MULTIFREQUENCY CROSS-CORRELATION PHASE FLUOROMETRY OF CHLOROPHYLL a FLUORESCENCE IN THYLAKOID AND PSII-ENRICHED MEMBRANES

- We prescnt here a comparative study on the decay of chlorophyll (Chl) u fluorescence yield in thylakoid membranes and photosystem 11 (PSI1)-enriched samples, measured with multifrequency cross-correlation phase fluorometry. These measurements confirm the general conclusions of Van Mieghem ef al. (Binchim. Biophys. Acta 1100, 198-206, 1992), obtained with a flash method, on the effects of reduction of the primary quinone acceptor (QA) on Chi a fluorescence yield of PSI. Different states of the reaction centers of PSII were produced by: (1) pretreatment with sodium dithionite and mcthyl viologen followed by laser illumination: the doubly reduced QA (Q4HZ) centers: (2) with laser illumination or pretreatment with diuron: QA- centers; and (3) the addition ofmicromolar concentration of dichlorobenzoquinonc (DCBQ): oxidized QA centers. The data were analyzed with Lorentzian distribution as well as with multiexponential fluorescence decay functions. The analysis with Lorcntzian distribution function showed that upon formation of QA . the major lifetime distribution peak shifted to longer lifetimes: from 0.25 ns to 1.66 ns (pea thylakoid membrancs) and from 0.24 ns to 1.31 ns (core PSII). However, when QAHZ was formed, the lifetime distribution peaks shifted back to shorter lifetimes (0.574.77 ns) both in thylakoids and PSIl membranes. Multiexponential analysis showed three lifetime components: fast (40-400 ps), middle (300-1 500 ps) and slow (5-25 ns). When QA was formed in PSI1 centers, the amplitude of the fast Component decreased, but both thc amplitude and the lifetime of the middle component increased severalfold. However, when Q,, was doubly reduccd, the amplitude of the fast component increased and the amplitude of the middle component decreased: in addition, the lifetime of the slow component increased. All of the above results are consistent with the conclusions that PSI1 charge separation is decreased when Q A ~ is formed and increased when doubly reduced QA is formed.


INTRODUCTION
In chloroplasts, light energy is captured by antenna pigments that are contained in chlorophyll (Chl)? a/h or Chl a-protein complexes. The excitation energy is transferred through a series of ultrafast energy transfer steps in tens of picoseconds before it reaches the reaction centers for photochemical charge separation. Most of Chl a fluorescence at room temperature originates in photosystem I1 (PSII). The yield of this PSII fliiorescencc is dependent on the redox state of the reaction center. When the electron acceptor QA of PSIl is in the oxi d i x d state. thc Chl a fluorescence yield is low (F,,). However, when the QA is reduced, the Chl a fluorescence yield is high The origins of each of these components are still not fully understood. It has been shown, from time-resolved emission spectra'5,'6 and from PSI and PSII mutant studies," that a part of the rapid decay arises from PSI (30 ps; 100-150 ps) and another (100-350 ps) from PSII. Klimov et a/.'H proposed that a slow component (1.3-2.5 ns) originates from PSI1 radical pair recombination between the oxidized reaction center Chl a of PSII, P680+, and reduced primary acceptor (pheophytin) of PSII, Pheo-, an idea that may not be valid for intact PSI1;'but may be applicable to reaction center I1 preparations for the 2-35 nscornponent (seet.g. Covindjee The idea that the complexity of the kinetic Components of fluorescence decay could be interpreted in terms of the existence of two types of PSII centers PSIIn and PSIIB (see e.g Mclis and Honiann,/Y and as done by Holzwarthj) was first suggested by Butler et a/.*u (also see other references2',-").
