17388
J. Phys. Chem. 1996, 100, 17388-17394
Femtosecond Spectroscopy of a 13-Demethylrhodopsin Visual Pigment Analogue: The Role of Nonbonded Interactions in the Isomerization Process Qing Wang,†,‡ Gerd G. Kochendoerfer,‡ Robert W. Schoenlein,† Peter J. E. Verdegem,§ Johan Lugtenburg,§ Richard A. Mathies,*,‡ and Charles V. Shank†,‡ Materials Sciences DiVision, Lawrence Berkeley National Laboratory, Berkeley, California 94720, Department of Chemistry, UniVersity of California, Berkeley, California 94720, and Leiden Institute of Chemistry, Leiden UniVersity, 2300 RA Leiden, The Netherlands ReceiVed: April 18, 1996; In Final Form: August 7, 1996X
The photoisomerization reaction of the visual pigment analogue 11-cis-13-demethylrhodopsin is investigated using femtosecond pump-probe techniques. Following excitation with a 40-fs pump pulse at 500 nm, differential transient absorption spectra are measured from 470 to 560 nm using a 10-fs probe pulse centered at 500 nm and from 560 to 650 nm using a 10-fs probe pulse centered at 620 nm. The persistence of the excited-state absorption, the recovery kinetics of the ground-state bleach, and the formation time of the photoproduct absorption all indicate that this photoisomerization reaction is complete on the 400-fs time scale. Comparison of the reaction dynamics of 11-cis-rhodopsin with 11-cis-13-demethylrhodopsin suggests that the removal of the nonbonded steric interaction between the C13-methyl group and the C10-hydrogen atom slows down the initial torsional dynamics and that this in turn results in a lower isomerization quantum yield and a slower overall reaction time. Our results support the hypothesis that the quantum yield and the speed of the isomerization reaction are coupled according to a dynamic potential-surface crossing mechanism.
Introduction The molecule responsible for light detection in the process of vision is the membrane protein rhodopsin. Rhodopsin consists of the 11-cis isomer of retinal covalently bound to a lysine residue of the protein opsin through a protonated Schiff base linkage. Absorption of a photon by rhodopsin initiates the photoisomerization of the chromophore which is followed by a series of thermal reactions that trigger the excitation of the retinal rod cell.1-3 Recently, we have shown that in room temperature solution this light-induced isomerization reaction is a vibrationally coherent photochemical process that is complete in only 200 fs.4-6 These results suggest that, following photoexcitation, the photoproduct is formed through an essentially barrierless transition from the excited-state potential energy surface of the reactant to the ground state of the photoproduct. The potential surfaces are schematically indicated in Figure 1. It has been postulated (1) that the surprisingly fast isomerization dynamics in rhodopsin are facilitated by the nonbonded steric interaction between the 13-methyl group and the 10-hydrogen atom in the 11-cis chromophore and (2) that the quantum yield in vision depends on the speed of isomerization because the isomerization mechanism involves a Landau-Zener dynamic internal conversion process.4-6 To test these hypotheses, we first measured the femtosecond photoisomerization dynamics of a rhodopsin analogue, isorhodopsin.7 Isorhodopsin is an isomer of rhodopsin containing a 9-cis-retinal chromophore in place of the natural 11-cis-retinal chromophore in rhodopsin. The nonbonded steric interaction between the C13-methyl group and the C10-hydrogen atom is absent in the 9-cis chromophore. Our measurements show that the isomerization reaction rate of isorhodopsin is 3-fold lower than that in rhodopsin. The reduced reaction rate correlates with the 3-fold reduction of the quantum yield of the pigment. †
Lawrence Berkeley National Laboratory. University of California, Berkeley. § Leiden University. X Abstract published in AdVance ACS Abstracts, October 1, 1996. ‡
S0022-3654(96)01150-1 CCC: $12.00
Figure 1. Schematic ground- and excited-state potential energy surfaces for the 11-cis f all-trans isomerization in rhodopsins. The dashed lines connecting the S1 and S0 potentials represent the nonadiabatic surface upon which the isomerization occurs.
