Spectrochimica Acta Part A 54 (1998) 1141 – 1156

ENDOR studies of substituted chlorophyll cation radicals Hanno Ka¨ß a, Wolfgang Lubitz a,*, Gerhard Hartwig b, Hugo Scheer b, Dror Noy c, Avigdor Scherz c a

Max – Volmer-Institut fu¨r Biophysikalische Chemie und Biochemie, Technische Uni6ersita¨t Berlin, Straße des 17. Juni 135, D-10623 Berlin, Germany b Botanisches Institut der Uni6ersita¨t Mu¨nchen, Menzingerstr. 67, 80638 Mu¨nchen, Germany c Weizmann Institute of Science, Biochemistry Department, Reho6ot, Israel

Abstract Cation radicals of substituted chlorophyll (Chl) a molecules were characterized by EPR and ENDOR spectroscopy. The effect of substitutions of the vinyl group at position 3 and the carbomethoxy group at 132 as well as the replacement of the central magnesium atom by zinc were investigated. All major hyperfine coupling constants could be determined and assigned to specific molecular positions. Comparison of the experimental results with data from related bacteriochlorophyll (BChl) a cation radicals shows that introduction of the same substitution in Chl a and BChl a causes a similar change of the electron spin density distribution in both radicals. Semi-empirical calculations of the RHF-INDO/SP type were performed on all systems and yielded the isotropic and the anisotropic part of the proton hyperfine coupling tensors. The experimental and theoretical results are in good agreement. The data obtained for the Chl a species investigated are discussed with regard to the structure of the primary donor cation radicals in plant photosystem I and II. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Photosynthesis; EPR; ENDOR; MO theory; Pigment radical cations; Electronic structure; Spin destiny distribution; Hyperfine couplings; Substituted chlorophylls; Zn-(bacterio)chlorophyll

1. Introduction Chlorophyll (Chl) and bacteriochlorophyll (BChl) molecules are important cofactors of photosynthetic proteins [1]. In the antenna complexes they take part in light harvesting and energy transfer processes; whereas, in the reaction centers (RCs) of plants and bacteria they are involved in charge separation by acting both as electron * Corresponding author. Fax: + 49 30 31421122; e-mail: [email protected]

donors and acceptors. In the RC of purple bacteria-whose structure is well known [2], electron transfer starts with a donation of an electron from a photoexcited BChl-dimer to an electron transport chain involving monomeric BChl, bacteriopheophytin (BPheo) and two quinones coupled to an Fe2 + [3]. In plants and cyanobacteria two photosystems—PS I and PS II—exist in which two separate light reactions occur. The structure of PS II, for which no X-ray crystallographic structure exists to date, is assumed to be similar to that of purple bacteria [4], in particular with

1386-1425/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S1386-1425(98)00065-1

1142

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

respect to the acceptor side. The structure of the primary donor (P680), a Chl monomer or dimer, is not yet understood [5 – 7]. This species which has an unusually high redox potential of \ +1.1 V is of central importance in the light-induced water splitting process that occurs in PS II [8,9]. For PS I a low resolution X-ray crystallographic structure is available [10]. Here, the primary donor P700 is a Chl a dimer; monomeric Chl acts as first electron acceptor [11,12]. In all bacterial RCs and plant photosystems radical ions are formed directly in the charge separation process. EPR is the method of choice to detect and characterize these species in vivo [13]. Using this technique Norris et al. proposed + ’ + ’ that P700 in PS I and P865 in bacterial RCs consist of dimeric (B)Chl species (‘special pair’ hypothesis) [14]. The chlorophyllous acceptors have also been identified by EPR (for a review see Ref. [13]). In conventional EPR spectra of protein-bound radicals the electron nuclear hyperfine couplings (hfc’s) are usually not resolved. This information is, however, required to determine the distribution of the unpaired electron in the valence orbitals. The hfc’s can be obtained by using electron nuclear double resonance (ENDOR) spectroscopy which provides a much higher resolution than EPR [15,16]. For the detection of dipolar and quadrupolar couplings of nuclei with small magnetic moments and small hfc’s electron spin echo envelope modulation (ESEEM) can conveniently be used [17]. ENDOR was instrumental in showing that P + ’ in purple bacteria is a supermolecule of two BChl molecules with a wave function extending over both BChl halves [18]. This was supported by the molecular orbital (MO) calculations of Plato et al. [18,19] based on the X-ray crystal structure of the RC. The asymmetry of the electron spin density distribution in the dimer was shown to strongly depend on the protein surrounding [20,21]. + ’ In PS I the dimeric nature of P700 was, however, questioned [22] on the basis of results achieved by ab initio CI MO calculations [23]. According to those results, the electronic doublet ground state, D0, and the first exited state, D1, of Chl a + ’ are almost degenerate in contrast to the

case of BChl a. Therefore, the two states could mix resulting in a hybrid wave function that is strongly influenced by geometrical details of the pigment structure and its immediate environment [22]. Based on this view a new interpretation of + ’ was given, and it the experimental data of P700 was regarded as a monomeric Chl a cation radical [23]. Recent nitrogen ENDOR/ESEEM data were also interpreted on the basis of this model [24]. On the other hand, there is evidence from similar work of other groups [25,26] and in particular from recent 1H ENDOR [27,28] and 14N ESEEM + ’ is [29] performed on PS I single crystals that P700 a Chl dimer with a strongly asymmetric spin density distribution over the two halves. The distinction between a disturbed monomer and a strongly asymmetric dimer is not easy to make. Clearly, for a final interpretation of the hyperfine data in vivo it is necessary to have a good model system in vitro. Consequently, the monomeric BChl a cation radical was studied in great detail in organic solvents (reviewed in Ref. [13]). The influence of different substituents on the electronic structure of BChl a cation radicals has also been investigated in detail and was very helpful-together with MO calculations of the spin density distributions [30]—to understand the factors determining structure and function of BChl pigments in the RC. Although an early solid state ENDOR study of several Chl a cation radical exists [31], detailed information about the effect of various substituents on the spin density distribution is still lacking for this system. Such an investigation should shed light on the question of whether the Chl a + ’ can indeed be described by a hybrid wave function [5,22]. Such a description would imply relatively large changes of the spin density distribution upon substitution, since the two mixed orbitals of the doublet state (D0 and D1) are supposed to have very different orbital coefficients (see Refs. [22,23]). A basic knowledge of the way how the electronic structure of chlorophyll-type cation radicals is changed by substituents and by the surrounding medium is not only important for a better understanding of the hyperfine data of + ’ P700 in PS I but also for the various oxidized chlorophyll species occurring in PS II including + ’ the primary donor, P680 [5–7,32].

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

1143

Fig. 1. Structures of investigated chlorophylls (a) and bacteriochlorophylls (b). The IUPAC numbering of molecular positions is used, for substituents see Table 1. Note: a-H-atoms are directly attached to the conjugated p-system, b (g)-H-atoms are one (two) bonds away from it.

In BChl and Chl species the central metal also plays a role, e.g. for the interaction of these species with their environment via axial ligands, for the structure (planarity) of the macrocyclic ring, for the redox behavior and other spectroscopic and functional properties. Among the divalent metals, zinc is believed to be quite similar to the magnesium and can replace it in many systems. Indeed, in a recent study, it has been reported that the photosynthetic acidophilic bacterium Acidiphilum rubrum contains Zn – BChl a as pigment in its antenna system and in the RC where it seems to fully replace the Mg – BChl a in its function [33]. As a model for the cation radical of the respective primary donor — probably a Zn–BChl a dimer — we have therefore studied in this paper the hyperfine structure of Zn – BChl a + ’ by EPR and ENDOR techniques and compared it with that of Zn – Chl a + ’. The results show that these molecules have electronic structures very similar to those of the respective Mg-containing pigments.

