Thrombin receptors define responsiveness of cholesterol-modified platelets.

The microviscosity of human platelet membranes was changed by incubating platelets with liposomes containing various ratios of cholesterol and lecithin. Binding of 125I-thrombin to the modified platelets was measured together with platelet aggregation and secretion. In cholesterol-normal platelets (mole ratio of cholesterol to phospholipid (C:PL) = 0.553; eta = 2.40 poise), weighted nonlinear least squares curve fitting indicated that a model involving two classes of sites was adequate to describe the binding isotherm (K1 = 8.3 X 10(8) M-1; R1 = 150 sites/platelet; K2 = 6.4 X 10(6) M-1; R2 = 16,000 sites/platelet). In cholesterol-enriched platelets (C:PL = 0.857; eta = 3.05 poise), the apparent affinities for the two classes of sites decreased to 55 and 53%, respectively, while the binding capacities increased to 170 and 160%, respectively. In contrast, in the cholesterol-depleted platelets (C:PL = 0.435; eta = 2.03 poise), the affinities increased to 220 and 180%, respectively, while the binding capacities decreased to 53 and 46%, respectively. In cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets, the thrombin concentrations required for half-maximal aggregation were 0.17, 0.35, and 0.52 nM, respectively, while the values for half-maximal secretion of [14C]serotonin were 0.17, 0.40, and 0.55 nM, respectively. Plots of receptor occupancy versus biological response showed that maximum response in cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets occurred with occupancy of 30, 50, and 70% of the high affinity sites, respectively. In all three treatment groups, occupancy of 40-50 high affinity sites results in 50% aggregation. These results show that (i) modification of platelet membrane microviscosity results in changes in the number and affinity of both high and low affinity thrombin receptors, (ii) the change in receptor number rather than affinity is the determinant for platelet responsiveness, and (iii) the changes in membrane microviscosity do not appear to alter the coupling between occupied receptor and subsequent bioresponse.

220 and 180%, respectively, while the binding capacities decreased to 53 and 46%, respectively. In cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets, the thrombin concentrations required for half-maximal aggregation were 0.17, 0.35, and 0.52 nM, respectively, while the values for half-maximal secretion of ['4C]serotonin were 0.17, 0.40, and 0.55 nM, respectively. Plots of receptor occupancy uersus biological response showed that maximum response in cholesterol-enriched, cholesterol-normal, and cholesteroldepleted platelets occurred with occupancy of 30, 50, and 70% of the high affinity sites, respectively. In all three treatment groups, occupancy of 40-50 high affinity sites results in 50% aggregation. These results show that (i) modification of platelet membrane microviscosity results in changes in the number and affinity of both high and low affinity thrombin receptors, (ii) the change in receptor number rather than affinity is the determinant for platelet responsiveness, and (iii) the changes in membrane microviscosity do not appear to alter the coupling between occupied receptor and subsequent bioresponse.
Elevated serum cholesterol is one of the most consistent risk factors for atherosclerosis and related vaso-occlusive disorders. Platelets from patients with hypercholesterolemia have an increased sensitivity to aggregating agents, especially * This work was supported by United States Public Health Service Grant HL 14697 and Biomedical Research Support Grant RR05737. This paper is Contribution 584 from the American Red Cross Blood Services Laboratories. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. epinephrine (1-3). A series of in vitro studies has sought to establish the basis for this hypersensitivity; when the choles-tero1:phospholipid ratio of platelets was elevated by exposure to cholesterol-rich liposomes, the platelets were found to have increased membrane microviscosity (4), increased sensitivity to epinephrine and ADP ( 5 ) , and increased basal levels of adenylate cyclase (6). Although thrombin-induced platelet aggregation was originally reported to be unaffected by changes in membrane cholesterol ( 5 ) , more recent studies, reported since the completion of our own work, have shown that increases in the cholestero1:phospholipid ratio result in increased platelet aggregation with thrombin and increases in the liberation of arachidonate (7,8) and the levels of thromboxane B2 secretion (9).
