1: %%2/25/00
2: \documentstyle[preprint,prb,aps]{revtex}
3: \tightenlines
4: \begin{document}
5: \draft
6: \title{The Functional Form of Angular Forces around Transition Metal Ions in Biomolecules}
7: \author{A. E. Carlsson and S. Zapata}
8: \address{Department of Physics\\
9: Washington University\\
10: St. Louis, Missouri 63130-4899}
11: \date{Submitted to Biophysical Journal, 28 February 2000}
12: \maketitle
13: \begin{abstract}
14: A method for generating angular forces around $\sigma$-bonded transition
15: metal ions is generalized to treat $\pi$-bonded configurations.
16: The theoretical approach is based on an analysis of a ligand-field
17: Hamiltonian based on the moments of the electron state distribution.
18: The functional forms that are obtained involve a modification of
19: the usual expression of the binding energy as a sum of ligand-ligand
20: interactions, which however requires very little increased in CPU time.
21: The angular interactions have simple forms involving sin and cos
22: functions, whose relative weights depend on whether the ligands are
23: $\sigma$- or $\pi$-bonded. They describe the ligand-field stabilization
24: energy to an accuracy of about 10\%. The resulting force field
25: is used to model the structure of small clusters, including fragments
26: of the copper blue protein structure. Large deviations from the typical
27: square copper coordination are found when $\pi$-bonded ligands are
28: present.
29: \end{abstract}
30: \pacs{Running title: Angular Forces around Ions\\
31: Keywords:\qquad Metalloproteins, Enzyme structure, Ligand-field
32: splitting,\\ Force fields, Metal complexes}
33:
34: %\bibliographystyle{prsty}
35:
36: \newpage
37:
38: \section{Introduction}
39:
40: Atomic-level simulations of proteins interacting with transition-metal
41: ions have the promise of elucidating a wide variety of biophysical
42: %%phenomena\cite{Kaim94}.
43: phenomena~(Kaim, 1994).
44: %For example, the function of both redox and non-redox enzymes often
45: %is centered on one or more metal ions. In these cases, simulations
46: %can elucidate the underlying structure and how it response to substrate
47: %molecules or to changes in the charge of the metal ions. Metal ions
48: %often served as structural ``glue" in stabilizing the tertiary structure
49: %of folded proteins (such as zinc
50: %%fingers\cite{}),
51: %fingers~(?..., 19..?)), and the secondary
52: %structure of shorter peptides. Simulations can aid in the prediction of
53: %the extent and specificity of such ``glue" effects.
54: Such simulations can help establish both the native structure and
55: reaction paths of metalloproteins, and explain the roles of metal
56: ions in protein folding and stability. Similar applications can
57: be seen for simulations of metal ions interacting with DNA and RNA.
58: However, the utility of such simulations depends critically on the
59: availability of accurate but computationally tractable force fields
60: for metals interacting with proteins. There are no quantitatively
61: accurate force fields for these purposes, and it is probably not
62: possible to construct force fields of a simple enough form to be
63: useful that have quantitative accuracy.
64: However, one can hope to generate force fields that include the basic
65: physical effects at a level of accuracy sufficient to ascertain
66: chemical trends as metals or ligands vary. The situation in this regard
67: is best for ``simple" metal ions that have no partly occupied shells.
68: Here an ionic picture based on electrostatics, supplemented by empirical
69: repulsive forces, may hope to treat the most important physical effects.
70: Even here though, care should be taken, since the bond is not completely
71: ionic but has some partial covalent character. For transition metals,
72: the situation is much more difficult. The ligand-field splitting
73: of the $d$-shell leads to important electronic contributions which
74: cannot be ignored. These are manifested, for example, in the typical
75: square or tetragonal coordination of Cu$^{2+}$ complexes.
76: Such effects cannot be described by radial interactions,
77: but rather require the introduction of angular forces describing
78: the energetics of the transition-metal $d$-shell.
79: Most existing methods for including angular forces in simulations
80: of metal ions have used assumed functional
81: %%forms\cite{Comba95,Allured91,Timofeeva95,Wiesemann94,Sayle95,Sayle97}
82: forms~(Comba et al., 1995, Allured et al., 1991, Timofeeva et al., 1995,
83: Wiesemann et al., 1994, Sayle et al., 1995, Sayle et al., 1997)
84: based, for example on the observed structures of small complexes.
85: A recently developed
86: %method\cite{Landis98}
87: method~(Landis et al., 1998)
88: treats the environment-dependent energetics of transition metals
89: via a valence-bond approach. This method appears promising for
90: covalently bonded metals. However, the bonding state of transition
91: metals in proteins is may be rather different from the $sd^n$
92: hybridized configuration assumed in that work. In the present analysis,
93: we seek to develop a method suitable for more ionic configurations.
94:
95: We have recently
96: %shown\cite{Carlsson98}
97: shown~(Carlsson, 1998) how the ligand-field splitting effects of
98: $\sigma$-bonded transition metal ions can be described by a force
99: field of a fairly simple form. One starts with an explicitly
100: quantum-mechanical form for the electronic energy, and then then
101: solves the quantum-mechanical problem with systematic approximation
102: methods to extract the real-space description of the electronic
103: bonding energy. A second-order treatment of the hybridization
104: terms between the ligand and metal orbitals is used to generate
105: a ligand-field Hamiltonian for the $d$-electrons. The electronic
106: bonding energy of this Hamiltonian is analyzed in terms its moments.
107: The moment analysis of the energy gives a ``semiclassical" energy
108: function with a simple trigonometric angular form.