2. Materials and methods Chl a (1) samples for EPR and ENDOR experiments were prepared from material purchased from Sigma and used without further purification. In addition, some samples were generated from Chl a that was obtained from spinach and purified by HPLC in order to remove remaining lipids [34]. The preparation of methyl-chlorophyllide a (2) is outlined in Ref. [35], that of pyro-Chl a (3); [3-acetyl]-Chl a (4); [3-acetyl]-pyro-Chl a (5); ZnChl a (6) and Zn–BChl a (8) is given in several references cited in a recent review [36]. BChl a (7) was prepared as described previously [30]. All pigments were chromatographically purified and characterized by their optical absorption spectra. The structures of the investigated pigments and abbreviations used in the text are given in Fig. 1(Table 1). All samples for EPR/ENDOR experiments were prepared in dim green light. The Chl-type pigments were dissolved in a mixture of CH2Cl2/

1144

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

Table 1 Investigated chlorophylls and bacteriochlorophylls, for molecular structures (see Fig. 1) Name

Abbreviation

Label

Me

R1

R2

Chlorophyll a Methyl-chlorophyllide a a Pyro-Chl a [3-acetyl]-Chl a [3-acetyl]-pyro-Chl a Zinc-chlorophyll a Bacteriochlorophyll a Zinc-bacteriochlorophyll a

Chl a Chlide p-Chl a 3ac-Chl a 3ac-p-Chl a Zn-Chl a BCh l a Zn-BChl a

1 2 3 4 5 6 7 8

Mg Mg Mg Mg Mg Zn Mg Zn

CHCH2 CHCH2 CHCH2 COCH3 COCH3 CHCH2 COCH3 COCH3

COOCH3 COOCH3 H COOCH3 H COOCH3 COOCH3 COOCH3

a

This molecule contains a methyl group instead of the phytyl side-chain.

THF (tetrahydrofurane) 10:1 (v/v) to a concentration of about 5 × 10 − 4 M. The solution was filled in a quartz capillary and bubbled with argon to remove dissolved oxygen. Cation radicals were generated chemically [25] by adding a 10-fold excess of iodine dissolved in CH2Cl2 together with a stoichiometric equivalent of AgClO4 dissolved in acetonitrile (CH3CN). During oxidation the color of the solution changed from green to brownish. The capillary was then sealed and immersed in liquid nitrogen in order to freeze-trap the final product and, thereby, prevent degradation of the radicals. The liquid solution ENDOR samples were prepared on a high-vacuum line [30]. For the BChl-type pigments a mixture of CH2Cl2/CH3OH 6:1 (v/v) was used as solvent. All solvents were of spectroscopic grade (Merck, Uvasol). The spectra were taken on a Bruker ESP 300 E EPR spectrometer with home-built ENDOR accessories [30,37]. Using either a nitrogen (Bruker ER 4111 VT) or a helium (Oxford A910) flow system temperatures between 2 and 300 K could be achieved. For the simulation of ENDOR spectra a self-written program was used that was described in Ref. [30]. MO calculations were performed using the RHF-INDO/SP program of Plato et al. [19,38]. It is based on the intermediate neglect of differential overlap (INDO) scheme [38] in the framework of a restricted Hartree-Fock (RHF) approach with subsequent treatment of spin polarization (SP) effects. In the present parametrization it can handle hydrogen, all second row elements and, in

addition, magnesium. In extension to Ref. [30] the complete hfc tensors can be calculated with the present program [40].

3. Results and discussion

3.1. EPR experiments EPR on the substituted Chl-type cation radicals in liquid solution at 200 K gave inhomogeneously broadened Gaussian lines centered at g= 2.0025, the linewidth DBpp was found to be in the range of 0.8 90.05 mT. No nuclear hyperfine structure could be resolved. In contrast to the situation found for BChl-type cation radicals [30] and also for Zn–BChl a + ’ a significantly higher microwave power (] 50 mW) was necessary to approach saturating conditions for all Chl-type species in liquid solution. In the solvent mixture used, the samples were liquid above 180 K, below 130 K they formed a polycrystalline solid powder. In between these limiting temperatures the anisotropic part of the hfc’s is partly averaged out (see ENDOR spectra below). Such an averaging process was also found for BChl-type cations and attributed to remaining molecular motions [41]. According to the work in Ref. [41], the Chl samples investigated here might be regarded as a soft glass between 130 and 180 K. In the following, this is called the glassy state of the samples. The described (partial) averaging of the dipolar part of certain hfc’s can also explain the temperature dependence of the EPR linewidth. For example,

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

in case of Chl a + ’ the linewidth was measured to DBpp =0.79 mT at 240 K in liquid solution, it increased to 0.89 mT at 140 K in the glassy state and finally reached 0.98 mT at 10 K in the powder. Effects of similar magnitude have also been found for the substituted Chl a radical cations. A range of different values was reported earlier for the EPR linewidth of Chl a + ’ (see Ref. [13] and references therein). This might be explained by a variation in preparation techniques. Depending on the solvent, the Chl a molecules tend to stack together and form larger aggregates. Delocalization of the unpaired electron within the aggregates can then lead to a pronounced narrowing of the EPR line [14,42]. In such a situation, monomers, dimers and higher aggregates with different linewidths might be superimposed and lead to varying EPR line shapes and widths. Comparison of published data with our experimental linewidths shows that the samples investigated in this work in the glassy and polycrystalline state essentially consisted of monomeric Chl-type cation radicals. At higher temperatures in the liquid state the strong affinity of the pigments could, however, lead to pigmentpigment interactions. This is known to decrease the spin relaxation times, make mw power saturation more difficult and prevent the detection of cw ENDOR.

1145

the anisotropic part of the hfc tensors determines the shape of the ENDOR lines. In particular, freely rotating methyl groups with their almost axially symmetric proton hfc tensors show a characteristic shape which allows a direct evaluation of the AÞ and A tensor components (see Fig. 3-2 in Ref. [43]). All methyl proton lines were assigned in this way. In case of a more general tensor symmetry or of overlapping ENDOR lines spectral simulations are necessary. Fig. 2 shows a representative set of 1H-ENDOR spectra for Zn–Chl a + ’ (6 + ’) as an example. The temperature was varied from 110 to

3.2. ENDOR experiments on samples in liquid and glassy solution Isotropic hfcs of BChl-type radicals can be obtained from ENDOR spectra in liquid solution using the ENDOR resonance condition [15,16,43] 69 ENDOR = 6n 9 a/2

(1) 9 ENDOR

with the ENDOR transition frequencies, V , the nuclear Larmor frequency, 6n, and the isotropic hfc a. However, most of the substituted Chl a cation radicals yielded ENDOR spectra only in frozen solution (powder) below 130 K and in the glassy state between 130 and 180 K. Partial averaging of hfc’s still occurs in the glassy state. Because of that, no significant anisotropic broadening was detected for the ENDOR lines of the methyl and b-protons [30,41]. In frozen solution

Fig. 2. 1H-ENDOR spectra of Zn-Chl a + ’ (6 + ’) in frozen polycrystalline solution (110, 130 K) and in the glassy state. Sample concentration 3 ×10 − 4 M, solvent CH2Cl2/THF 10:1 (v/v). Corresponding low- and high-frequency lines are numbered; 1,1% and 2,2% indicate the b-protons at positions 17 and 18; 3,3%, 4,4% and 5,5% the CH3 groups at positions 12, 2 and 7, respectively; see Fig. 1. Experimental conditions: frequency modulation 12.5 kHz, deviation 9100 kHz, time constant 320 ms, scan time 8 × 320 s, microwave power 20 mW in glassy state, 10 mW in frozen solution, RF power 200 W.