We have now studied the effect of changes in platelet membrane microviscosity on the number and affinity of receptors for thrombin with both increased and decreased ratios of cholestero1:phospholipid in comparison with cholesterolnormal platelets and have studied the bioresponse (both aggregation and secretion) in the three treatment groups. Our results show that the number of thrombin receptors increases with increasing membrane microviscosity while the affinity of binding decreases and that receptor number, rather than affinity, is the principal determinant of platelet responsiveness but that coupling between occupied receptor and bioresponse mechanism is unaffected by change in membrane microviscosity. Preliminary experiments leading to these studies have been published elsewhere (10).

EXPERIMENTAL PROCEDURES
Materiuls-L-a-Dipalmitoyllecithin, bovine serum albumin (essentially free of fatty acids), and 1,6-diphenyl-1,3,5-hexatriene were obtained from Sigma. Unesterified cholesterol (99%) was from Miles Laboratories. Garamycin (gentamycin sulfate) was from Schering-Plough Corp., Kenilworth, NJ. Human serum albumin USP (25% aqueous solution) was from Armour. Lecithin, 1,6-diphenyl-1,3,5hexatriene, and cholesterol were used without further purification. All other chemicals used were of reagent grade. Iodogen was obtained from Pierce Chemical Co. Human a-thrombin was purified by the procedure of Fenton et ul. (11) and had a specific activity in the clotting assay of 3400 NIH units/mg. Platelet Preparution-Fresh human blood anticoagulated with citrate/phosphate/dextrose/adenine was obtained from the Washington Region, American Red Cross Blood Services from volunteer donors. Platelet-rich plasma was obtained from 1 unit of whole blood by centrifugation at 3200 X g for 3.5 min at 22 "C. Platelet-rich plasma was transferred to the satellite bag and spun a t 3200 X g for 10 min to sediment the platelets and to obtain platelet concentrate, while the supernatant plasma was removed and used as platelet-poor plasma.
For the preparation of washed platelets, the pH of the platelet-rich plasma was adjusted to 6.5 with citric acid (150 mM), platelets were sedimented by centrifugation at 1600 X g for 10-12 min, and the pellet was resuspended in platelet wash buffer (11.0 mM dextrose, 128 mM NaCI, 4.26 mM NazP04, 7.46 mM NaHP04, 4.77 mM trisodium citrate, and 2.35 mM citric acid, pH 6.5) supplemented with 0.35% bovine serum albumin. Contaminating erythrocytes were removed by further centrifugation at 800 X g for 15 s. The platelet suspension was then centrifuged at 800 X g for 5 min, and the pellet was suspended and washed twice with platelet wash buffer supplemented with bovine serum albumin. Finally the pellet was suspended in modified Tyrode's buffer (136 mM NaCl, 5.5 mM dextrose, 2.7 mM KC1, 0.07 mM Na2HP04, 0.01 mM NaHC03). Platelets were counted FL) . electronically in a Coulter counter (Coulter Electronics Inc., Hialeah, Preparation of Liposome-Cholesterol-and lecithin-containing liposomes were prepared as described by Shattil and Cooper (4) in modified Tyrode's buffer. Weight ratios (milligram/mg) of cholesteroldecithin used to prepare lipid dispersions (liposomes) were 80:40 for cholesterol-rich, 23:40 for cholesterol-normal, and 0:40 for cholesterol-poor. After sonication for 1 h (Branson sonifier, model 350, setting 6), these dispersions were centrifuged at 500-600 X g for 10 min to remove undispersed lipid. The supernatant suspension was stored at 4 "C and was used within 1 week. Immediately before use, the dispersions were made 2.5 g/100 ml with respect to human serum albumin, incubated for 20-30 min at 37 "C, and centrifuged at 21,000 X g for 30 min to sediment any aggregated lipid.
Modification of Membrane Lipids-The following procedures were performed under sterile conditions. One unit of platelet concentrate was adjusted with platelet-poor plasma to a platelet concentration of 1-1.5 X 109/ml. Ten-milliliter portions of this diluted platelet concentrate were transferred to three separate 150-ml blood transfer packs and incubated with equal volumes (10 ml) of each lipid dispersion in Tyrode's buffer containing 2.5 g/100 ml human serum albumin, pH 7.2. Garamycin (30 pg/ml-final concentration) was added, and the mixtures were incubated for 18-22 h a t 22 "C in an incubator equipped with a horizontal shaking device (Forma Scientific platelet incubator) of the type used for storage of platelets for clinical use (12). This incubation resulted in an insignificant increase in the pH (0.05-0.1 pH unit) of the platelet suspension. The pH of the incubated platelet suspensions was adjusted to 6.5 with citric acid, and the platelets were pelleted by centrifugation at 1600 X g for 12 min at 22 "C, resuspended in platelet wash buffer, washed twice with platelet wash buffer, and finally suspended in Tyrode's buffer or thrombin binding buffer as required.
