Electronic structure of the oxidized and reduced blue copper sites: contributions to the electron transfer pathway, reduction potential, and geometry

i

d SSlg ELSEVIER

Inorganica Chimica Acta 243 (1996) 67-78

Electronic structure of the oxidized and reduced blue copper sites: contributions to the electron transfer pathway, reduction potential, 1 and geometry Edward I. Solomon*, Kevin W. Penfield, Andrew A. Gewirth, Michael D. Lowery, Susan E. Shadle, Jeffrey A. Guckert, Louis B. LaCroix Department of Chemistry, Stanford University, Stanford, CA 94305, USA

Abstract

This is a review of our group's studies on the electronic structure of the blue copper site and its contribution to function. It starts with the known crystallographic results and demonstrates how spectroscopy allows these results to be extended to obtain fundamental insight into reactivity. These studies: (1) demonstrate that the unique spectral features associated with the oxidized blue copper site reflect a novel ground state wavefunction with high anisotropic covalency which plays a key role in activating specific electron transfer pathways; (2) provide the first description of the electronic structure of the reduced blue copper site obtained using variable energy photoelectron spectroscopy as a new method of bioinorganic spectroscopy and (3) determine the change in electronic structure on oxidation which allows for an evaluation of the Jahn-Teller distorting forces present in the oxidized blue copper site. The last result provides insight into whether the reduced geometry is in fact imposed on the oxidized site by the protein (i.e. the 'entatic' or 'rack' state).

Keywords: Electronic structure; Oxidized blue copper sites; Reduced blue copper sites

1. Introduction Blue copper sites are found in many mononuclear copper proteins (plastocyanins, azurins, stellacyanin, etc.) and multicopper enzymes (laccase, ceruloplasmin, ascorbate oxidase, nitrite reductase, etc.) [1]. In the former they function in inter-protein electron transfer while in the latter they are involved in intra-molecular electron transfer to another copper center which is the site of small molecule reduction. These sites generally exhibit rapid electron transfer (k = 103--105 M -l s-Z) over a distance of -13/~,. They also have high reduction potentials (e.g. plastocyanin's is - 3 7 0 mV). A characteristic that has attracted much interest in these protein sites is their unique spectral features in the oxidized state (see below) relative to normal cupric complexes [2-4]. Such differences have

* Corresponding author. Tel.: +1 415 7239104; fax : +1 415 7250259. I This paper is dedicated to Professor Harry B. Gray.

0020-1693/96/$15.00 © 1996 Elsevier Science S.A. All fights reserved SSDI 0 0 2 0 - 1 6 9 3 ( 9 5 ) 0 4 8 9 3 - 1

lead to the idea of an 'entatic' or 'rack' state being imposed on the copper center by the protein [5-8]. In partitular the reduced geometry has been thought to be imposed on the oxidized site. Cu(I) often is found with an approximately tetrahedral geometry while Cu(II) tends to be tetragonally coordinated due to the Jahn-Teller effect. Thus, the entatic or rack state could be described as the protein opposing the Jahn-Teller distorting force present in the oxidized site. Our research has focused on three topics concerning the blue copper sites, each of which is reviewed below [4,9,10]. First, the unique spectral features of the oxidized site are now well understood. These reflect a highly covalent ground state wav.efunction where the covalency involves a specific ligand-metal bond which activates this bond as a pathway for rapid electron transfer. Second, the reduced site is d l° and thus has a closed shell configuration which is not accessible to the usual spectroscopic methods in bioinorganic chemistry. We have thus developed variable energy photoelectron spectroscopy as a new

68

E.I. Solomon et aL / lnorganica Chimica Acta 243 (1996) 67-78

A

6000 His 87

"

_

600 500

400O

400

3000

300

Cys84

. ~ N ~ ~

I I Blue Copper

C

gI= 2.00 cJix I

50O0

mOO His 37 ' - ~

i I Blue Copper

200

1000

A~r = 164X10"4cm'~

100

NormalCopper

I

25000 20000 15000 10000

2500

E n e r g y (cm "1)

I

I

/] ~J

3100 3300 Field (Gauss)

2700 2900

Fig. 1. Blue Copper proteins: (A) X-ray structure of poplar plastocyanin (figure adapted from Ref. [ 12]). (B) Absorption spectrum of plastoeyanin (left scale) and 'normal' D4h-CuCI42- (fight e scale). (C) X-band EPR spectrum of plastoeyanin and D4h-CuCI42-. spectroscopic probe of reduced copper sites and, combined with self consistent field-Xa-scattered wave (SCFXcz-SW) calculations, have used this technique to evaluate the electronic and geometric structure of the reduced blue copper site relative to normal ligand-Cu(I) bonding. This study indicates that the protein imposes a long thioether S--Cu bond on the site which reduces the donor interaction of this ligand and raises the reduction potential of the site. Finally, from the above studies we have evalu-

A

a

ated the change in electronic structure which occurs onoxidation of the blue copper site. The results o f this study have allowed us to determine the associated change in geometric structure which would occur for an unconstrained blue copper center. T h u s we have evaluated whether the limited geometric change observed upon oxidation, which contributes to rapid electron transfer, is in fact imposed by the protein on the copper center (i.e. an entatic or rack state) [5-8].

C

B

Experimental 0=5"

1

Simulated gll = z 5o CMet~

e=5~

CMet

0=85"

SMet

O= 85" ~1

I I

d x2__y2 I

2500 3000 3500 2500 3000 3500 Field (gauss) Field (gauss) Fig. 2. Single-crystal EPR of poplar plastocyanin; orientation of the dF _ yz orbital. (A) Unit cell and molecular orientation of poplar plastocyanin with respect to the applied magnetic field. 0 is the angle between the magnetic field (H) and the crystallographic c axis (e.g. 0 = 0 implies that H and c are coilinear). (13)EPR spectra and simulations for the crystal orientations shown. (C) Orientation of the gll direction and the dx2_ yz orbital superimposed on the blue copper site.

