Femtosecond time-resolved transient absorption spectroscopy of xanthophylls

Femtosecond Time-Resolved Transient Absorption Spectroscopy of Xanthophylls

Dariusz M. Niedzwiedzki,† James O. Sullivan,† Tomas Polıvka,‡,§ Robert R. Birge,† andHarry A. Frank* ,†

Department of Chemistry, UniVersity of Connecticut, Storrs, Connecticut 06269-3060,Institute of Physical Biology, UniVersity of South Bohemia, NoVe Hrady, Czech Republic, andBiological Centre, Czech Academy of Sciences, Ceske´ BudejoVice, Czech Republic

ReceiVed: April 12, 2006; In Final Form: July 20, 2006

Xanthophylls are a major class of photosynthetic pigments that participate in an adaptation mechanism bywhich higher plants protect themselves from high light stress. In the present work, an ultrafast time-resolvedspectroscopic investigation of all the major xanthophyll pigments from spinach has been performed. Themolecules are zeaxanthin, lutein, violaxanthin, and neoxanthin.â-Carotene was also studied. The experimentaldata reveal the inherent spectral properties and ultrafast dynamics including the S1 state lifetimes of each ofthe pigments. In conjunction with quantum mechanical computations the results address the molecular featuresof xanthophylls that control the formation and decay of the S* state in solution. The findings provide compellingevidence that S* is an excited state with a conformational geometry twisted relative to the ground state. Thedata indicate that S* is formed via a branched pathway from higher excited singlet states and that its yielddepends critically on the presence ofâ-ionylidene rings in the polyene system ofπ-electron conjugated doublebonds. The data are expected to be beneficial to researchers employing ultrafast time-resolved spectroscopicmethods to investigate the mechanisms of both energy transfer and nonphotochemical quenching in higherplant preparations.

Introduction

Xanthophylls are the oxygenated derivatives of carotenes andrepresent a large part of the group of naturally occurringpigments known as carotenoids. Carotenoids are derived fromphotosynthesis and are responsible for the abundance of yellow,orange, and red colors of many biological organisms.1 In higherplants, these molecules play particularly important roles inharvesting light, stabilizing protein structures, regulating energyflow, and dissipating excess energy not required by the organismfor photosynthetic growth.2 If this surplus energy is notdissipated, then deleterious reactions may occur betweenchlorophyll (Chl) and active oxygen species. Especially harmfulis the1∆g state of molecular oxygen that is generated by energytransfer from the Chl triplet state formed by intersystem crossingfrom the photoexcited Chl singlet state.3

Xanthophylls contribute to both short- and long-term adaptivemechanisms of protection of plants against high light stress.One such mechanism is termed nonphotochemical quenching(NPQ). NPQ has several components that work together to bringabout nonradiative dissipation of excess excited singlet statesof Chl that limits the photoinduced damage to the photosyntheticapparatus. The most rapid component of NPQ is denoted qEand is sometimes referred to as high-energy or feedback-regulated quenching.4 For qE to occur, a protein denoted PsbSmust be present, the chloroplast thylakoid lumen must beacidified, and the xanthophyll, violaxanthin, must be enzymati-cally de-epoxidated to zeaxanthin (Figure 1).4

De-epoxidation of violaxanthin to zeaxanthin has a profoundeffect on both the structure of the xanthophylls (Figure 1) and

their excited-state energy levels (Figure 2). Removing theepoxide groups from the terminalâ-ionylidene rings rendersthem more in the plane of the extendedπ-electron system ofconjugated carbon-carbon double bonds. This is evident inFigure 3, which shows different views of the computationally

* Author to whom correspondence should be addressed. Phone: (860)486-2844. Fax: (860) 486-6558. E-mail: [email protected].

† University of Connecticut.‡ University of South Bohemia.§ Czech Academy of Sciences.

Figure 1. Interconversion of violaxanthin and zeaxanthin accordingto the xanthophyll cycle.

Figure 2. Energy level diagram showing possible pathways of energytransfer to and from xanthophylls and chorophylls.

22872 J. Phys. Chem. B2006,110,22872-22885

10.1021/jp0622738 CCC: $33.50 © 2006 American Chemical SocietyPublished on Web 10/25/2006

optimized xanthophyll structures. De-epoxidation also lengthensthe conjugatedπ-electron system (Figures 1 and 3) andsignificantly lowers the energies of the excited states ofzeaxanthin compared to those of violaxanthin.

The structural alterations and the changes in positions of theexcited-state energy levels have been implicated separately indifferent models of how qE functions. The change in xantho-phyll structure (shape) upon conversion of violaxanthin tozeaxanthin has been postulated to increase the aggregation stateof the major light-harvesting pigment-protein complexes as-sociated with photosystem II (PS II) in higher plants.5-8 Chlaggregation is thought to generate low-energy exciton traps thatin turn may induce qE. This model is termed “indirectquenching”.5-8 In an alternative model, a lower-energy excitedS1 state of zeaxanthin compared to violaxanthin may provide adirect route of quenching by energy transfer from the excited-state Chl (Figure 2).9 This “direct quenching” model has beensupported by steady-state fluorescence spectroscopic measure-ments on model polyenes and carotenoids that have shown,either by extrapolation or by direct measurements, that theenergy of the S1 state of zeaxanthin is low enough to potentiallyquench the lowest excited singlet state of Chl, which lies at∼14 600 cm-1.9 However, the various reported values of theS1 state energy of violaxanthin in solution10-12 and in pigment-protein complexes13 range from 13 700( 300 to 15 580( 60

cm-1, indicating that there is still some uncertainty whetherviolaxanthin also has a sufficiently low energy S1 state to enablequenching. Additional evidence from femtosecond time-resolvedspectroscopic investigations carried out on thylakoid membranesfrom spinach andArabidopsis thalianareinforce this model ofzeaxanthin functioning as a direct quencher of Chl singletstates.14 However, the mechanism consistent with the ultrafastspectral and kinetic observations is complex and involvesexcitation transfer from bulk antenna-bound Chl to a specialChl-zeaxanthin heterodimer that undergoes ultrafast (0.1-1 ps)electronic charge separation to form a zeaxanthin cation-Chlanion radical pair (Zea+Chl-) that then recombines nonradia-tively in ∼150 ps.14 It still remains to identify where in the PSII reaction center such a heterodimer may be situated.

An energy level diagram that can be used to describe manyof the spectroscopic and photochemical properties associatedthe xanthophylls is shown in Figure 2. The xanthophyll groundstate, S0, is assigned Ag symmetry in the idealizedC2h pointgroup. This is in keeping with the convention derived from anabundance of studies on model polyenes and carotenoids.15,16

The first excited singlet state is also assigned Ag symmetry anddenoted S1 or 21Ag

-, where the minus sign designates thepseudoparity character of the state derived from orbital pairingrelationships when configuration interaction among singlyexcited configurations is included.17-19 In this notation the

Figure 3. Geometry-optimized structures of the all-trans configurations ofâ-carotene, zeaxanthin, lutein, and violaxanthin. The structure of9′-cis-neoxanthin was obtained from the coordinates of its crystal structure in the LHCIIb pigment protein complex.35

Femtosecond Spectroscopy of Xanthophylls J. Phys. Chem. B, Vol. 110, No. 45, 200622873

ground state S0 is 11Ag-. One-photon transitions between S0

(11Ag-) and S1 (21Ag

-) are forbidden by group theoretical (gT u) and pseudoparity (+ T -) selection rules. The state intowhich one-photon absorption from the ground S0 (11Ag

-) stateis strongly allowed is the 11Bu

+ state. The customary practiceis to denote this state S2, but both theoretical20,21 and experi-mental results (see Polivka and Sundstrom22 for a recent reviewof this topic) are suggestive of excited states lying near or belowthis state. (To avoid confusion regarding state ordering, in thiswork we shall adhere to the customary notation of S0 (11Ag

-),S1 (21Ag

-), and S2 (11Bu+), and when it is necessary to refer to

the positions of the other states, we shall do so using eithertheir symmetry representations or nonnumerical notation.) Inparticular, Koyama et al.23 have assigned spectroscopic featuresto a low-lying 11Bu

- state and have postulated that it providesa route of both deactivation from S2 and energy transfer to Chl.Also, van Grondelle and co-workers24 have invoked a differentexcited state called S*, to account for the ultrafast dynamics ofthe carotenoid, spirilloxanthin, in solution and in the LH1complex fromRhodospirillum rubrum, being different at dif-ferent probe wavelengths, and in the LH1 complex leading totriplet state formation. Initially, S* was thought to be formedonly in the very long (number of conjugated double bonds,N) 13) spirilloxanthin molecule24 and that it provides an alternatepath for the depopulation of S2 (11Bu

+). However, subsequentstudies on spheroidene,25 rhodopin glucoside,26,27 lycopene,28

zeaxanthin,28 andâ-carotene28 have suggested that S* may occurmore commonly.22,29-31

