Regulation by High Density Lipoproteins of Muscarinic Acetylcholine Receptor Function in Chick Heart Cells Cultured in Defined Medium* of

Activation cardiac muscarinic acetylcholine receptors (mAChR) on cultured chick heart cells results in a decrease in cellular cAMP levels and a stimulation of phosphoinositide breakdown. A serum-free culture system has been used to investigate the regulation of mAChR number and function by purified serum high density lipoprotein (HDL). Administration of HDL pu- rified from rooster serum to chick heart cells cultured in defined medium results in an attenuation of the ability of muscarinic agonist to inhibit forskolin-stim-ulated cAMP accumulation, with no change in its abil- ity to stimulate phosphoinositide hydrolysis or to mediate down-regulation of receptor number. The inclu- sion of HDL in the culture medium did not result in appreciable changes in mAChR number or affinity, nor were the levels of the inhibitory guanine nucleotide- binding regulatory proteins (G-proteins) altered. How-ever, the ability of guanine nucleotides to inhibit for- skolin-stimulated adenylate cyclase activity was reduced by HDL treatment, suggesting that HDL inter- feres with the capacity of G-proteins to interact with adenylate cyclase. In order to determine which component of native HDL mediates the decreased effec- tiveness of carbachol, the ability of lipid and apoprotein fractions to mimic the effect of HDL was tested. HDL lipid fractions were able to mimic the effect of native HDL, while protein fractions were not. This result suggests that the ability of

Activation of cardiac muscarinic acetylcholine receptors (mAChR) on cultured chick heart cells results in a decrease in cellular cAMP levels and a stimulation of phosphoinositide breakdown. A serum-free culture system has been used to investigate the regulation of mAChR number and function by purified serum high density lipoprotein (HDL). Administration of HDL purified from rooster serum to chick heart cells cultured in defined medium results in an attenuation of the ability of muscarinic agonist to inhibit forskolin-stimulated cAMP accumulation, with no change in its ability to stimulate phosphoinositide hydrolysis or to mediate down-regulation of receptor number. The inclusion of HDL in the culture medium did not result in appreciable changes in mAChR number or affinity, nor were the levels of the inhibitory guanine nucleotidebinding regulatory proteins (G-proteins) altered. However, the ability of guanine nucleotides to inhibit forskolin-stimulated adenylate cyclase activity was reduced by HDL treatment, suggesting that HDL interferes with the capacity of G-proteins to interact with adenylate cyclase. In order to determine which component of native HDL mediates the decreased effectiveness of carbachol, the ability of lipid and apoprotein fractions to mimic the effect of HDL was tested. HDL lipid fractions were able to mimic the effect of native HDL, while protein fractions were not. This result suggests that the ability of HDL to attenuate muscarinic receptor function is mediated by its lipid constituents. The effect of HDL and HDL lipid fractions were not correlated with changes in membrane cholesterol content.
Cardiac muscarinic acetylcholine receptors (mAChR)' mediate the decreased rate and force of contraction observed in response to stimulation of the vagus nerve. The decrease in *This work was supported in part by a grant-in-aid from the American Heart Association and by Grants HL30639 and GM07270 from the National Institutes of Health. 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 U.S.C. Section 1734 solely to indicate this fact.
$ Established Investigator of the American Heart Association. To whom correspondence should be addressed. The abbreviations used are: mAChR, muscarinic acetylcholine receptor; HDL, high density lipoprotein; G-protein, guanine nucleotide-binding regulatory protein; Gi, G-protein which inhibits adenylate cyclase; G,, G-protein which stimulates adenylate cyclase; Go, G-protein which neither stimulates nor inhibits adenylate cyclase; NMS, N-methylscopolamine; QNB, quinuclidinyl benzilate; IT& serumfree defined culture medium; SDS, sodium dodecyl sulfate; LDL, low density lipoprotein; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid Ins-Ps, total inositol phosphates; GppNHp, guanylyl-5'-imidodiphosphate. heart rate following mAChR activation is due in part to the opening of a receptor-gated inward rectifying potassium channel, while the decrease in cardiac contractility arises from diminished calcium entry (Loffelholz and Pappano, 1985). Biochemical consequences of mAChR activation include an increase in cellular cGMP, a decrease in cellular CAMP, and an increase in phosphoinositide hydrolysis (reviewed in Nathanson, 1987). Receptor-mediated changes in cellular cAMP and cGMP have been implicated in the mediation of the negative inotropic response caused by muscarinic agonists Hartzell and Fischmeister, 1986). Stimulation of phosphoinositide hydrolysis results in the production of additional second messengers, including diacylglycerol, which activates protein kinase C, and inositol trisphosphate, which causes the release of intracellular calcium (Berridge and Irvine, 1984). The mAChR produces these biochemical effects via interactions with the guanine nucleotide regulatory proteins (G-proteins) Gi, Go, and the as yet unidentified G-protein which couples to phospholipase C (Florio and Sternweis, 1985;Hepler and Harden, 1986;Martin et al., 1988).
