- Email: [email protected]
The signal transduction between β-receptors and adenylyl cyclase
Life Sciences, Vol. Printed in the USA
52, pp. 2093-2100
The signal transduction between cyclase
Pergamon Press
-receptors and adenylyl
A. Levitzki, I. Marbach and A. Bar-Sinai
Department of Biological Chemistry, The Alexander Silberman Institute of Life Sciences, Hebrew University of Jerusalem, Jerusalem, Israel
Summary The 13-adrenergic receptor-dependent adenylyl cyclase system is the most extensively studied G-protein-coupled system. Studies of the coupling between the receptor and effector can provide an insight into the nature of all these systems in general. In the activation of adenylyl cyclase by the receptor, the binding of an agonist to the stimulatory receptor (Rs) and the binding of GTP to the G-protein (Gs) are both required to activate the catalytic moiety (C). The active state decays as GTP is hydrolysed to GDP and inorganic phosphate (Pi), but reactivation occurs as GTP is replenished. The receptor acts as a catalyst, i.e. one agonist-bound receptor can activate numerous adenylyl cyclase molecules. Kinetic studies led to the formulation of the 'collision coupling' model of receptor activation and show that G s protein does not shuttle between the receptor and cyclase. The G~ protein appears to undergo conformational changes between an 'open' state in which it can bind with GTP, and a 'closed' state unable to achieve this binding. This mechanism of activation does not involve the dissociation of G s or of G i. A model which fits the experimental data suggests that Gi*GTP affects cyclase only in its Gs-activated state via the G~ 1 subunit, but that the oligomeric state of G i is required for inhibition. The site on C which interacts with G i is formed only when C is activated by G s.
Introduction The [~-adrenergic receptor-dependent adenylyl cyclase is still the most extensively studied Gprotein coupled system and the one we understand best. Thus understanding the details of receptor to effector coupling in this system is a useful model which can help investigators learn about other receptor G-protein systems. In the receptor-regulated adenylyl cyclase, all the protein components have been characterized, purified and reconstituted [ 1, 2]. The members of the [3-adrenoceptor family have all been found to possess seven putative transmembrane domains and, therefore, similar three dimensional structure and overall conformation. The effectors of this family of receptor G-protein coupled systems are less well characterized, with only the adenylyl cyclase having been recently cloned [3]. Since we are aware o f at least 16 different heterotrimeric G-proteins [4] it is clear that many effectors are waiting to be discovered or characterized more fully. The receptor-regulated adenylyl cyclase is composed o f 5 functional units: a) A stimulatory receptor Rs which binds the stimulatory hormone or the neurotransmitter. The [~-adrenergic receptor is the prototype and best characterized [1, 5]. This receptor is a transmembrane glycoprotein with 7 putative transmembrane-spanning sequences. It is still the only member of this large family of G-protein coupled receptors which has been purified and successfully reconsti-
Copyright
©
0024-3205/93 $6.00 + .00 1993 Pergamon Press Ltd. All rights reserved.
2094
Signal Transduction and Adenylyl Cyclase
Vol. 52, No. 26, 1993
tuted with other pure components of the adenylyl cyclase system [6-9]. b) The G-protein, G s, which binds GTP and activates the adenylyl cyclase enzyme, C. G s is composed of 3 subunits: a s, which possesses the GTP binding site and is the target for cholera toxin catalyzed ADP-ribosylation, and the 13and y subunits, which are closely associated with each other [10]. The heterotrimeric Gs-protein is localized in the inner leaflet of the plasma membrane, c) The catalytic unit, C, is a hydrophobic protein with twelve putative transmembrane sequences with the catalytic site on the cytoplasmic side of the membrane [3]. The system would be incomplete if we discussed only the activating branch of the adenylyl cyclase system. Thus we shall consider d), the inhibitory receptors, R i, which bind inhibitory ligands and, like the 13-adrenergic receptor, span the membrane 7 times; an example of this is the a2-adrenergic receptor, e) The inhibitory GTP-binding protein, G i. This protein, like G~, is composed of three subunits: a i, 13and y, where the 13ycomplex is highly similar or identical to the 13yin G~ [10]. The subunit cg is homologous to a s but is a substrate for ADP-ribosylation catalyzed by pertussis toxin, but not by cholera toxin [11]. The homology between a~ and a i is, as expected, high in sequences involved in the binding of GTP, but low at the carboxyl terminal regions which participate in G-protein receptor and effector (cyclase) interactions. Numerous studies of G i and G~ reveal that the 13"/subunits are functionally interchangeable.
