ARCHIVES

OF BIOCHEMISTRY

AND BIOPHYSICS

Vol. 296, No. 1, July, pp. 247-255, 1992

Theoretical Study of Oxyhemocyanin Active Site: A Possible Insight on the First Step of Phenol Oxidation by Tyrosinase 0. Eisenstein,*

C. Giessner-Prettre,P,l

J. Maddaluno,$

D. Stussi,§

and J. Weber$

*Laboratoire de Chimie Theorique, Universite’ de Paris Sud, URA 506 CNRS, B&iment 490, 91405 Orsay Cedex, France; tLaboratoire de Chimie Organique Theorique, Universite’ P. et M. Curie, URA 506 CNRS, Batiment F, 4, place Jussieu, 75252 Paris Cedex 05, France; SLaboratoire de Chimie Organique, Universite’ de Rouen, URA 464 CNRS, B.P. 118, 76134 Mont St. Aignan Ceder, France; and SDepartement de Chimie Physique, Universite’ de Geneve, 30, quai Ernest Ansermet, CH-1211 Geneva 4, Switzerland

Received December 16, 1991, and in revised form March 2, 1992

Extended Huckel theory calculations have been carried out on a model of the oxyhemocyanin active site that includes six imidazoles, the two copper cations, and a dioxygen molecule. The results obtained for the very likely ~-$:$ arrangement of the dioxygen molecule show that the most favorable orientation of Ox is such that the two long Cu - N coordination bonds are perpendicular to the plane formed by the two metal atoms and Oz. This arrangement leads to pentacoordinated coppers with a distorted square pyramidal geometry. The molecular electrostatic potential maps of the complexes exhibit a potential well located close to the peroxo anion midbond. The dependence of the energy and of the molecular electrostatic maps on the precise orientation and location of the imidazole rings has been investigated. These results, which show the important role played by the third remote imidazole ligand, are discussed in relation with the first step of tyrosinase-mediated phenol oxidation. 0 1992 Academic Press, Inc.

Hemocyanin and tyrosinase are extensively studied enzymes having a dinuclear copper active site with strikingly similar physicochemical features (1). However, while the former is a well known oxygen carrier protein found in molluscs and arthropods (2), the latter of these enzymes is widely spread among the living reign where its monooxygenase activity (3) contributes to phenols oxidation and biopolymerization into pigments (4). A large set of experimental data, concerning the enzymes themselves (5) or several model compounds (6), has given access to ’ To whom correspondence

should be addressed.

0003-9861/92 $5.00 Copyright 0 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.

important information about the arrangement of the histidine ligands and dioxygen binding mode on this type of bimetallic system. In addition, recent theoretical approaches on models (7) of these enzymes have indicated that dioxygen binds in a p-v2:v2 mode rather than a pdioxo one (Scheme l), in agreement with several experimental results. Despite the available information, none of the experimental work has led to the description of the relative orientation of dioxygen with respect to the various histidines because of the absence of any crystallographic data about the enzyme oxy forms. In this context, no conclusion about dioxygen activation mode or precise mechanism for phenol selective o&o-hydroxylation has been established to date (8), despite recent progress in the determination of structural requirements for the monooxygenase-like activity (6c, Bb). For these reasons, Extended Huckel theory (EHT)2 calculations have been performed on a realistic model of the complexes containing two copper cations, the dioxygen molecule, and six imidazoles (Im) to model the six histidines. Our aim being a better understanding of the role of the histidine ligands in the dioxygen coordination mode, we will study the influence of the orientation of O2 with respect to the imidazoles, on one hand, and of the arrangement of the six ligands, on the other, on the energy and electronic structure of the complex. Because of the semiempirical character of the computational method retained, the results can only be used for comparison between closely related situations. In addition, molecular electrostatic potential maps have been calculated (9) to gain insight into the proton transfer step from the phenol OH to the peroxo anion. ’ Abbreviations

used: EHT, extended Huckel theory; Im, imidizole. 241

248

EISENSTEIN

o-o C” _------___-_----------.C”

cis-p-dioxo

p-TJ2:lj2 SCHEME

COMPUTATIONAL

1

PROCEDURES AND INPUTS

The EHT calculations (10) reported hereafter use the weighted Hi; formula (11) with parameters for Cu (12a) and (C, N, 0, H) (12b) taken from the literature. Since EHT is known to give unrealistic bond lengths, distances between bonded atoms are kept constant in all computations. The molecular electrostatic potential is calculated from these EHT wavefunctions, the three center integrals being evaluated following Mulliken’s approximation (13). Thus for each point P on a map we have:

where pIiy and S,, are the @Yelement of the EHT density and overlap matrices, respectively (lob). The first summation runs over the atomic orbitals and the second one over the nuclei of the system. Such a computational procedure has been shown to give in the case of organometallic compounds molecular electrostatic potential maps very similar to those obtained from nonempirical approaches (13, 14). It is therefore expected to give meaningful results for the kind of complexes studied here. The lack of complete structural information led us to make several assumptions and to examine two models. First, it should be mentioned that the exact number (2 or 3) of histidine ligands coordinated to each Cu nucleus is still a matter of controversy. For reasons discussed below, our study has been undertaken on two (Im)GCuz(02)2+ complexes. The first model (Hc) makes use of the geometry obtained from the latest refinements of the crystal structure of Panulirus interruptus hemocyanin (15a) (Scheme 2). The Cu- Cu distance is equal to 3.42 A. Around each copper center there are two short and one long Cu - N bond. The four short ones are coplanar while the two long ones (CU-N,,,) are perpendicular to the previous plane (Scheme 2). The dioxygen bond length has been taken equal to 1.48 A according to our recent theoretical study (7~). This appears reasonable since it has been experimentally shown on model complexes that the 0- 0 bond length is not very sensitive to the Cu - Cu distance nor to the number and nature of the nitrogen ligands (6b,d). In a second model, we have considered that the imidazole ligands are displayed according to a “standard av-

ET AL.

erage” (Sa) arrangement around the copper cations. The three imidazole rings are the apices of a tetrahedron centered at the copper ion, the second metal being the fourth apex (Fig. 2). The Cu- Cu-N angles are all equal to 110”. One of the three Cu-N distances has been taken equal to 2.4 A (remote imidazole) while the two others have a shorter value (1.9 A) in agreement with X-ray (15a, 16) and EXAFS (5~) data. The Cu-Cu distance, now equal to 3.7 8, as in earlier experimental (15b) and theoretical (7~) work, is in agreement with the absorption spectrum (7~). It is worth emphasizing that while the Sa geometry described above could be considered an arbitrary choice, it remains fully consistent with all available experimental data on the tyrosinase active site. The very different chemical behaviors of hemocyanin and tyrosinase are likely to be linked to some differences between their active site geometries. We have studied these models to investigate to what extend the influence of subtle changes in the arrangement of the histidine ligands around the metal centers can influence the potential reactivity of the active site. The comparison between results obtained from these two models may then provide an evaluation of the geometric parameter-activity relationship in these enzymes. It has been recently shown that dioxygen binds in a ~-~2:~2 mode (Scheme 1) for synthetic model complexes (6b, d, 7) of tyrosinase or hemocyanin which have been thoroughly characterized. We have thus assumed that dioxygen is perpendicular to the Cu-Cu direction. Furthermore, the Cu centers have been kept in the planes of the imidazole rings as indicated by both ab initio quantum mechanical studies on complexes between metal ions and imidazole (17) and experimental data on different metalloenzymes (18) since deviation from planarity is always less than 20”. Another point of interest concerns the orientation of the planes of the histidines with respect to the Cu - Cu direction (for which no experimental data are available). that the In a first step, we have considered Cu - Cu - N, - C, torsional angle is equal to zero; as a result, the imidazole N-H bond points toward the internuclear space (Fig. 1). The Hc and Sa model structures, which both exhibit an inversion center, are summarized

NN Sa type SCHEME

2

STUDY

OF OXYHEMOCYANIN

in Scheme 3. To simplify the presentation of results we have introduced the following notations: (i) the plane containing the two copper atoms and the coordinatingnitrogens of both remote imidazoles (N, and N,r) is labeled the CuN plane; (ii) the plane containing the two copper cations and the two oxygen atoms is named the CuO plane (Scheme 3). The results of the various studies that we present in this paper concern three points, viz. (i) variation of the dihedral angle (Y (Scheme 3) between the CuN and CuO planes from 0 to 90’; (ii) variation of cp(Cu - N) according to three different modes (i.e., relative orientation of the imidazole rings); (iii) variation of the two long Cu - N distances labeled d(Cu - N,) (i.e., position of remote imidazoles). It should be mentioned that both models presented above are less restricted and more realistic than those recently studied using SCF-Xa-SW methods (7a, b, d) since: (i) they take into consideration different dioxygen orientations with respect to the CuN plane; (ii) they are more consistent with X-ray data. RESULTS

AND

DISCUSSION

Variation

of the Dihedral

Angle cu(N,CuCuO)

