INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Rhodopsin (Rh), the most widely found photoreceptor within the rod and cone cells in vertebrates, as well as in most invertebrates, is recognized as the fundamental molecular entity by which visual processes function. Rh transmembrane proteins (e.g., bovine Rh) typically contain 348 amino acids arranged in seven -helical structures. The biochemical function of Rh relies on the molecular properties of the retinal chromophore and its interactions with the surrounding amino acid environment, especially with those amino acid groups that form the retinal binding pocket. The characterization of the room-temperature Rh (Rh^RT) functionality in terms of the intermediates that comprise its photoreaction provides a molecular-level understanding of visual processes in both vertebrates and invertebrates. This information may also be relevant to an understanding of the corresponding intermediates that function in the much larger, heptahelical protein family (Oesterhelt, 1995).

Rh^RT contains an 11-cis retinal chromophore that is bonded to the apoprotein (opsin) via a protonated Schiff base linkage at Lys²⁹⁶ (Ovchinnikov, 1982). Light absorption by the 11-cis retinal in Rh^RT initiates isomerization to all-trans retinal (Wald, 1968). Recent models suggest that 11-cis to all-trans isomerization occurs on the 200-fs time scale (Wang et al., 1993, 1994; Peteanu et al., 1993; Schoenlein et al., 1991, 1993). The remainder of the Rh^RT photoreaction, spanning times scales from femtoseconds to milliseconds, comprises several intermediates that were initially identified via their respective absorption maxima: photo (~200 fs), batho (542 nm, ~5 ps), blue-shifted intermediate or BSI (477 nm, ~105 ns), lumi (497 nm, ~200 ns), meta I (478 nm, ~50 µs), and meta II (380 nm, ~1 ms). All of these intermediates appear and decay before the dissociation of the retinal chromophore from the apoprotein. Meta I forms an equilibrium with meta II, which is the active state for in vivo Rh^RT (Randall et al., 1991; Thorgeirsson et al., 1993; Lewis et al., 1992; Kliger et al., 1995). The activated Rh^RT functions via G-protein binding at the surface of the transmembrane protein to eventually create a synaptic signal (Franke, 1992).

The mechanistic role of lumi^RT in the overall Rh^RT energy storage/transduction mechanism has not been established, although in general it is thought to function to connect the initial 11-cis to all-trans isomerization with energy transduction to the protein environment. Lumi^RT is formed (105-ns time constant; Kliger and Lewis, 1995) from BSI^RT (in equilibrium with batho^RT) and decays directly to meta^RT I (>50 µs time constant; Kliger and Lewis, 1995). The rate of the lumi^RT to meta^RT I transformation exhibits a strong temperature dependence, which may reflect multiexponential decay pathways with different millisecond time constants (Kliger and Lewis, 1995).

The irreversibility of the Rh^RT photoreaction and the wide range of time scales over which intermediates appear (vide supra) have made it exceptionally difficult to characterize the molecular properties of Rh^RT intermediates. Thus lumi^RT had previously been identified only via transient absorption spectra. The vibrational spectrum of lumi at low temperature has been reported from Fourier transform infrared (FTIR) studies (Ganter et al., 1988, 1990, 1991), but before the PTR/CARS data presented here, no vibrational data on lumi^RT had been reported.

The experimental challenges associated with measuring the vibrational spectrum of lumi^RT are successfully addressed in this study with picosecond coherent anti-Stokes Raman spectroscopy (PTR/CARS). The PTR/CARS techniques and methodology were initially developed to record high-quality (S/N) vibrational data from intermediates in the room-temperature bacteriorhodopsin (BR^RT) photocycle (Weidlich et al., 1997; Jäger et al., 1996; Ujj et al., 1996). The subsequent application of PTR/CARS to the Rh^RT photoreaction independently proved to be exceptionally valuable, because it provided the first opportunities to measure high S/N vibrational spectra from intermediates in this irreversible protein reaction (e.g., batho^RT; Jäger et al., 1997; Popp et al., 1996).

