The Glucagon-sensitive Adenyl Cyclase System in Plasma Membranes of Rat Liver

The first in a now-classic series of five articles in which Rodbell concluded that GTP was likely the active biological factor
 in separating glucagon, a hormone that can act to increase blood glucose levels, from the cell's receptor, which had important
 implications for the treatment of various disorders and diseases.

Franz the Section on Nenzbrane IZegulatiou, Nationad Institute of Arthritis cd Jletabolic Diseases, ii'atio,lal Institutes of Health, Bethesda, Maryland 20014 SUMMARY Plasma membranes prepared from rat livers have been treated with digitonin or phospholipase A under conditions which result in substantial loss of glucagon-stimulated adenyl cyclase activity but no loss of fluoride-stimulated activity. These results are thought to reflect extensive modification of the structures responsible for hormone sensitivity without destruction of the catalytic component of the adenyl cyclase system in these membranes.
Corresponding decreases in binding of lz51-glucagon to the membranes are observed following digitonin or phospholipase A treatment. Both glucagon sensitivity of adenyl cyclase and binding of lz61glucagon can be partially restored by exposing treated membranes to aqueous suspensions of membrane lipids. The mechanism of the effects of these lipids has not been established.
Pure phosphatidylserine, phosphatidylcholine, and phosphatidylethanolamine are all capable of partially restoring glucagon-stimulated adenyl cyclase activity and binding of 1251-glucagon to phospholipase A-treated membranes. Of these, phosphatidylserine produces the greatest effects. Chromatographic fractions of membrane lipids produce positive and negative effects on control and treated membranes which are difficult to interpret.
However, it appears that there is some specificity in the effects of the lipid fractions. These results may have significant implications regarding the relationship of adenyl cyclase systems to membrane structure and attempts to purify the components of these systems. * Some of the studies have been reported in preliminary form (l-3).
The previous papers in this series are References 4 to 8.
A recent series of papers from this laboratory (4-8) described in detail the glucagon-sensitive adenyl cyclase system in plasma membranes prepared from rat liver. One set of observations led to the conclusion that glucagon and fluoride ion stimulate thr: same adenyl cyclase by distinctly different mechanisms (5). Using t.he fluoride-stimulated activity as a marker for the catalytic cornponent of the system, it was shown that treatment of the mcmbranes with digitonin or phospholipase 11 could destroy cornpletely the hormone sensitivity of the system without destroying the catalytic function (5). Furthermore, it was shown that the ability of the membranes to bind glucagon is reduced by treatment with digitonin or phospholipase A (6). These studies suggested that membrane lipids play an essential role in the process by which glucagon stimulates the liver membrane adenyl cyclase.
Studies with other tissues suggest that lipids are important, in all hormone-sensitive adenyl cyclase systems. Sutherland, Rall, and Menon (9) found that adenyl cyclase activity which had been "solubilized" by treating homogenates of brain and other tissues with Triton (alkyl polyphenoxypolyethoxyethanol) no longer rcsponded to hormone stimulation but' did respond to fluoride ion. Levey (10) has recently reported that cat heart ventricles homogenized in Lubrol-PX (a nonionic detergent) contains adenyl cyclase activity which is soluble by several criteria; this act,ivity also responds to fluoride but not to hormone stimulation. Wolff and Jones (11) have reported inhibition of hormone-stimulated adenyl cyclase activity but not fluoride-stimulated activity by chlorpromazine in several tissues. Treatment of fat cell ghosts with digitonin results in loss of the response of their adenyl cyclase activity to four different hormones but not, to fluoride ion (5). The agents used in these various studies would all be expected to interact with membrane lipids.
Elucidation of the relat,ionship bet'ween membrane lipids and adenyl cyclase will undoubtedly be required in order to obtain an understanding of the initial actions of polypeptide hormones.