Given the previously expressed geometrical parameters, the EHT optimization of the orientation of the dioxygen molecule with respect to the Cu(Im)B moieties leads for both Hc and Sa geometries to a configuration (displayed in Fig. 1) in which CuN and CuO planes are perpendicular to each other (a = 90”). The conformations for which a! = 0” lie higher in energy by 32 (Hc) and 8 (Sa) kcal/mol. In addition, among the different Sa conformations, the

ACTIVE

249

SITE

CuN plane

SCHEME

3

one with O2 perpendicular to a pair of short Cu - N bonds ((Y = 120’) is less stable by 2.3 kcal/mol than the (Y = 90” situation (long Cu-NN, bond perpendicular to the CuO plane). There is then a preference for a structure that can be viewed as a planar square-based pyramid in the case of Hc or a nonplanar one in the case of Sa. Although there is a significant difference between their rotation barriers, the reasons for the stability of the (Y = 90” conformation are identical. This can be understood by looking at the interaction of an O2 ligand with 2 Cu(Im)3f moieties following the type of analysis developed by Hoffmann to explain structure and reactivity of organic, organometallit, and inorganic systems (10). The orbitals of the Cu(Im)B fragment able to create bonds with an additional ligand are the high-lying orbitals shown in 1 and 2 (Scheme 4) in the case of the Hc geometry. Orbital 1, which is antisymmetrical with respect to the CuN plane, is mostly

A

FIG. 1. Perspective (Hc) X-ray geometry; Cu - Cu directions.

view, along the 0- 0 bond, of the optimized conformation of the (Cut-Im&-O2 (B) “standard average” (Sa) geometry. The two structures exhibit an approximate

complexes with: (A) “hemocyanin” inversion center at the middle of the

250

EISENSTEIN

ET AL.

3 SCHEME

in the plane including the short Cu- N2 and Cu-NN3 bonds and is extended away from these ligands. Orbital 2, which is symmetrical with respect to the CuN plane, is mostly extended opposite to the long Cu-NN1 bond. These two orbitals would have been degenerate if the three Cu - N bond lengths were equal. Orbitals 1 and 2 combine independently to give an inphase (l+, 2+) and out of phase combination (l-, 2-). For (Y = 90”, the aa0 orbital lying in the CuO plane combines with l- while the aao empty orbital combines with l+ as shown in 3 and 4 (Scheme 4). In the other orientation considered (LY= O”), the same orbitals of O2 combine respectively with 2- and 2+. The calculations show that the overlap between interacting orbitals is larger for a! = 90” than (Y = 0”. This is due to the fact that orbitals of type 1 are entirely contained in the plane of the short Cu - N bonds while type 2 orbitals are distributed along the three Cu-N bonds. This leads to electron transfer from the formally doubly occupied orbital ~6~ (the O2 ligand being in the peroxo form) and from the Cu centers into the high lying ~6~. The accepting role of uao has recently been pointed out by Ross and Solomon (7a). From our calculations, there is thus more bonding between the O2 ligand and the copper centers in the structure where a = 90”. Moreover, in the Sa case, orbitals 1 and 2 play a more equivalent role because of the fact that the three Im ligands are arranged in a more symmetrical manner. This results in more equivalent overlaps with orbitals of dioxygen as the (Y angle is varied, leading to a lower rotational barrier than in the Hc case. Examination of the atomic electronic populations shows a striking difference between the copper charges

4

obtained for the two modes of oxygen binding. The values reported in Table 1 indicate that in the (Y = 0” situation the positive charge of the copper ions is considerably smaller (and can even become negative) than for the (Y = 90” orientation. This result suggests that the (Y = 0“ geometry is not only energetically disfavored but also renders the metal ions much less attractive toward a nucleophilic sixth ligand such as the phenol oxygen atom of oxytyrosinase substrates (8). Possibilities for an artifact due to the simplified computational method used throughout this study prompted us to compare the copper atomic net charges obtained from, respectively, EHT and SCF ab initio calculations for (CU+)~--O~ (7~) and (Cu’ - Im), (19) complexes, using of course identical geometrical parameters. We obtain, respectively, 1.273e versus 1.325e for the former complex and 0.408e versus 0.832e for the latter one. These values show that, although EHT overestimates the ligand + cation charge transfer, it reproduces nicely the relative trends of the copper charges in the different complexes studied by nonempirical methods. From a topological point of view, it is easy to see from the two structures of Fig. 1 that, upon Oz binding, both copper ions become pentacoordinate and sit in a distorted square pyramidal geometry. The four basal ligands consist of the two oxygen centers plus two imidazole nitrogens, while the apical ligand is the more distant imidazole. This result is in complete agreement with Kitajima’s experimental data (6d) on a model complex with a Cu- Cu distance value of 3.56 A, which is close to that used in this work. At this stage, the major difference appearing between the Hc and the Sa situations is that in the first

