8420
J. Phys. Chem. 1995,99, 8420-8431
Comparison of Calculated and Experimentally Resolved Rate Constants for Excitation Energy Transfer in C-Phycocyanin. 2. Trimers Martin P. Debreczeny? and Kenneth Sauer* Structural Biology Division, Lawrence Berkeley Laboratory and Department of Chemistry, University of Califomia, Berkeley, Califomia 94720
Jianhui Zhou’ and Donald A. Bryant Department of Molecular and Cell Biology and Center for Biomolecular Structure and Function, Pennsylvania State University, University Park, Pennsylvania I6802 Received: November 1.5, 1994; In Final Form: February 2, 1 9 9 P
The light-harvesting protein C-phycocyanin (PC) in the trimeric aggregation state, isolated from the cyanobacterium Synechococcus sp. PCC 7002, is studied by absorption spectroscopy and by time-resolved anisotropic fluorescence spectroscopy with 1 ps time resolution. PC trimers isolated from the wild-type strain and a mutant strain, cpcB/CI55S, in which the chromophore is missing, are compared. The absorption spectra of the trimeric PCs, when compared with previously published spectra of the monomeric PCs [Debreczeny et al. J. Phys. Chem. 1993,97,9852-98621, lead us to conclude that the absorption spectrum of the P I 5 5 chromophore is similar when PC is in the monomeric and trimeric states. This means that the red shift of the absorption spectrum that occurs when PC aggregates from monomers to trimers is due to changes in the spectra of the a84 andor p g 4 chromophores. First-order exciton coupling between chromophores cannot alone be the cause of the red shift. Time-resolved fluorescence anisotropy measurements lead to assignments of the dominant rate constants for energy transfer between chromophores in PC trimers and provide information about the relative chromophore orientations. The angle between the transition dipoles of the ai4and pi4 chromophores on adjacent monomers in the trimer is estimated from the fluorescence anisotropy decay to be 52” (the chromophore numbering scheme follows the convention established by Schirmer et al. J. Mol. Biol. 1987, 196, 677-695). The angles between the C3 axis of symmetry in the trimer and the a84 and pg4 chromophore transition dipoles are restricted to four possible values. The observed anisotropic fluorescence decay times of 1.0, 50, and 40 ps are respectively assigned to energy transfer within the ai,-&, pairs on adjacent monomers, within the &55-pk4pairs on the same monomer, and between the ai&,, ai&,, and ai&, pairs around the trimer ring. These results are in excellent agreement with Forster calculations in the weak coupling limit which predict decay times of 1.4, 49, and 46 ps for these same three energy-transfer processes. In combination with previous results [Debreczeny et al., companion paper], these results indicate that the dominant energy-transfer processes in monomeric and trimeric PC are well described by Forster’s theory. This is the first detailed confirmation of Forster theory in a pigment protein.
I. Introduction In the preceding paper,’ it was demonstrated that excitation energy transfer in the light-harvesting protein C-phycocyanin (PC) in the monomeric aggregation state is well described by Forster’s model in the weak coupling limit. Here we move one step closer to the native state in cyanobacteria by investigating PC in the trimeric aggregation state. PC trimers stack along their 3-fold axis of symmetry in association with “linker” proteins to form the peripheral rods of the phycobilisome. The crystals from which the X-ray structures of PC were determined contained linker-free PC in the trimeric or hexameric states, depending on the organism from which they are i s ~ l a t e d .The ~.~ crystal structures (see Figure 8, for example) are suggestive that aggregation of PC monomers into trimers introduces new routes for energy transfer. In particular, in trimeric PC the ai4and pi4 chromophores on adjacent monomers are separated by a center-to-center distance of only 21 A, whereas the most closely Current address: Chemistry Division, Argonne National Laboratory, Argonne, IL 60439. Current address: Department of Plant Biology, University of California, Berkeley, CA 94720. ‘Abstract published in Advance ACS Abstracts, April 15, 1995.
0022-365419512099-8420$09.00/0
coupled chromophores in the monomer, the p 1 5 j - B S 4 pair, are separated by 34 A. The effect of aggregation on the fluorescence anisotropy decay of PC is dramatic. The residual anisotropy value decreases to nearly zero, and the rate of anisotropy decay increases as PC is aggregated from monomers to brimers. Some of the decay constants for energy transfer in trimeric PC are too fast to be resolved by the time-correlated single photon counting technique. Instead, the fluorescence upconversion technique is used to measure the decay of fluorescence anisotropy of PC trimers with 1 ps time resolution. In this paper, we pursue the question of whether Forster theory, so effective at describing the kinetic processes in PC monomers, is also applicable in the trimers. In addition to the dramatic increase in the transfer rates, a more subtle effect of the aggregation of PC from monomers to trimers is the modification of the spectroscopic properties of the individual chromophores in PC. That the chromophores are affected by the aggregation state is clear from the observed red shift of the absorption spectrum of trimers compared to monomers. The cause of the spectroscopic changes is less clear. One possibility is that coupling between chromophore pairs 0 1995 American Chemical Society
Excitation Energy Transfer in C-Phycocyanin
J. Phys. Chem., Vol. 99, No. 20, 1995 8421
introduced upon trimer formation is strong enough to influence total volume). The elution profile monitored by absorbance at the absorption spectra of the chromophores. If the geometry 614 nm, contained a major and a minor peak. Sodium dodecyl of the interaction is such that excitation preferentially induces sulfate polyacrylamide gel electrophoresis (SDS-PAGE) of transitions to the lower energy band of a pair of excitonically fractions collected during the first (minor) peak (at 30 mM (K) split states, this could explain the observed red shift of the phosphate) showed the characteristic a and ,8 PC bands and, in absorption spectrum. An altemative explanation of the red shift addition, bands at 33, 28, 27, and 22 kDa. These bands are is that the new protein-chromophore interactions formed when probably due to linker proteins and their degradation product^.^,^ monomers aggregate to trimers cause changes in the individual SDS-PAGE of fractions collected after the second (major) peak chromophore conformations. in the elution profile (at 60 mM (K) phosphate) showed only the characteristic 16 and 19 kDa bands assignable to the a and As with monomeric PC, it would be informative to resolve ,8 subunits of PC. These linker-free fractions were pooled, and the individual chromophore spectra in trimeric PC. Such a resolution could shed light on the question of exciton coupling after appropriate concentration by ultrafiltration (Amicon Div., in PC trimers as well as allow one to model the rate constants W. R. Grace and Co., type PM30 filters) and dialysis into 50 for energy transfer. The use of PC engineered to be missing a mM (Na) phosphate, pH 7.0, they were used in the PC trimer specific chromophore is a valuable tool for. effecting this experiments. spectroscopicresolution. Work described in this paper involves The above isolation procedure was also applied to the cpcB/ a comparison of the spectroscopic properties of PC trimers, in C155S mutant strain of Synechococcus sp. PCC 7002. The the absence of linker proteins, isolated from the wild-type strain elution profile contained only one peak, and this peak occurred and from the mutant strain PR6235 (cpcB/C155s) in which the at a slightly lower phosphate concentration (at 50 mM (K) ,8155 chromophore is absent. PC trimers isolated from the wildphosphate) than the major peak in the elution profile of PC type and mutant strains are referred to as (apc,8pc)3 and (apcp)3, isolated from the wild-type strain. SDS-PAGE of the eluted respectively. fractions showed little evidence of protein other than the a and Resolution of the absorption spectrum of the ,8l55 chro,8 subunits of PC. Of the four non-PC bands observed in the mophore in PC trimers is achieved by comparison of the steadySDS-PAGE of the chromatographic fractions of PC isolated state absorption spectra of (aKpPc)3 and (apcp)3. Comparison from the wild-type strain, only two very faint bands at 27 and of the anisotropy decays