For example, lipids may be involved in the binding of hormones to membranes, in translatior! of this binding reaction into an effect such as stimulation of adenyl cyclase activity, or both. Furthermore, it may not be possible to study the components of hormone-sensitive adenyl cyclase systems in simpler, nonmt~rnbrn-neous states unless the role of lipids is understood. For these reasons, we have examined further the lipid requirements of the liver membrane adenyl cyclnse system. The studies described in the present report show that the hormone-response of the adenyl cyclase system and the ability of the membranes to bind glucagon can be partially restored following digitonin or phospholipase A treatment by exposing the treated membranes to aqueous dispcrsons of lipids. Chloroform and methanol were reagent grade products of the J. T. Baker Chemical Company, and were used without redistillation. Glucagon (crystalline) was a gift from the Eli Lilly Company. Phosphatidylserine and phosphatidylcholine were obtained from Applied Science Labs and phosphatidylethanolamine from Supelco, Inc., Bellefonte, Pennsylvania. These lipids were found to yield one spot with the thin layer chromatographic system described below and were used as supplied.
The sources of other materials have been specified previously (4, 6).

Methods
Liver Membranes-Plasma membranes were prepared from rat livers by a minor modification of the procedure of Neville (13) as described previously (4). All preparations used for this study were partially purified membranes (4).
Extraction of Lipids-An antioxidant, 2,6-di-tert-butyl-4methylphenol (Aldrich Chemical Company), at 0.005°~0, was added t,o all solvents. One batch of liver membranes (80 to 120 mg of membrane protein) suspended in 1 mM KHCOI in a total volume of about 5 ml, was extracted with 90 ml of CHCl:s-CH30H, 2:l (v/v), for 2 hours in an ice bath. The insoluble material was removed by filtration through Whatman No. 1 filt,er paper. The filt,rate was extracted with 0.2 volume of 0.1 M CaC$. The upper phase was discarded and the lower phase was filtered to remove any remaining water droplets.
The solvents were removed by evaporation under a stream of Nz in a flash evaporator (Buchler Instruments, Inc., Fort Lee, New Jersey) at 40". Following evaporation, the lipid residue was dissolved immediately in 1 to 2 ml of CHClp.
Column Chromatography of Lipids-The lipids extracted from one batch of liver membranes were dissolved in 2 ml of CHCl,. One ml of this solution was rcmovcd for comparison with the subsequelit lipid fract,ions. The remainder was applied to a column (0.8 x 10 cm) of silicic acid (Unisil, Clarkson Chemical Company, Williamsport, Pennsylvania; 200 to 325 mesh) packed in CIIC'13. The lipids were washed into the column with 1 to 2 ml of CIICl~, and the column was elutcd at 22", with a flow rate of 0.5 ml per min, using 20 ml of the following eluents: Fraction I, CHCl,; Fraction II, CHC1&H30H, 9:l (v/v); Fraction III, CHC1&H30H, 4:l; Fraction IV, CHC13&H30H, 1:4; Fraction V, CHpOH.
Solvents were removed from these fractions, which were collected in tared vessels, by flash evaporation as described above, and the lipid residues were weighed and then dissolved immediately in 1 ml of CHC13.
Dispersion of Lipids in Buffer-Lipids dissolved in CHC13 were transferred to either glass test tubes (10 x 75 mm) or 13.ml conical glass centrifuge tubes depending on the volume required. The solvent was removed by evaporation with a stream of Ns, and the buffer, 25 mM Tris-HCl, pH 7.6, 1 mM EDTA, was added over the lipid film. In some experiments, the lipids were dissolved in diethyl ether which was layered on and evaporated from the surface of the buffer. The former procedure was found to be simpler. The lipids were then dispersed by sonication with a Biosonik III (Bronwill Scientific Company, Rochester, New York) at 35y0 intensity until continued sonication failed to clarify the solution further (2 to 10 min). The vessel in which the sonication was performed was immersed in an ice bath and a stream of Nz was directed over the surface of the buffer. The dispersed lipids were stored at 4" and were used within 1 week of preparation.
The membrane lipids used in this study were not weighed routinely.