Based 011 time-resolved emission and excitation spectral analyscs performed under conditions when reaction centers are opcn (F", QA is oxidized) and when they are closed (F,,,, QA is reduced), Holzwarth cif a/. " had proposed that an 80  ps component arises from PSI, a 180 ps component from   open PSiltv centers, the original "middle" component (500  ps) from open, and slow (1.2 ns) from closed PSlI[j centers, while the long-lived (2.2 ns) component is emitted by closed PSIIN centers. However, Holzwarth3 later suggested that the 1.2 ns component is from closed PSIla and the 2.2 ns from PSII closed PSIIp.
Usinga simpler system, i.c. PSI1 particles and low intensity of excitation, Schati and coworkers",-'4 observed that upon closure ol'reaction centers (QA -closed), lifetimes of fluorescence changed from about 80 ps and 520 ps to 220 ps and 1-3 ns. SchatL (it a/." suggested that QA controls primary charge separation; with QA present, the charge separation was slowed/decreased, as confirmed by Trissl rt al. 26: the 520 ps tinic was related t o the time of electron transfer from the rcduccd Pheo to QA.
Discrete component analysis of the fluorescence assumes that all thc radiating fluorophores decay with a well-defined set of lifetimes. However. in the case of heterogeneous systems containing proteins, membranes, P / C . , in which the electronic environments of the emitting molecules are far from being unique and can change during the excited state lifetime, such an approach has been questioned." The simulation study of Alcala el u / . , -'~ for protein fluorescence, showed that the discrete componcnt analysis with one or two exponentials, when used to study distributions of lifetimes, was very sensitive to the number and range of frequencies at which thc data are collected. In general a two exponential fit to a symmetric distribution yielded a nonsymmetric result. The result of the fit was symmetric only with very particular sets of frequencies whose values depended on the distribution shape. To distinguish among the different factors involved in the tlecay is impossible due to the limited resolvability of the data i n lifetime components provided by current instrumentation. Thus, the observed signal may, alternatively, be easily composed of a superposition of heterogeneous decays comprising individual lifetime values that are close to one another.'" As a result, the assignment of one or more exponcntials to describe the overall decay process can hide the true physical origin of lifetime heterogeneity. Photosystem I 1 is known to bc a Chl-protein-containing heterogeneous membrane system.3u It is well established that protein structural fluctuations can occur in the nanosecondpicosecond time scale.3'-33 The concept of distribution of lifetimc values has been introduced in fluorescence-'4 and has been successful in lifetime analysis of Chl a of reaction renter I 1 preparations" that lack QA.
We present here an analysis of fluorescence decay data in thylakoid membranes and in two diffcrent PSII-enriched membranes, using multifrequency phase fluorometry. A Lo-rent7ian analysis is made because the fluorescence decay may be a superposition of many similar exponential decays, as noted above. In particular. we have uscd PSII samples similar to thosc used by Van Mieghem r t u/? in order to test their conclusions regarding the influence of the redox state of QA on the charge transfer by the multifrequency cross-correlation phase fiuorometry. These data were analyzed for both open and closed PSI1 centers in terms of distributions of lifetimes based on the principles outlined by Alcala el u/.27-'y~*7 (also see Govindjec 01 a/. '4). Analysis of the data with both ni u 1 ti exponent i a 1 a n d Lo rent z i a n d i s t r i b u t i on fun c t i on s showed that upon closure of reaction centers (QA -closed centers), the lifetime peak of the major Lorentzian distribution shifted to longer lifetimes: from 0.25 ns to 1.66 ns in pea thylakoid membranes; and from 0.24 ns to 1.31 ns in PSII membranes. This change in fluorescence properties of PSII may be caused by a transmembrane electric field and the charge of QA ~ (see Keuper and Sauer" and Holzwarth3).
However, in samples in which QA was doubly reduced, the lifetime distribution (peak at 0.57 ns for spinach thylakoids and at 0.77 ns for PSII) had shorter lifetimes compared to that o f Q A -closed centers and longer lifetimes compared to that of the opcn centers (QA centers). In this case, the effect o f the charge disappears as doubly reduced QA may become quinol, as suggested by Van Mieghem et ul..i6 Thus, our data on our PSII samples, measured with an independent method, complement the published d a t P and support the earlier conclusions.