109: It is not precisely an additive function of ligand-ligand interactions,
110: but is nearly as simple computationally. This energy function gives
111: quite accurate energy results for $\sigma$-bonded ligands,
112: with less than 10\% error in the ligand-field stabilization energy
113: (defined below).
114:
115: This approach is extended in the present paper, in two ways.
116: First, we develop energy functions which treat $\pi$-bonded
117: ligands as well. This results in new and distinct angular forms.
118: Second, we present results for the lowest-energy structures
119: of small clusters, which are a guide to understanding the bonding
120: preferences of transition metals in proteins.
121:
122: The organization of the remainder of the paper is as follows.
123: Section~II develops the formalism underlying the angular
124: forces. Section~III gives tests of the functional form of
125: the angular forces by comparing with results from diagonalization
126: of simple cluster Hamiltonians. Section~IV gives the small-cluster
127: results for ideal examples, and for an analogue of the copper
128: environment in blue-copper proteins.
129:
130:
131: \section{Model}
132:
133: We model the electronic structure of the environment of a single
134: transition metal ion in terms of orbitals $|L,\mu\rangle$
135: localized on the ligands and $|M,\nu\rangle$ localized on the
136: single metal ion. The orbitals taken to be orthogonal for simplicity
137: of calculation. Here ``L" denotes a particular ligand atom and
138: the $|L,\mu\rangle$ are distinct orbitals on that atom.
139: The Hamiltonian takes the following form:
140: \begin{eqnarray}
141: \hat H =
142: && \sum_{L,\mu} \varepsilon_{L,\mu} |L,\mu \rangle\langle L,\mu |
143: +\sum_\nu \varepsilon_{M,\nu}
144: |M,\nu \rangle\langle M,\nu |\nonumber \\
145: %+ && \sum_{L,\mu ,\nu } [h_{LM}^{\mu \nu} |L,\mu ><M,\nu | +
146: && + \sum_{L,\mu ,\nu} \Bigl[ h_{LM}^{\mu\nu}
147: |L,\mu \rangle\langle M,\nu | + h_{ML}^{\nu \mu}
148: |M,\nu \rangle\langle L,\mu | \Bigr]\quad .
149: \label{eqnarray}
150: \end{eqnarray}
151: Here the $\varepsilon$ terms are on-site energies for the orbitals,
152: and the $h$ terms are hybridization energies between the ligand orbitals
153: and the metal orbitals. We ignore electron-interaction terms as
154: well as explicit electrostatic effects. The energy associated with
155: this Hamiltonian can then be obtained by diagonalizing its matrix
156: as given in the $|L,\mu \rangle -|M,\nu\rangle$ basis,
157: and taking the sum of the energies of the occupied eigenvectors.
158:
159: To obtain a real-space description of the energetics of this Hamiltonian,
160: we make the simplifying approximation that the dimensionless ratios
161: $|h_{LM}^{\mu \nu}|/(\varepsilon_{M,\nu}-\varepsilon_{L,\mu})$ are small
162: and can be used as expansion parameters. This will be the case if the
163: bonding is primarily ionic as opposed to covalent. In the case of
164: transition metal ions, this allows a great simplification by the
165: application of ligand-field theory, which gives an effective
166: Hamiltonian for the $d$-shell:
167: \begin{equation}
168: \hat H_{d}
169: = \sum_{\mu ,\nu}h_d^{\mu\nu} |M,\mu \rangle\langle M,\nu |\quad ,
170: \label{hd}
171: \end{equation}
172: where
173: \begin{equation}
174: h_d^{\mu\nu}
175: = \sum_{L,\eta}(\varepsilon_d - \varepsilon_{L,\eta })^{-1}
176: h_{ML}^{\mu \eta} h_{LM}^{\eta \nu}\quad .
177: \label{hlft}
178: \end{equation}
179: Note that we have assumed all of the $d$-orbitals on the transition metal
180: ion to have the same energies before ligand-field effects are ``turned on".
181:
182: In a fully quantum-mechanical ligand-field-theory calculation,
183: one would numerically diagonalize the matrix $\hat H_d$.
184: In order to obtain a force field with a nearly classical form,
185: we instead work with the moments of this matrix,
186: defined in terms the traces of its powers, as follows.
187: The first moment is defined as the average energy of the $d$-complex, or
188: \begin{eqnarray}
189: \bar \varepsilon
190: &=& (1/5) {\rm Tr}\hat H_d
191: =(1/5)\sum_{L,\eta}(\varepsilon_d -\varepsilon_{L,\eta})^{-1}
192: \sum_\mu h_{ML}^{\mu \eta} h_{LM}^{\eta \mu} \nonumber \\
193: &=& (1/5)\sum_{L,\eta}(\varepsilon_d -\varepsilon_{L,\eta})^{-1}
194: g_{\eta\eta}
195: \label{mu1}
196: \end{eqnarray}
197: where
198: \begin{equation}
199: g_{\eta ,\eta '}=\sum_\mu h_{L'M}^{\eta '\mu}h_{ML}^{\mu\eta}\quad .
200: \label{g}
201: \end{equation}
202: %In order to get a better feeling for these terms, we note
203: %that $g_{\eta\eta}=\langle L,\eta |\hat P_d | L,\eta\rangle$, where
204: %$\hat P_d=\sum_\mu |M,\mu \rangle\langle M,\mu |$ is the projection onto the
205: %$d$-subspace of the transition metal ion. Since $\hat P_d$ is a
206: %rotationally invariant operator, it follows that $g_{\eta \eta}$
207: %is a radial function of the distance between the ligand L and the metal ion.
208: Thus $\bar\varepsilon$ is given as a sum of independent contributions
209: from the ligands. Because the $d$-shell by itself has spherical symmetry,
210: the contribution from each ligand is a radial function (no angular
211: dependence) of the metal-ligand distance.