1146

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

165 K. Four pairs of narrow lines were detected in the soft glass at 165 K, denoted 1,1% (coinciding with 2,2%), 3,3%, 4,4% and 5,5%. On lowering the temperature to 150 K, the lines broaden slightly. In addition, a line centered at the 1H Larmor frequency 61H of 14.65 MHz appears at T 5 150 K. At 130 K the remaining molecular motions, which are still present in the glassy state, are frozen out. Now line pair 3,3% shows the typical anisotropic broadening of an axially symmetric hfc tensor [16,43]. Thus, it is assigned to the protons of a single methyl group [30]. Line pair 4,4% (5,5%) obviously contains contributions of two overlapping methyl proton tensors. Besides the matrix proton line at 14.65 MHz an additional line pair 6,6% shows up in the central region of the spectrum, which is due to purely dipolar coupled protons. At 110 K a pair of very broad lines (see amplified part (×4) of the spectrum) is seen at the position where the intense and narrow lines 1,1% (2,2%) were detected in the glassy state. These lines are therefore assigned to b-protons. They have non-axial hfc tensors of larger anisotropy than the methyl proton hfc tensors [30] and are subject to stronger line broadening in the solid state. It has to be noted that these lines 1,1% (2,2%) could not be resolved in the experiment performed at 130 K. They are most likely hidden under the broad wings of line pair 3,3%. At lower temperatures (110 K) the shape of these lines (3,3%) is slightly changed and some structure due to the outer lines 1,1% (2,2%) becomes visible. The assignments of 1H-hfc’s of the substituted Chl a cation radicals rely on the comparison of their ENDOR spectra in the glassy state (Fig. 3) and in frozen solution (Fig. 5), which allows to separate methyl proton lines from b-proton lines and other resonances. Furthermore, it is based on the fully analyzed ENDOR spectrum of Chl a + ’ [13,30,31,44]. Experimental hfc’s are given in Table 2. Unless otherwise stated all hfc’s were evaluated from ENDOR spectra in the glassy state. In these spectra the line near the Larmor frequency is due to small-mainly dipolar-proton hfc’s. Its amplitude depends sensitively on freezing, the exact temperature, mw and rf field strengths and rf modulation depth and thus varies in intensity for the different radicals 1 + ’ – 6 + ’ (Figs. 2 and 3).

Fig. 3. 1H-ENDOR spectra of Chl a + ’ (1 + ’), pyro-Chl a + ’ (3 + ’), [3-acetyl]-Chl a + ’ (4 + ’), [3-acetyl]-pyro-Chl a + ’ (5 + ’) and methyl-chlorophyllide a + ’ (2 + ’) in the glassy state at 140 K. Sample concentration 3 ×10 − 4 M, solvent CH2Cl2/THF 10:1 (v/v). (Corresponding low- and high-frequency lines are numbered as in Fig. 2.) For experimental conditions see Fig. 2 except for RF power 250 W.

3.2.1. Assignments for Chl a + ’ The spectrum of Chl a + ’ (1 + ’) in the glassy state is shown in Fig. 3. The line pair labeled 1,1% and 2,2% is due to the two b-protons at molecular positions 17 and 18 in ring D. Their individual lines overlap almost completely, the respective linewidth of 400 kHz is clearly larger than that of line pair 3,3% ( 300kHz). The latter is assigned to the protons of the methyl group at molecular position 12 in ring C. Pair 4,4% (5,5%) is mainly due to the two methyl groups at positions 2 and 7 in rings A and B. According to earlier deuteration experiments [31] the 1H-hfc’s of the CH2 group at position 8 are additionally assigned to this line

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

1147

Table 2 Experimental 1H-hfc’s [MHz] of substituted Chl-type cation radicalsa Line

Chl a+’ 1+’b

1

Chlide+’ 2+’c 10.29

9.96 2 3

p-Chl a+’ 3+’

8.58 7.11

6.83

3.01e

3.01e

2.98e

5 0.80h

Zn-Chl a+’ 6+’

9.00

10.83

Assignment 17

9.24

7.08

0.67

3ac-p-Chl a+’5+’

9.91 9.93

4

6

3ac-Chl a+’ 4+’d

18

6.93

6.48

7.32

121

4.68

4.38

3.29e

21

2.94e 1.24f

2.85e

3.87

71 5g

0.80h

1.02h 0.48f

20i

a Isotropic values evaluated from cw-ENDOR on samples in the glassy state at about 140 K; accuracy 9 25 kHz. Solvent CH2Cl2/THF 10:1 (v/v), for structures and abbreviations see Fig. 1and Table 1. b Values in liquid solution: pos. (= position) 17/18: 10.17 MHz; pos. 121: 7.12 MHz; pos. 71/21: 2.97 MHz; 132: 1.97; pos. 20: 0.57i MHz (spectrum not shown). c Values in frozen solution using pure THF: pos. 17/18: 10.08 MHz; pos. 121: 7.28 MHz; pos. 71/21: 2.99 MHz, pos. 5/20: 0.73 MHz. d Values in liquid solution using CH2Cl2/CH3OH 6:1 (v/v): pos. 17/18: 9.94 MHz: pos. 121: 6.86 MHz, pos. 21: 4.64 MHz; pos. 71: 2.83 MHz; pos. 132: 1.73 MHz; pos. 20: 0.54 MHz. e This hfc is additionally assigned to the protons at the molecular position 81 ([31,30]). RHF-INDO/SP calculations show that the hfc’s of the a-protons at positions 5 and 10 are of the same magnitude. f From experiments in the glassy state, using 200 kHz FM g An additional assignment to the H-atom at position 132 cannot be excluded, since no signs of hfc’s could be determined. h Estimated accuracy 9100 kHz, evaluated from the frozen solution spectrum (Fig. 5). i Most likely the other protons at positions 31, 32, 82, 171 and 181 are to be expected in this spectral region.

pair. From a comparison with the hfc’s found for BChl a + ’ and calculated by the RHF-INDO/SP method (Table 4) it is reasonable to assume that the methin protons at positions 5 and 10 also contribute to these lines. The remaining small hfc’s of the protons at positions 31, 32, 82, 171, 181 and 20 are all assigned to line pair 6,6%. ENDOR and general TRIPLE resonance on Chl a + ’ in liquid solution have shown that the proton at position 132 yields a negative hfc of about 2 MHz [13,30,44]. Most likely the glassy state spectra do not show the corresponding lines because of the quite large anisotropy expected for this hfc that is no longer averaged out. All 1H-hfc data for Chl a + ’ found earlier have been verified in this work, using the highly purified Chl a starting material. Furthermore, the samples used here yielded better resolved liquid solution ENDOR spectra (not shown) than those made earlier [30] from commercially available material. This is probably due to the fact that mate-

rial from a standard preparation contains remaining lipids which may cause aggregation effects. The ENDOR spectra of the substituted chlorophyll cation radicals 1 + ’ –6 + ’ in the glassy state are shown in Figs. 2 and 3. The respective 1 H hfc’s and their assignments are given in Table 2. In the following the effect of the substitutions on the electron spin density distributions are discussed and compared with the respective BChl’s with related substitution patterns.