MATERIALS AND METHODS
Thylakoid and PSI1 membranes were prepared from appressed membrane fragments of chloroplasts from Piszrm sarirum (peas), as described by others (see K and M preparation in Dunahay rI a/. Is).
A PSI1 sample, prepared by the method of Ghanotakis et a/.," labeled as "core PSII," was also used in this study. In addition, thylakoids and PSII-enriched membranes (Berthold ef a/.4") were prepared from spinach (Syinacia oleracra). The Chl concentration was determined using extinction coefficients published by Ziegler and Egle" for absorbance at 664 nm and 647 nm.
Samples were suspended in a reaction medium containing 0.4 M sorbitol, 5 mM 2-(N-morpholino)ethanesulfonic acid (MES)-KOH (pH 6 . 5 ) , 20 mM KCI, 2 mM MgCI, and 1 pM nigericin when fluorescence was measured. The Chl concentration was 5 p M . The F,, (open centers) condition was obtained by the addition of 15 p M 2,6-dichloro-p-benzoquinone (DCBQ) and the F,,,;,, by 5 p M 343.4dichlorophenylf-1.1 -dimethylurea (DCMU) (Q,, ~ -closed centers). In addition, in other samples Q,, was doubly reduced by adding I5 mM freshly prepared sodium dithionite and I50 pMmethyl viologen and leaving the sample to incubatc in the dark for about 2 h before use (modified after Van Mieghem el a1.j'; this is a much more drastic trcatmcnt than used for chemical reduction of Pheo; see e.g. Wasiclewski rt a/.4'). The double reduction of Q,, was tested as in Van Mieghem rt a/.4' by Govindjee (data not shown). The samples were kept in tightly covered cuvcttes during measurements.
To study the time-resolved fluorescence emission, a multifrequency cross-correlation phase fluorimeter was used. The light source consisted of a Coherent Antares 76-S neodymium yttrium-aluminum-garnet (Nd-YAG) laser, mode locked at 76 MHz. The picosecond optical pulse train generated by this system synchronously pumped a cavity-dumped, model 701-3 rhodamine 6G dye laser (Coherent). The repetition rate of the Coherent model 7200 cavity dumper was set at 2 MHz. The sample was excited under "magic angle" conditions at 610 nm with an attenuated, collimated I mW beam. The emission was observed at 680 nm through a UV/vis Fl3.5 monochromator (Instruments SA model H 10) equipped with a concave holographic grating with 1200 grooves/mm. Bandwidths of 8 nm full width at half maximum (FWHM) were used throughout the experiments. Both reference and sample detectors consisted of highly sensitive, low-dark-noise Hamamatsu R-928 photomultipliers operated at room temperature. Radiofrequency signals were obtained from a Marconi model 2022A signal generator and subsequently amplified by an Electronic Navigation Instruments model 603L R F power amplifier. The cross-correlation signal was set at 40 Hc (see details in earlier publications'l',Y"). -When a lluorcscence system is excited by a sinusoidally modulated light intensity at an angular frequency w, we have: where E, is the average intensity and M, is the modulation of the excitation. The Ruorcscence response of the system can be written in the form (2) whcrc F, and MI are the average fluorescence and its modulation. The fluorescence is phase shifted with respect to the excitation by a valuc ip and demodulated such that the ratio M = M,/M, < 1. At a givcn modulation rrequency, the mcasurablc quantitics 6 and M are related to thc physical parameters of fluorescence population by the following equations": . r il The function I, contains information on the distribution of components in the time domain. S(w) and G ( w ) are Ihc sine and cosine Fourier transformations of 1, and N , a normalization factor." Phasc arid modulation data were generated by using sets of components with amplitudcs determined according to a given distribution function. Analysis of the data with multiexponential function assumed that thc light emission by a fluorcscence sample upon delta-function excitation can he described as a superposition ofdiscrete cxponential dccays (see earlier publications" ?,y.44).