Chick heart cells represent a convenient model system in which to study the regulation of cardiac mAChR function because they are easily cultured, because they readily respond to muscarinic agonists, and because their environment can be controlled. We have previously demonstrated that chick heart cells can be cultured in fully defined, serum-free medium with retention of mAChR-mediated responses (Subers and Nathanson, 1988). Thus, cells cultured in defined medium respond to the agonist carbachol with an inhibition of cAMP accumulation, a stimulation of phosphoinositide metabolism, and a down-regulation of mAChR number. This culture system allows for the investigation of the role of serum components in the regulation of mAChR function, without the complications arising from the use of undefined serum-supplemented culture medium. The sensitivity to the muscarinic agonist carbachol can vary over two orders of magnitude depending on the serum lot used to supplement the medium in which the cells are cultured Nathanson, 1985, 1986) indicating that serum constituents may regulate mAChR function. In addition, cells cultured in serum-free defined medium tend to show a greater inhibition of cAMP accumulation and a greater stimulation of phosphoinositide metabolism in response to mAChR activation than do cells cultured in the presence of 5% fetal calf serum (Subers and . More specifically, Renaud et al. (1982) demonstrated that chick ventricular cells grown in fetal calf serum depleted of lipoproteins responded to the muscarinic agonist oxotremorine with a decrease in beating rate, while those cultured in medium supplemented with complete fetal calf serum did not. Haigh et al. (1988) reported that the increased negative chronotropic response of chick atrial cells cultured in lipoprotein-deficient serum could be reversed by 19685 the addition of bovine LDL. These results suggested that serum lipoproteins may decrease mAChR responsiveness in cultured chick heart cells. In order to test the role of serum lipoproteins on cardiac mAChR function in a defined, homologous system, we determined the effect of purified chicken lipoproteins on mAChR number and function in chick heart cells cultured in serum-free defined medium, using biochemical assays, which are more defined and specific than physiological measurements such as beating rate responses. In this report we demonstrate that serum high density lipoprotein (HDL) decreases the ability of carbachol to inhibit forskolinstimulated cAMP accumulation, without affecting its ability to stimulate phosphoinositide metabolism or agonist-mediated loss of cell surface receptors. HDL-mediated inhibition of mAChR responsiveness appears to result from a diminished ability of Gi to couple with adenylate cyclase in membranes prepared from HDL-treated cells. This effect of HDL was mimicked by the lipid fraction of native HDL and not by the apoprotein fraction, suggesting that alterations in membrane lipid composition may mediate the decrease in mAChR responsiveness observed after HDL treatment.

EXPERIMENTAL PROCEDURES
Materials-White leghorn chicken eggs were obtained from College Biological Supply (Bothell, WA) and maintained in a humidified 38 "C incubator until the ninth incubation day. White leghorn roosters, age 20-22 weeks, were obtained from H & N International (Redmond, WA), and fed a standard laboratory diet ad libitum, until the day before use at which time they were restricted to water only. Tissue culture materials were obtained as described previously (Subers and Nathanson, 1988 (1987). Antisera AS7 and RV3 were a generous gift of Dr. Allen Spiegel (National Institutes of Health). All other materials were purchased as described previously (Halvorsen and Nathanson, 1984;Luetje et al., 1987).
Cell Culture-Chick heart cells were cultured in serum-free defined medium from 9-day embryonic chickens as described previously (Subers and Nathanson, 1988). The medium, based on that of Libby (1984) and referred to as was M199, supplemented with penicillin-streptomycin (100 units/ml and 100 pg/ml final concentrations, respectively), insulin (5 pglml), transferrin (5 pg/ml), sodium selenite (5 pg/ml), testosterone (10 nM), and triiodothyronine (3 nM). The medium was changed on the third day in culture, and experiments were performed on the fourth. In general, four dozen hearts were dissected, dissociated, and plated on either eight 100-mm plates for assays requiring membrane preparations, twenty-four 60-mm plates for cAMP accumulation assays, or sixty to seventy 35-mm plates for phosphoinositide and intact cell binding assays.
Purification of Lipoproteins-Roosters were anesthetized with 2-3 ml of pentobarbital and 30-50 ml of blood was drawn from the jugular vein. Serum was prepared from clotted blood, and high density lipoproteins were purified from serum by flotation on KBr-NaCl gradients by the method of Chapman (1981) as described by Hermier et al. (1985). Fractions containing HDL or LDL were identified by SDS-polyacrylamide gel electrophoresis, pooled, dialyzed against phosphate-buffered saline (PBSA, 137 mM NaCl, 2.7 mM KCl, 8 mM Na2HP0,, 1.5 mM KHzPO,, pH 7.41, and sterilized by filtration through a 0.2-pm nitrocellulose filter prior to use. Purity was determined by silver-stained SDS-polyacrylamide gel electrophoresis as shown in Fig. 1. SDS-polyacrylamide gel electrophoresis was performed by the method of Laemmli (1970) and silver staining by the method of Heukeshoven and Dernick (1985). Pooled fractions were stored at 4 "C for up to 6-8 weeks. Lipid subfractions of HDL were prepared by extraction in chloroform/methanol/phosphate-buffered saline (PBSA containing 1 mM CaCl, and 0.5 mM MgC12) as described by Folch (1957). Apoprotein subfractions were prepared by extraction of HDL in chloroform/diethyl ether as described by Scanu and Edelstein (1971).