S* has yet to be fully characterized, and there is considerabledebate as to how it is formed.22,24-26,28,32The primary spectro-scopic characteristics of S* are that it is associated with atransition having a maximum in the wavelength region betweenthe S0 f S2 and S1 f Sn absorption bands and that it decays inseveral picoseconds. (In this paper, Sn should be taken to meana generic high-energy excited singlet state having a symmetrythat gives rise to strong allowedness of the transition with whichit is associated.) In light-harvesting pigment-protein complexes,S* is proposed to lead to triplet state formation via ultrafastsinglet-triplet homofisson.24-26 In solution, triplet states areapparently not formed from S*. An alternative view of the natureof the S* state has been published by Wohlleben et al.28 whocarried out pump-deplete-probe and transient absorptionspectroscopic experiments. Upon selective depletion of the S2

state population using a high-power laser pulse they observeda decrease in the intensity of the S1 f Sn transition but no effecton the S* population. On the basis of this observation and theposition and broadness of the S*f Sn transition they arguedthat, in solution, S* is a vibrationally excited, “hot” ground statepopulated by a combination of impulsive Raman scattering ofthe S0 f S2 pump pulse and internal conversion from S1. Theyproposed that the lifetime of S* corresponds to vibrationalrelaxation in the ground state and measured it to be a constant6.2( 0.4 ps for carotenoids havingN g 11 and equal to the S1lifetime for shorter molecules. A time constant in this rangehas been previously implicated in vibrational relaxation in theground state of carotenoids.33 However, one measure ofuncertainty with this assignment is that it implies that S* insolution differs from S* observed in LH2 proteins where it wasshown to serve as a donor state in energy transfer from acarotenoid to BChl a.25

Virtually all of the models proposing to explain howxanthophylls function in the qE component of NPQ have beenderived from observations of the spectroscopic and dynamicbehavior of the molecules. Given the complexity associated with

the spectroscopic properties of xanthophylls and uncertainty intheir system of energy levels, to make compelling assignmentsof their function, it is critical to have a clear understanding ofhow each of these molecules behaves in the ultrafast timeregime. This is particularly important for analyzing the ultrafastspectroscopic observables of xanthophylls present in the mul-ticomponent, spectrally congested, thylakoid and pigment-protein complex preparations from higher plants. In thosesamples it is essential to know precisely where the variousexcited-state transitions occur and how they are contributing tothe spectral and temporal line shapes.

In this paper we present the results of a systematic, ultrafast,time-resolved spectroscopic investigation of all the majorcarotenoid pigments in spinach:â-carotene and the xantho-phylls, zeaxanthin, lutein, violaxanthin, and neoxanthin. The datareveal the inherent spectral properties and ultrafast dynamicsof each of these pigments and address the molecular featuresof xanthophylls that control S* formation in solution. The resultsprovide compelling evidence for the origin of the S* state andwill be of use to researchers employing ultrafast time-resolvedspectroscopic methods to investigate the mechanisms of bothenergy transfer and NPQ in intact thylakoid membranes, isolatedpigment-protein complexes, and whole photosynthetic organ-isms.

Materials and Methods

All xanthophylls except zeaxanthin were extracted fromspinach obtained at a local market. Approximately 10 g of leaveswere ground in 50 mL of acetone/methanol (50/50 v/v technicalgrade), filtered, and dried with a gentle stream of nitrogen gasin the dark at room temperature. The dried pigment extract wasredissolved in 87/10/3 v/v/v acetonitrile (Fisher)/methanol(Fisher)/water (Sigma), filtered, and injected into a MilliporeWaters 600E high-performance liquid chromatography system(HPLC) employing a 3.9 mm× 300 mm Nova-Pak C18 column.The protocol featured a gradient mobile phase of 100% A to100% B in 40 min (A, 87/10/3 v/v/v acetonitrile (Fisher)/methanol (Fisher)/water (Sigma); B, ethyl acetate (Fisher)) witha flow rate of 1 mL/min. Zeaxanthin was obtained from F.Hoffman LaRoche, andâ-carotene was purchased from Sigma.Both molecules were purified using the above protocol. Thepurified pigments from HPLC were dried with a gentle streamof nitrogen gas in the dark at room temperature and stored at-40 °C until use.

Prior to the transient absorption measurements, the moleculeswere dissolved in 99.9% grade pyridine (J.T. Baker) to an opticaldensity (OD) of 0.1-0.3 at the excitation wavelength in a 2mm path length cuvette. Transient absorption spectra were takenusing a femtosecond spectrometer system described in detailpreviously.34 The xanthophylls andâ-carotene were excited intothe lowest energy vibronic band (0-0) associated with theirabsorption spectra in pyridine: 481 nm for neoxanthin, 485 nmfor violaxanthin, 491 nm for lutein, and 497 nm for zeaxanthinandâ-carotene. The excitation energy was typically 1µJ, butthe signals depended slightly on pump energy (see below). Theexcitation beam was focused into a spot of 1.2 mm in diameter,yielding excitation densities in the range of∼2 × 1014 photonspulse-1 cm-2 for the used excitation wavelengths. The excitationand probe beams were overlapped at the sample, and the relativepolarization of the beams was set to the magic angle. Also, apolarizer was placed before the CCD detector to minimizescattered signal from the pump beam. The time resolution(instrument response time) of the spectrometer was obtainedas one of the parameters of the global fitting procedure (Table

22874 J. Phys. Chem. B, Vol. 110, No. 45, 2006 Niedzwiedzki et al.

1). The samples were stirred using a magnetic microstirrer toprotect them from photodegradation. To confirm the sampleintegrity, absorption spectra were taken before and after thetransient absorption experiments at room temperature. Nochanges in the absorption spectra were evident. Surface XplorerPro 1.0.4 (Ultrafast Systems, LLC) software was used to correctfor dispersion in the transient absorption spectra using acorrection curve based on a set of initial times (t0) of the signalproduced from fitting the kinetics at several different wave-lengths. ASUfit 3.0 software provided by Dr. Evaldas Katilius

from Arizona State University was used for global fittingcalculations and for separation of artifacts in the transientabsorption spectra associated with the solvent response within100 fs of excitation.

Carotenoid structures (except 9′-cis-neoxanthin) shown inFigure 3 were constructed using ChemDraw Ultra 5.0 software(CambridgeSoft Corp.) and geometrically optimized usingHyperChem 5.1 (Hypercube, Inc.) software that employs anAM1 semiempirical method with a Polak-Ribiere algorithmin a vacuum environment. The structure of 9′-cis-neoxanthin

TABLE 1: Dynamics of the S1 (τ1), Vibrationally “Hot” S 1 (τ1′), S2 (τ2), and S* (τ3) States ofâ-Carotene, Zeaxanthin, Lutein,Violaxanthin, and Neoxanthina

moleculepumpλ(nm)

probeλ(nm)

τ1

(ps)τ1′(fs)

τ2

(fs)τ3

(ps)τs

b

(fs)fittingmethod solvent reference

â-carotene 497 cont.c 9.5( 0.1 366( 10 170( 2 3.4( 0.2 97 global fit pyridine this work497 594 9.2( 0.3 n.a.d n.a. 2.9( 0.4 n.a. singleλ pyridine this work450 460 7.9( 0.5 n.d.e n.d. n.e.f 4000 singleλ 3-methylpentane 63450 480 8.1( 0.6 n.d. n.d. n.e. 4000 singleλ 3-methylpentane 63450 550 10.0( 0.5 n.d. n.d. n.e. 4000 singleλ 3-methylpentane 63355 556 12.4( 0.5 n.d. n.d. n.e. 50 singleλ n-hexane 64481 cont. 8.9( 0.2 600( 100 180( 10 n.e. 220( 20 global fit n-hexane 50481 570 9.7( 0.4 n.a. 150( 50 n.e. 100( 50 singleλ n-hexane 50476 cont. 9.9 n.d. n.d. n.e. 250 n.d. n-hexane 65504 cont. 8.2( 0.2 n.d. 160( 40 n.e. 300 global fit n-hexane 66397 cont. 9.4( 0.2 n.d. 220( 50 n.e. 300 global fit n-hexane 66504 cont. 8.4( 0.2 n.d. 140( 40 n.e. 300 global fit benzene 66397 cont. 9.1( 0.2 n.d. 250( 50 n.e. 300 global fit benzene 66490 var.g 9.0 500-650 110-260 n.e. 120-160 singleλ n-hexane 52475 cont. 9.1( 0.5 400( 100 140( 30 n.e. 100 global fit hexane 51475 cont. 9.6( 0.5 300( 100 120( 30 n.e. 100 global fit ethanol 51475 cont. 10.7( 0.5 400( 100 130( 30 n.e. 100 global fit benzyl alcohol 51480 540 9.5 n.d. 250 n.e. 200 singleλ ethanol 67480 570 11.0 n.d. 200 n.e. 200 singleλ CS2 67352 n.e. 10( 2 n.d. n.d. n.e. 6000 singleλ var. 68793 cont. 8.7( 0.9 200( 20 163( 9 n.e. 80 singleλ cyclohexane 69794 var. 8.8( 0.2 600 350( 50 n.e. 300 singleλ benzene 70490 484 9.0 210 n.d. n.e. ∼150 singleλ n-hexane 52490 546 8.9 n.d. 260 n.e. ∼150 singleλ n-hexane 52490 580 9.0 650 140 n.e. ∼150 singleλ n-hexane 52490 481 9.0 180 n.d. n.e. ∼150 singleλ methanol 52490 544 9.0 n.d. 200 n.e. ∼150 singleλ methanol 52490 583 9.0 500 110 n.e. ∼150 singleλ methanol 52