of (aPCpPC)3 and (apc,8*)3 also greatly 22 kDa could be observed in any of the chromatographic aids in the assignment of the dominant kinetic processes in PC fractions of PC isolated from the mutant strain. Fractions trimers. A comparison is made of calculated Forster rate showing no evidence by SDS-PAGE of proteins other than the constants for energy transfer with those rate constants resolved a and ,8 subunits were pooled and concentrated to make PC experimentally in the PC trimers. Because the absorption and trimers for the experiments. When PCs isolated from the wildfluorescence spectra of the a 8 4 and ,884 chromophores in PC type and cpcB/C155S strains were run beside each other on SDStrimers have not yet been resolved, the Forster calculations in PAGE gels, the a subunits migrated at indistinguishable rates, PC trimers rely on the chromophore spectra resolved in PC whereas the ,8 subunit of PC from the mutant strain consistently monomers, as described in ref 4. migrated slightly faster than the wild-type ,8 subunit. A reduction in the molecular weight of the ,8 subunit by 0.6 kDa would be 11. Methods and Materials predicted from the substitution of a cysteine with a serine and the resulting absence of a pliycocyanobilin chromophore. Growth Conditions and PC Isolation. The wild-type and Site-Selected Mutant Strains. Spectroscopic studies were mutant strains of Synechococcus sp. PCC 7002 were grown as performed on wild-type PC and on PC isolated from mutant described by Gindt.5,6 The mutant strain PR6235 (cpcB/C155S) strain PR6235 (cpcB/C155s) in which the cysteine at the ,8155 was grown in a medium that contained kanamycin (100 mg L-I) position was substituted with a serine.I0 The chromosomal and ampicillin (2 mg L-l). copies of the cpcB and cpcA genes were deleted by interposon PC for the trimer studies was isolated according to a mutagenesis with the aph2 gene of Tn5 and trans-complemented procedure adapted from that developed by Yu et al.' All buffer with the biphasic shuttle vector pAQE19 which carries the wildsolutions contained 1 mM (Na) azide as a preservative. Cells, type cpcA gene and the mutant cpcB gene. harvested by centrifugation at 3500g, were resuspended in 1 Determination of Protein Aggregation State: TrimM (K) phosphate, 0.1 M NaC1, pH 7.0, and pelleted again by mers. Ultracentrifugationwas used to determine conditions for centrifugation. This washing step was repeated three times, and which PC is stable in the trimeric aggregation state. Samples the final pellet was resuspended in a roughly equal volume of (50-100 pL) were layered onto the top of a 5 to 20% linear the same buffer. Cells were homogenized and then ruptured sucrose gradient containing the same buffer as that used for the by passing them several times through a French pressure cell sample under investigation. The gradients were run at 20 "C, at 22 GPa. The broken cell suspension was centrifuged at in a Beckman SW50.1 swinging bucket rotor, at 45 000 rpm 35000g for 30 min at 4 "C. The blue supematant was (250000g) for 14 h. Sedimentation coefficients were extracted ultracentrifuged at 300000g for 60 min at 4 "C. The supematant from standard curves measured under the same conditions was dialyzed (or gel filtered on Sephadex G-25) into 1 mM (Beckman Instruments, Inc., publication DS-528A). The sedi(K) phosphate, 0.1 M NaC1, pH 7.0. The protein was applied mentation coefficients of PC in the monomeric, trimeric, and to a hydroxylapatite column equilibrated in the same buffer and hexameric states have been previously reported elsewhere* to washed with several column volumes of the same buffer. PC be 2.3, 5.4, and 11 S, respectively. was eluted with 35 mM (K) phosphate, 0.1 M NaC1, pH 7.0. Allophycocyanin remained on the column during this step. The Using linker-free preparations of PC in 50 mM (Na) pooled PC was dialyzed (or gel filtered on Sephadex G-25) into phosphate, pH 7.0, PC is stable in the trimeric state at protein 5 mM (K) phosphate, pH 7.0 and applied to a column (2.8 x concentrations of 0.1-0.4 mg mL-' (A,,, = 1-4 cm-I) and 8.5 cm) of DEAE cellulose equilibrated with the same buffer 1.5-6 mg mL-I (Amax 10-40 cm-I) when isolated from the at 4 "C. Several column volumes of the same buffer were wild-type and cpcB/Cl55S mutant strains, respectively. The passed through the column, and the PC was eluted with a linear high concentration of PC from the mutant strain is necessary gradient of 5 mM to 125 mM (K) phosphate, pH 7.0 (400 mL to prevent partial dissociation of trimers into monomers. On
(as)
8422 J. Phys. Chem., Vol. 99, No. 20, 1995
the other hand, the observed sedimentation coefficients indicated that PC from the wild-type strain undergoes partial conversion to hexamers when the protein concentration is '1 mg mL-'. Samples described as PC trimers were at concentrations of 0.2 mg mL-' (Amax= 2 cm-I) and 3 mg mL-' (Am= = 20 cm-I) for the PCs isolated from the wild-type and mutant strains, respectively. Steady-State Absorption. Absorption spectroscopy was performed on an Aviv 14DS W-VIS-NR spectrophotometer (Aviv, Inc., Lakewood, NJ). For the comparative absorption studies of PC trimers from the wild-type and cpcB/C155S mutant strains, it was necessary to use short-path cuvettes because of the high protein concentration required to keep the PC from the mutant strain stable in the trimeric aggregation state. Absorption spectra of PC trimers from the mutant and wildtype strains were measured using quartz sample cells with path lengths of 0.10 and 1.00 mm, respectively. Fluorescence Upconversion Instrument. Fluorescence upconversion is a technique used to resolve fluorescence decay on an ultrafast time scale.I2-l5 A train of laser pulses is split into two beams, one of which is used to excite the sample, while the other beam is used in conjunction with a nonlinear crystal to "gate" the fluorescence with laser-pulse-width limited time resolution. In the fluorescence upconversion experiments described in this paper, an Nd:YAG (Coherent Antares Model 76s) laser is frequency doubled in a heated KTF' crystal to produce up to 3 W of 90 ps pulses of 532 nm light at 78 MHz. The 532 nm light is used to pump a dye laser (Coherent Model 702) which employs dual jets, typically with pump dye rhodamine 590 and saturable absorber dye 3,3'-diethyloxadicarbocyanine iodide (DODCI). The output of the dye laser is cavity dumped at 3.8 MHz to increase the power per pulse and to allow adequate time for sample relaxation between laser pulses. With the above combination of dyes, the output is typically 70 mW of 1-1.5 ps pulses tunable between 575 and 625 nm. Approximately 70% of the laser is split into the gating path. After passing through a computer-controlled variable delay line, the gating beam is focused onto a nonlinear crystal using a 20 cm focal length achromatic lens. A fraction of the gating pulse is split off, optically chopped, photodiode detected, and lockin amplified for use as a monitor of the laser power. The upconverted signal is divided by the square of this reference signal to compensate for fluctuations in laser intensity. The other 30% of the laser (the pump beam) is sent through a fixed path during which it is amplitude and polarization modulated and then focused by a microscope objective (10 x , 0.25 NA). The sample is flowed through a short-path cuvette and recirculated so that approximately 3 mL of total sample is required. The path length of the flow cell is 0.1 and 1.0 mm for PC trimers isolated from the mutant and wild-type strains, respectively. The optical density of the samples is typically 0.2 or less at the excitation wavelength to limit self-absorption effects. The fluorescence is collected and collimated using another microscope objective ( l o x , 0.25 NA) and then focused onto the crystal using a 15 cm achromatic lens. Because the fluorescence is collected at 180" from the pump, the transmitted pump light is also collected and focused onto the crystal. This arrangement is advantageous because, by tuning the angle of the crystal and setting the monochromator to the sum frequency of the two laser pulses, we can record an autocorrelation trace (same as the instrument response function (IRF) in this experiment) by scanning the delay stage. The nonlinear element is a 1 mm-thick single crystal of LiI03, cut and mounted so that the angle between the incoming laser