The interaction between the receptor and its effector A fundamental observation in the 1970's was that a cyclase stimulatory receptor from one cell can hybridize with a heterologous adenylyl cyclase system when implanted into the membrane of the recipient cell. The 13-adrenocepter from turkey erythrocytes and the glucagon receptor from rat liver cells can be implanted into Friend erythroleukemia cells and activate the host adenylyl cyclase [12]. Kinetic experiments also demonstrated that two different cyclase-stimulatory receptors (A2 adenosine receptors and 131-adrenoceptors) in the same cell couple to a single pool of adenylyl cyclase molecules and, therefore, to the same pool of G~ proteins [13]. These experiments carried out almost 15 years ago suggested that pharmacologically different receptors possess common structural domains which participate in the interaction between receptor and G s. This hypothesis was indeed verified by recent cloning work [5]. G s seems also to be a universal molecule, since it can couple to a variety of C units from different species. Thus, a functional 13-adrenoceptor-dependent adenylyl cyclase complex can be reconstituted in $49 cyc-membranes using G s from various sources (e.g. wild type $49 lymphoma cells, rabbit liver, turkey erythrocytes and human erythrocytes) [14]. We have shown that the catalytic unit of adenylyl cyclase from two different species, bovine brain and rabbit myocardium, couple equipotently with G s from either turkey erythrocytes or rabbit liver [7]. These results suggest that C is also a highly conserved molecule, at least in the domains which interact with the other components of the signaling complex. The universality of the components has allowed more flexibility in reconstitution experiments in which components from various species have been co-reconstituted to produce functional interactions [7, 8, 9].
Activation of adenylyl cyclase by the receptor Agonist binding to R Sand GTP binding to G S at the as-subunit are both required to induce activation of the catalytic moiety, C. The active state of the system decays concomitantly with the hydrolysis of GTP to GDP and Pi at the G s regulatory site. Replenishment of G s with GTP and the continued presence of hormone at the receptor allow the system to regain its active cyclic adenosine monophosphate (cAMP)-producing state repeatedly. This type of 'on-off' cycle accounts
Signal Transductlon and Adenylyl Cyclase
Vol. 52, No. 26, 1993
2095
well for the properties of hormone-dependent adenylyl cyclase [1] and, in fact, for all processes of effector activation through G-proteins. Hormone-dependent adenylyl cyclase possesses a slow (kcat = 15 min-1) hormone-dependent GTPase activity. This activity has been demonstrated in ~ladrenoceptor-dependent cyclase [15], in glucagon-dependent cyclase, pancreoenzymin-dependent cyclase and in prostaglandin El-dependent cyclase [14], as well as in ~-adrenoceptor G~ reconstituted systems [5]. Thus, non-hydrolyzable GTP analogs, such as GPPNHP and GTPTS generate a constitutively active form of the enzyme, as the 'off' GTPase reaction is blocked [ 16]. Similarly, inhibition of the GTPase reaction by cholera toxin catalyzed ADP-ribosylation of G s activates the cyclase due to the slow-down of the 'off' GTPase reaction [11, 14]. From independent measurements of the rate of enzyme activation by hormone and guanyl nucleotide (the 'on' reaction) and the decay of the cAMP-producing state to its basal state (the 'off' reaction), one can compute the fraction of the total pool of cyclase which is in the active state in the presence of GTP. These measurements suggested to us, as early as 1977, that the receptor-cyclase system can be described by a two-state model [ 17], where the system oscillates between the GTP-active and the GDP-inactive state. In a further study we demonstrated that partial agonists differ from full agonists in that they promote activation of the cyclase with a reduced rate (kon), whereas the GTPase koff rate remains unchanged. The simplest explanation is that a smaller fraction of the receptor attains the active state in the presence of the partial agonist [18]. The alternative explanation is, of course, that partial agonists produce intermediate conformations of the receptor with altered kon, but unchanged koff values. This is a much less pleasing and less likely molecular model.