zeaxanthin 497 cont. 10.2( 0.2 370( 5 146( 2 2.8( 0.2 130 global fit pyridine this work497 596 10.3( 0.1 n.a. n.a. n.e. n.a. singleλ pyridine this work266 cont. 9.8( 1.0 700( 70 180( 36 4.9( 0.5 120-160 global fit methanol 32400 cont. 9.0( 0.9 350( 40 70( 14 2.8( 0.3 120-160 global fit methanol 32485 cont. 9.2( 0.9 350( 40 135( 27 n.e. 120-160 global fit methanol 32490 var. 8.6 or 8.8 220 or 230 nd. n.e. 200 singleλ methanol 10420 540 9.0 n.d. n.d. n.e. 300 singleλ hexane 9490 555 9.0 n.d. n.d. n.e. ∼150 singleλ methanol 71490 484 9.0 300 n.d n.e. ∼150 singleλ n-hexane 52490 547 9.6 n.d. 270 n.e. ∼150 singleλ n-hexane 52490 577 9.1 800 200 n.e. ∼150 singleλ n-hexane 52490 481 9.0 200 n.d. n.e. ∼150 singleλ methanol 52490 548 9.1 n.d. 280 n.e. ∼150 singleλ methanol 52490 579 9.0 540 120 n.e. ∼150 singleλ methanol 52485 560 9.3 n.d. 290 n.e. ∼100 singleλ ethanol 72

lutein 491 cont. 15.6( 0.1 435( 5 127( 3 2.9( 0.1 121 global fit pyridine this work491 571 15.8( 0.1 n.a. n.a. 2.2( 0.4 n.a. singleλ pyridine this work490 528 15 n.d. n.d. n.e. ∼150 singleλ methanol 71

violaxanthin 485 cont. 26.1( 0.1 582( 6 163( 1 5.0( 0.2 128 global fit pyridine this work485 543 25.5( 0.2 n.a. n.a. 2.0( 0.6 n.a. singleλ pyridine this work420 512 23.9 n.d. n.d. n.e. 300 singleλ n-hexane 9480 var. 24.6 or 25.3 320-380 n.d. n.e. 200 singleλ methanol 10

neoxanthin 481 cont. 37.6( 0.1 444( 5 110( 2 2.7( 0.1 120 global fit pyridine this work481 558 37.5( 0.8 n.a. n.a. 2.7( 0.4 n.a. singleλ pyridine this work467 n.a. 35( 2 n.d. n.d. n.e. 140 singleλ n-hexane 59467 n.a. 35( 2 n.d. n.d. n.e. 140 singleλ methanol 59

a All data were taken at room temperature.b τs is the instrument response time.c White light continuum.d Not applicable.e Not determined.f Not evident.g Various. The fitting of the data was initialized after 1 ps from t0, where component associated with S2 state is greatly diminished.

Femtosecond Spectroscopy of Xanthophylls J. Phys. Chem. B, Vol. 110, No. 45, 200622875

was obtained from its coordinates in the crystal structure in themajor light-harvesting complex (LHCIIb) from spinach.35 Forthe quantum mechanical computations carried out onâ-carotene,the structure was optimized using density functional methodsfor the ground state (B3LYP/6-31G(d)) and ab initio methods(CIS(D) and SAC-CI with a D95 basis set) for the low-lyingexcited singlet states. The excited singlet state calculations werelimited to the 32 highest-energy filled orbitals and the 32 lowest-energy unfilled orbitals, and the SAC-CI calculations werecarried out by selecting integrals using the level one approxima-tion. The spectroscopic properties were then analyzed usingMNDO-PSDCI molecular orbital theory using methods andprocedures described previously.36-38

Results

Room-temperature absorption spectra ofâ-carotene and thexanthophylls in pyridine are shown in Figure 4. Pyridine, whichhas a refractive index ofn ≈ 1.51 at 20°C and a polarizabilityof ∼0.3, was chosen because the absorption spectra of themolecules in this solvent are well-resolved and occur atwavelengths very close to those observed when they are boundin lipid membranes or in membrane proteins.39-43 â-Caroteneand zeaxanthin, both with 11 conjugated carbon-carbon doublebonds (N ) 11, Figure 3), display almost identical broadabsorption spectral line shapes having Franck-Condon maximaat 468 nm. A profound similarity is expected because thehydroxyl groups attached to the terminalâ-ionylidene rings ofzeaxanthin (Figure 3) do not perturb theπ-electron conjugatedsystem that controls the light-absorption characteristics of thesemolecules. The absorption spectrum of lutein (N ) 10) shownin Figure 4 is better resolved than those ofâ-carotene andzeaxanthin, and its Franck-Condon maximum is blue-shiftedby 7 nm to 461 nm. The blue shift is due to the decrease inN,and the improved vibronic resolution can be traced to a reductionin conformational disorder, which can cause broadening of thespectral profiles.31 It is well-known that the presence of theterminal â-rings broadens the distribution of conformationsalong theπ-electron conjugated chain, and that together withdisorder owing to variations in the solvent environment leadsto spectral broadening.44,45 The higher vibronic resolution oflutein compared those of toâ-carotene and zeaxanthin derivesfrom the fact that one of the rings in lutein, theε-ring on theright-hand side of the structure shown in Figure 3, has a doublebond removed from the extendedπ-electron polyene chain bytwo carbon-carbon single bonds. Hence, lutein, with one lessring in conjugation thanâ-carotene and zeaxanthin, has a lesseramount of ring-induced conformational disorder. The maximain the steady-state absorption spectra of violaxanthin (N ) 9)

and neoxanthin (N ) 9), which appear at 456 and 451 nm,respectively, are blue-shifted relative to the other moleculesowing to the presence of epoxide groups in the case ofviolaxanthin and an epoxide and allene group in the case ofneoxanthin (Figure 3). These molecules contain one lessconjugated double bond than lutein and show even moreimproved vibrational resolution because in both instances theterminal â-rings do not contain carbon-carbon double bondsin conjugation with the extended polyene chain.

Transient absorption spectra ofâ-carotene and the xantho-phylls in pyridine at room temperature were taken at variousdelay times after the excitation pulse. The spectral traces areshown in Figures 5A-E. Analogous to their steady-stateabsorption spectra shown in Figure 4, the transient absorptionspectra ofâ-carotene and zeaxanthin (Figures 5A and 5B) arevery similar. They both display broad negative signals in therange of 470-520 nm corresponding to the bleaching of thestrongly allowed S0 f S2 absorption band upon excitation andalso show a buildup of a strong transient absorption signal inthe region of 520-610 nm. This latter peak is associated withthe S1 f Sn transition. The fact that the transition is very intenseimplies that the Sn state has Bu+ symmetry. This is supportedby the quantum computations discussed below. The spectrumof this S1 f Sn transition is broad (56( 1 nm, full width athalf-maximum (fwhm) measured at a 2 psdelay time) for bothâ-carotene and zeaxanthin and shows only a slight differencein their maximum positions: 579 nm forâ-carotene and 576nm for zeaxanthin.

For lutein, as is observed in its steady-state absorptionspectrum (Figure 4) and attributed to reduced conformationaldisorder, the transient absorption spectrum of this xanthophyll(Figure 5C) in the S1 f Sn transition region is sharper (46( 1nm fwhm at 2 ps) than those ofâ-carotene and zeaxanthin(Figures 5A and 5B). Also, the main S1 f Sn band has amaximum at 558 nm, which is shifted to a shorter wavelengthcompared to those ofâ-carotene and zeaxanthin. This isconsistent with an increase in the energy of the Sn state brought

Figure 4. Steady-state absorption spectra ofâ-carotene (â-car),zeaxanthin (zea), lutein (lut), violaxanthin (viol), and neoxanthin (neo)taken in pyridine solvent at room temperature. The spectra were allnormalized at the maximum of their (0-1) vibronic bands and arbitrarilyvertically offset for clarity.

Figure 5. Transient absorption spectra taken at different time delaysafter excitation into the (0-0) vibrational level of the S2 state: (A)â-carotene, (B) zeaxanthin, (C) lutein, (D) violaxanthin, and (E)neoxanthin. The spectra were taken at room temperature from themolecules dissolved in pyridine.

