The early receptor current (ERC) represents molecular charge movement during rhodopsin conformational dynamics. To determine whether this time-resolved assay can probe various aspects of structure–function relationships in rhodopsin, we first measured properties of expressed normal human rhodopsin with ERC recordings. These studies were conducted in single fused giant cells containing on the order of a picogram of regenerated pigment. The action spectrum of the ERC of normal human opsin regenerated with 11-cis-retinal was fit by the human rhodopsin absorbance spectrum. Successive flashes extinguished ERC signals consistent with bleaching of a rhodopsin photopigment with a normal range of photosensitivity. ERC signals followed the univariance principle since millisecond-order relaxation kinetics were independent of the wavelength of the flash stimulus. After signal extinction, dark adaptation without added 11-cis-retinal resulted in spontaneous pigment regeneration from an intracellular store of chromophore remaining from earlier loading. After the ERC was extinguished, 350-nm flashes overlapping metarhodopsin-II absorption promoted immediate recovery of ERC charge motions identified by subsequent 500-nm flashes. Small inverted R2 signals were seen in response to some 350-nm flashes. These results indicate that the ERC can be photoregenerated from the metarhodopsin-II state. Regeneration with 9-cis-retinal permits recording of ERC signals consistent with flash activation of isorhodopsin. We initiated structure–function studies by measuring ERC signals in cells expressing the D83N and E134Q mutant human rhodopsin pigments. D83N ERCs were simplified in comparison with normal rhodopsin, while E134Q ERCs had only the early phase of charge motion. This study demonstrates that properties of normal rhodopsin can be accurately measured with the ERC assay and that a structure–function investigation of rapid activation processes in analogue and mutant visual pigments is feasible in a live unicellular environment.
Rhodopsin is the visual pigment of the rod photoreceptor and catalyzes the activation of the G-protein, transducin. Seven transmembrane segments of opsin form a pocket to bind 11-cis-retinal (11cRet),1 forming a chromophore with the lysine (K296) as a protonated Schiff base (PSB+-H). The chromophore isomerizes to all-trans-retinal within 200 fs (Schoenlein et al. 1991). Energy uptake by the pigment is efficient and related to the steric strain of isomerization and the charge separation of the cationic PSB+-H from its anionic counterion (E113−) (Birge et al. 1988). Rhodopsin then undergoes a series of thermal transitions over the time scale of picoseconds through milliseconds to achieve the metarhodopsin-II (Meta-II) spectral conformation. At least two critical and sequential proton exchange mechanisms occur during Meta-II formation: the net transfer of the Schiff base proton to the counterion at E113− and the uptake of two protons into the cytoplasmic membrane surface (Parkes and Liebman 1984; Arnis et al. 1994; Jager et al. 1994). These charge motions are reflected in transitions between at least two respective Meta-II states (Meta-IIa and Meta-IIb) that share the same absorption (λmax 380 nm). The transition from the Meta-I to the Meta-II state is the only endothermic state change that occurs during the thermal dark reactions. This indicates that the spontaneous transition into these states is associated with a large positive entropy. A significant molecular volume increase occurs during the lifetime of the Meta-II states (Lamola et al. 1974; Arnis and Hofmann 1993). The causes of volume expansion are likely to involve changes in sidechain interactions within the membrane, movement of α helices, and the configuration of the cytoplasmic loops that are temporally correlated with formation of the R* state, which allows transducin docking (Khorana 1992; Altenbach et al. 1996; Han et al. 1996). Several critical state transitions in rhodopsin activation are thus spectrally silent. Moreover, the underlying molecular biophysics and forces driving these conformational changes remain unsolved. Proton transfer from the Schiff base to E113− is sufficient to activate a chain of molecular events which result in R* (Longstaff et al. 1986).
Other tools such as Fourier transform infrared spectroscopy or electron spin resonance can sample conformation changes outside the chromophore environment. But, like time-resolved absorption studies, these are currently limited by the need for hundreds of micrograms or milligrams of detergent-extracted and purified rhodopsin, or cysteine mutagenic engineering to allow site-specific attachment of spin probes. Compared with Fourier transform infrared spectroscopy, only the more sensitive electron spin resonance technology can resolve environmental transitions on a millisecond time scale. These tools have nonetheless contributed greatly to our current understanding of the rhodopsin activation process (Farahbakhsh et al. 1993, Farahbakhsh et al. 1995; Fahmy et al. 1993; Altenbach et al. 1996; Farrens et al. 1996; Kim et al. 1997), but await the development of cellular expression systems where the harvest of milligram-order quantities of mutated visual pigments is less of an experimental limitation (Reeves et al. 1996; Sullivan and Satchwell 2000).
The early receptor potential (ERP) is a charge redistribution in rhodopsin associated with protein conformational changes (Cone and Pak 1971). This signal can be measured at the surface of the living eye, across the retina, or across the membrane of single isolated photoreceptor cells. The ERP depends on rhodopsin being uniformly oriented in the plasma membrane and occurs in two distinct phases (Cone and Brown 1967; Govardovskii 1979). The depolarizing R1 component correlates with flash presentation and likely reflects the molecular process of charge separation during isomerization (Trissl 1982; Birge 1990a,Birge 1990b), but its underlying mechanism is not yet established. The millisecond-order R2 phase is hyperpolarizing and correlates with conformational changes leading to the biochemically active R* or Meta-II intermediates (Ebrey 1968; Hodgkin and O'Bryan 1977; Spalink and Stieve 1980; Trissl 1982). Absorption of light by different “spectral” states of rhodopsin generates unique ERP signals (Cone 1967; Cone and Cobbs 1969). This suggests that the “electrical” states of rhodopsin reflect unique charge distributions during conformational changes. Thus, rhodopsin activation results in both spectral and electrical state transitions. A clear correlation has been established between spectral and electrical states (Ebrey 1968; Cone and Cobbs 1969; Spalink and Stieve 1980). However, the measurements of these states reflect different molecular properties. Absorption state transitions reflect the different environments of the chromophore. Since the chromophore lies predominantly in the membrane plane, these transitions reflect principally in-plane charge redistributions. Electrical transitions, which are measured across the membrane, reflect molecular events with vector contributions largely orthogonal to the dipole absorption state of the chromophore environment. Because of this, electrical measurements assess a vector of the activation process that is distinct. An additional advantage is that electrical measurements can simultaneously sample the entire coordinated ensemble of charge motions, dipole reorientations, or interfacial charge transfer reactions across the full thickness of the membrane plane and its local boundary surface layers. With modern cellular electrophysiological tools, charge motions can currently be examined with submillisecond time resolution. Supplementary techniques are likely to extend recording to the nanosecond time domain.
The early receptor current (ERC) of rhodopsin activation is the direct measure of charge flow that underlies the ERP. This nonlinear capacitative current shows saturation, dependent upon the amount of rhodopsin molecules available for activation (Hodgkin and O'Bryan 1977). Protein conformation-dependent currents are quite common (Honig et al. 1986), probably the best example being the gating currents of ionic channels that have features in common with the ERC such as signal waveform and bandwidth (Bezanilla and Stefani 1994). Studies of channel gating currents has clarified the role of the molecular structure during voltage-dependent gating. For example, the localization of gating charges to the S4 helices, and the motions of these elements in the electrostatic field, was advanced through gating current studies of expressed ionic channels. Similarly, the ERC might be used to study the forces that govern rhodopsin state transitions while it resides in a physiologically intact membranous environment, or to analyze electrical state transitions that may be spectrally silent to UV-visible or infrared absorption, or occur in environments not accessible to spin probe introduction.
The ERC of rhodopsin activation in intact photoreceptors has been elegantly studied using gigaohm-seal, whole-cell patch clamping techniques (Hestrin and Korenbrot 1990; Makino et al. 1991). Therefore, techniques were developed in which the ERC could be recorded in a unicellular expression system containing high levels of normal rhodopsin (Sullivan 1996, Sullivan 1998; Shukla and Sullivan 1998; Sullivan and Shukla 1999; Sullivan et al. 2000). This method allows assay of rhodopsin activation with seven to eight orders of magnitude less material than other time-resolved techniques. High fidelity ERC currents are routinely recorded from single fused giant cells containing on the order of a picogram of regenerated rhodopsin (1.5 × 107 molecules). In the current work, the ERC approach is used to investigate activation properties of expressed normal or wild-type (WT) human rhodopsin for comparison to known properties of the native pigment previously studied in situ. Human opsin is expressed and regenerated in a membrane environment of transformed HEK293S kidney cells at densities comparable with that in intact photoreceptors (Sullivan and Shukla 1999). Properties of the WT human visual pigment studied with the ERC are consistent with charge motions originating from activation of the ground state of rhodopsin, within the normal range of photosensitivity, and according to the principle of univariance. ERC signals can be regenerated not only with 11cRet but also with 9-cis-retinal (9cRet). This is a first step toward analogue pigment investigations with the time-resolved ERC tool. Finally, ERC signals were recorded from two mutant human opsin pigments (D83N, E134Q) that have R2 relaxation properties distinct from WT. Initial steps are taken toward the development of a quantitative and analytical approach to investigate rhodopsin structure–function relationships using this electrophysiological method. This is especially relevant because the millisecond-order conformational events during rhodopsin activation are now thought to be electrostatically mediated (Fahmy et al. 1995; Shieh et al. 1997). This study is the first to apply the ERC method to characterize rapid electrical processes during activation in expressed mutant and analogue visual pigments in comparison with the normal process. The ERC approach has the sensitivity and temporal resolution to significantly advance knowledge of the underlying molecular biophysical chemistry of rhodopsin activation.
Materials And Methods
Cell Culture and Fusion
Human opsin-expressing HEK293S cell lines were used for ERC recordings (Sullivan 1998). Stable mutant opsin HEK293S cell lines were developed using cytomegalovirus expression vectors that were transfected using calcium phosphate (Chen and Okayama 1988) and selected in G418 (500 μg/ml). Development and characterization of human opsin cell lines is described elsewhere (Sullivan and Satchwell, 1999). The WT human cell line used in these experiments was WT-12 (∼3 × 106 opsins/cell), the D83N mutant cell line was D83N-11 (∼8 × 105 opsins/cell), and the E134Q cell line was E134Q-7 (∼2 × 105 opsins/cell) (Nathans et al. 1989; Sullivan and Satchwell 2000). All cell lines were grown at 37°C and 5% CO2 in DMEM/F12 containing 10% calf serum, antibiotics, and 2.5 mM l-glutamine on poly-l-lysine–coated coverslips. Cells on coverslips were chemically fused to form giant cells by a 1–2 min exposure to 50% (wt/vol) polyethylene glycol (1,500 g/mol) in 75 mM HEPES, pH 8.0 (Boehringer-Mannheim Biochemicals) (Kersting et al. 1989). Giant cells with a single plasma membrane were evident within hours after fusion. ERCs were recorded between 1 and 3 d after fusion. Fused HEK293 cells amplify the amount of expressed integral plasma membrane proteins and allowed high fidelity measurements of sodium channel gating currents (Sheets et al. 1996) and rhodopsin ERCs (Sullivan and Shukla 1999; Sullivan et al. 2000).