The collision coupling mechanism Kinetic-mechanistic experiments [16, 19] demonstrated, rather surprisingly at the time, that the role of the receptor is catalytic [ 1]: namely, one agonist-bound receptor can activate numerous adenylyl cyclase molecules. When the number of [3-adrenoceptors on turkey erythrocyte membranes was progressively reduced using a specific ~-adrenoceptor-directed affinity label [20, 21 ], the rate of cyclase activation was proportionally reduced, but the maximum specific activity attainable remained unchanged when hormone and GPPNHP were used. The rate constant of cyclase activation (kon) was found to be linearly dependent on receptor concentration: kon=k[RToTAL]X[Agonist]/(KH+[H]) Where k is the intrinsic rate constant, K H is the receptor-agonist dissociation constant and R.ro. TALis the total receptor concentration in the experiment. This discovery led to the formulation of the 'collision coupling' model [1] where the steps of adenylyl cyclase activation were described as follows: HR+GsC HR-GsC HR-G~*C HR-Gs*C* Gs*C*
~ ---) ~ ~ ~
HR-GsC HR-G~*C(active G~) HR-G~*C*(active C) HR+Gs*C* GsC
where the intermediates HR-GsC, HR-Gs*C and HR-Gs*C* do not accumulate. This predicts: (I) strict first order kinetic of adenylyl cyclase activation by hormone and GPPNHP or GTP [ 19]; (II) linear dependence of the rate constant of activation kon on receptor
2096
Signal Transductlon and Adenylyl Cyclase
Vol. 52, No. 26, 1993
concentration [6, 7, 16, 19] and (III) a Michaelian (non-cooperative) dependence of kon on the hormone concentration [19, 22, 23]. From the hormone dependence curve of IConone can compute receptor-agonist dissociation constants. The numerical values for these constants should be identical to those obtained by direct binding measurements [24]. These predictions were indeed verified experimentally. A large series of agonists and partial agonists were tested and it was found that the better the agonist, the higher the kon, whereas the GTPase turn-off rate was identical for all agonists [19]. Detailed kinetic analysis has conclusively demonstrated [22, 25] that alternative kinetic models [26, 27] suggesting that the G s protein 'shuttles' between the receptor and the cyclase are invalid. These models predict non-first order complex kinetics of adenylyl cyclase activation on receptor concentration. These shuttle models also predict a dependence of the rate of enzyme activation on the concentration of G s and C. Direct biochemical experiments also show that G Sand C are closely associated with each other [28-30]. Conformational changes
The nature of the conformational changes which occur at the receptor, upon agonist binding, are not known. One can, however, speculate about the nature of the structural changes which occur. Since the catecholamine is bound to residues from different transmembrane domains (Fig. 1) [5] it is likely that the binding of the agonist pulls these domains together thus inducing movement at the cytoplasmic domains of the protein which interact with the G s protein. The binding of the catecholamine within a hydrophobic pocket accounts for the high affinity of hydrophobic agonists like propranolol and cyanopindolol to the receptor. Furthermore, an affinity label based on cyanopindolol labels a glycolipid moiety which is intimately associated with the receptor binding site [29]. It has not been determined if the glycolipid moiety plays any role in signal transduction, although arguments in favor of this have been presented [29]. The induced conformational change at G s is manifested by the 'opening' of the nucleotide binding site. As a result of this, GDP is removed from the site and is replaced by incoming GTP. The G-protein in its 'open', but nucleotide-free state has a longer life-time than in the presence of a nucleotide such as GPPNHP [23]. The conversion of the 'open' metastable conformation to the 'closed' conformation, which is unable to exchange the nucleotide, is facilitat-
OH I CHCH2NH3+
OH
-OOC-ASP 113(3)
"
HO-Ser 20415) HO-Ser 207(5)
Fig. 1: The binding of the catcholamine between transmembrane domains. The studies of Dixon and his colleagues revealed that the catecholamine agonist is bound to protein residues from three different transmembrane domains of the protein, quite remote from each other.
Vol. 52, No. 26, 1993
Signal Transductlon and Adenylyl Cyclase
2097
ed by the bound nucleotide. GTP, GPPNHP, and GTPTS facilitate the 'closure', reaction whereas G s can be trapped in the metastable 'open' state for a long time by GMP [23]. The 'open' conformation of G s can also be demonstrated by its increased affinity to the agonist-bound ~adrenoceptor. In the presence of [3-agonist and in the absence of GTP, one can separate an HR~G s (or G~C) complex [30] using molecular sieve chromatography. Upon addition of GTP, Gs decays to its 'closed' conformation, which displays a lower affinity towards HR s, and the complex dissociates.