22876 J. Phys. Chem. B, Vol. 110, No. 45, 2006 Niedzwiedzki et al.

about by the molecule having one less carbon-carbon doublebond in conjugation. The S1 state energy also increases withdecreasing conjugation length, but the increase of the S1 (21Ag

-)state energy is apparently less than that for the high-energy Bu

+

state into which the S1 f Sn transition occurs. The spectrum oflutein also displays a clearly formed shoulder near 525 nm onthe short-wavelength side of the main band. For the carotenoids,spirilloxanthin and spheroidene, in LH complexes, this shoulderhas been assigned to the S* state.24,25,27,29,46A less well-resolvedshoulder is seen in this region of the spectra fromâ-caroteneand zeaxanthin (Figures 5A and 5B). Upon close inspection, itis observed that the short-wavelength (shoulder) feature has adifferent time dependence than the main band and, as recentlyreported for the carotenoid rhodopin glucoside in the LH2complex fromRhodopseudomonas acidophila,27 also a slightlydifferent dependence on pump laser power. The data reveal aslight increase in relative signal intensity in the shoulder regionas the pump energy is increased from 600 nJ to 2µJ. (See FigureS1 in the Supporting Information and discussion below.)

The transient absorption spectrum of violaxanthin is shownin Figure 5D. Due to its shorterπ-electron conjugated chainand the absence of terminalâ-rings in conjugation comparedto â-carotene, zeaxanthin, and lutein, the main S1 f Sn transientabsorption peak is shifted even farther to the blue, appearing at533 nm as a fairly sharp peak with a fwhm of 24( 1 nm. Thisspectrum does show a short-wavelength shoulder at 510 nm,but it is much less intense compared to those of lutein and theother molecules. The transient absorption spectrum of neoxan-thin is shown in Figure 5E. Its main S1 f Sn absorption peakhas a maximum at 543 nm and a fwhm of 41( 1 nm, bothvalues of which are midrange between those for violaxanthinand the other molecules. The position of the maximum sug-gests that in the excited state neoxanthin has a longer effec-tive π-electron conjugation than violaxanthin but shorter thanthose of the other molecules. The value of the bandwidthsuggests that the extent of conformational disorder for neox-anthin is greater than that in in violaxanthin but less than thosein the other molecules. Neither of these two factors follow thesame trend seen for neoxanthin in the ground state where its S0

f S2 transition is the most blue-shifted of all the moleculesand its extent of vibronic resolution is comparable to that ofviolaxanthin.

One other significant spectral feature seen for all of themolecules is a broad, gradually sloping, positive signal thatappears on the long-wavelength side of the major S1 f Sn peak.This is observed in all of the transient profiles taken at a 500 fsdelay of the pulse beam. (See the dashed lines in Figures5A-E.) In all cases, this feature builds up and decays beforethe main S1 f Sn peak (solid line in Figures 5A-E) reaches itsfull intensity.

The transient data can be summarized as follows: In the timerange between 0 and 10 ps, upon photoexcitation all of themolecules display an immediate onset of bleaching of the S0

f S2 absorption transition in the wavelength range between450 and 525 nm, the subsequent build up and decay within 1ps of a broad, sloping, long-wavelength feature, followed bythe rise and partial decay of both a strong S1 f Sn transition inthe region of 520-610 nm and a variable-sized short-wavelengthshoulder in the region of 500-550 nm.

To gain more insight into the photophysical behavior of thesemolecules, the entire spectral and temporal datasets were fitsimultaneously using a global analysis procedure employing amultiexponential function,S(λ,t) ) ∑i Ai(λ) exp(-t/τi), whereAi(λ) is the preexponential amplitude factor associated with

decay componenti having a time constantτi. This sum ofexponentials model represents the dynamic behavior of a numberof parallel, noninteracting kinetic components, the amplitudefactors,Ai(λ), which are termed decay associated spectra (DAS),or more appropriately in the present context, decay associateddifference spectra (DADS) because difference absorption spectraare recorded.47,48 As is thoroughly described in the literature,DADS amplitudes do not represent real, physical, spectroscopicprofiles of the transient species. Real, physical spectra are termedspecies associated difference spectra (SADS), but DADS canbe expressed as linear combinations of SADS; i.e., DADSi )∑j

n cijSADSj where the ith DADS component, DADSi, isidentical to the preexponential factor,Ai(λ), in the multiexpo-nential function,S(λ,t), and j is the index for each one of anumber,n, of physically real (SADSj) spectra contributing tothe DADSi profile. Individual SADSj are often difficult to obtaindue to overlapping spectral profiles and comparable kineticbehavior among the transient species. An alternative method isto fit the datasets using a nonbranching, sequential, irreversiblescheme Af B, B f C, C f D, ... The arrows representincreasingly slower monoexponential processes, and the timeconstants of these processes correspond to lifetimes of thetransient species A, B, C, D, ... The spectral profiles of thesespecies are termed evolution-associated difference spectra(EADS). Although EADS in complicated systems do notnecessarily correspond to SADS of particular excited states, theyprovide information about the time evolution of the wholesystem.48 Thus, while DADS provide information about spectralprofiles of the preexponential factors, EADS give first ap-proximations to the real concentration profiles of the transientspecies. A detailed analysis of the global fitting methods andtheir application to various biological systems can be found inref 48.

For all the molecules examined in this work, four (n ) 4)DADS and EADS components were necessary to obtainsatisfactory fits based on a chi square (ø2) test and the smallnessof their equivalent residual matrices. Single-wavelength fitsdisplayed in Figures S2 and S3 show that the molecules exhibitdifferent time dependences at different probe wavelengths.Although three kinetic components lead to a satisfactory fit attheλmax of the S1 f Sn transition (Figure S2), four componentsare needed when one probes the wavelength region on the short-wavelength side of this band. This is most clearly illustrated inFigure S3 forâ-carotene, zeaxanthin, and lutein where a three-component fit is shown to be inadequate, but a four-componentfit works nicely. The DADS amplitudes resulting from fittingthe transient absorption data to a sum of exponentials kineticmodel are displayed in Figure 6. Forâ-carotene (Figure 6A), avery broad, negative-amplitude DADS component builds up in170( 2 fs. For the xanthophylls, the time for this fast, negativecomponent ranged from 110 to 170 fs (Figures 6B-E). A timeconstant in this range is associated with the lifetime of the S2

state of carotenoids.49 The various assorted negative featuresappearing in this first DADS component may be attributed tothe buildup of a vibrationally hot S1 f Sn transition, theformation of the S* state, the buildup of its associated S*f Sn

transition, stimulated fluorescence, and stimulated Raman bandsarising from the solvent. The second DADS component in allcases has a complex line shape featuring a broad positive (decay)profile at long wavelengths, a zero crossing, and a broadnegative (build-up) band that spans the region encompassingthe S1 f Sn transition and the S*f Sn short-wavelengthshoulder. Forâ-carotene this second amplitude spectrum has atime constant of 366( 10 fs, and for the xanthophylls, the

Femtosecond Spectroscopy of Xanthophylls J. Phys. Chem. B, Vol. 110, No. 45, 200622877

values range from 370 to 582 fs. The positive, long-wavelengthpart of this component has been attributed to the decay of avibrationally hot S1 state to form a vibrationally equilibratedS1 state.50-52 This is rationalized by the fact that the decay ofS2 to S1 is so rapid that it brings with it a significant amount ofvibrational energy that can only be dissipated after a fewhundred femtoseconds. The presence of this excess vibrationalenergy in S1, which populates the upper vibronic levels of thestate, is expected to give rise to a broad, red-shifted S1 f Sn

spectrum that then decays to a narrower, blue-shifted, S1 f Sn

transition associated with the vibrationally equilibrated system.The zero crossing and a strong, negative, shorter-wavelengthfeature observed in the second DADS component are compellingevidence that this is the case for all of the molecules examinedhere. Yet, it also has been suggested that this kinetic componentand the broad, long-wavelength feature correspond to an excited-state transition originating from the 11Bu

- excited electronicstate,23 theoretically predicted to be in the vicinity of S1 andS2.20,53,54 However, as pointed out by Billsten,32 if this werethe case, then such an intense signal implies that the final stateshould have Ag+ symmetry. The energy of the lowest Ag

+ stateis known from the location of the S0 f “A g

+” cis-peak. Forâ-carotene this occurs at∼29 000 cm-1.55 Subtraction of theapproximate energy (∼18 000 cm-1) corresponding to theobserved broad transition peaking at∼ 550 nm would put the11Bu

- state at∼11 000 cm-1, which is far below the S1 (21Ag-)

state energy ofâ-carotene known to be at∼14 500 cm-1.56 Thiswould contradict both experimental evidence23 and theoreticalpredictions53 for the position of the 11Bu

- state relative to S1.The second DADS component also shows a negative-amplitudeshoulder on the short-wavelength side of the strong negativefeature. This is most evident at 530 nm in the amplitudespectrum of lutein (dotted line in Figure 6C), to some extentnoticeable, but not well-resolved, in the amplitude spectra ofâ-carotene, zeaxanthin, and neoxanthin, and absent in the

amplitude spectrum of violaxanthin (dotted line in Figure 6D).This negative short-wavelength shoulder may be attributed tothe arrival of population from S2 into the S* state, giving riseto an associated S*f Sn transition.