Typically, one bat,ch of membranes yielded 60 mg of lipid which was dispersed in 2.5 ml of buffer. The lipid suspensions were used at this concentration except as specified in "Results" when they were diluted in 25 mrvr Tris-HCl, 1 m&r EDTA.
In early experiments, a small amount of sedimentable material was removed from the milky whole lipid suspensions by centrifugation for 15 min at 25,000 x g. However, this was found not to affect the subsequent effects of the lipids and was not continued.
Thin Layer Chromatography of Lipids-Glass thin layer chromatography plates precoated with Silica Gel H (Analtech, Inc., Wilmington, Delaware), were activated by heating for 30 min at, 110". Lipids were spotted in 2 to 10 ~1 of CHC13, and the plates were developed with CHCl&H30H-CH&OOH-HzO, 100: 62 : 18:24 (14). Detection of spots was accomplished by spraying the plates with 5y0 H.$Od in ethanol and heating for 10 min at 220". Purified phospholipids described above were used as standards.
ddenyl Cyclase Assay-Adenyl cyclase activity \VBS determined by following the conversion of ATP-ar-32P to cyclic hl\II'-32P' using the method of Krishna, Weiss, and Brodie (15). Standard incubation media and conditions have been specified previousl> (4) '2jI-Glucagon Binding Assay-The methods for preparation of IzjI-glucagon and measurement of binding have been described previously (6). The incubation rnedium contained 2.5% albumin, 20 InM Tris-HCI, pH 7.6, 0.2 to 1.0 mM EDTA, 0.5 to 2.0 X lo-!' III i*%glucagon, and 0.2 to 0.4 mg per ml of membranes in a volume of 0.125 ml. Incubation was for 15 min at 30".
Protein Determination-protein was measured by the method of Lowry et al. (16) using bovine serum albumin as standard.

Expression of Results and Terminology
In all figures and tables "ildenyl cyclase activity" refers to nanomoles of cyclic AMP formed in 10 min per mg of membrane protein and "Glucagon bound" refers to picomoles of r2%glucagon per mg of membrane protein.
Basal adenyl cyclase activity is the activity measured in the absence of glucagon or fluoride ion. Glucagon-and fluoride-stimulated activities are the activities measured in the presence of these compounds and from which basal activity has been subtracted.

RESULTS
Tn the present studies we have used the fluoride-stimulated adenyl cyclase activity only to indicate that treated membranes are capable of catalyzing the conversion of ATP to cyclic AMP. Conditions of treatment with digitonin or phospholipase A have been selected to produce substantial loss of glucagon-stimulated adenyl cyclase activity and glucagon binding but little or no loss of the fluoride-stimulated activity. This approach is justified by our previous finding that glucagon and fluoride ion stimulate the same adenyl cyclase activity in the liver membrane preparation but do so by different mechanisms (5).
Digitonin and phospholipase A, the agents which were found to produce a selective loss of the glucagon-stimulated adenyl cyclase activity and of glucagon bindin g, would be expected to interact primarily with membrane lipids.
Therefore, it was reasonable to suspect that lipids might be able to restore these functions to treated membranes.
Accordingly, membrane lipids were extracted and dispersed in buffer and added to suspensions of treated membranes.
A representative experiment with digitonin is shown in Fig. 1. The glucagon-stimulated adenyl cyclase activity has been reduced to 36% or 9% of the control by treatment of the membranes with 1 or 10 mg per ml of digitonin, respectively. These activities are increased to 65% or 22y0 of control by exposing the treated membranes to a suspension of membrane lipids.
The basal activity (no glucagon or fluoride, no digitonin) in this experiment was 0.20 nmoles of cyclic AMP per 10 min per mg of protein and was unaffected by the addition of the lipid dispersion. Digitonin treatment decreased the basal activity and it remained unaffected by lipids.
The control for this experiment is the glucagon-stimulated activity of the membranes which were carried through the entire treatment procedure but in media containing no digitonin.
As is apparent from Fig. 1, this activity is unaffected by the addition of the lipid dispersion.
Therefore, the lipid dispersion has partially restored the glucagon response of the adenyl cyclase activity in the treated membranes.