Although these results support our hypothesis, the difference in the structure between the rhodopsin and isorhodopsin chromophores may also play a role in changing the photophysical properties because of the altered spatial relationship between the cis double bond and the surrounding protein pocket. An alternative approach to testing our hypotheses is to replace the C13-methyl group of the rhodopsin chromophore with a much smaller hydrogen atom, thus removing the relevant nonbonded interaction in as isomorphous a manner as possible. The © 1996 American Chemical Society
13-Demethylrhodopsin Visual Pigment Analogue
J. Phys. Chem., Vol. 100, No. 43, 1996 17389 Experimental Section
Figure 2. Schematic diagram of the structural changes that occur upon photoisomerization of rhodopsin and 13-demethylrhodopsin. Steric interaction between the C13-methyl group and C10-hydrogen atom in rhodopsin is eliminated in the 13-demethyl analogue.
rhodopsin analogue in which the sterically hindered C13-methyl group is removed from the 11-cis-retinal chromophore is called 13-demethylrhodopsin (Figure 2). After photoexcitation, 13demethylrhodopsin isomerizes about the same C11dC12 double bond as rhodopsin and produces the all-trans photoproduct. Since 13-demethylrhodopsin and rhodopsin have the isomerizing double bond in the same position, the experiments should provide a better test of the role of the 13-methyl group nonbonded interaction in the photochemical process. Here we present the results of femtosecond pump-probe studies on 13-demethylrhodopsin. The isomerization reaction rate of this rhodopsin analogue is significantly slower than that of rhodopsin. The photoproduct formation time is ∼400 fs, which is about 2 times slower than rhodopsin. The slower isomerization speed of 13-demethylrhodopsin correlates well with the reduced reaction quantum yield compared to that of natural rhodopsin (0.67).8 These results suggest that the steric interaction between the 10-H and 13-methyl group is one of the forces that accelerates the molecule along the isomerization coordinate in this ultrafast reaction. Furthermore, our results support the idea that the speed of the torsional dynamics is an important factor that determines the isomerization quantum yield.
The femtosecond laser system is similar to that described in our previous publications.9 Briefly, the system consists of a colliding-pulse, mode-locked dye laser that produces 50-fs pulses at 620 nm. The femtosecond pulses are amplified in a dye amplifier pumped by the second harmonic of a Continuum YAG laser operating at 540 Hz. The amplified femtosecond pulses have energies of a few microjoules per pulse. Part of the amplified pulse is used to generate an ultrashort red probe pulse (10 fs, 560-650 nm) using an optical fiber with a sequence of gratings and prisms for phase compensation. The remainder of the amplified red pulse is used to generate a white-light continuum in a jet of ethylene glycol. The blue-green portion of the continuum is re-amplified to the microjoule level in another dye amplifier pumped by the third harmonic (355 nm) of the same YAG laser. Part of the amplified green pulse is sent through the fiber-grating-prism compressor to obtain the green probe pulse (10 fs, 450-580 nm). The pump pulse (40 fs, 500 nm) is taken directly from the second amplifier. The pump (∼0.5 mJ/cm2) and probe (∼0.04 mJ/cm2) beams are crossed in a 0.3-mm jet of flowing 13-demethylrhodopsin solution in CHAPSO detergent buffer. The flow rate is sufficient to replace the photolyzed sample between successive pump pulses. Kinetic information at specific wavelengths is obtained by filtering the probe pulse after the sample (∼7 nm band-pass) and combining differential detection with lock-in amplification as the pump-probe delay time is continuously varied. The signal (∆T/T) was a few percent or less and was found to be linear with pump power. Rod outer segments (ROS) are isolated from 200 bovine retinas by sucrose floation followed by sucrose density gradient centrifugation as described previously.10 The ROS are bleached in the presence of 20 mM NH2OH and washed with PIPES buffer three times to remove excess NH2OH. The