Electron−Phonon Coupling in Solubilized LHC II Complexes of Green

J. Phys. Chem. B 2001, 105, 7115-7124

7115

Electron-Phonon Coupling in Solubilized LHC II Complexes of Green Plants Investigated by Line-Narrowing and Temperature-Dependent Fluorescence Spectroscopy J. Pieper,*,†,§ R. Scho1 del,§ K.-D. Irrgang,| J. Voigt,§ and G. Renger*,‡,| Institute of Physics, Humboldt UniVersity, 10099 Berlin, Germany, and Max-Volmer-Institute for Biophysical Chemistry and Biochemistry, Technical UniVersity, 10623 Berlin, Germany ReceiVed: January 22, 2001; In Final Form: April 26, 2001

Line-narrowed and temperature-dependent fluorescence spectra are reported for the solubilized trimeric lightharvesting complex of Photosystem II (LHC II). Special attention has been paid to eliminate effects owing to reabsorption and to ensure that the line-narrowed fluorescence spectra are virtually unaffected by hole burning or scattering artifacts. Analysis of line-narrowed fluorescence spectra at 4.2 K indicates that the lowest Qy-state of LHC II is characterized by weak electron-phonon coupling with a Huang-Rhys factor of ∼ 0.9 and a broad and strongly asymmetric one-phonon profile with a peak frequency ωm of 15 cm-1 and a width of Γ ) 105 cm-1. The 4.2 K fluorescence data are further consistent with the assignment of the lowest Qy-state at ∼ 680.0 nm and an inhomogeneous width of ∼80 cm-1 gathered from a recent hole-burning study (Pieper et al. J. Phys. Chem. A 1999, 103, 2412). The temperature dependence of the fluorescence spectra of LHC II is simulated using the low-energy Qy-level structure reported in the latter study as well as the parameters of electron-phonon coupling determined in the present study. Up to a temperature of 120 K, the calculations reveal that this model satisfactorily describes the basic features of the fluorescence spectra such as thermal broadening and, especially, the blue-shift of the fluorescence peak with increasing temperature. An unexpected red-shift of the fluorescence peak above 150 K is attributed to conformational changes of the protein environment. The shape of the temperature-dependent fluorescence spectra indicates that the low-energy Qystates are populated according to a Boltzmann distribution representing the thermal equilibrium of excitation energy.

1. Introduction The primary steps of photosynthesis comprise two types of reaction sequences: (a) the generation of electronically excited states by light absorption in pigment-protein complexes which act as antenna and (b) rapid excitation energy transfer to photoactive pigments bound in reaction center complexes (for reviews see van Grondelle et al.1 and G. Renger2). The major antenna of green plants is the light-harvesting complex of photosystem II (LHC II), which binds approximately 65% of the chlorophyll associated with PS II (for a review see Paulsen3). Its structure has been analyzed by electron crystallography to a resolution of 3.4 Å.4 The LHC II protein crystallizes as a trimer of subunits, each of which contains 12 or 13 Chl molecules arranged in two layers close to the upper and lower surfaces of the thylakoid membrane and at least two carotenoid molecules. The planes of the chlorine rings are oriented almost perpendicularly to the plane of the thylakoid membrane. Within each of the two layers the shortest center to center distances between the Chl molecules are in the range of ∼9-14 Å. Three membrane spanning R-helices and a fourth amphipathic helix located at the lumenal side form the protein backbone. However, the present structural resolution does not permit an unambiguous identification of Chl a and Chl b molecules. A tentative assignment was achieved on the basis of functional properties. * Corresponding author. † E-mail: [email protected]. ‡ E-mail: [email protected]. § Institute of Physics. | Max-Volmer-Institute for Biophysical Chemistry and Biochemistry.

Taking into account the fast singlet transfer from Chl a to Chl b and the efficient quenching of chlorophyll triplets by carotenoids that requires a Dexter mechanism, the seven molecules in close contact to carotenoids were identified as Chl a. The remaining more distant five molecules are assumed to be Chl b. This model, however, has recently been questioned by theoretical simulations of time-domain experiments.5,6 Nevertheless, the identities of five Chl molecules (Chl a1, a2, a3, b5, and b6) could be confirmed by spectroscopic studies of LHC II-mutants lacking single pigments while Chl b3 was found to be a Chl a.7 The availability of a structural model for trimeric LHC II prompted a variety of time-resolved spectroscopic studies8-12 revealing that Chl b f Chl a excitation energy transfer (EET) is ultrafast with kinetic components of ∼ 150 fs, 600 fs, and 10 ps at room temperature.12 A detailed understanding of this ultrafast dynamics requires a determination of the energy level structure of the excited electronic states, which is governed by the Chl-Chl and Chl-protein interactions. In addition, the coupling of the purely electronic transition of a pigment to lowfrequency vibrational modes of the protein matrix (phonons), also known as electron-phonon coupling, is of particular relevance for light-harvesting and EET in antenna complexes. The simultaneous creation or annihilation of a phonon during an optical transition permits the absorption of light quanta that do not match the transition frequency of a given pigment and, therefore, leads to a considerable increase of the spectral absorption cross section of a single pigment.13 Furthermore, interaction with phonons may induce relaxation of the excited

10.1021/jp010229g CCC: $20.00 © 2001 American Chemical Society Published on Web 06/23/2001

7116 J. Phys. Chem. B, Vol. 105, No. 29, 2001 states of an antenna complex and, thus, plays an important role in dissipation of excess energy during EET.14,15 In the frequency domain spectral hole-burning (HB) and fluorescence linenarrowing (FLN) have been proven to be valuable experimental tools for analyses of the strength of electron-phonon coupling as well as of the line shapes of zero-phonon line (ZPL) and phonon sideband (PSB) in amorphous hosts (for reviews see Jankowiak et al.16,17). The first detailed investigation of the Qy-energy level structure of trimeric LHC II was based on 77 K absorption, linear dichroism and circular dichroism spectra.18 Nine Qy-states were assigned in the region between ∼650 and 676 nm. The presence of an additional, relatively weak transition at ∼680 nm was reported based on hole-burning spectroscopy at 4.2 K19 and pump-probe experiments at room temperature.20,21 Subsequently, reabsorption studies22 as well as fluorescence linenarrowing and temperature-dependent absorption spectroscopies23 confirmed the existence of a 680 nm state. Quite recently, more detailed hole-burning experiments24 revealed the presence of three low-energy Qy-states, which are located at (677.1 ( 0.2), (678.4 ( 0.2) and (679.8 ( 0.2) nm. The inhomogeneous width of their absorption bands is (80 ( 10) cm-1. Their combined absorption intensity is equivalent to that of three Chl a molecules per LHC II trimer so that each state is most likely quite highly localized on one Chl a molecule of a subunit. For intersubunit coupling strength of 5 cm-1,22 the energetic separations of ∼30 cm-1 are most likely attributed to structural heterogeneity. As to the electron-phonon coupling, results of 4.2 K holeburning19,24 and fluorescence line-narrowing experiments23 have already been reported for trimeric LHC II. The results of the hole-burning study24 led to the conclusion that the lowest Qystate of this complex is characterized by weak electron-phonon coupling to a one-phonon profile with a mean phonon frequency of ωm ) 18 cm-1, a width of ∼25 cm-1 and S ) 0.8. Inspection of the line-narrowed fluorescence spectra yielded a phonon frequency of ωm ) 22 cm-1 and a strongly asymmetric onephonon profile with a width of ∼65 cm-1. Because the zerophonon line was interfered with by scattered laser light in the line-narrowed fluorescence spectra, an S factor of ∼0.6 was determined by fitting the low-energy wing of temperaturedependent absorption spectra of LHC II using the latter onephonon profile and assuming that the lowest state is located at 676 nm. Weak electron-phonon coupling with S < 1 and ωm values of 20-30 cm-1 are typically observed for photosynthetic antenna complexes (see, e.g., Reddy et al.25). Nevertheless, the shapes of the one-phonon profile reported in the above studies differ significantly. Recently, theoretical simulations26 of the hole-burning results of Pieper et al.24 and preliminary line-narrowed fluorescence data proposed a model that is able to account for the phonon structure of both types of line narrowed spectra. The present paper reports on more detailed line-narrowed and non-linenarrowed fluorescence data measured at 4.2 K that provide further support for the model proposed in ref 26. Special attention has been paid to eliminate effects owing to possible systematic errors. The electron-phonon coupling parameters presented here and the low-energy level structure reported in ref 24 are also used to describe the temperature dependence of the non-line-narrowed fluorescence spectra up to 120 K. Finally, deviations from these model simulations at higher temperatures are attributed to conformational changes of the protein matrix. This interpretation is in line with studies of the temperaturedependent absorption spectra of bacterial antenna complexes.27,28