Debreczeny et al. light and the optic axis of the crystal is 60" when the front face of the crystal is at 90". The crystal is mounted on a computercontrolled rotation stage. By using a noncollinear geometry where the gating and fluorescence light cross inside the crystal at an angle of approximately 8", it is possible to geometrically discriminate against the laser fundamental, laser second harmonic, and fluorescence using an aperture. Due to scattering in the sample, optical imperfections, and room lights, it is helpful to use a W-pass filter to further suppress visible light. With this arrangement, only the sum frequency of the pump and probe (used as an IRF) and the upconverted fluorescence (the signal of interest) should be able to reach the detector (PMT) with significant intensity. Phase-matching will be found at different crystal angles for the IRF and the upconverted fluorescence; however, since the IRF signal is much more intense, it can dominate the signal if fluorescence near the excitation frequency is being upconverted. A 0.22 m double monochromator is helpful in further resolving the signal from the IRF. Time-Resolved Polarization Measurements. For the type I LiIO3 crystal used in this experiment, the input beams must be polarized parallel to each other to produce sum frequency generation.I3 With the polarization of the gating beam fixed, the nonlinear crystal will act analogously to a polarizer in that it will upconvert only parallel fluorescence. By rotating the pump polarization between parallel and perpendicular before it reaches the sample, the polarization of the fluorescence which is upconverted is selected between parallel and perpendicular (relative to the pump polarization). Because the monochromator and detector see only upconverted light of a fixed polarization and because the excitation power is the same for parallel and perpendicular polarizations, no "G factor" l 6 corrections need be applied. The polarization of the pump is modulated between parallel and perpendicular linear polarizations (relative to the polarization of upconverted fluorescence) at 80 kHz by a photoelastic modulator (Hinds PEM-80). An acousto-optic modulator (AOM) (Isomet, Model 12092-1) is also placed on the pump line to amplitude modulate the intensity between 0 and 100%. The AOM is driven at the same frequency as the PEM-80 but with a variable phase shift. The phase shifting is performed under computer control by routing the output of the PEM driver through a delay generator (Stanford Research Systems, Model DG535). By varying the relative phase of the PEM and the AOM, the polarization of the pump beam can be rapidly switched between parallel and perpendicular. The result is nearly simultaneous acquisition of parallel and perpendicular signals. By rotating the monochromator and nonlinear crystal angle in tandem, time-resolved fluorescence spectra with 1- 1.5 ps resolution can be recorded. To make the isotropic time-resolved fluorescence spectra independent of the instrumentation used to acquire them, correction factors would need to be applied. Fortunately, when calculating the fluorescence anisotropy spectrum (eq A1 in ref 1) these corrections are unnecessary because they are the same for the parallel and perpendicular cases and thus cancel out. 111. Results Absorption. The steady-state absorption spectra of (apcPpc) and (apc@pc)3 are shown in Figure la. The aggregation of the wild-type PC from the monomeric to the trimeric state results in a 10 nm red shift of the peak in the visible region of the absorption spectrum and a net increase in the oscillator strength of this peak. The steady-state spectra of (aPC@*) and (aPCP*)3 are shown in Figure lb. The effect of aggregation on the
Excitation Energy Transfer in C-Phycocyanin
300
-/
la
-26
J. Phys. Chem., Vol. 99, No. 20, 1995 8423
a
nm
wlld-type
-z 2
50-
0
. 0
0I
300
0 100
I
500 600 Wavelenglh (nm)
400
700
800
-
200
300 Delay (PSI
0.4 .......................t. .............................
ib
cpcB/ClSSS mutant Q h0.31 l
1
400
, 500
.*.................
monomers trimers
+--IFF
*....................................................................... I 1
0.0
I
0
I
I
I
300
400
I
500 600 Wavelength (nm)
I
I
700
800
Figure 1. Room temperature absorption spectra of PC monomers (.* -) and trimers (-) isolated from (a, upper) the wild-type and (b, lower) cpcBK155S mutant strains of Synechococcus sp. PCC 7002. Absorptivity values are given per mole of (90) monomer. I
O
O
:
:
nm
i
10 15 20 Delay (PS) Figure 3. Fluorescence anisotropy decay of PC monomers and trimers
5
isolated from the cpcBK155S mutant. The fluorescence upconversion technique was used to time resolve the fluorescence. The instrument response function (IRF) was 1.O- 1.5 ps FWHM. The laser excitation wavelength was 590 nm and emission was observed at 650 nm (trimers) or 659 nm (monomers).The measured time point interval for the trimer decay in this figure is 0.2 ps from -5 to 20 ps, 1.0 ps from 20 to 100 ps, and 10 ps for the remainder of the decay (measured to 1 ns).
l00-l
P,
absorbance In PC
.....
300
Trimers Monomers
400
600 Wavelength (nm)
500
700
800
Figure 2. Absorption spectrum of the p 1 5 5 chromophore at room temperature as determined in PC in the monomeric (.* .) and trimeric (-) aggregation states. Absorptivity values are given per mole of chromophore. absorption spectrum of PC isolated from the mutant strain is a slight red-shifting (4nm) and a large increase in the net oscillator strength of the band in the visible region. As was shown for PC monomers in ref 4,the (apcp*)3 spectrum can be subtracted from the (aK/?'")3spectrum to resolve the p155 absorption spectrum in PC trimers (Figure 2). The PISSabsorption spectrum resolved in PC trimers has a maximum at 598 f 0.5 nm with a molar absorptivity of (1.01 & 0.05) x lo5 M-I cm-I. This agrees closely with the p l 5 5 absorption spectrum resolved in PC monomers for which the absorption maximum is at 600 f 1 nm and the molar absorptivity is (1.12 f 0.05) x lo5 M-' cm-l. These p 1 5 5 absorption spectra, as resolved in PC monomers and trimers, are shown overlaid in Figure 2. Time-ResolvedFluorescence Anisotropy. The fluorescence anisotropy decay of (aPCp*)3, and that of (apcp*) for reference,
both measured by the fluorescence upconversion technique, are displayed in Figures 3a,b. The excitation wavelength is at 590 nm, and emission is observed at 650 or 659 nm. The data in Figures 3a,b are the same, but 3b is displayed over a narrower time window. The fluorescence anisotropy decay of (aPC/?*)3 starts at 0.26 f 0.02 and decays with an exponential time constant of 0.9 4Z 0.1 ps to an anisotropy of 0.20 f 0.01, followed by exponential decay with a 40 f 2 ps time constant to a final anisotropy of 0.07 4Z 0.02. This is dramatically different from the (aPC/?*) decay, which starts at 0.40 and decays to 0.33 with a single exponential time constant of 200 f 70 ps. A fluorescence anisotropy value of 0.4 is expected for an isotropic solution in the absence of energy transfer and prior to rotational diffusion for a chromophore whose absorption and emission transition dipoles are parallel. That the anisotropy decay of (aPCT) begins at 0.4 in Figure 3a is indicative of the fact that on this time scale we are resolving all energy-transfer steps between chromophores with nonparallel transition dipole moments. The anisotropy decay of the trimers, however, begins below the 0.4 level, and this is an indication that some kinetic processes are not being resolved in our measurement. The widths of the instrument response functions, recorded shortly before or after the decays, were between 1.O and 1.5 ps, FWHM, for all of the PC trimer decay measurements described in this paper. Thus, the -1 ps decay constant we measure is near the limit of our instrument resolution without deconvolution. Time-resolved anisotropy spectra of (apcp*)3, excited at 590 nm, are shown at several time delay points, relative to the excitation pulse, in Figure 4. The anisotropy spectrum measured at the peak of the IRF (labeled 0 ps) is flat as a function of wavelength within the uncertainty of the measurement. The anisotropy remains flat across the emission spectrum as a
Debreczeny et al.
8424 J. Phys. Chem., Vol. 99, No. 20, 1995
a
0.4>
6
0
03-
e
8c
i
0
0
::
o u u
0.2-
o o
0 0
o
* -
o
o
a
0
*
630
640
o . 0 4 1
1.3 ps.
.d
r = 0.19
o 20 ps. r =
0.1
*
r = 0.24
0 ps,
~
0.13
* * 200 ps, r = 0.07 l
I
I
650 660 670 680 Emission Wavelength (nm)
I
I
690
700
4
I
.loo
0
0.2
200 Delay (PI)
100
300
400
500
Figure 4. Time-resolved fluorescence anisotropy spectra of PC trimers isolated from the cpcB/CI55S mutant strain. The fluorescence upconversion technique was used to time resolve the fluorescence. The instrument response function (IRF) was 1.0-1.5 ps FWHM. The laser excitation wavelength was 590 nm.