A crossover experiment was carried out to establish that changes in platelet responsiveness and in binding data were not due to differing protective effects of the three liposome populations. In this experiment, platelets originally rendered cholesterol-enriched or cholesterol-depleted by incubation with liposomes for 5 h at 37 "C (4) were further incubated for 18 h at 22 "C with cholesterol-depleted and cholesterol-enriched liposomes, respectively.
Platelet Lipid Analysis-Thrice-washed platelets in platelet wash buffer were extracted by shaking with 15-20 volumes of chloroform:methanol (2:l) for 3-4 h. The organic layer was washed twice with one-tenth volume of 0.15 M KC1 in 50% methanol. The organic phase was evaporated under a stream of dry nitrogen at 40 "C. The contents of cholesterol and lipid phosphorus were measured in the dried residue by the methods of Zlatkis et al. (13) and Chen et al. (14), respectively.
Steady state fluorescence polarization intensity was measured with a Perkin-Elmer fluorescence spectrophotometer (MFP-44B) equipped with polarizers, an automatic polarizer (emission) flipping device, a thermostatted cell holder, and a variable speed stirring device. The excitation and emission monochromators were fixed at 355 and 429 nm, respectively. The temperature of the platelet SUSpension was measured with k O . 1 "C accuracy using a digital thermo-Brook, NJ). couple thermometer, model BAT-12 (Bailey Instruments Inc., Saddle In preliminary studies, it was found that platelet concentrations higher than 2 x lO'/rnl gave erroneous P values. However, concentrations between 1.5 and 0.7 X 108/ml gave constant values of P, ensuring that observed changes in fluorescence depolarization were not due to light scattering. For this reason, all experiments relating to the determination of fluorescence polarization were carried out at a platelet concentration of 1.5 X 10' ml. The output signals from platelets labeled with 1,6-diphenyl-l,3,5,hexatriene were more than 50-fold greater than those obtained with unlabeled platelets.
Aggregation and Release-Platelet aggregation was measured in an aggregometer at 37 "C with constant stirring at a platelet concentration of 2.5-3.0 X 108/ml in a volume of 450 pl. Thrombin was added in a total volume of 50 p1 to various final concentrations as specified.
To determine the extent of release, modified platelets in the mixture of platelet-rich plasma and liposomes were first incubated with ["C] serotonin (1 p M final concentration), washed two times with platelet wash buffer, and then resuspended in Tyrode's buffer. Following the induction of secretion by addition of various concentrations of thrombin as described under "Results," the reaction was stopped by the addition of 50 p1 of 3% formaldehyde and 50 mM EGTA,' pH 6.5, followed by rapid sedimentation of platelets at 15,000 x g for 1 min.
[14C]Serotonin present in the supernatant solution was counted in an Intertechnique liquid scintillation counter (model SL 4000) at an efficiency of 90% for "C.
Thrombin Binding Assays-1251-Thrombin was prepared using an Iodogen procedure in which 50 pg of human a-thrombin were combined with 65 p1 of buffer (0.1 M NaCI, 50 mM sodium phosphate, pH 7.0, and 10 mM benzamidine) and 2 mCi of Na'"I in a test tube coated with 5 pg of Iodogen. The iodination was performed at 4 "C and was terminated when the lZ5I incorporated into protein was 0.4-0.8 (0.3-0.6 molecules of '251/thrombin molecule; specific activity = 16-32 &/fig). The iodinatedprotein was separated from NaIz5I on a column of Sephadex G-25 using 50 mM sodium phosphate buffer, pH 7.0, at high ionic strength (0.75 M NaCI). The 1251-thrombin had 100% of the clotting activity of the unlabeled thrombin and was a single band on both reduced and nonreduced 10% gels as described by Laemmli (17). '2SI-Thrombin was stored at -70 "C in the presence of 8% bovine serum albumin. Under these storage conditions, the iodinated thrombin was stable for approximately 30 days as measured by radioactivity precipitable by 5% trichloroacetic acid and by the radioreceptor assay.
