Electron spin resonance of charge carriers in ... - Semantic Scholar

Proc. Nati. Acad. Sci. USA Vol. 85, pp. 1498-1502, March 1988 Biophysics

Electron spin resonance of charge carriers in chlorophyll a/water micelles (oxidized chlorophyll/one-dimensional conductor/electron transfer)

M. K. BOWMAN, T. J. MICHALSKI, R. L. TYSON*, D. L. WORCESTERt, AND J. J. KATZ Chemistry Division, Argonne National Laboratory, Argonne, IL 60439

Contributed by J. J. Katz, October 26, 1987

recognized that they have some long-range order (2), it was generally considered that the micellar systems under discussion here had such a wide distribution of sizes, shapes, and organization as to defy description in structural terms. This is very far from being the case. Small-angle neutronscattering (SANS) studies of chlorophyll/water micelles (dispersed in n-octane or n-octane/toluene mixtures) show that they have similar, regular shapes, and for any particular chlorophyll/water micelle, the shape, size, and mass distribution of the chlorophyll is very well defined (8). This work has made it possible to advance a fairly detailed model of the Chla/water micelle. In this paper, we shall be concerned exclusively with this highly ordered, well-characterized species. The Chla/water micelles are important not only for their role as models of in vivo chlorophyll properties but also because (i) these self-assembling, macromolecular Chla structures exhibit some properties of natural photosynthetic structures (8) and (ii) the high degree of order makes it possible to study electron and perhaps excitation-energy (exciton) transport in large arrays of chlorophyll with a well-defined structure. The micelles are long, hollow cylinders with an inner diameter of 11.4 nm. The surface of the cylinders consists of a monolayer of chlorophyll molecules containing chains of macrocycles crosslinked by intercalated water. The oxygen atom of a water molecule is coordinated to the central magnesium atom of one macrocycle, and one of its hydrogen atoms forms a hydrogen bond to the ring V keto carbonyl function of its neighbor in the chain. Within the chains, rings I and III of adjacent Chla molecules partially overlap. The strong ii-ir overlap in the chain and the hydrogen bonds provide a pathway for either electron delocalization or rapid intrachain electron transfer. Interchain hydrogen bonds formed between the ring V carbomethoxy group in one chain segment and the propionic ester carbonyl group of an adjacent chain segment crosslink the chains to form a long cylinder. Crosslinkage between chains to form a closed sheet does not connect the X systems of the macrocycles. Interchain interactions are expected to be much less effective than intrachain interactions in promoting either electron delocalization or electron transfer between the ir systems. The SANS data (8) are compatible with Chla chains parallel to the cylinder axis or oriented around the circumference of the cylinder as closed rings or as a shallow helix. From a chemical and physical point of view, the idea of a low-pitch helical arrangement of the chains is attractive. In any of these arrangements, all of the macrocycles in the surface of the micelle are fully equivalent.

Chlorophyll a/water micelles (P740) preABSTRACT pared in hydrocarbon media have been shown by small-angle neutron scattering to consist of hollow cylinders whose surface is formed of a monolayer of chlorophyll crosslinked by water. The micelles can be reversibly oxidized or reduced to generate highly mobile holes or electrons that undergo rapid, onedimensional transport along the chains of chlorophyll macrocycles comprising the surface of the micelles. Large 17-1 overlap within the chains facilitates the one-dimensional charge transport and is expected to do the same for energy transport. Structural defects in the micelle surface act as boundaries for charge transport, confining the spins to onedimensional domains of approximately 200 macrocycles. The one-dimensional transport within the limited domains results in motionally narrowed electron spin resonance lines with some residual inhomogeneous broadening. Although the chlorophyll a incorporated in micelles is more easily oxidized than is monomeric chlorophyll a, it is much more resistant to chemical alteration.

Chlorophyll/water micelles are readily formed by addition of stoichiometric amounts of water to a solution of chlorophyll in nonpolar organic solvents (1, 2). The micelles so produced have optical spectra that are strongly red-shifted, and, when oxidized, yield a variety of ESR signals (3-5). As these properties are also displayed by chlorophyll in vivo, the chlorophyll/water micelles have received considerable attention, and, indeed, the study of such chlorophyll species has provided much of the basis for the special-pair concept of photoreaction-center chlorophyll (6, 7). ESR signals in chlorophyll/water micelles have been generated either by illumination with red light or by small amounts (relative to the number of chlorophyll molecules) of oxidants such as 12, 02, or FeCl3 (6, 7). In many cases the ESR spectra obtained from chlorophyll a (Chla)/water micelles consist of a single line whose width is much less than that measured for monomeric Chlat cation radicals. The extraordinarily narrowed ESR signal of the oxidized micelles is assigned to radical cations of the Chia. The ESR line-

narrowing has been explained either by delocalization of the unpaired electron over many chlorophyll molecules or by rapid hopping of the unpaired electron over a large domain of macrocycles (6, 7). These two line-narrowing mechanisms result in two different limiting ESR lineshapes. In the case of electron delocalization, the ESR lineshape is expected to be Gaussian, whereas for itinerant electrons, the limiting lineshape is nearly Lorentzian. The shape of the very narrow ESR line in previous studies of oxidized Chla/water micelles was not often reported. Until recently, no structural information was available for any of the chlorophyll/water micelles, and although it was

Abbreviations: Chla, chlorophyll a; Chlb, chlorophyll b; SANS, small-angle neutron scattering. *Present address: Department of Chemistry, University of Saskatchewan, Saskatoon, Saskatchewan, S7N OWO Canada. tPermanent address: Biology Division, University of Missouri, Columbia, Columbia, MO 65211.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

1498

Biophysics: Bowman et al. The presence of ir-ir overlap within the chains and the absence of such interactions between adjacent chain segments suggest that transport of unpaired electrons may occur along an effectively one-dimensional chain of Chla molecules. The one-dimensionality would have important consequences not only for the ESR spectrum but also for excitation-energy transport and charge transport in these micelles. Models of photosynthetic structures based on the properties of these micelles would also share the distinctive one-dimensional character of the micelles.

MATERIALS AND METHODS Materials. Chla/water micelles were prepared under nitrogen as described by Worcester et al. (8). Our first ESR experiments were carried out on a part of the micelle sample used previously in the SANS experiments (8). Subsequent preparations were made specifically for ESR. The optical spectrum of each micelle preparation was recorded (Hewlett-Packard 8451A diode-array spectrophotometer) and compared with spectra recorded from samples of known structure. Periodically, aliquots of the sample were examined by SANS to verify the structural integrity of the

preparation. The chemical purity of the ESR samples was routinely determined by HPLC analysis [Beckman 110A dual-pump system with an Ultrasphere ODS (5 pum) column (4.6 mm x 25 cm)]. The Chla in some of the initial ESR samples contained small amounts of 10-hydroxy-Chla, chlorophyll b (Chlb), pyrochlorophyll a, pheophytin a, lipids, and carotenoids. Micelles made from this batch of Chla gave inconsistent ESR results. In our later ESR experiments, Chla was first purified by HPLC followed by precipitation from octane solution by addition of water. Chla prepared by HPLC contained only trace amounts of pyrochlorophyll a and pheophytin a. The ESR results then became much more consistent. The 0.05 samples of Chla micelles (10 mg of Chla per ml) were oxidized or reduced under nitrogen in quartz ESR tubes (4-mm outer diameter) by the addition of 1-10 1.l from a set of stock solutions of 12 in octane or saturated aqueous potassium dithionate. The solutions were mixed in the tubes by gentle agitation and by mild sonication for about 30 sec. The ESR tubes were sealed against the admission of oxygen by a gas-tight O-ring seal. ESR Measurements. ESR measurements were performed with a Varian E-9 spectrometer operating at X-band frequency. Sample temperatures below ambient were maintained by a variable-temperature nitrogen gas flow system inserted into the cavity. Measurements were made below 0.5-mW incident power, which avoided saturation effects. The 100-kHz magnetic field modulation amplitude was kept below 10% of the linewidth. Relative spin susceptibilities (numbers of unpaired spins) and lineshapes were measured by a convolutional fitting procedure for inhomogeneously broadened lines (M.K.B. and R.L.T., unpublished work). Pulsed ESR measurements were made on an electron spin echo spectrometer (9) of our own construction. Microwave pulses of 250 W with a duration of about 30 or 60 ns were used to rotate the sample magnetization by 90° or 1800, respectively. The interpulse spacing was varied, as needed, in equal steps with a resolution of 10 ns. A three-pulse inversion-recovery (180o-90°-180°) pulse sequence was used to measure the electron spin lattice relaxation rate Tj1-e, whereas a two-pulse Hahn echo (90°-180°) sequence was used to measure the phase memory time TM (10). The data were recorded with a Nicolet 1180E computer, which also extracted the relaxation rates by using a nonlinear leastsquares fitting routine.

Proc. Natl. Acad. Sci. USA 85 (1988)

1499

RESULTS ESR of Oxidized Micelles. Chla micelles treated in solution with 12 are oxidized to free-radical species. The ESR spectrum consists of a single line strongly narrowed in comparison to the ESR line of oxidized, monomeric Chla+. The ESR linewidth varies from 0.30 to 0.033 millitesla (mT), whereas that of the monomer Chlat radical cation is 0.93 mT (6, 7). The ESR linewidth varies with the purity of the Chla, but in each sample it decreases toward the same limiting linewidth as the oxidant concentration is increased. The ESR lines are symmetric except at the narrowest linewidths, where the ESR line develops a pronounced asymmetry characteristic of an axial g factor (Fig. 1). The spin susceptibility, or number of unpaired electrons, increases linearly with added oxidant concentration over a molar ratio of 3 x 10-4 to 0.5 [I2]/[Chla] (Fig. 2) before decreasing at high oxidant concentrations. Many samples, particularly those containing impurities, have a very weak ESR signal prior to the addition of oxidant. This weak signal often shows some reversible photoactivity, either increasing or decreasing in amplitude when illuminated with red light. Addition of small amounts (-10 pmol) of I2 sometimes increases and sometimes decreases this preexisting signal. In micelles made from Chla containing impurities, the ESR linewidths and lineshapes change slowly over the course of a few hours, particularly at low oxidant concentrations. When large volumes of solvent are introduced with the oxidant, the ESR linewidth changes, presumably reflecting a slow relaxation of the micelle structure in response to changes in total Chla concentration. ESR Temperature Dependence. The ESR linewidth varies smoothly as the sample temperature is reduced below 293 K (Fig. 3) In particular, there is no discontinuity in the linewidth when the sample freezes at 216 K. The spin susceptibility increases linearly with the inverse of the temperature, showing Curie-law behavior (Fig. 4). Pulsed ESR Measurements. ESR relaxation measurements were performed on Chla/water micelles that showed gfactor anisotropy. Values of Tj;1, the electron spin lattice relaxation rate, are plotted in Fig. 5 as a function of temperature. The spin lattice relaxation rate is proportional to the temperature. The measured phase memory relaxation rate TM1j is a weak function of the sample temperature (Fig. 5). No indication of instantaneous spectral diffusion was found in the pulsed ESR measurements, suggesting that electron spin-electron spin interactions are unimportant in determining phase memory relaxation (10). Reduction of the Micelles. ESR signals were also produced in Chla/water micelles by the addition of 1-5 ,ul of saturated aqueous dithionite solution, followed by illumination with red light. The spectra consisted of a single weak ESR line whose width varied from 0.06 mT to more than 0.40 mT,

FIG. 1. ESR spectrum of Chla/water micelles in octane/toluene, 1:1, oxidized by 12 at 293 K; microwave power, 1.0 mW; frequency, 9.135 GHz; modulation amplitude, 0.0016 mT. The g-factor difference is about 0.00028.

Proc. Natl. Acad. Sci. USA 85 (1988)

Biophysics: Bowman et al.

1500

500 -

400 .

300 U) C:

U)

c

C

U

200 -

100 .

4

3.5

0.1

[l2]/[Chla] FIG. 2. 12 concentration dependence of the spin susceptibility. Chia concentration was 1.12 mM. on the amount of dithionite added. The signals not very stable, disappearing in the dark in the course

depending were

of 30 min. Visual examination of the samples suggested that the saturated dithionite solution destroyed the micelle structure.

Micelle Stability. The dispersions of micelles are surprisingly stable with respect to chemical modification of the pigments. HPLC analyses were performed on micelle dispersions stored at room temperature for several days; samples that had been oxidized and stored at room temperature under nitrogen for several days; and samples that had been oxidized and then exposed for several minutes to the focused output of a 300-W Xe arc lamp (Eimac Division, Varian Associates) filtered by 9 cm of water. HPLC analysis showed no increase in the amount of altered chromophores-in particular, no increase in 10-hydroxy-Chla. Oxidation of the micelles is completely reversible over periods of several hours, and the Chla is recovered unchanged within the limit of detection.

DISCUSSION The ESR signals observed in the Chla/water micelles are produced by oxidation or reduction reactions. The weak, variable ESR signals observed in the absence of any added 0.25

5

4.5

1/Temperature (K)

5.5

6

FIG. 4. Temperature dependence of the spin susceptibility. The lines are least-squares fits of the Curie law to the data at two concentrations of 12: 131 AM (m) and 657 AM (o).

oxidant or reductant are attributed to small amounts of electron-acceptor or -donor impurities in the solvent and/or in the Chla. The weak, variable photoinduced signals are likewise attributed to light-driven redox reactions involving trace impurities. Very pure Chla/water micelles in the absence of electron donors or acceptors do not appear to have intrinsic, photoactive ESR signals easily detected in normal ESR experiments. At low concentrations of oxidants, the ESR spectra are influenced by signals from the small amounts of easily oxidized impurities. However, at higher oxidant concentrations, the impurity radicals contribute negligibly to the total ESR signal, and the limiting linewidths are those of the mobile electron spins on the Chla/water micelles. This limiting ESR linewidth, as the samples are cooled from 300 K to 77 K, shows a large, smooth variation that is difficult to explain if the unpaired spin species is a delocalized electronic state spread over many molecules. Were that the case, the half-occupied molecular orbital must expand and contract with temperature in order to produce the observed temperature-dependent linewidth. This is highly unlikely behavior. Rather, the ESR linewidth and lineshape variations are likely the product of electron spin motion on the micellar surface. The thermally activated transport of the 1001

-

10.

*

-

.

0.20-

N

:P

0.15-

0.10

0.100.001-

a1

150

200

250

300

Temperature, K FIG. 3. Experimental temperature dependence of the peak-topeak linewidth of an oxidized Chla/water micelle sample.

50

100

150

200

250

Temperature, K FIG. 5. Temperature dependence of TiC1 (a) and TMj1 (U). Smooth curves have been drawn through the experimental data.

Biophysics: Bowman et al. spins produces random changes in the electron spin-electron spin and electron spin-nuclear spin interactions experienced by the unpaired electron. These random changes narrow the ESR spectrum. Analogous effects are observed in the ESR spectra of the mobile spins in polyacetylene (11). Both the observed Curie-law behavior and the absence of any abrupt changes in ESR properties when the solvent freezes indicate that the spin motion is purely electronic in nature and does not depend on molecular or ionic transport. The pulsed ESR measurements show a small decrease in the phase memory relaxation rate at high temperatures. Highly mobile spins moving from macrocycle to macrocycle by a thermally driven process exhibit this temperature dependence when the electron-transfer rate (in MHz) exceeds the ESR linewidth (in MHz). These are precisely the same conditions for the observation of a strongly narrowed ESR spectrum. The weak temperature dependence of the electron spin lattice relaxation time is expected if the activation energy for spin motion is small (11). Rapid spin motion affects the ESR lineshape in two ways, producing (i) motional or dynamic narrowing of the ESR signal and (it) lineshapes that depend strongly on the detailed motion of the electron. While our results show strong dynamic narrowing of the ESR signal, which is characteristic of rapid motion of an electron, the lineshapes, by themselves, do not permit an unequivocal discrimination between one- and two-dimensional motion. However, lineshapes characteristic of one-dimensional motion can be absent in other spin systems such as polyacetylene (11) and linear Heisenberg chains (12), which are known to be onedimensional. In these systems, the characteristic "onedimensional" lineshape is not observed because the long "tails" of the one-dimensional spin correlation function are destroyed by interchain interactions or by impurities in the chains, and this could be true for our system as well. The incorporation of a small amount of Chlb in the Chla micelles results in a broader ESR line (0.08 mT linewidth from Chla micelles formed from a solution containing 33% Chlb). Chlb that gets incorporated into the Chla/water micelles can act as a barrier to electron transport because the redox potential of Chlb is not the same as that of Chla in the micelle. This disruption of long-range electron transport shortens the domains over which the electron can move, thereby broadening the ESR line. Other point defects in the micelle structure (such as a water-molecule vacancy, a Chla vacancy, or a substitutional impurity), or a major disruption in the micellar structure, will also disrupt long-range electron transport, destroying the characteristic low-dimensional lineshape and producing the observed spectrum. The partially narrowed ESR lineshape and its temperature dependence in the Chla/water micelle indicate the presence of mobile spin species, and the lineshape does not preclude either onedimensional or two-dimensional transport over limited domains on the micelle surface. The ESR lineshapes of the Chla/water micelles are described very well by a convolution of an inhomogeneous broadening function with an intrinsinc Lorentzian spinpacket lineshape (M.K.B. and R.L.T., unpublished work). The inhomogeneous broadening function is constant for each sample while the Lorentzian width is temperature dependent. This observation supports the model in which the ESR signals arise from mobile spin species rather than from delocalized states and in which spin motion is changing the spin packet width. The inhomogeneous contribution to the linewidth arises from the finite domain size, and it can be used to estimate the domain size. In the highly narrowed samples exhibiting g-factor anisotropy, the inhomogeneous broadening is 4.7 MHz (M.K.B. and R.L.T., unpublished work). This value is a factor of n /2 smaller than the value for

Proc. NaMl. Acad. Sci. USA 85 (1988)

1501

the Chlat cation of 65.27 MHz,* where n is the number of molecules in the one-dimensional domain (7). Thus, onedimensional electron transport occurs over domains with an average size of about 200 molecules. An axial g factor with a domain size of 200 molecules has important implications for the structure of the Chla/water micelles. For each Chlat cation, the ESR line center is shifted slightly, due to its anisotropic g factor, by an amount that depends on the orientation of the macrocycle relative to the ESR spectrometer magnetic field. The motion of the spin can dynamically average the g-factor shifts in a manner that depends on the path of the spin. If the electron transport is rapid about the circumference of the micelle, there will be only two distinguishable directions for the electron spins: parallel to the micellar axis and perpendicular to it. Electron transport parallel to the micellar axis will produce little averaging of the g factor and will yield three distinguishable directions. The observed g factor has axial symmetry and only two distinguishable directions. We conclude that the electron-transport pathway most likely follows a low-pitch helix on the surface of the micelle, yielding five turns on the circumference of the micelle for a domain size of 200 molecules. The axial g factor is also good evidence that the electron transport is predominantly one-dimensional. Since only 200 molecules are visited by the electron spin in the samples exhibiting resolved g-factor anisotropy, those 200 molecules must uniformly cover a region encircling the micelle. Given the dimensions of the micelle, this is a band 5 molecules wide and 40 molecules in circumference. The electron transport is limited to a tiny band on the surface of a much larger micelle. An alternative to one-dimensional electron transport is twodimensional transport in surface domains of the necessary size and shape. However, it is difficult to explain why such two-dimensional domains exist and why electron transfer is as efficient between chains as along a chain. The outline of a two-dimensional domain must consist of a string of defects preventing electron transport out of the domain and it is difficult to see why such a defect string should arise. On the other hand, a few scattered defects will break a onedimensional conductor into a series of finite domains. The picture that emerges from these ESR studies of Chla/water micelles is fairly detailed and remarkably consistent with the SANS structure. The ESR signals in oxidized or reduced Chla/water micelles arise from highly mobile electrons or holes, which hop from macrocycle to macrocycle on the surface of the micelles. Our data rule out closed rings and argue strongly for a shallow helical arrangement. The electron or hole transport is very nearly onedimensional in nature, proceeding almost exclusively along a chain of chlorophyll molecules wound in a helical fashion to form the walls of the micelles. Impurities, such as Chlb, and structural defects in the micelle wall act as boundaries to electron transport and confine the electrons to finite chains of Chla. The one-dimensional electron transport within the finite chains produces the inhomogeneously broadened ESR lineshapes we observe. The ESR signals in Chla/water micelles are much narrower than we have observed in micelles of any other chlorophyll. The narrowing implies rapid electron transfer over large one-dimensional domains of macrocycles in the Chla/water micelles. Both the rapid electron motion and the size of the domains depend on the equivalence of the macrocycles in the micellar structure. Both our SANS and ESR observations of a variety of chlorophyll/water micelles and chemical considerations lead us to conjecture that no

tThis value is derived from the second moment (w2) = 0.216 mT, as 65.27 MHz = 2-(w 2)1/2-(28 MHz/mT).(21r)1/2.

1502

Biophysics: Bowman et al.

chlorophyll possessing more than one strong nucleophilic group forms aggregates with water in which there are extensive domains of electronically equivalent macrocycles such as are present in Chla/water micelles. Bacteriochlorophylls a and b, which possess an auxiliary nucleophile in ring I, form micelles with water that have sharply restricted domains of equivalent macrocycles over which spin hopping or sharing can occur, as is evident from their broader ESR signals ( M.K.B., T.J.M., and J.J.K., unpublished data). The same is also true for micelles of 2-desvinyl-2-acetyl-Chla, a synthetic derivative that contains a supplementary nucleophilic group in ring I. We have already noted that the addition of small amounts of Chlb to a Chla/water micelle reduces the average size of the domains over which charges are mobile. This may, in fact, be an important function of Chlb in green plants. The presence of auxiliary nucleophiles increases the variety of coordination and hydrogen-bonding interactions and allows arrangements with differing 7r-ir overlaps, thus producing statistical

mixtures of orientations that break the micelle into very short domains of "equivalent" macrocycles. While the presence of additional or supplementary nucleophiles appears to prevent long-range electron transport, additional nucleophilic groups can provide an efficient mechanism for channeling electrons or electronic excitations to traps located at points along the micelle surface by fine-tuning energy levels and transfer rates along the chain.

Proc. Natl. Acad. Sci. USA 85 (1988) This work was supported by the Office of Basic Energy Sciences, Division of Chemical Sciences, U.S. Department of Energy, under contract W-31-109-Eng-38. 1. Katz, J. J. & Ballschmiter, K. (1968) Angew. Chem. 80, 283-284. 2. Jacobs, E. E., Vatter, A. E. & Holt, A. S. (1954) Arch. Biochem. Biophys. 53, 228-238. 3. Katz, J. J., Ballschmiter, K., Garcia-Morin, M., Strain, H. H. & Uphaus, R. A. (1968) Proc. Natl. Acad. Sci. USA 60, 100-107. 4. Garcia-Morin, M., Uphaus, R. A., Norris, J. R. & Katz, J. J. (1969) J. Phys. Chem. 73, 1066-1070. 5. Katz, J. J. (1973) in Inorganic Biochemistry, ed. Eichhorn, G. (Elsevier, Amsterdam), Vol II, Chap. 29, pp. 1022-1066. 6. Katz, J. J. & Norris, J. R. (1973) Curr. Top. Bioenerg. 5, 41-75. 7. Norris, J. R., Uphaus, R. A., Crespi, H. L. & Katz, J. J. (1971) Proc. NatI. Acad. Sci. USA 68, 625-628. 8. Worcester, D. L., Michalski, T. J. & Katz, J. J. (1986) Proc. Natl. Acad. Sci. USA 83, 3791-3795. 9. Norris, J. R., Thurnauer, M. C. & Bowman, M. K. (1980) in Advances in Biological and Medical Physics, eds. Lawrence, J. H., Gofman, J. W. & Hayes, T. L. (Academic, New York), Vol. 17, pp. 365-415. 10. Salikhov, K. M., Semenov, A. G. & Tsvetkov, Y. D. (1976) Electron Spin Echoes and Their Applications (Nauka, Novo-

sibirsk, U.S.S.R.). 11. Thomann, H. & Dalton, L. R. (1986) in Handbook on Conducting Polymers, ed. Skotheim, T. A. (Dekker, New York). 12. Richards, P. P. (1974) Phys. Rev. B 10, 805-813.