD, OEl GlyB56 N LeuBl07 CG GlyBl08 CA GluB55 CG CysB538 CB ProB373 CB ProB373 CB GlyB539 N, CA LysA196 CE, NZ AsnB548 ODI AsnB548 C, 0, CG, ODI, ND2 ProB549 C, 0 LysB559 N, CA, CB, CD AsnB548 CG, ND2 IleB546 CB, CG2, CD1 BsnB548 CB, CG, ND2 AlaB 0 IleB546 CD1 LeuB545 0 IleB546 CA, CG2 AsnB547 N TleB546 (‘Gl

2 2 2 2 2 I 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

AlaA GluA467 GluA467 AspA AspA LysA411 ProB552 AsnB551 AsnB547 ProA AsnB547 SerB465 GluB467 SerB470 SerB470 GluB467 GluB467 LysB411 ThrB412 AsnB357 LysB475 ProA AsnA551 BsnA547 ProA AsnA547

3 3 3 3 3 3 4 4 4 4 4 5 5 5 5 .5 6 6 6 6 9 7 7 7 7 7

CB CG, CA CD, OEI CA, CB ODI CE, NZ C’G 0, CG, ODl CG, ODI. ND2 CB ODI OG CO OG 0, CB, OG OEl CG, CD, OEl, OE2 C. 0 0 CG ODI, ND2 (‘E’ 0 C”G, 0, C, ODl, ND2 ND2 CB 01~1

code Translation

Symmetry

2 2 2

operation

Local dyad A -+ B -x,Y,Z x+1/2,--Y+1/2,-z -x,Y,Z -X+1/2,Y+1/2,-2 x+l/2,-Y+l/2,-z -x,Y,Z x+1/2,y+1/2,-z x+1/2,--Y+1/2,-2

TX 1 0 1 0

- -1 1 - -1 0

component

TY 0 0 0 -1 0 0 0 0

TZ 0 0 -1 1 1 1 0 1

A. Messerschmidt

194

Figure 8. Stereo drawing of the type-l

copper site in domain 3. The displayed bond distances are for subunit A

We examined by modelling whether by side-chain torsion angle rotations His62 and His450 could become bridging ligands between CU4 K3 and CU2 K3 or CU3 K3 with both ring nitrogen atoms as ligating but found this stereochemically unreasonable. The copper-copper and copper-ligand distances of the trinuclear copper site are displayed in Table 10 for both subunits. Their mean values (Cu-N, 2.09 ((T = 0.06); Cu-0, 2.02 (0 = 0.02)) are comparable to

Table 10 Copper-copper and copper-ligand distances in AOase compared to copper-ligand distances for the type-l copper site in plastocyanin and azurin Atom1 (‘c’l &ll CIJl

Kl Kl Kl

Atom 2 (‘U2 &J3 CU4

Distance (A) A B 12.20 12.20 12.73 12.65 1487 1486

K3 K3 K3

Type-l

2.10 2.13 2.05 2.90 478

copper site

GUI CZJl &I GUI WI

Kl K1 Kl Kl Kl

ND1 SG ND1 SD 0

4454 507 512 517 444

CU2 CU2 CU3

K3 K3 K3

CU3 CU4 CU4

K3 K3 K3

3.68 3.78 3.66

3.73 3.90 369

CU2 (‘U2 / CU2 CU3 CU3 CU3 CU4 CIT4 CU“1 CU3 CU4

K3 K3 K3 K3 K3 K3 K3 K3 K3 K3 K3

NE2 NE2 NE2 ND1 NE2 NE2 NE2 NE2 OH1 OH1 OH3

106 450 506 62 104 508 60 448 K3 K3 K3

2.16 2.06 207 1.98 2.19 2.14 2-00 2.09 2.00 2.06 2.02

2.17 2.09 2.09 2.13 2.14 2.08 2-02 2.05 2.00 2.00 2.03

Trinuclear

2.12 203 211 2.83 483

et al.

Azurint wtp5 wtp9 2.09 2.11 2.26 2.25 2.04 2.03 3.12 3.15 2.97 2.95

Plastocyanin 204 2.13 2.10 2.90 3.82

copper site

t Data from NW el al. (1991), wtp5, wtp9 recombinant wildtype structures from Pneudomoms aeruginosa azurin at pH 55 and 9.0. $ Plastocyanin from poplar leaves, data from Guss & Freeman (1983). 4 AOase numbering.

binuclear copper model compounds with nitrogen and oxygen copper ligands (Karlin et al., 1984: Chaudhuri et al., 1985) in the range of the accuracy of 0.1 A for the copper ligand bonds. There is no long Cu-N bond (about 2.7 A) as found in the bicopper centre of haemocyanin from nuclear interruptus determined by X-ray Panulirus crystallography at 3.2 A resolution (Volbeda dz Hol, 1989). The binuclear copper centre in haemocyanin has a trigonal antiprismatic arrangement of the six histidine ligands while the type-3 coppers in AOase show the trigonal prismatic co-ordination. The presence of the type-2 copper and its ligands would not, allow the former arrangement. The average coppercopper distance in the trinuclear copper site of AOase is 3.74 A (a = 0.08 A) and the individual distances do not deviate by more than 0.16 A from this mean value. The average copper-copper distance in haemocyanin is 3.54 A (Volbeda & Hol, 1989). The copper-copper distances are too long for copper-copper bonds but magnetic interactions are possible. The shortest distance between the type-l copper centre and the trinuclear copper centre is 122 A. The His506-Cys507-His508 amino acid sequence segment links the type-l copper centre and the type-3 copper centre as a bridging ligand (see Fig. 13). Aspects of the intramolecular electron transfer between the two redox crntres will be discussed in section (k) below.