resulting opsin suspension is dissolved in 20 mM CHAPSO buffer and mixed with a 3-fold excess of the retinal analogue for 4 h. The regeneration of the pigment is followed by monitoring the absorbance at 500 nm. After no further increase of absorbance is detected, the regeneration mixture is applied to a Concanavalin A-Sepharose 4B affinity column that has been equilibrated with 20 mM PIPES (pH 6.5) supplemented with 20 mM CHAPSO, 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, 1 mM MnCl2, and 0.1 mM EDTA. The column is thoroughly washed until the eluent does not show any residual absorption around 380 nm due to excess retinal and retinal oxine. The pigment is eluted with buffer supplemented with 200 mM R-methylmannose. The absorption maximum of the regenerated 13-demethylrhodopsin is at 498 nm, consistent with previously published results.11-13 The eluent is diluted by a factor of 4 to reduce the mannose concentration and concentrated to an optical density of 10-13 OD/cm. Results Time-resolved measurements of the 13-demethylrhodopsin photolysis at wavelengths between 470 and 650 nm are presented in Figure 3. Data from 470 to 560 nm are obtained using the 10-fs blue probe, while data from 550 to 650 nm are obtained using the 10-fs red probe pulse. There is no significant difference between measurements at 550 and 560 nm obtained with the blue and the red probe configurations. In the 470530-nm measurements, an absorption feature (∆T/T < 0) appears instantaneously after the excitation pulse. This absorption signal is attributed to an excited-state absorption from the first excited state to higher lying excited states (S1 f Sn). The excitedstate absorption decays quickly revealing the ground-state bleach
17390 J. Phys. Chem., Vol. 100, No. 43, 1996
Wang et al.
Figure 3. Transient absorption measurements of 13-demethylrhodopsin at various probe wavelengths (10-fs probe) following excitation by a 40-fs pump pulse at 500 nm. Panel A presents measurements recorded from 200 to 800 fs, and B presents measurements recorded from 0 to 6 ps.
(∆T/T > 0), which reaches a maximum value at ∼150-200 fs. Time-resolved measurements at 510 nm on a 6-ps time scale (Figure 3B) show that the recovery of bleach can be fit to a single-exponential decay with a 1.5-ps time constant. Approximately 60% of the initial ground-state bleach (determined by extrapolating the exponential decay curve to zero time) is recovered at 6 ps, which is consistent with current estimates of the reaction quantum yield (0.47).14 Measurements at 540 nm also show an instantaneous excited-state absorption. As the excited-state absorption disappears, the ground-state bleach signal (∆T/T > 0) is revealed,which is dominant at ∼100 fs. Beyond this time, the ground-state bleach is rapidly masked by the photoproduct absorption (∆T/T < 0), which grows in further on a 2-ps time scale. The photoproduct formation dynamics are probed at wavelengths from 540 to 650 nm. The absorption signal at 560570 nm, which is near the absorption maximum of the photoproduct, does not fully develop until 350-450 fs after the excitation pulse. The absorption signal at 550 and 560 nm decays very little from 400 fs to 6 ps. However, at the redder probe wavelengths (590, 610, 630, and 650 nm), the absorption signals reach maxima at earlier times and decay with time constants of 2.1, 1.3, 0.96, and 0.90 ps, respectively. Similar to what has been observed in rhodopsin and isorhodopsin, the wavelength dependence of the appearance and the decay of the photoproduct absorption feature is consistent with vibrational and/or conformational relaxation.5,6 Spectra are constructed from the time-resolved data at different delay times from -100 to 400 fs in order to more clearly reveal the photoproduct formation dynamics (Figure 4). The blue probe data are scaled to the red probe data according to the measurements at 560 nm. The data points are connected through a spline fit to guide the eye. The absorption feature
Figure 4. Differential absorption spectra of rhodopsin generated from the time-resolved traces in Figure 2. Spectra at various time delays of the probe pulse following excitation with a 40-fs pump pulse are shown.