69

E.L Solomon et al. / lnorganica Chimica Acta 243 (1996) 67-78

2. Electronic structure o f the o x i d i z e d blue c o p p e r site: c o n t r i b u t i o n s to electron transfer p a t h w a y s

Harry Gray, Jeff Hare and I predicted from spectroscopy that the oxidized blue copper site would have a distorted tetrahedral geometry with a cysteine thiolateCu(II) bond [11]. The first crystal structure of a blue copper site was that of oxidized plastocyanin by Hans Freeman et al. in 1978. The crystal structure determined by Guss and Freeman is shown in Fig. 1A [12]. Rather than the normal tetragonal geometry of Cu(II) the site has a distorted tetrahedral structure. It also has two particularly interesting bonds. A short thiolate S-Cu(II) bond at 2.13 ,/~ of the cysteine (Cys) 84 and a long thioether SCu(II) bond at 2.9/~ of methionine (Met) 92. The other two ligands are fairly normal imidazole N--Cu(II) bonds of histidine (His) residues. Associated with this unusual geometric structure are the unique spectral features of the blue copper site [2-4]. In the absorption spectrum in Fig. 1B, there is an extremely intense band at 600nm (16 000 cm -1) with an emax = 5000 M -! cm -1 where norreal cupric complexes usually have weak LaportE forbidden d---> d transitions with emax = 40 M -~ cm -l. In the electron paramagnetic resonance (EPR) spectrum shown in Fig. 1C, the parallel hyperfine (Au) splitting of the blue copper site is reduced by more than a factor of 2 relative to that of normal cupric complexes. As shown below, these unique spectral features reflect a novel ground state wavefunction. This is the Cu(II) half-occupied HOMO level which is the redox active orbital and thus plays a

A

B C

key role in the electron transfer function of this site [3,4,13]. As can be seen in Fig. 1C, gll > g_L > 2.00 for the blue copper site which from ligand field theory indicates that the half occupied orbital is dxa_y2. We determined the orientation of this orbital in the distorted protein site in Fig. 1A through single-crystal EPR spectroscopy [14]. As shown in Fig. 2A, the protein crystallizes in an orthorhombic space group with four symmetry-related molecules in the unit cell [12]. If the crystal is oriented such that the magnetic field (H) is in the b c plane and the crystal is rotated around the a axis, the set of spectra shown in Fig. 2B is obtained. A gll spectrum (i.e. one with four hyperfine lines) is observed with the magnetic field approximately along the c axis while a g.L signal (i.e. one with no resolvable hyperfine splitting) is observed with the field rotated 90 °. Thus, gll is approximately along the c axis which is the approximate orientation of the long thioether S--Cu bond of Met 92 (Fig. 2A). A series of different single-crystal rotations combined with simulations of the EPR spectra and ligand field calculations give the specific orientation of the g tensor (Fig. 2C). gl) is just 5 ° off the long thioether S-Cu(II) bond and since this defines the electronic z axis, the dx2_# orbital is perpendicular to this and within 15 ° of the plane defined by the three strong ligands; the thiolate S and two imidazole Ns. The next feature of the blue copper ground state wavefunction on which we focus is the origin of the small parallel hyperfine splitting in the EPR spectrum in Fig. 1C. A small All value is observed in a number of distorted

C

x2_y 2

astceyanin Z2 xy

? . . . . 4p z z ,[,

--dx2 y2!~'

. . . . 4Px,y z xz+yz

dx2__y2

xz-yz

, yz)

"( decreases A II

rhombic

elongated

C3v

increases A H

Fig. 3. Originof smallAftvalues: (A) D2d-CuCI42-;4pz mixingwith the dx2_ f orbital in D2d symmeCylowers All (13)Energylevels for blue copper site (left) and its axial C3v limit (right) determined through low temperature magneticcircular dichroismspectroscopy (LT MCD) and ligand field theory (LVF). (C) C3v (Blue Copper); effect of 4Px,y mixing with the dF _ y2orbital for a C3v distortionincreases An. Arrows on stick figuresindicate the natureof D2dand C3v distortionsfromTd.

E.L Solomon et aL / lnorganica ChimicaActa 243 (1996) 67-78

70

tetrahedral copper complexes, in particular D2d-CuCI42-. The explanation that was developed to explain this reduction in A, is that the distorted D2d geometry allows Cu 4pz to mix into the dx2_y2 orbital [15]. As pictured in Fig. 3A, this introduces a spin dipolar interaction with the nuclear spin on the Cu which opposes that of an electron in the dx2_y2 orbital and would reduce the parallel hyperfine splitting. 12% 4pz mixing would lead to the A, value of D2d-CuCI42- and that of the blue copper site [16,17]. However, a combination of ligand field calculations [14] and low temperature MCD data [18] (see below) had provided the d orbital splitting pattern in Fig. 3B (left), which is rhombic but close to the axial limit in Fig. 3B (right). This d orbital splitting is associated with an elongated C3v distorted tetrahedral structure. As shown on the top of Fig. 3C, this involves an elongation of one ligandmetal bond with the metal dropped towards the opposing

A

A----~ %OMO:J'~"~Cu dx2_y2 I~ + {~Cu 4p

il-~

Copper 1s

B :

Cu

c,

L/L/

CI

CI

/ CI

D4h- CuCI4-2 2.5

"~

1.0

8979 eV ls ~ > 3d / /

O.5

o.o

,

91o

.

i 180

,

A 275

,

3SO

(degrees) 8960

t 6960

' 9000

90 '2 0

' 9040

9060

E n e r g y (eY) Fig. 4. Polarized single-crystal Cu K-edge XAS of D4h-CuCl42- (A) Energy level diagram depicting the Cu Is to HOMO transition. (B) Experimental frame (top) showing orientation of light with respect to the sample (~ m 0 ° along a Cu-CI bond) and XAS spectra for D4hCuCI42" for ~ = 0 ° and 45 ° (bottom). Insert in the bottom panel shows the intensity of the 8979 eV peak as a function of ~p, the angle of rotation about the z axis.