Preparation of Membrane Homogenutes-Plates were washed three times in phosphate-buffered saline at 37 "C to remove the growth medium, and cells were scraped with a rubber policeman in 2 ml of the desired homogenization buffer at 4 "C (see individual methods for binding, adenylate cyclase immunoblots, and cholesterol assays, below). Cells were homogenized by 15-20 strokes of a glass-glass homogenizer, and the homogenates spun at 18,000 X g for 15 min. The crude membrane pellet was washed once by centrifugation and resuspended in the appropriate assay buffer (see individual methods, below).
mAChR Biding Assays-Membrane and intact cell binding assays were performed as described by Nathanson (1983). For membrane binding assays, cells were homogenized in 50 mM NaPO, at 4 "C, pH 7.4, membranes prepared (see above), and incubated with the indicated concentrations of [3H]QNB in a total volume of 1 ml of 50 mM NaPO,, pH 7.4, for 90 min. Dithiothreitol (0.1 mM) was included in the homogenization and resuspension buffer for membranes used in assays investigating the effect of GppNHp on carbachol binding (Halvorsen and Nathanson, 1984). Reactions were started with the addition of membranes and terminated by dilution with ice-cold 50 mM NaP04, followed by filtration through Whatman GF/C filters. All assays were performed in duplicate and nonspecific binding was defined as that remaining in the presence of 1 ~L M atropine. For intact cell binding assays, culture plates were rinsed three times with assay medium (116 mM NaCl, 1.8 mM CaCl2, 2.7 mM KCl, 0.81 mM MgSO,, 1 mM Na2HP0,, 5.56 mM glucose, and 25 mM HEPES, pH 7.4) at 37 "C, and allowed to equilibrate in this medium for 15 min prior to the addition of ligands. Plates were incubated at 37 'C with the indicated concentrations of [3H]NMS for 1 h. Incubations were stopped by rapid rinsing with assay medium at 4 "C followed by the addition of 0.5 ml of 1% Triton X-100 in phosphate-buffered saline. After 10-15 min, the detergent solution was removed and radioactivity determined by liquid scintillation counting. Nonspecific binding was defined as that occurring in the presence of 1 pM atropine. Total binding was measured in triplicate plates while nonspecific binding was measured in duplicate.
CAMP Accumulation Assay-CAMP levels were determined as described previously (Nathanson et al., 1978). Plates were rinsed three times with Earle's salts (116 mM NaCl, 1.8 mM CaC12, 1 rnM NaH,PO,, 5.4 mM KC1, 0.81 mM MgSO,, 5.56 mM glucose, 25 mM HEPES, pH 7.4) at 37 "C, and preincubated 20 min in M199 containing 25 mM HEPES, pH 7.4, and 5 mM theophylline. Plates were then incubated 5 min in 100 p~ forskolin, to stimulate cAMP formation, and the indicated concentrations of carbachol. Incubations were stopped by rinsing once with Earle's salts at 4 "C, immediately followed by the addition of 2 ml of ice-cold 5% trichloroacetic acid. cAMP was isolated by ion exchange chromatography according to Matsuzawa and Nirenberg (1975) and quantitated by the competitive protein binding method of Gilman (1970).
Phosphoinositide Hydrolysis Assays-Cells were incubated overnight in 1 ml of growth medium containing 1 pCi of [3H]myo-inositol on the third culture day, and experiments were performed the following day as described previously (Subers and Nathanson, 1988). Briefly, cells were rinsed three times with physiological saline solution (118 mM NaCl, 4.7 mM KCl, 3 mM CaC12, 1.2 mM KHJ'Or, 10 mM glucose, 0.5 mM EDTA, 20 mM HEPES, pH 7.4) at 37 "C, preincubated with the same containing 10 mM LiCl for 30 min, treated as indicated for an additional 15 min, and the inositol phosphates (Ins-Ps) produced were isolated and quantitated as described by Masters et al. (1984).