The role of G-protein subunits in signaling GTPTS in the presence of high Mg 2÷ and lubrol PX induces the irreversible dissociation of the heterotrimeric protein complex: GsGTP7S ---) (~s(GTP7S) + ~7 This finding has led to a model [31] suggesting that hormonal stimulation within the native membrane, in the presence of the natural ligand GTP, also leads to the reversible dissociation of G s. The c~s*(GTP) then associates with the cyclase C and activates the catalytic unit C to form c%*(GTP)C*. Upon hydrolysis of GTP to GDP, %(GDP) dissociates from C and reassociates with 13T (Fig. 2). According to this model (~s'(GTP) binds strongly to C but weakly to 137, whereas the situation is reversed for %(GDP). Accordingly 137-subunits compete with C for %. Hence, elevation of the level of ~7-subunits causes c% to be less accessible to C and therefore
C
~
13"/°%( G D P ) ~ r m ? -
GTP
~'- 137
137(~i (GDP) 0%(GDP)'C
"X
~
"/
.,o' \
\
J o~'s(GTP)
./ " ~'i(GTP) "
/
i / ~ H20 Fig. 2:
The Gs-G i dissociation model.
o~,s(GVTP).C '
~
/
GDP
2098
Signal Transduction and Adenylyl Cyclase
Vol. 52, No. 26, 1993
induces inhibition. When an agonist binds to an inhibitory receptor (Ri), it induces the dissociation of G i to o~i(GTP) and 137, the latter scavenging c~ and thus inhibiting C (Fig. 2) [31, 33]. This mechanism implies that a s shuttles between G~ and the catalytic unit C and that G s and C are physically separate entities within the membrane bilayer. Any mechanism, however, which implies GsC dissociation is not compatible with the kinetic data in native membranes and in reconstituted systems as described above. Furthermore, direct biochemical evidence has demonstrated that the holotrimeric G s protein is permanently associated with C [29, 30] throughout the activation cycle of adenylyl cyclase. Subsequent to activation of the enzyme by hormone and the non-hydrolyzable GTP analog, one can isolate and purify the undissociated species ~s(GPPNHP)137. This has been demonstrated for both the bovine brain adenylyl cyclase system [29] and the turkey erythrocyte system [30]. We have pointed out [30] that other G-protein systems may indeed operate via G-protein dissociation. Indeed, G i type proteins tend to dissociate to their a and 137 subunits more readily. Even in the adenylyl cyclase system, it seems that G i can dissociate upon its activation with an inhibitory receptor (see also discussion below.) M e c h a n i s m of G i action
In spite of the compelling evidence against G s dissociation, the most commonly accepted model for the action of G i is still the dissociation mechanism (Fig. 2). The fact that [37 subunits cannot play the major role assigned to them has also gained support from recent studies [34]. Our findings and many other observations summarized by us [29, 30] have led us to suggest an alternative model for cyclase inhibition [29, 35]. The model suggests that Gi*GTP affects the cyclase only in its G~ activated state through the Goq subunit: HR s G~(GDP)C
G~(GTP) ,
G~(GTP)C* active cyclase
- > Gs(GTP)C*Gi(GTP) inhibited cyclase
For the inhibitory effect one requires, however, the oligomeric state of the G i. The site on C which interacts with G i is formed only when C is activated by G s. Since G i, in contrast to G s, tends to dissociate to its subunits, it is expected that the addition of 137 will have a strong inhibitory effect on the enzyme, by pushing the equilibrium to the associated state of G i, which is the inhibitory species. This type of model takes into account all the experimental observations made on the hormone-dependent adenylyl cyclase over the last 20 years. References
1. Levitzki A. From epinephrine to cyclic AMP. Science 1988; 241: 800-6. 2. Birnbaumer L, Abramowitz J, Brown AM. Receptor-effector coupling by G proteins. Biochim Biophys Acta Rev Cancer 1990; 103: 163-224. 3. Krupinski J, Coussen F, Bakalyar HA et al. Adenylyl cyclase amino acid sequence: possible channelor transporter-like structure. Science 1989; 244:1558~4. 4. Bourne HR, Sanders DA, McCormick F. The GTPase superfamily: a conserved switch for diverse cell functions. Nature 1990; 348: 125-32. 5. Lefkowitz RY, Caron MG. Adrenergic receptors - - models for the study of receptors coupled to guanine nucleotide regulatory proteins. J Biol Chem 1988; 263: 4993-6. 6. Hekman M, Feder D, Keenan AK et al. Reconstitution of ~-adrenergic receptor with components of adenylate cyclase. Embo J 1984; 3: 3339-45.