Pieper et al. 2. Materials and Methods Sample Preparation. LHC II samples were isolated by solubilization of salt washed PS II membrane fragments of spinach in the presence of β-dodecyl maltoside (β-DM) and separation by sucrose density gradient centrifugation as described by Irrgang et al.29 The Qy-absorption spectrum at room temperature exhibited two bands with maxima at 652 ( 1 nm and 675 ( 1 nm. The room-temperature CD spectrum was identical to that of the trimeric LHC II reported by Peterman et al.30 A Chl a/Chl b ratio of 1.35 ( 0.05 was determined using the method of Porra et al.31 The polypeptide composition was checked by SDS/urea/PAGE as reported in Irrgang et al. 29 in combination with silver staining and immunoblotting experiments as described in detail by Vasiliev et al.32 Besides the major light-harvesting complex ascribed to Lhc b1-3 gene products, the β-DM solubilized LHC II also contains minor Chl a/b binding proteins known to be associated with the antenna system of photosystem II. According to densitometrical scanning of silver-stained polyacrylamide gels, about 87% of the apoproteins could be ascribed to Lhc b1-3 proteins and 13% to the minor pigment-protein complexes, mainly Lhc b5 (CP 24) and Lhc b6 (CP 26). The samples were diluted with a glass forming buffer solution containing 70% w/v glycerol. The total Chl concentration was 0.2 mg/mL. Experimental Setup. Fluorescence spectra were measured using a Nd3+-YAG laser pumped dye laser (LAS LDL 105) which delivered excitation pulses of about 3 ns duration, a repetition rate of 10 Hz and a spectral width of ∼0.002 nm. Focusing of the excitation spot and collection of the fluorescence signal were carried out as in Scho¨del et al.33 in order to ensure that the data are related to a well-defined excitation intensity (vide infra). The spot size was about 100 µm. The fluorescence signal was passing a double monochromator (GDM 1000, Carl Zeiss Jena) and monitored with an avalanche diode (Analogue Modules, Inc., type 712-A4). Signals were delivered into a boxcar integrator (gate width of 10 ns) and then read out by a personal computer. When measuring non-line-narrowed fluorescence spectra a resolution of the detection branch of 0.5 nm was used. The applied photon density was approximately 1013 photons per cm2 and pulse in order to avoid intensity dependent effects.33 The samples were contained in cuvettes of 10 µm optical path length. The resulting optical density of 0.02 at 676 nm and 4.2 K assured that the results remain virtually unaffected by reabsorption (see Results section). Unless otherwise noted, line-narrowed fluorescence spectra were measured using excitation wavelengths >680.0 nm, a photon density of approximately 1013 photons per cm2 and pulse as well as a spectral resolution of 0.2 nm. The samples were contained in cuvettes of 0.50 mm optical path length resulting in an optical density of 1.00 at 676 nm and 4.2 K. The latter parameters were carefully chosen in order to avoid contamination of the data with hole-burning or reabsorption artifacts (see Results section). The ZPL of the FLN spectrum is usually obscured by scattered laser light. To permit a subtraction of this contribution the spectral characteristics of the excitation beam was measured separately (see dashed line in Figure 3). This spectrum was normalized to the high-energy wing of the FLN spectra and then subtracted to yield the spectra shown in Figures 4 and 6. A commercial Shimadzu UV 3000 spectrophotometer with a spectral resolution of 2 nm was used for absorption measurements.

Solubilized LHC II Complexes of Green Plants

J. Phys. Chem. B, Vol. 105, No. 29, 2001 7117

The sample temperature was maintained using a Oxford Helium flow Cryostat (Optistat) and stabilized by an Oxford temperature controller. The sample was allowed to equilibrate for 20 min prior to fluorescence or absorption measurements. Theoretical Background. The theory describing the spectral properties of a pigment embedded in an amorphous host is well developed13,34 and has been extended to simulate holeburning35-37 and fluorescence line-narrowing34,38 spectra. The following paragraph gives a brief summary of the expressions used for the theoretical calculations of this study. For linear, harmonic Franck-Condon coupling to a distribution of delocalized phonons the homogeneously broadened absorption and fluorescence spectra in the low-temperature limit are given by35 ∞ e-S L(ω) ) e-Sl0(ω - Ω0) + SR lR(ω - Ω0 - Rωm) (1) R! R)1 ZPL PSB

∑

where -Rωm and +Rωm correspond to absorption and fluorescence, respectively. Here, l0(ω - Ω0) is the Lorentzian zerophonon line (ZPL) with peak position Ω0 and homogeneous width γ. The phonon sideband (PSB) consists of all R-phonon transitions (R ) 1, 2, ...) with normalized line shapes lR (ω Ω0 - Rωm) peaking at Ω0(Rωm for a mean phonon frequency of ωm. The intensity distribution is determined by a Poisson type weighting factor for every R where the HuangRhys factor S is a measure for the strength of electron-phonon coupling. The one-phonon profile l1(ω - Ω0 - ωm) represents the product g(ω)D(ω) with g(ω) being the density of states of the phonon modes and D(ω) an electron-phonon coupling term. Each profile lR (R > 1) is then completely determined by the shape of l1 and obtained by folding l1 R times with itself. Especially, for l1 being a Gaussian (Lorentzian) with a full width at half-maximum (fwhm) ΓGauss (ΓLorentz) the profile lR becomes a Gaussian (Lorentzian) with fwhm xR ΓGauss (RΓLorentz).35 The spectrum for the whole ensemble of pigments is obtained by convolution of the homogeneously broadened fluorescence spectrum with an inhomogeneous distribution function (IDF) N(Ω0 - ωC) peaking at ωC that is assumed to exhibit Gaussian shape and is characterized by its fwhm Γinh. Then, the inhomogeneously broadened low-temperature fluorescence spectrum is given by ∞

L(ω) )

∑

( )∫ SR

R)0

e-S R!

dΩ0 N(Ω0 - ωC)lR(ω - Ω0 + Rωm) (2)

On the basis of the approaches of Mc Colgin38 and Personov,34 the low-temperature FLN spectrum F(ω) is given by ∞

L(ω) )

∑

R,P)0

( )( )∫ SR

e-S R!

SP

e-S P!

dΩ0 N(Ω0 - ωC)lR(ω - Ω0 + Rωm)lP(ωE - Ω0 - Pωm) (3)

with ωE being the excitation frequency. The line shape function lP represents those ZPL which are resonantly (P ) 0) and nonresonantly (P * 0) excited, respectively, while lR is the homogeneously broadened fluorescence spectrum. For arbitrary temperatures thermal population of phonon levels according to the Bose-Einstein statistics has to be considered. Accounting for this effect, the following expression

can be obtained for the inhomogeneously broadened fluorescence spectrum37

L(ω) ) e

-

∑

jk+1) kSk(2n

∞

R

∑∑ ∏ k R)0 r ) 0

[Sk(njk + 1)]R-r[Sknjk]r (R - r)!r!