3.2.2. Substitution at position 3 Substitution of the vinyl group in 1 by acetyl leads to 3ac-Chl (4). In the ENDOR spectrum of 4 + ’ (Fig. 3) line pairs 1,1% and 2,2% are resolved in contrast to 1 + ’. The slight difference of the hfc’s of the two b-protons at positions 17 and 18 can be explained by somewhat different conformations of ring D in 1 + ’ and 4 + ’. The hfc of the methyl protons at position 12 is slightly reduced from 7.08 to 6.93 MHz, when

1148

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

going from 1 + ’ to 4 + ’. A similar reduction of the respective methyl hfc values (6.83 – 6.48 MHz) is detected for the pyro-compounds 3 + ’ and 5 + ’. Obviously, in Chl-type cation radicals, a change of substitution in ring A is also affecting the methyl hfc in ring C on the opposite site of the p-system1. Exactly the same effects have been found for the analogous substitutions in the BChl’s (see Ref. [30]). Line pair 4,4%, yielding a hfc of 4.68 MHz, shifts when going from the 3-acetyl compound 4 + ’ to the 3-vinyl compound 1 + ’ and is in the latter superimposed to line pair 5,5%. The strong intensity in the frozen solution ENDOR spectrum (Fig. 5) shows that line pair 4,4% is due to a methyl proton hfc. In analogy to the BChl-type cation radicals [30] it is assigned to the methyl group at position 2. Comparison of the corresponding pyro-compounds 5 + ’ and 3 + ’ shows the same situtation (Table 2). The ENDOR spectra of 4 + ’ and 5 + ’ in frozen solution (Fig. 5) show that line pair 5,5% has to be assigned to the third methyl group at position 7 in ring B. In case of 1 + ’ and 3 + ’ line pair 5,5% is overlapping with 4,4%. For the vinyl compounds 1 + ’ and 3 + ’ the methyl hfc at position 7 is slightly larger than for 4 + ’ and 5 + ’ carrying the acetyl group This effect has no counterpart in BChl a-type cation radicals which have no large CH3 hfc at position 7 due to hydrogenation of ring B. Concluding, both in Chl a- and BChl a-type cation radicals, the substitution of the vinyl group at position 3 by an acetyl group causes a pronounced increase of the methyl proton hfc at position 2 in ring A of about 1.5 MHz in Chl’s and 2.3 MHz in BChl’s, respectively. Furthermore, this substitution causes a decrease of the 1 H-hfc’s for BChl-type cations are given in Table 2 of Ref. [30]. Unfortunately, the published data set contains three typing errors, two of them being of interest for the present discussion: In case of 132-hydroxy-BChl a + ’ the hfc values for the protons at carbon position 121 and 18 have to be exchanged. Thus the correct values are + 8.90 MHz at pos. 18 and + 9.21 MHz at pos. 121 (these two typing errors are obvious, considering Fig. 9 of Ref. [30]). The third typing error occurred in case of [3-vinyl]-BChl a + ’. Here the 1H-hfc at carbon position 171 is −0.56 MHz and not − 0.80 MHz.

methyl proton hfc at position 12 in ring C of about 0.2 MHz for Chl’s and 0.6 MHz for BChl’s (averaged values). A quantitative evaluation of the increase and decrease of the methyl proton hfc’s must account for the fact that the conjugated system of BChl a is different from that of Chl a (conjugated ring B). As a consequence, the ‘averaged’ methyl proton hfc’s as well as their absolute changes induced by a specific substitution are larger in the BChl-type cation radicals.

3.2.3. Substitution at position 13 2 A comparison of the hfc’s of 1 + ’ and 4 + ’ with those of the respective pyro-compounds 3 + ’ and 5 + ’ (Table 2) reveals the effect of a substitution at position 132. In analogy to the related BChl a compounds a distinct decrease of about 0.3 MHz (average) is found for the methyl group hfc at position 12 when going from 1 + ’ to 3 + ’ or from 4 + ’ to 5 + ’. In case of the BChl-type cation radicals the corresponding decrease is again larger, i.e. about 0.6 MHz [30]. The methyl proton hfc at position 2 is also influenced by substitution at position 132. The effect is clearly observed for the 3-acetyl systems. Comparison of 4 + ’ and 5 + ’ or 1 + ’ and 3 + ’ reveals a decrease of 0.3 MHz for this hfc. Again, this agrees with the situation in the BChl a compounds. Here, the effect is more pronounced for the Chl-type cation radicals. 3.2.4. Combined substitution at positions 3 and 13 2 A comparison of the experimental hfc’s of 5 + ’ +’ (3 ) and 1 + ’ (4 + ’) shows that the effects induced by combined substitutions at positions 3 and 132 are additive. The results resemble closely those obtained for the BChl-type cations radicals [30]. The methyl group at position 12 in 1 + ’ yields a hfc of 7.08 MHz. In the 3-acetyl system 4 + ’ it is reduced to 6.93 MHz and further reduced to 6.48 MHz in the pyro compound 5 + ’. Exactly the same situation is found for the BChl a compounds. In [3-vinyl]-BChl a + ’ the respective hfc is 10.26 MHz. Going to the 3-acetyl system BChl a + ’ it reduces to 9.60 MHz, in pyroBChl a + ’ an even lower value of 9.02 MHz is measured.

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

The respective 3-acetyl and the 3-vinyl-pyro systems once again show close similarities between Chl- and BChl-type cation radicals. The total change of the methyl group hfc at position 12 for 4 + ’ and 3 + is − 0.10 MHz. Comparing BChl a + ’ and [3-vinyl]-pyro-BChl a + ’ it is + 0.04 MHz. In both cases these are quite small variations as compared with the changes observed for the 3-vinyl to the 3-acetyl-pyro cation radicals in which the differences amount to −0.60 MHz (Chl’s) and −1.24 MHz (BChls), see Table 2.

3.2.5. Substitution of the phytyl side chain The cation of methyl-chlorophyllide a (2 + ’) was investigated in order to elucidate possible effects induced by the phytyl side chain on the electronic structure of 1 + ’. In 2 + ’ the hfc’s of the two b-protons in ring D vary, depending on the solvent used. The sample prepared in the solvent mixture CH2Cl2/THF (10:1) shows two line pairs 1,1% and 2,2% (Fig. 3); whereas, in pure THF only one unresolved line pair is detected (not shown). An explanation could lie in a different geometry of ring D. Most likely the removal of the large phytyl chain increases the flexibility of the molecular skeleton of 2 + ’ as compared with Chl a + ’. Therefore, slightly different conformations of ring D can be more easily adopted in the case of 2 + ’, depending on the solvent cage. In contrast to that finding the hfc’s of the methyl protons essentially remained constant in both solvents. Furthermore, no significant differences of these hfc’s were found for 1 + ’ and 2 + ’ (see Table 2). The large dependence of the b-proton hfc’s on minor conformational changes of ring D has also been found for the substituted BChl-type cation radicals [30]. In the latter case, a pronounced steric effect of the side group at position 132 on the hfc’s of the b-protons at positions 17 and 18 was detected. In case of 2 + ’ a similar effect was induced by replacing the phytyl chain at position 17 by methyl instead of introducing a more voluminous side group at position 132, as it had been done in the BChl’s [30]. 3.2.6. Substitution of the central metal atom The exchange of the central Mg atom in Chl a + ’ by Zn leads to an increase of the b-pro-

1149

ton hfc’s from 9.96 to 10.83 MHz. In addition, the methyl proton hfc’s become slightly larger. Since this effect is different for the methyl groups at rings A and B the line pairs 4,4% and 5,5% — which overlap in 1 + ’ —are well resolved in case of 6 + ’ (Fig. 2). A comparison with the corresponding BChl-type cation radicals helps in assigning line pairs 4,4% and 5,5%. Liquid solution ENDOR spectra of BChl a + ’ +’ (7 ) and Zn–BChl a + ’ (8 + ’) are shown in Fig. 4, the elucidated 1H-hfc’s are given in Table 3. The assignments were done as in Ref. [30]. In general the detected differences between the two radicals are small. This is not unexpected since the ionic radii and electronegativities of Mg2 + and Zn2 + are not very different. A significant increase of the b-proton hfc’s is observed for molecular positions 8 and 7 in ring B together with a smaller increase for positions 17 and 18 in ring D when comparing 7 + ’ and 8 + ’. These hfc’s are, however, quite sensitive to even slight geometrical