212: Our major interest is in the angular terms resulting from the
213: ligand-field splitting, so we do not consider the $\bar\varepsilon$
214: term further.
215:
216: The width $W$ of the $d$-complex corresponding to the ligand-field
217: splitting can in the simplest picture be described in terms of
218: the second moment or variance $\delta\varepsilon^2$ of the
219: eigenvalues of $\hat H_d$. We expect that
220: $W\propto\sqrt{\delta\varepsilon^2}$. We can obtain a simple
221: real-space form for $\delta\varepsilon^2$ as follows: First we
222: note that
223: $\delta\varepsilon^2=(1/5)\sum_n(\varepsilon_n-\bar\varepsilon)^2
224: =(1/5){\rm Tr}(\hat H_d-\bar\varepsilon\hat I )^2$
225: where the $\varepsilon_n$ are the eigenvalues of $\hat H_d$.
226: From Eqs.~(\ref{hd}), (\ref{hlft}), and (\ref{g}), we see that
227: \begin{eqnarray}
228: {\rm Tr}\hat H_d^2
229: &=& \sum_{\mu ,\nu}h_d^{\mu\nu} h_d^{\nu\mu} \nonumber \\
230: &=& \sum_{L,\eta ,L',\eta '}(\varepsilon_d-\varepsilon_{L,\eta})^{-1}
231: (\varepsilon_d-\varepsilon_{L',\eta '})^{-1}
232: \sum_\mu h_{L'M}^{\eta ' \mu}h_{ML}^{\mu \eta}
233: \sum_\nu h_{LM}^{\eta\nu}h_{ML'}^{\nu\eta '} \nonumber \\
234: &=& \sum_{L,\eta ,L',\eta '}(\varepsilon_d-\varepsilon_{L,\eta})^{-1}
235: (\varepsilon_d-\varepsilon_{L',\eta '})^{-1} g_{\eta \eta '}^2\quad .
236: \end{eqnarray}
237: Then the variance is
238: \begin{eqnarray}
239: \delta\varepsilon^2
240: &=& (1/5){\rm Tr}\Bigl( \hat H_d-\bar\varepsilon \hat I\Bigr)^2
241: =(1/5)\Bigl( {\rm Tr}\hat H_d^2 -5\bar\varepsilon^2\Bigr)\nonumber \\
242: &=& (1/5)\sum_{L,\eta ,L', \eta '}
243: (\varepsilon_d-\varepsilon_{L,\eta})^{-1}
244: (\varepsilon_d-\varepsilon_{L',\eta '})^{-1}
245: \Bigl[ g_{\eta \eta '}^2-(1/5)g_{\eta\eta}
246: g_{\eta ',\eta '}\Bigr]\quad .
247: \label{variance}
248: \end{eqnarray}
249:
250: We will focus on the ligand-field stabilization energy, which is
251: defined as the sum of the energies of the occupied orbitals
252: relative to $\bar\varepsilon$, or
253: $E_{\rm LFSE}=\sum_{n}' (\varepsilon_n -\bar\varepsilon )$,
254: where the sum is over only the occupied eigenfunctions.
255: We assume that $E_{\rm LFSE}$ is proportional to the ligand-field
256: splitting $W$, and also identify $W$ with the standard deviation
257: $\sqrt{\delta \varepsilon^2}$.
258: Thus our expression for the ligand-field stabilization energy
259: has the form
260: \begin{equation}
261: E_{\rm LFSE} = -\biggl\{ \alpha /5
262: \sum_{L,\eta ,L', \eta '}
263: (\varepsilon_d-\varepsilon_{L,\eta})^{-1}
264: (\varepsilon_d-\varepsilon_{L',\eta '})^{-1}
265: \Bigl[ g_{\eta\eta '}^2-(1/5) g_{\eta\eta}
266: g_{\eta ',\eta '}\Bigr]\biggr\}^{1/2}\quad ,
267: \label{elfse}
268: \end{equation}
269: where $\alpha$ is a dimensionless constant which will later be
270: used as a fitting parameter.
271:
272: The accuracy of this form will be demonstrated in the next section
273: via specific numerical experiments. At present, we will show how
274: this form for $E_{\rm LFSE}$ results in an expression for the energy
275: in terms of simple angular interactions between he ligands.
276: We thus need to develop analytic forms for the $g_{\eta\eta '}$.
277: First we note that we can write
278: \begin{eqnarray}
279: g_{\eta \eta '}
280: &=& \langle L,\eta | \hat H P_d^2 \hat H |L', \eta '\rangle \nonumber \\
281: &=& \langle \Psi | \Psi '\rangle \quad ,
282: \end{eqnarray}
283: where $\hat P_d = \sum_\mu |M,\mu\rangle\langle M,\mu |$
284: is the projection onto the $d$-subspace of the transition metal ion
285: Here we define $|\Psi \rangle =\hat P_d H |L,\eta\rangle$
286: and we have have used the relation $\hat P_d^2 = \hat P_d$, which
287: follows from $\hat P_d$'s being a projection operator.
288:
289: Consider first the case where both the $|L,\eta\rangle$ and
290: $|L',\eta '\rangle$ orbitals have $\sigma$-character with respect to
291: the metal ion. Then they couple only to the $d$-orbitals
292: $|M,\sigma\rangle$ and $|M,\sigma '\rangle$ that have
293: $\sigma$-character with respect to the bond axes, so that
294: $|\Psi\rangle = h_\sigma |M, \sigma\rangle$ and
295: $|\Psi '\rangle = h_{\sigma }'|M, \sigma ' \rangle$, where
296: $h_\sigma$ and $h_{\sigma}'$ are the appropriate coupling strengths.