Quantitative Immunoblot Assays-Culture plates were rinsed, scraped, and homogenized on ice with 20 mM NaP04, 150 mM NaCl containing 0.4 mM phenylmethanesulfonyl fluoride, 1 mM lJ0-phenanthroline, 1 mM iodoacetamide, and 1 pM pepstatin A. Membranes were prepared as described above and stored at -70 "C until use. Aliquots of membrane samples were pelleted in a microcentrifuge, resuspended in SDS sample buffer, and run on 9% SDS-polyacrylamide gels according to the method of Laemmli (1970). G-protein standards were run in duplicate on the same gels as were triplicate samples of membranes from control and treated cells. Proteins were then electrophoretically transferred to nitrocellulose by the method of Towbin et al. (1979). The nitrocellulose was allowed to dry, and then either stained with Amido Black to determine the total protein present, or incubated in 10% bovine hemoglobin for 1 h. Hemoglobinblocked blots were rinsed with distilled water and incubated overnight at room temperature with affinity purified primary antisera diluted in 20 mM NaP04, 150 mM NaCl, and 0.5% Tween 20 (TPBS). Blots were rinsed three times with TPBS and incubated 2 h at room temperature with alkaline phosphatase-conjugated goat anti-rabbit IgG. Bound antibody was visualized as described by Smith and Fisher (1984) and quantitation of stained bands was performed by the method of Luetje et al. (1987). Briefly, immunostained bands were cut from the nitrocellulose, placed in microtiter wells, and incubated with 100 pl of 1 M Tris-HC1, pH 8.0, containing 0.2 mg/ml pnitrophenyl phosphate. After 10-20 min, color development was stopped by the addition of 100 pl of K2HP04, and the optical density determined at 410 nm. By comparing the optical density of known amounts of G-protein subunit standards to that of the samples, subjected to electrophoresis on the same gel, the amount of G-protein subunit in the sample could be determined.
Membrane Cholesterol Assays-Membranes were prepared in 50 mM NaP04 buffer as described above and stored at -20 "C until the day of use. Cholesterol content was determined by the cholesterol oxidase method as described by Oram (1986).

RESULTS
Purification of HDL-Lipoprotein fractions containing apoprotein A-I, the major apoprotein in HDL, were identified by SDS-gel electrophoresis and pooled to yield the final HDL or LDL preparation. The apoprotein content of these preparations was then analyzed by silver-stained SDS-polyacrylamide gel electrophoresis as shown in Fig. 1. Panel A demonstrates that the major HDL apoprotein band migrates at M , 24,300-25,700 on a 15% gel, agreeing with the reported MI of rooster apoA-I of 27,000-28,000 obtained in a different gel system (Hermier et al., 1985). The density of the fractions pooled for the final HDL preparation ranged from 1.08 to 1.17 g/ml.
Minor bands are observed at MI 14,000-15,000, reported previously to be apolipoproteins present in HDL fractions (Hermier et al., 1985). Faint bands are seen at M , > 60,000 but are likely to be contaminants in the gel buffers as they appear in many gel lanes, regardless of the sample content. These results indicate that HDL was purified to near homogeneity from rooster serum. Panel B, lune 1, shows the typical band profile for pooled LDL fractions, which ranged in density from 1.04 to 1.06 g/ml. The major LDL apoprotein is apoprotein B-100 (arrow), which has a reported MI of 410,000 (Hermier et al., 1985). HDL apoprotein A-I is present in the pooled LDL fractions. Lane 2 shows that the HDL preparations are free of contaminating LDL, as no apoprotein B-100 is observed in the HDL lane. LDL preparations devoid of contaminating HDL could not be obtained using this purification scheme.
Effect of HDL on mAChR-mediated Responses- Renaud et al. (1982) observed that chick heart cells cultured in the absence of lipoproteins responded to the agonist oxotremorine with a decrease in beating rate, while cells cultured in their presence did not. This result suggests that serum lipoproteins can inhibit mAChR-mediated responses. More recently, Haigh et al. (1988) demonstrated that LDL derived from fetal  calf serum could reverse the increased responsiveness of chick heart cells cultured in lipoprotein-deficient serum. We wanted to determine whether purified serum lipoproteins could alter mAChR responsiveness in a defined culture system that was amenable to biochemical manipulation. Purified LDL or HDL was added to defined medium and the ability of the agonist carbachol to inhibit forskolin-stimulated cAMP accumulation was tested. The final concentrations of LDL and HDL tested, 13 pg/ml and 25-26 pg/ml, respectively, were chosen because these would be equivalent to the final concentrations which would be attained if the cells were cultured in medium supplemented with 5% rooster serum. The results, shown in Fig.  2, indicate that both serum HDL and LDL inhibit the maximal response to carbachol and increase the concentration necessary for half-maximal inhibition (EC50). Because LDL was somewhat less effective than HDL in attenuating the response to carbachol, and because LDL was not available free of contaminating HDL, subsequent experiments were performed using HDL. Concentration-response curves for carbachol demonstrate that the maximum inhibition of cAMP accumulation is 62% in control cells, and 38% in HDL-treated cells, while the ECsos for carbachol in control and HDLtreated cultures are 0.3 and 3 p~, respectively (Fig. 3). In order to determine whether the ability of HDL to attenuate mAChR-mediated