The longest-time DADS component, in all cases, shows thefamiliar, strongly allowed, vibronically relaxed, positive-amplitude spectrum associated with the S1 f Sn transition. Forâ-carotene the time constant of this component is 9.5( 0.1 ps,and for the xanthophylls, zeaxanthin, lutein, violaxanthin, andneoxanthin, the values are 10.2( 0.2, 15.6( 0.1, 26.1( 0.1,and 37.6( 0.1 ps, respectively. These correspond well to thevalues of the S1 lifetime for these molecules reported in theliterature that tend to increase with decreasingπ-electronconjugation length (Table 1). This longest-time DADS com-ponent also has an associated strong negative signal at a shorterwavelength that represents the recovery of the ground-statebleaching as S1 decays.

For all of the molecules examined here, a third DADScomponent was observed having a time constant in the rangeof 2.7-5.0 ps, i.e., between the second and longest-timecomponents. This DADS component has a significant amplitudefor â-carotene, zeaxanthin, and lutein but is small and hardlynoticeable for violaxanthin and neoxanthin. The componentshows a wavy line shape with at least two positive and twonegative peaks spanning the entire probe wavelength region.For â-carotene, zeaxanthin, and lutein, it has a negativeamplitude on the red side and a positive amplitude on the short-wavelength side of the S1 f Sn transition profile. For violax-anthin and neoxanthin, although the signals are very small(Figures 6D and 6E), this appears to be reversed, with positiveamplitude on the long-wavelength side and negative amplitudeon the short-wavelength side of the S1 f Sn band. However, inall cases the negative-amplitude feature tracks precisely thestrong positive feature of the final DADS.

Figure 6. Decay associated difference spectra (DADS) obtained froma global fitting analysis using four kinetic components: (A)â-carotene,(B) zeaxanthin, (C) lutein, (D) violaxanthin, and (E) neoxanthin.

Figure 7. Evolution-associated difference spectra (EADS) obtainedfrom a global fitting analysis using four kinetic components: (A)â-carotene, (B) zeaxanthin, (C) lutein, (D) violaxanthin, and (E)neoxanthin.

22878 J. Phys. Chem. B, Vol. 110, No. 45, 2006 Niedzwiedzki et al.

The EADS components resulting from a global fitting analysisusing a sequential kinetic model are displayed in Figure 7. Inall cases, the initial EADS corresponds to the spectrum of theexcited S2 state. It is characterized by a large negative, ground-state bleaching signal between 475 and 525 nm accompaniedby a broad, sloping, negative feature at a longer wavelengthdue primarily to stimulated emission from S2. The first EADSdecays rapidly (110-170 fs) to form the second EADScomponent that for all the molecules displays a very broad,positive line shape extending significantly to long wavelengths.As mentioned above this feature is assigned to a transitionbetween a vibrationally hot S1 state and Sn.51,52The third EADSin the sequence rises in 366-582 fs and decays with a 2.7-5.0ps time constant and shows a very strong, broad, positive band,which narrows as the system evolves into a fourth and finalEADS. The line narrowing is most evident forâ-carotene,zeaxanthin, and lutein. This step is also accompanied by a slightwavelength shift of the major positive feature either to the blue(â-carotene, zeaxanthin, lutein) or to the red (violaxanthin,neoxanthin). Also, from the third to the fourth EADS, theground-state bleaching signal recovers slightly indicating aportion of the population has relaxed from an excited state tothe ground state.

Because the DADS components can be expressed as linearcombinations of various SADS approximated by the EADS (seeabove), the interpretation of the shape of the DADS profiles inFigure 6 is straightforward. The individual DADS in Figure 6can be generated by taking an arithmetic difference betweentwo sequential EADS with only a slight adjustment in coef-ficient, Cij. The larger the difference in the time constants ofthe EADS components, the more precise the agreement. Forexample, subtracting any fourth EADS component from anythird EADS component from the same molecule yields almostperfect agreement with its third DADS component given inFigure 6. This is clearly illustrated in an overlay of the EADSdifference spectra with the DADS components shown in FigureS4 in the Supporting Information. Thus, the reason for thewavy nature of the 2.7-5.0 ps (third) DADS components ofâ-carotene, zeaxanthin, and lutein (light solid lines in Figures6A-C) becomes clear. It is due to shifts in the wavelengthpositions of the peaks in the fourth EADS component comparedto those in the third (Figures 7A-C).

The time-resolved data were also analyzed using single-wavelength fits (Figure 8) taken at positions where the second

DADS component assigned to the hot S1 state crosses zero. Atthese wavelengths, the kinetics are free from a contribution ofS1 vibrational relaxation. The crossover wavelengths are 594nm for â-carotene, 596 nm for zeaxanthin, 571 nm for lutein,543 nm for violaxanthin, and 558 nm for neoxanthin (Figures6A-E). The fits to these specific single-wavelength responseprofiles are shown in Figure 8 and required two exponentialdecay components to satisfactorily reproduce the experimentaldata in all cases except for zeaxanthin where two were notneeded because of the vanishingly small amplitude of the seconddecay component at 596 nm. These two decay componentsemerging from the single-wavelength fits are associated withthe lifetimes of the S1 and S* states. The values of the S1/S*ratios of the preexponential factors were-4.3 (â-carotene),-15.7 (lutein), 16.5 (violaxanthin), and 6.8 (neoxanthin). Thechange in sign of the ratio for the latter two molecules is dueto the inversion of the amplitude spectrum of the kineticcomponent associated with S* (Figure 6).

The kinetics obtained from the global fitting and single-wavelength analyses have been collected with results fromprevious experiments on the same molecules in various solventsand are presented in Table 1.

Discussion

Spectral Features and Kinetic Components.The steady-state and transient absorption spectra of the molecules examinedhere follow the trends previously observed in both position andbroadness of the S0 f S2 and S1 f Sn transitions. The longerconjugated carotenoids absorb farther to the red than the shortermolecules, and the systems with terminalâ-ionylidene ringshaving a double bond in conjugation with the extended polyenechain show broader S0 f S2 and S1 f Sn spectra due toconformational disorder. The only exception is neoxanthin whereits S1 f Sn transition appears broader and more red-shifted thanexpected. This is the case because neoxanthin adopts a 9′-cis-configuration as its most stable geometric isomer. Previous workhas demonstrated that although the S0 f S2 transition for cis-isomers of carotenoids are generally blue-shifted compared totheir all-trans counterparts57 the S1 f Sn transitions of cis-isomers are typically red-shifted. Recent work in our laboratorycomparing the transient absorption spectra of cis- and trans-isomers ofâ-carotene and spheroidene have confirmed this tobe the case.50 Thus, neoxanthin would not be expected to followthe trends in position and width set by a series of trans-isomers,and indeed the lack of agreement for this molecule among theothers in the series is understandable on this basis.

The global fits reveal two kinetic components with lifetimeslonger than 1 ps. The first of these ranges from 9.5 to 37.6 psin going fromâ-carotene to neoxanthin and is clearly associatedwith the lifetime of the S1 state. A change in S1 lifetime isexpected based on variations in the conjugatedπ-electron chainlength that lead to changes in the S1-S0 energy gap. This effecthas been well-documented (Table 1). The second kineticcomponent spans a narrower range (2.7-5.0 ps) and representsthe lifetime of the S* state. The wavy, variable-amplitude spectraof the 2.7-5.0 ps DADS components (Figure 6) are suggestiveof spectral band shifts. In the wavelength region of theS1 f Sn transition, the shift appears to be to longer wavelengthsfor â-carotene, zeaxanthin, and lutein (see the third DADScomponent in Figures 6A-C), because a positive feature appearsbetween the S1 f Sn transition and the S0 f S2 transition anda negative feature appears at longer wavelengths. The effect isreversed for violaxanthin and neoxanthin (see the third DADScomponent in Figures 6D and 6E), but to understand the datamore thoroughly, the EADS components should be considered.

Figure 8. Transient absorption kinetic traces of (9) â-carotene, (0)zeaxanthin, (b) lutein, (O) violaxanthin, and ([) neoxanthin probedat the crossover wavelengths where the contribution from the S1 f Sn

transition involving vibrationally hot S1 is negligible. The amplitudeswere normalized to unity, and only every third data point is shown forclarity. The solid lines represent the fits obtained from a sum ofexponentials expression as described in the text.

Femtosecond Spectroscopy of Xanthophylls J. Phys. Chem. B, Vol. 110, No. 45, 200622879

In the longest (fourth) EADS, a shoulder is observed on theshort-wavelength side of the S1 f Sn transition forâ-carotene,zeaxanthin, lutein, and neoxanthin. This shoulder is reminiscentof that assigned to S* state in spirilloxanthin,24,27,29,46but thedata presented here show unequivocally that the shoulderobserved at the longer times must be associated with the S1

state for these xanthophylls. This is demonstrated by the factthat it persists in the longest DADS and EADS profiles, evenfor neoxanthin whose S1 lifetime is 37.6 ps, i.e., more than anorder of magnitude longer than the 2.7 ps lifetime assigned toS* state. This indicates that the shoulder seen in the longest-time DADS and EADS components must be associated withthe S1 state, because if it were associated with the S* state,then it would have already decayed away.