Table 1 Blue Copper covalency: quantitative analysis of g values

Plastocyanin experimental

gx gy gz

2.047 2.059 2.226

x2 _ y2 spin only

2.00 2.00 2.00

LFT d orbitals + 2 L'S

2.125 2.196 2.479

SCF-Xa-SW d levels, CT levels )'Cu L.S + 2 L L'S Norman

Adjusted

radii

radii

2.046 2.067 2.159

2.059 2.076 2.226

trigonal face. Referring to the blue copper site, the C 3 axis would correspond to the long thioether S--Cu bond. This geometry conflicts with the explanation for small A, above because in C3v symmetry the Cu 4px, y, rather than 4pz, mixes into the dx2_y2 orbital, and as is illustrated on the bottom of Fig. 3C, the spin dipolar interaction of an electron in a 4px, y orbital with the Cu nuclear spin would tend to increase instead of decrease the A, value. Thus, we wanted to experimentally determine whether Cu 4pz is mixed into the half occupied dx2_y2 HOMO level. This goal was accomplished by going up ~10 orders of magnitude in photon energy in performing X-ray absorption spectroscopy (XAS) at the copper K-edge [19]. As shown in Fig. 4A, the lowest energy pre-edge transition at the Cu K-edge is the Cu ls --->half-occupied HOMO which produces a weak feature at -8979 eV. Cu ls---> Cu 3dx2_y2 is electric dipole forbidden, therefore, any electric dipole intensity in this transition should reflect Cu 4p mixing due to the distorted geometric structure of the copper site. However, as shown in Fig. 4B square planar CuC142-, which has an inversion center and thus can have no 4p mixing into the dx2_y2 orbital, shows some intensity in the 8979 eV transition. In collaboration with Keith Hodgson, we first studied the origin of this intensity through polarized single-crystal XAS on square planar CuCI42- [20]. The configuration used in the experiment is shown at the top of Fig. 4B, where the E and propagating k vectors of light are oriented in the xy plane and the complex is rotated about the z axis. From the data shown at the bottom of Fig. 4B, the 8979 eV peak is clearly polarized and the intensity of this band is plotted as a function of the angle of rotation about the molecular z axis (Fig. 4B, insert). The intensity peaks every 90 ° which indicates that this is a quadrupole transition (a dipole transition would peak every 180°). This band has observable absorption intensity because at -9000 eV the wavelength of light is ~ 1.4 A and thus higher order terms in the multipole expansion of the interaction of radiation with matter can be significant. In Fig. 5A, we compare the orientation averaged pre-edge region of a powder of square planar CuCI42- to that of a frozen solution of plastocyanin [19]. The 8979 eV peak had increased in intensity by a factor of ~2 in plastocyanin indicating that

71

E.L Solomon et al. / lnorganica Chimica Acta 243 (1996) 67-78

B

A I

I

i

I

:y iI

8

D ..,

8 EIIz (1s -~ 4pz)

<.8

I

2

O~- CuCI4.

L........

J .......

E II (x,y) (ls --~ 4p)w) "'1"'"

I

8960 8970 8980 8990 9000

Energy (eV)

I

,I

I

8960 8970 8980 8990 9000

Energy (eV)

Fig. 5. X-ray absorption spectroscopy at tic Ca K-edge: (A) Orientation averaged XAS spectra of D4h-C'uCI42-and the blue copper site in poplar plastocyanin (inset, pre-edge region at higher sensitivity) and (B) polarized single-crystal XAS spectra for poplar plastocyanin (spectra taken from Ref. [21]).

there is now Cu 4t) mixing into the d~_~ due to the distotted tetrahedral geometry of the blue copper site. Polarized single-crystal XAS data were analyzed to determine the nature of this mixing (i.e. 4pz or 4px,y, see Fig. 3) [21]. Single-crystal EPR studies indicated that the z axis is collinear with the long thioether S-Cu bond (see above) [14]. The single-crystal XAS spectrum of plastocyanin taken with the E vector oriented along this z axis in Fig. 5B shows no 8979 eV intensity. This is the orientation for observing 4pz mixing into the d~_ ~ orbital and thus there is no measurable 4pz character. Alternatively, the polarized XAS spectrum taken with the E vector of light in the xy plane which would show the 4px,y mixing (plus the quadrupole component) shows all the 8979 eV peak intensity. This indicates that the Cu dx2_~ orbital only mixes with Cu 4px,y in the blue copper site and thus the origin of the small A, value is not Cu 4pz mixing. This

was an extremely important result because it allowed us to focus on an alternative mechanism for small hyperfine splitting which is a highly covalent site [18,22]. Covalency reduces A, by delocalizing the electron spin off the metal center and reducing its interaction with the nuclear spin on the copper. A

A--~%ouo d,a-~= T - (~ s(cys) 3p ~[--~

B

Copper 2p I

I

:'! D CUCI42 ''

¢-.

61%

% NHis i": plastocyanin Q

.N O

x -p

:

','

,16 '/o Scys p~

o Z

I ~ ,I 925 "'-" 4% NHis Fig. 6. Contour of plastocyanin ground state wavefunction (HOMO). Contour lines are at :e0.64, ~-0.32, ~--0.16, ~--0.08, ~-0.04, :e0.02, and ~0.01 (electrons/bohr3) !/2. The outermost contour encompasses 90% of the electron density.

930

935

.,I 940

Energy (eV) Fig. 7. Cu L-edge XAS as a probe of ligand-metal covalency: (A) Energy level diagram depicting the Cu 2p to HOMO transition. 03) XAS spectra for D4h-CuCI42- and plastocyanin (spectra taken from Ref. [29]). Values listed are the amount of Cud character in the HOMO.

72

E.L Solomon et al. / Inorganica Chimica Acta 243 (1996) 67-78

A • ... ^ = ~ C u |

- o~S(Cys) 3p

~/._.~

B

dx2 , 2

Sulfur ls

I

I plastocyanin

0o t,.t-"o N o~

/ 1o~2=38% tetb

\

,",

E

t-.

o

z

. . . . . . [" 2468 ~ 2469

I 2470

2471

Energy (eV) Fig. 8. S K-edge XAS as a probe 0fligand-metal covalency: (A) Energy level diagram depicting S I s to HOMO transition. (B) Orientation averaged XAS spectra for tet b and plastocyanin (spectra taken from Ref. [19]).