near 525 nm at 36 fs is likely due to the excited-state absorption S1 f Sn. At 100 fs, the excited-state absorption is replaced by the bleach of the ground state. From 100 to 200 fs, the highenergy end of the bleach spectrum shifts slightly toward the blue. This behavior is likely a result of the motion of the wave packet on the excited-state potential energy surface. As the wave packet moves away from the Franck-Condon region, the excited-state absorption shifts to the blue, revealing more of the bleach of the ground-state reactant. A similar behavior of the excited-state population dynamics has been observed in bacteriorhodopsin, rhodopsin, and isorhodopsin.6,7,15 The spectral data from 530 to 630 nm directly examine the dynamics of the photoproduct formation. At 36-fs delay time, the broad absorption feature beyond 550 nm can be attributed to the early
13-Demethylrhodopsin Visual Pigment Analogue appearance of the photoproduct. At later times, as more photoproduct forms and relaxes toward the bottom of the product potential well, the photoproduct absorption band grows in strength and shifts to the blue. Figure 4 shows that the area of the absorption continuously increases until 400 fs. Spectra at later times exhibit a blue-shift of the absorption peak and a slight decrease in the amplitude. The blue-shift of the photoproduct absorption is attributed to the vibrational and conformational relaxation of the photoproduct, and vibrational cooling of the returning hot ground-state reactant, which will be discussed later. The decrease of the amplitude can be caused by two effects, the cooling of the hot reactant in the ground state, and the blueshifting of the absorption spectrum of the photoproduct which is partially canceled by the reactant bleach signal. Discussion Femtosecond pump-probe data have been used to study the initial dynamics of the isomerization reaction in a rhodopsin analogue, 13-demethylrhodopsin. Comparison of the reaction dynamics between demethylrhodopsin, rhodopsin, and isorhodopsin provides important information for understanding the molecular mechanism of the isomerization in the first step in vision. Interpretation of 13-Demethylrhodopsin Femtosecond Data. Upon optical excitation of 13-demethylrhodopsin, an increase in differential transmission (bleach) should appear instantaneously. The bleach signal should be centered at the maximum of the reactant absorption band. However, in demethylrhodopsin, the initial bleach is masked by an absorption band which is centered at ∼525 nm (Figure 4). We attribute this feature to an excited-state absorption from the first to higher excited states (S1 f Sn) when the population is near the FranckCondon region of the excited-state potential energy surface. At 100 fs, the excited-state absorption feature disappears and the bleach signal is revealed. Subsequently, an apparent shift of the bleach signal toward the blue is observed from 100 to 400 fs. Because the differential transient absorption in this spectral region is due to the combination of excited-state absorption and bleach, the broadening of the blue side of the bleach signal may be explained by the shift of the excited-state absorption spectrum toward the blue. The evolution of the excited-state absorption spectrum directly reflects the initial dynamics of the wave packet on the excited-state surface near the Franck-Condon region. As the excited-state wave packet moves away from the FranckCondon region, the excited-state absorption band shifts to the blue, revealing first the red side of the bleach, then the whole bleach spectrum. Similar absorption features are also observed in rhodopsin and isorhodopsin.5,6 In rhodopsin, the excitedstate absorption maximum is at ∼500 nm. This absorption feature vanishes by 100 fs, and there is no significant shift of the bleach signal from 100 fs to 6 ps. This suggests that rhodopsin exhibits more rapid initial dynamics of the wave packet than 13-demethylrhodopsin, where this shift is complete in 200 fs. In isorhodopsin, the initial excited-state absorption feature is centered at 530 nm and it persists until after 100 fs. A broadening of the bleach signal toward the blue is observed from 200 to 600 fs. These results are consistent with the idea that the initial excited-state dynamics of 13-demethylrhodopsin are slower than those of rhodopsin, yet faster than isorhodopsin. Furthermore, the lifetime of the excited-state absorption near the Franck-Condon region sets a lower limit on the rate of the isomerization reaction. Stimulated emission appears as an increase in transmission in the red spectral region and is another important indication of the population dynamics on the excited-state. In 13-
