Optimization of Receptor-G Protein Coupling by Bilayer Lipid Composition I

The role of membrane composition in modulating the rate of G protein-receptor complex formation was examined using rhodopsin and transducin (Gt) as a model system. Metarhodopsin II (MII) and MII-Gt complex formation rates were measured, in the absence of GTP, via flash photolysis for rhodopsin reconstituted in 1-stearoyl-2-oleoyl-sn-glycero-3-phosphocholine (18:0,18:1PC) and 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine (18:0,22:6PC) bilayers, with and without 30 mol% cholesterol. Variation in bilayer lipid composition altered the lifetime of MII-Gt formation to a greater extent than the lifetime of MII. MII-Gt formation was fastest in 18:0,22:6PC and slowest in 18:0,18:1PC/30 mol% cholesterol. At 37 °C and a Gt to photolyzed rhodopsin ratio of 1:1 in 18:0,22:6PC bilayers, MII-Gt formed with a lifetime of 0.6 ± 0.06 ms, which was not significantly different from the lifetime for MII formation. Incorporation of 30 mol% cholesterol slowed the rate of MII-Gt complex formation by about 400% in 18:0,18:1PC, but by less than 25% in 18:0,22:6PC bilayers. In 18:0,22:6PC, with or without cholesterol, MII-Gt formed rapidly after MII formed. In contrast, cholesterol in 18:0,18:1PC induced a considerable lag time in MII-Gt formation after MII formed. These results demonstrate that membrane composition is a critical factor in determining the temporal response of a G protein-coupled signaling system.

In G protein-coupled signaling pathways, the stimulus is passed from the receptor to a G protein and subsequently, to an effector enzyme. The number of G protein-coupled signaling pathways has grown tremendously in the past 10 years, and now includes sensory pathways associated with vision, olfaction, taste, and receptors for dopamine, serotonin, ␥-aminobutyric acid, and histamine. Much of the basic information about G protein-coupled signaling pathways was obtained from work on the visual system (1,2), and it is the only G protein-coupled pathway for which there are three-dimensional structures for both the receptor, rhodopsin (3), and G protein, transducin (G t ) 1 (4). A unique aspect of the visual system is the rapid rate of G protein activation. Recent biochemical measurements establish that G t activation by rhodopsin occurs about 100 times faster than the activation of other G proteins (5). Two properties are believed to be responsible for this rapid rate of G t activation: 1) the high concentration of G t on the surface of the rod outer segment disc membrane and 2) the low degree of acyl chain packing order in the disc membrane, resulting from the high levels of phospholipid acyl chain polyunsaturation (5). Recent experiments with whole rod cells from transgenic mice that expressed half the normal amount of rhodopsin establish that the rate of phototransduction in the rod cell is determined by protein diffusion in the disc membrane (6).
Within a few milliseconds of photon absorption, a metastable equilibrium is established between two conformational states of photoexcited rhodopsin, MI and MII (7). MII is the conformation of photoisomerized rhodopsin that binds and activates G t (8 -10). G t is bound to the membrane by a combination of electrostatic forces (11) and hydrophobic forces associated with a farnesyl group on the ␥ subunit of G t (12,13). The binding interaction between rhodopsin and G t has been examined in some detail (for reviews, see Refs. 14 and 15). Binding of G t by MII catalyzes the exchange of GDP for GTP, and the visual "signal" is carried from rhodopsin to the effector enzyme, a cGMP-specific phosphodiesterase. The activated phosphodiesterase catalyzes the hydrolysis of cGMP, which triggers the closure of cGMP gated Na ϩ /Ca 2ϩ channels leading to the hyperpolarization of the rod outer segment (ROS) plasma membrane and the visual response. In the absence of GTP, the MII-G t complex is stable and has the same absorption spectrum as unbound MII. In the outer segment of the human rod cell there are about 10 rhodopsin molecules for every G t molecule. However, during normal physiological function, only ϳ1 rhodopsin of every 100,000 is photoisomerized and active at any one time. Consequently, the rate of MII-G t complex formation is governed by a two-dimensional, diffusion-controlled search in the plane of the disc membrane of G t for photoactivated rhodopsin, in its MII conformational state (6).