This paper presents the vibrational spectrum (650 cm¹-1750 cm¹) of lumi^RT and utilizes it to elucidate the retinal structure in lumi^RT and the structural changes in the earlier stages of the Rh^RT photoreaction that lead to lumi^RT formation. Comparisons of these PTR/CARS results for lumi^RT with low-temperature FTIR spectra reported previously clarify some vibrational mode assignments for retinal and provide new insight into the related retinal-protein interactions. As a result, vibrational mode assignments, together with a geometric configuration for retinal in lumi^RT, can be derived from these data. The nano/microsecond dynamics of lumi^RT formation are also determined independently from PTR/CARS data. Finally, PTR/CARS spectra recorded during the nanosecond appearance of lumi^RT contain vibrational features that confirm the presence of an intermediate with a retinal structure distinct from that in either batho^RT and lumi^RT (e.g., BSI^RT).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bovine retinas (Lawson Co., Lincoln, NE) are prepared according to published procedures (Papermaster, 1982). The washed rod outer segments of Rh membranes are suspended in 10 mM Tris buffer (pH 7) containing aprotinin (0.1% v/v) and dithiothreitol. These samples have a ratio of 280-nm and 500-nm absorption maxima of ~2 and an optical density (OD) at 500 nm (with background scattering subtracted) of ~3. Large particle scattering from the disc membranes contributes ~2 OD (at 800 nm) to the sample absorbance. To ensure a constant Rh concentration while recording a CARS spectrum (20-30 s), the sample volume was selected to be 25-30 ml. Because the Rh concentration remains constant (<5% decrease) over the several minutes required for each PTR/CARS experiment, the amplitudes and lineshapes of CARS features remain unchanged, and therefore two consecutive CARS measurements can be averaged to improve the resultant S/N.

The instrumentation and experimental procedures used to record PTR/CARS signals are discussed in detail elsewhere (Ujj et al., 1994, 1996), and therefore only a brief description of the fundamental issues is presented here. Several instrumental enhancements specifically important to the measurements of lumi^RT spectrum are described in detail. A schematic representation of the PTR/CARS instrumentation used is presented in Fig. 1.

View larger version (47K): [in this window] [in a new window]   FIGURE 1   (A) Instrumentation used to measure PTR/CARS spectra and PTA data. SHG, second harmonic generator; THG, third harmonic generator; M, mirror; BS, beam splitter; CD, cavity dumper; BE, beam expander; AC, auto-correlator; O, oscilloscope; PD, photo diode; A, aperture; PM, phase-matching adjustment; F, optical filter; G, optical grating; L, lens. The temporal widths of the excitation and the two probe lasers pulses were 3 ps, 5.5 ps, and 7 ps, respectively (400 kHz repetition rates). (B) Schematic representation, based on a perspective view of the sample region indicated by the square bracket in Fig. 1 A, of the laser beam arrangement used to generate PTR/CARS signals. The horizontal arrow indicates adjustable spatial positions of the pump laser beam. The spatial displacement is synchronized with the timing sequence of the respective laser pulse (e.g., p, 1, and s).

The 1053-nm output (30-ps pulses at 76-MHz repetition rate) of a cw, mode-locked Nd:YLF laser (Coherent, Antares 76) is used to generate second and third harmonic radiation from LBO (527 nm) and BBO (351 nm, Coherent 7950 THG) crystals, respectively. The 527-nm and 351-nm radiation is used to pump three independently controlled dye lasers (Coherent, model 700). Each dye laser is equipped with a cavity dumper (Coherent, models 7210 and 7220), three of which are synchronized to the 76-MHz (TEM mode) repetition rate of the Nd:YLF mode locker. The entire laser system is operated at a 400-kHz repetition rate to match the flow properties of the liquid Rh^RT sample jet. The velocity of the Rh^RT sample in the 400-µm-square nozzle is adjusted to 12 m/s (laminar flow) to ensure a complete replacement of the sample volume between the arrival of sets of laser pulses. Critical to this adjustment is the beam waist of the pump beam (20 µm) and the 400-kHz repetition rate of the dye lasers.