The thrombin binding assay was performed for 10 min at room temperature in a total volume of 0.1 ml. The final concentrations of reactants were 1251-thrombin, 0.1 nM; bovine serum albumin, 10 mg/ ml; and platelets/ml, 2-5 X 10' . All reactants were prepared in 136 mM sodium chloride, 0.6% polyethylene glycol 6000, and 25 mM Tris, pH 7.4. Unlabeled thrombin was added in increasing concentrations up to 1 PM as indicated. Cell-bound thrombin was separated from free thrombin by a centrifugation for 2 min in a Beckman microfuge. The supernatant (free thrombin) was removed, the bottom of the microfuge tube containing the cell pellet (bound thrombin) was cut off, and radioactivity was determined in an LKB 80,000 y-counter. Nonspecific thrombin binding was defined as the radioactivity associated with the cell pellet in the presence of unlabeled thrombin at 1 pM.
The integrity of the '251-thrombin was measured on the supernatant by determining the amount of radioiodine soluble in 5% trichloroacetic acid. Thrombin degradation was less than 5% in the course of the experiment. Under these conditions, specific binding reached a steady state and was linear with respect to cell concentration. Data from binding experiments were analyzed by weighted nonlinear least squares curve fitting, using total ligand concentration as the independent variable and bound ligand concentration as the dependent variable, assuming a constant percentage error in the concentration of bound ligand (18). Results were displayed graphically as competition curves or as Scatchard plots (19). Objective statistical criteria (F test, extra sum of squares principle) were used to evaluate goodness of fit and for discriminating between different models. Curves from multiple experiments were analyzed both individually and simultaneously using constrained curve fitting to obtain improved precision of parameter estimates (18). Nonspecific binding was treated as a parameter subject to error and was fit simultaneously with other parameters.  (4) and others (7)(8)(9) which involved incubation at 37 "C for 5 h with frequent inversion. For this reason, the extent of incorporation was determined by cholesterol and phospholipid analysis and changes in fluorescence polarization. The results obtained (Table I) show values for C:PL ratios and for microviscosity similar to those previously reported for cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets (4). The values obtained for platelets incubated with cholesterolnormal liposomes were identical with those obtained with unmodified platelets incubated with Tyrode's buffer under similar conditions showing that incubation, as such, did not affect the C:PL ratio. Platelet Aggregation and Secretion-In order to determine platelet responsiveness by aggregation and secretion, the three classes of cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets were subjected to aggregation by titrating them with increasing levels of thrombin from 0-1 nM (0-100 milliunits/ml). As shown in Fig. lA, a clear-  cholesterol-normal platelets a t 0.60 nM (60 milliunits/ml), and with cholesterol-depleted platelets at 0.75 nM (75 milliunits/ml). Values for half-maximal aggregation from the dose response curves of Fig. 1 were 0.17, 0.35, and 0.53 nM. While there were minor shifts in values between the 8 and 10 individual platelet preparations examined in this way, this general pattern was observed in all cases.

Membrane
Similar results were observed when the release of serotonin was examined at different thrombin concentrations in the three platelet populations. The maximum release for cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets was achieved at thrombin concentrations of 0.50, 0.70, and 1.0 nM, respectively, and the same type of differential curves were obtained as were seen in the case of the aggregation response (Fig. 1B).