(j) Substrate (i) Kinetic

binding

data for

sites and catalytic

mechanism

AOase and laecase

The numerous studies concerning the catalytic and redox properties of the blue oxidases are well documented in several reviews (see e.g. for lactase, 1984; for AOase, Mondovi & Reinhammar, Avigliano, 1984). Some important findings are summarized here. The type-l copper is the primary electron acceptor from the reducing substrate. The reduction of the type-l copper, monitored by the bleaching of the absorption band at 6OOnm, is a bimolecular second-order reaction. The rate constants of the anaerobic reduction are e.g. 1.6 x lo3 mol-’ ssr for t’ree lactase with hydroquinone as substrate at

Rejined Structure of Ascorbate Oxidase

Figure 9. Stereo plot of the overlay of the type-l copper sites for poplar plastocyanin aeruginoaa azurin (medium thick line), and AOase (thick line).

Figure

10. Rtrreo drawing of the trinuclear

copper site. The displayed

195

(thin line), Pwudomonas

bond distances are for subunit A

Figure 11. Stereo plot for HisB62 ligated to CU3 BK3 by its ND1 atom. At,omic model plus 2F, - F, electron density map contoured at I.0 o.

pH 7.4 (Andreasson & Reinhammar, 1976) and 1 x 10’ to 3 X 107 mol- ’ s- ’ for AOase from Zucchini with a one electron-reduced nitroaromatic (ArNO;~ ) as substrate at pH 7.5 (O’Neill et al.. 1983). Ascorbate as substrate for AOase should be at least as fast since the reaction is completed within the deadtime of the stopped flow instrument (3 to 5 ms; Kroneck et al., 1982). The k,,, and KMfor AOase with ascorbate as substrate are 8 x 103 s--l and 0.2 mM, respectively (Mondovi & Avigliano. 1984). kJK,,, is 4 x 10’ Me1 s-l, which indicates the low limit of the bimolecular enzyme-substrate reaction rate. One electron is transferred from ascorbate and the generated semidehydroascorbate spontaneously dismutates in solution (Y amazaki & Piette, 1961). The electron transfer from type-t to type-3 copper in AOase has been studied under anaerobic conditions (half-time for bleaching of optical type-3 signal, 330 nm band, about 5 to 10 min) and found to be too slow to fit the overall kinetics with a turnover number of 8 x lo3 s-l (Avigliano et al.. 1978). It may be that the presence of copperoxygen complexes significantly affects the intramolecular electron transfer by affecting the driving force (i.e. redox potential).

Figure 13. Stereo drawing trinuclear

copper

site between

The bleaching of the type-3 copper signal for tree lactase is monophasic at pH 6.5, with a bimolecular rat,e of 4.0 x 102 mole1 s 1 and biphasic at, higher with a bimolecular rate of pH-values 1.5 x 103 rnol~-’ s-l for the initial reduction and a unimolecular rate of 0.4 so ’ at pH 7.4, for the final reduction (Andreasson & Reinhammar. 1976). The reduction of the type-2 copper has been determined by rapid freeze quench experiments for tree laccasr (Andreasson & Reinhammar, 1979). It is as fast as the reduction of the type-l copper. It must br recalled however that the overall kinet’ics for t)his enzyme is significantly different from AOase. The unimolecular rates for the reduct’ion of the t’ype-3 copper are only relevant for the catalytic cycle assuming that the reducing substrate is only reacting with the type-l copper centre under turnover conditions. The slower reduction of the type-2 copper in AOase, determined by rapid freeze quench r.p.r. measurements (Kroneck et al., 1982), may be due to the binding of OH3 K3 (OH-- or H,O) to the type-2 copper (CL14 K3) changing the reduct’ion properties of the trinuclear copper- species and creating a resting form of the enzyme. The trinuclear copper site is the binding site for

of the region of the atomic model containing domain 1 and domain 3.

the type-l

copper

centre

in domain

3 and the

Rejned Structure of Ascorbate Oxidase the second substrate, the dioxygen. This has been concluded from reoxidation studies of the fully reduced enzyme by dioxygen or H,O, or from spectroscopic studies of the binding of azide t,o tree lactase (for azidr binding to tree lacacase see e.g. Spira-Solomon rt al., 1986; Cole et al., 1990). The reoxidation of t(he type-l copper (absorption band at’ 600 nm) and of the type-3 coppers (absorption band at 330 nm) occurs at equal rat’es of 5x10hmol~1s~’ both for tree lactase (Andreasson rt al.. 1976) and for AOase from Zucchini (Nakamura & Ogura, 1968). During the reoxidation is formed in with H,O,. an O--intermediate several minutes. which is document,ed for tree laccaw by changes in the c.d. spectrum (Farver et al.. 1978a) and for AOase from Zucchini in the formation of an absorption band at, 3.50 nm (Strot.hkamp & l)~~~son. 1978). The int.ermediates are stable for sev~al hours in both enzymes. An interwas found during the reoxidaesting int,ermrdiate t,ion of fully reduced tree lactase by dioxygrn (Andreasson e.f al., 1976). This intermediate caused the rapid formation of an absorption maximum at 360 nm and also affected signal. Tt deciiyeti slowly

the type-2

copper

in a first-order

e.p.r.