Previous studies in this laboratory (16) and others (17,18) demonstrate that the equilibrium concentration of MII is increased by the presence of phospholipids with one or more 22:6n-3 acyl chains. An important aspect of visual signal transduction is rapid response, and studies on rhesus monkeys show that electroretinagram b-wave implicit times (19) and a-wave * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  implicit times 2 are delayed in n-3-deficient animals. Studies on human infants show that light-adapted oscillatory potentials are delayed in infants fed standard infant formula without 22:6n-3 relative to infants fed 22:6n-3-supplemented formula (20). The present study was undertaken to determine the effects of 22:6n-3 and cholesterol on the kinetics of the first two steps in visual signal transduction, which are the formation of MII and the binding of G t to MII. The latter process represents the first stage in signal amplification in the visual transduction pathway.

EXPERIMENTAL PROCEDURES
Sample Preparation-Phospholipids 18:0,18:1PC and 18:0,22:6PC were purchased from Avanti Polar Lipids Inc. (Alabaster, AL), and their purity was ascertained by high performance liquid chromatography. Cholesterol was purchased from Calbiochem (La Jolla, CA). All preparation of phospholipids was carried out in an argon-filled glove box and in thoroughly degassed buffers because of the susceptibility to oxidation of polyunsaturated phospholipids. ROS were isolated from frozen retinas (James and Wanda Lawson, Lincoln, NE) as described previously (21). ROS was solubilized in octyl glucoside, and rhodopsin was purified on a concanavalin A affinity column (22). Large unilamellar vesicles containing rhodopsin at a ratio of 1 rhodopsin to 100 phospholipids were prepared using the rapid dilution technique (23). Following dialysis to remove detergent, all vesicle preparations were suspended in pH 7.5 Tris basal salt (TBS) buffer consisting of 10 mM Tris, 60 mM NaCl, 30 mM KCl, 50 M diethylenetriamine pentaacetic acid. G t was prepared from ROS as a hypotonic extract (24), the activity level was determined by a binding assay using 35 S-labeled GTP␥S (25), and stored in pH 7.5 TBS buffer with 30% glycerol at Ϫ20°C for no longer than 2 weeks. The phospholipid, cholesterol, and rhodopsin content of each reconstituted vesicle preparation were determined by independent phosphate (26), cholesterol (Waco Chemicals USA, Inc., Richmond, VA), and ⌬A 500 assays, respectively. Samples for flash photolysis measurements were prepared by diluting concentrated, reconstituted vesicle stocks with pH 7.5 TBS buffer, dividing the solution in half, and adding concentrated G t to one half and an identical amount of G t buffer to the other half. Samples were then incubated for 4 h on ice to ensure binding of G t to the bilayer (24). Final concentrations were 7.5 M rhodopsin and 0.5-1.5 M G t .
Flash Photolysis Measurements-Kinetics of MII and MII 158 G t formation were assessed by measuring the transient absorption at 380 nm using a flash photolysis system constructed in the laboratory (27) with the following modifications. Excitation was provided by a high pressure flash lamp (E.G. & G., pulse width ϭ 1s) filtered with a broad (Ϯ 25 nm) bandpass filter centered at 500 nm. The current from a thermoelectrically cooled photomultiplier tube (R928, Hamamatsu) was passed to a low noise current amplifier (Stanford Research). The amplifier output voltage was acquired at 2-10 s/point by a 1.25-MHz., 12-bit analog-to-digital converter (National Instruments) installed in a PC.