297: Thus
298: $g_{\eta\eta '}=h_\sigma h_{\sigma}'\langle M,\sigma | M,\sigma '\rangle$,
299: and the angular dependence is contained in the last inner product.
300: But this is simply a matrix element of a rotation about $M$ which
301: carries $L$ into $L'$. We note that $|M, \sigma\rangle$ is equivalent
302: to $|M,m=0\rangle$, where $m=0$ denotes the angular dependence of
303: the spherical harmonic $Y_{2m}$.
304: Choosing our coordinate system so that $L$ is along the
305: $z$-axis and $L'$ is in the $z$-$x$ plane, we find that
306: \begin{eqnarray}
307: \langle M,\sigma | M,\sigma '\rangle
308: &=& D_{00}^{(2)}(0,\theta , 0) \nonumber \\
309: &=& (3 \cos^2 \theta -1)/2
310: \end{eqnarray}
311: where the $D$-term is a matrix element of the $l=2$ representation
312: of the rotation group, and the second equality follows from the
313: explicit formulas of the $D$-terms given in
314: %%Ref.\onlinecite{Wigner59}.
315: Ref.~(Wigner, 1959).
316: In summary, for $\sigma$-bonded ligands,
317: \begin{equation}
318: g_{\eta\eta `}=h_\sigma h_{\sigma}' (3\cos^2\theta -1)/2\quad .
319: \label{gsigma}
320: \end{equation}
321:
322: For the case of a $\pi$-bonded orbital $L$ and a $\sigma$-bonded
323: orbital $L'$, we consider only the case in which the axis of the orbital
324: lies along the circle connecting $L$ and $L'$; if it is perpendicular
325: to this circle, $L$ and $L'$ have different inversion symmetries so
326: their coupling vanishes. We write
327: $| \Psi\rangle =h_{\pi} | M,\pi\rangle$ where
328: $|M,\pi\rangle =(1/\sqrt{2})(-|M,m=1\rangle +|M, m=-1\rangle )$ and
329: $m$ the usual aximuthal angular momentum index for the spherical harmonics.
330: (The ``$-$" sign comes from the definition of the spherical harmonics).
331: Then, again choosing a coordinate system in which $L$ is along the
332: $z$-axis and $L'$ is in the $z$-$x$ plane, and following reasoning
333: parallel to the $\sigma$-bonded case, we see that
334: \begin{eqnarray}
335: \langle M,\pi | M,\sigma '\rangle
336: &=& (1/\sqrt{2}) \Bigl[ -D_{10}^{(2)}(0,\theta ,0)
337: + D_{-10}^{(2)}(0,\theta ,0)\Bigr] \nonumber \\
338: &=& \sqrt{3} \sin{\theta}\cos{\theta}\quad .
339: \end{eqnarray}
340: and
341: \begin{equation}
342: g_{\eta\eta `}=\sqrt{3}h_{\pi} h_{\sigma}'\sin{\theta}\cos{\theta}\quad .
343: \label{gpisigma}
344: \end{equation}
345:
346: Finally we turn to the case of two $\pi$-bonded orbitals.
347: To obtain the subsequent results, it is sufficient to consider the
348: case~1) in which the orbitals are parallel to the arc connecting
349: $L$ and $L'$, and the case~2) in which they are perpendicular to it.
350: By reasoning similar to that above, one sees that in case~1)
351: \begin{eqnarray}
352: \langle M, \pi | M, \pi '\rangle
353: &=& (1/2)\Bigl[ D_{11}^{(2)}(0,\theta ,0) +D_{-1-1}^{(2)}(0,\theta ,0)
354: - D_{1-1}^{(2)}(0,\theta ,0)
355: - D_{-11}^{(2)} (0,\theta ,0)\Bigr] \nonumber \\
356: &=& \cos{2\theta}
357: \end{eqnarray}
358: and
359: \begin{equation}
360: g_{\eta\eta `} = h_{\pi} h_{\pi}' \cos{2\theta}\quad .
361: \label{gpipia}
362: \end{equation}
363: In case 2),
364: \begin{eqnarray}
365: \langle M,\pi | M,\pi '\rangle
366: &=& (1/2) \Bigl[ D_{11}^{(2)}(0,\theta ,0)
367: + D_{-1-1}^{(2)}(0,\theta ,0)
368: + D_{1-1}^{(2)}(0,\theta ,0)l
369: + D_{-11}^{(2)}(0,\theta ,0)\Bigr] \nonumber \\
370: &=& \cos{\theta}
371: \end{eqnarray}
372: and
373: \begin{equation}
374: g_{\eta\eta `} = h_{\pi} h_{\pi}' \cos{\theta}\quad .