STUDY

OF OXYHEMOCYANIN

ACTIVE

251

SITE

A

rcal/mol

Iccal/mol

rlIN=-88.49 -

xl.89

--

57.59

- 75.44 - - 52.55 - - 29.32 --- 5.755 .--- -17.51 -41.37 -54.83

FIG. 2. Molecular electrostatic potential maps (values in kcal/mol) in the CuN plane of the optimized conformation of the (Cu+ - Im& - 0, complexes with: (A) “hemocyanin” (Hc) X-ray geometry; (B) “standard average” (Sa) geometry. A phenol as a possible incoming sixth ligand on one of the coppers is also shown (see text).

case the copper cation is in the plane that includes the basal ligands while in the latter it is somewhat displaced out of this plane, toward the apex (Fig. 1). Perspective views of Fig. 1 show that, despite the important structural differences that are clearly seen from Scheme 2, the Sa and Hc forms of the complex have a common feature which is of primary importance for the substrate binding of oxytyrosinase. Indeed, for both geometries, the Cu cations are fully accessible for the proposed phenol oxygen coordination (8), provided that the incoming molecule is in the CuN plane and anti to the N1 or N’, remote imidazole, as shown in Fig. 2. As recently proposed from experimental results based on the crystal structure of a model complex (20), the substrate could be docked on the active site with a phenolic rather than a phenolate form. This arrangement would thus locate the OH proton in close proximity to both negatively charged oxygen atoms of the peroxo anion. This remark is to be considered in relation with Tyeklar and Karlin’s proposal of the occurrence, in tyrosinase, of a “proton transfer (from a phenol substrate?)” (6e) based on experimental results obtained with an hydroperoxo copper(I1) complex (21). Since the value of the molecular electrostatic potential provides information about the most probable proton binding site (17), mapping of this quantity for the various complexes studied here may provide helpful elements for

discussing the mechanism postulated above. The plots of the corresponding results are reported in Fig. 2 for the GUN plane perpendicular to the OX bond. We see that in the cases of both Hc and Sa there is a deep potential well located close by the peroxo anion midbond in the average plane of the phenol ring and remote imidazole. This pit’s depth is two orders of magnitude larger than in free dioxygen (Fig. 3) and will undoubtedly constitute a driving force for the proton transfer from phenol to the favorably

~ ” :

kcal/mol

,,.,

”. .y:._ .-.‘I.:. ‘.. .l:. ,:,.. ,:’ ‘“--. ,,:,

: .’

MIN=-1.451

-

100.0

--

75.00

- - 50.00

‘.

“.

.._..-’ ,:’

:’

---

25.00

-.--.

10.00 0 -1.ooo

3. Molecular electrostatic potential map of an isolated dioxygen molecule in the plane perpendicular to the 0 - 0 bond. FIG.

252

EISENSTEIN

located peroxo moiety. Such a finding may appear somewhat surprising if we consider the overall +2 charge carried by the complex. However, the analysis of the atomic electronic population (Table I) shows that the presence of the imidazole ligands reduces considerably the total formal charge carried by the (CU+)~-O~ substructure, decreasing it from +2.00e to +O.O8e and -0.44e, respectively, for the Hc and Sa geometries of the complete (Cu’ - Im& - O2 complex. In addition, the nitrogens coordinated to the metal atoms carry a negative charge in all cases larger than -0.40e. The reactivity of the complex toward phenol oxygen derives from a LUMO concentrated on copper (orbital 2-). The comparison of the molecular electrostatic potential maps obtained for the two geometries shows (Fig. 4) that, despite the important differences between the imidazole’s position in the two geometries, the isocontours are remarkably similar, the two maps differing only by the absolute value of the pit depth near the O2 ligand. This similarity suggests that the location and the shape of the potential well are primarily determined by the electronic populations on the copper cations, the coordinated nitrogens, and OZ. Relative Orientation

of the Imidazole

Rings: Variation

of dcu--N) Since the X-ray data have shown that the six N&H protons are engaged in hydrogen bonds with the carbonyl oxygen of different residues or with water molecules (5), the energy optimization of the complex with respect to the Cu - Cu - N, - C, torsion angles could appear of little interest. We could indeed be dealing with energy variations of the same order of magnitude (20-30 kcal/ mol) as those of the six intermolecular bonds neglected in the absence of the corresponding measured geometrical data (5). But again we thought that optimizing imidazole ring torsion could give us information about the sensitivity of our model to this aspect of their structure and provide a numerical estimation of the rotational barrier values for comparison with the possibly antagonistic hydrogen bond energies. It would also check if the overall potential map shape and pit depth obtained above could be signif-