×

∫ dΩ0N(Ω0 - ωC)lR,r[ω - Ω0 + ∑(R - 2r)ωk]

(4)

k

where njk ) [exp(pωk/kT) - 1]-1 denotes the thermal occupation number of phonons with the frequency ωk. The product over k allows for inclusion of different phonon modes. The factors Sk(njk + 1) and Sknjk represent phonon creation and annihilation, respectively. As in eq 1, R denotes the total number of phonon transitions (R ) 1, 2, ...) while r gives the number of annihilated phonons (0 ) r ) R). Accordingly, optical transitions with more phonons being created than annihilated (R - r > r) give rise to the Stokes-part while those with more phonons being annihilated than created (R - r < r) constitute the anti-Stokes part of the PSB. These spectral features appear on the low- and high-energy side of the ZPL, respectively. Those transitions with the same number of created and annihilated phonons (R - r ) r) are coincident with the ZPL. The profile lR,r (R > 1) is obtained by folding the one-phonon profile l1,0 |R - 2r|-times with itself. Then, for l1,0 being a Gaussian (Lorentzian) with a width ΓGauss (ΓLorentz) the profile lR,r becomes a Gaussian (Lorentzian) with a width of |R-2r|1/2 ΓGauss (|R - 2r| ΓLorentz).37 In the lowtemperature limit (T f 0 K) njk tends to zero and eq 4 reduces to eq 2. As in eq 2 the homogeneously broadened spectrum is folded with the IDF N(Ω0 - ωC) to account for the heterogeneity of the amorphous host. Calculation programs to simulate the low-temperature nonline-narrowed fluorescence (eq 2) and FLN (eq 3) spectra as well as the inhomogeneously broadened fluorescence spectrum for arbitrary temperature (eq 4) were written in Mathematica 2.2.1. The infinite sums were cut at R values sufficient to account for at least 99% of the full fluorescence (absorption) intensity at a given temperature. Most calculations of this study were carried out for a temperature of 4.2 K where a summation up to R ) 5 for S ) 0.9 (see Discussion section) assured that 99.99% of the full intensity is considered. In eqs 2 and 4 the proper folding of the narrow ZPL is omitted. Rather, the inhomogeneously broadened ZPL is taken to be a Gaussian having a width of Γinh. This approximation requires that Γinh . Γhom. This appears to be justified for values of Γhom ) 0.073 cm-1 and Γinh ∼ 80 cm-1 for the lowest (fluorescing) state of LHC II determined in a recent hole-burning study.24 When calculating FLN spectra (eq 3), only terms for R > 0 are considered because the ZPL is obscured by scattered laser light in the experimental data (see Results section). Finally, guided by the previous work of Hayes et al.37 and the experimental data of this study (see Results section) the one-phonon profile l1 is generally assumed to be asymmetric with a Gaussian shape at its low-energy wing and a Lorentzian shape at its high-energy wing. Then, the full profile has a peak frequency of ωm and a width of Γ ) ΓGauss/2 + ΓLorentz/2. 3. Results To analyze the emission properties of isolated trimeric LHC II three different types of experiments were performed: (a) absorption and emission spectroscopy at 4.2 K with special emphasis on the elimination of interference by reabsorption effects, (b) temperature-dependent fluorescence spectroscopy, and (c) fluorescence line-narrowing spectroscopy at 4.2 K.

7118 J. Phys. Chem. B, Vol. 105, No. 29, 2001

Figure 1. The 4.2 K absorption (dashed line) and normalized fluorescence spectrum (full line) of trimeric LHC II. The inset shows 4.2 K fluorescence spectra obtained for samples of different optical density (different optical path length of cuvettes), i.e., (a) 0.02 (0.01 mm), (b) 0.2 (0.1 mm), (c) 0.4 (0.2 mm), (d) 1.0 (0.5 mm), and (e) 2.0 (1.0 mm) at 676 nm and a constant chlorophyll concentration of 0.2 mg/mL (Spectrum a is the same as that shown in the main part.). The spectra are normalized at 740 nm.

Absorption and Fluorescence Spectra at 4.2 K. Figure 1 shows the 4.2 K absorption and fluorescence spectra of trimeric LHC II. The absorption spectrum (see dashed line) exhibits maxima at 649.5, 661.4, 671.3, and 676.1 nm. The fluorescence spectrum (see full line) was nonselectively excited at 430 nm (Spectra obtained under such conditions have to be distinguished from those selectively excited within the main fluorescence band. In the following, the former and latter will be referred to as non-line-narrowed and line-narrowed fluorescence spectra, respectively.). The optical density (OD) of the sample was 0.02, which is sufficiently low to avoid systematic errors due to reabsorption (vide infra). The non-line-narrowed fluorescence spectrum is characterized by a broad and strongly asymmetric band peaking at 680.3 ( 0.2 nm with a full width at half-maximum (fwhm) of 5.5 ( 0.2 nm. The main fluorescence band is accompanied by a number of partially resolved satellite features (e.g., at ∼697, ∼715, ∼741, and 769 nm) which were shown to correspond to intramolecular vibrations of Chl a.23,39 In general, both absorption and non-line-narrowed fluorescence spectra, are similar to those presented in refs 24 and 40. This correspondence confirms that the preparation studied here is comparable to the solubilized LHC II complexes used in the former studies. The peak positions and the absence of a shoulder at ∼ 695 nm in the fluorescence spectrum clearly indicate that this preparation is virtually free of aggregation effects.32,40 The main fluorescence band at 680.3 nm overlaps the sharply decreasing low-energy wing of the absorption spectrum. A recent hole-burning study revealed that the lowest (fluorescing) Qy-state of LHC II is located at ∼ 679.8 nm,24 i.e., the energetic position of this state is almost resonant to the fluorescence maximum. In this case strong reabsorption of the emitted fluorescence can be expected. To address this essential problem fluorescence measurements were performed using LHC II samples with different optical density. The inset of Figure 1 shows spectra obtained varying the optical path length at a constant chlorophyll concentration. To allow for direct comparability, the data were normalized at 740 nm where effects due to reabsorption can be neglected. An inspection of the results reveals that two striking features emerge with decreasing OD: (i) a marked increase of emission in the peak region and (ii) a small but discernible blue-shift as well as a narrowing of the

Pieper et al.

Figure 2. Normalized fluorescence spectra of LHC II excited at 430 nm for different temperatures (see labels). The spectra are separated by an equidistant offset of 0.2. The inset shows the position of the fluorescence maximum as a function of temperature.

main fluorescence band. Variation of the chlorophyll concentration at constant optical path length led to consistent results (not shown). As a consequence of these findings it is of key relevance to select experimental conditions which ensure a sufficient elimination of effects owing to reabsorption. A comparison of the fluorescence spectra measured at samples with OD values of 0.02 and 0.2, however, reveals no discernible changes, within experimental uncertainty. Therefore, the 4.2 K fluorescence spectra obtained for LHC II samples with an OD of 0.02 appear to be virtually unaffected by reabsorption. Consequently, measurements of non-line-narrowed fluorescence spectra were performed for samples with an OD of 0.02 and the spectra obtained are suitable for comparison with model calculations and line-narrowed spectra (see Discussion section). Temperature Dependence of Fluorescence Spectra. Figure 2 shows fluorescence spectra of LHC II measured at temperatures between 4.2 and 300 K. The most striking feature of these data is the temperature dependence of the fluorescence maximum (see inset of Figure 2). Two characteristics are observed with increasing temperature: (a) at low temperatures (T < 120 K) there is a blue shift from 680.3 nm at 4.2 K to 678.9 nm at 120 K and (b) at higher temperatures (T > 120 K) the fluorescence maximum shifts continuously back to the red to a peak position of 682.0 nm at 300 K. In addition to these shifts, the main fluorescence band broadens from 5.5 nm at 4.2 K to 8.4 nm at 80 K and 17.5 nm at room temperature and its shape becomes more symmetric with increasing temperature. These results are similar to those of Peterman et al.23 Line-Narrowed Fluorescence Spectra at 4.2 K. Linenarrowed fluorescence (FLN) spectra were obtained at a temperature of 4.2 K using laser pulses with a narrow line width (fwhm of ∼0.002 nm) and excitation wavelengths (λE) located within the main fluorescence band (λE > 680 nm). A typical FLN spectrum excited at 680.5 nm with a photon density of 1013 photons per cm2 and pulse is shown as the upper trace in Figure 3 and labeled by ”FLN”. Before describing the results in detail, however, the discussion will be focused on possible systematic errors such as scattering, hole burning, or reabsorption artifacts. As to the first, it is obvious from Figure 3 that the ZPL of the FLN spectrum is obscured by scattered laser light. This effect is characteristic for FLN experiments because the excitation wavelength λE is coincident with the spectral position of the ZPL. To quantitatively analyze the contribution of the scattered laser light to the entire FLN spectrum separate experiments were