Vol. 52, No. 26, 1993
Signal Transductlon and Adenylyl Cyclase
2099
7. Feder D, Im M-J, Klein HW et al. Reconstitution of 131-adrenoceptor-dependent adenylate cyclase from purified components. Embo J 1986; 5: 1509-14. 8. Feder D, Im M-J, Pfeuffer T et al. The hormonal regulation of adenylate cyclase. Biochem Soc Symp 1986; 52:145-51. 9. May DC, Ross EM, Gilman AG, Smigel MD. Reconstitution of catecholamine-stimulated adenylate cyclase activity using three purified proteins. J Biol Chem 1985; 260: 15829-33. 10. Casey PJ, Gilman AG. G protein involvement in receptor-effector coupling. J Biol Chem 1988; 263: 2577-80. l l . Gilman AG. G proteins: Transducers of receptor-generated signals. Ann Rev Biochem 1987; 56: 615-49. 12. Schramm M. Transfer of glucagon receptor from liver membranes to a foreign adenylate cyclase by a membrane fusion procedure. Proc Natl Acad Sci USA 1979; 76:1174-8. 13. Tolkovsky AM, Levitzki A. Coupling of a single adenylate cyclase to two receptors: Adenosine and catchecholamine. Biochemistry 1978; 17:3811-7. 14. Levitzki A. [3-adrenergic receptors and their mode of coupling to adenylate cyclase. Physiol Rev 1986; 66:819--42. 15. Cassel D, Selinger Z. Activation of turkey erythrocyte adenylate cyclase and blocking of the catecholamine stimulated GTPase by guanosine 5'-(y-thio) triphosphate. Biochem Biophys Res Commun 1977; 7: 868-73. 16. Arad H, Rimon G, Levitzki A. The reversal of the Gpp (NH)p-activated state of adenylate cyclase by GTP and hormone is by the 'collision coupling' mechanism. J Biol Chem 1981; 25: 1593-7. 17. Levitzki A. The role of GTP in the activation of adenylate cyclase. Biochem Biophys Res Commun 1977; 74:1154-9. 18. Arad H, Levitzki A. The mechanism of partial agonism in the II-receptor dependent adenylate cyclase of turkey erythrocytes. Mol Pharmacol 1979; 16: 748-56. 19. Tolkovsky AM, Levitzki A. Mode of coupling between the II-adrenergic receptor and adenylate cyclase in turkey erythrocytes. Biochemistry 1978; 17:3795-810. 20. Atlas D, Steer ML, Levitzki A. Affinity label for the 13-adrenergic receptor in turkey erythrocytes. Proc Natl Acad Sci USA 1976; 73: 1921-5. 21. Atlas D, Levitzki A. Tentative identification of the ~-adrenoreceptor subunits. Nature 1978; 272: 5651-2. 22. Tolkovsky AM, Levitzki A. Theories and predictions of models describing sequential interactions between the receptor, the GTP regulatory unit and the catalytic unit of hormone-dependent adenylate cyclase. J Cyclic Nucl Res 1991; 1: 139-50. 23. Braun S, Tolkovsky AM, Levitzki A. Mechanism of control of the turkey erythrocyte II-adrenoceptor dependent adenylate cyclase by guanyl nucleotides: A minimum model. J Cyclic Nucl Res 1982; 8: 133-47. 24. Levitzki A, Sevilla N, Atlas D, Steer ML. Ligand specificity and characteristics of the I~-adrenergic receptor in turkey erythrocyte plasma membrane. J Mol Biol 1975; 91: 35-53. 25. Tolkovsky AM, Braun S, Levitzki A. Kinetics of interaction between 13-receptors, GTP protein, and the catalytic unit of turkey erythrocyte adenylate cyclase. Proc Natl Acad Sci USA 1982; 79: 213-7. 26. Citri Y, Schramm M. Resolution, reconstitution and kinetics of the primary action of a hormone receptor. Nature 1980; 287: 297-300. 27. De Lean A, Stadel JM, Lefkowitz RJ. A ternary complex model explains the agonist-specific binding properties of the adenylate cyclase-coupled I~-adrenergic receptor. J Biol Chem 1980; 255:7108-17. 28. Arad H, Rosenbusch J, Levitzki A. Stimulatory GTP regulatory unit Ns and the catalytic unit of adenylate cyclase are tightly associated. Mechanistic consequences. Proc Natl Acad Sci USA 1984; 81 : 6579-83. 29. Bar-Sinai A, Aldouby Y, Chorev M, Levitzki A. Association of the I~-adrenoceptor with a specific lipid component. Embo J 1986; 5:1175-80. 30. Bar-Sinai A, Marbach I, Short RGL, Levitzki A. The GppNHp-activated adenylyl cyclase complex from turkey erythrocyte membranes can be isolated with its [37 subunits. Eur J Biochem. In press. 31. Gilman AG. G-proteins and dual control of adenylate cyclase. Cell 1984; 36: 577-9.