A new spatial configuration of the three laser pulses used to generate PTR/CARS signals, together with a two-dimensional multichannel array detector used to measure CARS signals, is introduced into these experiments. PTR/CARS signals are generated simultaneously from two separate sample compartments, reference and Rh (Fig. 1 B). This new sample configuration is intended to minimize long-term intensity drift and spectral shape changes associated with the laser pulses. The reference compartment is a capillary containing a static water sample placed next to the nozzle through which the Rh^RT sample flows. This configuration permits the simultaneous measurement of the nonresonant background from water (Ujj et al., 1994b) and the CARS signal from Rh^RT, thereby minimizing the effects of any variability in the spectral intensity of the laser pulses or in the flow dynamics of the Rh^RT sample. The two sets of probe laser pulses required for these measurements are produced by amplitude division with a pellicle beam splitter (Fig. 1 A). Each pair of probe laser pulses is focused separately into the reference and sample compartments by using the same microscope objective (f = 5 cm). The angle between the laser beams is selected to ensure that the horizontal separation of the focal regions is ~0.8 mm. The signals from both compartments are collimated and focused onto the entrance slit of a triple monochromator (Spex, Triplemate). The wavelength-dispersed CARS signals are focused onto two separate, parallel stripes of a liquid-nitrogen-cooled, CCD multichannel array (Princeton Instruments, LN/CCD-1024-F/1 UV). The spectral resolution of these CARS measurements, determined primarily by the bandwidth of the ₁ laser (vide infra), is <2 cm¹.

The Rh^RT photoreaction is initiated via optical excitation (500 nm, <3-ps laser pulse, 7.5 nJ/pulse). Other than changing the relative concentrations of intermediates, no alteration in the Rh^RT photoreaction is found when the pumping pulse energy varies from 1 nJ to 7.5 nJ. This constancy is determined by monitoring the batho^RT CARS signal 100 ps after excitation of Rh^RT as a function of the pump pulse energy (data not shown). The excitation conditions are selected to minimize any secondary photochemistry involving Rh^RT intermediates by utilizing 1) pump laser pulses that are short (<3 ps) relative to the appearance of batho^RT (~5 ps), 2) probe laser wavelengths (600 nm and 626 nm-670 nm) that are on the low-energy (red) side of the absorption bands for Rh^RT (500 nm) and lumi^RT (497 nm), and 3) low energies of the probe laser pulses (2 nJ and 4 nJ for ₁ and _s, respectively) (Ujj et al., 1996).

Because the spectral region (542-577 nm) in which the CARS signal (_a = ₁ + (_1  s)) appears is ~50 nm to the red of the one-photon absorption maximum of Rh^RT (500 nm with a 50-nm bandwidth, half-width half-maximum), the major resonance enhancement can be attributed to the one-photon transition at _a. No signals are observed that can be attributed to resonances induced by dephasing (i.e., excited-state vibrational resonances) (Andrews et al., 1981). This result is expected because _s is >150 nm removed from the one-photon transition in Rh^RT.

The temporal synchronization of the laser pulses, monitored throughout the CARS experiment with a autocorrelator (FR-103XL; Femtochrome Research), is characterized by a cross-correlation time (CCT) for the two probe laser pulses of ~9 ps. The temporal pulsewidths for ₁ and _s are 5.5 ps and 7 ps, respectively (assuming Gaussian pulse envelopes). The timing jitter between the excitation pulse (₁) and the two probe pulses (₁ and _s) is measured to be <2 ps.

The timing sequences and delays between the three dye laser pulses are selected by three separate but correlated optical delay lines for the picosecond time scale and by electronic delays involving the three independently synchronized cavity dumpers for the nanosecond time scale (Weidlich et al., 1997). Because lumi^RT appears in <500 ns and decays within 50 µs, a 2.5-µs delay is chosen for recording the PTR/CARS data from lumi^RT presented here. Because the velocity of the flowing Rh^RT sample is adjusted to ensure that the irradiated volume is completely replaced in <2.5 µs, it is necessary to spatially displace the positions of the pump and probe laser beams within the flowing Rh^RT sample. Specifically, the two probe beams are displaced downstream from where the pump beam intersects the Rh^RT sample (Fig. 1 B). Coincidentally, the time interval between laser pulses is also 2.5 µs (400 kHz). Consequentially, the pump and probe laser pulses can be synchronized simply by using a pulse from the pump laser to excite the Rh^RT and two successive pulses from the two synchronized probe pulse trains to generate the CARS signal.