There are two possibilities for how the shoulder may arise.The first is that it may be associated with a (0-1) vibronictransition accompanying the major (0-0) spectral origin of theS1 f Sn transition. The intensity ratio and energy separationbetween the (0-0) major feature and the (0-1) shoulder arenot inconsistent with this assignment. The energy separation isobserved to be in the range of 900-1100 cm-1, which is in

agreement with that expected for the difference between the(0-0) spectral origin and a (0-1) vibronic band for thesemolecules. See, for example, the∼25 nm separation of the(0-0) and (0-1) vibronic peaks in the steady-state absorptionspectra ofâ-carotene and zeaxanthin (Figures 4A and 4B),which corresponds to an energy separation of∼1100 cm-1.

The second possibility is that the shoulder represents atransition from S1 to a different higher-energy electronic statethan that giving rise to the major S1 f Sn absorption band. Toexamine this option more thoroughly, quantum computationswere carried out.

Quantum Chemical Computations.The geometry ofâ-car-otene was optimized using density functional methods forthe ground state (B3LYP/6-31G(d)) and ab initio methods(CIS(D) and SAC-CI with a D95 basis set) for the low-lyingexcited singlet states. The spectroscopic properties of the mole-cule were then analyzed using MNDO-PSDCI molecular orbitaltheory.36-38

Experimental and theoretical studies are in agreement thatthe â-ionylidene ring in both the short-chain retinal polyenesand the longer-chain carotenoids selects a 6-s-cis-conformationin the ground state.37,58This observation also applies to the low-lying strongly allowed 11Bu

+ state. In contrast, the approximateadiabatic surfaces for ring torsion in the ground and first twoexcited singlet states ofâ-carotene given in Figure 9 show thatlowest-lying 21Ag

- state selects preferentially a 6-s-trans-conformation. The ground-state surface minimum has a distorteds-cis geometry, but the lowest excited singlet state has a planars-trans geometry. The origin of this conformational selection isexamined in Figure 10 where the key molecular orbitals thatparticipate in the configurational description are shown. Ex-

Figure 9. Approximate adiabatic surfaces for ring torsion in the groundand first two excited singlet states ofâ-carotene. The ground-statesurface minimum has a distorted s-cis geometry, but the lowest excitedsinglet state has a near-planar s-trans geometry whose conformation islabeled S1*.

Figure 10. Key orbitals that make up the configurational descriptionof the first excited singlet state ofâ-carotene. This state is an opticallyforbidden1Ag

--like state characterized by both MNDO-PSDCI andSAC-CI theory as having a high degree of doubly excited character(∼55%). This character produces significant bond order reversal andpreferential stabilization of the 6-s-trans geometry, thus providing amore highly correlated singlet state.

22880 J. Phys. Chem. B, Vol. 110, No. 45, 2006 Niedzwiedzki et al.

amination of Figure 10 indicates that transferring an electronfrom the filled to the open orbitals creates significant bond orderreversal near the center of the polyene chain. Indeed, the bondlengths of the polyene carbon atoms near the center of thepolyene chain are within 0.01 Å of one another. Such a highdegree of bond order reversal is unexpected in such a long-chain polyene.

Figure 11 shows the calculated (MNDO-PSDCI) S1 f Sn

and S2 f Sn spectra of â-carotene based on geometriesminimized to the specified (cis- or trans-) 6-s-conformation.Solid vertical bars indicate transitions to the1Bu-like states, andgray bars represent transitions to1Ag-like states. The computa-tions show thatâ-carotene optimized to a 6-s-cis-conformationdisplays two separate but neighboring electronic transitions fromthe S1 (21Ag

-) state to different high-energy Bu-like states (top

panel in Figure 11). The higher-energy transition of the two iscomputed to have approximately half of the oscillator strengthof the major transition, which is entirely consistent with theexperimental observations of a major S1 f Sn band and a smallerblue-edge shoulder (Figures 5-7). The optimization to the 6-s-cis-conformation also predicts an S2 f Sn transition in the near-IR region at∼850 nm (bottom panel in Figure 11). We haveobserved strong transient absorption signals from xanthophyllsin this wavelength region (data not shown). The signals buildup and decay within the time duration of the excitation laserpulse and thus are entirely consistent with a transition originatingfrom S2. The details of these observations will be included in aforthcoming paper soon to be submitted for publication.

At present it is not possible to distinguish between the twopossibilities, but if either of the interpretations is correct, then

Figure 11. Calculated (MNDO-PSDCI) S1 f Sn and S2 f Sn spectra ofâ-carotene based on geometries minimized to the specified (cis- ortrans-)6-s-conformation. Solid vertical bars indicate transitions to the1Bu-like states, and gray bars represent transitions to1Ag-like states. Calculationsincluded full single and double configuration interaction within theπ-electron manifold.

Femtosecond Spectroscopy of Xanthophylls J. Phys. Chem. B, Vol. 110, No. 45, 200622881

the shoulder observed in the longest-time DADS and EADSline shapes should grow in at the same rate as the major S1 fSn transition upon vibronic relaxation of the hot S1 state. Thesecond DADS components (dotted lines in Figure 6) show thatthis is the case. Both the shoulder and the main band featureappear in the same negative (build-up) amplitude DADS forall of the molecules except violaxanthin. For this molecule theamplitude of the DADS spectrum in the region of the shoulderis essentially zero. However, violaxanthin behaves less like axanthophyll with terminal rings and more analogous to open-chain carotenoids where the majority of the S1 f Sn oscillatorstrength appears to be built into a single, narrow, strongband.50,52,59-61 Thus, the shoulder observed on the short-wavelength side of the S1 f Sn transition at early times (<5ps) contains a contribution from S*. Any remaining ampli-tude in this region at longer times (>15 ps) is associated solelywith S1.

Nature of the S* State. Currently, there have been twomodels proposed to explain the origin and behavior of S* forcarotenoids in solution: the excited-state model and the hotground-state model (Figure 12). In the excited-state model, S*is formed in∼100 fs after photoexcitation either via a brancheddecay pathway from S2 which also forms S1 (red dashed linesin Figure 12) or, to account for the nonlinear dependence ofthe signal amplitudes on pump energy,27 from a higher electronicor vibronic state populated via a second photoexcitation fromS2 which then branches to form S* and S1 (green dashed linesin Figure 12). It is important to note, however, that S* is formedwith a substantial yield even at very low excitation intensi-ties.24,27Subsequently, the S*f Sn and S1 f Sn signals decayindependently in several picoseconds as S* and S1 are depopu-lated. In the hot ground-state model, S* is a vibrationally excitedground state populated in part instantaneously by impulsiveRaman scattering during the time course of the pump laserexcitation.28 In this case, photoexcitation can be thought of asyielding a population of hot ground-state molecules from whichabsorption still occurs to the S2 state, but the spectral line shapeof the S* f S2 transition is expected to be broader and red-shifted relative to the normal S0 f S2 transition.

The DADS and EADS global fits in conjunction withquantum computational modeling hold the key to understanding

which of these models represents the true behavior of thesemolecules. Because the DADS are very well approximated bydifferences in sequential EADS (Figure S4), it is clear that thewavy shape of the 2.7-5.0 ps DADS kinetic components(Figure 6) occurs due to the formation and decay of transientspecies having distinct spectra in the regions corresponding toboth the S1 f Sn and the S0 f S2 transitions. To see this moreclearly, Figure 13 shows the spectral profiles of the 2.7-5.0 psDADS component overlaid with both the S0 f S2 steady-stateabsorption spectra and the longest-time DADS component,which is dominated by the spectrum of the strongly allowed S1

f Sn transition. Note that the spectra in Figure 13 are allnormalized to the amplitude of their largest positive feature.The 2.7-5.0 ps DADS line shapes seen in Figure 13 revealespecially clearly for zeaxanthin, lutein, violaxanthin, andneoxanthin a very broad transition associated with S* and havinga significant positive rise starting at 650 nm and gaining intensityas one goes to shorter wavelengths. Built on this broad spectralline shape are features corresponding to the major S1 f Sn

transition and the peaks of the ground-state S0 f S2 vibronictransitions. These features appear because neither the DADS(parallel) model nor the EADS (sequential) model perfectlyrepresent the photophysics of decay of the xanthophylls fromtheir excited states. Thus, the third DADS components orequivalently the difference between the third and fourth EADScomponents (Figure S4) contains a contribution from theS1 f Sn excited-state absorption and ground-state bleaching inaddition to the S*f Sn′ transition. Nevertheless, as shown inprevious work,28 the S*f Sn′ line shape is broad and featurelessin the region of 450-650 nm, and the data presented hereindicate that the intensity of this band depends critically on thestructure of the xanthophyll. A clear distinction is observedbetween two groups of molecules:â-Carotene, lutein, andzeaxanthin have a higher yield of S* and a stronger S*f Sn′transition than violaxanthin and neoxanthin. The major struc-tural difference between these two groups is the extensionof conjugation to the terminal ring(s) in the first group (Figure3), and these data indicate the key role of the conjugatedâ-ionylidene rings in controlling the dynamics of deactivationfrom the excited states of the xanthophylls.