We probed the possibility o f a highly covalent ground state wavefunction for the' blue copper site through a quantitative consideration of its g values. The experimental g values obtained from Q-band EPR spectroscopy on plastocyanin [22] are given in the first column of Table 1. They are anisotropic and all are greater than 2.00. If the half-occupied HOMO were in fact a pure d~_y2 orbital there would be only spin angular momentum and all g values would be 2.00. In ligand field theory (LFT) [23], one allows for spin-orbit coupling of excited d levels into the ground state which mixes orbital angular momentum into this level and leads to the g values in the third column in Table 1. All are greater than 2.00 due to an orbital contribution to the g values, however, they now deviate from 2.00 to a greater extent than the experimental values indicating that there is too much angular momentum in the LFT ground state wavefunction. This deviation from experiment arises from the fact that the ligand field calculation given in Table 1 uses pure d orbitals, while covalency reduces the orbital angular momentum associated with a molecular orbital. In order to obtain a description of the ground state of the blue copper' site which includes covalency, we performed SCF-Xa-SW [24] calculations on the structure in Fig. 1A [18,22]. We then allowed for metal and ligand spin-orbit coupling over all the resulting wavefunctions and calculated the associated g values of the ground state. The values ~obtained are given in the fourth column of Table 1. They are indeed reduced from the ligand field values due to covalency, however, they

are closer to 2.00 than the experimental values in column 1 indicating that the calculation has produced too covalent a description of the ground state. There is one set of adjustable parameters in the SCF-Xa-SW calculations which is the sphere sizes used to define the scattered wave solutions. The standard sphere radii used in most SCF-Xa-SW calculations, Norman radii [25], were used in these calculations of the blue copper site. Our studies on the CuCI42- model complex [26-28] have indicated that the Norman radii give too covalent a description of the bonding relative to the spectral data. However, if the spheres are adjusted so that the calculations reproduce the ground state g values, very good agreement is obtained between the SCF-Xa-SW calculations and spectral data covering many orders of magnitude in photon energy [26-28]. Adjusting the sphere radii in the blue copper calculations to obtain reasonable agreement with the experimental g values gives the ground state wavefunction pictured in Fig. 6. Fig. 6 is a contour in the xy plane containing the coordinating cysteine S and 2 histidine N atoms. This ground state is extremely interesting in that it is highly covalent with only 42% Cu dx2_y2 character (accounting for the small hyperfine coupling) and strongly anisotropic, dominantly involving the p~ orbital of the thiolate sulfur. Due to the importance of this redox active orbital we have evaluated its key features experimentally using spectral methods covering many energy regions• In collaboration with Steve Cramer, we were able to experimentally evaluate the Cu dx2_~ character in the half occupied HOMO using Cu L-edge spectroscopy [29]. As shown in Fig. 7A, there is a transition at ~930 eV involving the Cu 2p---> half occupied HOMO excitation. The intensity of this transition reflects the 2p --->3d component which is electric dipole allowed and thus is a measure of the 3dx~_y2 character in the HOMO. As determined from the data in Fig. 7B, the blue copper site has 0.67 of the Cu L-edge intensity of square planar CuCI42-. Our studies on the latter complex have indicated that it has 61% Cu dx~_y= character in the ground state. Thus the intensity difference in Fig. 7 gives -41% Cu dx=_y2 character in plastocyanin, in good agreement with the SCF-Xa-SW calculated wavefunction shown in Fig. 6. The fact that this covalency dominantly involves the thiolate S-Cu(II) bond was demonstrated by sulfur Kedge spectral studies conducted with Keith Hodgson and Britt Hedman [19]. The S ls ---) half-occupied HOMO transition occurs at ~2470 eV (Fig. 8A). The electric dipole component is S ls --->S 3p and thus the intensity of this transition reflects S 3p character in the half-occupied HOMO. As can be seen from Fig. 8B, plastocyanin has 2.6 times the intensity of the pre-edge feature in tet b, a model complex prepared by Harvey Schugar et al. [30] which has a fairly normal thiolate S--Cu(II) bond with -15% covalency. Thus the intensity ratio from Fig. 8B indicates that the blue copper ground state wavefunction

73

E.L Solomon et al. / lnorganica Chimica Acta 243 (1996) 67-78

B

A 6000 A

I

5000

¥

4000

o "7,

3000

I

Normal Copper -~

I

orption_-

dx2-y2

s _L~. strongc

2000 1000

,o

Cys4=A LT AbsorptionCy~ t I " pseuao-o I; '~ d %., 6

E 4ooo "7, 3000 2000 ~o 1000

~let ~.

a

*

3/.'

I 101 ) A ~ace II a ~ / \

e'ID-

~5~

I

d

xy dz2 I

Single Crystal Polarized Abs.

Plastocyanin

"~

~Cu~

.e

<

dx2-y2

~:0S J'~~~-str°ng~: I ~ weak

_LT MCD

"7b~

O°

-1ot 1

-20

25000

3 4 ~ 41

5~ 7

8'

I 20000 15ooo loooo 5o00 Energy(cm"1)

0

'°

O

Fig. 9. Blue Copper excited-state spectra and orientation of plastocyanin ground state wavefuncfion. (A) Room temperature, low-temperature absorption, single crystal polarized absorption, and low-temperature MCD spectra of plastocyanin. Band 8 has been scaled by a factor of five in the MCD spec~m. The LT absorption spectrum has been Gaussian resolved into its component bands as in reference 18. (13) Cu--cysteine bonding interactions; top section illustrates a 'normal' bonding mod¢ with weak n and strong o charge transfer transitions while the bottom section depicts the strong ~r and weak o charger transfer spectra in plastocyanin (which results from a 45 ° rotation of the Cu dx2 _ y2 orbital due to strong ~t bonding). Note that the 6 here refers to the pseudo o valence orbital (see Fig. 11B).