2100
Signal Transductlon and Adenylyl Cyclase
Vol. 52, No. 26, 1993
32. Katada T, Bokoch G, Smigel MD et al. The inhibitory guanine nucleotide-bindingregulatory component of adenylate cyclase - - subunit dissociation and the inhibition of adenylate cyclase in $49 lymphoma cyc- and wild type membranes. J Biol Chem 1984; 259: 3586-95. 33. Katada T, Northup JK, Bokoch GM et al. The inhibitory guanine nucleotide-bindingregulatory component of adenylate cyclase - - subunit dissociation and guanine nucleotide-dependent hormonal inhibition. J Biol Chem 1984; 259: 3578-85. 34. Hildebrandt JD, Kohnken RE. Hormone inhibition of adenylyl cyclase - - differences in the mechanisms for inhibition by hormones and G protein 137.J Biol Chem 1990; 265: 9825-30. 35. Levitzki A, Bar-Sinai A. The regulation of adenylyl cyclase by receptor operated G proteins. Pharmacol Ther 1991; 50: 271-83.
52, pp. 2093-2100
The signal transduction between cyclase
Pergamon Press
-receptors and adenylyl
A. Levitzki, I. Marbach and A. Bar-Sinai
Department of Biological Chemistry, The Alexander Silberman Institute of Life Sciences, Hebrew University of Jerusalem, Jerusalem, Israel
Summary The 13-adrenergic receptor-dependent adenylyl cyclase system is the most extensively studied G-protein-coupled system. Studies of the coupling between the receptor and effector can provide an insight into the nature of all these systems in general. In the activation of adenylyl cyclase by the receptor, the binding of an agonist to the stimulatory receptor (Rs) and the binding of GTP to the G-protein (Gs) are both required to activate the catalytic moiety (C). The active state decays as GTP is hydrolysed to GDP and inorganic phosphate (Pi), but reactivation occurs as GTP is replenished. The receptor acts as a catalyst, i.e. one agonist-bound receptor can activate numerous adenylyl cyclase molecules. Kinetic studies led to the formulation of the 'collision coupling' model of receptor activation and show that G s protein does not shuttle between the receptor and cyclase. The G~ protein appears to undergo conformational changes between an 'open' state in which it can bind with GTP, and a 'closed' state unable to achieve this binding. This mechanism of activation does not involve the dissociation of G s or of G i. A model which fits the experimental data suggests that Gi*GTP affects cyclase only in its Gs-activated state via the G~ 1 subunit, but that the oligomeric state of G i is required for inhibition. The site on C which interacts with G i is formed only when C is activated by G s.
Introduction The [~-adrenergic receptor-dependent adenylyl cyclase is still the most extensively studied Gprotein coupled system and the one we understand best. Thus understanding the details of receptor to effector coupling in this system is a useful model which can help investigators learn about other receptor G-protein systems. In the receptor-regulated adenylyl cyclase, all the protein components have been characterized, purified and reconstituted [ 1, 2]. The members of the [3-adrenoceptor family have all been found to possess seven putative transmembrane domains and, therefore, similar three dimensional structure and overall conformation. The effectors of this family of receptor G-protein coupled systems are less well characterized, with only the adenylyl cyclase having been recently cloned [3]. Since we are aware o f at least 16 different heterotrimeric G-proteins [4] it is clear that many effectors are waiting to be discovered or characterized more fully. The receptor-regulated adenylyl cyclase is composed o f 5 functional units: a) A stimulatory receptor Rs which binds the stimulatory hormone or the neurotransmitter. The [~-adrenergic receptor is the prototype and best characterized [1, 5]. This receptor is a transmembrane glycoprotein with 7 putative transmembrane-spanning sequences. It is still the only member of this large family of G-protein coupled receptors which has been purified and successfully reconsti-
Copyright
©
0024-3205/93 $6.00 + .00 1993 Pergamon Press Ltd. All rights reserved.