J. Phys. Chem., Vol. 100, No. 43, 1996 17391 demethylrhodopsin, similar to rhodopsin, no obvious stimulated emission signal is observed after the excitation pulse.6 The stimulated emission is either masked by the rapid formation of the absorbing photoproduct in the same spectral region or evolves out of the probing spectral window very quickly. In isorhodopsin, whose isomerization reaction rate is 3-4 times slower than that of rhodopsin, a distinct stimulated emission signal from 570 to 670 nm lasts for at least 100 fs. Nevertheless, the comparison between different analogues indicates that the excited-state wave packet dynamics in 13-demethylrhodopsin are faster than those of isorhodopsin. Important information is obtained by comparing the shorttime and long-time dynamics of the bleach signal in 13demethylrhodopsin, isorhodopsin, and rhodopsin. Analysis of the time-resolved data shows that the bleach recovery at 495 nm can be fit to a single-exponential rise of ∼1.5 ps. The single-exponential fit is extrapolated to 0 fs, and the recovery of the bleach at 6 ps is estimated to be ∼60%, consistent with the 0.47 quantum yield of the reaction.14 This observation suggests that the excited-state population returns to the ground state of the reactant and fills in the ground-state hole on the picosecond time scale. Consistent with this interpretation, experiments on octopus rhodopsin (which has a 400-fs photoproduct formation time) reveal a decay of the excited-state population on a ∼2-ps time scale.16 Other effects may also contribute to the filling-in of the bleach signal, such as the cooling of the hot reactant molecules returning from the excited state and the shifting of the photoproduct absorption band from the red toward the blue, thereby canceling the bleach. To determine the formation time of the photoproduct, we need to correctly identify the formation time of the hot reactant which absorbs in the same spectral region as the photoproduct. If the hot reactant forms quickly on a few hundred femtosecond time scale, followed by the cooling of the population on a picosecond time scale, as was observed in isorhodopsin, we would expect a fast recovery of the bleach at early times due to the hot reactant formation, followed by a slow recovery of the bleach at later times due to cooling.7 The single-exponential filling-in of the bleach signal at 495 and 500 nm for the 13-demethyl pigment and the agreement between the recovery of the bleach and the quantum yield strongly indicate that the filling-in of the hole occurs on the picosecond time scale, similar to the time scale of the cooling of the hot reactant.17 The most direct measurement of the dynamics of the photoproduct formation is made in the spectral region where the initial photoproduct absorbs. Figure 5 presents a comparison of rhodopsin and 13-demethylrhodopsin isomerization kinetics probing at 570 and 610 nm. At 570 nm, where the bulk of the initial photoproduct absorbs, rhodopsin shows a fast increase in photoproduct absorption which is complete in only 200 fs. On the other hand, the photoproduct absorption of demethylrhodopsin is not fully developed until 400 fs, which is consistent with the temporal evolution of the area of the photoproduct absorption spectra shown in Figure 4. This comparison suggests that the isomerization reaction of demethylrhodopsin is about two times slower than that of rhodopsin. At 610 nm, which is on the red edge of the primary photoproduct absorption spectrum, we can examine the population dynamics closer to the crossing region. Both rhodopsin and demethylrhodopsin exhibit a fast appearance of the photoproduct absorption. The rhodopsin data also exhibits a sharp peak at 200 fs, which is absent in 13-demethylrhodopsin. The differences in the rhodopsin and 13-demethylrhodopsin measurements may be interpreted using the following model: The photoexcited rhodopsin molecules maintain a relatively
17392 J. Phys. Chem., Vol. 100, No. 43, 1996
Figure 5. Comparison between the time-resolved traces of 13demethylrhodopsin and rhodopsin at probing wavelengths of 570 and 610 nm.