Fig. 4. 1H-ENDOR spectra of BChl a + ’ (7 + ’) and ZnBChl a + ’ (8 + ’) in liquid solution at 235 K. Sample concentration 3×10 − 4 M, solvent CH2Cl2/CH3OH 6:1 (v/v). Corresponding low- and high-frequency lines of the major hfc’s are numbered. 1,1% to 4,4% indicate the b-protons at positions 8, 7, 17 and 18; 5,5% and 6,6% the CH3 groups at positions 12 and 2. The smaller hfc’s are not assigned here (see [13]). Experimental conditions: FM (12.5 kHz) deviation 9 100 kHz, time constant 640 ms, scan time 4 × 800 s, microwave power 12.6 mW, RF power 200 W.

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

1150

Table 3 1 H-hfc’s [MHz]a of BChl a+’ and Zn- BChl a+’ lineb

BChl a+’ 6+’

Zn-BChl a+’ 8+’

assignmentc

1 2 3 4 5 6 7 8 9 10 11

16.48 13.59 13.06 11.63 9.60 4.84 2.35 1.66 1.29 0.59 0.15

17.40 14.46 13.48 11.88 9.36 4.82 2.46 1.51 1.51 0.59 0.30

8 7 17 18 121 21 5 132 10, 20 171, 181 71, 81

are due to the large b-proton hfc’s are no longer visible. The reason is that these couplings contain significant anisotropic contributions from non-axial hf tensors. This causes line broadening and results in weak and poorly resolved spectral features. Furthermore, a significant part of these lines is hidden in the wings of the intense line pair 3,3%. In view of the possible occurrence of temperature-dependent changes of ENDOR linewidths, which might impede the detection of specific lines at certain temperatures, frozen solution spectra were taken between 130 and 10 K. However,

a

Obtained from cw-ENDOR spectra in liquid solution, accuracy 925 kHz. b In Fig. 4 only line pairs 1.1% to 6,6% are labeled that belong to the larger hfc’s. c For structures and numbering scheme see Fig. 1.

changes of the hydrogenated rings [30]. Obviously, substitution of Mg by Zn affects the conformation of ring B somewhat more than that of ring D. The changes of the other resolved hfc’s are all less than 5%. The structure of ring B is different for 6 + ’ and +’ 8 because of the hydrogenated, D 7,8-bond in the BChl’s. Thus the methyl group at position 7 in 6 + ’ has no direct counterpart in 8 + ’. Nevertheless, in analogy to the situation of BChl’s it appears reasonable to conclude that the effects induced by exchange of Mg to Zn in Chl’s should be most pronounced for protons close to ring B. Thus, in 6 + ’ line pair 5,5% is assigned to the methyl group at position 7 and 4,4% to that at position 2. Furthermore the analogy between the Zn-substituted Chl-type and BChl-type cation radicals is reflected in the increase of the b-proton hfc’s of ring D in 6 + ’ and 8 + ’ as compared with 1 + ’ and 7 + ’, respectively (see Tables 2 and 3).

3.3. ENDOR experiments on samples in the solid state. ENDOR experiments in frozen solution below 130 K were performed on all investigated cation radicals. A set of spectra from Chl-type systems is shown in Fig. 5. The line pairs 1,1% and 2,2% that

Fig. 5. 1H-ENDOR spectra of Chl a + ’ (1 + ’), pyro-Chl a + ’ (3 + ’), [3-acetyl]-Chl a + ’ (4 + ’), [3-acetyl]-pyro-Chl a + ’ (5 + ’) and methyl-chlorophyllide a + ’ (2 + ’) in frozen solution at about 120 K (for 6 + ’ see Fig. 2). Sample concentrations and solvents as in Fig. 3. Experimental conditions: FM deviation 9100 kHz, time constant 640 ms, scan time 4 ×800 s, microwave power 10 mW, RF power 200 W.

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

1151

Table 4 H-hfc’s [MHz] of substituted Chl a-type cation radicals calculated with the RHF-INDO/SP methoda

1

Positionb

Chl a+’ 1+’

p-Chl a+’ 3+’

3ac-Chl a+’ 4+’

3ac-p-Chl a+’ 5+’

17 18 121 21 71 132 5 10 20 31 32 81 171 181

+11.08 +10.28 +5.78 +5.00 +4.18 −1.50 +4.40 +5.51 +0.83 +1.67 −0.36 +1.74 −0.60 −0.27

+10.73 +10.29 +4.65 +5.16 +4.16 −4.53 +4.57 +5.46 +0.80 +1.67 −0.36 +1.81 −0.59 −0.26

+11.08 +10.73 +5.79 +6.66 +4.06 −1.57 +4.72 +5.41 +0.52 — −0.05 +1.73 −0.59 −0.27

+10.79 +10.74 +4.67 +6.79 +3.98 −4.66 +4.89 +5.34 +0.51 — −0.05 +1.82 −0.58 −0.26

a

The hfc’s are obtained from calculated s-spin densities using QH =1420 MHz (see [38]). In addition to the results given in ref. [30] the calculated values presented here are averaged with respect to the free rotation of the methyl groups. b For structures and abbreviations see Fig. 1. The phytyl side chain was replaced by a methyl group in all calculations (see text).

contributions of line pair 1,1% (2,2%) could only be detected in the frozen solution ENDOR experiment on 6 + ’. Here the b-proton hfc’s are somewhat larger than in the other Chl-type cation radicals (see Fig. 2, bottom). Local inhomogeneities in the polycrystalline frozen solvent may be the reason for the severe broadening of the methyl proton line pairs 3,3%, 4,4% and 5,5%, erasing the typical patterns. This effect has been observed earlier by comparison of ENDOR spectra from Chl a + ’ in frozen polycrystalline + ’ solvents and P700 in a frozen protein matrix [25,45]. The experiments in the glassy state (Fig. 3) revealed also lines in the central part of the spectrum (see e.g. 6,6% for 1 + ’). In frozen solution a single line pair 6,6% is also emerging for all Chl-type cation radicals (for 5 + ’ only shoulders on the matrix line are observed). This line pair will probably contain a whole set of small hfc’s that can arise from the protons at positions 31, 32, 82, 171, 181 and 20. Furthermore, coupled protons from the surrounding solvent matrix are expected in this region. Because of the strong overlap of all these individual lines and due to anisotropic line broadening a detailed evaluation of these hfc’s is not possible by ENDOR.