has -38% thiolate S character. This result is also in good agreement with the SCF-Xa-SW calculations in Fig. 6. The final key feature of the ground state wavefunction pictured in Fig. 6, its orientation, was confirmed from the assignment of the unique absorption spectrum of the blue copper site. Fig. 9A shows the effects of going to low temperature on this spectrum. Bands are better resolved and by examining a combination of low temperature (LT) absorption, CD and MCD data (each method governed by a different selection rule) [18], a minimum of eight transitions are required to fit the absorption spectrum, labeled bands 1-8 in Fig. 9A. Band 4 is the intense 600 nm absorption feature characteristic of blue copper sites, and at lower energy band 6 also has moderate intensity. Polarized absorption data on single crystals of plastocyanin [14] show that both bands have the same polarization ratio which associates these with the cysteine-copper bond. As depicted in Fig. 9B, normal Cu(II) complexes exhibit a characteristic charge transfer pattern consisting of a low energy weak ~r and a high energy intense o

charge transfer transition. This derives from the fact that charge transfer intensity reflects the overlap of the orbitals involved in the transition. In normal Cu(II) complexes (Fig. 9B, top), a lobe of the d~_ y2 orbital is oriented along the ligand-metal bond and thus the ligand o orbital, which is at deep energy due to the o bonding, also has the most charge transfer intensity due to overlap. Thus the blue copper absorption spectrum has generally been assigned as hand 6 being the cysteine :r and band 4 the cysteine o---> Cu(II) dF_y~ charge transfer transition [2,11]. Significant insight into this possible assignment came from LT MCD data [18] shown at the bottom of Fig. 9A. Bands 5-8 which are all relatively weak in absorption are the most intense in the LT MCD spectrum. LT MCD involves a C-term mechanism which requires spin-orbit coupling hence d character. Thus bands 5-8 are d --~ d transitions with the specific assignments indicated at the top of the low temperature absorption spectrum. Band 4 which is intense in the absorption spectrum is relatively weak in the LT MCD spectrum and thus is as-

74

E.I. Solomon et aL / lnorganica Chiraica Acta 243 (1996) 67-78

v

T3 Cu His

T3 Cu T2 Cu

His

His(

cys

3" T3Cu {~

His

~

L Y

T2 Cu O _..-O T3Cu

Fig. 10. Proposed long-rangeelectron transferpathway in blue copper proteins [3,4,13]. The plastoeyanin wavefunctioncontours have been superimposed on the blue copper (type 1) site in ascorbateoxidase (structure taken from Ref. [31]). The contour shows the substantial electron delocalization onto the cysteine SPn orbital that activates electron transfer to the trinuclear copper cluster at -12.5 A from the blue copper site. This low-energy, intense Cys S~r -~ Cu charge transfer transition provides an effective hole superexchange mechanismfor rapid long-rangeelectron transfer between these sites. The space filling figure is drawn with the xy plane in the paper and the stick figure at the bottomhas been rotated 90°. signed as the lowest energy cysteine ~ - - > C u dx2_y2 charge transfer transition. A higher energy, weak band is associated with the cysteine a --->Cu d~_ f charge transfer transition. (Note that the Cys a here refers to the pseudo a valence orbital in Fig 11B (see below) and the moderate absorption intensity in band 6 derives from the fact that it involves the dxz+yz orbital which has some mixing with the cysteine ~ orbital). Thus, we do not observe the normal low energy weak/high energy intense charge transfer pattern of most cupric complexes (Fig. 9B, top). Since charge transfer intensity derives from overlap, the blue copper intensity pattern (~ intense/a weak) requires that the d~_y2 orbital be rotated by 45 ° such that its lobes are bisected by the Cys(S)--Cu(II) bond (Fig. 9B, bottom). This rotation of the dx2_p orbital derives from the strong ~ antibonding interaction with the thiolate sulfur at the unusually short 2.1/~ Cys(S)--Cu(II) bond length in the blue copper site. These studies have defined a novel electronic structure for the oxidized blue copper site which can make significant contributions to long range electron transfer. Fig. 10 gives part of the crystal structure of ascorbate oxidase determined by Messerschmidt et all. [31]. The cysteine ligand of the blue copper is flanked on either side in the sequence by histidines which are ligands at a trinuelear copper cluster (first defined by our MCD studies on laccase) [32]. Electrons enter at the blue copper site and are

transferred rapidly (k > 103 s -l) over -13 A to the trinuclear cluster where dioxygen is reduced to water [33]. We have superimposed the blue copper ground state wavefunction from Fig. 6 on the crystal structure of the ascorbate oxidase in Fig. 10 and find an important correlation to reactivity. The high anisotropic covalency of the Cys(S)-Cu(II) d:_y2 bond activates this ligand for directional electron transfer to the trinuclear copper cluster. Further, the low energy intense Cys ~r ~ Cu(II) d~_y2 charge transfer transition in Fig. 9A provides an efficient hole superexchange pathway for rapid electron transfer [3,4,13]. Thus while the unique spectral features of the blue copper center certainly reflect the unusual geometric structure and ligation of the site in Fig. 1A, they, in fact, derive from an electronic structure tuned for rapid electron transfer to a specific position in or on the protein.

3. Electronic structure of the reduced blue copper site: Hgand-metal bonding and its contribution to reduction potentials For the oxidized blue copper site the most challenging aspect for understanding the site was the presence of unique spectral features while for the reduced blue copper site the major obstacle was that it has a d l° configuration and is therefore inaccessible to the usual spectroscopic

75

E.L Solomon et al. / lnorganica Chimica Acta 243 (I 996) 67-78

A

. . . .

I

. . . .

I

. . . .

1

. . . .

I

rBC

. . . .

Cu 3d2.#~ -2

,

Cu 3d~y~

\

Cu3d '

\\\

"Cu 3 d , z + v z ~ ~ x ~ xz-yz

>

q) Cys ~ His ~1.-=

>,

~-4 (1) ¢UJ C--H psuedoo t ~ / ~

u

. . . .

25

I

. . . .

20

I

~Vexposed.clean (x2) . . . .

15

I

. . . .