Cellular Pigment Regeneration
Cells on coverslips were washed in regeneration buffer and placed in a light-tight container in the darkroom at room temperature (22–25°C). Regeneration buffer was (mM): 140 NaCl, 5.4 KCl, 1.8 CaCl2, 1.0 MgCl2, 10 glucose, 10 HEPES-NaOH, pH 7.2, and 2% (wt/vol) (∼290 μM) fatty acid–free bovine serum albumin (FAF-BSA; Sigma Chemical Co.). Concentrated (mM) 11cRet or 9cRet stocks (in ethanol) were added in small volumes to this solution to a final concentration of 50 μM in preliminary experiments and 25 μM in the final protocol with 0.025% (vol/vol) α-d-tocopherol (vitamin E) added as an antioxidant. Vitamin E is found in high concentrations in photoreceptor cells and may serve as an antioxidant (Dilley and McConnell 1970). Vitamin E appeared to improve the longevity and quality of the recordings, likely because of its antioxidation effects on membranes and retinals. Primary regeneration was conducted for at least 30 min and coverslip fragments were then washed in bath recording buffer (without 11cRet or 9cRet) and transferred to the recording chamber, which contained E-1 buffer (see below). FAF-BSA molecules are a pool of chromophore receptors that efficiently solubilize hydrophobic retinals into aqueous solvents and deliver 11cRet loads rapidly and stoichiometrically to regenerate rhodopsin from opsin (Szuts and Harosi 1991; McDowell 1993). Optically clear solutions and effective solubilization result when ethanolic stocks of chromophore are introduced into 2% FAF-BSA buffers, whereas without FAF-BSA a scattering dispersion results that is indicative of chromophore partition into aggregates or micelles. Delivery of chromophore to cells is facilitated because a solubilized chromophore can partition more efficaciously into cellular membranes while chromophores bound to FAF-BSA may also be absorbed by pinocytosis. 11cRet was a gift of the National Eye Institute and Dr. Rosalie Crouch (Medical University of South Carolina, Charleston, SC) and 9cRet was obtained from Sigma Chemical Co.
Rhodopsin regenerated in fused giant cells was activated by an intense flash microbeam apparatus described in detail elsewhere (Sullivan 1998). In brief, light from a xenon flash tube is collimated, filtered, and condensed into a 1-mm-core fused silica fiber optics for transmission to the epifluorescent port of the microscope (Diaphot; Nikon Inc.). The objective lens is used to condense the fiber output into a microbeam spot parafocal with the specimen plane where the giant cell is situated. The spot-size diameter [full-width-half-maximum (FWHM)] in these experiments is 228 μm, which is about three times the size of the largest giant cell used. In routine flash photolysis, three-cavity bandpass filter elements (350, 430, 500, and 570 nm) were used that had 70-nm bandpass (FWHM) centered on peak transmission wavelength. To acquire action spectra data, 30 nm FWHM bandpass filters (centered at 400, 440, 480, 500, 520, 580, and 620 nm, and a 540-nm filter with a 10 nm FWHM bandwidth) were used. The throughputs of all filters, except those at 350 (70) and 400 (30) nm, do not overlap with the absorption spectra of free chromophore (peak ≈ 374 nm) such that isomerization (cis → trans or trans → cis) of any free chromophore is not expected during flashes used to elicit ERCs. Unless otherwise mentioned, flashes were delivered at the maximum capacity of the instrument. Intensities were 108–109 photons/μm2 across the near UV/visible band. Flash microbeam intensities were measured using a calibrated photodiode placed over the specimen plane of the microscope. To regulate flash intensity output (see Fig. 5), the voltage on the flash tube energy storage capacitor was adjusted. Flash duration was only ∼14 μs, insuring that the Meta-I ⇔ Meta-II equilibrium (milliseconds) generated at room temperature in these experiments was not perturbed by photoregeneration to other states (Sullivan 1998). Flash duration did overlap with lifetimes of bathorhodopsin, blue-shifted intermediate, and lumirhodopsin such that photoregeneration from these states is possible. Since we were largely concerned with the millisecond-order R2 phase of the ERC, any photoregeneration from early bleaching intermediates should not perturb the charge motions occurring during R2, which correlate with the time scale of the Meta-I ⇔ Meta-II transition. Shielding and fiber optic transmission prevent contamination of the patch-clamp electronics with flash-associated noise.
Photosensitivity (Pt) is the product of quantal efficiency (γ) and the wavelength-dependent absorbance cross section (αλ). The absorbance cross section of wild-type human rhodopsin is 1.53 × 10−8 μm2 (calculated from an extinction coefficient of 40,000 M−1 cm−1 at 493 nm (Wald and Brown 1958; Dartnall 1968; Knowles and Dartnall 1977) and γ is 0.67, leading to a Pt of 10−8 μm2 for normal human rhodopsin at peak extinction (493 nm). Pt can be used to estimate the fraction of rhodopsin molecules absorbing at least one photon per flash using the zero-order term of the Poisson equation [1 − Po = 1 − exp(−Pt · i)], where i is the flash intensity (photons/μm2) and Po is the fraction that absorbs no photons [Poisson Eq.: Pn = (Pt*i)n*exp(−Pt*i)/n!, where n is the number of absorptions per chromophore]. In this calculation, one adjusts αλ by the ratio of absorbance at the wavelength of interest to that at peak extinction. γ is assumed to be constant and independent of wavelength. For the 70-nm bandpass filters used in these experiments (centered at 350, 430, 500, and 570 nm), the fraction of molecules absorbing at least one photon were estimated to be 0.159, 0.716, 0.963, and 0.273, respectively. For the 30- and 10-nm bandpass filters used in these experiments (centered at 400, 440, 480, 500, 520, 540, 580, and 620 nm), the fraction of rhodopsin molecules absorbing at least one photon were estimated at 0.226, 0.626, 0.831, 0.80, 0.733, 0.44, 0.122, and 0.013, respectively. These calculations assume no orientational factors, no self-screening effects, and transparent cellular media. Thus, microbeam flash intensities were not expected to be experimentally limiting for flash photolytic stimulation of expressed rhodopsin pigments, except perhaps for the 620-nm stimulus. The maximum extent of rhodopsin bleaching (i.e., formation of Meta-II) after a single flash is 50% (Williams 1965, Williams 1974) because of second (or even-numbered) photon reabsorptions during the lifetimes of early states that have high quantal efficiency to photochemically regenerate the ground state (e.g., bathorhodopsin, lumirhodopsin). Flashes at 400, 580, and 620 nm were likely to elicit only single photon absorptions (>90%). Flashes at other wavelengths (440, 480, 500, 520, and 540 nm) were more likely to include even-numbered absorptions (relative fraction of total for even numbered absorptions 0.31, 0.415, 0.405, 0.367, and 0.22, respectively). The absolute flash intensities (108 photons/μm2) at the various center wavelengths (parentheses) used in action spectra experiments were as follows: 1.20 (400 nm), 2.29 (440 nm), 2.27 (480 nm), 1.96 (500 nm), 2.02 (520 nm), 1.49 (540 nm), 2.49 (580 nm), and 1.68 (620 nm). The relative ratios of absolute flash intensities relative to that at 500 nm were 0.61, 1.16, 1.15, 1.0, 1.03, 0.76, 1.27, and 0.85, respectively. To scale charge motions for action spectra (see Fig. 3), the reciprocal of these scale factors were used to multiplicatively scale the integrated charge motions.
Whole-Cell ERC Recording
Cells on coverslips were imaged using infrared light (high pass cutoff 830 nm) at 80–160× by an inverted microscope (Diaphot; Nikon Inc.) equipped with Nomarski differential interference contrast, a CCD camera, and a TV monitor. The microscope was housed in a Faraday cage in a dark room. Microelectrodes were fashioned from borosilicate glass using two stage pulls and coated with Black Sylgard (Dow Corning Corp.). Electrodes were routinely filled with one of two intracellular solutions (with or without 10 mM HEPES-CsOH) containing (mM) 70 tetramethylammonium (TMA)-OH, 70 Mes-H, 70 TMA-F, 10 EGTA-CsOH, 10 HEPES-CsOH, pH 6.5; these solutions yielded ERCs with no qualitative changes and are called I-1. The internal pH was chosen to be 6.5 to forward bias the Meta-I ⇔ Meta-II equilibrium strongly in favor of Meta-II (Parkes and Liebman 1984). In the E134Q experiments, intracellular buffers of otherwise identical composition to I-1 were formulated (I-2, I-3, I-4, I-5) to have intracellular pHs of 6.0, 7.0, 7.5, and 8.0, respectively. Bath solution contained (mM): 140 TMA-OH, 140 Mes-H, 2.0 CaCl2, 2.0 MgCl2, 5.0 HEPES-NaOH, pH 7.0 (E-1). pH was titrated by addition of HCl. Gigaohm seals formed readily with these solutions, electrode/patch/seal capacitance was compensated electronically, and further suction was used to enter whole-cell recording. Inward and outward rectifying conductances are present in HEK293 cells but were suppressed with permeant ion replacement solutions used in the pipette and bath. Without serum used in the growth medium, fused giant cells in E-1 recording buffer rounded up into approximately spherical shapes that remained well attached to the treated coverslips at their base.
The patch-clamp instrument was an Axopatch 1C with a CV-4 resistive feedback headstage and the later was used with a gain of 1 (0.5 Gigaohm feedback resistor; Axon Instruments). Since the ERC is a capacitative current, whole-cell capacitance (Cmem) and series resistance were not compensated, because this has the potential to alter the waveform. Membrane holding potential was clamped at 0 mV unless otherwise noted in the legends. Whole-cell capacity current was measured by a +20 mV/4 ms test pulse from a holding potential of −80 mV and Cmem was computed by integrating the capacitative current waveform to obtain charge (Q) (Q/V = Cmem) after substraction of ohmic current. Cell surface area was computed from the measured Cmem (1 μF/cm2). Ramp voltage clamps were delivered to test for a high resistance membrane and low level leakage. Cells with large leakage were discarded. Whole-cell currents were recorded at 5 kHz bandwidth using an eight-pole Bessel filter. Flashes were controlled and ERC and flash stimulation data acquired using pCLAMP 5.51 (CLAMPEX) and digitized (200 μs/point) by a Labmaster (100 kHz) interface board (Scientific Solutions Inc.). This acquisition rate was selected to provide the best possible representation of the R1 signal, which is still undersampled, while critically allowing the full time course of the R2 component to be acquired out to 100 ms. All ERC data was processed and analyzed using the Origin4.1 package (MicroCal Software, Inc.). Nonlinear least squares fitting was conducted using a Levinberg-Marquardt algorithm.
Giant Cells Generate Large ERCs that Recover Spontaneously with Dark Adaptation
Polyethylene-glycol–fused giant HEK293S cells are used to amplify the amount of regenerable plasma membrane opsin (WT or mutant) in single large cells to improve the signal-to-noise ratio (SNR) during ERC data acquisition. These are prepared from single cells that are stably transformed to constitutively express WT or mutant human opsins in the range of 1–10 × 106 molecules (Sullivan and Satchwell 2000). Giant cells share with unfused cells a uniform distribution of immunocytochemically reactive opsin in their plasma membranes (not shown). The increased plasma membrane surface area of fused cells contains larger numbers of opsin molecules in linear proportion to the number of cells that participated in the fusion events (Sullivan and Shukla 1999). R2 charge motion is linearly dependent on the number of light-activated rhodopsin molecules in the plasma membranes (Cone and Pak 1971). Therefore, the fused cell system was a rational choice for recording larger ERC signals at greater SNR. The largest fused cells had total WT human ERC R2 charge (Q) about an order of magnitude greater (2.0 vs. 0.29 pC) than amphibian rod photoreceptors (Makino et al. 1991). Fused cells increased total ERC charge, improved SNR, and allowed spontaneous regeneration of pigment after bleaching by simple dark adaptation without further addition of chromophore beyond the primary regeneration.
Fig. 1 A shows a giant cell ERC obtained on the first flash series (500 nm) after a 30-min regeneration. Flash stimuli were given at 500 nm to extinguish the ERC signal into background whole-cell noise. Both the submillisecond negative R1 current and the millisecond-order larger, and positive R2 current were routinely observed and essentially identical to those recorded in photoreceptors (see Hestrin and Korenbrot 1990; Makino et al. 1991; Sullivan and Shukla 1999; Sullivan et al. 2000). While not detectable in single unfused cells, in giant cells the R1 phase of the ERC was recordable, apparently due to increased total rhodopsin and improved SNR. After subtraction of baseline current, integration of each ERC lead to the charge motion (Qi) in femtocoulombs attributable to each phase. Successive flashes progressively extinguished both phases of the ERC until no responses were observed above background noise. This was consistent with bleaching of plasma membrane rhodopsin due to photolysis. The observation that giant cells spontaneously recovered ERC signals after a complete bleach by simple dark adaptation for 10–15 min without added chromophore was quite surprising. Unless otherwise stated, cells were not exposed to additional 11cRet once the coverslip was removed from regeneration buffer and placed in E-1 buffer in the recording chamber. Subsequent flash photolysis after post-bleach dark adaptation lead to robust ERC signals that were again extinguished by additional flashes at 500 nm (Fig. 1 B). Another period of dark adaptation promoted spontaneous recovery of the ERC that was again extinguished by successive 570-nm flashes (Fig. 1 C). The R2 but not R1 signals were recorded after spontaneous regeneration of visual pigment that occurs during 10-min dark adaptation (see discussion).
Fig. 1 D shows the kinetics of the spontaneous recovery of total ERC charge upon dark adaptation for two similarly sized fused cells (67.5 and 65.6 μm diameter). Cells were bleached and dark adaptation was allowed to occur for variable time periods before additional 500-nm flashes were given in rapid succession to extinguish the ERC signal. The total charge (Q∞) present at each time point of dark adaptation was obtained by summing all Qi values to yield Q∞ at that time point. Q∞ values were normalized to the maximum charge determined for each cell so that regeneration data could be compared. The data were fit by a single exponential accumulation curve. The initial regeneration rate is fast and the process begins to approach steady state by 10 min of dark adaptation. The half time of ERC regeneration was 3.2 ± 1.1 min. If primary regeneration in 2% BSA fully equilibrates WT-HEK293S cells with 11cRet at a concentration (25 μM) in considerable excess over the expected molarity of opsin in fused giant cells (≈2.4 μM), then spontaneous regeneration can be described as a process with pseudo–first order kinetics from a single compartment containing chromophore. We routinely dark adapted cells for either 10 or 15 min between successive extinctions to allow visual pigment and ERC signal recovery to the stationary plateau or, at a minimum, to permit a criterion level of regeneration for any comparison of charge motions under different conditions tested on a single cell.
The Source of Chromophore during Dark Adaptation Is Intracellular
The source of chromophore that promoted spontaneous ERC recovery was investigated. Single fused giant cells regenerated with 11cRet were subjected to primary extinction by successive flashes at 500 nm in normal bath solution (E-1 without chromophore). Dark adaptation promoted spontaneous pigment regeneration and allowed for a secondary extinction of the ERC. Immediately after the second extinction and throughout the next dark adaptation, the chamber volume was replaced with E-1 containing 10 mM hydroxylamine (NH2OH, pH 7.0) and an additional 5 min of dark adaptation preceded the next series of successive 500-nm flashes. ERCs were elicited and had Qi values comparable with those seen during the primary and secondary extinctions without NH2OH. Qi extinctions before and after NH2OH are shown (Fig. 2 A). When NH2OH was present in the bath many more flashes (in this cell 23 flashes) were required to completely extinguish the ERC R2 charge in comparison to three to five flashes during the primary and secondary extinctions. This suggested that 10 mM NH2OH decreased photosensitivity and this effect will be examined fully in a subsequent study. Two dark adaptations (10 min each) in 10 mM NH2OH were followed by some but not full ERC signal recovery, which was then subsequently extinguished by additional flashes. ERC signals continued to be recordable on the time scale of tens of minutes in the presence of constant 10 mM extracellular NH2OH, a concentration far greater than the initial loading of chromophore (25 μM). Fig. 2 B shows the exhaustion course of Qi vs. flash number for another giant cell before and after introduction of 10 mM NH2OH into the bath solution. In NH2OH total R2 charge decreased with successive extinctions and dark regenerations until no significant signals remained. Bath solution was then replaced with fresh E-1 containing 25 μM 11cRet in 2% FAF-BSA (without NH2OH). Strong ERC R2 signals were regenerated comparable in total charge with that seen during the primary extinction. These experiments demonstrate strong evidence that the source of 11cRet during spontaneous dark regeneration is internal to the cell that is recorded.
Additional observations support an intracellular origin for 11cRet during dark adaptation. First, the cell repeats dark-adaptive ERC regeneration until the signals expire, and then they do not again regenerate unless 11cRet is reapplied to the cell. This rundown is evident in the extinctions in NH2OH in Fig. 2 and suggests that cells exhaust their store of 11cRet. Second, in some cells the plasma membrane appeared wrinkled or the cytoplasm appeared optically smooth after expiration of ERC signals. This suggested a change in the structure of internal membranes associated with regeneration ability. Finally, the greatest number of “visual cycles” in this study was eight and was encountered in the largest fused giant cells (≈80 μm diameter) and, although not yet systematically investigated, the number of regenerations appeared to scale in proportion to cell size. Given the hydrophobic nature of 11cRet, it is likely that it is stored by partitioning into internal cellular membranes, and then repartitions back to the plasma membrane during dark adaptation to regenerate visual pigment (see discussion).
The ERC Action Spectrum Is Consistent with the WT Human Rhodopsin Photopigment
A previous study of ERC spectral sensitivity used 70-nm bandpass filters and demonstrated a broad action spectrum consistent with the ground state of human rhodopsin pigment (Sullivan and Shukla 1999). However, the possibility remained that other intermediates such as Meta-III465, pseudophotoproducts (Meta470) or even isorhodopsin could contribute to the action spectrum. To investigate this issue further, the spectral sensitivity of ERC R2 charge motion was measured with stimuli generated with relatively narrow 30-nm band-pass (FWHM) interference filters. In these experiments, the ERC R2 charge from only the first flash after each 10-min recovery period was obtained between different filter settings. Rather than extinguish the signal, we did a criterion dark adaptation for 10 min between each single flash at a given wavelength. Fig. 3 shows first-flash ERC signals at several different wavelengths from a single giant cell. All ERC signals in the figure were multiplicatively scaled by the ratio of absolute flash intensities measured relative to 500 nm so that the spectral sensitivity can be more readily appreciated by observation. The ERC currents vary in amplitude depending on the wavelength of stimulation. To generate the action spectrum, the ERC charge (Qi) was obtained at each wavelength and the charges were corrected for differences in photon density delivered with different filters so that action spectra at equal photon exposure would be generated. To allow comparison of action spectra from different fused cells containing different amounts of rhodopsin and ERC signal size, spectral sensitivity data were normalized to the maximum integrated R2 charge at whatever wavelength it occurred. For all three cells, the largest R2 charge occurred with 500-nm stimulation. In Fig. 3, the mean ERC spectral sensitivity data (±SEM, n = 3) are plotted versus wavelength of peak stimulation. By stimulating with only a single flash at each wavelength and allowing sufficient time for regeneration, less stress is placed on the retinal load in the cell and the relative percent of bleaching is considerably less than when the ERC was extinguished by successive flashes at a single wavelength. This should allow greater sensitivity to detect differences because the fraction of pigment that must regenerate between each single flash bleach is smaller and data can be obtained from a uniformly regenerated population of pigment molecules with low variability.
Instead of fitting spectral sensitivity data with a Lorenztian/Gaussian peak function (e.g., Voigt) (Sullivan and Shukla 1999), the absorbance spectrum of WT human rhodopsin was normalized with respect to the maximum visible band at 493 nm (α band) and overlaid with the ERC action spectrum. WT rhodopsin was immunoaffinity purified from the same WT-expressing cell line used in these ERC experiments (Sullivan and Satchwell 2000). The absorbance spectrum of WT human rhodopsin (α band peak at 493 nm) provides an excellent fit to the ERC action spectrum obtained from cells expressing WT human opsin that was regenerated with 11cRet. The ERC action spectrum peaked around 493 nm, consistent with the major absorption band of ground state human rhodopsin regenerated with 11cRet. Moreover, the bandwidth of the α peak of the human rhodopsin absorbance spectrum also fits the ERC data well. It is important to mention that all of the action spectra data were obtained from cells after they had undergone a primary bleach; that is, under conditions where long-lived bleaching intermediates or photoregenerated pigments might have been present. Normal human rhodopsin has an absorbance spectrum that is slightly blue shifted (≈493 nm) with respect to the bovine pigment (498 nm) (Wald and Brown 1958), and this is maintained on expressing human rhodopsin in HEK293S cells. Assuming that absorption maxima of spectral states of human rhodopsin would scale relative to the bovine pigment (isorhodopsin487, Meta-III465, and Meta470), one would expect peaks at 482, 460, and 465 nm, respectively, in the human pigment. Therefore, if there are spectral states of bleaching intermediates or isorhodopsin present during ERC data acquisition, then they must represent a very small unresolved component of the charge motion. From this we conclude that the pigment underlying the expression ERC is WT human rhodopsin, and that during dark adaptation 11cRet from within the cell reacts with bleached opsin to form a PSB+-H and a normal ground state pigment.
ERC Extinction Measures Bleaching of Rod Rhodopsin
Successive flashes at a single wavelength and intensity promoted progressive loss of ERC R2 charge until no further signal was obtained above background current noise. Since the interstimulus intervals between successive flashes were only ∼10 s, pigment regeneration was minimal between stimuli, and regeneration should not contribute to the extinction progression. The spectral sensitivity of the ERC governs not only the efficacy of successive bleaches, but also single bleaches. Fig. 4 A shows the Qi extinction of R2 in response to successive flashes at 570 and then 500 nm in a single giant cell. Cumulative flash intensity delivered is used as the dependent variable. After ERC charge was effectively extinguished into noise by 570-nm flashes, 500-nm flashes of greater effective intensity were immediately delivered. The additional extinguishable ERC charge found with 500-nm flashes indicated residual ground state rhodopsin in the cell after the 570-nm flashes. This resulted because 570-nm flashes are not as effective at eliciting ERC currents as are flashes at 500 nm given the relative absorbance of WT human rhodopsin at 570 vs. 493 nm (peak absorbance) (OD570/OD493 = 0.436) (Wald and Brown 1958). Similar findings occur when bleaching at other wavelengths is followed by flash photolysis at 500 nm. The relative probability of activating rhodopsin, taken as the ratio of absorbance cross sections, is only ∼12% at 570 nm relative to 500 nm (α570/α493 = 0.115), assuming equal photon density at the two wavelengths. At the maximal flash strengths used, the fraction of rhodopsin molecules absorbing at least one photon at 500 vs. 570 nm was estimated to be ∼0.96 and 0.27, respectively. Thus, the flash system does not deliver sufficient photons at 570 nm to compensate for the lower probability of activation. This illustrates that detection of rhodopsin charge motions depends on the unitary charge motion, which should be the same at any wavelength (see univariance below), and the number of activated rhodopsin molecules that mobilize charge and sum into an ensemble ERC current. Even at peak wavelength (≈500 nm), a given flash intensity may not be sufficient to generate ERC currents above noise because the ensemble ERC current lies within the noise band.
Single exponential exhaustion curves are fit for both data sets at the two respective wavelengths. ERC extinction data for 570-, 500-, or 430-nm flashes from many experiments were always fit by single exponential curves. This is consistent with the ERC charge motion being proportional to the amount of rhodopsin that remains unactivated before each flash is given. Extinction data were fit by the following model using a nonlinear least squares method:
where Qi is the charge motion resulting from a single flash, Q∞ is the total charge before any flashes are given, I is the cumulative flash intensity, and Pt is the photosensitivity. The bleaching process follows an exponential extinction that was further tested by a natural logarithmic transform of Qi values and linear fitting (Fig. 4 B). was used to determine Pt for several giant cells subjected to flash photolytic exhaustion under conditions of different flash stimulation wavelength. The charge extinction data for the initial (primary) and three subsequent (secondary) bleaches at 500 nm and single bleaches at 430 and 570 nm in a single large giant cell (≈80 μm) are shown in Fig. 4 C. The charge extinction data sets for each bleach were normalized to the maximum charge on the first flash, and then was fit to each data set and overlaid. The fit by the single exponential model implies that each flash promoted activation and extinction (bleaching) of a fraction of remaining ground state pigment. In some flash stimulation series, there was a residual content of charge that was slow to extinguish (10–20%). Pt for the primary bleach at 500 nm was estimated to be ∼3.3 × 10−9 μm2, but this is less reliable as there were only three points to fit. The Pt values obtained from fitted curves at 500 nm for the three secondary bleaches were 3.28 × 10−9, 3.04 × 10−9, and 2.24 × 10−9 μm2. Pt was stable over successive secondary bleaches and similar to that of the primary bleach. This is evidence in addition to the action spectrum that the ground state of rhodopsin is regenerated with 11cRet during dark adaptation between successive flash cycles and there was no accumulation of other intermediates with significantly different photosensitivities. The mean (±SEM) for Pt values from several cells are shown in Fig. 4 D at 500(70), 430(70), 570(70), and 500(30) nm, where the filter bandwidth is indicated parenthetically. Both parametric and nonparametric (Kruskal-Wallis) analysis of variance tests were used to evaluate whether the means of the five conditions were different. No statistically significant difference was found between Pt estimates of human rhodopsin during the initial and secondary extinctions after spontaneous regeneration. However, the trend toward a lower Pt at 430 nm is consistent with the ratio of absorbance of rhodopsin at 430 vs. 500 nm (α430/α493 = 0.423) (Schneider et al. 1939).
The Pt of WT human rhodopsin is estimated to be ∼1.0 × 10−8 μm2 (see materials and methods). The mean value of Pt [500(70) nm] determined from 11 extinctions in 6 cells, all after spontaneous regeneration was 2.6 ± 0.4 × 10−9 μm2 and the maximum value measured was 5.0 × 10−9μm2. Pt values measured using the extinction of R2 charge are consistent with but lower than that expected of a rhodopsin chromophore. This is likely due to the suppressive effect on Pt of photoregeneration resulting from multiple photon absorptions per rhodopsin molecule at the flash intensities used. For example, with each 500(70)-nm flash, the estimates on even numbered absorptions are ∼50%, which would underestimate Pt by a similar amount (see discussion). In recent experiments, stimulus intensity (at 500 nm) was reduced (by 85%) to decrease the probability of multiple hits per molecule, allowing Pt estimation to a mean value of 8.5 × 10−9 μm2, which approximates that expected from the extinction coefficient (Brueggemann and Sullivan, manuscript in preparation).
Tests of Linearity and the Univariance Principle with ERC Measurements
As first demonstrated by Makino et al. 1991, the ERC of photoreceptors has invariant kinetics regardless of the wavelength or intensity of the flash, although the amplitude of the signal scales in linear proportion to the amount of rhodopsin activated, provided that the flash strength is below saturation. Any rhodopsin molecule absorbing a single or odd number of photons will have a finite probability (constant γ = 0.67) of successful activation and will contribute to the kinetics of ERC charge flow. At fixed wavelength, variation in stimulus intensity is expected to affect the probability of absorption if each activated rhodopsin molecule makes an independent and additive (linear) contribution to the ERC, and photoreversal to other states by second photon absorptions is minimal. Fig. 5 shows ERCs resulting from 500-nm stimulation at two different flash strengths. When these responses are normalized and overlaid for comparison, the kinetics of the R2 relaxation at different intensities are not distinguishable. The ERC response versus flash intensity was measured. The Qi response of the first flash in an extinction series at a constant intensity is plotted versus intensity. A linear fit of the charge motion versus absolute intensity was found. This indicated that the flash intensities used were below saturation for the cellular expression system. This result is consistent with known properties of the ERC/ERP. Each activated rhodopsin molecule undergoes conformational changes to contribute a quantum of charge motion to the overall R2 signal (Cone and Pak 1971; Hodgkin and O'Bryan 1977).
According to the univariance principle, the energy of the photon (wavelength) should not affect the activation kinetics of independent rhodopsin molecules. Photon energy only affects the probability of absorption because the molecular cross section is a function of wavelength. Fig. 6 shows ERCs acquired from a giant cell at 430, 500, and 570 nm. ERC waveforms were normalized to the smoothed peak of the R2 current for comparison. A double exponential curve was generated to fit the relaxation kinetics of the R2 signal from the 500-nm stimulus. This template was then overlaid with the R2 signals from the 430- and 570-nm responses. The 500-nm template fits the R2 relaxations of the 430- and 570-nm responses rather well, even though the SNR was lower with 570-nm stimuli because the absolute response was smaller. Although the amplitudes of the ERCs and total charge motion of the R2 signal vary with wavelength, the kinetics of R2 relaxation are similar. Similarly, the large 500-nm R2 response in Fig. 3 was fit with a double exponential function, and this template was overlaid with the large 440-, 480-, and 520-nm ERC responses and provided a good fit. The 400- and 540-nm ERC responses were of lower SNR and were not fit well by the template. ERC data in Fig. 3 was collected with 30 nm FWHM stimuli. In the rhodopsin expression system, photon energy does not affect the kinetics of the state transitions in rhodopsin, which is consistent with the univariance principle.
Ground State Rhodopsin ERC Can Be Photoregenerated from Metarhodopsin-II
Previous studies have shown that the ground state of rhodopsin can be photoregenerated from Meta-II380 by near UV flashes delivered concurrent with its lifetime (Williams 1964, Williams 1968; Williams and Breil 1968; Cafiso and Hubbell 1980; Drachev et al. 1981; Arnis and Hofmann 1995). Photolysis of Meta-II generated inverted ERP signals (Cone and Cobbs 1969; Ebrey 1968; Cafiso and Hubbell 1980; Drachev et al. 1981). To investigate whether photoregeneration could be assayed with ERC measurements, 500-nm flashes were used to extinguish the ERC into background cell noise, whereupon 350-nm flashes (70 nm FWHM) overlapping the Meta-II absorption (350 nm, 8.02 × 107 photons/μm2) were delivered immediately and in rapid succession. The number of near UV flashes (1–5, 5, 10, and 20) delivered was varied to affect an increasing dose of photoregeneration stimulus to the Meta-II remaining in the cell. Immediately after the near UV flashes were delivered, additional flashes at 500 nm were used to measure the level of regeneration as the ERC R2 charge. Under the conditions of these experiments, the expected lifetime of Meta-II at room temperature in a membrane environment is on the order of several minutes (van Breugel et al., 1979; Parkes and Liebman 1984). Fig. 7 A shows ERC responses from an experiment where 5, 10, or 20 350-nm flashes were given to promote photoregeneration of ground state rhodopsin from Meta-II. The top trace in each case is the ERC response to the first 500-nm flash in the series used to initially extinguish the signal. The middle shows the response to the last 500-nm flash indicating the level of extinction into the whole-cell current noise. The bottom shows the first flash at 500 nm after 5 (left), 10 (middle), or 20 (right) UV flashes. As the number of UV flashes is increased, the size of the recovered 500-nm ERC signal becomes similar to that found before the UV flashes. The ERC R2 current waveform in response to 500-nm flashes after the UV flashes is comparable in size and kinetics to the 500-nm signal that preceded bleaching whether 5, 10, or 20 UV stimuli were delivered. This data suggested that the ground state of the pigment (11cRet) is regenerated during the photoconversion process. In this cell, the amount of ERC R2 charge increased over the time period of successive extinctions. The underlying mechanism is not yet clear (see discussion). Fig. 7 B shows the percent of ERC charge recovered after 5 (65%), 10 (74%), or 20 (87%) UV flashes, measured against the UV flash dose. The period of time needed to deliver 5, 10, or 20 UV flashes was 27, 67, or 182 s, respectively, in this experiment. Since the kinetics of spontaneous regeneration has a time constant of ∼3 min (Fig. 1 D), ERC responses after 10 or 20 UV flashes could be significantly affected by chemical regeneration with chromophore rather than photoregeneration. Therefore, the charge recovery in response to one to five UV flashes was examined (Fig. 7 B). By the third UV flash, the process of regeneration had stabilized (∼55%) at the UV intensity used and no significant additional charge recovery was found when two additional UV flashes were given. The amount of charge regenerated by three to five UV flashes was similar to that which regenerated after five flashes of the same intensity in the experiment described above. Only 11 s were required to deliver the three UV flashes, making it highly unlikely that any significant chemical regeneration had contributed to the 55% charge recovery. Thus, UV stimuli promoted a large fractional recovery of R2 charge in a time frame that is much more rapid than the kinetics of spontaneous ERC regeneration by dark adaptation. These results strongly suggest that the ERC can be photoregenerated by additional photons overlapping the Meta-II bandwidth. The quantal efficiency appears favorable to studying the mechanism in greater detail.
In most cells receiving UV stimuli, no apparent ERC charge motion occurred above the noise level of the cell. However, in a few large giant cells, apparent UV flash–induced ERC signals were identified even without any signal smoothing to suppress noise. An example is shown in Fig. 7 C, where a response to a UV (350-nm) flash is generated immediately after extinction of ERC signals with 500-nm stimuli. The UV-induced ERC signal is small and has a negative (inverted) R2-like response. An unconstrained third order polynomial fit the UV R2 signal and also demonstrated the inverted R2 signals. An ERC generated with a 500-nm flash after dark adaptation in the same cell is shown to demonstrate the magnitude of the normal R2 charge motion. The time to peak of the positive 500- and negative 350-nm–induced R2 signals was 5.9 and 12.2 ms, respectively. The ratio of the inverted to noninverted R2 charge was 0.175 (0.099 in another smaller cell).
9-cis-Retinal Regeneration Results in ERC Signals Consistent with Isorhodopsin
Cells were regenerated with 9cRet to test the feasibility of ERC investigation of rhodopsin activation in analogue visual pigments. Analogue visual pigments are usually formed from WT opsin and a synthetic retinal known to have unique properties (e.g., to block Meta-II formation), but could also be formed from synthetic retinals and site-specific opsin mutants. The naturally occurring 9cRet analogue forms isorhodopsin, a stable ground state pigment that is generated in a photostationary state with rhodopsin and bathorhodopsin (Birge et al. 1988). Once isorhodopsin is photoactivated to bathorhodopsin, the same sequence of bleaching intermediates occur as compared with normal rhodopsin activation. ERCs were recorded in three of four fused cells regenerated in 9cRet and signals were uniformly small. This may in part be related to the cell sizes used [Cmem 85.7, 11.6 (probably a single cell), and 48.6 pF]. Fig. 8 shows primary (left) and secondary (middle) extinctions of ERCs with 500-nm flashes for a fused cell regenerated with 9cRet. ERC signals in fused WT-HEK293 cells regenerated with 9cRet were smaller than those regenerated with 11cRet in the population of cells studied. However, the ERC R2 waveform was similar. Pt was determined by flash series extinction at 500 nm (70 nm) for this cell and found to be 1.03 × 10−9 μm2, which is similar to the Pt we calculate for isorhodopsin at peak extinction at 483 nm (5.55 × 10−9 μm2). This value is arrived at by first calculating the molecular cross section (αλ) from the extinction coefficient for isorhodopsin at 483 nm (44,000 M−1 · cm−1) (αλ = 3.82 × 10−21 * ∈λ) (αλ = 1.68 × 10−8 μm2) and multiplying αλ by the quantal efficiency of isorhodopsin of 0.33 (Crouch et al. 1975). At the flash intensities used in these experiments (4.08 × 108 photons/μm2), ∼90% of isorhodopsin pigment molecules should absorb at least one photon, with ∼55% of these being odd-numbered isomerizations that would proceed forward to bleaching and ∼45% being even-numbered isomerizations that would result in photoconversion to ground state species. Thus, Pt could be suppressed by photoregeneration at the flash intensities used.
Kinetic Comparison of Mutant and WT Visual Pigments with the ERC
One way to exploit the sensitivity of the ERC technique is to investigate charge motions in mutant rhodopsins. We generated stable, high-level producing (∼106 opsins/single cell), HEK293S cell lines of several human rod opsin mutants altered at single amino acids that could support proton exchange processes in the membrane region (Sullivan and Satchwell 2000). Here we demonstrate the nature of the ERC signals obtained from D83N and E134Q opsin pigments regenerated with 11cRet. Our previous work suggested that the kinetic relaxation of the R2 phase of WT human rhodopsin in WT-HEK293S giant cells was kinetically complex (Sullivan and Shukla 1999). In that study, we found that double exponentials were typically required to reliably fit large R2 relaxations >100 ms after the flash. In this study, we further our kinetic analysis of WT R2 relaxation and compare R2 relaxations from the two mutant pigments to WT.
Fig. 9 shows ERC signals in response to the first 500-nm flash after primary regeneration and secondary recovery in fused giant cells containing WT, D83N, or E134Q human rhodopsins. The WT pigment generated strong R1 signals during the primary bleach. R1 signals are rarely seen during the secondary or subsequent extinctions indicating that, if present, the size is below the limits of detection at the flash intensities used. Large (>40 pA) WT ERC signals typically require two exponentials to fit the time course of the R2 relaxation over the first 100 ms. Residuals are shown beneath the ERC waveforms. R1 signals were not observed in D83N rhodopsin during primary extinction in cells with R2 charges of the same order as seen in fused WT cells that had R1 signals. Like WT, the D83N R2 relaxation typically requires two exponentials to reliably fit its relaxation. However, D83N signals appear to lack the “stretched” exponential appearance seen in many large WT signals during the 100 ms after the flash. The E134Q ERC signal was distinctly different from the WT signal. R1 signals were not observed during primary extinctions. Moreover, the outward R2 signals in E134Q rhodopsin-expressing cells were markedly simplified in comparison with WT or D83N ERCs. The relaxation was brief and required only a single exponential to fit its decay. Like WT ERCs, D83N and E134Q ERC signals extinguished with successive flashes and had spectral sensitivity consistent with pigments absorbing ∼500 nm (Sakmar et al. 1989; Nathans 1990).
To begin to characterize R2 relaxation, single or double exponential functions were fit to a large number of WT, D83N, and E134Q R2 signals from many cells of similar size range. Time constants associated with R2 relaxation, but not R1, are essentially independent of Cmem (Sullivan and Shukla 1999). Time constants were extracted from the first and second exponential terms (τa, τb). Since the identification of the respective components is dependent upon their weighting, it is possible that the τb constant could be assigned to the τa data set if the weighting of the τa component is small or unreliable (e.g., lower SNR). Therefore, all the time constants obtained for R2 relaxation were placed into a total ensemble, and histograms were generated from these populations for the WT, D83N, and E134Q datasets (Fig. 10, A–C). The WT pigment demonstrated a broad skewed distribution suggestive of density around three time constant ranges, whereas the E134Q distribution was simple and symmetrical and the D83N distribution was intermediate. To quantitatively characterize the ensemble of time constants, Gaussian distribution functions were fit to each histogram. The WT histogram was reliably fit by a sum of three gaussian distributions, centered at 4.1, 12.5, and 26.4 ms (Table). There were also residuals with time constants longer than the three fitted distributions. This analysis demonstrates the kinetic complexity of the WT R2 relaxation and supported the conclusion of a minimum of three charge states with distinct lifetimes (see discussion). The D83N histogram was reliably fit to the sum of two Gaussian functions centered at 3.7 and 10.7 ms, still leaving some residuals. These time constants are comparable with the two fastest time constants measured in the first 100 ms of WT R2 relaxations, whereas the third τ found in WT appears to be missing (Table). The E134Q histogram was distinctly different from both WT and D83N, requiring only a single Gaussian function with a peak centered at 4.4 ms. This single peak overlaps with the fastest time constant seen in the WT and D83N pigments (Table). This work establishes an initial approach to parameterize the R2 relaxation. The intent is to use this approach as a means to quantify differences between WT and mutant ERC kinetics during the biochemically important time period of the R2 signal.
In earlier work, we established that, after regeneration with 11cRet, ERC signals could be recorded from both single and fused HEK293S cells expressing high levels of WT human opsin to their plasma membranes (Sullivan 1998; Sullivan and Shukla 1999). The ERC originated from plasma membrane opsin that was regenerated with chromophore and the amount of charge motion was proportional to the size of the fused giant cell, consistent with the quantity of opsin expressed in the plasma membrane. The fused cell technique offered the advantage of a larger SNR, permitting more complex experiments and observations of greater complexity in the ERC signal (e.g., the R1 phase) when compared with the single cell system. In this early work, an action spectrum for the R2 signal was consistent with a normal rhodopsin pigment, but the bandwidth of the filters used (70 nm) made it difficult to conclusively rule out the contribution to the ERC response of other spectral pigment states with absorption ∼500 nm. Critically, it was demonstrated that the application of time-resolved ERC recording to expressed rhodopsin improved measurement sensitivity between 107- and 108-fold when compared with other contemporary methods used to study expressed visual pigments.
In the current experiments, we applied the expression ERC tool to investigate physical properties of WT rhodopsin expressed in HEK293S cells. ERC measurements are consistent with normal properties of WT human rhodopsin such as the absorbance spectrum, photosensitivity, univariance, and photoreversibility from Meta-II. The utility of the expression ERC tool was expanded on in the experiments reported here by demonstrating successful measurements of ERCs in cells regenerated with the analogue chromophore 9cRet. Moreover, ERCs of two mutant pigments D83N and E134Q demonstrate qualitative and quantitative differences with respect to WT ERCs. These results strongly suggest that the expression ERC approach could be productively expanded to investigate a broad range of rhodopsin activation properties of analogue visual pigments and mutant visual pigments, thus embracing a structure–function approach applied to both the chromophore and remote environments of rhodopsin.
On Recording ERCs from Expressed and Regenerated Visual Pigments in Giant HEK293S Cells
When fused giant cells are regenerated (primary) with 11cRet in the dark, ERC signals are obtained on high intensity flash stimulation (108 photons/μm2) with essentially identical waveform to the ERCs of vertebrate amphibian photoreceptors (Hestrin and Korenbrot 1990; Makino et al. 1991; Sullivan and Shukla 1999; Sullivan et al. 2000). The total charge motion in fused giant cells typically exceeds that found in amphibian rod photoreceptors. Both the R1 and R2 signals of the WT ERC were observed. The R1 signal is not well resolved kinetically and is not likely to be since R1 is thought to report very early charge separation events in the chromophore environment that are orders of magnitude outside the bandwidth of whole-cell patch clamp recording. Thus, only the amplitude of R1 charge could be measured, but not its rise time. The R2 signal is completely resolved in these experiments given that the rise of R2 to its peak is observable in most cells, as is the R1 to R2 transition during primary bleaching. The ERC time scale that was well resolved in these experiments extends out from ∼400 μs after the flash. This is sufficient to cover the entire time course of the Meta-I to Meta-II equilibrium at room temperature. At higher bandwidth and lower temperatures, the formation of Meta-I from lumirhodopsin could probably be examined. The temporal limits of investigation of rhodopsin electrical state transitions with whole-cell recording will be the rise time of the flash stimulus, the speed and sensitivity of the amplifier and digitization hardware, cellular capacitance and access resistance, and the decreased SNR expected due to greater noise at wider bandwidths.
To regenerate rhodopsin from opsin apoprotein constitutively expressed in these cells, a FAF-BSA technique previously used to regenerate rhodopsin from opsin in intact rod outer segments (McDowell 1993) was adapted to the HEK293S fused cell expression system. The regeneration kinetics (≈3 min) are similar to rates in intact photoreceptors (τ1/2 ≈ 1 min, complete by 10 min; McDowell 1993). The combination of fusing single cells to form giant cells and the use of FAF-BSA regeneration permitted much larger ERC signals and the spontaneous recovery of the ERC signal during post-bleach dark adaptation.
Experiments reported here indicated that the source of 11cRet during spontaneous dark regeneration is internal to the cell that is recorded. Several bleach and recovery cycles were required to completely exhaust the source of chromophore in the presence of extracellular NH2OH. If regenerating 11cRet chromophore originated from the outer surface of the plasma membrane or external to the cell (e.g., released by other cells on the coverslip or from recording chamber surfaces), then NH2OH should rapidly prevent ERC recovery by converting retinaldehydes to oximes, which do not form visual pigments. The only other potential source of chromophore is internal to the cell under recording, shielded from reaction with extracellular 10 mM NH2OH. Once NH2OH was applied to the bath, plasma membrane visual pigment was partially bleached and spontaneously regenerated over several cycles before the source of intracellular chromophore became exhausted and ERCs did not recover. Prompt recovery followed washout of NH2OH and perfusion of 25 μM 11cRet plus FAF-BSA in recording buffer. These experiments suggest that 11cRet can enter the retinal binding pocket when presented from either side of the membrane. The data also provides indirect evidence that NH2OH is impeded from permeating the plasma membrane of the HEK293 cells in E-1/I-1 solutions because if it had, then the spontaneous regeneration should have been promptly quenched. That ERCs are recordable when NH2OH has access to the extracellular surface of membrane-oriented rhodopsin indicates that visual pigment regenerated in the dark is not reacting with this agent, consistent with the resistance of the ground state of rod rhodopsin to NH2OH. NH2OH does not affect the rate of formation of Meta-II (Johnson 1970). Persistence of ERC R2 signals in NH2OH also suggests that buildup of thermal bleaching intermediates such as Meta-II, Meta-III, or M470 cannot be contributing to the charge motions of the ERC signals during secondary extinctions because the all-trans-retinal in the ligand pocket of these rhodopsin conformations would react rapidly with NH2OH to form opsin and the oxime adduct and extinguish the signal entirely.
One of the major advantages of the FAF-BSA/α-d-tocopherol regeneration technique in fused giant cells is the spontaneous recovery of visual pigment and ERC signals without the need for reinstallation of 11cRet into the recording chamber. A similar process has been reported for retina isolated away from pigment epithelium and apparently reflects limited photoreceptor stores of 11cRet (Cone and Brown 1969; Knowles and Dartnall 1977). Makino et al. 1991 also found spontaneous regeneration of ERC signals and used extracellular NH2OH (10 mM) to quench the process, indicating that a major fraction of chromophore originated from outside the cell (e.g., partitioned onto chamber surfaces) during active delivery of retinals to the recording chamber. In the experiments reported here, no 11cRet or 9cRet was unintentionally added to the chamber in our experiments other than perhaps a small amount on the small coverslip fragment holding the giant cells. The FAF-BSA/vitamin E technique appears to be an efficient way to load chromophore into HEK293 cells that express visual pigments. Similar success was found by McDowell 1993 for intact rod photoreceptors using FAF-BSA and for Jones et al. 1989, where interphotoreceptor retinoid-binding protein was used. The experiments reported by McDowell 1993 were conducted with a 10-fold molar excess of 11cRet (25 μM) over opsin and demonstrated pseudo–first-order kinetics. Rotmans et al. 1974 demonstrated that the regeneration rate of opsin in bleached outer segment fragments (τ1/2 ≈ 10 min) was similar to the decay of Meta-II at 25°C (≈7 min half-life) and proposed that 11cRet enters the ligand-binding pocket only after all-trans-retinal has vacated the environment. FAF-BSA can be viewed as a nonspecific retinoid binding protein (Noy and Xu 1990; Livrea et al. 1991; Winston and Rando 1998). The likely mechanism of its action is that FAF-BSA effectively solubilizes chromophore in an aqueous environment and permits efficient transfer into lipid membranes. Once partitioned into cellular membranes, retinoid transfer into bleached pigments can be rapid (Noy and Xu 1990; McDowell 1993; Stecher et al. 1999). Curiously, the rate limiting enzyme in the visual cycle of rodent rod photoreceptors, all-trans-retinol dehydrogenase/reductase, is also expressed in kidney (Haeseleer et al. 1998). This raises the interesting hypothesis that HEK293S cells may have some capacity for retinoid metabolism. HEK293S cells are a suspension-adapted clone of transformed human embryonic kidney cells. As such, they are likely to express both general human cell housekeeping genes as well as some genes specific for the kidney. This cellular system has been a popular expression environment for heterologous genes including human and bovine opsins over the last decade (Nathans et al. 1989; Nathans 1990).
Spontaneous regeneration of visual pigment after 11cRet loading in FAF-BSA/vitamin E provides parsimonious support of ERC experiments. The regeneration of ERCs in WT-HEK293S cells from an internal source of 11cRet provides a clear advantage for complex experiments where charge motions must be compared against different conditions (e.g., action spectra determination). Two properties of ERCs measured in this environment are under further active investigation (Brueggemann and Sullivan, manuscript in preparation). First is the loss of the inward R1 signal after post-bleach spontaneous regeneration. The time to peak of the R2 signal is slowed approximately twofold when the R1 signal is present (Sullivan and Shukla 1999), possibly suggesting that the R1 signal represents additional state transformations. In cells that were primarily loaded with 11cRet and studied under whole-cell recording for periods of time >30 min, a slow accumulation of R2 charge (approximately twofold) occurred (see Fig. 7 A). The mechanism of this process is not yet clear.
The Ground State of Rhodopsin Is the Source of the ERC Signal
Strong evidence is demonstrated that ERC signals result from activation of the ground state of human rhodopsin in the plasma membrane of fused giant cells and that no other additional spectral states are contributing to these measurements. Evidence supports the conclusion that the identity of the chromophore regenerating visual pigment is 11cRet, which forms a PSB+-H with K296 of the opsin apoprotein. The action spectrum of the ERC of WT human rhodopsin, when stimulated through 30-nm bandpass filters, was well fit by a scaled normalized template of the absorbance spectrum of WT human rhodopsin purified from the same cell line used for ERC studies. The action spectrum was obtained from cells that had already had their ERCs extinguished and had undergone spontaneous recovery. These experiments allow the conclusion that the ground state of rhodopsin is the one that regenerates upon dark adaptation after primary ERC extinction. Thus, during dark adaptation, the Meta-II state must decay with hydrolysis of the Schiff base to form opsin, which then can subsequently react with 11cRet to form a protonated Schiff base at K296. From our previous work on spectral sensitivity, which was conducted with 70-nm bandpass filters, the action spectra was much broader than rhodopsin absorbance. Data from the experiments reported here shows that this outcome was the result of stimulus bandwidth, and not the presence of additional spectral states that were contributing to charge motion. Specifically, there were three pigments that could have contributed to “recovery” of the ERC when later stimulated with 500-nm stimuli: Meta-III465 and Meta470, both resulting from thermal Meta-II380 decay (Hofmann et al. 1992), and isorhodopsin resulting from photoregeneration from early intermediates. The formation and lifetime of Meta-III and Meta470 are consistent with the kinetics found for ERC recovery (≈3.0 min). However, if these states contributed substantially to the ERC, the action spectrum would not have been fit by the absorbance of rhodopsin and the signals should have been promptly extinguished by 10 mM NH2OH (Hofmann et al. 1992; Lewis et al. 1997). Given the difference in peak absorbance of bovine and human rhodopsin, these three pigments in isolation would have produced action spectra with peaks around 460, 475, and 483 nm. The human rhodopsin absorbance spectrum, peaking at 493 nm, provided a good fit to the action spectra peak and there was no evidence of a significant blue shift of the ERC data. In addition, the bandwidth of the ERC action spectrum was consistent with rhodopsin absorbance. Thus, we conclude that the pigment underlying the ERC after primary regeneration and secondary recovery is rod rhodopsin, which was not sensitive to NH2OH in the dark. The rapid recovery of ERC signal over 10-min periods suggests that long-lived photointermediates such as Meta-III or Meta470 are not extensively populated in this expression system. If these or other pigment states are present, their contribution to the ERC must be minimal. This conclusion is substantiated by the similar Pt values for ERC extinction after primary regeneration and secondary recovery. If additional states were involved it is likely that the Pt parameter would have changed significantly.
R2 Extinction Photosensitivity Is Consistent with Rhodopsin Activation
The ERC signal can be extinguished with successive flashes in a fashion similar to rhodopsin bleaching in photoreceptors and with similar photosensitivity. The mean and maximum Pt values (2.6 × 10−9 and 5.0 × 10−9 μm2, respectively) determined from exponential extinction decays are consistent with a rhodopsin photopigment but are lower than expected (1.0 × 10−8 μm2), taking the product of the molecular cross section of human rhodopsin at 493 nm (α493 = 1.528 × 10−8 μm2, see materials and methods) and the known quantal efficiency (γ = 0.67) (Dartnall 1968). The mechanism for the approximately twofold suppression of Pt is considered. One would not expect any significant pigment orientational effect in the fused cell system because rhodopsin is expressed in a membrane with an essentially spherical geometry at the time of recordings and any orientation should increase Pt. The Pt values obtained from ERC extinctions of R2 in amphibian rods (7.6 ± 2.2 × 10−9 μm2) originate predominantly from an A2 pigment with a lower peak extinction coefficient than human rhodopsin (≈30,000 vs. 40,000 M−1 · cm−1) (Makino et al. 1991). Since cells were regenerated with 11cRet, an A1 pigment is expected and found upon purification of rhodopsin from the same cell line. Therefore, the suppressed values of Pt must have a different origin. Our findings may reflect, in part, the variability of extracting the Pt parameter from extinction experiments (Makino et al. 1991). The accumulation of bleaching intermediates is unlikely because the extinction coefficients of Meta-III or Meta470 are similar to that for rhodopsin and would not be expected to suppress Pt. Moreover, the accumulation of these intermediates was essentially excluded by ERC persistence in NH2OH (and further Pt suppression) and the fitting of the ERC action spectrum by the absorbance spectrum of WT rhodopsin.
Photoregeneration from early intermediate states with high quantal efficiency for photoconversion (e.g., bathorhodopsin, lumirhodopsin) can definitively lower Pt estimates for the absorbing pigment (Williams 1964, Williams 1965). Makino et al. 1991 found no photoregeneration in ERC experiments on rod and cone photoreceptors when Pt was measured using flash strengths that were typically about an order of magnitude less intense than used here but with durations ∼20-fold longer (300 μs) that probably overlapped with the Meta-I lifetime and the Meta-I ⇔ Meta-II equilibrium. Flashes of this duration were, however, unlikely to overlap significantly with the lifetimes of earlier intermediates that have higher quantal efficiency of photoconversion (Williams 1964, Williams 1965, Williams 1974). At room temperature in these experiments, the only bleaching intermediates that could overlap with the 14-μs flash duration are bathorhodopsin, the blue-shifted intermediate, and lumirhodopsin. Lumirhodopsin is completely formed by 1 μs at room temperature (Lewis et al. 1989), and is likely to have been the most populated state during the flashes presented. The flash duration in these experiments would not overlap the rate of Meta-I formation at room temperature (100 μs) (Sengbusch and Stieve 1971). The stimuli used in these experiments should have no influence on the Meta-I ⇔ Meta-II equilibrium, which correlates in part with the R2 time course (Ebrey 1968; Spalink and Stieve 1980). In extracting Pt from R2 extinction, we are assessing early photochemical processes of absorption through examination of charge motions of later conformational states that are thermal and not photochemical in origin. If we use the mean Pt obtained at 500 nm (2.6 × 10−9 μm2), we can calculate the quantal efficiency assuming realistically that the molecular cross section is constant (1.23 × 10−8 μm2). This results in an estimate for γ of 0.21 that is only ∼32% of the known value of 0.67 determined from small bleaches.
How could γ be decreased to explain the suppression of Pt? In fact, Williams 1964, Williams 1965, Williams 1966, Williams 1974 have shown that γ is a function of light intensity. When the flash intensity is infinitely strong, γ should approach zero. γ is expected to decrease in proportion to the intensity of the flash stimulation because photoconversion becomes increasingly likely (even-numbered absorptions) from intermediates with lifetimes of the same order as the flash duration. Hagins 1955 and Williams 1964, Williams 1965, Williams 1974 have shown that the maximum fraction of rhodopsin bleached in a single brief flash can be only 50% because of this process. At the flash intensities used in these experiments (e.g., 500 and 70 nm FWHM, 4.08 × 108 photons/μm2), we estimate the fraction of rhodopsin molecules absorbing at least one photon to be ∼0.963 using the crude estimates obtained from the Poisson distribution (see materials and methods). However, only odd-numbered absorptions go on to bleaching, whereas even numbered absorptions photoconvert. The fraction of even-numbered photoregenerating absorptions is ∼50% at the flash intensity used. Therefore a maximum of 50% of rhodopsin molecules bleach with each flash, although there are sufficient photons in a single flash to bleach all molecules present. The amount of rhodopsin bleaching with each flash, then, is a constant fraction of unbleached molecules remaining. This explains why the extinction followed a single exponential decay function. It also explains why Pt is suppressed by at least 50% in our measurements. The R2 signal reports only those rhodopsin molecules that absorb an odd number of photons and move forward through later intermediates to bleaching. It is important to mention that the ERC R2 kinetics should not be influenced at all by photoregeneration because the flash duration is about two orders of magnitude more brief than the electrical state transitions measured. The shape of the R2 waveform does not change with sequential flashes even though the amplitude decreases and no evidence for photoregeneration was found when examining the extinction of the R2 signal or the homogeneous kinetics of R2 relaxation with different flash intensities. Thus, our findings are consistent with a significant suppression of Pt of WT rhodopsin, likely because of the high probability of early state photoconversion using the stimuli employed in these experiments. The intermediate likely to be significantly populated during the highest photon fluxes in these experiments is lumirhodopsin since it is completely formed by 1 μs and is the only state present during the bulk of photon irradiance (rise time ≈ 5 μs). High intensities were used to obtain good estimates of the cross section (α) in the spectral measurements and were maintained throughout because of the high SNR obtained in ERC recordings. In fact, Pt approximates the value expected when stimulus intensities (500 nm) are nearly a log-fold lower (see results). In future experiments, the effects of a range of flash intensities on Pt and quantal efficiency will be investigated. In addition, we will also investigate the mechanism for the marked suppressive effect (≈1 log) of NH2OH on Pt, which appears to be a novel finding. We speculate that NH2OH has the capacity to bind in retinal binding pocket and perhaps alter the local chromophore environment in the ground state in such a manner as to result in altered photochemical properties.
Photoregeneration of Meta-II Can Be Detected with ERC Measurements
Experiments were conducted to test whether the ERC can be used to study rhodopsin photoregeneration from the Meta-II state. These experiments were motivated by earlier studies that reported reversed ERP signals from the Meta-II state upon near UV stimulation (Cone and Cobbs 1969; Ebrey 1968; Drachev et al. 1981; Cafiso and Hubbell 1980). As discussed above, photoregeneration from early intermediates is likely to be responsible for the suppressive effect on estimation of Pt using the ERC method. To further explore photoregeneration as a molecular conformational process, we tested whether the ground state of rhodopsin could be achieved by photolytic activation of Meta-II, which has a lifetime accessible to xenon flash stimulation and cellular ERC measurement. Near UV flashes overlapping with Meta-II380 absorption and lifetime were presented immediately after extinction of 500 nm ERC responses and promoted rapid recovery of the ERC signal in a manner inconsistent with the normal chemical regeneration by 11cRet. Meta-II is known to have an extinction coefficient slightly larger than rhodopsin, and photocoversion of Meta-II to rhodopsin has been reported to occur with a quantal efficiency of ∼0.2 (Williams 1968; Williams and Breil 1968). Using estimates of the extinction coefficient for Meta-II between 33,000 and 49,200 M−1 cm−1 and the Poisson equation, the fraction of Meta-II molecules absorbing at least one photon would be 25–35%/flash with 80–90% of these being single photon absorptions at the 350-nm photon densities delivered (8.02 × 107 photons/μm2). The inverted ERC signals that were only seen in large cells are consistent with inverted ERP signals that have been recorded from the retina (Arden et al. 1966; Cone 1967; Ebrey 1968; Cone and Cobbs 1969). These results are consistent with a net reversal of charge flow during photoregeneration from Meta-II. The reverse state path is likely to be different compared with the forward activation pathway given that the amount of charge is much smaller (10–20%) and appears to have a considerably slower time course with respect to the peak times of the positive R2 signals that associate with Meta- II formation. Complete reversibility of the transitions would be expected to result in charge motions of the same magnitude but different polarity. Arnis and Hofmann 1995 recently showed that photoregeneration of rhodopsin from the active Meta-II state proceeds by isomerization of the chromophore and Schiff base reprotonation, but the ground state was achieved on a much slower time scale and was due to thermal and not photochemical conformational changes. Such slow events would not be seen in these current recordings, but might be reflected in ERP (voltage) recordings where the charge flow can be integrated on the membrane capacitance. Therefore, the small inverted ERC signals that we see with UV flashes could reflect dipolar rearrangements in the chromophore pocket or the reprotonation of the Schiff base, perhaps from the E113 protonated counterion, which is located more toward the extracellular surface of rhodopsin (Baldwin 1993; Jager et al. 1994). Or it could reflect release of protons from the cytoplasmic surface (Ostroy 1974). Future experiments will focus on the rate of thermal disappearance of the Meta-II state from which photoregeneration most likely occurs and the molecular origin of the inverted charge flow (e.g., pH dependence). The quantal efficiency of the photoregeneration process appears favorable to study the mechanism in greater detail in even larger cells.
9-cis-Retinal Regenerates ERCs from WT Opsin
When WT-HEK293S cells are regenerated with 9cRet, ERC signals are recordable upon 500-nm flash photolysis. 9c-Ret regenerates a visual pigment, presumed to be isorhodopsin (peak ≈ 483 nm), which would broadly overlap with the 70-nm band stimulus. That ERC signals can be recorded from a “natural” analogue pigment provides evidence to support a role for the ERC in investigation of a wide variety of analogue visual pigments that are known to have unique properties. For example, certain locked analogues are known to prevent energy uptake by preventing isomerization (Bhattacharya et al. 1992). Others block the bleaching sequence at discrete spectral states, for example, 9-cis-desmethyl-retinal blocks the thermal bleaching pathway at Meta-I (Ganter et al. 1989). Investigation of analogue pigments can now be extended to the vectorial charge flows that are orthogonal to the chromophore plane. Similarly, analogue retinals could regenerate opsins with engineered side-chain mutations in the ligand-binding pocket to investigate environmental interactions that shape chromophoric properties and activation behavior.
Kinetic Analysis of R2 ERC Signals of WT and Mutant (D83N and E134Q) Pigments
A major challenge in structure–function studies of visual pigments has been the preparation of sufficient mutant protein in expression systems for biophysical or biochemical analysis. The ERC method improves detection sensitivity by 107–108-fold, allowing measurements of conformational activation in an ensemble of regenerated rhodopsin molecules in the physiologically intact environment of a single fused giant cell (Sullivan and Shukla 1999). In applying the ERC to mechanistic studies of conformational activation, a number of rhodopsin mutants were screened at sites of potential proton exchange reactions in the membrane environment of the protein. The time course of the expression ERC R2 signal overlaps with the temporal scale of Schiff base deprotonation and Meta-II formation, as well as the related proton uptake into the cytoplasmic face of the pigment and the conformational transition to the biochemically active Meta-IIb state (Arnis and Hofmann 1993). Since known mutants can affect various steps in this activation process, a quantitative structure–function investigation on opsin mutants was initiated in this ERC study. A mutation in the PSB+-H environment (D83N) on the second α helix was found to perturb the kinetics of the R2 ERC phase. A mutation on the cytoplasmic face of the third α helix (E134Q), at a residue known to be a gatekeeper to both proton uptake and the related generation of the transducin docking environment, results in loss of most of the ERC R2 signal except for the fast initial process.
The R2 signal is well resolved in cellular ERC measurements and overlaps temporally with critical events leading to biochemical activation of rhodopsin. WT rhodopsin R2 relaxation kinetics are an easily accessible aspect of the total ERC signal and are invariant to intensity or wavelength of stimulation as found in these experiments. Therefore, the R2 relaxation should be a reliable parameter to evaluate for understanding conformational dynamics. A quantitative analysis of R2 relaxation in WT rhodopsin should serve not only to begin characterization of the state complexity of charge motion on the millisecond time scale, but also serve for comparison of mutant pigments that might be affected in some aspect of R2 electrical state transitions. WT rhodopsin R2 relaxation was first characterized by fitting sums of exponentials to a set of WT ERC waveforms and generating an ensemble of time constants (τa,b). We assumed that the ensemble of time constants would represent the kinetic aspects of the signal even with any slight heterogeneity (e.g., due to cell size) that might affect only the fastest aspects of the relaxation (Sullivan and Shukla 1999). Treating the entire time constant data set as an ensemble is reasonable because the assignment of a value to τa or τb is somewhat arbitrary given that the short time constant may not have been weighted or fit in a particular ERC signal and a longer time constant was then assigned to τa. In fact, there is some overlap in the assignments (Sullivan and Shukla 1999). Also, analysis of an ensemble of ERC signals is more likely to identify additional complexity for quantification than can be obtained in fittings to single ERC waveforms. A histogram was generated from the ensemble of the total WT time constants data set (Fig. 10 A). The histogram is not consistent with a single distribution of time constants. While there was a prominent peak around 4 ms, there is significant weighting outside this band leading to a skewed distribution with additional peaks being apparent. To characterize the distribution of time constants in the R2 relaxation, we fit sums of Gaussians to the (skewed) histogram of the entire τa,b ensemble. We attempted fits of one, two, three, and four Gaussians to the data. With this data set, a reliable fit of the sum of three Gaussians was obtained that was independent of bin width. Three Gaussian peaks were centered around 4.1, 12.45, and 26.4 ms, and the errors in the fitting results are shown in Table. The natural conclusion from this analysis is that there are a minimum of three distinct electrically active states in the WT R2 relaxation at room temperature in WT rhodopsin. We speculate that the underlying molecular basis of the three fitted time constants are likely to represent processes related to Schiff base deprotonation, proton uptake, and α helical movements associated with conformational activation. However, other interpretations are possible. For example, the capacitative, or AC-coupled nature of the ERC could reflect forward and reverse movements of charges (e.g., protons, sidechains), in particular molecular environments, leading to a series of coordinated charge separation and recombination events, each satisfying a zero-time integral expected for pure capacitative components (e.g., see Hong et al. 1992). We have not demonstrated a net DC component to the ERC. Further investigation will be necessary to examine these hypotheses and to determine the uniqueness of the time constants determined. We anticipate that dissection of the ERC R2 signal can be achieved with mutations at protein side chains likely to participate in charge transfer events.
ERC signals from D83N rhodopsin (regenerated with 11cRet) appear slightly simplified in comparison with WT. The R2 signal appears WT in nature, but shorter in comparison to the broad stretched relaxation of the WT pigment. D83N ERCs were subjected to R2 relaxation time constant analysis, and the sum of two Gaussian distributions was required to best fit the total time constant histogram. Peaks were identified at 3.67 and 10.67 ms, and the errors in the fitting are shown in Table. These times appear comparable to the fast and medium time constants in the WT pigment that was measured under identical conditions. The third and longest time constant of the R2 relaxation in WT (26.4 ms) is missing in the D83N pigment. We suspect this is likely to represent a distinct property of the D83N pigment since R2 relaxation in D83N ERCs have a different waveform compared with WT. In summary, D83N loses the R1 signal during primary bleaches and its R2 relaxation loses the “stretched exponential” characteristic of the WT R2 relaxation.
The D83N mutation is known to slightly blue-shift the absorbance of rhodopsin, which is consistent with an alteration of the Schiff base environment in the ground state pigment, resulting from the isomorphic loss of the protonated carboxyl group of aspartate (Fahmy et al. 1993; Rath et al. 1993). However, D83N pigment in detergent has the capacity to form Meta-II and to activate transducin, but these properties have not been examined under rapid time-resolved conditions (Sakmar et al. 1989; Zhukovsky and Oprian 1989; Nathans 1990; Cohen et al. 1993). Weitz and Nathans 1993 concluded that the D83N mutation leads to a more favored formation of Meta-II compared with WT rhodopsin by analysis of difference spectra. The D83 sidechain is located in the immediate environment of the protonated Schiff base and is a highly conserved residue in G-protein–coupled receptors (Baldwin 1993). Therefore, it could play an as yet unrealized role in rapid activation processes. Fahmy et al. 1993 found a shift in carboxyl signals in the WT Meta-II/rhodopsin infrared difference spectra that was lost in the D83N mutant. They suggested that the D83 sidechain undergoes an increase in hydrogen bonding during Meta-II formation, although they could not explicitly rule out a transient deprotonation/reprotonation reaction on the path to Meta-II. Ganter et al. 1988, Ganter et al. 1989 suggested that the formation of Meta-I from lumirhodopsin was associated with carboxyl groups in the membrane environment undergoing such reactions. This is an area where ERC recordings are likely to be useful because of the extended bandwidth and profound relative sensitivity compared with most spectroscopic techniques. The apparent loss of the R1 signal is particularly interesting because this signal is believed to originate from charge separation of the protonated Schiff base from its counterion during isomerization (Trissl 1982; Birge 1990a,Birge 1990b). It may also be that later thermal transitions contribute to the inward R1 charge flow and that the D83N mutation may prevent these states during primary extinction, where we usually see R1 signals. Since the recording conditions are otherwise constant between WT-, D83N-, and E134Q-expressing cells (e.g., similar access resistance and similar ranges of membrane capacitance), it is unlikely that the loss of the R1 signal is purely due to changes in the measurement under whole-cell voltage clamp (Okajima and Hong 1986). The loss of R1 signal in the D83N pigment leads us to consider that side-chain interactions in the ligand binding pocket near the Schiff base could contribute to R1. The loss of R1 in the WT pigment after primary extinction may simply mean that the spatial order of such interactions is not rapidly achieved under the conditions of these experiments. Finally, the apparent simplification of the R2 relaxation in D83N may indicate that the path to Meta-II is slightly different with respect to the WT pigment, even though Meta-II ultimately results. Further experiments will be necessary to investigate these ideas as we use a variety of approaches to dissect the R2 relaxation into quantifiable biophysical processes.
ERC signals from E134Q rhodopsin have little complexity in comparison with WT and demonstrate no R1 and a brief R2 relaxation always well fit by a single exponential. When R2 relaxation time constant analysis was applied to the E134Q pigment, a single Gaussian was needed to fit the time-constant histogram. The value of the peak was 4.4 ms, which was comparable within error to WT fast phase. The timing of the fast phase did not appear to change when the cytoplasmic pH was held at different constant values over the range from 6.0 to 8.0. The E134 sidechain is known to be essential to proton uptake into the cytoplasmic face of rhodopsin during normal Meta-II formation given that the E134Q mutant binds zero protons in comparison with the two adsorbed by the WT pigment (Arnis et al. 1994; Parkes and Liebman 1984). The sites of binding of the protons are unknown, but the E134 sidechain could titrate one proton. The loss of the medium and slow ERC time constants and the stretch beyond 50 ms in the WT R2 relaxation suggest that these components might represent proton movements or other charge motions gated by proton uptake. E134Q, when regenerated with 11cRet in the dark, is capable of activating transducin in an enhanced manner in comparison with the WT pigment after light stimulation (Franke et al. 1992; Cohen et al. 1993). The E134/R135 charge pair is known to be an essential switch to transducin docking and activation and is screened from solution once transducin is bound to R* (Franke et al. 1992; Ernst et al. 1995; Fahmy and Sakmar 1993; Acharya and Karnik 1996). During photoactivation to R*, a rigid body motion of α helix VI relative to α helix III (the location of E134/R135) has been detected. This results in helix VI being displaced outward from the disk membrane into the cytoplasm, although the magnitude of the vectorial movements are not yet clear (Farrens et al. 1996; Altenbach et al. 1996; Han et al. 1996). The E134Q mutation promotes movements of the III and VII α helices in the dark that are not affected by light and do not affect the subsequent movement of α helix VI. What results is a partial activation of the receptor in the darkness or constitutive activity (Kim et al. 1997). The movements of α helices may be normally coupled, but can still occur independently in the presence of environmental perturbation (e.g., mutation). Since spectral Meta-II formation is not perturbed in E134Q, we suggest that the residual fast ERC signal in this mutant represents the charge motion associated with Schiff base deprotonation. In preliminary experiments, the residual R2 time constant of E134Q did not demonstrate sensitivity to intracellular pH over the range of 6.0 to 8.0 and might reflect the pH-insensitive component of the ERP related to Schiff base deprotonation (Bennett et al. 1980; Lindau and Ruppel 1985; Arnis and Hofmann 1993). By elimination, the remaining medium and slow time constants of the WT R2 relaxation could be associated with proton uptake into the cytoplasmic face and the subsequent conformational transformation of the rhodopsin that is gated by this chemical potential (e.g., macrodipole movements of α helices). If so, then these later time constants in the WT pigment should be responsive to changes in intracellular pH or to molecular conditions that prevent the conformational activation of the molecule that allows transducin binding. Further experiments will be necessary to decipher the molecular origin of the charge motions during R2 and how local environment affects these transitions.
We were surprised that no R1 signals were found in fused E134Q giant cells given that the E134 sidechain resides at the cytoplasmic face of the third α helix remote from the Schiff base environment where R1 is thought to originate. However, total charge motion in E134Q cells is smaller than in WT cells and the R1 signal may simply have been lost in noise, as we found for R1 in single unfused cells expressing WT pigment regenerated with 11cRet (Sullivan and Shukla 1999). However, in a separate cell line of E134Q (E134Q-12) that expresses three to four times more mutant opsin (Sullivan and Satchwell 2000), the R1 signal was also missing and the R2 relaxation is still simple and single exponential in character (Brueggemann and Sullivan, manuscript in preparation). In future efforts, the stoichiometry of charge motion in the E134Q pigment will be compared with WT. Given that R1 is likely to be definitively lacking, a model should be considered wherein the membrane-buried Schiff base environment near the middle of the membrane and the proton uptake environment on the cytoplasmic face of the membrane might communicate on a very rapid time scale during activation.
Although we do not yet have a molecular understanding of the simplification of the ERC R2 signal in D83N or E134Q, such mutant pigments with uniquely different ERC signals constitute a substrate to understand the components of the normal ERC signal in WT rhodopsin. At this point, our efforts have clearly established the feasibility of using the ERC to study time-resolved and electrically active conformational changes during rhodopsin activation in WT and two mutant pigments. Future studies will focus on the molecular mechanism of R* formation and the forces that govern the molecular volume increase of the pigment during the Meta-IIa to Meta-IIb transition in WT and mutant pigments (Arnis and Hofmann 1993).
Summary and Conclusions
The sensitivity of the expression ERC method is greatly improved when compared with other contemporary time-resolved techniques applied to rhodopsin activation and allows measurement of rhodopsin activation in single cells containing about a picogram of visual pigment. The sensitivity of this approach is dependent upon the use of gigaohm-seal, whole-cell patch clamping techniques that are capable of resolving small currents with fast time resolution. The ERC is a conformation-dependent charge motion of the same family as ionic channel gating currents. Gating current analysis of expressed mutant ionic channels has significantly extended knowledge of molecular processes of conformational activation of ionic channels, and for this reason we are applying the same approach to visual pigments that are known to be electrically active proteins. In this work, we apply the expression ERC approach to measure properties of WT rhodopsin and found that the ERC signal results from activation of the ground state of rhodopsin, given the action spectra and bleaching photosensitivity. ERC kinetics are independent of wavelength, as expected for the univariance principle, and rhodopsin can be photoregenerated from the Meta-II state. The major impact of this study resides in our success at recording ERCs from an analogue visual pigment (isorhodopsin) and from two mutant human opsins regenerated with 11cRet. This study establishes the feasibility of an ERC structure/function investigation applied to the ligand binding pocket, as probed by artificial chromophores or site-specific mutations, or remote regions of the pigment, as probed by engineered mutations or chemically reactive ligands. The ERP has been similarly used in expressed mutant bacteriorhodopsin pigments to understand proton exchange mechanisms (e.g., Moltke et al. 1992; Misra 1998). The similarity of the visual rhodopsin ERC/ERP and the bacteriorhodopsin ERP is striking and suggests that similar underlying mechanisms of conformational transitions may be conserved over evolutionary time. The ERC measures a vector of activation across the membrane plane in a population of oriented molecules under physiologically intact conditions not dissimilar to the environment of activation in a photoreceptor. We anticipate that this approach will lead to further understanding of the energies, forces, and molecular processes of rhodopsin activation on a rapid time scale.
The authors thank Drs. Michael Sheets and J. Hugh McDowell for sharing their methods on recording ionic channel gating currents from fused transfected HEK293 cells and on FAF-BSA rhodopsin regeneration, respectively. The authors acknowledge the technical assistance of Michael F. Satchwell with cell cultures. The authors appreciate the critical review of the manuscript by Drs. Bob Birge, Ken Foster, and Barry Knox before its submission and the comments of the reviewers before publication.
This work was funded by an R29 award (EY11384) from the National Eye Institute to J.M. Sullivan. Also supported, in part, by start-up funds from the Department of Ophthalmology at SUNY (which is a recipient of a Research to Prevent Blindness Award).
1used in this paper: 11cRet, 11-cis-retinal; Cmem, membrane capacitance; ERC, early receptor current; ERP, early receptor potential; FAF, fatty acid–free; FWHM, full-width-half-maximum; PSB+-H, protonated Schiff base; Pt, photosensitivity; SNR, signal-to-noise ratio; WT, wild type