Similarly, the lipid dispersion partially restores the binding of lZ51glucagon to digitonin-treated membranes ( Table I). The fluoride-stimulated adenyl cyclase activities are included in Fig. 1 to illustrate the fact that the mechanism through which fluoride acts on adenyl cyclase can also be modified by agents which interact with membrane lipids. However, for reasons mentioned above and because the mechanism of action of fluoride on the system is unknown, the fluoride-stimulated adenyl cyclase activity can only serve to demonstrate the integrity of the catalytic site and cannot serve as a control for interpreting effects on the glucagon-stimulated activity.
In general, digitonin and phospholipase A tend t'o iucrease the fluoride-stimulated activity; however, relatively high concentrations of these agents decrease this activity.
Addit,ion of lipid dispersons mimics the effects of digitonin and phospholipase A and, as illustrated in Fig. 1 snake venom (17). As report,ad previously (5), treatment of the membranes with a relatively low concentration of phospholipase A slightly but significantly increases the glucagon-stimulated adenyl cyclase activity (Fig. 2). In addition, treatment with a lower concent,ration of phospholipase A increases the binding of 2. Effects of phospholipasc A on xdcnyl cyclase :tctivit,y and reversal by lipids. Liver membranes, 3 mg of membrane protein per ml, w&C: incubated ill 0.38 ml of a medium containing 25 rn~ Tris-HCl. 1111 7.6. and 1 rn~ C&I? with uhosoholioase A at the indicated C(;I1Celltr:,tiOlls for 5 min :it 30".-The re&ion was stopped by adding 0.02 ml of 20 mM EDTA. Aliqlmts of the treated membrane sllspellsion were incubated with an equal volume of either 25 mM Tris, 1 mM IGlYlYA, or membrane lipid suspension, 1:8 (set "Methods") in 25 IIIM Tris, 1 mM EDTA for 5 min at 22'. Adenyl cyclase activities in the absence or presence of 10 pg per ml of glllcagon and protein concentrations were measnred as described nnder "Xethods." Values are the increase in adenyl cyclasc activity due to the addition of glucagon to the assay medium alld are t,he meal, f half the range of d\lplicate determinations.
Exposing phosl~holipase A-treated membranes to membrane lipids partially restores the response of adcngl cyclnse :Ictivit,y to glucagon.
In Fig. 2, this effect is most, striking at, the highest phospholipase ,I concentration ; under this condition, no stimulation of adenyl cyclase by glucagon was observed unless the membranes were exposed to lipids.
The positive effect of the lipids on the control (no phospholipase A) membranes is probably due to the fact that diluted membrane lipids were used in this experimcnt (see below).
In other experiments using 4-to g-fold higher lipid coucelltr:Ltions, the glucagoll-stiniulated ndenyl cyclase activity of the control membranes was unaffected or reduced.
In all experiments, basal adenyl cyclsse act.ivity was either unaffected or reduced by either phospholipase B or lipids, or both. As with digitonin, either phospholipase A treatment or lipids slightly increased fluoride-stimulated adenyl cyclase activity, but both together failed to produce an additional effect (data not s110w11). Exposing l~liosl~liolil~ase X-treated membranes to lipids also partially restores the ability of the membranes to bind glucngon. In Fig. 3, at the highest concentration of phospholipasc A, lipids have restored glucagon binding from 20 to 89% of control.
It should be noted that the membranes, phospholipase I\, and treatment conditions were identical in the experiments shown in Figs. 2 and 3. As shown previously (6), the effects of varying concentrations of phospholipase d on glucagon-stimulated ndcnyl cyclase and on glucagon binding do uot correlate perfectly.
The binding data presented in this paper must be considered to apply to the characteristics of a glucagorl-specific binding site in the same membranes which contain the glucagoa-stimulated adcnyl cyclase system and not neressarily to the characteristics of th:lt site which actually mediates the stimulation of adenyl c~cl:~sc activity by glucagon.