Data Analysis-The detailed kinetics of MII formation were extracted from the changes in absorbance observed at 380 nm in the absence of G t via analysis in terms of the microscopic rate constants of the square (28) and branched (27) photoreaction models shown in Fig.  1. The observed absorbance increase at 380 nm was directly analyzed in terms of the microscopic rate constants of the two models using NON-LIN (29) with subroutines specifying each model written by the authors.
The dilution reconstitution procedure produces rhodopsin-containing vesicles with half of the rhodopsin oriented with its cytoplasmic face external to the vesicle and half with the cytoplasmic face internal to the vesicle (23). G t that has been added to the vesicle suspension can only interact with half of the rhodopsin; thus, the absorbance change observed in the presence of G t is the result of three distinct species: MII inside , MII outside , and MII-G t . To isolate the absorbance increase caused by formation of MII-G t , it is necessary to remove the contributions resulting from MII inside and MII outside . The absorbance change recorded in the absence of G t was multiplied by the following scaling factor and subtracted from the absorbance change recorded in the presence of G t .
Scaling factor ϭ ͓͑K eq ϪG ϩ 1͒/͑K eq ϩG ϩ 1͒ ϩ 1͔/2 (Eq. 1) K eq ϪG and K eq ϩG are the MI-MII equilibrium constants measured in the absence and presence of G t , respectively, and were determined from equilibrium measurements on the same samples used for the kinetic measurements (30). Absorbance increases at 380 nm observed in the presence of G t were analyzed with the two microscopic photoreaction models shown in Fig. 1 plus a sum-of-two exponentials process for the binding of G t to MII.

RESULTS
The rate of formation of MII and MII-G t complex in 18:0,22: 6PC and 18:0,18:1PC with and without 30 mol% cholesterol was measured by flash photolysis. The flash-induced increase in absorbance at 380 nm, near the absorbance maximum for MII, was monitored from 20 s to several hundred ms. For samples containing G t , it was necessary to separate the rate of MII formation from that of the MII-G t complex by characterizing the kinetics of MII formation observed in the absence of G t . The increase in absorbance corresponding to the formation of MII was analyzed using the two microscopic photoreaction models shown in Fig. 1, which characterize the portion of the rhodopsin photoreaction cascade from the decay of lumirhodopsin to the formation of MII. An example of the flash-induced increase in absorbance at 380 nm for rhodopsin in 18:0,22: 6PC/30 mol% cholesterol with and without added G t at 37°C is shown in Fig. 2. The results of the analysis for the G t -free sample using the branched model (Ref. 27; Fig. 2A) and the square model (Ref. 28; Fig. 2B) demonstrate the difference between these two models. In the branched model, both MII fast and MII slow are present at equilibrium; thus, the MI-MII equilibrium for this model is calculated according to K eq ϭ (MII fast ϩ MII slow )/MI. The two models produced values of K eq for the MI-MII equilibrium that agreed with each other, as shown by the similar amplitudes for both MI and total MII for both models. The significant difference between the two photoreaction models is the manner in which they account for the most rapidly appearing species with an absorbance at 380 nm. For the branched model, the most rapidly forming species is MII fast (dashed curve, Fig. 2A), which rises to a maximum at ϳ 200 s and decays to a level such that it constitutes ϳ25% of total MII at equilibrium. In the square model, the most rapidly forming species with an absorbance at 380 nm is MI 380 (dashed curve, Fig. 2B), which reaches a maximum at ϳ200 s and then decays to zero by 6 ms. The decay of MI 380 to zero results in the time course for formation of total MII being shifted to the right for the square model (solid curve, Fig. 2B) relative to the branched model (solid curve, Fig. 2A). An exponential rise to a plateau is characterized by the lifetime, , denoting the time The significance of the 0.79-ms difference in the rate of formation of MII for the two models is illustrated by the analysis of the flash-induced kinetics in the presence of G t , shown in Fig. 2C. The upper curve in Fig. 2C was obtained for a sample identical to the one shown in Fig. 2 (A and B), except that it contains 1 M G t . The lower curve in Fig. 2C is the isolated formation of the MII-G t complex, as determined by an empirical separation of the raw data. The separation was accomplished by scaling the transient recorded without G t according to Equation 1 and subtracting it from the transient observed in the presence of G t . Inspection of this curve shows that it rises very rapidly, and the hash mark denotes the time, 0.74 ms, when it reaches to within (1 Ϫ 1/e) of its equilibrium value. Thus, the lifetime for MII formation, according to the square model, is greater than the lifetime for the MII-G t complex formation. This physically improbable situation is simply a result of the provision in the square model that the most rapidly forming species at 380 nm is a MI species that decays to zero at equilibrium. In the branched model, the most rapidly forming species at 380 nm is a kinetically distinct form of MII. As a result, formation of total MII, according to the branched model, is approximately coincident with formation of the MII-G t complex. The analysis and comparisons presented in Fig. 2 make it clear that at 37°C the fastest component appearing at 380 nm must correspond to a "normal" MII species that is capable of binding G t . Because of the consistency generated for the kinetics of the MII and MII-G t formation, the branched model for the Lumi-MI-MII portion of the rhodopsin photoreaction cascade was employed in all analysis.