The spatial displacement of these pump and probe laser pulses is determined by monitoring the picosecond transient absorption (PTA) (Blanchard et al., 1991) signal from a BR^RT sample placed in the reservoir of the circulating liquid system. At 2.5-µs delay, the L-550 intermediate of the BR^RT photocycle is present (e.g., Lohrmann and Stockburger, 1992). The PTA signal recorded at 600 nm and at a 2.5-µs delay has an opposite sign relative to the signal assignable to the K-590 intermediate present earlier (50-100 ns) in the BR^RT photocycle (Ujj et al., 1996). An isobestic point for the absorption bands involved appears on the higher energy side (blue) of 600 nm. The pump/probe laser beam alignment is achieved by maximizing the PTA signal at 2.5-µs delay to ensure that the Rh^RT sample initially pumped is probed via CARS.

Procedurally, CARS spectra from Rh^RT are recorded (i.e., picosecond resonance CARS or PR/CARS) when the Rh^RT concentration remains unchanged. PTR/CARS data are subsequently recorded for specific time delays between the pumping pulse (_p) and the two probing CARS pulses (₁ and _s). PR/CARS spectra from Rh^RT are taken in an alternating order with PTR/CARS data to compensate for the decrease in Rh^RT concentration due to the irreversibility of the Rh^RT photoreaction.

	THEORETICAL

CARS signals have complex, dispersive lineshapes that originate from interferences among the several scattered coherent waves attributable to each component of the sample (e.g., nonresonant background of the water solvent, vibrational modes within each molecular species, and resonances emanating from electronic transitions energetically accessed by the excitation wavelengths). The successful modeling of the third-order susceptibility (i.e., ⁽³⁾ or the symmetrical response function in the frequency domain) of retinal protein samples utilizes two major assumptions: 1) complex Lorentzian lineshapes are accurate representations of the measured vibrational bands and 2) the electronic phase factor is constant over the relatively narrow (>300 cm¹) spectral region in which CARS data are measured during a single experiment.

In general, the second assumption is not valid for electronically resonant CARS (e.g., resonance excited-state CARS; Kamalov et al., 1989), but is correct when there is correlation among the phases assigned to different vibrational modes. Because all vibrational phases are normally zero for nonresonant scattering (excitation wavelengths are energetically well removed from electronic resonances), the second assumption is correct when excitation wavelengths are on the low-energy ("red") side of the one-photon electronic absorption transitions. To a high degree of accuracy, these relationships have been confirmed empirically by using data from the ethylenic bands of BR^RT (Ujj et al., 1997). Based solely on the derivative lineshape in CARS data, some information, however, is lost upon ⁽³⁾ analysis.

The S/N in the CARS data presented here is sufficiently large (100/1) and the spectral region measured in one experiment wide enough (~700 cm¹) such that the ⁽³⁾ analysis reveals some quantitative deviations from assumption 2 when the spectra are fit with a common phase value. These deviations over the entire spectral region can be readily included in the ⁽³⁾ analysis by introducing a monotonic phase shift via the anti-Stokes frequencies. Based on ⁽³⁾ theory (Mukamel, 1995), such an _a dependence is expected for a simplified, condensed-phase molecular system. To accurately treat this dependence, a second-order, polynomial function of _a can be introduced in the ⁽³⁾ susceptibility expression. An analytic formalism previously utilized [Equation 3 in (Ujj et al., 1994a)] is modified here:

(1)

where a reactive mixture of Rh^RT (index Rh) and lumi^RT (index lumi) is treated; _Rh = _Rh,0(1 + a_Rh(_a  0) + b_Rh(_a  0)²) and _lumi = _lumi,0(1 + a_lumi(_a  0) + b_lumi(_a  0)²) are the differences between the vibrational and background phases; a and b are the polynomial coefficients; the index 0 refers to the reference frequency (₀); phases (_Rh,0, _lumi,0) are chosen for the spectral region measured; µ is a scaling factor; and the summations run over the vibrational bands of Rh^RT (N_Rh) and lumi^RT (N_lumi), respectively. A_j (A_k) is the amplitude and _j (_k) is the detuning parameter for the jth (kth) vibrational mode with a frequency _j (_k) and a bandwidth 2_j (2_k), where (_j = (_j  (_1  s))/_j.). The relative concentration of lumi^RT is given by  (0    1, but  = 0 when Rh^RT is not optically excited).