inhibition of cAMP accumulation arose from the increased protein content of the HDL-supplemented culture medium, the effect of the addition of a nonspecific protein, hemoglobin, was tested. When equivalent amounts of bovine hemoglobin were added to the culture medium, no significant attenuation in carbachol-mediated inhibition of forskolin-stimulated cAMP accumulation was observed (data not shown). This result indicates that a general increase in the protein content of the culture medium does not explain the capacity of serum lipoproteins to inhibit mAChR responsiveness. Cultured chick heart cells respond to mAChR activation with an increase in phosphoinositide metabolism. In order to by High Density Lipoproteins   I   I   l  ,  I  ,  I  ,  I  ,  I  ,  I  determine whether this pathway was also affected by HDL treatment of the cells, we measured the ability of carbachol to stimulate PI turnover in cells that had been cultured in the presence and absence of HDL. The maximum increases in [3H]Ins-Ps produced by cells cultured in the absence and presence of HDL were 378 and 421%, respectively (Fig. 4). While cells cultured in the presence of HDL tended to produce more [3H]Ins-Ps in response to carbachol than did control cells, the difference was not statistically significant. These results suggest that mAChR-mediated stimulation of phosphoinositide turnover is not appreciably altered by HDL. Long term exposure of chick heart cells to muscarinic agonist results in a decrease in both the total and cell surface number of mAChR present, a process known as down-regulation (Galper and Smith, 1980;Hunter and Nathanson, 1986). In order to determine whether lipoprotein treatment of cultured chick heart cells altered agonist-mediated regulation of mAChR number, the ability of carbachol to produce down-regulation in control and in HDL-treated cultures was examined. The maximum decrease in cell surface mAChR number, as measured by [3H]NMS binding to intact cells, was 68% in control cultures and 67% in HDL-treated cells, and the ECSOs for carbachol were also equivalent (Fig. 5). These results indicate that HDL treatment does not alter the ability of agonist to regulate mAChR number.

30
The observations that HDL treatment does not affect carbachol-mediated stimulation of phosphoinositide turnover or down-regulation suggest that HDL treatment does not cause a general decrease in receptor-mediated function. Thus, HDL appears to selectively attenuate mAChR-mediated inhibition of cAMP accumulation.
Determination of the Site of Action of HDL-Muscarinic receptor-mediated inhibition of cAMP accumulation requires the functional coupling of mAChR and Gj, and of Gi and adenylate cyclase (Murayama and Ui, 1983;Martin et al., 1985). Thus, HDL could inhibit signal transduction by decreasing mAChR number or affinity for agonist, decreasing the ability of mAChR to interact with Gi, decreasing the level of Gi, or decreasing the ability of Gi to couple with adenylate cyclase.
The possibility that mAChR-mediated inhibition of CAMP accumulation is attenuated in HDL-treated cells due to a decrease in mAChR number was tested by comparing total receptor number and antagonist affinity in cells cultured in the absence and presence of HDL. As shown in Fig. 6 A decrease in mAChR affinity for agonist could explain the increase in the ECso for carbachol-mediated inhibition of cAMP accumulation in HDL-treated cells. We therefore tested the ability of carbachol to compete for [3H]NMS binding on intact heart cells that had been cultured in the presence and absence of HDL. The results, shown in Fig. 7, demonstrate that carbachol competed for [3H]NMS binding identically in cells cultured under both conditions, indicating that mAChR affinity for agonist was the same in HDL-treated and control cells.
The affinity of mAChR for agonist can be controlled by the degree to which the receptor and the G-proteins with which it interacts are coupled. Thus, guanine nucleotides and their analogs have been shown to cause a decrease in agonist affinity (Florio and Sternweis, 1985), and the magnitude of this shift in affinity has been taken as a measure of coupling between G-proteins and mAChR. The decreased ability of carbachol to inhibit cAMP accumulation resulting from HDL treatment could arise from an HDL-mediated decrease in coupling between the receptor and Gi. In order to determine whether HDL treatment results in a diminished coupling between mAChR and the G-proteins with which it interacts, we determined the ability of carbachol to compete for [3H] QNB binding in the presence and absence of the non-hydrolyzable GTP analog, GppNHp, in membranes prepared from HDL-treat.ed and control cells. The results shown in Fig. 8  demonstrate that carbachol competition for [3H]QNB binding in the absence of GppNHp is similar in membranes prepared from HDL-treated and control cultures. The ECsos for carbachol inhibition of [3H]QNB binding in the absence of GppNHp in membranes from control and HDL-treated cells are 3 and 5.6 p~, respectively. The ECIOs for carbachol inhibition of [3H]QNB binding in the presence of 10 p~ GppNHp in membranes from control and HDL-treated cells are 18 and 70 p~, respectively. Therefore, GppNHp shifted the ECbO for carbachol 6-fold in control membranes, and 12.5-fold in HDLtreated membranes. These data indicate that the ability of GppNHp to regulate agonist binding is not diminished in HDL-treated cells and that the coupling between the mAChR and G-proteins is somewhat greater for HDL-treated cells.