2094
Signal Transduction and Adenylyl Cyclase
Vol. 52, No. 26, 1993
tuted with other pure components of the adenylyl cyclase system [6-9]. b) The G-protein, G s, which binds GTP and activates the adenylyl cyclase enzyme, C. G s is composed of 3 subunits: a s, which possesses the GTP binding site and is the target for cholera toxin catalyzed ADP-ribosylation, and the 13and y subunits, which are closely associated with each other [10]. The heterotrimeric Gs-protein is localized in the inner leaflet of the plasma membrane, c) The catalytic unit, C, is a hydrophobic protein with twelve putative transmembrane sequences with the catalytic site on the cytoplasmic side of the membrane [3]. The system would be incomplete if we discussed only the activating branch of the adenylyl cyclase system. Thus we shall consider d), the inhibitory receptors, R i, which bind inhibitory ligands and, like the 13-adrenergic receptor, span the membrane 7 times; an example of this is the a2-adrenergic receptor, e) The inhibitory GTP-binding protein, G i. This protein, like G~, is composed of three subunits: a i, 13and y, where the 13ycomplex is highly similar or identical to the 13yin G~ [10]. The subunit cg is homologous to a s but is a substrate for ADP-ribosylation catalyzed by pertussis toxin, but not by cholera toxin [11]. The homology between a~ and a i is, as expected, high in sequences involved in the binding of GTP, but low at the carboxyl terminal regions which participate in G-protein receptor and effector (cyclase) interactions. Numerous studies of G i and G~ reveal that the 13"/subunits are functionally interchangeable.
The interaction between the receptor and its effector A fundamental observation in the 1970's was that a cyclase stimulatory receptor from one cell can hybridize with a heterologous adenylyl cyclase system when implanted into the membrane of the recipient cell. The 13-adrenocepter from turkey erythrocytes and the glucagon receptor from rat liver cells can be implanted into Friend erythroleukemia cells and activate the host adenylyl cyclase [12]. Kinetic experiments also demonstrated that two different cyclase-stimulatory receptors (A2 adenosine receptors and 131-adrenoceptors) in the same cell couple to a single pool of adenylyl cyclase molecules and, therefore, to the same pool of G~ proteins [13]. These experiments carried out almost 15 years ago suggested that pharmacologically different receptors possess common structural domains which participate in the interaction between receptor and G s. This hypothesis was indeed verified by recent cloning work [5]. G s seems also to be a universal molecule, since it can couple to a variety of C units from different species. Thus, a functional 13-adrenoceptor-dependent adenylyl cyclase complex can be reconstituted in $49 cyc-membranes using G s from various sources (e.g. wild type $49 lymphoma cells, rabbit liver, turkey erythrocytes and human erythrocytes) [14]. We have shown that the catalytic unit of adenylyl cyclase from two different species, bovine brain and rabbit myocardium, couple equipotently with G s from either turkey erythrocytes or rabbit liver [7]. These results suggest that C is also a highly conserved molecule, at least in the domains which interact with the other components of the signaling complex. The universality of the components has allowed more flexibility in reconstitution experiments in which components from various species have been co-reconstituted to produce functional interactions [7, 8, 9].
Activation of adenylyl cyclase by the receptor Agonist binding to R Sand GTP binding to G S at the as-subunit are both required to induce activation of the catalytic moiety, C. The active state of the system decays concomitantly with the hydrolysis of GTP to GDP and Pi at the G s regulatory site. Replenishment of G s with GTP and the continued presence of hormone at the receptor allow the system to regain its active cyclic adenosine monophosphate (cAMP)-producing state repeatedly. This type of 'on-off' cycle accounts
Signal Transductlon and Adenylyl Cyclase
Vol. 52, No. 26, 1993
2095
well for the properties of hormone-dependent adenylyl cyclase [1] and, in fact, for all processes of effector activation through G-proteins. Hormone-dependent adenylyl cyclase possesses a slow (kcat = 15 min-1) hormone-dependent GTPase activity. This activity has been demonstrated in ~ladrenoceptor-dependent cyclase [15], in glucagon-dependent cyclase, pancreoenzymin-dependent cyclase and in prostaglandin El-dependent cyclase [14], as well as in ~-adrenoceptor G~ reconstituted systems [5]. Thus, non-hydrolyzable GTP analogs, such as GPPNHP and GTPTS generate a constitutively active form of the enzyme, as the 'off' GTPase reaction is blocked [ 16]. Similarly, inhibition of the GTPase reaction by cholera toxin catalyzed ADP-ribosylation of G s activates the cyclase due to the slow-down of the 'off' GTPase reaction [11, 14]. From independent measurements of the rate of enzyme activation by hormone and guanyl nucleotide (the 'on' reaction) and the decay of the cAMP-producing state to its basal state (the 'off' reaction), one can compute the fraction of the total pool of cyclase which is in the active state in the presence of GTP. These measurements suggested to us, as early as 1977, that the receptor-cyclase system can be described by a two-state model [ 17], where the system oscillates between the GTP-active and the GDP-inactive state. In a further study we demonstrated that partial agonists differ from full agonists in that they promote activation of the cyclase with a reduced rate (kon), whereas the GTPase koff rate remains unchanged. The simplest explanation is that a smaller fraction of the receptor attains the active state in the presence of the partial agonist [18]. The alternative explanation is, of course, that partial agonists produce intermediate conformations of the receptor with altered kon, but unchanged koff values. This is a much less pleasing and less likely molecular model.