On the other hand, the studies prcsentcd here do not prove that the observed binding sites are not those which rnediate the stimulation of ndenyl cyclase activity. The problem of establishing or rulin, 0 out a cause-effect relationeliil~ between observed binding of glucagon and activation of adeny cyclase has been treated in a preliminary way elsewhere (3, 6) and is beyond the scope of this paper.
In general, the restoration effects of lipids on phospholipase *Itreated membranes have been larger than the effects on digitonintreat,ed membranes.
For this reason, phospholipase A treat,mcnt was selected for studies of the specificity of the effects of liljids. In Figs. 1 to 3, effects of membrane lipids on treated membranes were emphasized by selecting conditions in which lipids produced little or no effect on control membranes.
Under these conditions, exposure to lipids restored adenyl csyclase and binding properties of treated membranes toward the values of relatively unaffccttd controls.
However, the control membranes may also be affected by exposure to lipids.
For cxamplr, Fig. 4 sho1T-s both positive Issue of July 25, 1971 AS. L. Pohl, H. M. J. Kram, V. Kozyq$, L. Birnbaumer, and M. Rodbell and negative effects of lipids on control and phospholipase ht rtnted membranes dcpcnding on lipid concentration. It is clear, therefore, that addition of an aqueous dispersion of lipid?; can affect the system in more than one way. This fact greatly complicates the problem of establishing t.he specificity or nonspecificity of the effects of these lipids since both kinds of effccts may occur simultaneously and may either mimic or counteract each other. l'l~onl~l~olipids have surface-active properties which may be either strong or weak depending on the composition of the phospholipid and may, therefore, mimic the effects of detergents on the system. Furthermore, addition of a lipid dispersion t,o the system adds a new phase, lipid micelles, to a system which already contains two phases, an aqueous medium and membranes.
The physical properties of the micelle phase and its interact,ions with the other two phases will depend on the composit ion of the micelle phase. Finally, although the membranes are free of significant contamination by other organelles (4), they are still a highly complicated material for biochemical investigation. The components of the hormone-sensitive adenyl cyclase system probably represent only a small fraction of the total protein and lipid in the membranes, and alterations in the properties of the membrane as a whole may be reflected secondarily by changes in the adenyl cyclase system. In view of these considerations, it is not surprising t,hat the question of the specificity of t,he effects of adding lipids to the system cannot be answered easily.
Dcspitjc these problems, we have obtained some information regarding the effectivrncss of different lipid classes. In R typical experiment, one bst,c:h of partially purified membranes, npproximately 120 mg of membrane protein, was extracted as described in "Mel hods." _ I L One-half of this extract was dried aud weighed. Tile remainder was applied to the column and &ted as described in ".\It%hods." The amounts of lipid in these fractions, determinctl gravimetrically, were as follows: total allplied, 30 mg; Fraction I, 6.3 mg; Fraction 11, 3.6 mg; Fraction 111, 2.6 mg; l!ra&~n IV, 14.2 mg; Fraction V, 1.0 mg. By thin layer chromatography, neutral lipids were seen exclusively in Fraction I, and no spots corresponding to polar lipids were seen in this fraction. l'hosphatidylethanolamine was seen primarily in Fraction II, l~llospl~atidylseriue in Fraction III, and phosphatidylcholine in Fraction IV. However, there was considerable overlap of lipids in adjacent fractions.
These results agree well with analyses of rat liver plasma membrane lipids reported from other laboratories (14, 18).
The effects of the membrane lipid fractious on 1~hosl~holil~ase A treated and control membranes are shown in Table II. For t,his csperimcnt, the amount of lipid and buffer were adjusted so that the concentration of lipids in the suspension of each fraction would approximate t,he concentration of those lipids in the undiluted membrane lil'id suspensions used in experimeut,s above. These amounts could have been chosen so that the total lipid concent ration would have been the same for all fractiorls.
However, for re~tsons mentioned above, either choice is arbitrary and subject to the same crit,icisms. In order to compare the effects of fr:tctions of membrane lipids to those of whole membrane lipids it WIW intuitively more appealing to use the former condition. The high concentrations of lipids probably accounts for the negative cffccts of the various fractions on the control membranes (see Fig. 4). Fraction I reduced the glucagon-stimulated adenyl cyclnse act,ivity of treated membranes in proportion to its effect 011 control membranes.