The degree of optical conversion from Rh^RT into lumi^RT (), estimated to be 30-35% of the original Rh^RT sample, is determined by the amplitude decrease in CARS bands assigned to Rh^RT (i.e., the decrease in the corresponding ⁽³⁾ fit parameter (1  ) for the Rh^RT concentration in the reactive mixture). For these quantitative ⁽³⁾ analyses, the CARS spectra from Rh^RT and the buffer alone are normalized to the nonresonant background signal measured from the water alone (Eq. 1).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

PR/CARS spectra of Rh^RT and PTR/CARS data of lumi^RT measured in two spectrally overlapping (~700 cm¹ wide) regions are shown together with their respective ⁽³⁾ fits in Figs. 2-5. The corresponding background-free CARS spectra (with Lorentzian lineshapes) are also presented in these figures.

View larger version (29K): [in this window] [in a new window]   FIGURE 2   PR/CARS spectrum of RhRT (three OD sample) in the 640 cm1 to 1340 cm1 region. The nonresonant CARS background signal from water only, indicated by the horizontal dashed line, is used to normalize the PR/CARS signal. The phase parameters are Rh (1100 cm1) = 67°, a = 104, and b = 108. The (3)-fit function (Eq. 1 with  = 0) is shown as a solid line overlapping the PR/CARS data (). The corresponding background-free (Lorentzian lineshapes) vibrational spectrum of RhRT, as derived from the (3) fit, is shown at the bottom. The wavenumber positions of selected bands are also presented. The ordinate for the lower spectrum is scaled arbitrarily.

View larger version (34K): [in this window] [in a new window]   FIGURE 3   PTR/CARS spectrum in the 640 cm1 to 1340 cm1 region of the reaction mixture containing RhRT and lumiRT (37 ± 3% relative concentration) taken at 2.5 µs after photoexcitation of RhRT. The phase factors for lumiRT are found to be lum i (1100 cm1) = 67°, a = 104, and b = 108. The nonresonant background signal from water and opsin, indicated by the horizontal dashed line, is used to normalize the PTR/CARS signal. The (3) fit function (Eq. 1) is shown as a solid line overlapping the PTR/CARS data (). The corresponding vibrational spectrum of lumiRT, derived from the (3) fit by using Lorentzian lineshape functions, is shown at the bottom. The origin positions of the lumiRT bands are indicated by arrows on each spectrum. The ordinate for the lower spectrum is scaled arbitrarily.

View larger version (36K): [in this window] [in a new window]   FIGURE 4   PR/CARS spectrum of RhRT (3OD sample) in the 970 cm1 to 1760 cm1 region. The nonresonant CARS background signal from water only, indicated by the horizontal dashed line, is used to normalize the PR/CARS signal. The phase parameters are Rh (1300 cm1) = 70°, a = 104, and b = 108. The (3) fit function (Eq. 1 with  = 0) is shown as a solid line overlapping the PR/CARS data (). The corresponding background-free vibrational spectrum of RhRT derived from the (3) fit is shown at the bottom. The wavenumber positions of selected bands are also presented. The ordinate for the lower spectrum is scaled arbitrarily. The asymmetrical band at 1545 cm1 is fitted with two bands (see text for details).