TABLE

I

Copper and Oxygen Atomic Net Charges in the (Cu’- Im& - O2 Complexes for the Two Oxygen Binding Modes’ HCb

Sa

cu 0

a = o”

a = 90”

a = 0”

a = 90”

-0.032 -0.831

+0.414 -0.634

-0.008 -0.758

f0.538 -0.577

a See Scheme 3 for 01values. b In the case of the Hc geometry (approximate C,,, symmetry) we report average values for the two Cu+ and 0 centers.

ET AL. so iv 60_ 6040-

I I I I I 1 \ \ \

,

I

,

I

,

/

'

,

I

,

I

\

\

~---------------------~~ A

_______ - _________ -_-__---

IS

’ 2

I

\

FE 30: 8 620 w I

-30

,

’ 3

1

’ 4

’

’

I

C ______ - ______ -_------

’

5

6

d(Cu-N1)

(A)

I

’ 7

0

_____ -

’ 6

0 9

FIG. 4. Relative energy (in kcal/mol, arbitrary origin) of the (Cu’Im& - O2 complexes vs remote imidazole distance. A, Hc type, (Y = 0”. B, Hc type, 01 = 90”. C, Sa type, 01= 0”. D, Sa type, (Y = 90’.

icantly altered by modifying the relative arrangement of the ligands. Thus, three different types of rotation have been imposed on the imidazole ring groups. Starting from the previous cp(Cu-N) = 0 situation, they are defined as follows. -Type A rotation: synchronous and equal clockwise rotation of the three imidazoles belonging to the same copper coordination set around each Cu - N direction. This motion is applied simultaneously and counterclockwise to the opposite group to preserve the overall symmetry of the system. Rotation angle: (PA(Scheme 5A). -Type B rotation: equal and opposite rotation of the two remote imidazole rings only. Rotation angle: (Pa (Scheme 5s). -Type C rotation: opposite rotations of the two “close” imidazoles belonging to the same coordination set, the remote one remaining unaltered. An opposite and synchronous motion is applied to both ligand groups. Rotation angle: cpc (Scheme 5,9. The main features (22) of this study are reported in Table II. The B type rotation results are not presented since this motion did not affect significantly the stability of the complex for either Hc or Sa situations. On the other hand the A type rotation goes through two energy minima for c~A(Hc) = 30” and 150’ and cp*(Sa) = 60” and 120”. These are sterically relaxed situations where the interaction among the three heterocycles is minimized. The corresponding gains remain in all cases within the

STUDY

OF OXYHEMOCYANIN

,\ 7

/, \a

[email protected] of the Remote Imidazole Position: NJ

5

energy range of the hydrogen bonds that we could not take into account. In addition these situations correspond to particularly deep potential wells, especially in the Sa case for which an absolute minimum is then reached (-138 kcal/mol). The C type rotation leads to three minima for the Hc situation and two for the Sa. Strong deepening of potential pits is also observed in this case, leading to a -125 kcal/mol deep well for cpc(Sa) = 60”. Electrostatic potential maps drawn for several energy minima all share an almost identical shape, very similar to those presented above. The main alteration consists of well depth value modifications and therefore modulations of the possible attraction strength toward a labile proton coming to proximity. We noticed that deeper pits are obtained with the Sa geometries when compared to those of Hc. The origin of the functional differences between

TABLE

In absence of any crystallographical datr, for tyrosinase and since the third imidazole in the coordination shell of the copper cations does not show up with EXAFS data, the importance of this remote ligand could appear questionable. It thus became interesting to consider the effect of the variation of the Cu - Ni distance on the energy of the global (Cu - Im& - O2 complex (Fig. 4) and on the corresponding electrostatic potential map. We have performed, for both Hc and Sa geometries, two sets of calculations (a = 0 and cy = 90”) stretching simultaneously the Cu-N, and Cu-N’i distances from 2.0 A (i.e., almost identical to the two 1.9-A short Cu - N) to infinity. It appears from the results reported in Fig. 4 that for both Hc and Sa structures, the (Y = 90” conformation is more stable than its (Y = 0’ counterpart for all d values. We can notice that the repulsion which is calculated for d(Cu-Ni) =G3 A with the Hc arrangement is due to the steric repulsion between the two fixed imidazoles and the corresponding incoming one. This destabilizing interaction, which counterbalances the copper attraction, does not take place in this range of distance for the Sa situation because of the larger intramolecular distances then observed (compare Figs. 1A and 1B). More striking is the influence of d(Cu-Ni) on the electrostatic potential maps of the complex. Not only does the overall shape of the corresponding MEP progressively