Because the mAChR is known to interact with a variety of G-proteins in heart cells (Martin et al., 1985), it is possible that changes in coupling of mAChR with Gi specifically might not be apparent in binding studies such as those performed above. Therefore, we measured the actual level of Gi present in membranes prepared from control and HDL-treated cells using monospecific antibodies which recognize Gi,. Previous studies by Luetje et al. (1987) demonstrated that two forms

Muscarinic Receptor Regulation by High Density Lipoproteins
of Gia, one of M, 39,000 and one of M, 41,000, were present in chick heart. The 39,000 form of Gi, co-migrates with Go, but is immunologically distinct from Go, (Luetje et al., 1987). Quantitative immunoblot analysis using monospecific antisera demonstrates that the amounts of both the 41,000 and 39,000 forms of Gi, are similar in membranes prepared from control and HDL-treated cells (Table I). These results suggest that alterations in the level of Gi do not occur in response to HDL treatment and are unlikely to account for the diminished ability of carbachol to mediate inhibition of cAMP accumulation. Reports that certain responses could be due in part to the release of GB which could then act on effector proteins (Logothetis et al., 1987;Kim et al., 1989) raised the possibility that HDL might act by altering the level of GB in the membrane. However, membranes prepared from control and HDLtreated cells contained the same amount of GB, making this hypothesis untenable. Because HDL did not appear to act by altering receptor number, affinity, or coupling or by altering Gi, or GB levels, it was possible that Gi coupling with adenylate cyclase was less effective in HDL-treated cells. This theory was tested by analyzing the ability of GppNHp to inhibit, via activation of Gi, forskolin-stimulated adenylate cyclase activity in membranes prepared from control and HDL-treated cells. Because forskolin increases adenylate cyclase activity independently of stimulatory receptor activation, this protocol has been extensively used to quantitate Gi-adenylate cyclase coupling (Seamon and Daly, 1981;Halvorsen and Nathanson, 1984;Hunter and Nathanson, 1984;Martin et al., 1987;Luetje et al., 1987). GppNHp may also interact with G. to stimulate adenylate cyclase, resulting in an upward swing of a concentration-response curve at higher concentrations of GppNHp. Inhibition of adenylate cyclase activity by GppNHp occurs at lower concentrations of GppNHp than does activation because the on rate for GppNHp binding to Gi is thought to be greater than that to G, (Seamon and Daly, 1982). As demonstrated in Fig. 9, GppNHp was less effective in inhibiting adenylate cyclase activity in membranes prepared from HDLtreated cells than in control membranes. The maximum inhibition of cyclase activity by GppNHp was 32% in membranes from control cells and 23% in membranes from HDLtreated cells, while the ECBOs for GppNHp were 10 and 20 nM, respectively. The reversal of GppNHp-mediated inhibition of adenylate cyclase at 0.1 PM or greater GppNHp is most likely a reflection of the ability of GppNHp to stimulate adenylate cyclase activity via G. activation. These changes in GppNHp effectiveness are qualitatively similar to those observed for carbachol-mediated inhibition of forskolin-stimulated cAMP accumulation and could account for the decreased responsiveness seen in HDL-treated cells.
Determination of the Component of HDL Necessary for Inhibition of 4 C h R Responsiveness-A number of reports suggest that the effect of native lipoproteins can be accounted for in some cases by the apoprotein (Chen et al., 1986;Tour-

Effect of HDL on G-protein subunit levels in membranes prepared
from cultured chick heart cells 9-day embryonic chick heart cells were cultured in defined medium in the absence (Control) and presence of 25 pg/ml HDL, membranes prepared, and the levels of G i , (41,000 and 39,000 forms) and GB determined by quantitative immunoblot analysis as described under "Experimental Procedures."   HDL lipid (0, panel B ) . Cells were incubated in 100 p~ forskolin and the indicated concentrations of carbachol, and the cellular cAMP content was measured. Data are the average of at least two experiments, each performed in triplicate, and are expressed as a percent of the control cAMP content which is that measured in the absence of carbachol (m). nier et al., 1984;Wu et al., 1988) and in others by the lipid (Van Sickle et al., 1986;Cuthbert and Lipsky, 1986a) subfractions. We therefore tested each of these HDL subfractions for the ability to attenuate carbachol-mediated inhibition of forskolin-stimulated cAMP accumulation. Cells cultured in the presence of HDL apoprotein were as sensitive to carbachol as were control cells indicating that HDL apoprotein alone is not sufficient to explain theaction of native HDL (Fig. 1OA).

-
FIG. 11. Effect of HDL lipid on GppNHp-mediated inhibition of adenylate cyclase activity. 9-day embryonic chick heart cells were cultured four days in IT$ in the absence (W) and presence (0) of HDL lipid (in the amount equivalent to that contained in 25 wg/ml HDL), membranes prepared and incubated in 100 pM forskolin and the indicated concentrations of GppNHp for 10 min at 37 "C, and adenylate cyclase activity determined as described. Results are expressed as the percent of control activity which is that measured in the absence of GppNHp (na) and are the average of three experiments each performed in triplicate.