~
0,340
0 .
0 0
o o o o o o o o o o o o o
F
0'4i
OPS
0
1
I
100
200
I 300
I
I
400
500
Delay (PS)
Figure 6. Fluorescence decay of PC trimers isolated from the wildO,O{
*
)
620
***** )
640 660 680 Emission Wavelength (nm)
**;200ps
700
Figure 5. Time-resolved fluorescence anisotropy spectra of PC trimers isolated from the wild-type strain. The fluorescence upconversion technique was used to time resolve the fluorescence. The instrument response function (IRF) was 1.0-1.5 ps FWHM. The laser excitation wavelength was 590 nm. function of time, so that at all emission wavelengths, the anisotropy decays from 0.24 f 0.02 to 0.20 f 0.01 with a 0.9 ps exponential time constant, followed by further decay to 0.07 f 0.02 with a 40 ps time constant, as was seen already at the particular emission wavelength of 650 nm in Figure 3. In contrast to the time-resolved fluorescence anisotropy spectra of (aPC/3*)3, the anisotropy spectra of (apcppc)3, shown in Figure 5, display a strong wavelength dependence. At the earliest time delay relative to the excitation pulse (0 ps, the peak of the IRF) at the shortest wavelengths (1625 nm), the observed anisotropy is at the 0.4 level within the uncertainty of the measurement. This is an indication that in the short-wavelength region, the energy-transfer processes between chromophores with nonparallel transition dipoles in (apcppc)3are completely time-resolved, whereas in the long wavelength region, as was the case for (aKp*)3,the kinetics are only partially resolved with our 1 .O- 1.5 ps IRF. Because the only difference between the (aPCpPC)3 and (aK/Y)3 samples is in the respective presence and absence of the chromophore,the high anisotropy values at short wavelengths and early times are attributable to emission from the p i 5 5 chromophore. That the anisotropy is at a high value in the region of the ,8155 emission and low elsewhere is an indication that the PI55 chromophore is more weakly coupled to the other chromophores than the a g 4 and p84 chromophores are coupled to each other. Figures 6a,b display the parallel and perpendicular fluorescence decays, and the anisotropy calculated from these decays,
type strain. The laser excitation wavelength was 590 nm and emission was observed at 650 nm. (a, top) Decay of fluorescence polarized parallel and perpendicular to the excitation. (b, bottom) Fluorescence anisotropy decay calculated from the parallel and Perpendicular traces in (a). The time point interval of the measurement is as described in the caption of Figure 3.
for (aPCpPC)3 excited at 590 nm and emitting at 650 nm. The anisotropy decays from an initial value of 0.27 f 0.02 to a residual anisotropy of 0.01 f 0.01 with two well-separated exponential decay constants of 1.3 f 0.4 and 54 f 10 ps. The perpendicular fluorescence trace in Figure 6a clearly shows a rising component with a lifetime of the same magnitude as the 54 ps decay component observed in the anisotropy decay. Contrasting the behavior of the anisotropic fluorescence decay of (aPCpPC)3 observed at 650 nm is the decay at the emission wavelength of 624 nm shown in Figure 7. This emission wavelength is in the wavelength region of the anisotropy spectrum (Figure 5) that shows high anisotropy at early times. The anisotropy at 624 nm decays from an initial value of 0.37 f 0.01 to a final value of 0.03 f 0.01 with a single decay constant of 70 f 20 ps. No faster decay component was detected within the uncertainty of the measurement. Notice that, whereas the perpendicular fluorescence trace at the emission wavelength of 650 nm showed a rising component (Figure 6), at the emission wavelength of 624 nm (Figure 7), the perpendicular trace shows monotonic decay. Fiirster Calculations in PC Trimers. Table 1 shows the results of calculations of the Forster rate constants for energy transfer between the chromophores in PC trimers isolated from Synechococcus sp. PCC 7002. These calculations are based on the chromophore absorption spectra, fluorescence spectra, fluorescence quantum yields, fluorescence lifetimes, and extinction coefficients resolved in PC monomers as described in refs 1 and 4. The numbering scheme for the chromophores follows that established by Schirmer et aL2 as depicted in Figure 8. Notice in Table 1 that the calculations predict that only a few chromophores are coupled by energy transfer to a significant extent (> 1 ns-I). The most strongly coupled pair by far is the
J. Phys. Chem., Vol. 99, No. 20, 1995 8425
Excitation Energy Transfer in C-Phycocyanin 0.54
a
perpendicular
0.0 -100
I 0
I
100
1
I
200
300
I 400
I 500
Delay (PI)
0
I
I
100
200
I
1
I
300
400
500
Delay (PS)
Figure 7. Fluorescence decay of PC trimers isolated from the wildtype strain. The laser excitation wavelength was 590 nm and emission was observed at 624 nm. (a, top) Decay of fluorescence polarized parallel and perpendicular to the excitation. (b, bottom) Fluorescence anisotropy decay calculated from the parallel and perpendicular traces in (a).
a:,-@;, pair on neighboring monomers, followed by the /3i55-/3:4 and the a:,-&, pairs on the same monomer, and the
,$,-pi4
pair on neighboring monomers.
IV. Discussion Absorption. The absorption spectrum of the p i 5 5 chromophore was resolved by comparing the absorption spectra of PC trimers isolated from the wild-type and cpcB/CISSS strains. A comparison of the absorption spectra resolved in PC monomers vs in PC trimers (Figure 2) indicates that the differences between these two spectra are too small to account for the differences in the absorption spectra of (aPCpPC) and (aPC/3pc)3 (Figure la). This narrows the cause of the spectral shift in (apcpPc)upon aggregation to the a 8 4 and/or ,884 chromophores or to a change in coupling between them. On the basis of the absorption spectra of (apcp*) and (apc/3*)3 (Figure lb), one can rule out first-order exciton coupling as the dominant cause of the change in the absorption spectrum of wild-type PC upon aggregation. First-order exciton theory can explain a splitting or shift of the peak position of an absorption band, but the overall oscillator strength of the transition is conserved. The absorption spectrum of (aPC@*) shows only a small shift in peak position upon aggregation to trimers but a large increase in the oscillator strength of the band in the visible region. Since this behavior cannot be explained by first-order exciton theory, a change in the protein-chromophore interaction or chromophore conformation upon aggregation is a more plausible explanation. Alternatively, if higher electronic levels are considered (fistorder exciton theory considers only the first excited electronic state), the oscillator strength of any one transition can change as long as the sum of the oscillator strengths of all transitions is conserved (the sum rule of oscillator strengthsI7). Such a theory has been successfully used to explain the hypochromic
effect observed when random-coil DNA assembles into helical form.I8 Whereas the interaction between the helically stacked bases in DNA induces hypochromism, the a 8 4 and /384 chromophores on adjacent monomers of PC trimers interact headto-tail, an orientation which is conducive to hyperchromism. The increase in the absorption of DNA upon melting is about 40%,19so hyperchromism on the order of that seen in (apcp*)3 is not outside the range of possibility. Notice in Figure l b that the Sz transition in the W region of the spectrum loses oscillator strength as the SI transition gains oscillator strength upon aggregation of (apcp*) from monomers to trimers. This behavior is consistent with a hyperchromism effect but does not constitute definitive proof, because changes in conformation of the phycocyanobilin chromophore have been shown to produce similar effects.20.21 Investigations of the transition dipole moments to states higher than the first excited electronic level in PC trimers are needed if the hypochromism effect is to be modeled quantitatively. Time-Resolved Fluorescence Anisotropy. Two time constants, 0.9 f 0.1 and 40 f 2 ps, are observed in the anisotropic fluorescence decay of (apcp*)3that are not present in the monomer decay. These new modes of decay must be due to energetic exchange between either like or unlike chromophores on adjacent monomers. Since only two chromophore types are present in (aPC/3*)3 and the monomers are arranged into trimers with C3 symmetry, the possibilities for energy transfer between adjacent monomers are limited to ai4-pi4, pairs as shown in Figure 8. This figure, or a84-ai4 1 based on the crystal structure of PC from Mastigocladus laminosus, shows how the a g 4 , p84, and ,8155 chromophores are arranged in (apc/3pc)3. It is clear from the crystal structure that the ai,-p& chromophore pair, separated by 21 8, center-tocenter, should be the most strongly coupled pair in the trimer. The 1.0 ps decay of the anisotropy of (aPCp)3 is assigned to the partially resolved energy transfer processes occurring within the a:,-pi, chromophore pair. The 1.3 f 0.4 ps decay component observed in the time-resolved fluorescence anisotropy of (apcpPc)3 agrees within the uncertainty with the 0.9 f 0.1 ps decay constant resolved in (aPC,8*)3.That this rapid decay component is seen in the anisotropy decays of both (aPC/3*)3 and (apcppc)3 confirms the notion that the ,8155 chromophore is not involved. In the Appendix of the companion paper,’ a model is developed for the anisotropic fluorescence decay expected from a pair of nonidentical chromophores. This model is applied to the pair with the assumption that energy transfer within this pair is sufficiently rapid to be treated separately from energy transfer between any of the other chromophores in (apcp*)3.This assumption is supported both by the wide separation of the two measured exponential decay constants in the anisotropy of (aPcp*)3and by the calculated Forster rate constants (Table 1). The parallel and perpendicular decays can be described as sums of exponentials
/3i4-@i4,
ZpeT(t) = xAperPe-r’rz i
The decay times, zi,are the same in the I,,, and ZpeT functions, but the amplitude factors are different except for the constraint that at t = 0 the ratio of ZPza to ZpeT is 3 (giving an initial anisotropy of 0.4). As shown in ref 1 (eq A3), only two exponential terms are required to describe the parallel and perpendicular decays of a system of two weakly coupled
Debreczeny et al.