View larger version (38K): [in this window] [in a new window]   FIGURE 5   PTR/CARS spectrum in the 970 cm1 to 1760 cm1 region of the reaction mixture containing RhRT and lumiRT (37 ± 3% relative concentration) taken at 2.5 µs after photoexcitation of RhRT. The phase factors for lumiRT are found to be lumi (1100 cm1) = 67°, a = 104, and b = 108. The nonresonant background signal from water and opsin, indicated by the horizontal dashed line, is used to normalize the PTR/CARS signal. The (3) fit function (Eq. 1) is shown as a solid line overlapping the PTR/CARS data (). The corresponding vibrational spectrum of lumiRT, derived from the (3) fit by using Lorentzian lineshape functions, is shown at the bottom. The origin positions of the lumiRT bands are indicated by arrows on each spectrum. The ordinate for the lower spectrum is scaled arbitrarily.

Most of the features assignable to lumi^RT are spectrally resolved in the CARS spectra measured from the reactive mixture containing Rh^RT and lumi^RT (vertical arrows indicate bands in Figs. 3 and 5). The PR/CARS spectrum of Rh^RT (Figs. 2 and 4) agrees well with the corresponding resonance Raman spectrum (Mathies et al., 1977; Callender et al., 1976; Callender and Honig, 1977) and with the PR/CARS spectrum reported previously (Jäger et al., 1997). The vibrational parameters attributed to Rh^RT and lumi^RT (i.e., band origin positions (_i), bandwidths (_i), and amplitudes (A_i)) are presented in Table 1. These band parameters are comparable to those observed in the low-temperature FTIR spectrum of lumi (Ganter et al., 1988) and in the PTR/CARS spectrum of batho^RT (Jäger et al., 1997). A comparison with the batho^RT spectrum is of interest, because it contains an all-trans, strongly twisted retinal configuration. Specific results for lumi^RT are described for selected spectral regions.

Hydrogen out-of-plane and CH₃ rock vibrational region

The 650 cm¹ to 1000 cm¹ spectral region of lumi^RT contains at least seven vibrational bands (Fig. 2) assignable to the hydrogen out-of-plane (HOOP) modes of retinal, with the most intense band located at 944 cm¹. Two bands (940 cm¹ and 946 cm¹) appear near the 944 cm¹ position in the low temperature (173K) FTIR spectrum (Ganter et al., 1988). Only one of these bands (947(6) cm¹) is found in a recent low-temperature FTIR study in which special emphasis is placed on excluding the contribution from isorhodopsin (Maeda et al., 1993; Ohkita et al., 1995). The batho^RT band located at 921 cm¹ (Jäger et al., 1997; Eyring and Mathies, 1979) and the meta I band at 950 cm¹ (Doukas et al., 1978) appear to correlate with the 944 cm¹ lumi band. Although a definitive band near 912 cm¹ is not identified in FTIR spectra, a low-intensity feature appears to be present. The position and width (12 cm¹) of the CARS band at 892 cm¹ are in agreement with this low-temperature FTIR result.

The low-intensity bands below 832 cm¹ found in the CARS data (Fig. 3) have not been observed previously, although the batho^RT CARS spectrum also contains bands in this region. It is important to note that the 752 cm¹/766 cm¹ doublet appears in the CARS spectra of Rh^RT, batho^RT, and lumi^RT at essentially the same position (±2 cm¹).

Two bands are also observed at 1011 cm¹ and 1031 cm¹, with the latter appearing as a shoulder (Fig. 3). Based on the vibrational spectra assignable to the all-trans, protonated Schiff-base retinal in solution (Curry et al., 1982), in BR-570 (Ujj et al., 1994b; Lohrmann and Stockburger, 1992; Smith et al., 1984, 1987; Smith et al., 1987), and in batho^RT (Jäger et al., 1997), these two bands can be tentatively assigned to the CH₃-rocking modes.

C-C stretching (fingerprint) vibrational region

The C-C stretching region contains three prominent vibrational bands at 1162 cm¹, 1189 cm¹, and 1217 cm¹. The corresponding FTIR bands, observed as derivative features, appear at 1160 cm¹, 1205.5 cm¹, and 1222.5 cm¹. The assignment of these bands is made by means of isotopic (¹³C) substitutions at the C₁₄ and C₁₅ positions (Ganter et al., 1988). The 1162 cm¹ band contains mainly C₁₀-C₁₁ stretching character, the 1189 cm¹ band is assigned as a C₁₄-C₁₅ stretching mode, and the 1217 cm¹ band is attributable to the C₈-C₉ stretching mode. In the batho^RT CARS spectrum, a band appearing at 1241 cm¹ is assigned to the C₁₂-C₁₃ stretching mode (Deng et al., 1994). The analogous band for lumi^RT is not evident, but a feature with low intensity (at least 10 times less than that of the 1162 cm¹ band) may be present. The fingerprint bands at 1162 cm¹, 1189 cm¹, and 1217 cm¹ are indicative of the all-trans retinal chromophore (Curry et al., 1982; Smith et al., 1987; Mathies et al., 1987).

The spectral region between 1250 cm¹ and 1400 cm¹ contains bands assigned to the C-C-H in-plane bending modes. At least three such bands are found at 1282 cm¹, 1322 cm¹, and 1363 cm¹. All of these bands have a counterpart in the batho^RT CARS spectrum (Jäger et al., 1997). In contrast, analogs to the bands near 1450 cm¹ observed in lumi^RT are not found in the batho^RT spectrum, which rather contains bands at 1427 cm¹ and 1453 cm¹, the later being more intense.

CC stretching (ethylenic) vibrational region

The most intense CC stretching band in the lumi^RT spectrum is located at 1548 cm¹ (Fig. 5). The excellent reproduction of the 1548-cm¹ bandshape by a single Lorentzian function suggests that the band is assignable to a single isolated mode. Three smaller vibrational bands are identified at 1597 cm¹, 1635 cm¹, and 1659-cm¹.

The ⁽³⁾ analysis of these bands reveals that the 1545 cm¹ feature in Rh^RT (Fig. 4) is composed of two bands (at 1536 cm¹ and 1547 cm¹ (Jäger et al., 1997)). Regardless of whether one or two bands are assumed to be present in the 1545 cm¹ feature of Rh^RT, the lumi^RT band extracted from the PTR/CARS data remains unchanged (±0.5 cm¹) at 1548 cm¹, and only the amplitude of the band is altered, by ~15%. The appearance of a 1548 cm¹ band in the lumi^RT spectrum is consistent with the correlation between the absorption shift (i.e., "opsin shift") and the CC stretching frequency (Jäger et al., 1997; Kakitani et al., 1983). For Rh^RT and lumi^RT, the respective absorption maxima are not significantly different (<5 nm), and therefore the small (<2 cm¹) shift observed is expected. The CN stretching (Schiff base) mode appears to be shifted ~3-7 cm¹ (1659 ± 1.5 cm¹) relative to the band position found in batho^RT (1653 ± 1.5 cm¹) (Jäger et al., 1997).

Low-temperature FTIR spectra of lumi contain several bands in the 1500-1800 cm¹ region (Ganter et al., 1988). Based on the CARS data presented here, it is evident that, other than the bands described here, all of the 1500-1800 cm¹ bands are assignable to vibrational modes in the peptide chain (e.g., amide-I and -II and carboxylic acids).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Retinal configuration in lumi^RT

The mechanistic contribution made by lumi^RT to the overall Rh^RT photoreaction involves the transfer of the initial 33 kcal/mol (Schick et al., 1987; Birge et al., 1989; Cooper, 1979) stored in batho^RT to the protein. The specific retinal/protein interactions utilized are not yet fully characterized, but may involve even the tertiary structure of the protein, including changes in the cytoplasmic LOOP region of Rh^RT. The formation of intermediates late in the Rh^RT photoreaction such as lumi^RT are thought to be key elements in this transfer of energy from the retinal structure to the protein.

The retinal configuration in lumi^RT can be characterized primarily via the CARS spectrum presented here:

1. HOOP bands suggest that the retinal has a planar geometry and is not significantly twisted out of plane.

2. Fingerprint bands suggest that retinal configuration is all-trans.

3. The CC band frequencies indicate that the charge delocalization within the lumi^RT binding pocket is similar to that in Rh^RT, but different from that in batho^RT.

These general conclusions are based on the following specific observations and comparisons.

Vibrational band assignments in lumi^RT

1

1

1

1

Comparison between lumi^RT and batho^RT

1

1

1

1

1

1

1

1

Comparison of lumi^RT and Rh^RT

View larger version (34K): [in this window] [in a new window]   FIGURE 6   PR/CARS spectrum (640 cm1 to 1300 cm1) of RhRT in water (upper) and PTR/CARS spectra of reactive mixtures containing RhRT and its photoreaction intermediates recorded 210 ns, 500 ns, and 2.5 µs after photoexcitation (500 nm, 3 ps, 7.5 nJ) of RhRT. The vertical dashed lines indicate the most prominent, time-dependent spectral changes.

View larger version (27K): [in this window] [in a new window]   FIGURE 7   Background-free (Lorentzian lineshape) vibrational spectrum of RhRT (- - -) and lumiRT () over the 750 cm1-1750 cm1 region. The ordinate scale for the lumiRT spectrum is normalized to 1. The intensity of the 972-cm1 band in the spectrum of RhRT is normalized to the intensity of the 944-cm1 band in the lumiRT spectrum. The origin positions of the most significant bands are indicated by arrows.

1

1

1

Appearance of lumi^RT and BSI^RT

Transient absorption studies have shown that lumi^RT is formed with a 105-ns time constant. Lumi^RT is preceded in the Rh^RT photoreaction by the batho^RT/BSI^RT equilibrium (batho^RT  BSI^RT (74 ns) and BSI^RT  batho^RT (107 ns) (Kliger and Lewis, 1995)). The >50-µs time constant with which lumi^RT decays to meta^RT I exhibits a strong temperature dependence that may reflect multiexponential decay processes extending into the millisecond time regime (Kliger and Lewis, 1995).

PTR/CARS spectra measured with time delays of 210 ns, 500 ns, and 2.5 µs (Fig. 6) reveal significant spectral differences (highlighted by dashed lines) that indicate time-dependent changes in the reactive mixture of Rh^RT intermediates monitored. Specifically, the PTR/CARS spectrum recorded at 210 ns should contain vibrational band(s) assignable to batho^RT and BSI^RT, whereas the spectrum recorded at 2.5 µs should contain a vibrational structure assignable to lumi^RT and Rh^RT. Only lumi^RT and Rh^RT bands are also anticipated to contribute to the 500-ns PTR/CARS spectrum, because the lumi^RT concentration should be significantly less at 500 ns than at 2.5 µs.

Because the 210-ns spectrum contains both batho^RT and BSI^RT contributions, the CARS lineshapes are especially complex. From the quantitative utilization of the PTR/CARS spectrum of batho^RT measured previously (Jäger et al., 1997), it is evident that BSI^RT bands appear near 944 cm¹, 1162 cm¹, and 1000 cm¹. The 210-ns spectrum also contains bands in the 850-950 cm¹ region that are not found in Rh^RT, batho^RT, or lumi^RT spectra. The intensities of bands common to the 210-ns and batho^RT spectra are significantly reduced in the former.

Some information on the retinal structure in BSI^RT can be derived from even these few vibrational features. Like batho^RT, BSI^RT appears to have an all-trans retinal. The twisted retinal chain found in batho^RT is relaxed in BSI^RT, but is not completely planar and static relative to out-of-plane motion. The change in the CH₃-rocking region (1000-1050 cm¹) suggests that retinal/protein interactions are stronger in BSI^RT than in batho^RT, an effect that may be important in establishing the batho^RT  BSI^RT equilibrium.

Mechanistic role of lumi^RT

The PTR/CARS spectra of lumi^RT presented here provide new insight into the role of this intermediate in the Rh^RT photoreaction. The formation of lumi^RT from BSI^RT appears to primarily involve a spatial accommodation for the relaxed (nontwisted) all-trans retinal within the protein binding pocket. It should be noted that low-temperature FTIR data suggest that the retinal in lumi^RT interacts strongly with the peptide chain (Ganter et al., 1991). Most of the optical energy absorbed by Rh^RT has previously been transferred to/stored within batho^RT and utilized in the batho^RT  BSI^RT equilibrium. The latter may involve mostly the transfer of energy from the retinal structure into the protein, although this model requires more detailed examination via methods such as PTR/CARS.