Thrombin Binding Studies-The thrombin binding data could be fit successfully using a model involving two independent classes of binding sites for all experiments and for each of the three platelet treatment groups, but they could not be fit satisfactorily using a one-site model. All of the data from five separate binding studies for each of the three separate treatment groups of cholesterol-enriched, cholesterolnormal, and cholesterol-depeleted platelets were submitted to computer curve fitting ( Fig. 2 and Table 11). The shapes of the Scatchard plots for each of the three platelet treatment groups differ significantly. For cholesterol-normal platelets, the association constant for high affinity sites (8. Since the results of the binding studies indicated that both the affinity of the receptors and the number of receptors change upon modification of membrane cholesterol, we examined the relationship between thrombin receptor OCCUpancy and biological response, that is, thrombin-induced aggregation. Results of this analysis are shown in Fig. 3 for the high affinity sites. Cholesterol-enriched platelets were more responsive to the thrombin, as indicated by the leftward shift in the curve, than were cholesterol-normal platelets which were, in turn, more responsive than cholesterol-depleted platelets. The maximum response in cholesterol-enriched platelets occurred with 30% occupancy of the high affinity receptors; the maximum response in cholesterol-normal platelets occurred with 50% occupancy of the high affinity receptors; while the maximum response in cholesterol-depleted platelets occurred with 70% occupancy of the high affinity  Receptor numbers are given as sites/platelet. Errors are the standard errors obtained from LIGAND pooling over five experiments with a total of 65 data points (each a mean of triplicates) for each treatment group. Nonspecific binding is the bound to free ratio for lZ5I-thrombin in the presence of M unlabeled thrombin. receptors. Thus, the change in receptor number appears to be the determinant for platelet responsiveness.
Since biological response is related to the number of occupied receptors (23), we calculated the number of occupied high affinity receptors necessary to elicit 50% aggregation. In the three treatment groups, occupancy of 40-50 high affinity receptors resulted in 50% aggregation; specific values for cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets were 36, 50 and 48, respectively. Therefore, modification of platelet membrane microviscosity by changing the C:PL ratio appears to alter the number of thrombin receptor sites which in turn alters the responsiveness of the platelet, but this does not appear to alter the coupling between the occupied high affinity receptor and the bioresponse mechanism.
When similar calculations were made for all receptors, that is, both high and low affinity, the number of occupied receptors necessary to elicit 50% aggregation for the cholesterolenriched, cholesterol-normal, and cholesterol-depleted were 156, 112, and 95, respectively. This relationship is contrary to the observed dependence of platelet responsiveness on membrane microviscosity and suggests that the low affinity sites are not involved in this response.
Crossouer Experiment-The results of the crossover exper-  iment are shown in Table 111. When platelets were first depleted of cholesterol and then enriched, the C:PL ratio was raised to the value normally found in platelets directly enriched in cholesterol. The thrombin concentration for halfmaximal aggregation and the affinities and numbers of high and low affinity sites also were in the same range as the values in the directly enriched platelets.
In the case of the cholesterol-normal platelets, the C:PL ratio was unchanged after further incubation with cholesterolnormal liposomes although there was a slight increase in the thrombin concentration required for half-maximal aggregation. High affinity binding was in the expected range although the number and affinity of low affinity sites were decreased. When platelets were first enriched with cholesterol and then depleted by a further 18-h incubation, the C:PL ratio remained high, suggesting that the loss of cholesterol from the enriched platelets was relatively slow. Both high and low affinity receptors also gave values similar to those of platelets with high C:PL ratios. The increase in the concentration of thrombin required for 50% aggregation following crossover may reflect a generalized inhibition of platelet responsiveness due to the prolonged incubation.

DISCUSSION
The present study clearly indicates the advantages of an objective computer-assisted statistical analysis of complex ligand binding systems. Casual inspection of the Scatchard plots (Fig. 2) fails to convey the necessary information because it is impossible to view the entire binding isotherm (over 5 orders of magnitude) on one graph. Examination of the computer-generated curves indicates a consistent effect of either cholesterol enrichment or cholesterol depletion relative to cholesterol-normal platelets. Analysis using LIGAND permitted us to obtain estimates for the affinity constants and receptor number for both the high and low affinity sites, for nonspecific binding in each experiment, as well as the best scaling factor to correct for variations in platelet number and in density of receptors/platelet. The logarithmic mean of the affinity constants and receptor numbers showed a statistically significant difference between the three treatment groups. This was confirmed by simultaneously fitting multiple experiments within a treatment group giving rise to more precise estimates of parameters because of the increased number of observations relative to the number of fitted parameters. Although graphical analysis of the Scatchard plots (Fig. 2) might suggest gross similarities of shape, this hypothesis could be unequivocally rejected ( P < 0.001), fitting all curves from the three treatment groups simultaneously with the same affinity constants ( K1 and K2) and the same ratio of receptor number (It1/&) resulted in a drastic increase of the sum of square and appearance of severe nonrandomness of residuals.