8426 J. Phys. Chem., Vol. 99, No. 20, 1995
Figure 8. Arrangement of the chromophores in PC trimers based on the crystal structure coordinates. Chromophores in the same monomer have the same superscript number. The numbering convention is as established by Schirmer et aL2 Our assignments of the observed fluorescence decay constants are also shown (see Table 2 and text for details). TABLE 1: Calculated Fdrster Rate Constants (ns-') for Energy Transfer between Chromophores in Trimeric C-Phycocyanin. See Firmre 8 for the Chromophore Numbering Convention acceptors donors aA4
DL Pt55 ai4 Pi4
PL
ai4
*
2.50 0.3 0.99 f 0.2 0.26 f 0.03 275 i 31 1.08 f 0.2
&'I
3.85 f 0.4
Pls5
0.15 f 0.01 1.82 f 0.2
18.8 i 3 0.59 f 0.06 3.38 i 0.4 0.32 & 0.06
ai4
Pi4
Pf55
0.26 & 0.03 0.38 f 0.04 0.04 f 0.01
423 f 42 3.38 i 0.4 0.45 i 0.08
0.16 f 0.02 0.03 & 0.003 0.18 f 0.03
0.006 f 0.0006
0.04 f 0.005 0.18 f 0.03
chromophores. The two decay constants (inverse of the decay times) are (i) the overall rate constant for excited-state depopulation and (ii) the sum of the rate constants for forward and back energy transfer and the overall decay of excited-state depopulation. In the present case, the data are fit to a sum of three exponentials (convoluted with the IRF) to take into account the -40 ps decay component (whose origin is discussed below), which also contributes to the anisotropy decay. The lifetimes determined by the fits are 1.0 k 0.2, 39 f 1, and 940 f 10 ps. The 1.0 ps lifetime corresponds to the inverse of the sum of the forward plus back rate constants for energy transfer within the ai,-&, chromophore pair. The rate constant for overall excited-state depopulation will make a negligible contribution to this 1.0 ps lifetime because the excited-state lifetimes of the chromophores in PC are 1-2 ns. The 940 ps lifetime derives from this excited-state lifetime; because the fluorescence decay was measured only to 1 ns, the lifetime is fit to an artificially low value. The residual value of the time-resolved anisotropic fluorescence decay of a two-chromophore system contains information
about the relative orientation of the transition dipole moments of the two chromophores. To extract the residual anisotropy of the 1.0 ps decay within the ai4-&, chromophore pair, the amplitudes of the two longer lifetime components (39 and 940 ps) were summed. The result is a residual anisotropy value of 0.205 f 0.01. If the relative absorbances and emission intensities of the two chromophores at the excitation and fluorescence emission wavelengths are known and the ratio of forward to back energy transfer between the chromophores is also known, the angle between the transition dipoles of the chromophores can be directly calculated (eq A3d in ref 1). From the chromophore properties determined in PC monomers, we know that 6,,,(590 nm) = 0.601, ~ ~ ~ ( nm) 5 9 0= 0.399, faM(650 nm) = O.475,fa8,(650nm) = 0.525, and kp&I-a&I/ka&I-pM = 0.65. cx andfx are the relative absorbance and fluorescence of chromophore x at the excitation and emission wavelengths, respectively. Using these values, the angle between the transition dipoles of the d4and pi4chromophores is calculated to be 52". By fitting a straight line through the conjugated
Excitation Energy Transfer in C-Phycocyanin portion of the corresponding chromophores in the crystal structure of PC, Schirmer et ale2predict a value of 67". Next, we consider the assignment of the 40 ps decay component observed in the fluorescence anisotropy decay of (apcp*)3.From the crystal structure, one predicts the second most strongly coupled pair of chromophores between adjacent monomers in the trimer to be the &,-pi, pair (see Figure 8 and Table 1). The pi,-p;, pair, separated by a center-tocenter distance of 36 8, in the center of the ring-shaped PC trimer, is much closer together than the a;,-ai4 pair, separated by 69 A. The only other possibility for energy transfer in (apc/3*)3, not present in the monomer, is within the ai,-&, pair (see Figure 8). These chromophores are separated by 56 A, and from the Forster calculations (Table l) one would predict the sum of the rate constants for forward and back energy transfer to be < 1 ns-', making this pair an unlikely contributor to the observed anisotropy decay. Because the ai4-&, chromophore pair undergoes very rapid equilibration by energy transfer, on the time scale of the 40 ps decay, it is reasonable to assume that the ai4-&, pair is excited-state population equilibrated. The observed 40 ps decay time of (aKp*)3 is assigned to energy transfer between the energetically degenerate ai&$, ai&,, and a$i4 pairs around the trimer ring (see Figure 8). The major routes of energy transfer that contribute to this decay are between the ,884 chromophores on adjacent monomers and between the a84 and ,884 chromophores on the same monomer (see Table 1). Evidence that the 40 ps time constant is due to energy transfer among like chromophores (or like pairs of chromophores) is provided by the time-resolved fluorescence anisotropy spectra of (apc,8*)3 shown in Figure 4. Within the uncertainty due to the signal to noise ratio, the anisotropy is wavelength independent at time zero and remains flat as the anisotropy decays to a final value of 0.07. If energy transfer were to occur on this time scale between chromophores with different emission spectra, the fluorescence anisotropy at long wavelengths would become lower than the anisotropy at short wavelengths as a function of time. Lyle and StruveZ2have shown that the time-resolved anisotropy of a trimer of identical chromophores arranged with C3 symmetry decays as a single exponential with a decay constant that is 3 times the rate constant for energy transfer between chromophores. Lyle and Struve also showed that the residual anisotropy exhibited by a trimer of identical chromophores with C3 symmetry is given by
1 cos2 e - 112 r, = -(3 10 where 8 is the angle between the transition dipole of the chromophore and the C3 axis. To extend Lyle and Struve's model to (apc/3*)3, one can assume that the energetic equilibration within each a:,-&, pair is instantaneous on the time scale of energy transfer between these pairs. In this case, the model consists of three identical pairs of chromophores arranged with C3 symmetry. As shown in the Appendix (eq A2), the fluorescence anisotropy decay can still be described as a single exponential with a decay constant that is three times the rate constant for energy transfer between chromophore pairs. Thus, on the basis of the 40 ps decay constant observed experimentally,the inverse of the rate constant 3 for energy transfer between ai4-&,, ai4-&,, and a,,-&, pairs in the trimer is predicted to be 120 f 6 ps. Unlike Lyle and Struve's model, however, the residual anisotropy is no longer a simple function of the angle between the C3 axis
J. Phys. Chem., Vol. 99, No. 20, 1995 8427 and the transition dipole of a single chromophore. Rather, as shown in eq A2c, the residual anisotropy of (apcp")3is a function of the relative absorbances and emission intensities of the a84 and 8 8 4 chromophores and the ratio of forward to back energy transfer, as well as the cosines of the angles between the C3 axis of symmetry and the transition dipoles of the a84 and p g 4 chromophores (Yas4and ypS4,respectively). Using only the relative absorbances and emission intensities and the ratio of forward to back energy transfer determined in monomeric PC (as listed above), there is not enough information to extract both yas4 and ypS4from the residual anisotropy. However, the cosine of the angle between the a84 and p84 chromophores on the same monomer ( Y ~ ~ , _ ~and ~ , )that between the a:4 and &, chromophores on adjacent monomers (yag4-pa,)can be related to yau and yps4 according to
Yai4-44
-
(1 - YasJ2(1 - YB,J
2
- (yag4-pi4- Ya,,Yp,J +
yag4-pQ4 was determined to be 0.891 from the residual fluorescence anisotropy of (apc,8*) (see companion paper'), and yab4-p84was determined to be 0.618 in the above analysis of the residual anisotropy of the decay within the ai4-&, chromophore pair. Solving eqs 3 and A2c simultaneously for yQ4 and ypg4results in four real solutions:
yas4= cos(29.7") and ypg4= cos(47.1")
= cos(47.1") and yp,, = ~ o ~ ( 3 0 . 8 " )
ya,, = cos(69.3") and yp,, = cos(85.2") ya,, = ~ o ~ ( 8 4 . 2and " ) yp,, = ~ 0 ~ ( 7 0 . 1 " ) From the crystal structure, Schirmer et aL2 predict that Yaw = cos(75") and yps4= cos(61"). The fourth solution above shows the best agreement with the predictions of Schirmer et al., but if correct, this solution indicates that the transition dipoles of the ag4 and p 8 4 chromophores are tilted 9" further from the C3 axis than estimated from the crystal structure. In the above model, it was assumed that excitation is localized on one chromophore at any given time. Alternatively, if the interaction energy within the ai,-,8i4 pair is sufficiently strong to treat the pair as excitonically coupled, the emission might be observed from an exciton state. Beck and S a ~ e r ~ ~ treat the absorption anisotropy decay due to energy transfer around the trimer ring in allophycocyanin (thought to be structurally analogous to (aKp*)3) with such a model. However, it is unlikely that the coherence of the exciton state within the ai4-,8i4 pair would still exist on the time scale for which energy transfer around the trimer ring of (aKp*)3is observed (40 ps). Measured and predicted coherence dephasing times for condensed phase systems are in the subpicosecond Thus, even if the chromophore pair is coherently coupled during the absorption process, after a few picoseconds the state from which emission is observed will be incoherent, and the above treatment of the ai,-& pair should remain valid. Next we consider the time-resolved anisotropic fluorescence of (apc/3pc)3, for which the presence of the ,8155 chromophore must be taken into account in addition to the a 8 4 and ,884
8428 J. Phys. Chem., Vol. 99, No. 20, 1995 chromophores present in (aPC,8*)3. A comparison of the anisotropic fluorescence spectra of (aK,8*)3 and (aKPPc)3 shows that the presence of the ,8155 chromophore has a strong effect on the anisotropy of the emission at wavelengths less than 640 nm. It was shown previously (ref 4, Table 3) that in PC monomers the wavelength of maximum emission from the ,8155 chromophore is about 20 nm shorter than those of the a g 4 or ,884 chromophores. The high anisotropy observed at short wavelengths in the spectrum of compared with the spectrum of (aPC/Y)3 indicates that in PC trimers, as in monomers, the 8155 chromophore emits at higher energies than the other chromophores. To compare the shape of the fluorescence spectrum of the chromophore as emitted from PC in the monomeric and trimeric states, the emission spectrum of the ,8155 chromophore resolved from TCSPC studies of pPc is used to model the anisotropy spectrum of (aPCpPC)3 at the earliest measured delay time (t = 0 ps). The modeling relies on the fact that the observed fluorescence anisotropy from a multichromophore system is the population and emission intensity weighted sum of the anisotropies of each of the different emitting species.26 The time-resolved fluorescence anisotropy spectrum, r(t,A), of (aPC,8'")3 can thus be described as
Debreczeny et al.
a 0.451
T
0
li
-
0.30
I I
620
I
I
640 660 Emission Wavelength (nm)
I
r
680
700
b
I
620
where A,, is the wavelength of the observed emission,& is the fluorescence emission spectrum of chromophore x weighted by the fluorescence quantum yield of chromophore x, and Px is the excited-state population of chromophore x as a function of time. Notice that the a 8 4 and ,884 chromophores are grouped into a single term, (apc,8*)3. The emission due to the a g 4 and ,884 chromophores can be approximated by the steady-state emission spectrum of (aK/Y)3 because the a 8 4 and ,884 chromophores undergo very rapid energetic equilibration by energy transfer and the anisotropy spectrum3of )*pcpa( does not change its shape (Figure 4) at times > 1 ps. At time zero, the anisotropy of the ,8155 chromophore is assumed to be 0.4, as would be the case if this chromophore had not yet undergone energy transfer, and as evidenced by the fact that the anisotropy is 0.4 at the shortest wavelengths (Figure 5 , also supported by the Forster calculations shown in Table 1). The initial relative excitedstate populations of the ,8155 chromophore and the a534-,884 pair are determined by their relative absorbances at the exciting wavelength (590 nm). Earlier in this paper, it was shown how the ,8155 absorption spectrum could be resolved from the absorption spectrum of the a84-,884 pair by comparing the absorption spectra of (aPCB*)3 and (aPC/393. The results showed that the relative absorbance at 590 nm of the ,8155 chromophore and the a84-,884 pair are 0.407 f 0.005 and 0.593 f 0.005, respectively. The square of the difference between the experimental and simulated time zero (aPC,8PC)3 anisotropy spectra was minimized by varyicg the relative fluorescence quantum yields of the ,8155 chromophore and the a84-,884 pair and also varying the t = 0 anisotropy of (aPC/3$)3 (assumed to be the same at all Aem). The resulting simulation agrees with the experimental data within the error range, as shown in Figure 9a. The t = 0 anisotropy of (aK,8*)3 derived from the fit is 0.27 f 0.01, in good agreement with the value of 0.26 f 0.02 observed directly in the anisotropy decay of (aK,8*)3 (Figure 3b). The emission spectra of the ,8155 chromophore and of
experiment simulation
I
I
640 660 Emission Wavelength (nm)
I
I
680
700
Figure 9. (a, top) Simulated and experimentally observed time-resolved fluorescence anisotropy spectra, at the earliest resolved time, of PC trimers isolated from the wild-type strain. (b, bottom) The fluorescence emission spectra used in the simulation in (a). The p 1 5 5 emission spectrum was resolved in TCSPC experiments on the /3 subunit of PC (see ref 4). The (aPCp)3 emission spectrum was measured directly by steady-state spectroscopy (excitation at 590 nm). (a'Cp*)3 used in the fit are shown in Figure 9b. The good agreement between the experimental points and the simulation in Figure 9a indicates that, within the experimental error of our anisotropy measurement, the emission spectrum of the ,8155 chromophore is the same in PC trimers as that resolved by TCSPC measurements of pPc. Finally, the assignment of the rate constant for energy transfer between ,8l55 and the other chromophores in (apc,8pc)3 is discussed. The perpendicular polarized fluorescence of (aK,8K)3 at the emission wavelength of 624 nm (Figure 7) decays monotonically, whereas at 650 nm (Figure 6) the perpendicular trace contains a rising component. In contrast, the (aPC,8*)3 sample, containing no ,8155 chromophore, has very low relative intensity at 624 nm and shows no rising component in its perpendicular polarized fluorescence at 650 nm. These observations are consistent with the idea that energy transfer occurs from the short wavelength (620-630 nm) emitting ,8155 chromophore to a longer wavelength emitting chromophore (640660 nm) with a differently oriented transition dipole. As with the (apc,8*)3 decays, the parallel and perpendicular fluorescence decays of (apc,8pc)3 were simultaneously fit to sums of exponential terms convoluted with the IRF (eq 1). At the emission wavelength of 650 nm, the data were fit to a sum of three exponentials with decay times of 1.4 f 0.6, 52 f 5, and 970 f 10 ps, with a negative amplitude for the perpendicular component with the 52 ps lifetime. At the emission wavelength of 624 nm, the data were fit to a sum of two exponentials with decay times of 48 f 1 and 880 f 10 ps, both with positive amplitudes. The 1.4 ps decay component observed at 650 nm
Excitation Energy Transfer in C-Phycocyanin
J. Phys. Chem., Vol. 99, No. 20, 1995 8429
0,30j1
experimental points in Figure 10. The 50 ps lifetime is the inverse of the summed rate constants for forward and back energy transfer between the ,8155 chromophore and the other two chromophore types (a84 and ,884) in (aPCpPC)3.