375: \label{gpipib}
376: \end{equation}
377:
378: Thus, combining Eqs.~(\ref{elfse}), (\ref{gsigma}), (\ref{gpisigma}),
379: (\ref{gpipia}), and (\ref{gpipib}), we find the following explicit
380: form for $E_{\rm LFSE}$ as a sum of ligand-ligand interactions:
381: \begin{equation}
382: E_{\rm LFSE}
383: = -\sqrt{\sum_{L,\eta ,L',\eta '}} U(L,\eta ;L',\eta ')
384: \label{efinal}
385: \end{equation}
386: where the ligand-ligand interaction $U$ is given as follows:
387: \bigskip
388:
389: \noindent{For $\sigma$-$\sigma$ interactions,}
390: \begin{equation}
391: ~~~~~~~~U(L,\eta ;L',\eta ')
392: = (\alpha /5) e_{\sigma}(r) e_{\sigma}(r') u_{\sigma\sigma}(\theta )
393: %[(h_{\sigma}^2
394: %/(\varepsilon_d - \varepsilon_{L,\eta })]
395: %[{h_{\sigma}'}^2 /(\varepsilon_d - \varepsilon_{L,\eta '})]
396: %(9\cos^4{\theta}-6\cos^2{\theta}+1/5)/4,
397: \label{usigma}
398: \end{equation}
399: \noindent{for $\pi$-$\sigma$ interactions,}
400: \begin{equation}
401: ~~~~~~~~U(L,\eta ;L',\eta ')
402: = (\alpha /5)e_{\pi}(r) e_{\sigma}(r') u_{\pi\sigma}(\theta )
403: %[(h_{\pi}^2
404: %/(\varepsilon_d - \varepsilon_{L,\eta })]
405: %[{h_{\sigma}'}^2 /(\varepsilon_d - \varepsilon_{L,\eta '})]
406: %(-3\cos^4{\theta}+3\cos^2{\theta}-2/5),
407: \label{upisig}
408: \end{equation}
409: \noindent{and for $\pi$-$\pi$ interactions,}
410: \begin{equation}
411: ~~~~~~~~U(L,\eta ;L',\eta ')
412: = (\alpha /5) e_{\pi}(r) e_{\pi}(r') u_{\pi\pi}(\theta )
413: %[(h_{\pi}^2
414: %/(\varepsilon_d - \varepsilon_{L,\eta })]
415: %[{h_{\pi}'}^2 /(\varepsilon_d - \varepsilon_{L,\eta '})]
416: %(4\cos^4{\theta}-3\cos^2{\theta}+1/5),
417: \label{upipi}
418: \end{equation}
419: where
420: \begin{eqnarray}
421: e_{\sigma}(r)
422: = && {h_{\sigma}}^2 /(\varepsilon_d - \varepsilon_{L,\eta }), \\
423: e_{\pi}(r)
424: = && {h_{\pi}}^2 /(\varepsilon_d - \varepsilon_{L,\eta '}), \\
425: u_{\sigma\sigma}(\theta)
426: = && (9\cos^4{\theta}-6\cos^2{\theta}+1/5)/4, \\
427: u_{\pi\sigma}(\theta )
428: = && (-3\cos^4{\theta}+3\cos^2{\theta}-2/5), \\
429: u_{\pi\pi}(\theta )
430: = && (4\cos^4{\theta}-3\cos^2{\theta}+1/5)\quad .
431: \label{eandu}
432: \end{eqnarray}
433: In each case, $U$ is given as a product of radial terms involving
434: the two ligands, and a simple angular function. (Note that although
435: denoted an interaction here, $U$ does not have units of energy because
436: of the square root in Eq.~(\ref{efinal})). In the calculation of the
437: terms involving $\pi$-$\sigma$ interactions, each term involves
438: a sum over two $\pi$-orbitals, parallel and perpendicular to the
439: arc connecting the two ligands. In the latter case, $g_{\eta\eta '}$
440: in Eq.~(\ref{elfse}) vanishes, but $g_{\eta\eta}$ and
441: $g_{\eta ' \eta '}$ do not. In the calculation of the terms involving
442: $\pi$-$\pi$ interactions, one has a similar scenario except that one
443: sums over two pairs of $\pi$-orbitals.
444:
445: These forms are plotted out in Figure~1a. Note that
446: the $\sigma$-$\sigma$ interactions are fairly similar in form
447: to the $\pi$-$\pi$ interactions, both having pronounced minima
448: at $180^{\circ}$ (as well as the physically irrelevant one at
449: $0^{\circ}$), and a shallower minimum at $90^{\circ}$. The
450: $\sigma$-$\pi$ interaction is complementary to these,
451: having minima at $45^{\circ}$ and $135^{\circ}$. We shall
452: see later that these differences lead to large differences
453: in ground-state structures of small clusters.
454: For comparison, we show in Figure 1b an empirical angular interaction
455: %curve\cite{Comba95}
456: curve~(Comba et al., 1995)
457: assumed in some previous calculations of small transition-metal structures.
458: The angular dependence is based on the observed square structure of small
459: complexes of the transition metals of interest, and is proportional to
460: $\sin^2{2\theta}$. Since the metals which have square coordination
461: generally have predominantly $\sigma$-bonds to their neighbors,
462: the most relevant comparison is to the $\sigma$-$\sigma$ curve in Figure~1a.
463: We see that the behavior is quite different. The empirical curve has
464: equivalent minima at $180^{\circ}$ and $90^{\circ}$, while in the
465: theoretical curve the $180^{\circ}$ minimum is much deeper.
466:
467: \newpage
468: \section{Tests of Functional Form}
469:
470: In order to evaluate the accuracy of the semiclassical form of
471: Eq.~(\ref{elfse}) for the ligand-field stabilization energy, we
472: have performed explicit tests for small clusters. These clusters
473: consist of a central transition-metal ion with four neighbors
474: placed at random orientations at random distances relative to
475: the central ion.
476: We consider only the minority-spin orbitals, as these determine
477: $E_{\rm LFSE}$ for high-spin late transition metals. Four
478: electrons are placed in these states, corresponding to
479: Cu$^{2+}$; other band filling values gives similar results.
480: The random distances are taken into account by
481: varying the couplings $h_{\sigma}$ and $h_{\pi}$ uniformly over
482: a finite interval ranging from zero to $h_{\rm max}^\sigma$ or
483: $h_{\rm max}^\pi$. We take $h_{\rm max}^\pi = 0.5h_{\rm max}^\sigma$.