reaction

197

with a halftime of 20 seconds. A new e.p,r. signal at low temperature (10 K) due to this radical could be detected. The nature of this radical has been described as an O’- radical. The trinuclear copper site may store three electrons and transfer them effect,ively to the bound dioxggen followed by a final one electron transfer. Small anionic inhibitors like F ~. (‘1 N3 etc. do not affect the reduction of the t,ypr-I copper but, the reduction of the type-3 copper is strongly decreased. Tt has been concluded that these small inhibitors bind to the type-2 copper because they alter the type-2 e.p.r. signal. On the basis of these results. the slow intramolecular type-2 copper reduction rate found for lactase was explained by the existence of a rtIrluc~tion-inactive form caused by thr binding of an OF-l- to the type-2 copper. This binding is pHdtll)endent and different, in fungal and I we lawase. (ii) Binding

site of the reducing

orgwnic

s~~h.stmtr

.A (!onnollF surface (Connolly, 1983) of’ a subunit of’ AOase was calculated in order t,o identify a possible binding pocket for the organic substrate nfkar the type-l copper centre and possible c+hannels

Figure 14. Stereo drawing of the binding site of the organic substrate near the type-1 uopprr site viewed parallel to the ND1 His61N’Ul KI bond. Atomic model plus Connolly dot surface. (a) Empty brndinp pocket. (b) Rinding pocket plus

Figure 15. Stereo drawing of the binding site near the type-l copper plus docked I,-ascorbate. Atomic model plus Connolly dot surface. (a) Viewed perpendicular to the NI)l His512-CC1 Kl bond. (b) Water molecules bound in the crystal included.

to the trinuclear copper centre allowing access for dioxygen and release of the product water There is a marked depression in the Connolly surface (see Fig. 14(a)) prdviding access to the type-l copper centre from the solvent. Tn the this depression is filled with water crystal, molecules. Docking experiments with r.-ascorbate show that this pocket is very well designed t,o accommodate it. The energetically most favourable docking is obtained if the lactone-ring of L-ascorbate is oriented parallel to the aromatic ring of Trp163 and the exocyclic side-chain points away from Trp163 as depicted in Figure 14(b). Three residues of the pocket seem to be of particular importance for the binding of the substrate and the electron transfer from the substrate to the type-l copper: His512, located at the bottom, is bonded by its ND1 atom to the type-l copper and its SE2 atom is accessible for solvent. The aromatic ring of’ Trp362 is approximately parallel to the imidazole ring of His512. Its NE1 atom is also accessible for

hydrogen bonding. The aromatic ring of Trp163 forms t’he left wall of the pocket as shown in Figure 14. Figure 15 is a view of the binding pocket and the docked L-ascorbate molecule from inside the molecule. The type-l copper site is located at the bottom of this Figure. The side-chain of His512, Trp362 and Trp163 and the lactone ring of the r>-ascorbate molecule form a staggered system of cyclic molecules with conjugated bonds. Trp362 is conserved in lactase from Neurospora crassa but not in human ceruloplasmin and Trp163 is not, conserved in either of them (Messerschmidt. & Huber, 1990). It is interesting to note that three of the water molecules (SOLlS, SOL638, SOL1045) bound in the pocket in the crystal structure are close to the oxygen functions of the la&one ring in the optimally docked L-ascorbate molecule (see Fig. 15(b)). O-l and O-2 of L-ascorbate are in hydrogen bonding distances to NE2 of His512 and NE1 of Trp362, respectively in the hypothetical complex.

Re&ed Structure of Ascorbate Oxidase

Figure Atomic

16. Stereo drawing of the channel leading model

plus Connolly

dot surface.

The water

to the CUP K3-CU3 K3 copper molecules present in the crystals

(iii) Channe1.s to the trirmclear copper site Two channels could be found in the Connolly surface that provide access for solvent molecules to the trinuclear copper species. The first channel leads to t,he type-3 CU2 K3-CU3 K3 copper pair and is depicted in Figure 16. It is filled with water molecules in the crystals. This channel is very broad. The second channel is directed towards the CIJ4 K3 copper at,om (the spectroscopic type-2 copper) and is shown in Figure 17. This is narrow but filled by water molecules in the crystals, which form a hydrogen bonded chain connected directly to the oxygen ligand OH3 K3 or CXJ4 K3. The location of the channels providing access to the trinuclear copper centre is shown in Figure 18 for the dimer present in solution. All channels are well accessible from the solvent,. The binding pocket for the organic substrate (not specially labelled) approaches the type-l copper sites from the top of t,he subunit,s in t.his Figure.

Figure 17. Stereo drawing model

plus (Connolly

199

pair of the trinuclear are displayed. too.

copper

site.

(iv) Conclusions concerning the catalytic mechanism A model for t.he catalytic cycle as suggested in Figure 19 is based on assumptions about the structures of reaction intermediates found in reoxidation experiments of dioxygen with fully reduced lactase and on the course and rates of reduction and reoxidat,ion of the copper centres. During its reaction with fully reduced lactase dioxygen binds to the trinuclear copper species and three electrons are very rapidly transferred to it resulting in the formation of an “oxygen intermediate” with a characteristic: optical absorption band near 360 nm (Andreasson et al.. 1973, 1976) and a broad low temperature r.p.r. signal near g = 1.7 (Aasa et al., 1976a,b). The type-l copper is concomitantlv reoxidized when the low temperat.ure e.1j.r. signal “is formed. The oxygen intermediate decays very slowly (tli2 - 1 to 15 s) correlated with the reoxidation of the type-2 copper (HrB;nden & Drinum, 1978). The release of the produczt water has

of the channel leading to the CU4 K3 (type-2 copper) of the trinuclear dot surface. The water molecules present in the crystals are displayed. too.

copper

site. Atomic

200

A. Mmserschmidt

et al.