In the absence of G t , all kinetic data were well described by the branched model (Fig. 1), as shown by the smooth curves through the data obtained in the absence of G t , as shown in Fig.  3. The flash-induced increase in equilibrium absorbance at 380 nm is proportional to the MI-MII equilibrium constant, K eq ϪG .

The values of K eq
ϪG calculated from the branched model agreed with those determined from equilibrium difference spectra obtained for the same samples (30). Agreement between the values of K eq ϪG determined from equilibrium measurements and from analysis of kinetic data with the branched model was reported previously (31,32). In the presence of G t , all kinetic data were well described by using the solution to the branched model obtained in the absence of G t , rescaled according to Equation 1, plus the sum of two exponential terms, as shown by the smooth curves through the kinetic absorbance changes acquired in the presence of G t (Fig. 3).
The two pairs of curves in Fig. 3 demonstrate one of the fundamental differences between the kinetic processes observed at 37°C and those observed at 20°C. The two kinetic traces observed at 37°C (Fig. 3A) diverge within ϳ100 s of the activating flash, demonstrating the rapid formation of the MII-G t complex. In contrast, the two kinetic traces observed at 20°C (Fig. 3B), do not diverge until 2 or 3 ms after flash. This difference demonstrates that, at 37°C, the MII-G t complex forms very quickly after MII formation, whereas, at 20°C, there is a lag time in the formation of the MII-G t complex relative to MII.
The increase in absorbance at 380 nm was measured in the absence and presence of G t at physiological temperature for rhodopsin in vesicles with four different bilayer compositions: 18:0,22:6PC and 18:0,18:1PC, with and without 30 mol% cholesterol. Changes in phospholipid acyl chain composition and cholesterol altered the kinetics of both MII formation (Fig. 4A) and MII-G t complex formation (Fig. 4B). The data in each panel in Fig. 4 have been scaled to the same maximum change in absorbance to facilitate comparison of the differences in kinetics. Comparison of the curves in panels A and B of Fig. 4 show that, whereas both decreased acyl chain unsaturation and increased bilayer cholesterol delay the formation of MII, variation in these parameters cause an even greater delay in the formation of the MII-G t complex.
Complete analysis of the kinetic data acquired at 37°C, summarized in Fig. 5, shows that cholesterol has relatively little effect on the kinetics of either MII or MII-G t formation in a bilayer consisting of 18:0,22:6PC. This is consistent with a number of measurements that show that cholesterol has its smallest effect on bilayer properties in bilayers that contain 22:6n-3 acyl chains (33). In contrast, in an 18:0,18:1PC bilayer, cholesterol increases the time required for MII formation by 50%, and the time required for MII-G t formation by 400%. Reducing sn-2 acyl chain unsaturation from 22:6 to 18:1 doubles the time required for both MII formation and MII-G t formation. At physiological temperature, the rates of MII-G t complex formation approach those of MII formation in all bilayer compositions examined, with the exception of 18:0,18: 1PC/30 mol% cholesterol, as shown in Fig. 5.
To isolate the effects of bilayer composition on the lateral diffusion process that leads to formation of the MII-G t complex, it is informative to compare the lifetimes for MII formation and MII-G t complex formation. Comparison of the ratio of the average lifetimes, ͗͘ MII-G t/͗͘ MII , eliminates the effect of delayed MII formation on the diffusional search process. At physiological temperature, MII formation and MII-G t complex formation are nearly coincident in 18:0,22:6PC, even in the presence of 30 mol% cholesterol, as shown in Table I. Because MII cannot react with G t any faster than it is formed from MI, the rate of formation of MII-G t complex in 18:0,22:6PC appears to be maximal. In 18:0,22:6PC, the rate of MII-G t complex formation, which is limited by lateral diffusion, is only slightly altered by the presence of 30 mol% cholesterol, further suggesting that At 20°C the rates of MII and MII-G t formation in bilayers were slower than those at 37°C. The greatest effect of bilayer composition was on the lifetime for diffusion-dependent MII-G t complex formation, as shown in Table I Table I. In pure 18:0,18:1PC, the MII-G t complex formed 60% slower than in 18:0,22:6PC, and the addition of 30 mol% cholesterol slowed MII-G t complex formation by a factor of 2.5 in 18:0,18:1PC, whereas it had relatively little effect in 18:0,22:6PC. These differences suggest that a high level of polyunsaturation at the sn-2 position increases the rate of G t lateral diffusion. DISCUSSION The bilayer compositions examined in this study produced changes in the rates of formation of both MII and MII-G t complex, demonstrating that the diffusion of receptor and G protein are sensitive to the lipid composition of the membrane. G t is anchored to the membrane by an isoprenoid moiety; thus, the effects of bilayer composition on G t lateral diffusion are likely to be similar to those observed for phospholipid lateral diffusion. The present results are consistent with changes in the rate of lateral diffusion of fluorescent phospholipids caused by changes in bilayer cholesterol (34) and changes in polyunsaturated acyl chain oxidation (35). These two studies reported a positive correlation between slower lateral diffusion and more ordered acyl chain packing. Previous studies demonstrate that increased phospholipid acyl chain unsaturation leads to more disordered acyl chain packing and increased bilayer free volume (36). The present results are also in agreement with a detailed analysis of lateral diffusion in di14:0PC/cholesterol bilayers that determined that cholesterol slows phospholipid lateral diffusion by decreasing phospholipid acyl chain packing free volume (37).
A unique feature of this study is the comparison of the kinetics of the unimolecular MI-to-MII reaction and the diffusiondependent MII-G t at physiological temperature and at a reduced temperature. The results presented in Fig. 3 and summarized in Table I demonstrate that these two kinetic processes are slowed to different degrees by reduced temperature. In a bilayer consisting of 18:0,18:1PC, at 37°C, the lifetime for MII-G t formation is only 20% greater than the lifetime for MII formation. However, at 20°C, MII-G t formation trails MII formation by a factor of 3. The implication of this difference is that studies of MII-G t binding must be carried out at physiological temperature if the ultimate goal of the investigation is provide an explanation for processes observed in vivo.