HtisuLrs l.ifivitw qf ('hl a jluoreseetice qf P S I I when Q.,, is reduced We measured lifetimes of fluorescence at the two extremes: ( I J when all QA was oxidized and (F,,,,) when all QA was rcduced because the kinetics of decay are simplified and the analyses are unaffected by the organization of the PSII units ( I c' . they are independent of whether the matrix [i.e. "lake"] or separate package [i.e. "isolated puddles"] model is appropriate). As noted earlier, the samples were excited at 6 10 nm and the C'hl u fluorescence was measured at 680 nm. Figure  I shows the phase shift, @, and relative modulation, M, as a function of frequency in MHz in open and QA -closed PSI1 centers in pea thylakoid membranes (Fig. 1A) and pea core PSII (Fig. 113). The open (F,) condition4' was ascertained by the addition of 15 fiM DCBQ, whereas the QA -closed condition was obtained by the addition of 5 p M DCMU in both pea thylakoid membranes and PSII samples. Closure of PSI1 ccntcrs causes large changes in both demodulation M and phase shift 11) (compare data with open and closed symbols). The results of multicxponential model fits are shown in Table  I . In order to fit the data, a minimum of three exponential components are necessary at both F, (open centers) and F,,, ((& -closed centers) as judged by the low residuals (Fig. 2) and the low x 2 values. A single or double exponential fit is not sufficient to describe the data (data not shown). These rlwdts are qualitatively, but not quantitatively, in agreement with thosc published earlicr.'3,'6 Analysis with four decay components led to only a slight lowering of the x2 and produced a n additional component with an insignificant amplitude.  When QA remained mostly oxidized in the presence of DCBQ, thrcc resolved lifctimc components in the open PSII centers in the core PSII were approximately 40 ps (32Y1, fractional intensity), 480 ps (65%) and 8.5 ns (negligibly small, 3"/0), and those in pea thylakoid membranes were approximately 40 ps (30'%)), 600 ps (64%) and 5.9 ns (again, very small, 6"/0). Upon closure of the PSI1 centers (QA -closed), the three lifetime components were approximately 50 ps (only 8%), -1.3 n s (85Yo) and -5.6 ns (only 7%) in core PSII; in thylakoid membranes these components were approximately 140 ps (1 Io/o), 1.6 ns (78%) and 5.1 ns (1 lo/n). Thus closure of the reaction centers, i.r. formation of Q A -, leads to twoto three-fold increase of the lifetime of the middle component, but the fractional intensity increases only by a factor of 1.  of light, used here, also reduces all QA to QA (data not shown). Now, we show results with the altcrnate lifetime distribution method, the focus of this paper. Analysis with Lorentrian functions gives lifetime distributions shown in Fig.  3 ; the parameters for Lorentzian fit functions are presented in  with the three exponential decay analysis. In closed PSII centers, i.e. with all QA in the reduced state, a double Lorentzian lifetime distribution with a dominant distribution peaking at 1.3 ns (PSII core; curve D) or 1.7 ns (thylakoid membranes; curve C) was observed; a picosecond peak has a negligibly small fractional intensity of 5% or 7%. A shift from a shorter lifetime distribution to a longer lifetime distribution occurred upon closure of PSI1 (QA-closed), as expected.

P S I ccntcrs with doubfe reducriot? of QA
It appears that DCBQ opens most of the PSII centers by oxidiying the primary quinone acceptor QA. On the other hand, pretreatment with benzyl or methyl viologen and sodium dithionite and light leads to double reduction of QA (see Van Mieghem et aI.j6). Figure 4 shows the phase shift, a, and relative modulation, M, as a function of frequency in MHz in QA-and in doubly reduced QA PSII centers in spin-   ach thylakoid membranes (Fig. 4A) and PSI1 membranes (Fig. 4B). Samples with doubly reduced QA have significantly ditlerent demodulation M and phase shift @ from those with singly rcduccd QA.