-___

-_-.------.

Figure 18. Stereo C”-plot of the dimer of AOase present in solution with residues labelled. located at the entry of the channels leading to the trinuclear copper centre.

been studied by mass spectroscopy of the reaction between ‘*O, and reduced tree lactase by Brlnden et al. (1978). The enzyme is fully reduced by addition of four equivalents of reducing substrate under

anaerobic conditions and 1802 is added. Mass spectrometry shows that one water molecule is released fast but the second oxygen remains bound (as water or OH-) to the type-2 copper as equatorial ligand, determined from the e.p.r. spectrum (Branden & Deinum, 1977) and is slowly exchanged with the bulk solution in a half-time of 30 minutes. The mode of dioxygen binding to the trinuclear copper species is essential for an understanding of the catalytic mechanism. Several binuclear copper

Figure

19. Proposal for the catalytic

mechanism of AOase

compounds have been synthesized as models for dioxygen or azide binding to haemocyanin or tyrosinase. Reversible binding of dioxygen to a binuclear copper compound has been reported by Karlin et al. (1987). It is a phenoxo-bridged binuclear Cu(l)-complex with three nitrogen ligands for each copper. The copper-copper distance is 3% to 3.7 A and the binuclear centre is accessible for small molecules like dioxygen. Dioxygen binds as cis 1,2-p-peroxo species and the coppers are penta-coordinated with tetragonal pyramidal symmetry. A novel mode of dioxygen binding to a binuclear copper complex is found in the compound syntheet aE. (1989). The complex sized by Kitajima

Refined Structure of Ascorbate Oxidase

0

END-ON

X EXTERNAL BRIDGE, p-1,2 @I INTERNAL BRIDGE, p-1,2 .

SIDE-ON, cl-r1*:$

@ ALTERNATIVE EXTERNAL BRIDGE,p-12 Figure 20. Selected putative binding modes of dioxygen to the trinuclear copper species in .40ase.

contains the peroxide bound in ,u-1’ : 1’ mode i.e. side-on between the two copper(I1) ions. The copper co-ordination is a distorted tetragonal pyramid with the N,02 ligand donor set. An alkoxide-bridged binuclear complex with azide bound in 1,3-bridging mode is reported by McKee et al. (1984). The copper(copper(H) distance is 3.62 A. The copper (II) ions are penta-co-ordinated by four nitrogen atoms and one oxygen atom with tetragonal pyramidal symmetry. Other binding modes of dioxygen to mononuclear and binuclear metal centres are also seen in coordination compounds such as side-on and end-on binding to mononuclear centres. Such geometries are also possible with binuclear metal centres. Tram-p-1,2-bridging and p-1 ,l -bridging binding modes are also found in co-ordination compounds. An end-on binding of dioxygen to a binuclear iron centre has been established in the X-ray structure of oxy hemerythrin (Holmes et al., 1991). The dioxygen is bound as protonated peroxide. Model compounds with trinuclear copper centres in a triangular arrangement as in AOase have been prepared (see e.g. Ahlgren et al., 1980; Comarmond et al., 1985; Karlin et al, 1989). None of the known triangular trinuclear copper complex compounds, however, has been shown to bind dioxygen or reduce dioxygen to water. In the absence of an experimental structure we modelled dioxygen and azide binding to the tri-

201

nuclear copper centre in AOase. Dioxygen and azide molecules were added to the AOase co-ordinate data set and positioned interactively. At these positions, stereochemistry was checked for close contacts (for explanation of the binding modes discussed below see Fig. 20). We checked 1,3-bridging and end-on arrangements for azide binding. A 1,3-bridge is possible only between CU2 K3 and CU3 K3 by substituting the bridging oxygen OH1 K3 ligand. External 1,3bridging of CU2 K3 and CU4 K3 or CU3 K3 and CC4 K3 would interfere with surrounding histidine residues (distances less than 2-O A). Any of the three internal 1,3-bridging arrangements of the azide seems impossible as the central N-atom of the azide molecule would be in close contact of about 1.5 A to the opposite copper atom. A most plausible binding mode for azide seems therefore end-on binding at CU4 K3 as depicted in Figure 21. Azide would replace OH3 K3 and one or both water molecules 84, 219 (see Fig. lo)?. The type-3, type-2 copper bridging arrangement suggested by Cole et al. (1990) to explain spectroscopic properties seems sterically impossible. Binding of azide in AOase crystals was found quite similar to that in solution with an overall similar affinity (Merli et aE., 1988). The trinuclear copper species is a coupled electronic system and binding of azide to the type-2 copper will affect the spectroscopic and magnetic properties of the whole system to explain the observations of Cole et al. (1999). Modelling of the binding of the smaller dioxygen to the trinuclear copper site shows less steric restrictions than azide. Dioxygen may bind end-on at’ CU4 K3 in analogy to the probable binding of azide or it may form internal &s-1,2bridges between any pair of the copper triangle and a side-on bond to the opposite copper. This latter mode with dioxygen bridging CU2 K3 and CU3 K3 is depicted in Figure 21. If the bridging oxygen ligand OH1 K3 is absent in the fully reduced enzyme, dioxygen may also be bound as either external 1,2-bridge or side-on between the type-3 1,2-bridges to the type-2 copper ions. External copper sterically interfere with the His-ligands. Dioxygen binding in the copper triangle allows close int,eraction and rapid electron transfer from all three reduced copper ions, the negative charge developing at the dioxygen or oxygen intermediates may be balanced by the copper ions and the protein protected from oxygen radicals. The proposed catalytic model is schematically outlined in Figure 19. We assume that we observe in the crystal structure the resting form of the fully oxidized enzyme. Oxygen ligand OH3 K3, which may be OH- or Hz0 is presumably the inactivating ligand of the tvpe-2 copper (Fig. 19(a)). Upon reduction OH3 K3 k removed