Figure 11 shows that the calculated (MNDO-PSDCI) S1 fSn spectrum ofâ-carotene based on a geometry minimized to a6-s-cis-conformation predicts two close-lying electronic transi-tions originating from S1. In contrast, the geometry minimizedto a 6-s-trans-conformation shows three relatively strongelectronic transitions in the region between 450 and 650 nm,the combined maximum of which is shifted approximately 20nm to the shorter wavelengths of the 6-s-cis-conformationspectrum. This is the same magnitude of shift observed for theS* f Sn′ transition compared to the S1 f Sn transition (Figure13). The spectrum of the S*f Sn′ transition may also be broaderbased on the finding of significant bond order reversal near thecenter of the polyene chain that could lead to spectral hetero-geneity due to conformational disorder.44 The 6-s-trans-conformation also has a narrower energy gap with the groundstate (Figure 9), which, based on the energy gap law forradiationless transitions,62 could account for its faster (3.4(0.2 ps) lifetime compared to the (9.5( 0.1 ps) value for the S1state ofâ-carotene. Thus, we propose that a branched decaypathway of the photoexcited xanthophylls leads to the pop-ulation of either a 6-s-cis-conformation (the S1 state) or a6-s-trans-conformation (the S* state), which then decay in-dependently at different rates back to the 6-s-cis ground state.This is an important point because if the 6-s-trans ground state

Figure 12. Two hypothetical schemes for the origin of S*, the excited-state model and the hot ground-state model, and associated photopro-cesses: a, absorption; ta, transient absorption; ir, impulsive Ramanscattering; esa, excited-state absorption. The green dashed arrowsrepresent branching from Sn as suggested by Papagiannakis et al.,27

and the red arrows show branching from S2.

22882 J. Phys. Chem. B, Vol. 110, No. 45, 2006 Niedzwiedzki et al.

were populated, then the barrier between it and the more stable6-s-cis ground state would prevent the molecule from reformingthe original conformation and would result in a nondecayingbleaching signal. Because this is not observed, all of the excitedmolecules must ultimately end up back in the original 6-s-cisground state. The 6-s-trans-conformation accounts for thespectroscopic features and dynamics attributed to the S* stateincluding the transition appearing to the blue of the major S1

f Sn transient absorption band (Figure 5), its broad featurelessspectral line shape (Figure 13), and the partial recovery of theground-state bleaching in 2.7 to 5.0 ps (Figure 7). The remainderof the ground-state bleaching recovers as S1 decays. Thepresence of carbon-carbon double bonds in theâ-ionylidenerings in conjugation with theπ-electron polyene systems ofâ-carotene, zeaxanthin, and lutein facilitate the conformationalchange and lead to higher yields of S* for those molecules.The fact that the S* state was first detected and described forspirilloxanthin, which has the longestπ-electron conjugatedsystem (N ) 13) of any naturally occurring carotenoid, isconsistent with this apparent requirement for extended, flexible,terminalπ-bonds to produce S* in a significant yield.

The model being proposed here is similar to that describedby de Weerd et al.51 in which distortion ofâ-carotene is invokedto account for the time-resolved changes in excited-state

absorption during relaxation. If these models are correct, thenthe yield of twisted molecules should be influenced by themedium and reduced in high viscosity or frozen solvents.Preliminary studies in our laboratory on zeaxanthin and luteinin 5/5/2 v/v/v diethyl ether/isopentane/ethanol (EPA) glassesat 77 K indicate that this is indeed the case (unpublished results).

It should be mentioned that the assignment of the spectro-scopic features of S* to an excited state in the first place24 wasbased in part on the fact that spirilloxanthin showed similarbehavior whether it was in solution or bound in the LH1complex ofRs. rubrum. In particular, the analysis of the S*dynamics revealed it to be a precursor to carotenoid triplet stateformation in the LH1 complex, but this photochemical processdoes not occur for the molecule in solution. It is thus likelythat in light-harvesting complexes conformationally perturbedcarotenoids may facilitate triplet state formation.

The results of this study have detailed the positions and timedependence of the complex, ultrafast spectroscopic featuresassociated with the major xanthophyll pigments in higher plantsand have elucidated the origin of the S* state in xanthophylls.The higher yields of S* forâ-carotene, zeaxanthin, and luteincompared to those for violaxanthin and neoxanthin and the fasterdynamics of S* decay compared to that of S1 suggest S* andits associated twisted conformation may facilitate nonradiative

Figure 13. Overlay of the third and fourth DADS components with the steady-state spectra of all the molecules: (A)â-carotene, (B) zeaxanthin,(C) lutein, (D) violaxanthin, and (E) neoxanthin. The spectra were all normalized to the amplitude of their largest positive components.

Femtosecond Spectroscopy of Xanthophylls J. Phys. Chem. B, Vol. 110, No. 45, 200622883

relaxation to the ground state as a means of dissipating excessenergy in the process of NPQ. In any case, the data are expectedto be useful in the ongoing analysis of the ultrafast spectroscopicobservables of these same molecules in thylakoid and pigment-protein complex preparations from higher plants. These studiesseek to understand what precise relaxation pathways in proteinspertain to the mechanism of NPQ and control the manner inwhich plants adapt to high light stress.

Acknowledgment. The authors thank George Gibson forexpert advice on the operation of the laser spectrometer system.This work is supported in the laboratory of H.A.F. by theNational Science Foundation (Grant No. MCB-0314380) andthe University of Connecticut Research Foundation and in thelaboratory of R.R.B. by the National Institutes of Health (GrantNo. GM-34548) and the National Science Foundation (GrantNos. BES-0412387 and CCF-0432151). Partial support forcomponents of the ultrafast laser spectrometer system was alsoprovided by a grant to H.A.F. from the National Institutes ofHealth (Grant No. GM-30353). T.P. thanks the Czech Ministryof Education for financial support (Grants Nos. MSM6007665808and AV0Z50510513).

Supporting Information Available: Plot of the ratio of theS1 and S* amplitudes obtained from the third and fourthEADS as a function of pump pulse energy for lutein, compari-son of the three- and four-component fits at theλmax of theS1-Sn transition, comparison of the three- and four-componentfits at the region of the S*-Sn transition, and overlay of thethird DADS components with the difference spectra gen-erated by subtracting the third and fourth EADS compo-nents. This material is available free of charge via the Internetat http://pubs.acs.org.

References and Notes

(1) Carotenoids; Britton, G., Liaaen-Jensen, S., Pfander, H., Eds.;Birkhauser Verlag: Basel, Switzerland, 1995; Vol. 1B.

(2) Demmig-Adams, B.; Adams, W. W., III. The xanthophyll cycle.In Carotenoids in Photosynthesis; Young, A. J., Britton, G., Eds.; Chapmanand Hall: London, 1993; p 206

(3) Krinsky, N. I. Function. InCarotenoids; Isler, O., Guttman, G.,Solms, U., Eds.; Birkha¨user Verlag: Basel, Switzerland, 1971; p 669.

(4) Holt, N. E.; Fleming, G. R.; Niyogi, K. K.Biochemistry2004, 43,8281.

(5) Ruban, A. V.; Phillip, D.; Young, A. J.; Horton, P.Biochemistry1997, 36, 7855.

(6) Horton, P.; Ruban, A. V.; Rees, D.; Pascal, A. A.; Noctor, G.;Young, A. J.FEBS Lett.1991, 292, 1.

(7) Horton, P.; Ruban, A. V.; Walters, R. G.Plant Physiol.1994, 106,415.

(8) Horton, P.; Ruban, A. V.; Wentworth, M.Philos. Trans. R. Soc.London, Ser. B2000, 355, 1361.

(9) Frank, H. A.; Cua, A.; Chynwat, V.; Young, A.; Gosztola, D.;Wasielewski, M. R.Photosynth. Res.1994, 41, 389.

(10) Polıvka, T.; Herek, J. L.; Zigmantas, D.; Akerlund, H. E.;Sundstro¨m, V. Proc. Natl. Acad. Sci. U.S.A.1999, 96, 4914.

(11) Frank, H. A.; Bautista, J. A.; Josue, J. S.; Young, A. J.Biochemistry2000, 39, 2831.

(12) Josue, J. S.; Frank, H. A.J. Phys. Chem. A2002, 106, 4815.(13) Polıvka, T.; Zigmantas, D.; Sundstro¨m, V.; Formaggio, E.; Cinque,

G.; Bassi, R.Biochemistry2002, 41, 439.(14) Holt, N. E.; Zigmantas, D.; Valkunas, L.; Li, X. P.; Niyogi, K. K.;

Fleming, G. R.Science2005, 307, 433.(15) Hudson, B.; Kohler, B.Annu. ReV. Phys. Chem.1974, 25, 437.(16) Hudson, B. S.; Kohler, B. E.; Schulten, K. Linear polyene electronic

structure and potential surfaces. InExcited States; Lim, E. D., Ed.; AcademicPress: New York, 1982; Vol. 6, p 1.

(17) Birge, R. R.Acc. Chem. Res.1986, 19, 138.(18) Pariser, R.J. Chem. Phys.1955, 24, 250268.(19) Callis, P. R.; Scott, T. W.; Albrecht, A. C.J. Chem. Phys.1983,

78, 16.(20) Tavan, P.; Schulten, K.J. Chem. Phys.1979, 70, 5407.(21) Schulten, K.; Karplus, M.Chem. Phys. Lett.1972, 14, 305.