Solubilized LHC II Complexes of Green Plants

Figure 3. Selectively excited fluorescence spectra at 4.2 K (full lines). The upper FLN spectrum (labeled by “FLN”) was excited at 680.5 nm with a photon density of ∼1013 photons per cm2 and pulse. The lower spectrum (labeled by “FLN & HB”) was obtained under identical conditions but after previous exposure of the sample to laser pulses with a photon density of ∼1014 photons per cm2 and pulse for a total illumination time of 300 s. The dashed line gives the spectral characteristics of the scattered laser light.

performed replacing the LHC II sample by a reflecting, nonemitting paper surface. The spectral characteristics of the excitation beam determined in these experiments is shown as a dashed line in Figure 3, where it is normalized to the high energy wing of the FLN spectrum. On the basis of this result it can be concluded that the phonon sideband of the FLN spectrum is only slightly contaminated with scattering artifacts. Subtraction of the dashed line from the FLN spectrum readily yields the PSB (R-phonon transitions with R > 0). Furthermore, hole-burning was shown to be very efficient for burn wavelengths λB > 675 nm.19,24 Consequently, excitation of fluorescence in this spectral region may lead to simultaneous hole-burning and, as a result, to a subsequent decrease of the fluorescence yield. To address this problem, experiments were performed at different photon densities. Typical results obtained are shown in Figure 3. The upper FLN spectrum of Figure 3 (labeled by ”FLN”) was excited at 680.5 nm with a photon density of 1013 photons per cm2 and pulse. No changes in fluorescence yield were observed during a series of 10 consecutive spectral scans within a total illumination time of up to 20 min, i.e., hole-burning effects appear to be negligible. In contrast to this, the lower spectrum of Figure 3 (labeled by “FLN & HB”) was gained from a similar FLN measurement but after previous exposure of the sample to laser light with a wavelength of 680.5 nm and a higher photon density of 1014 photons per cm2 and pulse for a total illumination time of 300 s. The latter spectrum exhibits a similar shape as the upper spectrum but reduced fluorescence intensity due to hole-burning at the excitation wavelength. On the basis of the above-mentioned results, the wavelength dependence of the FLN spectra was studied using a photon density of 1013 photons per cm2 and pulse. Finally, effects owing to reabsorption were studied in an analogous way as described for the non-line-narrowed fluorescence spectrum above. Figure 4 shows FLN spectra obtained for an excitation wavelength of 680.5 nm and a photon density of 1013 photons per cm2 and pulse, but different OD values of the sample (see figure caption). The scattered laser light was subtracted before normalizing the spectra. Obviously, variation of the OD did not lead to changes of the spectral shape of the line-narrowed fluorescence. Thus, it can be concluded that the FLN spectra shown in Figure 4 are virtually free of reabsorption effects. The best possible signal-to-noise ratio as well as the

J. Phys. Chem. B, Vol. 105, No. 29, 2001 7119

Figure 4. FLN spectra generated with an excitation wavelength of 680.5 nm and a photon density of ∼1013 photons per cm2 and pulse at a temperature of 4.2 K. The spectra were obtained for samples with an optical density (different optical path length of cuvettes) of 0.2 (dashed line), 0.4, and 1.0 (full lines) at 676 nm and a constant chlorophyll concentration of 0.2 mg/mL. The scattered laser light was subtracted before normalizing the spectra.

Figure 5. FLN spectra (full lines) obtained at a temperature of 4.2 K with laser pulses of a photon density of ∼1013 photons per cm2 and pulse and excitation wavelengths of 680.5, 681.5, and 682.5 nm from left to right, respectively. Arrows indicate the presence of a shoulder in the FLN spectra at about 90 cm-1 relative to λE. The 4.2 K nonline-narrowed fluorescence spectrum (dashed line) excited at 645 nm is given for comparison.

lowest contribution from scattered laser light, however, can be expected for the highest tested OD of 1.0 at 676 nm and 4.2 K. In summary of the results presented above, FLN measurements were routinely performed employing excitation wavelengths >680 nm, a photon density of 1013 photons per cm2 and pulse as well as using LHC II samples with an OD of 1.0 at 676 nm and 4.2 K. Figure 5 shows line-narrowed fluorescence spectra of LHC II obtained for different excitation wavelengths (λE) within the fluorescence origin band at 4.2 K under the above-described and carefully chosen experimental conditions. For the sake of direct comparability the spectra are normalized by the factor Iexc(680.5 nm)/Iexc(λE), where Iexc(680.5 nm) and Iexc(λE) symbolize the photon densities at 680.5 nm and a particular excitation wavelength λE. The non-line-narrowed fluorescence spectrum excited at 645 nm is given for comparison as a dashed line in Figure 5. FLN experiments were restricted to λE > 680 nm in order to minimize the extent of nonresonant excitation.41 In addition, for λE < 680 nm, excitation of electronic states lying at higher energies followed by excitation energy transfer to the lowest state has to be taken into account. As a result, the FLN spectrum would be superimposed with contributions from

7120 J. Phys. Chem. B, Vol. 105, No. 29, 2001

Pieper et al.

the non-line-narrowed fluorescence spectrum causing an additional loss of selectivity. An inspection of the data reveals that the FLN spectra peak about 24 cm-1 to the red of the excitation wavelength (λE). This finding indicates that the lowest (fluorescing) Qy-state of LHC II is mainly inhomogeneously broadened and the FLN spectrum is predominantly due to electron-phonon coupling. Furthermore, the shape of the FLN spectra is characterized by a remarkable asymmetry (width of ∼100 cm-1) which resembles that of the non-line-narrowed fluorescence spectrum. Tuning the excitation wavelength to the red, the fluorescence intensity decreases drastically. This observation is qualitatively consistent with the relatively narrow inhomogeneous width of the absorption band of the lowest state of 80 cm-1 24. 4. Data Fits and Discussion When describing the spectral properties of a pigment embedded in a protein matrix in the low-temperature limit basically two effects have to be taken into account: (a) the electronic transition of a pigment at a single site of the matrix is coupled to the delocalized vibrations (phonons) of the protein matrix, and (b) the differences of the microenvironment at each site give rise to inhomogeneous broadening (see, e.g., Rebane13 and Personov34). The lowest (fluorescing) state of trimeric LHC II, which is investigated here, was shown to be quite highly localized on one Chl a molecule at 4.2 K24 so that Chl-Chl interactions within the LHC II complex (see, e.g., Th. Renger14,42 and references therein) will not be taken into account. A thorough characterization of electron-phonon coupling requires (see Theoretical background) the determination of the position of the lowest state ωC (or λC, respectively), its inhomogeneous width Γinh, the Huang-Rhys factor S as well as the spectral shape of the one-phonon profile given by ωm, ΓG, and ΓL. Not all of these parameters can be simultaneously obtained from the measured FLN spectra. Especially, because of the superposition of the ZPL with scattered laser light and the complex composition of the phonon wing in FLN spectra (see, e.g. Pieper et al.26 and references therein) the values of λC, Γinh, and S are not directly reflected in the experimental data. As a consequence, the simulation of the fluorescence and FLN data will be carried out in three steps. First, the one-phonon profile (ωm, ΓG, and ΓL) will be derived from the 4.2 K FLN spectra by using values of λC, Γinh, and S which were determined in a recent hole-burning study. 24 In a second step, the 4.2 K fluorescence spectrum is calculated in order to verify all parameters including λC, Γinh and the coupling strength S. Finally, the parameters of electron-phonon coupling determined in the first two steps as well as the low-energy level structure reported by Pieper et al.24 will be used to simulate the temperature dependence of the fluorescence spectra of LHC II. Simulations of Fluorescence Spectra at 4.2 K. Recent holeburning experiments24 revealed that the lowest state of trimeric LHC II is located at λC ) (679.8 ( 0.2) nm and possesses an inhomogeneous width of Γinh ) (80 ( 10) cm-1. In addition, a Huang-Rhys factor S ) 0.8-0.9 was gathered from the saturated hole depth at the low-energy wing of the absorption profile of the lowest state. To derive the one-phonon profile from the 4.2 K FLN spectra, the values of λC, Γinh, and S were varied within the range of the experimental uncertainty while ωm, ΓG, and ΓL defining the one-phonon profile were treated as free parameters. The pronounced asymmetry of the FLN spectra suggests that the one-phonon profile is also characterized by an asymmetric shape (see Theoretical background).