II

Energy and Electrostatic Potential Minima of the (Cuf-Im3)2-02 upon Different Imidazole Rotations (Values in kcal/mol) Hc type Rotation We A

C

0 30 150 0 60 240 330

Variation

of d(Cu -

c: SCHEME

253

SITE

tyrosinase (phenols/catechols oxidizer) and hemocyanin (0, carrier) could then be related to such a structural difference and consequently to geometrical variations analogous to those existing between our Hc and Sa models. This hypothesis, further supported in the conclusion of the next paragraph and by recent experiments (23), could constitute an alternative explanation to the “high accessibility for organic substrates” (lc) of the tyrosinase active site in regard to that of hemocyanin, previously proposed to explain their very different biological roles.

A:

B:

ACTIVE

Complexes

Sa type

Energy

Well depth

0.00 (ref.) -18.5 -18.5 0.00 (ref.) -17.0 -17.3 -10.8

-71 -109 -70 -71 -90 -78 -90

k&J 0 60 120 0 60 240

Energy

Well depth

0.00 (ref.) -3.9 -3.5 0.00 (ref.) -3.9 -3.7

-89 -138 -112 -89 -125 -105

254

EISENSTEIN

ET AL.

B

A

kcal/mol

kcal/mol

MIN=-24.77

MN=-32.78

-

7809

-

--

57.36

-.-

al06

---

14.77

76.44

. ..-. 4.699 -fl,w ‘, -49.15

FIG. 6. Molecular electrostatic potential maps (values in kcal/mol) in the CuN plane of the (Cu+---Im,),-Oa complexes with extreme stretch of Cu - Ni and Cu - Ni, bonds; (A) “hemocyanin” (Hc) X-ray geometry; (B) “standard average” (Sa) geometry. The isoenergy lines correspond to potential values identical to those depicted in Fig. 2.

change with this distance but so does the pit depth that decreases upon d(Cu - N,) lengthening. The final point with an infinite d(Cu - N,) value corresponds to the maps of Fig. 5, on which the potential well is considerably reduced. The major difference between the maps of Figs. 2 and 5 is due to the through-space contribution of the remote imidazoles to the potential well. Because of the scalar character of this quantity, the electrostatic potential of this ligand, which is maximum in the molecular plane close to N1 and N;, deepens considerably the potential well of the peroxo anion since they are located in the same plane. The shrinking of the proton attracting area in the modified active site may illustrate how the actual value of this distance parameter could contribute to the major functional differences between enzymatic systems considered here. Increasing d(Cu- N,) thus weakens the tyrosinase-like character of our active site model. Suggested Mechanism

and Conclusions

The results obtained in the present study are definitely in favor of a geometrical arrangement of the oxy form active site of hemocyanin in which the coordination bonds between the coppers and the remote imidazoles are perpendicular to the plane containing the metal ions and the dioxygen

molecule

(cu = 90”,

Scheme

3). The

calculations

carried out for different values of the torsion angle about the Cu- N coordination bond show that neither energies (in the context of neglected local intermolecular hydrogen bonds) nor potential maps are heavily altered upon imidazole rotation. On the other hand, the influence of the Cu-remote imidazole distance value appears to be very important for the depth of the electrostatic potential well associated to the dioxygen midbond.