Parameters obtained by a triple exponential lit for the above fluorcsccncc data arc prcscnted in Tablc 3. I n Q,--closed Table 3. Lifetime ( 7 ) and fractional intensity (f) obtained by a triple cxponcntial fit to the fluorescence decay of thylakoid membranes (spinach) and PSII membranes* -T I   Figure 5 shows double Lorentzian function lifetime distributions of spinach thylakoid membranes (Fig. 5A,C) and PSII membranes (Fig. 5B,D); the fit parameters for Lorcntzian functions arc presented in Table 4. In QA--closed centers, the majority offractional intensity is localized at a Lorentzian function center at 2 ns (thylakoid membranes; Fig. 5A) or 1.37 n4 (I'SII-enriched membranes; Fig. 5B). In samples with doubly reduced QA, however, the distribution peak is shifted to shorter lifetimes. The main centers ofthe Lorentzian funetion fit for thylakoid membranes and for PSII are a t 0.57 ns (Fig. 5") and at 0.77 ns (Fig. 5D).

DISCUSSION
In this paper, we have presented our mcasurements on Chl  )). In open PSIl centcrs the halftime of elcctron transfer from Pheo to Q A is 300-500 PS.'~.~' In closed centers the long-lived (larger than 2 ns) lifetime components for the radical pairs contained less than 10% of total fluorescence. The increase in Chl u fluorescence yield was considercd to be caused by a lengthening of the excited state lifetime due to a decreased yield of charge separation. In cont rast, M a u~e r a l l~~ concluded that the long-lived (about 2 ns) fluorescence from closed PSI1 centers is recombination luminescence as proposed by JSlimov ct al. However, recent experimcntal results d o not support the hypothesis of recombination luminescence in Q,-containing PSII:' Such recombinational luminescence, however, O C C U~S~. '~ in PSII reaction centers devoid of QA. In PSII samples that contain doubly reduced QA, lifetime distribution is shifted to shorter lifetimes from that in samples that contain reduced QA (Van Mieghem rt this paper) but with somewhat longer lifetimes than those with open centers. This is explained by a n increased probability ofcharge separation and by an increased probabilityofcharge recombination. The electrostatic effect of Q A -on the P680' Phco-radical pair is lost and these centers would resemble the open centers with short fluorescence lifetime.
The fluorescence decay is customarily resolved in terms of exponential components, and the values of the decay rates and preexponential factors of each component are associated with a particular conformation and with the relative population of each conformation. However, the preexponential factors cannot be related to the fraction of molecules in each conformation. The fluorescence lifctime distribution is determined by the multitudc of conformational substates in a protein and by the dynamics of thc protein (Gratton et ~1 .~~) .
T h e lifetime distribution mcthod, as used here, provides a good approach for the study of conformational substates and of the energetics of such substates. In the limiting case of the frozen protein (negligible dynamics), one may consider that the fluorescence is determined by a set of exponentials o f which the lifetimes and amplitudes are characteristic of the set of environments of the excited residues in the protcin.
However, as the dynamic nature o f the protein is allowed to play its role, the excited chlorophylls become exposed to electronic environments, the nature of which vary with time.
Here in this study, the changes in the lifetime distribution ofChl a fluorescence decay in QA--closed centers and centers with doubly reduced Q A have provided a newer and alternate view on the changes that occur when PSII reaction centers are closed.
.4c~knowledgements-The experiments and analyses of the data produced were performed at the Laboratory for Fluorescence Dynamics (LFD) at the University of Illinois at Urbana-Champaign (UIUC). The LFD is supported jointly by the National Center for Research Resources of the National Institutes of Health (RR03 155) and UIUC. Covindjee was supported by a grant from the NSF (91-16838). We thank Dr. Michael Seibert for his interest in the carly phase of this work. A preliminary version of the work presented here first appearedasan Appendix in thePh.D. thesisofJ. Cao(1992), submitted under the direction of Govindjee.