10

Cys pseudo-o

I

5

0

Ionization Energy (eV)

"x \ \

Cu 3dx2 \~ . / C u 3dxr \~ //His ~1 \ k~ ~ / ~ , ~ ~ __k~.----------d//~u 3dz2

\

\

/

!I

Cu 3dxz.yz Cys n

\

\----

Met b 1

\

\

-6

Cys pseudo-o

His n 2 ~

03

. . . .

l

~

Met b I

e-)

o , _~J

•

oBC

"~" ~ H i s

"10 f-

o (.9 -8

Cys a Met a i

n2

\ ~

Met al

\ Cu

rc

Cu

pseudo-o

Hiso~

Cu

cr

Fig. 11. CH3S--Cu(I ) bonding. (A) Valence band PES of clean Cu20(lll), Cu20(l 11) exposed to methanethiol, and their difference spectrum with Gaussian/Lorentzian resolution of the low-energy region. (B) valence orbitals of methanethiolate.

methods in bioinorganic chemistry. We have thus been developing variable energy photoelectron spectroscopy (PES) using synchrotron radiation as a new method to define the bonding and its change with ionization (i.e. oxidation) in transition metal complexes [28,34,35]. Since we are interested in electrons emitted from valence orbitals involved in bonding, we study model complexes rather than the proteins, and since we want to understand normal ligand-metal bonding and its relation to the blue copper site, the models we have chosen to study are the blue copper relevant imidazole, dimethylsulfide and methylthiolate ligands bound to coordinatively unsaturated Cu(I) sites on oxide and chloride single crystal surfaces in ultra-high vacuum. As an example, the photoelectron spectral data [10] for methylthiolate bound to cuprous oxide is shown in Fig. 11A (where we have used the surface oxide to deprotonate methylthiol) [36]. The dashed spectrum gives the valance band region of clean Cu20, and the solid spectrum is that of methylthiolate bound to the surface. The difference spectrum at the bottom gives the valance orbitals of the thiolate-Cu(I) bond. Varying the photon energy changes the intensity of the peaks [37,38] and allows the specific assignments of the

~N

Met b2

~ ~

}

Cyso

~xMet b2 His o

-10

Fig. 12. Valence molecular orbital levels from PES calibrated SCF-XaSW calculations on reduced (rBC) and oxidized (oBC) plastocyanin, See Ref. [10l for details.

levels indicated in the difference spectrum. Three valence orbitals of thiolate are involved in bonding to the Cu(I): the at, pseudo o and o shown in Fig. 1 lB. The at and pseudo o, which are the sulfur p orbitals perpendicular to the S--C bond, dominate the bonding and split in energy as the angle ~ in Fig. l i B deviates from 180 °. Thus the energy splittings and intensity changes with photon energy of the peaks in the difference spectrum in Fig. 11A give the geometric and electronic structure of the uncon/R

Met S_ R

T

2.9A

N~Cu__ .R N/ 2.1"~"S / Cys Fig. 13. Axial ligand effects in plastocyanin. The long S(Met)-Cu(I) bond is imposed by the tertiary protein structure on the blue copper site and results in a shortening of the S(Cys)--Cu(1) bond.

76

E.I. Solomon et aL / Inorganica Chimica Acta 243 (1996) 67-78

A Qi < F e x l ( b ' V / / Y Q i ) o l F ~ ~

!~ol

[ (dil'-

< rexl(OW/~,)Jrex >

ki

distortioncoordinate,Qi

B

~His87 "0"19~-~+9-

~Met92 Idlh

O reduced

•

/

_

(~) oxidized

Fig. 14. Distortingforcesand geometricchanges in the blue coppersite on oxidation. (A) Configurationcoordinatediagram of the linear coupling term for the distorting force along the ith normal mode of vibration, Qi, on the oxidizedCu(II) site relativeto the reducedCu(I) ground state, dV is the energy change calculated for a distortion dQi of the molecule. AQ i is the expecteddistortionalong the normal mode for the calculated linear coupling term. (13) Crystallographieallydetermined structural changes in the bond lengths and angles in the bINSequatorial plane (left) and in the S(Met)CuL angles (fight) (taken from Ref.

[41,42]). strained thiolate--Cu(I) bond of the surface complex. Our strategy was to use variable energy PES to experimentally define the geometric and electronic structure of each normal ligand--Cu(/) bond for the surface complexes. We then use these data to evaluate and calibrate SCF-Xa-SW electronic structure calculations of the unconstrained surface complexes. These experimentally calibrated calculations were then used to generate an electronic structure description of the reduced blue copper site and to evaluate the change in electronic structure with oxidation. These studies which are described in detail in Ref. [10], demonstrate that the bonding in the reduced blue copper site is dominated by ligand p to Cu(I) 4p donor interactions. Thus, even though the bonding involves a reduced copper center which results in the d orbitals being higher in energy by 1-2 eV due to the lower effective nuclear charge on the copper relative to the oxidized site (see Fig. 12), there is no evidence for back-bonding for any of the blue copper ligands. We further find that the

long thioether S-Cu(I) bond length of 2.9/~ does not derive from the electronic structure of the reduced blue copper site, but is imposed on the site by the tertiary structure of the protein. As described in the next section, this appears to be the key feature of the blue copper site structure that can be ascribed to an entatic or rack state. This elongated thioether S-Cu(I) bond reduces the donor interaction of this ligand with the copper. This is compensated for by the thiolate, leading to the short strong S(Cys)Cu(I) bond (Fig. 13). The length of the thioether S--Cu bond (i.e. the fact that the axial ligand--Cu bond is particularly weak) destabilizes the oxidized site more than the reduced site and is found to be the dominant ligand contribution to the high reduction potential of the blue copper center. This is referenced to the potential of a tetraimidazole complex, bis[2,2'-bis(2-imidazolyl)biphenyl] copper [39] where the imidazoles are constrained by bridging groups to undergo only a limited geometric change on oxidation as is observed for the blue copper site (see below). This geometric constraint also contributes to the high reduction potential of the copper site. 4. C h a n g e in electronic structure on oxidation: c o n t r i b u t i o n s to g e o m e t r y

Comparison of the electronic structure descriptions of the oxidized and reduced blue copper sites indicates that the ligand donor interactions with the unoccupied copper 4p (and 4s) levels change little upon oxidation and that the dominant change in bonding involves the hole produced in the dx2_y2 derived HOMO in Fig. 6. Thus on oxidation the blue copper site loses a strong antibonding interaction with the thiolate sulfur and weaker antibonding interactions with the imidazole nitrogens. Having obtained an experimentally calibrated electronic structure description of the reduced site, we could evaluate the changes in geometric structure which would occur for a blue copper site unconstrained by the protein. This involves evaluation of the electron-nuclear linear coupling terms of the oxidized site in the reduced geometry along all the normal modes, Qi, of the blue copper site. As illustrated in Fig. 14A, a non-zero slope of the linear