The molecular electrostatic potential maps are able to account for the proton transfer scenario involved in the first step of phenol oxidation by oxytyrosinase. The positive area surrounding the copper ion (reflecting its positive charge) could lead, initially, to phenol oxygen attraction (8,20), which would then become the sixth ligand of a Jahn-Teller distorted octahedron (seeFig. 4) expected for the Cu(I1) ions (24) [the probable ionization state of the metal cations at the active site oxyform (3, 7, 20)]. This OH proton is now in proximity to the negative potential well which, acting in turn as an attractor, makes the proton transfer possible provided that the Cu - N1 distances and therefore well depth are appropriate. The proton transfer will be further facilitated by the phenolCu(I1) binding. Ab initio calculations have indeed shown that water deprotonation energy is decreased from 398 to 94 kcal/mol upon complexation between HZ0 and Cu(I1) (25). It is worth emphasizing that these results support the second [Fig. 14B of Ref. (7b)] of the two substrate complexation mechanisms recently proposed by Ross and Solomon. Going further in this mechanistic description would call for computational methods able to deal with transition states involving diradical species. Such techniques are, for systems of this size, far beyond the scope of the present study. ACKNOWLEDGMENTS O.E., C.G.-P., and J.M. thank the Groupement Scientifique Modelisation Moleculaire CNRS-IBM France for computational support, Dr. Y. L. Pascal for providing us with the molecular graphic program used throughout this work and D. Girou for his help in getting the various plots presented herein. D.S. and J.W. were supported by the Swiss National Science Foundation (Project 20-29856,901.

STUDY

OF OXYHEMOCYANIN

Note added in proof. After this paper was submitted, a new complex presenting main features of our Sa model has been published (27).

REFERENCES 1. (a) Schoot Uiterkamp, A. J. M., and Mason, H. S. (1973) Proc. Natl. Acad. Sci. U.S.A. 70,993-995; (b) Eickman, N. C., Solomon, E. I., Larrabee, J. A., Spiro, T. G., and Lerch, K. (1978) J. Am. Chem. Sot. 100,6529-6531; (c) Himmelwright, R. S., Eickman, N. C., Lu Bien, C. D., Lerch, K., and Solomon, E. I. (1980) J. Am. Chem. Sot. 102,7339-7344. 2. Lontie, R., Vanquickenborne, L. (1974) in “Metal Ions in Biological Systems” (Siegel, H., Ed.), Vol. 3, pp. 183-200, Dekker, New York. 3. Mason, H. S. (1956) Nature

177, 79-81.

4. (a) Korner, A., and Pawelek, J. (1982) Science 217,1163-1165; (b) Lerch, K. (1981) in “Metal Ions in Biological Systems” (Siegel, H., Ed.), Vol. 13, pp. 143-186, Dekker, New York/Basel. 5. For lead references see (a) Solomon, E. I. (1983) Struct. Bonding 53, 2-57; (b) Solomon, E. I. (1983) Pure Appl. Chem. 55, 10691088; (c) Woolery, G. L., Powers, L., Winkler, M., Solomon, E. I., Lerch, K., and Spiro, T. G. (1984) Biochim. Biophys. Acta 788, 155-161; (d) Vigato, P. A., Tamburini, S., and Fenton, D. E. (1990) Coord. Chem.Reu. 106,25-170. 6. (a) Blackburn, N. J., Strange, R. W., Cruse, R. W., and Karlin, K. D. (1987) J. Am. Chem. Sot. 109, 1235-1237; (b) Jacobson, R. R., Tyeklar, Z., Farooq, A., Karlin, K. D., Liu, S., and Zubieta, J. (1988) J. Am. Chem. Sot. 110, 3690-3692; (c) Sorrell, T. N. (1989) Tetrahedron 45, 3-68; (d) Kitajima, N., Fujizawa, K., and Moro-Oka, Y. (1989) J. Am. Chem. Sot. 111,8975-8976; (e) Tyeklar, Z., and Karlin, K. D. (1989) Act. Chem. Res. 22,241-248; (f) R&glier, M., Jorand, C., and Waegell, B. (1990) J. Chem. Sot. Chem. Comm., 1752-1755; (g) Kitajima, N., Koda, T., Iwata, Y., and Moro-Oka, Y. (1990) J. Am. Chem. Sot. 112,8833-8839; (h) Sanyal, I., Strange, R. W., Blackburn, N. J., and Karlin, K. D. (1991) J. Am. Chem. Sot. 113,4692-4693; (i) Ngwenya, M. P., Chen, D., Martell, A. E., and Reibenspies, J. (1991) Inorg. Chem. 30, 2732-2736. 7. (a) Ross, P. K., and Solomon, E. I. (1990) J. Am. Chem. Sot. 112, 5871-5872; (b) Ross, P. K., and Solomon, E. I. (1991) J. Am. Chem. Sot. 113, 3246-3259; (c) Maddaluno, J., and Giessner-Prettre, C. (1991) Inorg. Chem. 30, 3439-3445; (d) Baldwin, M. J., Ross, P. K., Pate, J. E., Tyeklar, Z., Karlin, K. D., and Solomon, E. I. (1991) J. Am. Chem. Sot. 113,8671-8679. 8. (a) Winkler, M. E., Lerch, K., and Solomon, E. I. (1981) J. Am. Chem. Sot. 103, 7001-7003; (b) Paul, P. P., Tyeklar, Z., Jacobson, R. R., and Karlin, K. D. (1991) J. Am. Chem. Sot. 113,5322-5332. 9. Weber, J., Fluekiger, P., Morgantini, P.-Y., Schaad, O., Goursot, A., and Daul, C. (1988) J. Comp. Aided Mol. Des. 2, 235-253. 10. (a) Hoffmann, R. (1982) Angew. Chem. Znt. Ed. Eng. 21, 711-724; (b) Lowe, J. P. (1978) in Quantum Chemistry, pp. 283-299, Academic Press, London.