Fractions II and III contained low and roughly equal concentrations of lipids and affected adenyl cyclase activity of the control membranes to about the same extent. Fraction III, however, produced a much larger positive effect on the glucagon-stimulated ndenyl cyclase activity of t,he treated membranes than did Fracation II. Fraction IV had a substantial negative effect on both control and treated membranes but differed from the other phospholipid fractions in having a much larger total concentration of lipid.
Fraction III had the largest positive effect on glucagon binding in treated membranes.
However, in contrast to the adenyl cyclase findings, all of the fract,ions except Fraction I produced positive effects on binding in the treated membranes.
Because of the difficulties involved in purifying lipids and the apparent multiplicity of the effects of lipids, further fract,ionation of membrane lipids was not undertaken.
Instead, three of the principal membrane phospholipids were obtained in purified form from other sources. The effects of these phospholipids on phospholipase A-treated and control membranes are illustrated in Table III.
It is clear that all three lipids tested are capable of partially restoring glucagon-stimulated adenyl cyclase activity and bindiug of glucagon to treated membranes.
Of the three, phosphatidylserine is clearly the most effective. Phosphatidylethanolamine differs from the others in consistently showing a marked positive effect on control membranes.
Combinations of lipids do not produce additional effects; if anything, these combinations product less effect than the individual lipids.
Although the specificity experiments are difficult to interpret, four conclusions seem warranted.
(a) Different lipids may affect the system in different ways; (b) more than one phospholipid are capable of restoring glucagon-stimulated adenyl cyclase activity and glucagon binding of treated membranes toward the values of a relatively unaffected control; (c) certain lipids do not produce this restoration; (d) the effects of lipids on adenyl cyclase and on glucagon binding are not always quantitatively or qualitatively the same. The question of specificity of the role of lipids in hor- mone-sensitive adenyl cyclase systems probably will not be answered until the system has been simplified and separated from the bulk of the extraneous membrane material.

DISCUSSION
The structures responsible for hormone-sensitivity of a mammalian adenyl cyclase system can be extensively modified by agents which modify membrane lipids without destroying the catalytic site of the system and these effects can be partially reversed by exposing the treated membranes to aqueous dispersions of lipids. These findings illustrate further the complexity of mammalian adenyl cyclase systems and are consistent with the hypothesis (19, 20) that these systems are multimolecular with separate hormone-sensitive and catalytic components. While the present study is, to our knowledge, the first of its type reported for a hormone-sensitive adenyl cyclase system, effects of lipids on detergent and phospholipase-treat,ed preparations have been reported for (r\ra+ + K+)-dependent ATPase (21, 22), mitochondria (23), and several microsomal enzymes In no case has the mechanism of action of lipids been established.
The most attractive hypothesis is that lipid suspensions act by replacing essential membrane lipids which have been either removed or destroyed by detergents or phospholipase treatment (26). However, Emmelot and Bos (27), in reference to deoxycholate-treated preparations of (Na+ + K+)-dependent ATPase, have suggested that lipids act by removing bound detergent.
Zakim (24) has presented evidence that one effect of phospholipase A on microsomal glucose 6-phosphatase is to produce an unstable form of the microsomal enzyme and that lipids act by stabilizing this form. Any one or combination of these mechanisms could explain the effects of lipids reported here. The concentration-dependent positive and negative effects of lipid suspensions on control membranes may be due to detergent properties of phospholipids in these suspensions. The molecular basis of the involvement of membrane lipids in adenyl cyclase systems is far from clear. Hormone-specific binding sites appear to contain lipoproteins since they are sensitive to both trypsin (28)   changes in adenyl cyclase activity produced by manipulating the membrane lipids undoubtedly reflect changes in the state of the hormone-binding site. However, at least three observat.ions suggest that lipids are involved at multiple sites in the system: the increase in fluoride-stimulated adenyl cyclase activity produced by relatively low concentrations of digitonin or phospholipase A (5), the complete loss of fluoride-stimulated activity produced by sufficiently high concentrations of these agents,2 and the lack of absolute correlation between the effects of digitonin, phospholipase il, and lipid suspensions on glucagon binding and activation of adenyl cyclase. The ability of more than one kind of phospholipid to reverse the effects of phospholipase A indicates that part of the effect of membrane lipids may be nonspecific, i.e. related only to maintaining a generally hydrophobic environment for the components of the adenyl cyclase system. However, the differences in the effects of the various membrane lipid fractions on phospholipase A-treated membranes suggests that specific lipids may be required at specific sites in the system.