narrow wave packet at the crossing region while the demethylrhodopsin wave packet is more spread out. The narrow peak of rhodopsin signal at 610 nm and redder wavelengths is due to a relatively narrow wave packet moving across the spectral windows. However, the broad wave packet in 13-demethylrhodopsin washes out the coherent features that are observed in rhodopsin. The difference in the width of the wave packet between rhodopsin and 13-demethylrhodopsin may be due to the difference in their excited-state potential energy surfaces. However it is also likely that the steric interaction between the C10-hydrogen atom and the C13 methyl group in rhodopsin produces a much tighter ground-state torsional nuclear distribution than that in 13-demethylrhodopsin. The spectra presented in Figure 4, compared with previously published data of rhodopsin, also support the above model. In 13-demethylrhodopsin, the photoproduct absorption band starts to appear in the red wavelength region at very early times, suggesting that a portion of the population quickly crosses through the transition state to reach vibrationally excited levels of the product well. This absorption band keeps growing and shifting to the blue until 400 fs, which indicates that the formation of photoproduct is not complete until 400 fs. In rhodopsin, after the photoproduct absorption reaches a maximum at 200 fs, there is no significant change in the shape or the amplitude of the photoproduct absorption band thereafter. Furthermore, the width of the demethylrhodopsin photoproduct absorption band from 36 to 400 fs is much broader than that of rhodopsin in the red spectral region at comparable times, consistent with the picture that demethylrhodopsin molecules cross the transition region in a much broader wave packet. The dispersed wave packet in 13-demethylrhodopsin likely washes out any vibrational coherence of the photoproduct. However, the relatively narrow wave packet found for rhodopsin is maintained during the transition process, and vibrational coherence of the photoproduct is observed, supporting the nearly barrierless nature of the curve-crossing process in rhodopsin.4
Wang et al. The picosecond kinetics of the induced absorption reveal the dynamics of the conformational relaxation of the photoproduct and of the hot reactant cooling in the ground state. Population cooling on a harmonic surface will cause the spectra to become narrower,7 and vibrational cooling on an anharmonic surface will shift the spectrum to the blue. In rhodopsin and its analogues, the photon energy is stored in the form of conformational distortion.18 The excess vibrational energy in the photoproduct is less than one-half that of the hot reactant. Thus, the vibrational cooling of the hot photoproduct has a smaller effect on the spectral dynamics than the vibrational cooling of the hot reactant. This is apparent in the lack of spectral narrowing of the photoproduct absorption in rhodopsin. Conformational relaxation is another factor that affects the kinetics of the induced absorption band. After photoisomerization, the primary photoproduct is conformationally distorted and the protein pocket adjusts to accommodate the isomerized chromophore. The conformational relaxation of the chromophore and the protein will cause the photoproduct to lower its energy to a more relaxed ground-state form, which may shift the photoproduct absorption spectrum to the blue. Conformational changes of the molecules returning to the ground state of the reactant will be small since the molecules return to their original configuration and conformation. In rhodopsin, because 70% of the photoexcited molecules isomerize to form the photoproduct, the dominant picosecond dynamics in the red region are due to the conformational relaxation of the photoproduct, whose absorption maximum shifts to the blue. The cooling of the hot reactant has a small effect on the red part of the spectrum, since only 30% of the population returns to the ground state of the reactant. In isorhodopsin, on the other hand, the cooling of the hot reactant has a large effect on the dynamics of the red absorption band, since 78% of the excited population forms hot reactant. In 13-demethylrhodopsin, about 47% of the excited population isomerizes to form the photoproduct and ∼53% of the population returns to the reactant ground state. Both vibrational cooling of the reactant and the conformational relaxation of the photoproduct are important to the picosecond dynamics in the red spectral region. Since the nature of these two processes and their effects on the absorption spectra are not entirely clear, it is difficult to decompose their contributions to the transient absorption spectra. Nevertheless, we observe a wavelength-dependent decay of the induced absorption from 570 to 650 nm on the picosecond time scale. This behavior is consistent with either a narrowing of and/or a shifting of the absorption spectrum due to the conformational relaxation of the photoproduct and the cooling of the hot reactant. Implications for the Photochemical