ACTIVE

SITE

255

11. Ammeter, J. H., Burgi, H.-B., Thibeault, J. C., and Hoffmann, R. (1978) J. Am. Chem. Sot. 100, 3686-3706. 12. (a) Hay, P. J., Thibeault, J. C., and Hoffmann, R. (1975) J. Am. Chem. Sot. 97,4884-4899; (b) Hoffmann, R. (1963) J. Chem. Phys. 39,1397-1412. P. Y., and Weber, J. (1990) Znt. 13. Daul, C., Goursot, A., Morgantini, J. Quark Chem. 38,623-640. 14. See, e.g., Scrocco, E., and Tomasi, J. (1978) Adu. Quantum Chem. 11, 115; Politzer, P., and Truhlar, D. G., (Eds.) (1981) Chemical Applications of Atomic and Molecular Electrostatic Potentials, Plenum, New York. 15. (a) Volbeda, A., and Hol, W. G. J. (1989) J. Mol. Biol. 209, 249279; (b) Gaykema, W. P. J., Volbeda, A., and Hol, W. G. J. (1985) J. Mol. Biol. 187, 255-275. 16. Volbeda, A., Hol, W. G. J. (1988) in Oxydases and Related Redox Systems (King, T. E., Mason, H. S., Morrisson, M., Eds., pp. 291307, A. R. Liss Inc., New York. 17. (a) Basch, H., Krauss, M., and Stevens, W. J. (1987) Int. J. Quantum C/rem. 31,405-415; (b) Demoulin, D., and Pullman, A. (1978) Theor. Chim. Acta 49, 161-181; (c) Giessner-Prettre, C., and Jacob, 0. (1989) J. Comp. Aided Mol. Des. 3, 23-37. 18. See, e.g., (a) Holmes, M. A., and Matthews, B. W. (1982) J. Mol. Biol. 163, 623-639; (b) Tainer, J. A., Getzoff, E. D., Richardson, J. S., and Richardson, D. C. (1983) Nature 306,284-287; (c) Christianson, D. W., Lipscomb, W. N. (1989) Act. Chem. Res. 22, 6269. C., and Maddaluno, J., unpublished results. 19. Giessner-Prettre, 20. Masuda, H., Odami, A., Yamauchi, 0. (1989) Inorg. Chem. 28,624625. 21. Karlin, K. D., Ghosh, P., Cruse, R. W., Farooq, A., Gultnehy, Y., Jacobson, R. R., Blackburn, N. J., Strange, R. W., and Zubieta, J. (1988) J. Am. Chem. Sot. 110,6769-6780. 22. Due to space limitations, the whole set of corresponding electrostatic maps could not be reproduced here but is available upon request to C. Giessner-Prettre. 23. Casella, L., Gulloti, M., Bartosek, M., Pallanza, G., and Laurenti, E. (1991) J. Chem. Sot. Chem. Comm. 1235-1237. 24. Ohtaki, H., Yamaguchi, T., and Maeda, M. (1976) Bull. Chem. Sot. Japn. 49, 701-708; (b) Garcia, J., Benfatto, M., Natoli, C. R., Bianconi, A., Fontaine, A., and Tolentino, H. (1989) Chem. Phys. 132,

295-307. C., unpublished results. This large decrease (304 25. Giessner-Prettre, kcal/mol) of the deprotonation energy is consistent with corresponding results for Zn2+ complexation of water (17c, 26) and methanol (26). 26. Soli, M., Lledos, A., Duran, M., and Bertran, J. (1991) Inorg. Chem. 30,2523-2527. 27. Kitajima, N., Fujisawa, K., Fujimoto, C., Moro-oka, Y., Hashimoto, S., Kitagawa, T., Toriumi, K., Tatsumi, K., and Nakamura, A. (1992) J. Am. Chem. Sot. 114, 1277-1291.