The apparent involvement of lipids at multiple sites in the liver membrane adenyl cyclase system is consistent with certain concepts of adenyl cyclase organizat,ion and membrane structure. Davoren and Sutherland (29) first proposed that t,he hormone binding component of the adenyl cyclase system might be located on or near the extracellular surface and the catalytic component on or near the intracellular surface of the plasma membrane. This hypothesis has been supported by the finding that trypsin treatment of isolated fat cells results in loss of sensitivity of these cells to the effects of the lipolytic hormones (28) without disrupting the integrity of the plasma membrane (30, 31) or destroying the fluoride-stimulated adenyl cyclase activity in "ghosts" prepared from these treated cells (28). However, trypsin treatment of ghosts, a hypotonic lysate of isolated fat cells, inactivates bot.h the hormone-and the fluoride-stimulated adenyl cyclase activities (32). These results probably reflect a difference in accessibility of trypsin to the outer and inner surfaces of the plasma membrane in isolated fat cells and ghosts. Blthough numerous objections have been raised (33), it is generally accepted that the unit of membrane structure is the extended phospholipid bilayer (34). Even if the "unit membrane" theory is not correct in detail, there is evidence for extended, lipid-rich regions in cell membranes (35). Consequently, the hormone binding and catalytic components of mammalian adenyl cyclase systems may be separated by a lipid-rich layer.
Zakim (24) has observed increases in microsomal glucose 6phosphatase activity due to treatment with phospholipase A and albumin and, on this basis, has suggested that membrane lipids "constrain" the maximum activity of the enzyme. The increases in glucagon-or fluoride-stimulated adenyl cyclase activities produced by certain conditions of treatment with digitonin, phospholipase A, and lipid suspensions suggests a similar hypothesis for the adenyl cyclase system. The reartion between glucagon and its binding site on the membrane might then activate adenyl cyclase by changing the state of the membrane lipids in such a way that a limitation on the activity of the cat,alytic component is relieved.
If such a mechanism is operative, it should be possible to activate adenyl cyclase in the absence of glucagon by modifying membrane lipids.
Such a phenomenon has not yet been observed with the liver membrane system. However of adenyl cyclase by detergents in bhe absence of added hornlone has been observed with homogenates of brain (36) and nvian erythrocytes (37) and in fat cell ghosts (3).
The studies with lipids described here have important imy)lica:Ltions for future work with adenyl cyclase systems. In attempting to purify the components of these systems, it has generally been assumed that these structures might retain functional tollfigurations after separation from membrane lipids.
Such au espectation may be unreasonable since the factors which govc~n protein configurations in membranes are poorly understood. However, it may be possible to purify components of adeny cyclase systems and retain or restore function by performing purification steps in media containing phospholipids or by adding phospholipids to membrane protein fractions. Pastan, Pricer, and Blanchette-Mackie (38) have made a step in this dire&on by devising a method for preparing a particulate but nonsedimentable adrenocorticotropin-sensitive adenyl cyclase activity from an adrenal tumor; the method requires the addition of phospholipid to media. Very recently Levey has shown (39) that addition of phosphatidylserine to a hormone-insensitive adenyl cyclase, which has been solubilized from cat heart with Lubrol-PX and then freed of detergent by ion exchange chromatography, restores the sensitivity of the enzyme to glucagon.