These changes in surface expression of thrombin receptors were accompanied by differences in thrombin-induced aggregation and release between the three platelet treatment groups. Thus, the thrombin concentrations required for halfmaximal aggregation in cholesterol-enriched, cholesterol-normal, and cholesterol-depleted platelets were 0.17, 0.35, and 0.52 nM, respectively. These results confirm and extend the observations of Kramer et al. (7) who showed that cholesterolenriched platelets exhibit greater aggregation than cholesterol-normal platelets at thrombin concentrations in the range 2-10 nM.
The differences we are observing are at approximately 1 order of magnitude lower concentrations of thrombin and are observed for both cholesterol-enriched and cholesterol-depleted platelets. The differences observed are in the range of the high affinity association constants for the thrombin receptor, and no differences were observed between the three platelet treatment groups at higher concentrations of thrombin. We have previously observed that differences in reactivity between normal and Bernard-Soulier platelets are most readily detected at low thrombin concentrations (-0.3 nM) (23). Thus, titration of the aggregation response of platelets with low concentrations of thrombin may have value in detecting a variety of clinical platelet defects.
The crossover experiment designed to determine differential effects of the three liposome populations was not entirely satisfactory because of the additional periods of incubation and increased handling of all samples. However, the results with the platelets which were first depleted and then enriched are clear with respect to C:PL ratio, high and low affinity binding, and thrombin sensitivity. These experiments show that the observed differences are not due to differential effects of the liposome preparations.
Similar results to those observed for thrombin-induced aggregation were obtained for the thrombin concentrations required for half-maximal release of serotonin in the three platelet treatment groups at values of 0.17,0.40, and 0.55 nM, respectively. These results show an exact parallel between aggregation and release in the thrombin-induced stimulation of platelets.
Cholesterol affects membrane microviscosity by changing lipid-lipid interactions and affecting the packing density within the membrane (16). These changes in membrane microviscosity can influence expression of receptors by two different mechanisms. First, changes in membrane fluidity could affect the association of receptors to form oligomeric groups; second, changes in fluidity may cause passive modulation resulting from vertical deflection of receptors in relation to the plane of the membrane. At present, we cannot differentiate between thse two possible mechanisms with regard to platelets although both receptor number and affinity are affected in an inverse relationship.
Interestingly, similar trends with regard to fluidity and binding have been observed in the interaction of ['4C]serotonin with mouse brain membranes where increases in receptor number a t higher membrane microviscosities were associated with decreases in receptor affinity (24). In this case, the changes were interpreted as possibly being due to passive modulation by vertical displacement of the high and low affinity sites along the vertical axis of the receptor molecule. Whether similar relationships exist for high and low affinity receptors for thrombin in platelet membranes or whether the observed changes are due to combined effects on receptor association and vertical displacement remain to be determined.
The mechanism of the interaction of platelets with thrombin remains unresolved. We have proposed that thrombin binds first to a platelet receptor which, in a second step, interacts with a platelet effector leading to platelet aggregation and release (25). The alternative hypothesis suggests that platelet activation is entirely due to the proteolytic activity of thrombin and that thrombin binding, as such, is not a determinant of platelet responsiveness (26). Although it is possible that changes in membrane fluidity could affect the susceptibility of the proteolytic substrate to thrombin proteolysis, the present results show that platelet response is directly proportional to the amount of thrombin bound for each of the three platelet treatment groups examined.
The present work on dose response to thrombin suggests that half-maximal aggregation and secretion occur when the same number (40-50) of high affinity receptors are occupied in each of the three treatment groups, but that progressively higher concentrations of thrombin would be required to occupy this number in the cholesterol-enriched, cholesterolnormal, and cholesterol-depleted populations. The extensive metabolic changes observed in cholesterol-modified platelets in previous work (5-9) have been carried out a t thrombin concentrations far above the threshold thrombin concentra-tions reported here and may represent secondary changes due to elevated levels of thrombin rather than the primary effects defined by effects on receptor expression described here.