Figure 6. Experimental FLN spectra of LHC II at 4.2 K (full, noisy lines) generated with excitation wavelengths of 680.5, 681.5, and 682.5 nm from left to right, respectively. Scattered laser light has been subtracted from the FLN spectra. Calculated spectra (dashed lines) were obtained for the parameters given in Table 1. The dashed-dotted line represents the IDF profile. Experimental (full line) and calculated (dashed line) non-line-narrowed fluorescence spectra for a temperature of 4.2 K as well as the one-phonon profile a (see bottom) are shown in the inset.

TABLE 1: Parameters of the One-Phonon Profiles Used to Fit the 4.2 K FLN Spectra (Profile a, See Figure 6) and the Temperature-Dependent Fluorescence Spectra (Profiles b and c, See Figure 8), the Sum of Profiles b and c Equals Profile a profile a profile b profile c temperature lowest state(s) Huang-Rhys factor peak phonon frequency fwhm of Gaussian wing fwhm of Lorentzian wing fwhm, one-phonon profile

T [K]

4.2

S ωm [cm-1] Γg [cm-1] ΓL [cm-1] Γ [cm-1]

0.9 15 10 200 105

>4.2 see Table 2 0.22 15 10 40 25

0.68 60 60 200 130

The full, noisy lines in Figure 6 represent the phonon wing of the 4.2 K FLN spectra of Figure 5. The scattered laser light (see dashed line in Figure 3) was subtracted by the procedure given in the Results section. The calculated FLN spectra (see dashed lines in Figure 6) were obtained according to eq 3 for ωC ) 14 705 cm-1 (680.0 nm), ωE ) 14 695.1 cm-1 (680.5 nm), Γinh ) 80 cm-1, S ) 0.9 and a one-phonon profile with ωm ) 15 cm-1 and Γ ) 105 cm-1 (ΓG ) 10 cm-1, ΓL ) 200 cm-1). The parameters are also given as profile a in Table 1. A plot of the one-phonon profile is shown in the inset of Figure 6. In general, the fits are in reasonable agreement with the experimental data obtained for different excitation wavelengths. At the first glance the value of ωm ) 15 cm-1 seems to be surprisingly small compared to the peak frequency of the measured FLN spectra. For the relatively high S value of 0.9, however, higher phonon transitions (R > 1) contribute significantly to the entire phonon wing. Thus, the observed peak of the FLN spectra at 24 cm-1 results mainly from the superposition of the one and the two-phonon transitions at 15 and 30 cm-1, respectively. Therefore, the fit is very sensitive to the position of ωm. The width (Γ ) 105 cm-1) and, especially, the strong asymmetry (ΓL ) 200 cm-1) of the one-phonon profile are essential to account for the shape of the FLN spectra. The inset of Figure 6 shows the 4.2 K non-line-narrowed fluorescence spectrum of LHC II calculated according to eq 2 for the values given above (see profile a in Table 1). In general, the agreement with the experimental data significantly improved in comparison with previous studies.23,43 Obviously, the fit accounts for the peak position and the width of the fluorescence

Solubilized LHC II Complexes of Green Plants

Figure 7. Part A shows a saturated hole-burned spectrum of LHC II (from ref 24) obtained with λB ) 681 nm and a burn fluence of 15 J/cm2. The real- and pseudo-PSB hole at +24 and -18 cm-1 relative to the ZPH are indicated by dashed and full arrows, respectively. Part B gives a magnification of the hole-burned spectrum of Frame A for ease of inspection of the PSB features.

origin band as well as for its pronounced asymmetry. Thus, it provides further support for the electron-phonon coupling parameters determined by fitting the 4.2 K FLN spectra. Moreover, the assignment of the lowest Qy-state at ∼680 nm and an inhomogeneous width of ∼80 cm-1 reported by Pieper et al.24 can be confirmed. Slight deviations at the low-energy wing may originate from different effects. First, the arbitrarily chosen Gaussian/Lorentzian shape may not exactly describe the actual one-phonon profile. Especially, the FLN spectrum calculated for an excitation wavelength of 682.5 nm, i.e., the spectrum providing the highest spectral selectivity, appears to be too broad in the peak region while it does not fully account for the weak shoulder at ∼90 cm-1 (vide infra). Another possibility is that vibrational satellites resulting from lowfrequency intramolecular Chl modes (>250 cm-1) may contribute to the fluorescence in this spectral region. Such modes are not considered in the calculations. As mentioned in the Introduction, the phonon sideband structure observed in FLN and hole-burning spectra of LHC II exhibits apparent differences in shape and peak position. The noisy line in part A of Figure 7 is a typical hole-burning spectrum taken from ref 24. It was obtained for a burn wavelength λB ) 681 nm and a burn fluence of 15 J/cm2 at a temperature of 4.2 K. This spectrum is saturated, i.e., the holes have attained maximum depth. The real-phonon sideband hole indicated by the dashed arrow is interfered with by the antihole. Therefore, the analysis should be based on the pseudo-phonon sideband hole marked by the full arrow. The latter peaks ∼18 cm-1 to the red of the zero-phonon hole (ZPH) and has a width of only ∼25 cm-1 (see part B of Figure 7 for a magnification). At the first glance these data seem to be in contrast with the much broader (Γ ) 105 cm-1) one-phonon profile obtained from the FLN data. Surprisingly, a hole-burning spectrum calculated for the above data set (profile a in Table 1) is in complete agreement with the experimental data (see full, smooth line in Figure 7). An explanation for this effect has recently been given in a theoretical study.26 It was based on the fact that the inhomogeneous broadening of ∼80 cm-1 is smaller than the width of the one-phonon profile of ∼105 cm-1 for LHC II. Briefly, in this case (Γ J Γinh) the low-energy wing of the pseudo-PSB hole was shown to be determined by the IDF rather than by the shape of the one-phonon profile. Consequently, the pseudo-PSB hole appears to be narrower than the FLN spectrum.