484: The energies are obtained by explicit diagonalization of a
485: tight-binding Hamiltonian for this cluster. All of the ligand
486: orbitals are taken to have the same value of
487: $\varepsilon_{L,\eta}$. The only dimensionless
488: variable that enters the results is then
489: $\gamma = |h_{\rm max}/(\varepsilon_d -\varepsilon_{L,\eta})|$.
490: For small values of $\gamma$, the bonding is primarily ionic,
491: and for larger values it acquires more covalent character.
492: We use $\gamma = 0.1$ for our results, but even when $\gamma$ is
493: significantly larger we find that the accuracy of the
494: semiclassical form is essentially the same.
495:
496: We have fitted the energies of these clusters to the following
497: semiclassical form for the energy:
498: \begin{equation}
499: E_{\rm LFSE}^2
500: = {\sum_{ L,\eta ,L',\eta '}} U(L,\eta ;L',\eta ') +\beta \quad ,
501: \label{efit}
502: \end{equation}
503: involving two parameters $\alpha$ and $\beta$ (where $U$ contains
504: $\alpha$). We use a database
505: of 10,000 clusters to determine the parameters, and then test them on
506: a set of 1000 clusters not included in the ``training" set. Typical
507: results are shown in Fig.~2, which shows the semiclassical energies
508: vs. exact energies for 1000 clusters. Figure~2a corresponds to
509: three $\sigma$-bonded ligands and one $\pi$-bonded one, while Fig.~2b
510: corresponds to four $\pi$-bonded ligands. The rms errors in
511: $E_{\rm LFSE}$ for these two cases are 9\% and 14\%, respectively.
512: For clusters with two and three $\pi$-bonded ligands, the rms
513: errors are 10\% and 11\%. This is to be compared with rms errors
514: of 25\%--30\% that are obtained with empirical force
515: %%fields\cite{Carlsson98}.
516: fields~(Carlsson, 1998).
517:
518: \newpage
519: \section{Small Cluster Minimum-Energy Structures}
520:
521: In this section, we describe some of the implications
522: of the angular forms developed above for the structure
523: of small model clusters consisting of a metal atom and
524: four ligands. Such cluster calculations cannot treat the protein
525: environment accurately; this would await parametrization and
526: incorporation of the force field into protein codes, which
527: is in progress in our group. However, from the small-cluster
528: calculations it is possible to see the structural preferences
529: of the angular forces by themselves. These are an important
530: factor in the final structure adopted by a protein.
531:
532: To keep the calculations as simple as possible, we place the
533: ligands at frozen bond lengths from the central metal ion.
534: Here ``frozen means" that they are fixed at a given set of values,
535: but these values are not necessarily the same for all of the ligands.
536: The energy terms include the ligand-field stabilization energy as
537: described by our angular terms, as well as a repulsive radial interaction
538: between ligands forbidding close approaches. The latter term is needed
539: because otherwise spurious structures involving
540: $45^{\circ}$ bond angles can appear when $\pi$-bonding ligands
541: are present. For simplicity, we assume that the values of
542: $h_{\sigma}$ and $h_{\pi}$ at a given distance are the same.
543: To evaluate the magnitudes of these couplings, we assume
544: a value of 100~kJ/mole~$=23.9$~kcal/mole~$= 1.04$~eV for
545: the ligand-field stabilization energy of a cluster with
546: four $\sigma$-ligands in square coordination; this number
547: would correspond to the more strongly stabilized
548: %complexes\cite{Cotton72}.
549: complexes~(Cotton, 1972).
550: For clusters with unequal bond lengths, we assume
551: an exponential decay, so that
552: $h_{\sigma}(r)=h_{\sigma}(r_0)\exp{[-\kappa (r-r_0)]}$,
553: where $r_0$ is the reference distance and $\kappa$ is a decay parameter.
554: Since the most prominent case that we consider of a distant ligand is
555: sulfur, we identify $\kappa$ with the spatial decay rate coming from
556: the measured first ionization energy $E_1$ of sulfur,
557: 10.4~eV (=~240~kcal/mole), using the formula
558: $\hbar^2 \kappa^2/2m = E_1$. This yields
559: $\kappa = 1.65${\AA}$^{-1}$. The ligand-ligand terms
560: contain an exponential term taken as the repulsive part
561: of the van der Waals interactions as given in the ``MM2"
562: %force field\cite{Allinger87}.
563: force field~(Sprague et al., 1987).
564: In our model clusters, we use parameters and bond lengths typical
565: for copper or nickel interacting with nitrogen and/or sulfur ligands.
566: This is because copper and nickel have the largest ligand-field energies
567: among the 3d transition metals, and typical ligands for these metals
568: are nitrogen and sulfur.
569: %First ionization energy of S is 10.4 eV. This gives
570: %WAVE FUNCTION decay length of 0.605 AA, or
571: %0.303 AA for squared overlap.
572: We have examined four simple geometries;
573: the corresponding minimum-energy clusters are shown in Fig.~3.
574:
575: a) A cluster with all four ligands $\sigma$-bonded and placed at
576: equal distances of 2.0{\AA}. We take repulsion parameters appropriate
577: for nitrogen. The result is a square-planar cluster ({\it cf.} Fig.~3a),
578: consistent with the observed structures of many copper and nickel complexes.
579: With this parameterization, the square structure is quite strongly favored
580: over the tetrahedral one, by 0.31~eV or 7.0~kcal/mole. The square
581: structure remains stable under increases of the ligand-ligand
582: repulsion strengths. of up to a factor of over three. We note,
583: however, that the stability of the square structure is strongly
584: sensitive to the choice of repulsive terms. For example, we find
585: that when the van der Waals term from the OPLS-II force
586: %field\cite{Jorgensen88}
587: field~(Jorgensen et al., 1988)
588: is used instead of the MM2 form, an increase of the repulsion strengths
589: of only 20\% is enough to stabilize the tetrahedral structure.