Td --~ ~

--I-- 2T2

C3v

Cs

"f X~ -2 2,~' -7-Y ~>10000crn-1 , . ~ "-1--ZE ~ .~--xy [ (from MCD)

_.-~.__.~_

Fig. 15. Possible Jahn-Teller distorting forces in blue copper sites. Energy level diagrams are shown for an idealized Td, C3v (inclusionof long thioether bond), and Cs (further effect of short thiolate bond) geometries. The large splitting of the dF_ y2, dxy levels by the site structure eliminates the Jahn-Teller effect that would normally be present in the oxidizedblue copper site.

E.L Solomon et al. / Inorganica Chimica Acta 243 (1996) 67-78

coupling term of the oxidized site in the reduced geometry, Qi(Fexl(6V/6Qi)olFex), corresponds to a distorting force along normal mode Qi. [38,40]. The significant distorting forces we find to be present are consistent with the change in electronic structure on oxidation described above. There is a large distorting force toward contraction of the thiolate S-Cu bond. This, however, is opposed by a large force constant, ki, associated with the short, strong S(Cys)--Cu bond leading to only limited contraction (AQ~=(Fexl(6V/6Qi)olFex)lki). There are also distorting forces along the N(His)--Cu bonds. Fig. 14B depicts the geometric structural changes on oxidation of a blue copper site determined through crystallography [41,42]. Consistent with the linear coupling terms, the thiolate SCu bond contracts by 0.04/~ on oxidation. The N(His)Cu bonds are found from crystallography to contract by 0.2/~ [41,42] while from extended X-ray absorption fine structure (EXAFS) this contraction is more limited at 0.07 A, [21,43]. Importantly, no significant distorting force is present in any of the bending modes and no significant angle change is observed in the crystal structures with oxidation of the blue copper site in plastocyanin. An entatic or rack state where the Cu(I) geometry is imposed on the Cu(II) site by the protein would correspond to a large Jahn-Teller distorting force along a bending mode. The lack of a Jahn-Teller distorting force in the oxidized blue copper site can be understood from the scheme in Fig. 15. In an idealized tetrahedral structure there would be a 2T2 ground state that is orbitally degenerate which leads to the Jahn-Teller distortion. In the reduced blue copper site there is an elongated thioether SCu bond which produces C3v effective symmetry at the copper leading to a 2E ground state in the oxidized site which corresponds to the unpaired e- being in the degenerate (dx2_y2, dxr) set of orbitals. The long thioether S-Cu bond further results in contraction of the thiolate S--Cu bond (see above) which splits the orbital degeneracy of the 2E. In particular, from the MCD assignment in Fig. 9A the dxy level is > 10 000 cm -1 above the dx2_y2ground state; it is a linear coupling term between these levels that would lead to a Jahn-Teller distorting force. Thus there is little geometric change on oxidation and a low FranckCondon barrier to electron transfer [42,44].

5. Concluding comments The unique spectral features of the oxidized blue copper site derive from the high anisotropic covalency of the ground state wavefunction involving the thiolate, which activates this ligand as an efficient superexchange pathway for long range electron transfer. Variable energy photoelectron spectroscopy combined with SCF-Xa-SW calculations have been used to determine the electronic structure of the reduced blue copper site for the first time and to evaluate the change in electronic structure which occurs upon oxidation. This change in electronic structure

77

is consistent with the limited geometric structural change observed leading to a low Franck-Condon barrier to electron transfer. Thus the reduced geometry is not imposed on the oxidized site by the protein. The entatic or rack state in blue copper proteins would appear to involve the protein tertiary structure imposing the long methionine S--Cu (i.e. weak axial) bond on the reduced site. This leads to the high reduction potential, the short thiolate S--Cu bond and therefore the efficient superexchange pathway and the lack of a Jahn-Teller distorting force in the oxidized site which results in little geometric change and rapid electron transfer.

Acknowledgements Ed Solomon would like to thank Harry Gray for introducing him to the field of Bioinorganic Chemistry and to the unique spectral features of the Blue Copper site. Our best regards on your birthday! This work has been supported by the NSF (CHE-9217628).

References [1] (a) E.T. Adman, in P.M. Harrison (ed.) Topics in Molecular and Structural Biology: MetaUoproteins, Vol. 6, Part 1, Verlag Chemic, Weinheim, 1985. (b) E.T. Adman, Adv. Prot. Chem., 42 (1991) 145. [2] E.I. Solomon, K.W. Penfield and D.E. Wilcox, Struct. Bonding, 53 (1983) 1. [3] E.I. Solomon, M.J. Baldwin and M.D. Lowery, Chem. Rev., 92 (1992) 521. [4] E.I. Solomon and M.D. Lowery, Science, 259 (1993) 1575. [5] R. Lumry and H. Eyring, J. Phys. Chem., 58 (1954) 110. [6] R.J.P. Williams, J. Mol. Catal., 30 (1985) 1. [7] B.G. Malmstr6m, Eur. J. Biochem, 223 (1994) 711. [8] H.B. Gray and B.G. Malmstr6m, Comments Inorg. Chem., 11 (1983) 203. [9] E.I. Solomon and M.D. Lowery, in A.J. Welch and S.K. Chapman (eds.) The Chemistry of Copper and Zinc Triads, The Royal Society of Cbemistry, Cambridge, UK, 1993, p. 12. [10] J.A. Guckert, M.D. Lowery and E.I. Solomon, J. Am. Chem. Soc., 117 (1995) 2817. [11] E.I. Solomon, J.W. Hare, and H.B. Gray, Proc. Natl. Acad. Sci. USA, 73 (1976) 1389. [12] J.M. Guss and H.C. Freeman, J. Mol. Biol., 169 (1983) 521. [13] M.D. Lowery, J.A. Guckert, M.S. Gebhard and E.I. Solomon, J. Am. Chem. Soc., 115 (1993) 3012. [14] K.W. Penfield, R.R. Gay, R.S. Himmelwright, N.C. Eickman, V.A. Norris, H.C. Freeman and E.I. Solomon, J. Am. Chem. Soc., 103 (1981) 4382. [15] C.A. Bates, W.S. Moore, K.J. Standley and K.W.H. Stevens, Proc. Phys. Soc, 79 (1962) 73. [16] M. Sharnoff, J. Chem. Phys., 42 (1965) 3383. [17] J.E. Roberts, T.G. Brown, B.M. Hoffman and J. Peisach, J. Am. Chem. Soc., 102 (1980) 825. [18] A.A. Gewirth and E.I. Solomon, J. Am. Chem. Soc., 110 (1988) 3811. [19] S.E. Shadle, J.E. Penner-Hahn, H.J. Schugar, B. Hedman, K.O. Hodgson and E.I. Solomon, J. Ant Chem. Soc., 115 (1993) 767. [20] J.E. Hahn, R.A. ScoU, K.O. Hodgson, S. Doniach, S.R. Dejardins and E.1. Solomon, Chem. Phys. Len., 88 (1982) 595. [21] R.A. Scott, J.E. Hahn, S. Doniach, H.C. Freeman and K.O. Hodgson, J. Am. Chem. Soc., 104 (1982) 5364.