It is noteworthy that, among the bilayer compositions examined, the two compositions that most closely resemble the na-tive ROS disc membrane are 18:0,22:6PC with and without 30 mol% cholesterol. Like the native membrane, 50% of the acyl chains in these bilayers are 22:6n-3, and the basal disc membranes contain 30 mol% cholesterol (38), and the predominant acyl chain configuration of 22:6-containing phospholipids in the ROS disc membrane is sn-1 saturated, sn-2 22:6n-3 (39). The results presented here suggest that the delays in electroretinagram response observed in dietary n-3 deficiency 2 (19,20) could be partially the result of a slower rate of MII-G t formation in the absence of 22:6n-3 phospholipid acyl chains. At physiological temperature and the conditions of our measurements, MII-G t complex formed twice as slowly in 18:0,18:1PC relative to 18:0,22:6PC. The measurement conditions differ from those found in vivo in several respects. The two most significant differences are the low level of rhodopsin stimulation in vivo and the large contrast in acyl chain composition between 18: 0,18:1PC and 18:0,22:6PC. In the measurements reported here, 15% of the rhodopsin was bleached, resulting in an average distance between MII molecules of 15 nm. The physiological range of rod cell function involves activation of about 1 rhodopsin of every 100,000, resulting in an average distance between MII molecules of about 1000 nm. Thus, the experimental conditions employed in the present study would tend to lessen the effects of altered receptor and/or G protein lateral diffusion on the rate MII-G t formation relative to what would be observed under physiological conditions. The comparison of 18:1 and 22:6 at the sn-2 position is a greater alteration than that produced by dietary n-3 fatty acid deficiency, which generally leads to replacement of 22:6n-3 with 22:5n-6 (40). Thus, the bilayer compositions examined in the present study exaggerate the differences in membrane composition produced by dietary n-3 deficiency. The increase in ͗͘ MII-G caused by exchanging 22:6 for 18:1 at the sn-2 position, corresponds to a 50% reduction in the rate of binding of G t to MII. The initial rising phase of the rod cell photoresponse is a parabolic function of time and a linear function of the rate of activation of phosphodiesterase catalytic subunits per unit time (5). Biochemical analysis shows that the rates of G t activation and phosphodiesterase catalytic subunit activation are approximately equal (5). Assuming that the rates of G t binding and G t activation are similar, this means that a reduction in the rate of MII-G t complex formation by 10% will delay the rod cell photoresponse by 5%. We report a 50% reduction in the rate of MII-G t complex formation, under conditions that minimize the effects of lateral diffusion relative to physiological conditions. These results indicate that the effect of membrane composition on the rates of rhodopsin and/or G t lateral diffusion would be sufficient to account for the 5% delay in a-wave implicit time activity observed in dietary n-3 deficiency. 2 Since the classical photoreaction cascade of Lumi 3 Meta I 3 Meta II was introduced by Wald and co-workers (7), a number of additional photointermediates on the microsecond to millisecond timescale have been proposed based on observed kinetic complexity (27,28), proton motions (41), effects of pH (42,43), and interaction with G t that requires GTP (44). Several investigations have determined that the full complexity of the transitions between photointermediates occurring on the MI-to-MII time scale become most apparent when examined at physiological temperature (27,45). However, no previous study has examined the interaction of G t with MII, or any other 380 nm-absorbing species at 37°C. The data and analysis presented in Fig. 2 provide strong evidence that the most rapidly forming photointermediate with an absorbance at 380 nm is part of the classical MI-MII equilibrium and binds G t . The measurements presented in this study do not address issues raised by other investigations such as the effect of pH, proton uptake, or possible effects of GTP on the MI-to-MII transition. All measurements were made in the absence of GTP, a condition that leads to the formation of a G t -MI b complex in measurements at Ϫ35°C (44). However, the measurements reported here for samples at 37°C showed no evidence for binding of G t to a state other than MII. The visual signal transduction pathway has been optimized by evolution to be one of the most rapid and highly amplified signal transduction pathways in the human body. The present results demonstrate that the lipid composition of the rod outer segment disc membrane makes a significant contribution to the optimal kinetic functioning of this prototypical G protein-coupled pathway. The excitable membranes of the nervous system all contain high levels of phospholipids with polyunsaturated acyl chains, thus it is expected that the results presented here will be applicable to G protein-coupled systems associated with neurotransmitters, taste, and olfaction and should be generally applicable to other members in the G protein-coupled superfamily.