590: We expect that input from fully quantum-mechanical total-energy
591: calculations will be necessary to precisely pin down the competition
592: between the electronic energy and the repulsion terms in determining
593: the structural energy differences.
594:
595: b) A cluster similar to that of a), but with one $\pi$-bonded ligand.
596: This leads to a substantial deviation from planarity,
597: as shown in Fig.~3b, where the $\pi$-bonded ligand is atom~4.
598: As expected from the maximum of the $\pi$-$\sigma$ potential in Fig.~1,
599: the angle between atom 4 and atoms 1--3 is greater than
600: $90^{\circ}$, about $123^{\circ}$. The angle between the
601: $\sigma$-bonded atoms 1, 2, and 3, is slightly greater than
602: the ideal value of $90^{\circ}$, because of the repulsive energy term.
603:
604: c) A cluster with with two $\sigma$-bonded nitrogen ligands 1 and 2, one
605: $\sigma$-bonded sulfur ligand 3, and a $\pi$-bonded sulfur ligand 4.
606: The distances are 2.04{\AA} for the nitrogen ligands,
607: 2.18{\AA} for the $\pi$-bonded sulfur ligand, and
608: 2.64{\AA} for the $\sigma$-bonded sulfur ligand.
609: This geometry is motivated by the observed geometry of copper
610: sites in ``blue-copper" proteins. In these proteins,
611: the nitrogen ligands and a cysteine sulfur ligand are close
612: to the copper, and it is believed that the cysteine sulfur
613: ligand is primarily $\pi$-bonded. An additional $\sigma$-bonded
614: methionine sulfur ligand is farther from the copper. The values
615: of the ligand distances here are taken from cluster calculations
616: for blue-protein
617: %%models\cite{Ryde96}.
618: models~(Ryde et al., 1996).
619: The van der Waals parameters are taken from the OPLS parameter set
620: but we do not have a reliable procedure for determining the differences
621: between the ligand-ion interactions of the nitrogen and the sulfur ligands.
622: Since the greater distance of the sulfur at 2.18{\AA} from the copper
623: will be compensated to some extent by the greater size of sulfur
624: relative to nitrogen, we simply assume that the three close ligands
625: have the same coupling strength to the central atom.
626: We have varied the ratio of the sulfur to nitrogen coupling strengths
627: by up to 30\% in both directions and found changes of only a few degrees
628: in the bond angles of the cluster. For the far ligand, we use the scaling
629: procedure described above, which leads to a coupling that has 50\%
630: of the strength of the close ligands. We have varied this coupling
631: from 0\% to 75\% of the close-ligand value, and again found bond-angle
632: changes of only a few degrees. The lowest-energy structure for
633: this cluster is indicated in Fig.~3c. The geometry is trigonal,
634: with bond angles of $125^{\circ}$ between the cysteine sulfur
635: and the nitrogens; the nitrogen-nitrogen bond angle is $91^{\circ}$.
636: By comparison, the optimal values for protein models obtained by
637: quantum-chemical
638: %calculations\cite{Ryde96}
639: calculations~(Ryde et al., 1996)
640: are $125^{\circ}$ for the cysteine sulfur-nitrogen bond angle
641: and $103^{\circ}$ for the nitrogen-nitrogen bond angle.
642: The results obtained by the present method are do of course not
643: have quantitative accuracy, but the overall structure of the
644: coordination shell is quite similar to that obtained in
645: %%Ref. (\onlinecite{Ryde96}),
646: (Ryde et al., 1996),
647: which is shown in Fig.~3d. The stability of this structure is not
648: predetermined by the bond lengths that we used as input.
649: To demonstrate this, we have considered a cluster with the same
650: bond lengths as the blue-protein fragment, with with all ligands
651: $\sigma$-bonded. In this case, the structure is planar,
652: as shown in Fig.~3e. Thus the formation of the trigonal structure
653: is directly related to the special character of the angular interactions
654: associated with $\pi$-bonding.
655:
656: \newpage
657: \section{Conclusion}
658:
659: We have seen that it is possible to extract the functional form
660: of angular forces around transition metals in biomolecules by
661: using an approximate treatment of the ligand-field Hamiltonian.
662: The approximations made here, involving primarily ionic bonding,
663: are complementary to those used in
664: %%Ref. (\onlinecite{Landis98}).
665: (Landis et al., 1998).
666: The resulting angular forms have simple trigonometric forms,
667: which are very different for $\pi$-bonding as compared to $\sigma$-bonding.
668: Ligand-field energies are represented well by these angular forces,
669: and much better than by force fields with assumed functional forms.
670: Future work should aim to include the functional forms derived here
671: in widely used biomolecular simulation packages.
672:
673: \acknowledgments
674:
675: This work was supported by the National Science Foundation under
676: Grant DMR-9971476. We appreciate informative conversations with
677: Garland Marshall and Jay Ponder.
678:
679: \newpage
680: %\bibliography{bpj}
681: %\begin{thebibliography}{10}
682: \centerline{\bf REFERENCES}
683:
684: \begin{itemize}
685: \item{}
686: %\bibitem{Allured91}
687: Allured, V.~S., C.~M. Kelly, and C.~R. Landis. 1991.
688: Shapes Empirical Force Field: New Treatment of Angular Potentials
689: and its Application to Square-Planar Transition-Metal Complexes.
690: {\it J. Am. Chem. Soc.} 113:1-12.