In contrast, cells cultured in the presence of the HDL lipid subfraction were less responsive to carbachol (Fig. 1OB). The maximum inhibition of cAMP accumulation caused by carbachol was 61% in control cells and 41% in cells treated with HDL lipid. This diminution of the maximum response is similar to that seen with native HDL (cf. Figs. 2 and 3), suggesting that HDL lipid can fully account for the decrease in mAChR responsiveness arising from HDL treatment. If the ability of HDL to attenuate carbachol-mediated inhibition of cAMP accumulation arises from a decreased effectiveness of Gi to inhibit adenylate cyclase and if HDL lipid is responsible for the effect of native HDL, then HDL lipid should also diminish the ability of GppNHp to inhibit adenylate cyclase. The results shown in Fig. 11 demonstrate that GppNHp was less effective in inhibiting forskolin-stimulated adenylate cyclase activity in membranes prepared from HDL lipid-treated cells than in those prepared from control cells. This result is consistent with the hypothesis that HDL acts to inhibit Giadenylate cyclase interaction via its lipid content.
It has been suggested that the increased mAChR responsiveness in heart cells cultured in lipoprotein-depleted serum result from decreases in the level of cellular cholesterol (Haigh et al., 1988). Because HDL is known to alter the cholesterol level of cultured cells (Karlin et al., 1987), we wanted to determine if this could account for the effect of HDL on mAChR-mediated responses. The membrane cholesterol content of chick heart cells which had been cultured in the absence and presence of either native HDL or of HDL lipid was tested in order to determine whether alterations in membrane cholesterol could account for the effect of HDL. Native HDL reduced the cholesterol content of heart cell membranes from 27.2 f 2 pg/mg membrane protein to 18.4 f 1.2 pg/mg. The cholesterol content of membranes prepared from cultures treated with HDL lipid was 25.6 k 0.4 pg/mg protein. Thus, while HDL lipid mimicked the ability of HDL to attenuate mAChR-mediated inhibition of cAMP accumulation, HDL lipid did not produce the changes in membrane cholesterol levels observed after treatment with native HDL. These results suggest that changes in membrane cholesterol do not account for the effect of native HDL.

DISCUSSION
Chick heart cells cultured in defined serum-free medium have been previously shown to respond to mAChR activation with an inhibition of cAMP accumulation and a stimulation of PI metabolism (Subers and Nathanson, 1988). Renaud et al. (1982) and, more recently, Haigh et al. (1988) observed that cultured chick heart cells have a greater negative chronotropic response to muscarinic agonist when the cells were cultured in the absence of serum lipoproteins. This result suggests that serum lipoproteins can regulate the sensitivity of cultured chick heart cells to mAChR activation. Using a fully defined culture system, the ability of HDL purified from rooster serum to alter mAChR function was investigated. Consistent with the results of Renaud et al. (19821, we observed that cells cultured in the presence of HDL were less sensitive to muscarinic agonists than were untreated cells. Interestingly, this diminished sensitivity to mAChR activation was restricted to carbachol-mediated inhibition of CAMP accumulation; carbachol stimulation of PI turnover and of agonist-induced down-regulation were unaffected by exposure of the cells to . This result suggests that HDL selectively decreases the function of the signal transduction pathway responsible for mAChR-mediated inhibition of cAMP accumulation, without reduction of other mAChRcoupled pathways. Recent reports that a decrease in intracellular cAMP is involved in the mAChR-mediated regulation of the pacemaker current, if (DiFrancesco and Tromba, 1988), suggest that the observed ability of lipoproteins to decrease the negative chronotropic response of cultured chick heart cells (Renaud et al., 1982) could be related to our observation that HDL can decrease the capacity of muscarinic agonist to cause a inhibition of cAMP accumulation.
The mAChR, the inhibitory guanine nucleotide-binding regulatory protein of adenylate cyclase, Gi, and the enzyme adenylate cyclase, must be present and functional for the decrease in cellular cAMP levels produced by mAChR activation. One goal of this work was to determine where in this signal transduction pathway HDL exerted its effect. The possible sites of action of HDL include, but are not restricted to, a decrease in mAChR number or affinity, a decrease in membrane Gi content, a decrease in the ability of Gi to interact with the receptor, or a decrease in the ability of Gi to inhibit adenylate cyclase activity. Each of these potential sites of action of HDL was investigated.
Agonist and antagonist binding data indicate that HDL did not cause a decrease in either mAChR number or affinity for agonist, eliminating the possibility that HDL treatment caused a decrease in mAChR sensitivity by altering receptor binding properties (Figs. 6 and 7). Agonist binding to cell surface receptors on intact chick heart cells was measured because this assay most closely duplicates the conditions under which functional properties of mAChR were studied. The ability of guanine nucleotides to shift agonist binding curves in membrane homogenates can be taken as a measure of the coupling between the receptor and the G-proteins with which it interacts. The non-hydrolyzable GTP analog, GppNHp, produced greater shifts in competition curves between carbachol and [3H]QNB in membranes from HDLtreated cells, suggesting that mAChR was more efficiently coupled to the G-proteins with which it interacts following HDL treatment. Thus, a decreased efficiency of interaction between mAChR and Gi is unlikely to account for the diminished ability of carbachol to inhibit cellular cAMP accumulation in HDL-treated cells.