J. Phys. Chem. B, Vol. 105, No. 29, 2001 7121 In summary, the 4.2 K FLN and non-line-narrowed fluorescence data of the present study as well as the hole-burning data from ref 24 can be simultaneously described by weak electronphonon coupling with a Huang-Rhys factor S of 0.9 to a broad and strongly asymmetric one-phonon profile with a peak frequency ωm of 15 cm-1 and a width of Γ ) 105 cm-1. Weak coupling strengths (S < 1) and one-phonon profiles peaking at ∼20 cm-1 are generally reported for photosynthetic antenna complexes (for a review see, e.g., Reddy et al.25). however, a one-phonon profile having a width of 105 cm-1 cannot be understood in terms of a mean phonon frequency but rather represents a broad distribution of modes.44 As to the nature of the modes constituting the one-phonon profile, it is wellestablished that intramolecular vibrations of Chl a appear in the frequency range of >250 cm-1 23,45,46 with S factors lower than 0.04.46 Thus, the former are most likely attributed to delocalized vibrations of the amorphous protein matrix (see e.g. Hayes et al.47). On the other hand, it is possible that localized vibrations contribute to the high-frequency tail of the onephonon profile (>50 cm-1). The best example for such a mode is probably the marker mode (115-135 cm-1) observed for bacterial reaction centers which was assigned to an intermolecular vibration of the special pair.48 Similar modes with frequencies of about 80 cm-1 were reported from FLN spectra of the PS II reaction center49 and the CP47 core antenna complex.50 Recent hole-burning studies revealed a mode with a frequency of about 70 cm-1 for the PS I of the cyanobacterium Synechocystis51 and several localized modes with frequencies in the range of 20-165 cm-1 for the FMO complex of Chlorobium tepidum.52 In this regard it is interesting to note that a weak shoulder is observed at about 90 cm-1 in the FLN spectra of LHC II (see arrows in Figure 5). Unfortunately, the present resolution as well as the wide and structureless form of the one-phonon-profile determined above preclude an identification of distinct modes in the range of 50-200 cm-1. Thus, it has to be emphasized that all modes constituting the highfrequency tail of the one-phonon profile cannot be unambiguously assigned to phonon modes of the protein or to localized vibrations. Simulations of Temperature-Dependent Fluorescence Spectra for T < 120 K. An inspection of the experimental data reveals that two temperature regions have to be distinguished with respect to effects on the fluorescence spectra. A marked change arises at a temperature of 120 K. Therefore, the ranges below and above this characteristic temperature will be discussed separately. In general, for temperatures higher than 4.2 K thermal occupation of both higher electronic states and phonon modes has to be accounted for. Both effects are qualitatively observed in the fluorescence spectra shown in Figure 8. The blue-shift of the fluorescence peak observed for increasing temperature up to 120 K suggests that higher electronic states are subsequently populated with increasing temperature. Moreover, the onset of the blue-shift at a temperature as low as 30 K (kT ∼ 21 cm-1) points to a very small energetic difference between the adjacent state(s). The more symmetric shape of fluorescence origin band at temperatures g60 K can at least partly be attributed to thermal occupation of phonon modes giving rise to the anti-Stokes part of the PSB. The simulations presented in Figure 8 were based on the lowenergy Qy-level structure and corresponding oscillator strengths reported from recent hole-burning experiments.24 It was found in the latter study that the lowest Qy-state at ∼680.0 nm is accompanied by two higher states at 678.4 and 677.1 nm,

7122 J. Phys. Chem. B, Vol. 105, No. 29, 2001

Pieper et al. TABLE 2: Parameters of the Low-Energy Qy-Levels of Trimeric LHC II Used to Obtain the Fits Shown in Figure 8 adjacent Qy-state

Qy-state position homogeneous width inhomogeneous width absorption intensity per LHC II trimer

λC [nm] γ [cm-1] Γinh [cm-1] chl a

∼ 676.0 a ∼80 6

lowest-energy Qy-states of the trimer subunits 677.1 678.4 680.0 80 1

80 1

80 1

a The fit procedure used (see Theoretical Background) requires that γ , γinh.

Figure 8. Normalized experimental (full lines) and calculated (dashed lines) fluorescence spectra of LHC II at different temperatures. The spectra were given equidistant offsets for ease of inspection. The calculated spectra were obtained for the parameters of Table 1 and the low-energy level structure of Table 2. The one-phonon profiles b and c are shown at the bottom and labeled by their peak frequencies of 15 and 60 cm-1, respectively.

respectively. These three states were assumed to be associated with the three subunits of the LHC II trimer and the energetic separations of ∼ 30 cm-1 were ascribed to structural heterogeneity. Each of the three states was reported to carry the absorption intensity of approximately one Chl a molecule. The absorption intensity of the next higher Qy-state at ∼676.0 nm was estimated to be equal to that of six Chl a molecules per LHC II trimer. In the following it will be assumed that rapid thermal equilibration of excitation energy occurs among these states. Consideration of only these four Qy-states appears to be justified because kT ∼ 83 cm-1 at 120 K which almost equals the energetic difference between the 680 and the 676 nm-state. Thus, the 676 nm-state is the highest significantly populated Qy-level at 120 K. To reduce the number of parameters it will be assumed that all four states are characterized by an inhomogeneous broadening of 80 cm-1 and the electron-phonon coupling parameters of the lowest state reported in the previous paragraph. The theory of Hayes et al.35,37 (eq 4) is used to simulate the temperature dependence of electron-phonon coupling for each state. In the temperature range between 4.2 and 120 K the value of kT increases from 3.5 to 83 cm-1 while the one-phonon profile has a width of 105 cm-1. To partly account for the varying thermal occupation of the phonon modes, the onephonon profile (see bottom of the inset in Figure 6) was arbitrarily divided into two profiles (see bottom of Figure 8) with maxima at 15 and 60 cm-1 (see profiles b and c in Table 1), respectively. Summation over the two profiles yields almost exactly the one-phonon profile shown in the inset in Figure 6. The separate profile peaking at 60 cm-1 has no special physical meaning but appeared to be a good intermediate value between kT at 4.2 and 120 K, respectively. The 4.2 K fluorescence spectrum calculated for the double profile (see Figure 8) is fully equivalent to that shown in the inset in Figure 6. On the basis of these assumptions (see Table 2), the fluorescence spectrum of each state was calculated for a given temperature according to eq 4 and multiplied by its individual oscillator strength and thermal population according to the Boltzmann distribution. After summation of the fluorescence bands of the four states the spectrum is finally normalized and compared to the experimental data. It has to be emphasized that no adjustable parameters enter into the calculations. Figure 8

shows the measured non-line-narrowed (full lines) and calculated fluorescence spectra (dashed lines) for temperatures between 4.2 and 120 K. The calculated spectra reflect the basic features of the temperature-dependent fluorescence spectra such as thermal broadening, more symmetric shape and blue-shifting of the fluorescence peak with increasing temperature. Especially, the blue-shift of the maximum is in almost perfect agreement with the experimental data. For temperatures g 60 K, however, the simulated spectra appear to be too broad. There are two possible reasons for this phenomenon. First, the oscillator strength of the 676 nm-state might be slightly overestimated in ref 24. Second, the arbitrarily divided one-phonon profile represents, strictly speaking, a two phonon frequency approximation, only. While the thermal population is exactly calculated for the mean frequency of each of the two profiles, phonon modes of lower/higher frequency within the profile attain a population that is too low/ too high at a given temperature. This inaccuracy is more critical for the profile peaking at 60 cm-1 because of its width of more than 100 cm-1. Consequently, the calculated spectra with this mode being significantly populated, i.e., for T g 60 K become too broad. Nevertheless, it can be concluded that the temperature-dependent fluorescence spectra of LHC II are satisfactorily described between 4.2 and 120 K based on the Qy-level structure of ref 24 as well as the electron-phonon coupling parameters determined in this study. Effects of Temperatures above 120 K. The temperature dependence of the LHC II fluorescence spectrum undergoes a drastic change above 120 K. A further blue-shifting would be expected with increasing temperature and subsequent population of higher Qy-states. In contrast to this, a red-shift of the fluorescence peak is observed between ∼150 K and room temperature. A similar phenomenon was reported for the B850 absorption band of the LH2 antenna complexes of Rb. sphaeroides and Rps. acidophila27 as well as the B875 absorption band of the LH1 antenna complex of Rb. sphaeroides.28 In these systems, the energetic location was found to be invariant, within experimental uncertainty, below ∼150 K, but shifting linearly to the blue at higher temperatures. In addition different rates of thermal broadening were observed above and below ∼150 K, respectively. Because these changes set in close to the glass temperature (Tg) it was suggested that conformational changes of the protein environment can arise due to a decrease in viscosity above Tg.27,28 Such structural changes may alter both, the pigment-protein interactions as well as excitonic coupling between the pigments. The latter effect was found to be more important for the B850 and B875 bands in refs 27 and 28. In the case of LHC II, theoretical calculations within the density matrix formalism14 revealed that excitonic interactions between Chl molecules have a small effect on the Qy-energies but may result in a remarkable redistribution of oscillator strengths. A redistribution of absorption intensity with increasing tempera-