78

E.L Solomon et al. / lnorganica Chimica Acta 243 (1996) 67-78

[22] K.W. Penfield, A.A. Gewirth and E.I. Solomon, J. Am. Chem. Soc., 107 (1985) 4519. [23] For general references on ligand field theory see: (a) C.J. Balihansen, Introduction to Ligand Field Theory, McGraw-Hill, New York, 1962. (b) D.S. McClure, Electronic Spectra of Molecules and Ions in Crystals, Academic Press, New York, 1959. (c) J.S. Griffith, The Theory of Transition Metal Ions, Cambridge University Press, London, 1964. (d) S. Sugano, Y. Tanabe and H. Kamimura, Multiplets of Transition Metal Ions in Crystals, Academic Press, New York, 1970. (e) B.N. Figgis, Introduction to Ligand Fields, Interscience, New York, 1967. [24] For general references on SCF-Xa-SW theory see: (a) K.H. Johnson, Adv. Quant. Chem., 7 (1973) 143. (b) K.H. Johnson, J.G. Norman, Jr. and J.W.D. Connolly, in F. Herman, A.D. McLean and R.K. Nesbet (eds.) Computational Methods for Large Molecules and Localized States in Solids, Plenum, New York, 1973, p. 161. (c) J.W.D. Connolly, in G.A. Segal (ed.) Semiempirical Methods of Electronic Structure Calculations, Part A: Techniques, Plenum, New York, 1977. (d) N. Rosch, in P. Phariseu and L. Scheire (eds.) Electrons in Finite and Infinite Structures, Wiley, New York, 1977. (e) J.C. Slater, The Calculation of Molecular Orbitals, Wiley, New York, 1979, p. 104. [25] J.G.J. Norman, Mol. Phys., 31 (1976) 1191. [26] A.A. Gewirth, S.L. Cohen, H.J. Schugar and E.I. Solomon, lnorg. Chem., 26 (1987) 1133. [27] E.L Solomon, A.A. Gewirth and S.L. Cohen, in J. Avery, J.P. Dahl and A.E. Hansen (eds.) Understanding Molecular Properties, D. Reidel, Dordrecht, 1987, p. 27. [28] S.V. Didziulis, S.L. Cohen, A.A. Gewirth and E.I. Solomon, J. Am. Chem. Soc., l IO (1988) 250. [29] S.J. George, M.D. Lowery, E.I. Solomon and S.P. Cramer, J. Am.

Chem. Soc., 115 (1993) 2968. [30] J.L. Hughey, IV, T.G. Fawcett, S.M. Rudich, R.A. Lalancette, J.A. Potenza and H.J. Schugar, J. Am. Chem. Soc., 101 (1979) 2617. [31] A. Messerschmidt, R. Ladenstein, R. Huber, M. Bolognesi, L. Avigliano, R. Petruzzelli, A. Rossi and A. Finazzi-Agro, J. Mol. Biol., 224 (1992) 179. [32] M.D. Allendorf, D.J. Spira and E.I. Solomon, Proc. Natl. Acad. Sci. USA, 82 (1985) 3063. [33] J.L. Cole, D.P. Baliou and E.I. Solomon, J. Am. Chem. Soc., 113 (1991) 8544. [34] J. Lin, P.M. Jones, J.A. Guckert and E.I. Solomon, J. Am. Chem. Soc., 113 (1991) 8312. [35] J. Lin, P.M. Jones, M.D. Lowery, R.R. Gay, S.L. Cohen and E.I. Solomon, lnorg. Chem., 31 (1992) 686. [36] J. Lin, J.A. May, S. Didziulis and E.I. Solomon, J. Am. Chem. Soc., 114 (1992) 4718. [37] J.J. Yeh and I. Lindau, At. Data Nucl. Data Tables, 32 (1985) 1. [38] E.I. Solomon, Comments lnorg. Chem., 3, (1984) 225. [39] S. Knapp, T.P. Keenan, Z. Xiaohua, R. Fikar, J.A. Potenza, and H.J. Schugar, J. Ar~ Chem. Soc., 112 (1990) 3452. [40] R.B. Wilson and E.I. Solomon, J. Am. Chem. Soc., 102 (1980) 4085. [41] J.M. Guss, P.R. Harrowell, M. Murata, V.A. Norris and H.C. Freeman, J. Mol. Biol., 192 (1986) 361. [42] J.M. Guss, H.D. Bartunk and H.C. Freeman, Acta. Crystallogr., B48 (1992) 790. [43] L.M. Murphy, S.S. Hasnain, R.W. Strange, I. Harvey and W.J. lngledew, in S.S. Hasanin (ed.) X-ray Absorption Fine Structure, Ellis Horwood, Chichester, 1990, p. 152. [44] A.G. Sykes, Adv. lnorg. Chem., 36 (1991)377.