slowly

causing

a lag phase

t Preliminary experiments with &de-treated crystals showed no significant difference electron densities in accord with a replacement of bound groups by azide.

202

.4 MrsserCschmzidt et al

Figure 21. Stereo drawing of the atomic model around the trinuclear copper site plus modelled azide binding in end-on mode to CU4 K3 and dioxygen binding as trans-1,2-p-peroxo to CU2 K3 and CU3 K3. Atom N-2 of the azide molecule is not labelled. of the enzymatic reaction. After reduction with four equivalents of reductant the fully reduced enzyme is obtained (Fig. 19(c)), which is able t,o bind dioxygen. It may be bound as three-electron reduced \0,],3species (Fig. 21 and Fig. 19(d)). This species must accept one or two protons to release one mol of OH- or water. The protons may be provided by bulk solvent in the channels (Fig. 10) perhaps mediated by the histidine copper ligands. After release of the first water molecule the oxygen radical intermediate is formed (Fig. 19(e)). This will be reduced by the remaining electron on the type-2 copper. The oxidation of the type-2 copper is fast under normal turnover conditions. However, the oxygen intermediate decays slowly when stoichiometric amounts of dioxygen are reacted with fully reduced lactase. Fast decay of the intermediate may require that’ re-reduction of the oxidized copper centres has already occurred as may be the case in turnover. Oxygen ligands show indeed weaker binding to copper(I)-ions (see e.g. Karlin 8 Gultneh, 1987). Figure 19(f) shows the last step of the catalytic cycle. All four copper centres are molecule accepts two protons and oxidized, the 0’ is released as second water. Figure 19 was drawn assuming a p-1,2 internal11 bridging dioxygen. In this configuration, partially reduced dioxygen species seem t,o be most stabilized by interaction with all three copper ions. On the other hand, end-on binding of dioxygen at the type-2 copper could very well explain the formation of the resting enzyme form whereby the first water molecule is released rapidly from the dioxygen bound enzyme, but the second oxygen atom remains bound presumably as the OH3 K3 ligand of the type-2 copper. Side-on binding of dioxygen between the type-3 coppers as proposed by J. Reedijk (personal communication) would imply the

removal of the OH1 K3 bridging turnover. (k) Electron (i) Electron

transfer

transfer

ligand

during

the

processes

to the type- I copper redox centrr

As previously mentioned, the electron transfer from one electron-reduced nitroaromates (,4r?u‘O; ) to the type-l copper centre takes place in a bimolecular second-order reaction with rates compatible with the turnover number with ascorbate as reducing substrat.e. The electron transfer from ascorbate to the type-l copper centre can be even faster and is completed within the deadtime of the stopped flow instrument. It is therefore not the ratelimiting step in the overall reaction. The factors governing electron transfer may be described wit’hin the framework of Marcus electron transfer theory (see Marcus & Sutin, 1985). They are expressed in terms of driving force, distance of redox centres, reorganization energy, etc. The driving force. calculated from the difference in the redox potentials ( + 350 mV for t,he type-l copper in AOase: Mondovi & Avigliano, 1984; Kroneck et al., 1982; +295 mV for the couple ascorbate/ascorbat,e free radical: Farver et al., 19786), is 55 mV. In thr proposed modelled encounter complex there is a short distance of about 7 a between t’he two redox centres (distance CUl AKl-01 ASCl = f?Sa: distance CUl AKl-02 ASCl = 7.5 8) and an effective parallel arrangement of the rings with good overlap of the n-electron density systems facilitating a rapid electron transfer (see Fig. 15). It is well documented in small blue copper proteins, such as plastocyanin that there are minimal structural changes on reduction and reoxidation (Guss et al., 1986). The reorganization energy is probably small. therefore.

Refined Structure of Ascorbate Oxidase (ii) Intramolecular electron transfer from the type-l copper centre to the trinuclear copper centre Long-distance intramolecular electron transfer can be described in the frame of the theory of Marcus (see Marcus & Sutin, 1985). In the formulat,ion of Lieber et al. (1987), the intramolecular electron rate constant k,, can be written as k,, = v,T exp( -AG*/RT),

(1) with v,. nuclear frequency factor, normally 1Ol3 s- ‘; r, electronic factor; AG*, activation free energy for ET. The electronic factor, r, is unity when the donor and acceptor are strongly coupled but’ is much smaller at long donor-acceptor distances. In such cases. r is expected to fall off with distance (d):

r = W,kxpl-BPd,)l,

(2)

with d,. van der Waals contact distance, normally taken as 3 A (Marcus & Sutin, 1985); /?, electron coupling factor, decreases with increasing distance and depends on the nature of the intervening medium. AG* depends on the reaction free energy, AG’: and the nuclear reorganization energy 2 (Marcus & Sutin, 1985) according to the equation: AG* = (AC0 + i)2/41.