(22) Polıvka, T.; Sundstro¨m, V. Chem. ReV. 2004, 104, 2021.(23) Koyama, Y.; Rondonuwu, F. S.; Fujii, R.; Watanabe, Y.Biopoly-

mers2004, 74, 2.(24) Gradinaru, C. C.; Kennis, J. T. M.; Papagiannakis, E.; van Stokkum,

I. H. M.; Cogdell, R. J.; Fleming, G. R.; Niederman, R. A.; van Grondelle,R. Proc. Natl. Acad. Sci. U.S.A2001, 98, 2364.

(25) Papagiannakis, E.; Kennis, J. T. M.; van Stokkum, I. H. M.; Cogdell,R. J.; van Grondelle, R.Proc. Natl. Acad. Sci. U.S.A2002, 99, 6017.

(26) Wohlleben, W.; Buckup, T.; Herek, J. L.; Cogdell, R. J.; Motzkus,M. Biophys. J.2003, 85, 442.

(27) Papagiannakis, E.; van Stokkum, I. H.; Vengris, M.; Cogdell, R.J.; van Grondelle, R.; Larsen, D. S.J. Phys. Chem. B2006, 110, 5727.

(28) Wohlleben, W.; Buckup, T.; Hashimoto, H.; Cogdell, R. J.; Herek,J. L.; Motzkus, M.J. Phys. Chem. B2004, 108, 3320.

(29) Papagiannakis, E.; Das, S. K.; Gall, A.; Stokkum, I. H. M.; Robert,B.; van Grondelle, R.; Frank, H. A.; Kennis, J. T. M.J. Phys. Chem. B2003, 107, 5642.

(30) Polıvka, T.; Pullerits, T.; Frank, H. A.; Cogdell, R. J.; Sundstro¨m,V. J. Phys. Chem. B2004, 108, 15398.

(31) Christensen, R. L.; Barney, E. A.; Broene, R. D.; Galinato, M. G.I.; Frank, H. A.Arch. Biochem. Biophys.2004, 430, 30.

(32) Billsten, H. H.; Pan, J.; Sinha, S.; Pascher, T.; Sundstro¨m, V.;Polıvka, T. J. Phys. Chem. A2005, 109, 6852.

(33) Andersson, P. O.; Gillbro, T.J. Chem. Phys.1995, 103, 2509.(34) Ilagan, R. P.; Christensen, R. L.; Chapp, T. W.; Gibson, G. N.;

Pascher, T.; Polivka, T.; Frank, H. A.J. Phys. Chem. A2005, 109, 3120.(35) Liu, Z. F.; Yan, H. C.; Wang, K. B.; Kuang, T. Y.; Zhang, J. P.;

Gui, L. L.; An, X. M.; Chang, W. R.Nature (London)2004, 428, 287.(36) Dolan, P. M.; Miller, D.; Cogdell, R. J.; Birge, R. R.; Frank, H. A.

J. Phys. Chem. B2001, 105, 12134.(37) Shima, S.; Ilagan, R. P.; Gillespie, N.; Sommer, B. J.; Hiller, R.

G.; Sharples, F. P.; Frank, H. A.; Birge, R. R.J. Phys. Chem. A2003, 107,8052.

(38) Ilagan, R. P.; Shima, S.; Melkozernov, A.; Lin, S.; Blankenship,R. E.; Sharples, F. P.; Hiller, R. G.; Birge, R. R.; Frank, H. A.Biochemistry2004, 43, 1478.

(39) Ruban, A. V.; Pascal, A. A.; Robert, B.; Horton, P.J. Biol. Chem.2001, 276, 24862.

(40) Das, S. K.; Frank, H. A.Biochemistry2002, 41, 13087.(41) Peterman, E. J. G.; Gradinaru, C. C.; Calkoen, F.; Borst, J. C.; van

Grondelle, R.; van Amerongen, H.Biochemistry1997, 36, 12208.(42) Peterman, E. J. G.; Dukker, F. M.; van Amerongen, H.Biophys. J.

1995, 69, 2670.(43) Pascal, A.; Gradinaru, C.; Wacker, U.; Peterman, E.; Calkoen, F.;

Irrgang, K.-D.; Horton, P.; Renger, G.; Van Grondelle, R.; Robert, B.; VanAmerongen, H.Eur. J. Biochem.1999, 262, 817.

(44) Christensen, R. L.; Kohler, B. E.Photochem. Photobiol.1973, 18,293.

(45) Kohler, B. E. Electronic structure of Carotenoids. InCarotenoids;Britton, G., Liaaen-Jensen, S., Pfander, H., Eds.; Birkha¨user Verlag: Basel,Switzerland, 1995; Vol. 1B, p 3.

(46) Papagiannakis, E. Shedding light on the dark states of carotenoids.Ph.D. Dissertation, Free University of Amsterdam, Amsterdam, 2004.

(47) Holzwarth, A. R. Data analysis of time-resolved measurements. InBiophysical Techniques in Photosyntesis; Amesz, J., Hoff, A. J., Eds.;Advances in Photosynthesis 3; Kluwer Academic Publishers: Dordrecht,The Netherlands, 1996; p 75.

(48) van Stokkum, I. H. M.; Larsen, D. S.; van Grondelle, R.Biochim.Biophys. Acta2004, 1657, 82.

(49) Macpherson, A. N.; Arellano, J. B.; Fraser, N. J.; Cogdell, R. J.;Gillbro, T. Biophys. J.2001, 80, 923.

(50) Pendon, Z. D.; Gibson, G. N.; van der Hoef, I.; Lugtenburg, J.;Frank, H. A.J. Phys. Chem. B2005, 109, 21172.

(51) de Weerd, F. L.; van Stokkum, I. H. M.; van Grondelle, R.Chem.Phys. Lett.2002, 354, 38.

(52) Billsten, H. H.; Zigmantas, D.; Sundstro¨m, V.; Polıvka, T. Chem.Phys. Lett.2002, 355, 465.

(53) Tavan, P.; Schulten, K.J. Chem. Phys.1986, 85, 6602.(54) Tavan, P.; Schulten, K.Phys. ReV. B 1987, 36, 4337.(55) Tsukida, K.Methods Enzymol.1992, 213, 291-298.(56) Sashima, T.; Koyama, Y.; Yamada, T.; Hashimoto, H.J. Phys.

Chem. B2000, 104, 5011.(57) Carotenoids; Isler, O., Gutmann, H., Solms, U., Eds.; Birkha¨user

Verlag: Basel, Switzerland, 1971.(58) Dreuw, A.J. Phys. Chem. A2006, 110, 4592.(59) Frank, H. A.; Bautista, J. A.; Josue, J.; Pendon, Z.; Hiller, R. G.;

Sharples, F. P.; Gosztola, D.; Wasielewski, M. R.J. Phys. Chem. B2000,104, 4569.

(60) Polli, D.; Cerullo, G.; Lanzani, G.; De Silvestri, S.; Hashimoto,H.; Cogdell, R. J.Biophys. J.2006, 90, 2486.

(61) Polli, D.; Cerullo, G.; Lanzani, G.; De Silvestri, S.; Hashimoto,H.; Cogdell, R. J.Synth. Met.2003, 139, 893.

(62) Englman, R.; Jortner, J.Mol. Phys.1970, 18, 145.

22884 J. Phys. Chem. B, Vol. 110, No. 45, 2006 Niedzwiedzki et al.

(63) Wasielewski, M. R.; Johnson, D. G.; Bradford, E. G.; Kispert, L.D. J. Chem. Phys.1989, 91, 6691.

(64) Hashimoto, H.; Koyama, Y.; Hirata, Y.; Mataga, N.J. Phys. Chem.1991, 95, 3072.

(65) Nagae, H.; Kuki, M.; Zhang, J. P.; Sashima, T.; Mukai, Y.; Koyama,Y. J. Phys. Chem. A2000, 104, 4155.

(66) Kosumi, D.; Yanagi, K.; Nishio, T.; Hashimoto, H.; M., Y.Chem.Phys. Lett.2005, 408, 89.

(67) Shreve, A. P.; Trautman, J. K.; Owens, T. G.; Albrecht, A. C.Chem.Phys. Lett.1991, 178, 89.

(68) Bondarev, S. L.; Bachilo, S. M.; Dvornikov, S. S.; Tikhomirov, S.A. J. Photochem. Photobiol., A1989, 46, 315.

(69) McCamant, D. W.; Kukura, P.; Mathies, R. A.J. Phys. Chem. A2003, 107, 8208.

(70) Yoshizawa, M.; Aoki, H.; Hashimoto, H.Phys. ReV. B 2001, 63,18301.

(71) Billsten, H. H.; Bhosale, P.; Yemelyanov, A.; Bernstein, P. S.;Polıvka, T. Photochem. Photobiol.2003, 78, 138.

(72) Billsten, H. H.; Sundstro¨m, V.; Polıvka, T.J. Phys. Chem. A2005,109.

Femtosecond Spectroscopy of Xanthophylls J. Phys. Chem. B, Vol. 110, No. 45, 200622885