3.4. Calculation of hyperfine coupling constants Using the RHF-INDO/SP program 1H-hfc’s were calculated for all Chl-type cation radicals investigated in this work. An exception is the Zn-substituted compound 6 + ’ for which the parametrization for zinc is lacking. The basic structure used in the calculations was that of chlorophyllide a since the removal of the phytyl side chain-which is six bonds away from the conjugated system-has no effect on the spin densities in the p-system. Essentially, the molecular skeleton was built up using standard bond lengths and angles [39] except for some modifications of the exocyclic ring E [30]. Ring B was assumed to be coplanar with the porphyrin macrocycle, ring D was twisted out of the molecular plane by 5.5°. Our earlier attempts to employ a molecular geometry based on the Xray structure coordinates of chlorophyllide a [46] did not improve the calculated hfc’s. In general, the agreement of the calculated isotropic hfc’s given in Table 4 and the experimental data (Table 2) is satisfying, however, it is not as good as in case of the substituted BChltype cation radicals investigated previously [30]. The resulting b-proton hfc’s are reproduced. Sim-

1152

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

ilarly, the trends of changes, induced by the substitutions of the macrocycle, are fully reproduced for the methyl proton hfc’s (vide infra). Nevertheless, the calculated values of the methyl hfc’s are too close to each other, and the experimental values are larger than the theoretical ones for position 121 and smaller for positions 21 and 71. Liquid solution ENDOR on 1 + ’ yielded also hfc’s for the methin positions [30]. As already found for the BChl-type cation radicals the calculated methin proton hfc’s at positions 5 and 10 are overestimated in the calculations. Furthermore, in case of the pyro-compounds 3 + ’ and 5 + ’ the theoretical 1H-hfc’s at position 132 are too large. This effect has also been observed in the RHF-INDO/SP calculations of the substituted BChl-type cations. As possible reason it has been proposed that ring E in chlorophyll species cannot be modeled by standard bond lengths and angles [30].

3.4.1. Substitution at position 3 In the experiment, the methyl proton hfc at position 2 increases when substituting the vinyl group at position 3 by an acetyl group. This is also found in the calculations. The calculated 1 H-hfc’s vary slightly when the side groups are rotated relative to the molecular plane. As in case of the BChl-type cations [30] out-of-plane angles a of 50 and 75° were employed in the optimum calculations for 1 + ’ and 4 + ’, respectively. Here a is the angle between the molecular plane and the vinyl CC bond in 1 + ’ and between this plane and the acetyl CO bond in 4 + ’. 3.4.2. Substitution at position 13 2 As compared with the cation radicals 1 + ’ and +’ 4 the experiments show a clear decrease of the methyl proton hfc at position 12 in the related pyro-compounds 3 + ’ and 5 + ’, respectively. The 1 H-hfc’s at the other positions — except position 132 —should essentially remain constant. This is in good agreement with the RHF-INDO/SP calculations. The results presented above show that RHFINDO/SP yields good estimates for the 1H-hfc’s in the substituted Chl-type cation radicals. The deviations of the calculated data from the experi-

mental ones are larger than in case of the related substituted BChl-type systems. Nevertheless, the predicted qualitative trends induced by substitutions are clearly the same in Chl-type and BChltype cation radicals and are in good agreement with the experiment.

3.4.3. Anisotropic hfc’s To first approximation the hfc of methyl group protons can be described by an axially symmetric tensor with its A principal axis oriented along the CCH3 bond [43]. However, depending on the spin density residing on the neighboring atoms, a non-axial hfc tensor as well as a non-zero angle between its principal axis of this component and the methyl bond are generally expected. We have tried to model this situation by the RHF-INDO/ SP program [40]. For this purpose 1H-hfc tensors-averaged for rotating methyl groups-have been calculated for the methyl groups at molecular positions 2, 7 and 12 in 1 + ’ and for 2 and 12 in 7 + ’. Axes definitions are given in Fig. 6. The difference of the

Fig. 6. Schematic drawing of Chl a to illustrate the in-plane angle Df between the C – CH3 bond direction in standard geometry (see text) and the A33-axis of the respective 1H-hfc tensor in Chl a + ’ (1 + ’), calculation based on RHF-INDO/SP spin densities. The angles Df for the three ‘marker’ CH3 groups are 4.0° (pos. 12), 5.2° (pos. 7) and 9.4° (pos. 2). The calculated hf tensor values are (sequence: A33, A22 and A11); 7.0, 5.3 and 5.1 MHz (for CH3 at pos. 12); 6.4, 4.7 and 4.5 MHz (for pos. 7); 6.2, 4.6 and 4.5 MHz (for pos. 2); x and y depict the molecular axes.

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

resulting principal values of the CH3 tensors are 5 2 MHz. Consequently, the frozen solution ENDOR spectra, in which the individual CH3 tensor components are not resolved, should exhibit a line width in the order of 1 MHz. This agrees well with the experiment (see Fig. 5). The calculated principal axis corresponding to the largest principal value A332 of the respective hfc tensors lies always in the molecular plane, as expected for planar p-radicals. However, this axis is not always exactly parallel to the CCH3 bond. The values for the angular deviations Df are given in Fig. 6. The calculation clearly shows that these differences are due to a variation in the distribution of pz spin densities residing on the neighboring carbon atoms. For example, RHFINDO/SP yields a difference of almost two orders of magnitude for next neighbor spin densities at C1 and C3 for 1 + ’ (0.1000 and 0.0024, respectively). A somewhat different situation is found for the methyl groups at molecular positions 7 and 12. A similar result has been obtained for BChl a + ’ for which Df :6.3° both for position 2 and 12. The geometry used for the RHF-INDO/SP calculations is only an approximation. Nevertheless, in case of the methyl bonds the maximum orientational difference between the standard model used in our calculations and the experimental X-ray structure determined for ethyl-chlorophyllide a [46] is about 19 0.2°. Thus the calculated results may be regarded at least as a reasonable estimate of the real differences. Differences of the same magnitude were found for the difference between hf tensor directions of the oxidized primary donor + ’ measured by ENDOR in single crystal and P865 the respective CCH3 bond directions in the Xray structure of the RC of Rhodobacter sphaeroides [18].

3.5. Orbital mixing in Chl a cation radicals SCF-MO calculations performed by Davis et al. [5] on the cation radical of a simplified metal2

The A33 principal component of a non-axial tensor is equivalent to the A component of an axial methyl-proton tensor.

1153

lochlorine model yielded a separation of electronic ground state (2A2) MO and first excited state (2B2) MO of only about 0.25 eV. In this work it was assumed that interaction with ligands could further reduce this difference which in turn might cause a mixing of the two states. Thus, for Chl a+’ a dependence of experimental hfc’s from the solvent was expected and some data were interpreted in this way [5]. In a different approach, total energies and spin density distributions of the molecular doublet ground state (D0) and the first excited doublet state (D1) were calculated for ethyl-chlorophyllide a + ’ and ethyl-bacteriochlorophyllide a + ’ by an ab initio CI method [23]. The energetic difference between D0 and D1 was estimated to be 0.65 eV in the case of Chl a + ’ and 0.88 eV for BChl a + ’. Both investigations yielded a significant difference of the spin density distribution between the calculated states. This led to a new interpretation + ’ in PS I [22] of the experimental 1H-hfc’s of P700 that is based on a mixing of the D0 and the D1 state in a monomeric Chl a cation radical, due to interactions with the protein matrix. Thereby the 1 + ’ H-hfc’s of P700 could be modeled with-in generalsatisfying agreement [22]. Recently, the occurrence of a nitrogen hfc that was assigned to a histidine + ’ ligand of P700 in PS I has also been explained in the framework of this model [24]. The validity of the orbital mixing model can be probed by detection of the effect of different substituents on the spin density distribution of Chl a + ’. It could be shown that environmental effects like single H-bonds shift the redox potential of the primary donor cation in bacterial RCs by an average value of 40 mV and a maximum of 125 mV [47]. A comparable change of redox properties was measured for changes of the substitution pattern of monomeric chlorophylls in solution [48]. This shows that energetic shifts caused by substitution will be of the same order of magnitude as shifts introduced by the solvent surrounding or the protein matrix. The theoretical calculations showed that the spin density distribution over the molecule is different in the ground and excited doublet state [23]. If these states are indeed very close in energy, substitutions could thus affect the mixing and lead to drastic changes