(3)

The ET rate is maximal when -AG” = 1. The shortest distances, d, between the type-l copper and the copper ions of the trinuclear copper site are 12.2 a (CC1 Kl-CU2 K3) and about 12.7 A (CUl Kl-CU3 K3) (see Table 10). Figure 13 shows that His506-Cys507-His508 serve as bridging ligand between the two redox centres providing a bifurcated pathway for ET from the type-l copper centre to the trinuclear copper species. The difference in redox potential of the type-l copper centre and the type-3 copper ions, the driving force, measured at lO”C, is -AC” = 41 mV (Kroneck et aE., 1982). However the binding of dioxygen to the partly reduced protein and the presence of reduction intermediates may affect this redox potential (Avigliano et al. (1978) ha,ve found a very slow equilibration between t,ype-1 and type-3 coppers in AOase in the absence of dioxygen). For the reorganization energy, 2, and the electronic coupling factor, p, no estimates can be derived for AOase but reasonable values for proteins are 3, = 1 eV and Jl = 1.2 A-‘, according to Gra’y & Malmstram (1989). These values inserted into equation (1) yield k,, - lo5 s- ‘. Changing p t,o I.6 &’ gives k,, - 4x lo3 s-l, a value closer t,o the observed turnover number of 8x103s-‘. McLendon (1988) suggested t’hat the electron transfer in proteins may not be designed for very fast intramolecular ET with the exception of lightinduced ET in photosynthetic reaction centres. They could even be designed to slow down these rapid rates, which might otherwise lead to biological “short circuits”. Related to this point is the observation that. maximal rates for intramolecular

203

ET in organic donor-acceptor molecules with rigid spacers are significantly faster than those for Ru-labelled protein systems at similar distances (Mayo et al., 1986). In the case of AOase, the observed ET rates for the reduction of the type-3 copper ions with reductate as substrate are in the range of the turnover number (Kroneck et al., 1982). Similar results have been obtained by Meyer et al. (1991), who studied the anaerobic reduction and subsequent reoxidation of the type-l copper by lumiflavin semiquinone using laser flash photolysis. These findings indicate that the intramolecular ET from the type-l copper centre to the trinuclear copper site is rate-limiting in the catalytic cycle. ET from the type-l copper to the type-3 copper pair of the trinuclear copper site may be throughbond, through-space or a combination of both. A through-bond pathway is available for both branches, each with 11 bonds (see Fig. 13). The alternative combined through-bond and throughspace pathway from the type-l copper CUl Kl to CU2 K3 of the trinuclear centre involves a transfer from the SG atom of Cys507 to the main-chain carbonyl of Cys507 and through the hydrogen bond of this carbonyl to the ND1 atom of the His508. (iii) Electron transfer within the trimuclear copper site Electron exchange within the trinuclear copper site is expected to be very fast due to the short distances between the copper atoms (about, 3.7 A) as will be ET to the bound dioxygen. The exact geometry of bound dioxygen seems to be a major unsolved quest’ion, which we will approach by structure analysis of derivatives like the H,O, species. We will prepare this in the crystals and study its structure. A further objective for our studies would be the elucidation of the structure of the fully reduced enzyme. A.M. thanks t,he Deutsche Forschungsgemeinschaft’ (Schwerpunktthema: Bioanorganische Chemie) for financial support. We thank D. Turk for making his program MAW available for displaying and plotting the Connolly surfaces and for the docking experiments. The technical assist,ance of Mrs K. Epp is acknowledged.

References Aasa. R., BrB;nden, R., Deinum, J., Malmstrijm,

B. G.,

Reinhammar. B. & Vgnngard, T. (1976u). A paramagnetic intermediate in the reduction of oxygen by reduced lactase. FEBS Letters, 61, 115-119. Aasa. R.. BrSnden, R., Deinum, J., Malmstriim, B. G., Reinhammar, B. & Wnngard, T. (19766). A 170-effect on the EPR spectrum of t,he intermediate in the dioxygen-lactase reaction. B&hem. Biophys. Weu. Commun. 70, 1204-1209. Ahlgren. M., Turpeinen, U. & Smolander, K. (1980). The structure of bis[p-acetato-(O,O’)-diacetatop3 - (2 - diethylaminoethanolato - ru‘,p,-0) - p(2-diethylaminoethanolato - N,p - 0) - p3 - hydroxo - tricopper(II)] hydrate. Actu Crystallogr. sect. B, 36, 1091-1095. Andreasson. L.-E. & Reinhammar, B. (1976). Kinetic studies of rhus vernicifera lactase. Role of the metal

centrrs in electron transfrr. Biochmc. Eiqhys. =Ic/n. 445. 579~-597. Andreasson. L.-E. & Rrinhammar, B. (1979). The mecahanism of elrct,ron transfer. in lactase-catalysrd reinCons. N&Aim. Biophys. Acta, 568. 145-156. Atrdreasson. L-E.. Kriinden, R.. blalmstrtim, B. (:. & VSnngard. T. (1973). An intermediate in the reaction of reduced lactase with oxygen. FEHA’ Lettrrs. 32. 187-1x9.

.4ndrrasson. L.-E.. ISnden, R. $ Reinhammar. B. (1976). Kinetic studies for R~US oernicif~ra lactase. Evidence for mult.ielectron transfer and oxygen interreact ion Ki0clr.i lib. mediate in the reoxidat.ion Biophzyx

Actu,

438.

370~-379.

Avigliano. I,.. Rotilio. G.. lirbanelli. 6.. Mondovi. I?,. & Finazzi-Agrh, A. (1978). Anaerohica reaction of ascaorBiophys. bate oxidasr with ascorbatr. Arch,. Biochun.

1)‘Andrea. (l.. Kowst,ra. .J. B., Kamrriing. .I I”. A \‘licqrnt,hart. F. (:. (198X). Primary htru(.t urt’ of tht, xvlost,-c,ontainin# X-linked c*arbohyclrat,c. moic~t~ from asc~orbat,c~oxidasr of C‘~P/r?liitfr /j~/)f) rn~flullosfr.

185. 41!)-422.

R3i --Sk I

Baker. E. N. (1988). Struct,ure of azurin from Alcaliyun~~ dunitrificnns. Refinement of I.8 A resolution and cornparison of the two cr)-stallographically independent molecules. .I. Mol. Biol. 203. 1071-1095. Baker. E. S. & Hubbard. R. E. (1984). Hydrogen bonding in glokular proteins. Prog. Hiophys. Mol. Eol. 44. !J7 -1i!). Barlow, I). J. & Thornton. J. M. (1983). Ton pairs in proteins. ./. Nol. Biol. 168, 867-885. Barlow, I). J. & Thornt’on, *J. M. (1988). Helix geometr) in proteins. ,I. Mol. Viol. 201, 601-619. Bernst’rin. F. (1.. Kortzle, T. F.. Williams. (:. J. H.. Meyer. E. F.. -Jr, 1Srice. M. D.. Rodgers. .J. R., Krnnard. 0.. Shimanouchi. T. 8: Tasumi, M. (1977). The protein data bank: a computer-based archival file for macromolecular structures. J. Mol. Bid. 112. 535~54”. Kr%nden. R. & IIeinum, ,J. (1977). Type-2 copper as a c.omponent of the dioxygen reducing site in laccaase: evidence from epr experiments with “0. FEnA’ fAtrrs, 73. 1444146. Hr%nden. R. & Deinum, J. (1978). The t+Jec and electronic, strurture. J. Arrcer. (‘hem. Sot. 112. 9534 Y548. (‘omarmond. ,J., IIiet,rich. IS., Lrhn, ,J.-%I. & Louis. 1%. ( 19%). Cu” complexation by a trit.opic* macrocyclic ligand. A doubly p,-hydroxo hridped trinuclrar

Farvrr. 0 . (:oldberg. M. & I’rcaht. I. (l9780). (‘irc~ulitl dichroic~ spe&rurn of the lac,~ase-~)rroxiclr drrivativc>. FEHS I,Pttrrs, 94. :3x:3-386. Farver. 0.. Goldbt~rg, 11.. MIirrland. S. & I’c~c~ht. I. (197X6). Rrtiu~tant-tlrJ)t~ntlrnt electron distrihutioll around rrdos sites of tacbcasti. I’TOC .Vnl. .Ac&. SC;.. 17.S..4. 75. 5245-524n. Fw. .J. .I. ( 1975). (‘oJjJ)t’r-J)rot,t‘in systems (.ontaining thus ‘blue’ c~oppt~rc*rnt,rr. Str~&. Iiond. 23, I 60. Germann. 1’. .A.. Miiller. (i.. Hunziker. I’. It. Cy:I,c~r&. K. (1988). t ‘haract’erization of two allelic fi)rms ot Xrurospom XX6 X!NL

c’mssn

lac~xsr.

.I.

11iol.

(‘hrn~.

263.

(iray.

H. IS. Hr Malmstriim. IS. (:. (lSH9). Lolu-rangr transfer in multisitr metallo~jrot~rins. Bic~chumi.stry, 28. i49!1 7506. GUS. .I. 11. & FreemaIl. H. (‘. (l!M). Structurr ot oxidized poplar ptast,ocavanin at 1.li A r.c*solution. .J. Mol. Niol. 169. ,521~+ja. (:uss. .J. 11.. HarrowelI. I’. I?,.. Murat,a. M.. Xorris. I’. .\. K: Freeman. H. C’. (1986). C’rystal st.ruc:turc. analyses 01’ reducwl (f ‘u’) poplar plast,ocyanin at six pH values. rlectIYm

.I. Mol.

Bid.

192. 361 Xii.

Holmes. N. A.. LrTrong. I.. Turltry. S.. Siektar,. I.. k Sttankanrf). R. E. (1991). Struct’urrs of droxy and ox! hemer?,thrin at PO A resolution. ,J. Nol. Nini. 218. 5x3 5!X3,

Huber, R. & St,eigrmann. \Y. (1974). Two c.is-prolines in the Bencr-#Jones protein rpi and the As-Pro-bend. FEJLT

Lrttrrs.

48.

“X5

237.

Hvosleff. .I. (1!+69). C’hangrs in c~onforrnation and bonding of ascaorhic.actid by ioniza,tioll. The crystal strucsturc, of sodium asc*orbwtr. .-l&1 ( ‘rptcrllogr. .wct. I{, 25. 29,i 4 r”““~. b&V

#Jones.‘I’. ;\. (I!JiX). A graphics rnodrl building xystrm for m:r~roniolrc~rrlrs. merit oy.slal/oyT.

and rrtintz-

.I

‘4[‘PI

15. “4 31,

Kabsch. IV. & Sander, S. (19X3). I)ictionary of J)rot,ein secondary struc*turtl: pat,trrn recognition of hydra gen-ljontled and geomrt,riVal featurt>s. Hio~d!/mws. 22. 26ii 2637. Karlin. K. I). 6 (:ultnrll. 1’. (l!kii). Binding anti wtiv;Ltion of molec*ular oxygen by copper c~miplexes. Pry. Inorq.

C’hc,vn. 35,

21%327.

Karlin. K. I).. Hayes. .I. C’.. (:ultneh. Y., C’rus~. II. IV.. PvlcKnown. ,I. IV., Hutchinson. .I. I’. Kr Zubieta. .I. (198-l). (‘oJ)J)er-mrdiatrt1 hydroxvlntion of MI arenr: modrl ,s.vstem for the ac+ion of copper ~iioIwos~genases. St.ruc*tures of a binuclrar (‘u(I) ~omplrx and its cPsygPn;rtrd product. ./. .4n,~ C’hPn?. SW. 106. “I”1

2IPX.

Re$ned Structure

of Asrorbate

Karlin. K. 11.. (‘ruse. R. LV,, Gultneh. Y.. Farooq. 8.. Hayes. ,J. (‘. bz Zubieta, ,J. (1987). Dioxygen-copper reactivity. Reversible binding of 0, and CO to a phenoxo-bridged dicopper(T) complex. J. Amw. (‘hem. Sot. 109. 2668.-2679. Karlin, K. 11.. (Gan. 4.. Farooq, M.. Lin. S. & Zubieta. .J. (1989). Synthesis and X-ray crystal structure of a trinuclcar copper(T) cluster. Inorg. C’hiru. dcta, 165. 37-m

Karlsson, (:.. Aasa. R.. Malmstriitn. H. (:. & Lundberg, I‘. (:. (1989). Rack-induced bonding iu blue copper proteins: sprct,roscopic properties and reduct.iotr potent.iaJ of’ t,hr azurin mutant &Jet-l 21 -T,ru. FEBS htfrrs. 253. !49- IO”. Kitajitne. N.. Fujisawa. K. & Moro-oka. 1.. (l!J89). p-q2 : q2-peroro binuclear copper complex. [(‘l~(HH(:S..i-il’r2~)z)3]202. .J. Amer. C’hrm. Sot. 111. X!f’ih---.897ti. Kronrc*k. P. M. H.. Armstrong. F. A.. Merkle. H. & Marcthesini. -1. (1982). Xscorbat,e oxidasr: molecular proJ)rrties and catalytic activity. In ArllNncrS it/ (‘hemistr!y

Srrics.

So. NO. Ascorbic

=Icid- C’hrmistry.

,J/f~trrbo/is,,~.n71.dI -SPS(Srib. P. A. HLTol bert. R. $1.. rds). J)J). 123 648. The American (‘heuticA Society. NW York. Lee. IJarcus. R. A. Rr Sutin. X. (198.5). Electron chemistry a11d biology. Biochim,. Biophy.s. 265

transfer Acta.

in

811.

X22

Mayo. S. I,.. Ellis. 11’. Ii.. *Jr, Crutchley, R. ,I. & Gray. H. 13. (19%). Long-range electron transfer in hrme proteins. S’cicncr, 233, 94X-9%. McKtv. \... Zvagulis. >I.. Dagdigian. ,J. \‘.. f’at(sh. $1. G. bi Rertl. (‘. .\. ( 1984). Hemocyanin models: synthesis. structure. ancl magnetic properties of a binucleating c:~~J)per(ll) system. .J. .4nLrr. f’hra/. ,Sov. 106. 4765 4772. hlcJ,entlon. (;. ( t!488). Long-distance electron transfer in prc,tcins and model systems. Arc. C’hr~n. Rus. 21. IN-- 167. Merli. A.. Jtossi. (:. I,.. Bolognesi. 11.. Gat,ti. I).. MorJ)urgo. I,. & Finazzi-Agro. A. (1988). Single crystal ahsorJ)tion spect,ra of ascorbate oxidase from green zucchini squash. FEBS Letters, 231. 89-94. MmsrrschmitJt. .A. Xr Huber, R. (1990). The blue oxidases. asc.orbatr oxidase. taccase and ceruioplasmin. Jlodelling and structural relationships. Eur. J. RirJchem. 187. 341J3.52. SJessrrschmidt. .\.. Rossi. A.. Ladenstein, I~.. Hubrr, R,.. Holoynrsi. $1.. (Patti. G.. Marchesini. A.. Petruzzelh,

Oxidn,+e

205

R. & Finazzi-Agro, A. (1989). X-ray cryst,al structure of t’hr blue oxidase ascorbate oxidase horn zucchini. Analysis of the polypeptide fold and model of the copper sites and ligands. J. A’ol. Biol. 206. 513-529. Mry~r. T. E.. Jlarchesini. A., (‘usanovirh. M. .I. & Tollin. (4. (1991 ). Direct measurement of intramolecular electron transfer between type-i and type-3 copper centres in the multi-copper enzyme. ascorbate oxidase, and its type-2 copper-depleted and cyanide 30. 4619 4623. inhibited forms. Riochemistry, Mondovi. J3.K: Avigliano. I,. (1984). .&orbat,r oxidase In (‘oppvr Prrdrins and Copper F:nz?jnce.s ( Lontie. I,., cd.). vol. 3. pp IO- 118. CR