Alternatively, an HDL-mediated decrease in the level of Gi in the membrane could explain the diminished responsiveness of HDL-treated cells. When the amount of Gi, and Go were measured by quantitative immunoblot analysis in membranes prepared from control and HDL-treated cells, no difference Regulation by High Density Lipoproteins was observed (Table I). This result suggests that neither a decrease in Gi, nor in GB can account for the effect of HDL.
In order to determine whether HDL acted to reduce the efficiency of coupling between Gi and adenylate cyclase, a comparison of the ability of GppNHp to inhibit forskolinstimulated adenylate cyclase activity in membranes prepared from control and HDL-treated cells was performed. GppNHpinduced inhibition of adenylate cyclase activity was attenuated in membranes prepared from HDL-treated cells (Fig.  9), indicating that the coupling between Gi and adenylate cyclase was reduced by HDL-treatment. These experiments suggest that HDL causes a decrease in mAChR responsiveness by interfering with the interaction between Gi and adenylate cyclase.
Apoprotein fractions of HDL were unable to cause a decreased sensitivity to carbachol in cultured chick heart cells, suggesting that HDL apoproteins cannot substitute for native HDL. However, when the ability of lipid fractions of HDL to decrease mAChR responsiveness was examined, it was found that they could mimic the effect of native HDL. Because the degree of attenuation of the response to carbachol was similar for native HDL and its lipid fractions, it appears that the effect of the lipid fractions of HDL can fully account for the effect of native HDL. Addition of HDL lipid fractions to the culture medium also decreased GppNHp-mediated inhibition of adenylate cyclase activity, which is consistent with the conclusion that HDL acts to decrease the efficiency of coupling between Gj and adenylate cyclases.
Because the ability of Gi to inhibit adenylate cyclase and adenylate cyclase activity itself is sensitive to the membrane lipid composition (Engelhard et al., 1978;Murphy, 1986;Murphy et al., 1987), it seems likely that HDL acts to alter cellular membrane lipid composition, thus inhibiting mAChR-mediated reductions in cellular cAMP accumulation. In a related study, Haigh et at. (1988) reported that removal of lipoproteins from fetal calf serum used to supplement the medium resulted in an enhanced negative chronotropic response in cultured chick atrial cells exposed to muscarinic agonists. The increased sensitivity to mAChR activation was reversed when bovine LDL, but not bovine HDL, was added back to the culture medium. The heightened mAChR responsiveness was correlated with decreases in cellular cholesterol content, suggesting that the increased level of cholesterol in atrial cells can decrease mAChR responsiveness. The bovine HDL used in that study did not alter membrane cholesterol content, suggesting that it did not interact effectively with chick heart cells. In contrast, the data presented here demonstrate that both chicken HDL and LDL decrease mAChR sensitivity in heart cell cultures. Because HDL is able to stimulate bidirectional cholesterol flux in cultured cells (Karlin et al., 1987), the effect of HDL observed here could have been due to an increase in membrane cholesterol content, analogous to those reported by Haigh et al. (1988). However, the results reported here indicate that treatment of cultured chick heart cells with rooster HDL reduces membrane cholesterol content. Moreover, the lipid fraction of HDL did not alter membrane cholesterol but did mimic the effect of native HDL on mAChR sensitivity to agonist, suggesting that changes in membrane cholesterol do not mediate the effect of native HDL. Haigh et al. (1988) also observed a significant increase in mAChR number and in Gi, subunit levels in atrial cells cultured in serum in the absence of lipoproteins, which could explain the increased sensitivity of these cells to mAChR activation. It appears that lipoproteins can regulate mAChR function by more than one mechanism because addition of HDL to cells cultured in defined medium changed neither mAChR nor Gi, levels (Fig. 6, Table I). It should be noted that Haigh et al. (1988) measured only the ability of mAChR to mediate a decrease in beating rate following the removal of lipoproteins; mAChR-mediated inhibition of cAMP accumulation and stimulation of phosphoinositide metabolism were not examined. Here, we clearly demonstrate that lipoproteins do not affect all mAChR signal transduction pathways, so that it is possible that different pathways, e.g. beating rate response and cAMP response, are regulated by different lipoproteins with different mechanisms of action.
The work presented here shows that mAChR function is regulated by serum lipoproteins, especially by HDL. HDL causes an attenuation of the ability of carbachol to inhibit cAMP accumulation in cultured chick heart cells, with no change in its ability to stimulate phosphoinositide turnover, indicating that a general decrease in cellular, membrane, or receptor function does not occur. HDL apparently produces this effect by interfering with the ability of Gi to inhibit adenylate cyclase activity. The lipid fraction of HDL can mimic the effect of native HDL, suggesting that HDL may exert its effect by altering cellular lipid content.