1154

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

of the electron spin density distribution over the whole molecule [22]. This would lead to changes of the electronic g factor and the measured hyperfine couplings. Furthermore, since such effects have never been detected in BChl a-type cation radicals, the observed changes should be significantly different from those observed in the BChl a + ’ series [30]. A close inspection of the alterations observed in the differently substituted Chl-type cation radicals studied in this paper yield the following results: 1. The g factor of Chl a + ’ is close to 2.0025(1), whereas that of Chl a − ’ is 2.0029(1) [13]. To first order, these values are typical for an unpaired electron delocalized in the HOMO and LUMO of a BChl a radical that has a large HOMO/LUMO gap. Orbital mixing should therefore lead to a change of the g value. The measured g factor is, however, constant within experimental error for all Chl-type cation radicals studied in this paper. 2. The hyperfine couplings measured for the different Chl-type radicals are also not drastically changed. Furthermore, it was noted that a substituent influences to first order only the hfc’s in its immediate surrounding, long range effects are generally small. In the orbital mixing model one would expect larger alterations over the whole macrocycle. The changes of hfc’s detected are also quite similar in Chl- and BChl-type cation radicals. These findings do not support a significant orbital mixing in the Chl-type cation radicals. 3. Molecular orbital calculations (RHF-INDO/ SP) fully reproduce the observed trends in the spin density shifts of the various substituted Chl- and BChl-type cation radicals. Even the quantitative agreement between experiment and calculation is satisfying. The molecular orbital in which the unpaired electron resides in all systems is typical for the HOMO of chlorophyll cation radicals [19]. 4. The EPR saturation behavior of all Chl-type radical cations is similar — but it is different from that in the BChl a series. This must not necessarily be due to an orbital mixing but can be explained by a difference in aggregation behavior in solution. The tendency for Chl

cation radicals to aggregate in various solvents leads to spin exchange phenomena that severely aggravate mw power saturation and impede the detection of cw-ENDOR. This is the reason that ENDOR detection of Chl a + ’ and related systems is only possible for highly purified (lipid-free) material in solvents with special ligation properties (e.g. THF) at low temperatures where aggregation and spin exchange are effectively suppressed. That these arguments are correct is nicely demonstrated by the behavior of Chl a cation radicals in a protein matrix that prevents aggregation. Such species are easily saturated by mw and rf power and strong and narrow ENDOR signals + ’ in PS I is a particularly can be obtained. P700 good example of this behavior [25,26,28,45]. The above arguments show that orbital mixing of the D0 and D1 state as proposed in [5,22] does actually not play a role in Chl-type cation radicals. We suggest that the difficulties experienced by various groups in the past to explain their observations can essentially be traced back to pigment-pigment interactions in solution that are particularly severe at the high concentrations (10 − 4 –10 − 3 M) used for EPR/ENDOR experiments. If such effects can be eliminated, monomeric Chl cation radicals can be obtained that are good model systems for the respective species occurring both in PS I and PS II of plant photosynthesis.

4. Conclusions In this paper we have studied Chl-type cation radicals in glassy and powder solutions with different substitution patterns around the macrocycle supplementing an earlier study on similar BChl-type cation radicals [30]. For all species the g-factor and 1H-hyperfine coupling constants were measured by EPR and ENDOR, respectively. From the assigned hfc’s an experimental spin density distribution was obtained and compared with that derived from MO calculations. These RHF-INDO/SP calculations clearly supported the experimental finding that Chl-type and BChl-type cation radicals are closely related. In

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

general the method yielded good agreement between MO-theoretical and experimental results. However, the calculated hfc values were more accurate for the BChl-type systems [30]. This is probably due to a less well-defined molecular structure in case of Chl a. Furthermore, the anisotropic hfc’s of the three methyl groups at the Chl macrocycle were calculated including the tensor axes. These serve as structural markers for studies on chlorophyll radical ions in proteins. The tensor information for Chl a + ’ is important for studies of this species in photosynthetic proteins, in particular in protein single crystals, e.g. PS I [27,28]. Zn–Chl a + ’ has also been included in this study. The spin density distribution is not significantly altered as compared to the parent radical Chl a + ’. Furthermore, Zn-BChl a + ’ has also been investigated by ENDOR in liquid and frozen solution yielding a very detailed map of the spin density distribution of this species. This shows that the orbital distribution is very similar to that of BChl a + ’. The optical and redox properties of Zn – BChl are also not very different from those of BChl a [49–52]. This will enable the Zn-compound to replace the Mg – BChl a in its function in vivo, as was recently found for the bacterium Acidiphilium (Ac.) rubrum [33]. In the reaction center of this species a pigment absorbing at 855 nm was detected that bleaches under actinic light [33]. It was tentatively assigned to the primary donor which is possibly a Zn – BChl a dimer. The monomeric Zn–BChl a + ’ studied in this work is a good model system for future investigations of + ’ the electronic and geometrical structure of P865 in Ac. rubrum. Such experiments are under way in our laboratory. The analysis of our EPR and ENDOR data of the various substituted Chl-type radical cations showed that an electronic model for such species, based on mixing of ground and excited states [5,22], is probably not correct. The chemical and spectroscopic peculiarities of these radicals could be explained by the unusual aggregation behavior of these species in solution. The monomer model + ’ for P700 is therefore not supported by our experiments. ENDOR and ESEEM measurements on + ’ P700 recently performed in our laboratory suggest

1155

that this species is a Chl a dimer with a strongly asymmetric spin density distribution over the dimer halves [27,28]. Support for this model is expected from future studies of site-directed mutants in which amino acids in the surrounding of P700 are altered [53]. In this work we have determined the electronic structure of monomeric Chl a + ’ in (frozen/glassy) solution and could show that it is very probably a good model system for ligated chlorophyll cation radicals in proteins. This is of particular importance for PS II in which several different species of this kind might be present created by the + ’ strongly oxidizing P680 . The spatial and electronic + ’ structure of P680 is studied by EPR, ENDOR spectroscopy and related techniques by several researchers including our own group.

Acknowledgements The authors wish to thank Dr Martin Plato for providing us with the latest release of his RHFINDO/SP program. The work was supported by the ‘Deutsche Forschungsgemeinschaft’, Sfb 312 (TP A4), Sfb 143 (TP A6), and Az Sche 140/9-2 by the Avron-Minerva Center for Photosynthesis (Israel) and by ‘Fonds der Chemischen Industrie’ (to W. Lubitz). Supporting Information: Cartesian coordinates of the molecular structures of 1, 3, 4, 5 and 7 employed in the RHF-INDO/SP calculations of Tables 4 and 5 can be obtained from the corresponding author.

References [1] H. Scheer, (Ed.), Chlorophylls, CRC Press, Boca Raton, 1991. [2] C.R.D. Lancaster, U. Ermler, H. Michel, in: R.E. Blankenship, M.T. Madigan, C.E. Bauer (Eds.), Anoxygenic Photosynthetic Bacteria, Kluwer, Dordrecht, 1995, p. 503. [3] N.W. Woodbury, J.P. Allen, in: R.E. Blankenship, M.T. Madigan, C.E. Bauer (Eds.), Anoxygenic Photosynthetic Bacteria, Kluwer, Dordrecht, 1995, p. 527. [4] J. Deisenhofer, H. Michel, Biochemistry 27 (1988) 1. [5] M.S. Davis, A. Forman, J. Fajer, Proc. Natl. Acad. Sci. USA 76 (1979) 4170.