0.25
V. Conclusions
I :\
~
.-cIn
0.20-
0
a
-
0
experiment exponential fit
-2
0
0.150
e
-3 LL
I." 0.00
I
I
1
I
I
0
200
400
600
800
i 1000
Delay (PS)
Figure 10. Isotropic fluorescence decay of PC trimers isolated from the wild-type strain. The decay was calculated from the parallel and perpendicular decays shown in Figure 7a using eq 5 . The laser excitation wavelength was 590 nm and emission was observed at 624 nm. A twoexponential fit (with decay times of 50 and 900 ps) is shown overlaid with the experimental points.
but not at 624 nm is again assigned to the inverse of the sum of the forward and back rate constants for energy transfer within the ai4-&, pair. The approximately 50 ps decay time observed at both wavelengths is due at least in part to energy transfer between the ,8155 chromophore and the other chromophores in (apcppc)3. However, from the analysis of the fluorescence anisotropy decay of (apc,8*)3, one would also expect to observe a 40 ps decay time due to energy transfer between ai&,, ai&,, and pairs around the trimer ring. Fitting the (apcPPc)3decays to an additional exponential component did not resolve this lifetime from that due to energy transfer from the chromophore. Isotropic fluorescencedecay is not sensitive to energy transfer between identical chromophores. Therefore, fitting the isotropic fluorescence decay of (aPCppc)3should allow us to observe the energy transfer between the ,8155 chromophore and any other chromophore with a different emission spectrum, while being insensitive to the transfer of energy between identical pairs of chromophores around the trimer ring. The isotropic fluorescence decay can be calculated from the parallel and perpendicular decays using
The isotropic fluorescence decay of (apcpPc)3 excited at 590 nm and observed at an emission wavelength of 624 nm is shown in Figure 10, as calculated from the parallel and perpendicular traces displayed in Figure 7. The isotropic fluorescence decay of (apcpPC)3 at 624 nm was fit to a sum of two exponentials with lifetimes of 50 f 1 and 900 f 10 and relative amplitudes of 0.61 and 0.39, and the fit is shown overlaid with the
The absorption spectrum of the ,8l55 chromophore resolved in PC trimers is similar in peak position and oscillator strength to that resolved in PC monomers (Figure 2). The red shift of the visible absorption peak in wild-type PC upon aggregation from monomers to trimers (Figure la) appears in large part to be due to an increase in oscillator strength within the as4-ps4 pair of chromophores. This result argues against first-order exciton coupling being the dominant cause of the red shift. Instead we speculate that this effect is due to either hyperchromism between a;, and &, chromophores on neighboring monomers or to new protein-chromophore interactions introduced upon trimer formation. The emission spectrum of the ,8155 chromophore is also similar in the monomeric and trimeric aggregation states. The experimental assignments of the energy-transfer rate constants between chromophores in trimeric PC are summarized in the first row of Table 2 and in Figure 8. The three decay times derived from fits to the anisotropic and isotropic fluorescence decays of (aPcp*)3and (apcpPc)3 are assigned to the following energy-transfer processes. The rapid 1.0 f 0.2 ps decay, observed in the anisotropic decays of both the (apcp*)3 and (aPCpPC)3samples, is assigned to the inverse of the sum of forward and back rate constants for energy transfer within the 2 ai4-,884 pair. This assignment agrees well with the Forster calculation shown in row 2 of Table 2. The 50 & 1 ps decay observed in the anisotropic and isotropic decays of (aPCpPC)3 but not in the (aK,8*)3 data is assigned to the inverse of the sum of forward and back rate constants for energy transfer between the ,8155 chromophore and the other two chromophore types in PC trimers. The close match of this decay time with that assigned to energy transfer between the pl55 and ,884 chromophores in PC monomers (52 ps, ref 4) suggests that the ,884 chromophore on the same monomer is still the primary energy transfer partner for the ,8155 chromophore in PC trimers. This is supported by the Forster calculations, which predict that the inverse of the summed rate constants for forward and back energy transfer between the ,8155 and p 8 4 chromophores is 49 ps. The inverse of the summed rate constants for forward and back energy transfer between the P I 5 5 and all other chromophores (besides the ,884 chromophore on the same monomer) is '800 ps. The 40 & 1 ps anisotropy decay observed in (aPC,8*)3 is assigned to energy transfer between the identical chromophore pairs ai&4, and The Forster calculations indicate that two main processes are responsible for this mode of depolarization: energy transfer between ,884 chromophores on adjacent monomers and energy transfer between a 8 4 and ,884 chromophores on the same monomer. The observed and predicted rate constants for this process are again in close agreement. Some previous experimental assignments of the rate constants in trimeric PC are also summarized in Table 2. Until re~ e n t l y ? the ~ . ~time ~ resolution of most experiments was insufficient to allow observation of the fastest decay constant in (aPC,8PC)3resulting from energy transfer within the a~,-,8~, pair. The assumption in some of these earlier studies that all rate constants were being resolvedz9led to some confusion over the assignments of the rate constants for the .:,-pi4 and ,8t5,-,8A4 pairs. In some of the studies only isotropic fluorescence d e ~ a y l lor- ~absorption ~ recovery3' was observed, in which
Debreczeny et al.
8430 J. Phys. Chem., Vol. 99, No. 20, 1995 TABLE 2: Rate Constants for Energy Transfer between Chromophores in Trimeric C-Phycocyanin
organism
ref
Syn. 7002"
time res
4 4
1 ps
Syn. 7002"
M.laminosus" A . halophytica' W .prolijicad M.laminosus" Syn. 6301'
28 27 31,32 35 I1
Syn. 7002" M.laminosus"
33 33
F. diplosiphod P. luridumk'
3
-
PL
Pl,
-P:,
a
4
-
Our Experimental Results 1.0 f0.2 50f 1 Our Forster Calculations 1.4 f 0.1 49 f 8 Previous Experimental Results 0.5 f0.1 30-100
0.1 ps 0.1 ps
44&-
@:,
40 f 2 46 f 5
0.55
3-lops 5 - 10 ps, 0.4 PS 40-60 ps
31 f 5 27 f 4 120f 10 35 f 3 Previous Forster Calculations 0.37 22 0.33 17 6.2 530 0.5 1.8
200 f 60 14 12 330 0.24
34 a Synechoccus sp. PCC 7002. Mastigocladus laminosus. Aphanotheca halophytica. Westiellopsis prolfica. 'Synechococcus sp. PCC 6301.
f Fremeyella
diplosiphon. 8 Phormidium luridum.
case the 40 ps decay constant here assigned to energy transfer between identical chromophore pairs should be invisible. This, combined with the fact that the predicted and observed lifetime for energy transfer between the SI,, and pi4chromophores is so similar to the 40 ps decay time due to energy transfer between the 44Pi4, d$I$,and d-&4 pairs probably explains why the latter has not previously been resolved. The (aK/?*)3 sample, genetically engineered to be missing the pi55 chromophore, made the resolution of this kinetic component feasible. The assignment of the 1 ps anisotropy decay time in PC trimers to energy transfer between the and pi4 chromophores still holds in the event that these chromophores are excitonically coupled. In this case, the 1 ps decay time would be attributed to interexciton relaxation processes rather than interchromophore processes. However, to date, there has been no convincing evidence of exciton coupling in PC trimers. Xia et uL3*recently invoked exciton coupling within the pair to explain their time-resolved polarized absorption experiments on PC trimers. A 33 ps decay time is attributed to direct ground state relaxation from the upper exciton state. The authors suggest that interexciton relaxation processes also occur but are not resolved by their measurements (10 ps autocorrelation pulse widths). In the same paper, PC monomers were found to decay with a characteristic 52 f2 ps exponential time constant (our results are in agreement). It was not explained why the same decay constant is not observed in the trimers, and no experimental basis was provided for assigning the 33 ps decay constant in trimers to an exciton relaxation process. An alternative interpretation is that the 33 ps decay process observed in the polarized absorption of PC trimers is due to a combination of energy transfer between and pw chromophores on the same monomer and energy transfer between the and d$h pairs on adjacent monomers. Previous calculation^^^^^^^^ of the Forster rate constants for energy transfer in PC trimers are also shown for comparison in Table 2. In these calculations, the spectral differences between the three chromophore types in PC were either neglected3 or approximated by deconvolution procedure^.^^,^^ The improved agreement between our calculated Forster rate constants and the experimentally resolved rate constants is due to a more accurate resolution of the chromophore properties. In combination with previous results,' these results indicate that the
d4
d4-&
Qi4,4$i4,
dominant energy-transfer processes in monomeric and trimeric PC are well described by Forster theory. This is the first detailed confirmation of Forster's theory in a pigment protein. Acknowledgment. This work was supported by the Director, Office of Energy Research, Division of Energy Biosciences, of the U.S. Department of Energy under Contract No. DE-AC0376SF-OOO98 (M.P.D. and K.S.) and by U.S. Public Health Service Grant GM-3 1625 (D.A.B.). Appendix Consider three identical pairs of chromophores arranged with C3 symmetry where the chromophores within each pair are not identical.
Q Treating the energy transfer within each pair as instantaneous on the time scale of the measurement, the excited-state populations of the chromophores as a function of time, if chromophore ul is initially excited, are
Excitation Energy Transfer in C-Phycocyanin
Following the methodology developed in the Appendix of ref 1, the fluorescence anisotropy decay can be described by a single exponential with a rate constant that is three times the rate constant for energy transfer between pairs:
where
ya and y b are the cosines of the angles between the C3 axis of symmetry and the transition dipole moments of chromophores a and b, respectively. yab is the cosine of the angle between the transition dipole moments of the a and b chromophores within a single ab pair (such as al-bl in the figure above).
References and Notes (1) Debreczeny, M. P.; Sauer, K.; Zhou, J.; Bryant, D. A. J. Phys. Chem. 1995, 99, 8412. (2) Schirmer, T.; Bode, W.; Huber, R. J. Mol. Biol. 1987, 196, 67795. (3) Duemng, M.: Schmidt, G. B.; Huber, R. J. Mol. Biol. 1991, 217, 577-592. (4) Debreczeny, M. P.; Sauer, K.; Zhou, J.; Bryant, D. A. J. Phys. Chem. 1993, 97, 9852-9862. ( 5 ) Gindt, Y. M.: Zhou, J.; Bryant, D. A,; Sauer, K. J. Photochem. Photobiol. B 1992, 15, 75-89. (6) Gindt, Y. M. Ph.D. Thesis, University of California, Berkeley, Lawrence Berkeley Laboratory Report LBL-33932, 1993.
J. Phys. Chem., Vol. 99, No. 20, 1995 8431 (7) Yu, M.-H.; Glazer, A. N.: Williams, R. C. J. Biol. Chem. 1981, 256, 13130-13136. (8) Yu, M.-H.: Glazer, A. N. J. Biol. Chem. 1982, 257, 3429-3433. (9) Gottschalk, L.; Fischer, R.: Lottspeich, F.; Scheer, H. Photochem. Photobiol. 1991, 54, 283-288. (10) Zhou, J. Ph.D. Thesis, Pennsylvania State University, 1992. (1 1) Holzwarth, A. R.; Wendler, J.; Suter, G. W. Biophys. J. 1987, 51, 1-12. (12) Shah, J. IEEE J. Quantum Electron. 1988, 24, 276-288. (13) Doust, T. A. M. In Picosecond Chemistry and Biology: Doust, T. A. M., West, M. A., Eds.; Science Reviews: Northwood, Middlesex, U.K., 1982; pp 1-34. (14) Kahlow, M. A.; Jarzeba, W.; DuBruil, T. P.; Barbara, P. F. Rev. Sci. Instrum. 1988, 59, 1098-1109. (15) Debreczeny, M. P. Ph.D. Thesis, University of California at Berkeley; Lawrence Berkeley Laboratory Report LBL-35672, 1994. (16) Lakowicz, J. R. Principles of Fluorescence Spectroscopy: Plenum Press: New York, 1983. (17) ThiBry, J. J. Chem. Phys. 1965, 43, 553-560. (18) Tinoco, I., Jr. Adv. Chem. Phys. 1962, 4 , 113-160. (19) Bush, C. A. In Basic Principles in Nucleic Acid Chemistry; Ts’o, P. 0. P., Ed.; Academic Press: New York, 1974; Vol. 11, pp 91-169. (20) MacColl, R.: Guard-Friar, D. Phycobiliproteins; CRC Press, Inc.: Boca Raton, FL, 1987; pp 58-61. (21) Scheer, H.: Kufer, W. Z. Naturforsch. 1977, 32c, 513-519. (22) Lyle, P. A,: Struve, W. S. Photochem. Photobiol. 1991, 53, 359365. (23) Beck, W. F.;Sauer, K. J. Phys. Chem. 1992, 96, 4658-4666. (24) Kenkre, V. M.: Knox, R. S. Phys. Rev. Lett. 1974, 33, 803-806. (25) Knox, R. S. In Photosynthesis 111: Photosynthetic Membranes and Light Harvesting Systems: New Series ed.; Staehelin L. A., Amtzen, C. J., Eds.: Springer-Verlag: New York, 1986; Vol. 19, pp 286-298. (26) Cross, A. J.; Waldeck, D. H.; Fleming, G. R. J. Chem. Phys. 1983, 78, 6455-6467. (27) Xie, X.; Du, M.; Mets, L.; Fleming, G. R. In Proceedings of SPIE-The Intemational Society for Optical Engineering: SPIE-The Intemational Society for Optical Engineering: Los Angeles, CA, 1992; pp 690-706. (28) Gillbro, T.; Sharkov, A. V.; Kryukov, I. V.; Khoroshilov, E. V.; Kryukov, P. G.; Fischer, R.: Scheer, H. Biochim. Biophys. Acta 1993,1140, 321-326. (29) Holzwarth, A. R. Q. Rev. Biophys. 1989, 22, 239-326. (30) Wendler, J.: John, W.; Scheer, H.: Holzwarth, A. R. Photochem. Photobiol. 1986, 44, 79-85. (31) Xia, A. D.: Zhu, J. C.; Jiang, L. J.; Li, D. L.: Zhang, X. Y. Biochem. Biophys. Res. Comm. 1991, 179, 558-564. (32) Xia, A. D.: Zhu, J. C.; Wu, H. J.; Jiang, L. J.; Zhang, X. Y.; Sudha, M.; Maruthi Sai, P. S. J. Photochem. Photobiol. B: Biol. 1993, 19, 111117. (33) Sauer, K.; Scheer, H. Biochim. Biophys. Acta 1988, 936, 157170. (34) Grabowski, J.; Bjom, G. S. In Photosynthetic Light-Harvesting Systems: Organization and Function; Scheer, H., Schneider, S., Us.: Walter de Gruyter: Berlin, New York, 1988; pp 491-506. (35) Sandstrom, A.; Gillbro, T.; Sundstrom, V.; Fischer, R.; Scheer, H. Biochim. Biophys. Acta 1988, 933, 42-53.
JP94307OZ