Solubilized LHC II Complexes of Green Plants

J. Phys. Chem. B, Vol. 105, No. 29, 2001 7123 to a Boltzmann distribution representing the thermal equilibrium of the excitation energy. This is in line with conclusions from pump-probe experiments at room temperature.20 Concluding Remarks

Figure 9. Semilogarithmic plot of normalized fluorescence spectra of LHC II (high-energy wing). The straight lines represent linear fits for temperatures of a) 80 K (85 ( 2), (b) 120 K (115 ( 2), (c) 180 K (169 ( 2), (d) 240 K (215 ( 5), and (e) 300 K (278 ( 7). Temperatures in brackets were calculated from linear regression.

ture, i.e., a gain of oscillator strength of the low-energy Qystates, has already been proposed from a study of the temperature dependence of the LHC II absorption spectrum53 and would also qualitatively account for the red-shift of the fluorescence peak above ∼150 K observed in the present study. Furthermore, additional factors such as increasing heterogeneity due to thermally accessible conformational substates as well as quadratic and/or anharmonic electron-phonon coupling can be expected at higher temperatures.37 This interplay of different effects precludes a further detailed analysis. Another interesting effect emerges for the high-energy wing of the LHC II fluorescence spectra. Figure 9 shows a semilogarithmic plot of the normalized fluorescence signal (650680 nm) as a function of wavenumber and wavelength (see bottom and top scale, respectively). Surprisingly, linear ranges over at least 200 cm-1 appear for all temperatures between 80 and 290 K. This observation suggests that the high-energy wing of the LHC II fluorescence spectrum can be characterized by an exponential law. If the slope is identified with kT/hν the linear regression yields approximately the sample temperature for T > 80 K (see caption of Figure 9). This feature is remarkable since emission of a given fluorescent state should be proportional to the product of oscillator strength and thermal population, i.e., the finding of an exponential law seems to be in contradiction with the oscillator strengths assigned to the low-energy (676, 677.1, 678.4, and 680.0 nm) states in ref 24 (vide supra). A closer inspection of Figure 9 reveals, however, that the linear ranges in the semilogarithmic plot are observed for wavelengths of 676-665 nm (80 K) and 676-655 nm (300 K). For example, at a temperature of 80 K the thermal population of a state in the vicinity of 665 nm is about 2 orders of magnitude lower than that of the 676 nm state. On the other hand, from lowtemperature absorption spectra of LHC II (cf. e.g. ref 23) it is reasonable to assume that the oscillator strengths of the electronic states in the given wavelength region do not differ by more than a factor of 3. Thus, for only a small variation of the oscillator strengths for the electronic states considered the line shape of the fluorescence should be mainly determined by the Boltzmann law. At a temperature of 300 K, however, the thermal population of a state at about 655 nm would be only 10 times lower than that of the 676 nm state, i.e., the variation of oscillator strengths leads to deviations from linearity at higher temperatures. Nevertheless, the results presented here indicate that the Qy-states of trimeric LHC II are populated according

Low-temperature non-line-narrowed fluorescence and FLN data were used to analyze electron-phonon coupling of the trimeric light-harvesting complex of Photosystem II (LHC II). Special attention has been paid to eliminate effects owing to reabsorption and to ensure that the line-narrowed fluorescence spectra are virtually unaffected by hole-burning or scattering artifacts. The 4.2 K fluorescence origin band shows a strong asymmetry with a width of ∼120 cm-1 and a peak at 680.3 ( 0.2 nm. The 4.2 K FLN spectra obtained by selective excitation within the low-energy wing of the fluorescence origin band exhibit a strongly asymmetric shape with a width of ∼100 cm-1 and peak ∼24 cm-1 to the red of the excitation wavelengths. Theoretical simulations of both types of spectra reveal that the lowest Qy-state of LHC II is characterized by weak electronphonon coupling with a Huang-Rhys factor of ∼0.9 to a broad and strongly asymmetric one-phonon profile with a peak frequency ωm of 15 cm-1 and a width of Γ ) 105 cm-1. In addition, the assignment of the lowest Qy-state at ∼680.0 nm and its inhomogeneous broadening of ∼80 cm-1 in a recent hole-burning study24 can be confirmed. The temperature dependence of the fluorescence spectra of LHC II can be understood up to a temperature of 120 K using the parameters of electron-phonon coupling reported above as well as the low-energy Qy-level structure and corresponding oscillator strengths reported in ref 24. The calculated spectra reflect the basic features, such as thermal broadening, varying shape and, especially, the correct blue-shift of the fluorescence peak with increasing temperature. Conformational changes of the protein environment due to a decrease in viscosity above the glass temperature (Tg) are assumed to alter excitonic coupling as well as pigment-protein interactions. These effects may be responsible for the red-shift of the fluorescence peak above 120 K as has been suggested, e.g., for the B850 complex of Rb. sphaeroides (see, e.g., Wu et al.27). The shape of the temperature-dependent fluorescence spectra indicates that the lowenergy Qy-states are populated according to a Boltzmann distribution representing the thermal equilibrium of excitation energy. Acknowledgment. We thank G. J. Small and J. M. Hayes for stimulating discussions and a calculation program provided for comparison. J.P. gratefully acknowledges support by NaFo¨G, the German Academic Exchange Service (DAAD), the FAZITStiftung in Frankfurt/Main and the International Institute of Theoretical and Applied Physics (IITAP) at Iowa State University in Ames, Iowa. K.-D.I. and G.R. as well as J. V. acknowledge financial support from Deutsche Forschungsgemeinschaft (SFB 429, TP A3 and SFB 312, TP A6, respectively). References and Notes (1) Van Grondelle, R.; Dekker, J. P.; Gillbro, T.; Sundstrom, V. Biochim. Biophys. Acta 1994, 1187, 1. (2) Renger, G. In Concepts in Photobiology and Photomorphogenesis; Singhal, G. S., Renger, G., Sopory, K., Irrgang, K.-D., Govindjee, Eds.; Narosa Publishing House: New Delhi, India, 1999; p 52. (3) Paulsen, H. Photochem. Photobiol. 1995, 62, 367. (4) Ku¨hlbrandt, W.; Wang, D. N.; Fujiyoshi, Y. Nature 1994, 367, 614. (5) Trinkunas, G.; Connelly, J. P.; Mu¨ller, M. G.; Valkunas, L.; Holzwarth, A. R. J. Phys. Chem. B 1997, 101, 7313.