Mechanism in Vision. Comparison of femtosecond pump-probe measurements on 13demethylrhodopsin with corresponding measurements on rhodopsin demonstrates that there is a correlation between the reaction quantum yield and the formation time of the photoproduct. It has been suggested that the transition of rhodopsin molecules from the excited state of the reactant to the ground state of the product cannot be described by a static model, since such a model assumes internal conversion from a thermalized population on the excited state. Instead, the transition quantum yield from the excited state to the ground state of the photoproduct is strongly influenced by the speed of the nonstationary nuclear wave packet along the reaction coordinate at the crossing region. This process is a complicated multidimensional electronic and nuclear relaxation that is best described, for example, by the semiclassical theory developed by Miller and George.19 In the absence of more detailed modeling, intuition may be provided
13-Demethylrhodopsin Visual Pigment Analogue by the one-dimensional Landau-Zener model where PLZ ∼ exp(-1/V) is the adiabatic curve-crossing probability (probability of the wave packet remaining on the diabatic surface) and V is the nuclear velocity along the reaction coordinate at the crossing region. According to this formalism, given the quantum yield of the reaction in rhodopsin (0.67), 13-demethylrhodopsin (∼0.47), and isorhodopsin (0.22)20 and assuming that the coupling between the excited state and the ground state are the same, then the speed of reaction in demethylrhodopsin and isorhodopsin would be respectively ∼2 and ∼4 times slower than that of rhodopsin, which is very close to the experimental observation. Given the roughness of the estimation and the fact that the potential energy surface of isorhodopsin is not isomorphous with that of rhodopsin,3 the agreement between the model and the experiments is quite good. Furthermore, our experimental measurements suggest that the initial excited-state dynamics out of the Franck-Condon region, which are probed by the excited-state absorption and stimulated emission, are correlated to both the overall speed and the quantum yield of the reaction. The initial excited-state dynamics in these systems have also been probed by fluorescence quantum yield measurements. These quantum yield measurements result in excitedstate fluorescence lifetimes of ∼50 fs for rhodopsin and ∼100 fs for isorhodopsin,21 consistent with the correlation discussed above. Comparison of the ultrafast dynamics of rhodopsin and 13demethylrhodopsin also provides insight into the mechanism of the isomerization reaction in the first step in vision. Our data suggest that the wave packet of 13-demethylrhodopsin is broad when it reaches the crossing region and that the rate of the reaction is slower than that of rhodopsin. Figure 1 depicts schematic potential energy surfaces for the isomerization of rhodopsins. Due to the C10-hydrogen and C13-methyl group steric interaction, the initial ground state of rhodopsin is distorted along the C11dC12 double bond and the angular distribution is confined. Due to the same steric interaction, the excited state potential energy surface has a steep slope along the isomerization coordinate. In 13-demethylrhodopsin, where the steric potential has been eliminated, the ground-state wave packet is broader and the excited-state potential energy surface has a smaller slope along the isomerization coordinate. Thus, the removal of the steric potential results in a broader initial excited-state wave packet and slower initial dynamics out of the Franck-Condon region. This model is supported by resonance Raman experiments that show that the groundstate reactant of rhodopsin is distorted along the C11dC12 double bond due to the steric interaction between the C13-methyl group and C10-hydrogen atom.22 On the other hand, the geometry of 13-demethylrhodopsin near the C11dC12 double bond should be closer to planar. This is consistent with the reduced C11dC12 hydrogen out-of-plane wagging intensity in 13-demethylrhodopsin.14 QCFF/PI calculations confirm this structural difference between the 11-cis-retinal protonated Schiff base (11-cis-PSB) and 13demethyl-11-cis-retinal PSB.14,23-25 In addition, resonance Raman data suggest that, in rhodopsin, the torsional mode localized at the C11dC12 double bond has significant displacement between the ground state and the excited state due to the initial twist of this double bond. This displacement can accelerate the excited molecules along the reaction coordinate. However, this effect should be absent in 13-demethylrhodopsin because its ground-state geometry is more relaxed.25 Consistent with this expectation, the resonance Raman spectrum of 13demethylrhodopsin exhibits a dramatic reduction of the C11dC12 torsional intensities.14
J. Phys. Chem., Vol. 100, No. 43, 1996 17393 The speeds of reaction in both 13-demethylrhodopsin and isorhodopsin are significantly slower than that in rhodopsin, yet they are still on the ultrafast time scale (