691: \item{}
692: %\bibitem{Carlsson98}
693: Carlsson, A.~E. 1998. Angular Forces Around Transition Metals
694: in Biomolecules. {\it Phys. Rev. Letters} 81:477-480.
695: \item{}
696: %\bibitem{Comba95}
697: Comba, P., T.~W. Hambley, and M. Str{\"o}hle. 1995.
698: The Directionality of d-Orbitals and Molecular-Mechanics
699: Calculations of Octhedral Transition-Metal Compounds.
700: {\it Helv. Chim. Acta} 78:2042-2047.
701: \item{}
702: %\bibitem{Cotton72}
703: Cotton, F.~A., and G. Wilkinson. 1972. Advanced Inorganic Chemistry.
704: Interscience, New York.
705: \item{}
706: %\bibitem{Jorgensen88}
707: Jorgensen, W.~L. and J. Tirado-Rives. 1988.
708: The OPLS Potential Functions for Proteins, Energy Minimizations
709: for Crystals of Cyclic Peptides and Crambin.
710: {\it J. Am. Chem. Soc.} 110:1657-1666.
711: %\bibitem{Kaim94}
712: \item{}
713: Kaim, W., and B. Schwederski. 1994. Bioinorganic Chemistry: Inorganic
714: Elements in the Chemistry of Life. Wiley, New York.
715: \item{}
716: %\bibitem{Landis98}
717: Landis, C.~R., T. Cleveland, and T.~K. Firman. 1998.
718: Valence Bond Concepts Applied to the Molecular Mechanics
719: Description of Molecular Shapes. 3. Applications to Transition
720: Metal Alkyls and Hydrides. {\it J. Am. Chem. Soc.} 120:2641-2649.
721: \item{}
722: %\bibitem{Ryde96}
723: Ryde, U., M.~H.~M. Olsson, K. Pierllot, and B.~O. Roos. 1996.
724: The Cupric Geometry of Blue Copper Proteins is not Strained.
725: {\it J. Mol. Biol.} 261:586-596.
726: \item{}
727: %\bibitem{Sayle95}
728: Sayle, D.~C., M.~A. Perrin, P. Nortier, and C.~R.~A. Catlow. 1995.
729: Simulation Study of Copper(I) and Copper(II) Species in ZSM-5
730: Zeolite. {\it J. Chem. Soc. Chem. Commun.} 9:945-947.
731: \item{}
732: %\bibitem{Sayle97}
733: Sayle, D.~C., {\it et~al.}, 1997.
734: Computer Modeling of the Active-Site Configurations within the NO
735: Decomposition Catalyst CU-ZSM-5. {J. Phys. Chem.} 101:3331-3337.
736: \item{}
737: %\bibitem{Sprague87}
738: Sprague, J.~T., J.~C. Tai, Y. Yuh, and N.~L. Allinger. 1987.
739: The MMP2 Calculational Method. {\it J. Comput. Chem.} 8:581-603.
740: \item{}
741: %\bibitem{Timofeeva95}
742: Timofeeva, T.~V., J.~H. Lii, and N.~L. Allinger. 1995.
743: Molecular Mechanics Explanation of the Metallocene Bent Sandwich
744: Structure. {\it J. Am. Chem Soc.} 117:7452-7459.
745: \item{}
746: %\bibitem{Wiesemann94}
747: Wiesemann, F., S. Teipel, B. Krebs, and U. Howeler. 1994.
748: Force-Field Calculations on the Structures of Transition-Metal
749: Complexes: 1. Application to Copper(II) Complexes in
750: Square-Planar Coordination. {\it Inorg. Chem.} 33:1891-1898.
751: \item{}
752: %\bibitem{Wigner59}
753: Wigner, E.~P. 1959. Group Theory and its Application to the Quantum
754: Mechanics of Atomic Spectra. Academic Press, New York.
755: \end{itemize}
756: %\end{thebibliography}
757: %\end{thebibliography}
758: \newpage
759: \begin{figure}
760: \caption{ (a) Interaction between ligands of central transition metal
761: ion. The dimensionless quantity $u$ ({\it cf.} Eqs.~(19)--(21))
762: is either $u_{\sigma,\sigma}$ (solid line),
763: $u_{\pi,\sigma}$ (dashed line), or $u_{\pi,\pi}$ (dotted line).
764: The quantity $-u$ is plotted so that negative values will
765: correspond to lower energies.
766: (b) Empirical force field from (Comba et al., 1995).}
767: \end{figure}
768:
769: \begin{figure}
770: \caption{Energies obtained by semiclassical force field vs.
771: exact energies for small model clusters. Energies given in
772: units of $h_{\rm max}$, maximal coupling strength between ligands and
773: transition metal. (a)~One $\pi$-bonded ligand and three
774: $\sigma$-bonded ligands. (b)~Four $\pi$-bonded ligands.}
775: \end{figure}
776:
777: \begin{figure}
778: \caption{Lowest-energy structures obtained in relaxations of
779: five-atom cluster using angular forces. (a)~Four equivalent
780: $\sigma$-bonded neighbors. (b)~Three $\sigma$-bonded neighbors
781: and one $\pi$-bonded neighbor~4, all at the same distance.
782: (c)~Two $\sigma$-bonded nitrogen neighbors~1 and 2, a near
783: $\pi$-bonded sulfur neighbor~3, and a far $\sigma$-bonded
784: neighbor~4. (d)~Structure for blue protein model obtained by
785: quantum-chemical calculations (Ryde et al., 1996). (e)~Three near
786: $\sigma$-bonded neighbors and a far $\sigma$-bonded neighbor.}
787: \end{figure}
788:
789: \end{document}
790: