Physics Today The Primary Steps of PhotosynthesisGraham R. Fleming and Rienk van Grondelle Citation: Physics Today 47(2), 48 (1994); doi: 10.1063/1.881413 View online: http://dx.doi.org/10.1063/1.881413 View Table of Contents: http://scitation.aip.org/content/aip/magazine/physicstoday/47/2?ver=pdfcov Published by the AIP Publishing

[[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

THE PRIMARY STEPSOF PHOTOSYNTHESIS

The two important initial steps of photosynthesis—electron transferand energy transfer—occur with great speed and efficiency. Newtechniques in laser optics and genetic engineering are helping us tounderstand why.

Graham R. Fleming and Rienk van Grondelle

Photosynthesis, the process by which plants convert solarenergy into chemical energy, results in about 10 billiontons of carbon entering the biosphere annually as carbo-hydrate—equivalent to about eight times mankind's en-ergy consumption in 1990. The apparatus used by plantsto perform this conversion is both complex and highlyefficient. Two initial steps of photosynthesis—energytransfer and electron transfer—are essential to its effi-ciency: Molecules of the light-harvesting system transferelectronic excitation energy to special chlorophyll mole-cules, whose role is to initiate the directional transfer ofelectrons across a biological membrane; the electrontransfer, which takes place in a pigment-protein complexcalled the reaction center, then creates a potential differ-ence that drives the subsequent biochemical reactionsthat store the energy. (Higher plants use two differentreaction centers, called photosystems I and II, whilepurple bacteria make do with a single reaction center.The difference is that the bacteria do not generate oxygenin the photosynthetic process.) Both the elementaryenergy transfer and the primary electron transfer areultrafast (occurring between 10"13 and 10"12 seconds),leading to the trapping of excitation energy at the reactioncenter (on a 100-picosecond timescale) and subsequentelectron transfer in about 3 picoseconds with almost 100%quantum yield.

In this article, we describe current theoretical andexperimental efforts aimed at understanding photosyn-thetic electron and energy transfer. Quantum dynam-ics—particularly the subtle interplay between timescalesset by electronic coupling between molecules, their inter-nal nuclear motions and the coupling of the system tothe dissipative bath (in this case, the protein scaffold-Graham Fleming is a professor of chemistry at the Universityof Chicago. Rienk van Grondelle is a professor of physics atthe Free University of Amsterdam, in The Netherlands.

ing)—is emerging as a key factor in understanding theefficiency and rapidity of these two processes. Such anunderstanding has implications for fields ranging fromthe design of herbicides to the construction of molecularelectronic devices.

Electron transferIn 1989 Hans Deisenhofer and Hartmut Michel deter-mined the molecular structure of a bacterium's photosyn-thetic reaction center to atomic resolution (figure I).1This remarkable achievement spurred great interest inthe mechanism of the primary electron-transfer step inphotosynthetic bacteria. Developments in ultrafast la-sers allow experimenters to observe electron transfer inreal time; genetic engineering allows controlled changesto be made in the protein structure; and advances intheory and computer simulation have brought into focusmany challenging questions, some of which we hope toillustrate—but not resolve—in this article.

Pioneering work by P. Leslie Dutton, Peter Rentzepis,William Parson and Maurice Windsor in 1975 establishedthe sequence of events in the reaction center: Followingoptical excitation of the "special pair," labelled P in figurelb, charge separation occurs within about 3 psec to givethe oxidized special pair P+ and a pheophytin anion HA~.The electron then hops to a quinone molecule QA in about200 psec and on to a second quinone QB in about 100jusec. This entire sequence is repeated (after the specialpair has been restored to neutrality by a cytochromemolecule), and the QB molecule leaves the protein as QBH2to take part in the chemical reactions that lead to thegeneration of an electrochemical potential gradient acrossthe biological membrane in which the complex sits. Even-tually the two electrons of the reduced quinone moleculeare returned to the cytochrome, leading to the transportof four protons across the membrane. The cumulativeenergy stored in the electrochemical gradient is sufficient

48 PHYSICS TODAY FEDRUARY 1994 1994 American Insrirure of Physics [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

Structure of the reaction center protein of the purplebacterium Rhodobacter sphaeroides. a: The complexconsists of three protein subunits called M, L and H (shownin red, green and blue, respectively), with the moleculesresponsible for electron transfer indicated in black. Thecomplex spans the biological membrane, b: The activeconstituents of the reaction centers (shown here with theprotein scaffold stripped away) are the special pair P (blue),the two bacteriochlorophyll molecules B (green), the twopheophytin molecules H (pink) and the two quinonemolecules Q (orange). The reaction center is made up ofactive branch A and the inactive branch B. Two aminoacids at sites L181 and M208 have been modified bygenetic manipulation; in the natural (wild-type) system theyare phenylalanine and tryptophan, respectively. The rates ofthe forward electron-transfer steps in the wild-type systemare indicated by arrows. (Courtesy of James R. Norris,Chong-Hwan Chang, Ossama El-Kabbani, David Teide andMarianne Schiffer, Argonne National Laboratory.) Figure 1

Inactive branchB

Active branchA

Q.

for green plants and bacteria to synthesize adenosinetriphosphate and other molecules used as energy sourcesby living organisms. If unused, the charge returns fromQB to P within a few seconds.

There are many remarkable aspects about the proc-ess just described. The quantum yield of electron transferis close to 100%, and about 40% of the input photonenergy is stored in the transmembrane charge separation.As figure lb shows, there appear to be two equally goodpaths for the electron to travel, yet no evidence has beenfound for electron transfer along the inactive branch(labelled B in the figure). Experiments by David Tiedehave shown that the active branch A is favored by atleast 200:l.2 The center-to-center distance between Pand HA (or HB) is 17 A, and the observed electron-transferrate between the two molecules is at least 1000 timesfaster than the expected rate over that distance in avacuum. Naturally, these findings have focused attention

on the role played in electron transfer by the bacterio-chlorophyll molecule BA and the protein scaffolding, aswe shall discuss below. Another striking finding is thatthe primary electron-transfer step, far from being ther-mally activated, actually speeds up slightly as the tem-perature is lowered from 300 to 10 K. Furthermore, theforward and backward electron-transfer rates differ bymany orders of magnitude. For example, if QB is re-moved, electron recombination from QA to P takes 100msec, six orders of magnitude slower than the forwardprocess P -» QA (via HA).

Electronic statesA look at the reaction center's active constituents,stripped of their protein scaffold, raises the question ofhow to describe the electronic states of such a system:Are they localized on a single molecule or delocalized overseveral? Almost 20 years before the x-ray work of Deis-enhofer and Michel, James Norris and Joseph J. Katzconcluded from electron paramagnetic resonance datathat the primary donor cation was indeed a dimer, a pairof molecules. Steven Boxer, Gerald Small and theircoworkers have used hole-burning spectroscopy at 1.5 K(see the article by Dietrich Haarer and Robert Silbey,PHYSICS TODAY, May 1990, page 58) to probe the electronicstates of the reaction center and their interactions withthe protein.3 The special pair is indeed "special" in itsspectroscopy, showing a strong coupling of the opticalexcitation to low-frequency vibrational modes, which isnot observed in the spectra of monomeric chlorophylls.Resonance Raman spectroscopy by Boxer, RichardMathies and their coworkers reveals that these modesare likely intramolecular in origin.4 It is, of course,tempting to speculate that the nuclear motions coupledto optical excitation also mediate electron transfer, butno one has yet confirmed this hypothesis.

Small has analyzed nonphotochemical hole burning

PHYSICS TODAY FEBRUARY 1994 4 9 [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

2 (P+BA"HA)

)?AG,

3 (P+BAH

Electron transfer. Left: Free-energy surfacesfor the three states believed to be involved inthe primary electron-transfer step from thespecial pair. The location of state 2 (P+BA~HA)is not yet known: It may lie above or belowstate 1 (P*BAHA). The free-energy gap AG Bbetween states 1 and 3 (P+BAFV) isapproximately 2000 cm"1. Right: Thediagrams illustrate the distinction between thesequential and superexchange mechanismsfor electron transfer. The left and rightnumbers designate the ket and bra of states 1,2 or 3. The angle brackets denote ensembleaveraging; the sequential process involvestwo uncorrelated steps, whereas thesuperexchange process is fully coherent.Figure 2

using an effective Hamiltonian, proposed by YoungdoWon and Richard Friesner,5 in which all six pigmentsare coupled excitonically. This model implies that onlythe pheophytin states can be considered monomeric.However, many fundamental aspects of the quantumdynamics in such a complex system remain unclear.

The most basic (and certainly the most controversial)question concerns the two possible mechanisms for the

electron transfer from P to HA. One possible mechanismis a sequential transfer process: P*BAHA -» P+BAHA~ -»P+BAHA". The second is a virtual process (called super-exchange) in which BA acts to mix the electronic statesof P and HA. This mechanism must operate if the stateP+BA"HA is higher in energy than P+BAHA~, and coherencemust be maintained until the final population state isreached. (See figure 2.) The competition between the twoproposed mechanisms depends on both the energetics andthe interaction of the system with its environment (theprotein).

Rudolph Marcus, Shaul Mukamel, Julian Joseph andWilliam Bialek,6 and Chi Mak have advanced the theo-retical description of virtual electron transfer in con-densed phases. The interest in these two mechanismshas stimulated development of the spin-boson model withthree electronic states. Detailed analysis suggests thatthe spectral features of the intermediate state might beobservable in the fully virtual process, which means thatthe sequential and superexchange mechanisms may bedifficult to distinguish experimentally.

Other unresolved issues concern the influence oftemperature on the ratio of coherent and incoherenttransfer and on the expected form of the initial populationdecay. Much remains to be learned about the interplaybetween coupling to the environment and the coherentand incoherent processes in three-state systems. Mor-dechai Bixon, Joshua Jortner and Maibe Michel-Byerleoffer an intriguing speculation on why the plant maybother with both mechanisms: They argue that thecoexistence of superexchange and sequential mechanismsallows for efficient electron transfer over a broad range

Glycerol/buffer glass Fourier transformfor 945-nm probe

100 200 300 400FREQUENCY (cm"1)

Oscillatory parts

1 2TIME (psec)

1 2TIME (psec)

Decay of the excited state ofthe special pair of the DLLmutant produced by stimulatedemission from a probe beamshows that no electron transferoccurs in this mutant. Theoscillations result fromvibrational wavepackets createdby the excitation pulse. Becauseof the coherent vibrationalmotion, the probability forinteraction with the probe lightshows a periodic modulationthat depends on the wavelengthof the probe light. The Fouriertransformation of the oscillatorycomponents reveals frequenciesat 15 cm"1 and 77 cm"'.(Adapted from ref. 14) Figure 3

50 PHYSICS TODAY FEBRUARY 1994 [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

of free-energy gaps.7 Thus the element of redundancyresults in a system that is less susceptible to environ-mental perturbations.

Computer simulations based on classical mechanicshave provided important insights into the functioning ofthe reaction center. Klaus Schulten, Arieh Warshel,William Parson, David Chandler and their coworkers allhave carried out largescale simulations.8 These calcula-tions show that the response of the protein to the changeseparation is extremely rapid (50—100 fsec) even at lowtemperature. Chandler's group has shown that al-though the free-energy surfaces are harmonic, the pro-tein response does appear to show detectable nonlinearbehavior.

Unfortunately, findings related to the energy levelsof the relevant states are less clear cut: Warshel, Parsonand coworkers find that the states P*BAHA and P+BA~HAare very similar in energy, whereas Chandler and co-workers find that P+BA HA is significantly higher inenergy than P*BAHA, which would point to the coherentprocess as the only viable mechanism. Chandler's simu-lations further imply that electron transfer along the"inactive" branch to HB is only slightly endothermic,whereas Warshel and Parson find that P+BB~HB liessignificantly above P*BH, which would allow one to useeither the sequential or the superexchange mechanismto rationalize the lack of electron transfer along theinactive branch.

And the experimenters have fared no better at reach-ing a consensus: Wolfgang Zinth concludes that a two-step mechanism is consistent with his data,9 whileChristine Kirmaier and Dewey Holten find no evidencefor an intermediate state and so favor the superexchangeprocess.10

Experiments by Boxer and coworkers exploit thesensitivity of electronic absorption bands to external elec-tric fields to probe the effective dielectric constant in thevicinity of the chromophores. Their results support theargument that intermediates such as P+BA~ and P+HA~are lower in energy than P+BB~ and P+HB~. Boxer's groupsuggests that the dielectric constant is higher along thefunctional, or active, pathway and argues that this de-termines the directionality of electron transfer. The ef-fect appears to be a collective one involving many aminoacids, so that changing a small number of amino acidswould be unlikely to alter the direction of electron flow.

Mutant reaction centersThe methods of genetic engineering, which allow thecontrolled modification of specific amino acids, make thereaction center a laboratory for studying the protein'srole in electron transfer. Douglas Youvan and coworkersswapped an entire transmembrane helix on the activebranch with the homologous helix from the inactivebranch, which contains 13 different amino acids. Theresulting mutant (known as DLL) is missing the pheo-phytin HA. In a remarkable demonstration of the elec-tron's reluctance to proceed down the inactive branch,

-50 50 100 150 200TIME (psec)

250 300 350

Trapping dynamics in photosynthesis of amutant of the photosynthetic purplebacterium Rhodobacter sphaeroicles; themutant lacks the peripheral light-harvestingantenna. This process can be studied usinglow-intensity, single-wavelength picosecondpump-probe measurements. Overlappingabsorption bands and relaxation processes inthe antenna give rise to complicateddynamics: At 860 ran there is a very smalltransmission increase followed by a strongabsorption increase due to the excited stateabsorption of the antenna, which decays,roughly monoexponentially, in 50 psec to astate with a residual transmission increase.The 50-psec decay time represents theaverage time for an excitation to reach thereaction center and initiate electron transfer.This final transmission increase results from aspectral change in the reaction center whenthe charge-separated state is formed.(Adapted from ref. 1 7) Figure 4

Jean-Louis Martin and Jacques Breton showed that inDLL, the electron refuses to go down either branch.11 Theexcited state of the special pair simply relaxes back toits ground state, and the quantum yield of electrontransfer is zero. This mutant, however, has proved ex-tremely valuable for studies of vibrational coherence inthe reaction center, as described below.

Graham Fleming's group (Chi-Kin Chan, TheodoreDiMagno and Yi-Wei Jia), in collaboration with JamesNorris, Deborah Hansen, Marrianne Schiffer and theircoworkers, have investigated the role of two particularamino acids, L181 and M208 (figure lb).12 In the naturalsystem, M208 (on the active branch) is a tyrosine residueand L181 (on the inactive branch) a phenylalanine. Na-ively supposing that the tyrosine might control the direc-tionality of the electron transfer, we chose to reverse thelocations of the two amino acids. We found that theelectron-transfer rate is essentially unchanged by thismodification, still proceeding along the active branch.Even more surprisingly, the electron-transfer rate in-creased slightly when "symmetry" was restored to thereaction center by making both L181 and M208 tyrosineresidues—in spite of the fact that this change modifiedonly the "inactive" side. Studies of a set of ten differentmutations on these two sites led us to conclude that theamino acids in these positions affect the redox potential

PHYSICS TODAY FEBRUARY 1994 5 1 [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

Light harvesting system in a photosynthetic bacterium. Theproposed arrangement of the a/3-heterodimer shows twobacteriochlorophyll molecules (green) and themembrane-spanning helices (blue). The tryptophan residues(red) significantly influence the position of the absorptionspectrum of the bacteriochlorophyll molecules. (Courtesy ofNeil Hunter, Sheffield University, UK.) Figure 5

of the special pair—that is, the difference in free energybetween P and P+. The dependence of the electron-trans-fer rate on the redox potential of the special pair impliesthat changes in the protein structure as a result ofelectron transfer (and hence the reorganization energy)are extremely small. Nuclear magnetic resonance studiesby Huub de Groot and coworkers also suggest that thereaction center protein is extraordinarily rigid. In fact,the electron-transfer theory of Rudolph Marcus (see PHYS-ICS TODAY, January 1993, page 20) requires a rigid non-polar protein if the electron transfer is to proceed at anoptimal rate, given the very low free-energy drop betweenP*BAHA and P+BAHA". If the protein reorganized sub-stantially as a result of the electron transfer, the transferwould be slow and strongly thermally activated.

It is now possible for experimenters to generate lightpulses that are short compared to the timescales ofvibrational motion in molecules. This allows abrupt op-tical excitation of electronic transitions in which vibra-tional wavepackets are generated in both ground andexcited electronic states. The subsequent motion of theexcited state and ground state wavepackets imposes pe-riodic modulation on the transmission of a probe beam,permitting one to probe the nuclear motion in real time.Ahmed Zewail and coworkers have used this techniqueextensively in the gas phase, and Fleming, SandfordRuhman and others have used it to study small moleculesin solution. It was a considerable surprise, then, whenMartin, Breton and their collaborators first reported ob-serving coherent nuclear motion in such a large, multi-dimensional system as the reaction center.13 Figure 3shows their results for the stimulated emission of thespecial pair in the DLL mutant.

Observation of an oscillatory contribution to the sig-

nal calls into question the conventional assumption thatvibrational dephasing and relaxation occur on muchshorter timescales than does the electron-transfer step.Jose Onuchic and coworkers have suggested14 that theprimary electron transfer occurs in a different regimefrom the traditional nonadiabatic regime conventionallydescribed by the Golden Rule. (See the article by Flemingand Peter Wolynes, PHYSICS TODAY, May 1990, page 36.)They propose that the electronic mixing between theinitial and final states is of the same order of magnitudeas the width (due to relaxation and dephasing) of thefinal states, thus making the electron-transfer rate com-petitive with relaxation in the final state. They showthat in this regime the rate is rather insensitive tochanges in the system parameters (free-energy gaps,coupling strengths, relaxation rates or vibrational fre-quencies).

John Jean, Richard Friesner and Fleming also inves-tigated coherence effects and the influence of slow energyflow between the reaction coordinate and its surround-ings. They quantified the breakdown of the Golden Rulethat results from slow relaxation and quantum interfer-ence effects. A full description of these effects, incorpo-rating a realistic model of the system-bath interaction(for example, at the so-called spin-boson level), is aformidable problem and remains an area of active theo-retical development.

Energy transferThe reaction centers of bacteria and plants are highlyoptimized devices that leave little room for improvement.However, on their own they would be of limited signifi-cance to the plant or bacterium, because the solar energyflux is too low to justify the investment a cell must maketo synthesize this complex equipment. In nature, reac-tion centers are surrounded by a light-harvesting system,generally consisting of chlorophyll and carotenoid mole-cules complexed to proteins. The light-harvesting systemallows the cell to greatly improve the absorption crosssection of each reaction center and make optimal use ofits energy-converting capacity. The functions of the light-

5 2 PHYSICS TODAY FEBRUARY 1994 [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

Transmembrane helices and chlorophyllmolecules in photosystem I. The green discsrepresent the light-harvesting chlorophyllmolecules surrounding the electron-transfercomponents, shown in yellow. Only abouthalf of the total number of chlorophyllmolecules have been located to date, whichmeans that the complete structure must beextremely densely packed with antennachlorophyll molecules. Once an antennamolecule is excited, the probability of theexcitation reaching the primary electrondonor (a pair of chlorophyll moleculesknown as P700) is at least 99%. (From ref.22.) Figure 6

harvesting system are to absorb light over a broad rangeof wavelengths and to transfer the excitation energy tothe special pair of the reaction center. The light-harvest-ing system typically contains between 24 and 300 (andin some exceptional cases more than 1000) pigments perreaction center. The size of the light harvesting system,often called the antenna, implies that the excitation mustvisit many molecules before finally being trapped at thereaction center, and thus individual transfer steps mustbe highly optimized.

Intermolecular energy transfer is generally describedby the Forster dipole-dipole resonance mechanism. Therate of energy transfer scales with the inverse of the sixthpower of the distance between pigments, along with afactor that accounts for the orientation of the dipoles inspace and another factor that measures the overlap be-tween the emission spectrum of the excitation donor andthe absorption spectrum of the excitation acceptor. Fora pair of chlorophyll a molecules separated by a distanceof 15 A (not unlike the values found in nature), the rateof energy transfer may be faster than 1012 per second.For these short distances and fast rates, the Forsterequation has only limited applicability. The point-dipoleapproximation is likely to be inadequate and, as in thecase of electron transfer, the excitation transfer possiblyoccurs from a vibrationally unrelaxed state. Additionally,in densely packed chlorophyll protein systems, excitonicinteractions between the pigments leads to a distinctsplitting of the excited state energies. In this case it ismore appropriate to describe the short-time portion ofthe energy transfer as a relaxation between the differentexciton levels of the system. Hole-burning studies havesuggested that these times can be 100 fsec or less.15

Light trappingBecause the excited state of an individual chlorophyllmolecule lives for about 10~9 seconds, trapping of theexcitation energy must occur in less than 100 picosecondsto produce a quantum yield greater than 90%. ArviFreiberg and coworkers (using fluorescence techniques)and Villy Sundstrom, Rienk van Grondelle and coworkers

(using absorption methods) have made direct measure-ments of the excitation trapping time in photosyntheticpurple bacterium, which have a relatively simple light-harvesting antenna. These measurements showed thatfor most of the systems studied, the trapping time isabout 50 picoseconds.16'17 Figure 4 shows a typical ex-ample of the kinetics of excitation trapping, measured ina bacterium consisting of only the reaction center sur-rounded by a ring of antenna pigments (the core antenna)absorbing, like the special pair, near 870-900 nanome-ters. Others have found the timescale of excitation trap-ping in other, more complex photosynthetic systems tobe rather similar: Fleming's group measured a trappingtime of about 30-40 psec for the core antenna of photo-system I; Alfred Holzwarth and coworkers measured an80-psec trapping time for the core antenna of photosystemII.18 Based on these findings, along with the knownnumbers of light-harvesting pigments per reaction centerand their spectral distribution and using a theoreticalexpression derived by Robert Pearlstein, the estimatedsingle-step energy-transfer time in these light-harvestingantennas is about 200 fsec or less. Analyzing the effi-ciency of biexcitonic annihilation processes following ex-citation with an intense laser pulse, van Grondelle andcoworkers and Leonas Valkunas obtained a similar esti-mate for the single-step energy-transfer time.19

No high-resolution crystal structure is available forthe core light-harvesting antenna of the photosyntheticbacterium. However, through a comparison of the aminoacid sequences of the polypeptides binding the bacterio-chlorophylls, invaluable structural information is nowavailable. Herbert Zuber and coworkers discovered thatin all photosynthetic purple bacteria, the light-harvestingchlorophyll molecules are noncovalently bound to a pairof small polypeptides called a and (3. Zuber suggestedthat the a/3BChl2 heterodimer is the basic structuralelement for the bacterium's light-harvesting antenna.Although the precise structure of the unit is not known,biochemical, spectroscopic and electron microscopy dataprovide a detailed view of the arrangement of the twopolypeptides and the bacteriochlorophyll molecules bound

PHYSICS TODAY FEBRUARY 1994 5 3 [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

to them. Figure 5 depicts the structure of the light-har-vesting unit. In vivo, heterodimer units assemble intoring-like aggregates that surround the reaction centers.

One phenomenon that so far remains unexplained isthe dramatic shift in color that the pigments experienceupon assembly of the light-harvesting antenna. Freebacteriochlorophyll in an organic solvent absorbs near770 nm, while the intact light-harvesting antenna maybe red-shifted to 870-900 nm. Although excitonic inter-actions between the quasidegenerate excited states con-tribute to the red shift, a large part of the shift is notexcitonic, but is probably due to specific electronicchanges in the immediate environment of the pigments.

Similar concepts apply for many other light-harvest-ing pigment proteins. As early as 1975 Roger Fenna andBrian Matthews crystallized the BChl a complex of agreen photosynthetjc bacterium,20 and today we know itsstructure to a 1.9-A resolution. The complex consists ofa trimer of identical subunits assembled with C3 (three-fold rotational) symmetry; each subunit contains sevenBChl a molecules arranged in a rather nonsymmetricfashion. The absorption properties of the complex arecharacteristic of strong excitonic coupling. The BChl amolecules within each monomer are excitonically coupled,with interaction energies up to 200 cm"1; the distancesbetween BChl a molecules in adjacent subunits are suf-ficiently short to give rise to excitonic interactions on theorder of 20 cm"1. Hole-burning spectroscopy by Small15

and singlet-triplet spectroscopy by van Mourik21 haveshown that all the absorption bands of the complex areindeed coupled. Small identified at least eight excitoniccomponents and concluded that the excitonic splittingexceeds by far the distribution of aborbance wavelengthsof the various sites (inhomogeneous broadening). More-over, both Small and van Mourik were able to distinguishin the lowest-energy absorption band of the complex,excitonic components parallel and perpendicular to theC3-axis. On the basis of these results, Small and cowork-ers have suggested that in this complex, the phonon-in-duced scattering between various exciton levels is thedominant mechanism for energy transfer.

Structural information is beginning to appear for thephotosynthetic proteins of plant systems. Photosystem Iof plants and cyanobacteria consists of an antenna sys-tem, which contains about 100 chlorophyll moleculescombined with the reaction center.22 (See figure 6.)Horst Witt, Wolfram Saenger and coworkers have located45 chlorophyll molecules in the 6-A-resolution structure.Among those is the primary electron donor, a chlorophyll

Q_

oO

OO

0.3-

0.2-

0.1-

-

•

1400-

1200-

1000-

800-

600"

400"

200-

0-

'y

\ -

. . . . . . . .

\fifjj

If-0.6 0.2 0.6

TIME (psec)1.4

Femtosecond spontaneous fluorescencesignals observed from the LHCII antennasystem of green plants. Energy transferbetween differently oriented chlorophyllmolecules leads to depolarization of thefluorescence. The bottom panel shows theparallel (upper) and perpendicular (lower)decay data from which the fluorescenceanisotropy (top panel) is constructed. Theanisotropy clearly contains both ultrafast(250-fsec) and slower (11-psec) decaycomponents. The 250-fsec decay mayrepresent energy transfer between nearestneighbors, while the longer timescalerepresents transfer over longer distances.Figure 7

dimer known as P700.Werner Kiihlbrandt has used electron microscopy to

obtain the structure of the major light-harvesting complexassociated with the oxygen-evolving photosystem II ofplants.23 Known as LHCII, this complex binds about halfthe chlorophyll on Earth and is composed of trimericassemblies of monomer units, each containing 14±1 chlo-rophyll molecules: 8+1 Chi cr's and 5±1 Chi b's. Withinthe monomer the chlorophylls are organized in two layers,and the center-to-center distances between the chloro-phylls fall in a narrow range of 9-14 A. From structuralstudies it is evident that, as for the green bacterium,excitonic interactions between pigments have a decisiveinfluence on the spectral properties of LHCII.

All the structural information available to date isconsistent with ultrafast energy transfer within chloro-phyll protein complexes. Recently, advances in lasertechnology have allowed researchers to monitor directlythe early energy-transfer events. By measuring time-dependent fluorescence depolarization, Mei Du, Sunney Xieand Yiwei Jia in Fleming's lab observed the elementary

54 PHYSICS TODAY FEBRUARY 1994 [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12

-0.10820 840 860 880 900

WAVELENGTH (nm)920 940

Transient absorption spectra of thelight-harvesting antenna of the photosyntheticbacterium Rs. Rubrum. The spectra arerecorded at 0 fsec (black), 166 fsec (green),333 fsec (purple), 500 fsec (blue), 1166 fsec(yellow) and 1833 fsec (red). Following therise in the first four spectra, resulting from theapparatus response time, the spectra shift toprogressively longer wavelengths as theexcitation begins to approach a Boltzmanndistribution among the excited states. Similarshifts are seen in the stimulated emission(positive optical density change, plottednegative). Figure 8

timescale of energy transfer for the photosystem I an-tenna and the major light-harvesting complex of plants,LHCII.24 In both cases, depolarization occurs on a times-cale of 150-300 fsec. Figure 7 shows some of Du's data:Depolarization of the fluorescence results from excitationsjumping from the initially excited chlorophyll moleculeto other molecules with different spatial orientations. Inthe LHCII system further depolarization is observed on a5-psec timescale, perhaps corresponding to energy trans-fer between different units of the whole assembly. Simi-larly, Matthieu Visser, in van Grondelle's lab, has dem-onstrated recently that equilibration of an initiallycreated distribution of excitations occurs on a timescaleof a few hundred femtoseconds: Figure 8 shows thetime-resolved difference spectra (after a 200-fsec laserpulse) of the light-harvesting antenna of purple bacteria.Effective localization of excitation on the red-most frac-tion of the pigments causes the dynamic red shift of thespectrum and the increase in excited-state absorption.

Thus, energy transfer in the complete photosyntheticapparatus spans a range of three orders of magnitude intime, from perhaps a few tens of femtoseconds up tohundreds of picoseconds. The process may initially in-volve coherent migration, in which the excitation is de-localized over several molecules, while the longer time-scales correspond to incoherent hopping from molecule tomolecule. As for electron transfer, current theories willneed to be extended to describe and interpret that process.

References1. J. Deisenhofer, H. Michel, EMBO Journal 8, 2149 (1989).2. E. C. Kellogg, S. Kolaczkowski, M. R. Waseliewski, D. Tiede,

Photosynth. Res. 72, 47 (1989).3. T. Middendorf, L. Mazzola, D. Gaul, C. Schenck, S. Boxer, J.

Phys. Chem. 95, 10142 (1991). N. Raja, S. Reddy, S. V. Ko-laczkowski, G. J. Small, J. Phys. Chem. 97, 6934 (1993).

4. A. P. Schreve, N. J. Cherepy, S. Franzen, S. G. Boxer, R. A.Mathies, Proc. Natl. Acad. Sci. U. S. A. 88, 11207 (1991).

5. Y. Won, R. A. Friesner, J. Phys. Chem. 92, 2208 (1988).6. R. A. Marcus, R. Almeida, J. Phys. Chem. 91, 2973; 2978

(1990). Y. Hu, S. Mukamel, J. Chem. Phys. 94, 6973 (1989).J. S. Joseph, W. A. Bialek, J. Phys. Chem. 97, 3245 (1993).

7. M. Bixon, J. Jortner, M. E. Michel-Byerle, Biochim. Biophys.Acta 1056,301 (1991).

8. K. Schulten, M. Tesch, Chem. Phys. 158, 421 (1991). A.Warshel, W. W. Parson, Annu. Rev. Phys Chem. 42, 279(1991). N. Marchi, J. N. Gehlen, D. Chandler, M. Newton, J.Amer. Chem. Soc. 115, 4178 (1993).

9. W. Holzapfel, U. Finkele, W. Kaiser, D. Oesterheldt, H. Scheer,H. U. Stilz, W. Zinth, Chem. Phys. Lett. 160, 1 (1989); Proc.Natl. Acad. Sci. 87, 5168 (1990).C. Kirmaier, D. Holten, Biochem. 30, 609 (1991).J. Breton, J.-L. Martin, J. C. Lambry, S. J. Robles, D. C.Youvan, in Reaction Centers of Photosynthetic Bacteria, M. E.Michel-Byerle, ed., Springer-Verlag, New York (1990), p. 293.C. K. Chan, T. J. DiMagno, L. X. Q. Chen, J. R. Norris, G. R.Fleming, Proc. Natl. Acad. Sci. U. S. A. 88, 11202 (1991). 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, 13180(1993).M. H. Vos, F. Rappaport, J.-C. Lambry, J. Breton, J.- L.Martin, Nature 363, 320 (1993).S. S. Skourtis, A. J. da Silva, W. Bialek, J. N. Onuchic, J. Phys.Chem. 96, 8034 (1992). J. M. Jean, R. A. Friesner, G. R.Fleming, J. Chem. Phys. 96, 5827 (1992).S. G. Johnson, G. J. Small, J. Phys. Chem. 95, 471 (1991).A. Y. Borisov, A. Freiberg, V. Godik, K. K. Rebane, K. E.Timpmann, Biochim. Biophys. Acta 807, 221 (1985).

17. V. Sundstrom, R. van Grondelle, H. Bergstrom, E. Akesson,T. Gillbro, Biochim. Biophys. Acta 851, 431 (1986).

18. T. G. Owens, S. P. Webb, L. Mets, R. S. Alberte, G. R. Fleming,Proc. Natl. Acad. Sci. U. S. A. 84, 1532 (1987). G. H. Schatz^H. Brock, A. R. Holzwarth, Proc. Natl. Acad. Sci. USA84 8414(1987).

19. J. G. C. Bakker, R. van Grondelle, W. T. F. den Hollander,Biochim. Biophys. Acta 725, 508 (1983). L. Valkunas, s!Kudzmauskas, V. Liuolia, Sov. Phys. Coll. 26, 1 (1986).

20. R. E. Fenna, B. W. Matthews, Nature 258, 573 (1975).21. F. van Mourik, R. R. Verwijst, J. M. Mulder, R. van Grondelle

J. Lumin. 53, 499(1992).22. N. Krauss, W. Hinrichs, I. Witt, P. Fromme, W. Pritzkow Z

Dauter, C. Betzel, K. S.Wilson, H.T.Witt, W Saenger Nature361,326(1993).W. Kohlbrandt, D. N. Wang, Nature 350, 326 (1991).M. Du, X. Xie, Y. Jia, L. Mets, G. R. Fleming, Chem PhvsLett. 201, 535(1993). ' -

1011

12

13

15

1516

2324

PHYSICS TODAY FEBRUARY 1994 5 5 [[[This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded

to ]]] IP: 128.32.208.2 On: Wed, 13 Nov 2013 23:33:12