7124 J. Phys. Chem. B, Vol. 105, No. 29, 2001 (6) Gradinaru, C. C.; O ¨ zdemir, S.; Gu¨len, D.; van Stokkum, I. H. M.; van Grondelle, R.; van Amerongen, H. Biophys. J. 1998, 75, 3064. (7) Rogl, H.; Ku¨hlbrandt, W. Biochemistry 1999, 38, 16214. (8) Du, M.; Xie, X.; Mets, L.; Fleming, G. R. J. Phys. Chem. 1994, 98, 4736. (9) Savikhin, S.; van Amerongen, H.; Kwa, S. L. S.; van Grondelle, R.; Struve, W. S. Biophys. J. 1994, 66, 1597. (10) Bittner, T.; Irrgang, K.-D.; Renger, G.; Wasilewski, M. R. J. Phys. Chem. 1994, 98, 11821. (11) Bittner, T.; Wiederrecht, G. P.; Irrgang, K.-D.; Renger, G.; Wasilewski, M. R. Chem. Phys. 1995, 194, 311. (12) Connelly, J. P.; Mu¨ller, M. G.; Hucke, M.; Gatzen, G.; Mullineaux, C. W.; Ruban, A. V.; Horton, P.; Holzwarth, A. R. J. Phys. Chem. 1997, 101, 1902. (13) Rebane, K. K. Impurity Spectra of Solids; Plenum Press: New York, 1970. (14) Renger, T.; Voigt, J.; May, V.; Ku¨hn, O. J. Phys. Chem. 1996, 100, 15654. (15) Ku¨hn, O.; Renger, T.; May, V.; Voigt, J.; Pullerits, T.; Sundstro¨m, V. Trends Photochem. Photobiol. 1997, 4, 213. (16) Jankowiak, R.; Small, G. J. Chem. Res. Toxicol. 1991, 4, 256. (17) Jankowiak, R.; Hayes, J. M.; Small, G. J. Chem. ReV. 1993, 93, 1471. (18) Hemelrijk, P. W.; Kwa, S. L. S.; van Grondelle, R.; Dekker, J. P. Biochim. Biophys. Acta 1991, 1098, 159. (19) Reddy, N. R. S.; van Amerongen, H.; Kwa, S. L. S.; van Grondelle, R.; Small, G. J. J. Phys. Chem. 1994, 98, 4729. (20) Schro¨tter, T.; Voigt, J.; Kehrberg, G. Photosynthetica 1994, 30 (4), 567. (21) Kehrberg, G.; Voigt, J.; Schro¨tter, T.; Renger, G. Biochim. Biophys. Acta 1995, 1231, 147. (22) Voigt, J.; Renger, T.; Scho¨del, R.; Schro¨tter, T.; Pieper, J.; Redlin, H. Phys. Stat. Sol. (b) 1996, 194, 333. (23) Peterman, E. J. G.; Pullerits, T.; van Grondelle, R.; van Amerongen, H. J. Phys. Chem. B. 1997, 101, 4448. (24) Pieper, J.; Ra¨tsep, M.; Jankowiak, R.; Irrgang, K.-D.; Voigt, J.; Renger, G.; Small, G. J. J. Phys. Chem. A. 1999, 103, 2412. (25) Reddy, N. R. S.; Lyle, P. A.; Small, G. J. Photosynth. Res. 1992, 31, 167. (26) Pieper, J.; Voigt, J.; Renger, G.; Small, G. J. Chem. Phys. Lett. 1999, 310, 296. (27) Wu, H.-M.; Ra¨tsep, M.; Lee, I.-J.; Cogdell, R. J.; Small, G. J. J. Phys. Chem. B 1997, 101, 7654. (28) Wu, H.-M.; Ra¨tsep, M.; Jankowiak, R.; Cogdell, R. J.; Small, G. J. J. Phys. Chem. B 1998, 102, 4023. (29) Irrgang, K.-D.; Boekema, E. J.; Vater, J.; Renger, G. Eur. J. Biochem. 1988, 178, 209. (30) Peterman, E. J. G.; Hobe, S.; Calkoen, F.; van Grondelle, R.; Paulsen, H.; van Amerongen, H. Biochim. Biophys. Acta 1996, 1273, 171.

Pieper et al. (31) Porra, R. G.; Thompson, W. A.; Kriedemann, P. E. Biochim. Biophys. Acta 1989, 975, 384. (32) Vasil’ev, S.; Irrgang, K.-D.; Schro¨tter, T.; Bergmann, A.; Eichler, H.-J.; Renger, G. Biochemistry 1997, 36, 7503. (33) Scho¨del, R.; Hillmann, F.; Schro¨tter, T.; Voigt, J.; Irrgang, K.-D.; Renger, G. Biophys. J. 1996, 71, 3370. (34) Personov, R. I. In Spectroscopy and Excitation Dynamics of Condensed Molecular Systems; Agranovich, M., Hochstrasser, R. M., Eds.; Elsevier/North-Holland, Amsterdam: The Netherlands, 1983; p 555. (35) Hayes, J. M.; Gillie, J. K.; Tang, D.; Small, G. J. Biochim. Biophys. Acta 1988, 932, 287. (36) Lee, I.; Hayes, J. M.; Small, G. J. J. Chem. Phys. 1989, 91 (6), 3463. (37) Hayes, J. M.; Lyle, P. A.; Small, G. J. J. Phys. Chem. 1994, 98, 7337. (38) McColgin, W. C. Ph.D. Thesis, The University of Rochester, 1975. (39) Kwa, S. L. S.; Vo¨lker, S.; Tilly, N. T.; van Grondelle, R.; Dekker, J. P. Photochem. Photobiol. 1993, 59, 219. (40) Pieper, J.; Irrgang, K.-D.; Ra¨tsep, M.; Jankowiak, R.; Schro¨tter, T.; Small, G. J.; Voigt, J.; Renger, G. J. Phys. Chem. A 1999, 103, 2422. (41) Kikas, J. Chem. Phys. Lett. 1978, 57, 511. (42) Renger, T.; May, V. Phys. ReV. Lett. 2001. In press. (43) Pieper, J.; Irrgang, K.-D.; Ra¨tsep, M.; Schro¨tter, T.; Small, G. J.; Voigt, J.; Renger, G. In Proceedings of the International Conference on Ultrafast Processes in Spectroscopy (UPS 97); Institute of Physics, University of Tartu: Tartu, Estonia, 1997. (44) At 4.2 K the mean frequency approximation remains valid for a broad distribution of phonon modes, because for kT ∼ 3.5 cm-1 the thermal occupation of phonon levels is negligible. In this case, the one-phonon profile reflects the density of states of the protein phonons multiplied by a frequency dependent coupling strength. (45) Avarmaa, R. A.; Rebane, K. K. Spectrochim. Acta 1985, 41, 1365. (46) Gillie, J. K.; Small, G. J.; Golbeck, J. H. J. Phys. Chem. 1989, 93, 1620. (47) Hayes, J. M.; Jankowiak, R.; Small, G. J. In Persistent Spectral Hole-Burning: Science and Applications; Moerner, W., Ed.; SpringerVerlag: Berlin, 1988; p 153. (48) Johnson, S. G.; Tang, D.; Jankowiak, R.; Hayes, J. M.; Small, G. J.; Tiede, D. M. J. Phys. Chem. 1989, 93, 5953. (49) Peterman, E. J. G.; van Amerongen, H.; van Grondelle, R.; Dekker, J. P. Proc. Natl. Acad. Sci. U. S .A. 1998, 95, 6128. (50) den Hartog, F. T. H.; Dekker, J. P.; van Grondelle, R.; Vo¨lker, S. J. Phys. Chem. B 1998, 102, 11007. (51) Hayes, J. M.; Matsuzaki, S.; Ra¨tsep, M.; Small, G. J. J. Phys. Chem. 2000, 104 (23), 5625. (52) Matsuzaki, S.; Zazubovich, V.; Ra¨tsep, M.; Hayes, J. M.; Small, G. J. J. Phys. Chem. B 2000, 104 (40), 9564. (53) Voigt, J.; Schro¨tter, T. Z. Phys. Chem. 1999, Z11, 181.