1156

H. Ka¨ß et al. / Spectrochimica Acta Part A 54 (1998) 1141–1156

[6] J.H.A. Nugent, A. Telfer, C. Demetriou, J. Barber, FEBS Lett. 255 (1989) 53. [7] S.E.J. Rigby, J.H.A. Nugent, P.J. O’Malley, Biochemistry 33 (1994) 10043. [8] D.R. Ort, Ch.F. Yocum, (Eds.), Oxygenic Photosynthesis: The Light Reactions. Kluwer, Dordrecht (1996). [9] G. Renger, Physiol. Plantarum (1997) (in press). [10] N. Krauß, W.-D. Schubert, O. Klukas, P. Fromme, H.T. Witt, W. Saenger, Nature Struct. Biol. 3 (1996) 965. [11] P. Se´tif, in: J. Barber (Ed.), Topics in Photosynthesis, Elsevier, Amsterdam, 1992, p. 471. [12] H.T. Witt, in: D.R. Ort, C.F. Yocum (Eds.), Oxygenic Photosynthesis: The Light Reactions, Kluwer Academic Publishers, Dordrecht, 1996, p. 363. [13] W. Lubitz, in: H. Scheer (Ed.), Chlorophylls, CRC Press, Boca Raton, Florida, 1991, p. 903. [14] J.R. Norris, R.A. Uphaus, H.L. Crespi, J.J. Katz, Proc. Natl. Acad. Sci. USA 68 (1971) 625. [15] K. Mo¨bius, W. Lubitz, M. Plato, in: A.J. Hoff (Ed.), Advanced EPR, Applications in Biology and Biochemistry, Elsevier, Amsterdam, 1989, p. 441. [16] W. Lubitz, F. Lendzian, Advances in Photosynthesis, Biophysical Techniques in Photosynthesis, Kluwer, Dordrecht, 1996, p. 255. [17] S.A. Dikanov, Y.D. Tsvetkow, Electron Spin Echo Envelope Modulation (ESEEM) Spectroscopy, CRC Press, Boca Raton, Florida, 1992. [18] F. Lendzian, H. Huber, R.A. Isaacson, B. Endeward, M. Plato, B. Bo¨nigk, K. Mo¨bius, W. Lubitz, G. Feher, Biochim. Biophys. Acta 1183 (1993) 139. [19] M. Plato, K. Mo¨bius, W. Lubitz, in: H. Scheer (Ed.), Chlorophylls, CRC Press, Boca Raton, Florida, 1991, p. 1015. [20] J. Rautter, F. Lendzian, C. Schulz, A. Fetsch, M. Kuhn, X. Lin, J.C. Williams, J.P. Allen, W. Lubitz, Biochemistry 34 (1995) 8130. [21] M. Huber, R.A. Isaacson, E.C. Abresch, D. Gaul, C.C. Schenck, G. Feher, Biochim. Biophys. Acta 1273 (1996) 108. [22] P.J. O’Malley, G.T Babcock, Proc. Natl. Acad. Sci. USA 81 (1984) 1098. [23] J.D. Petke, G.M. Maggiora, L.L. Shipman, R.E. Christoffersen, Photochem. Photobiol. 31 (1980) 243. [24] M. Mac, X.-S. Tang, B.A. Diner, J. McCracken, G.T. Babcock, Biochemistry 35 (1996) 13288. [25] H. Ka¨ß, E. Bittersmann-Weidlich, L.-E. Andre´asson, B. Bo¨nigk, W. Lubitz, Chem. Phys. 194 (1995) 419. [26] S.E. Rigby, J.H.A. Nugent, P.J. O’Malley, Biochemistry 33 (1994) 10043. [27] H. Ka¨ß, P. Fromme, H.T. Witt, W. Lubitz, Biophys. J. 66 (2) (1994) 228. [28] H. Ka¨ß, PhD thesis, Technische Universita¨t Berlin, 1995. [29] H. Ka¨ß, P. Fromme, W. Lubitz, Chem. Phys. Lett. 257 (1996) 197. [30] H. Ka¨ß, J. Rautter, W. Zweygart, A. Struck, H. Scheer, W. Lubitz, J. Phys. Chem. 98 (1994) 354. .

[31] H. Scheer, J.J. Katz, J.R. Norris, J. Am. Chem. Soc. 99 (1977) 1372. [32] A.-F. Miller, G.W. Brudvig, Biochim. Biophys. Acta 1056 (1991) 1. [33] N. Wakao, N. Yokoi, N. Isoyama, et al., Plant Cell Physiol. 37 (1996) 889. [34] E. Nabedryk, M. Leonhard, W. Ma¨ntele, J. Breton, Biochemistry 29 (1990) 3242. [35] L. Fiedor, V. Rosenbach-Belkin, A. Scherz, J. Biol. Chem. 267 (1992) 22043. [36] P.H. Hynninen, in: H. Scheer (Ed.), Chlorophylls, CRC Press, Boca Raton, Florida, 1991, p. 145. [37] W. Zweygart, R. Thanner, W. Lubitz, J. Magn. Reson. A 109 (1994) 172. [38] M. Plato, E. Tra¨nkle, W. Lubitz, F. Lendzian, K. Mo¨bius, Chem. Phys. 107 (1986) 185. [39] J.A. Pople, D.L. Beveridge, Approximate Molecular Orbital Theory, McGraw-Hill, New York, 1978. [40] M. Plato, F. Lendzian, W. Lubitz, K. Mo¨bius, in: J. Breton, A. Vermeglio (Eds.), The Photosynthetic Bacterial Reaction Center II: Structure, Spectroscopy and Dynamics, Plenum Press, New York, 1992, p. 109. [41] T.L. Horning, E. Fujita, J. Fajer, J. Am. Chem. Soc. 108 (1986) 323. [42] M.K. Bowman, T.J. Michalski, R.L Tyson, D.L. Worcester, J.J. Katz, J. Proc. Natl. Acad. Sci. USA 85 (1987) 1498. [43] H. Kurreck, B. Kirste, W. Lubitz, Electron Nuclear Double Resonance Spectroscopy of Radicals in Solution, VCH, Weinheim, 1988. [44] M. Huber, F. Lendzian, W. Lubitz, E. Tra¨nkle, K. Mo¨bius, M.R. Wasielewski, Chem. Phys. Lett. 132 (1986) 467. [45] F. Lendzian, PhD thesis, Freie Universita¨t Berlin (1982). [46] H.-C. Chow, R. Serlin, C.E. Strouse, J. Am. Chem. Soc. 97 (1975) 7230. [47] X. Lin, H.A. Murchison, V. Nagarajan, W.W. Parson, J.P. Allen, J.C. Williams, Proc. Natl. Acad. Sci. USA 91 (1994) 10265. [48] Y. Jia, T.J. DiMagno, C.-K. Chan, Z. Wang, M. Du, D.K. Hanson, M. Schiffer, J.R. Norris, G.R. Fleming, M.S. Popov, J. Phys. Chem. 97 (1993) 13180. [49] C. Geskes, G. Hartwich, H. Scheer, W. Ma¨ntele, J. Heinze, J. Am. Chem. Soc. 117 (1995) 7776. [50] G. Hartwich, C. Geskes, H. Scheer, J. Heinze, W. Ma¨ntele, J. Am. Chem. Soc. 117 (1995) 7784. [51] M. Meyer, G. Fritsch, U. Ermler, C. Geskes, H. Michel, J. Heinze, H. Scheer, in: P. Mathis (Ed.), Photosynthesis: From Light to Biosphere, Kluwer Academic Publishers, Dordrecht, 1995, p. 919. [52] A. Struck, E. Cmiel, I. Katheder, W. Scha¨fer, H. Scheer, Biochim. Biophys. Acta 1101 (1992) 321. [53] A.N. Webber, H. Su, S.E. Bingham, H. Ka¨ß, L. Krabben, M. Kuhn, E. Schlodder, W. Lubitz, Biochemistry 35 (1996) 12857.