The collision coupling mechanism Kinetic-mechanistic experiments [16, 19] demonstrated, rather surprisingly at the time, that the role of the receptor is catalytic [ 1]: namely, one agonist-bound receptor can activate numerous adenylyl cyclase molecules. When the number of [3-adrenoceptors on turkey erythrocyte membranes was progressively reduced using a specific ~-adrenoceptor-directed affinity label [20, 21 ], the rate of cyclase activation was proportionally reduced, but the maximum specific activity attainable remained unchanged when hormone and GPPNHP were used. The rate constant of cyclase activation (kon) was found to be linearly dependent on receptor concentration: kon=k[RToTAL]X[Agonist]/(KH+[H]) Where k is the intrinsic rate constant, K H is the receptor-agonist dissociation constant and R.ro. TALis the total receptor concentration in the experiment. This discovery led to the formulation of the 'collision coupling' model [1] where the steps of adenylyl cyclase activation were described as follows: HR+GsC HR-GsC HR-G~*C HR-Gs*C* Gs*C*
~ ---) ~ ~ ~
HR-GsC HR-G~*C(active G~) HR-G~*C*(active C) HR+Gs*C* GsC
where the intermediates HR-GsC, HR-Gs*C and HR-Gs*C* do not accumulate. This predicts: (I) strict first order kinetic of adenylyl cyclase activation by hormone and GPPNHP or GTP [ 19]; (II) linear dependence of the rate constant of activation kon on receptor
2096
Signal Transductlon and Adenylyl Cyclase
Vol. 52, No. 26, 1993
concentration [6, 7, 16, 19] and (III) a Michaelian (non-cooperative) dependence of kon on the hormone concentration [19, 22, 23]. From the hormone dependence curve of IConone can compute receptor-agonist dissociation constants. The numerical values for these constants should be identical to those obtained by direct binding measurements [24]. These predictions were indeed verified experimentally. A large series of agonists and partial agonists were tested and it was found that the better the agonist, the higher the kon, whereas the GTPase turn-off rate was identical for all agonists [19]. Detailed kinetic analysis has conclusively demonstrated [22, 25] that alternative kinetic models [26, 27] suggesting that the G s protein 'shuttles' between the receptor and the cyclase are invalid. These models predict non-first order complex kinetics of adenylyl cyclase activation on receptor concentration. These shuttle models also predict a dependence of the rate of enzyme activation on the concentration of G s and C. Direct biochemical experiments also show that G Sand C are closely associated with each other [28-30]. Conformational changes
The nature of the conformational changes which occur at the receptor, upon agonist binding, are not known. One can, however, speculate about the nature of the structural changes which occur. Since the catecholamine is bound to residues from different transmembrane domains (Fig. 1) [5] it is likely that the binding of the agonist pulls these domains together thus inducing movement at the cytoplasmic domains of the protein which interact with the G s protein. The binding of the catecholamine within a hydrophobic pocket accounts for the high affinity of hydrophobic agonists like propranolol and cyanopindolol to the receptor. Furthermore, an affinity label based on cyanopindolol labels a glycolipid moiety which is intimately associated with the receptor binding site [29]. It has not been determined if the glycolipid moiety plays any role in signal transduction, although arguments in favor of this have been presented [29]. The induced conformational change at G s is manifested by the 'opening' of the nucleotide binding site. As a result of this, GDP is removed from the site and is replaced by incoming GTP. The G-protein in its 'open', but nucleotide-free state has a longer life-time than in the presence of a nucleotide such as GPPNHP [23]. The conversion of the 'open' metastable conformation to the 'closed' conformation, which is unable to exchange the nucleotide, is facilitat-
OH I CHCH2NH3+
OH
-OOC-ASP 113(3)
"
HO-Ser 20415) HO-Ser 207(5)
Fig. 1: The binding of the catcholamine between transmembrane domains. The studies of Dixon and his colleagues revealed that the catecholamine agonist is bound to protein residues from three different transmembrane domains of the protein, quite remote from each other.
Vol. 52, No. 26, 1993
Signal Transductlon and Adenylyl Cyclase
2097
ed by the bound nucleotide. GTP, GPPNHP, and GTPTS facilitate the 'closure', reaction whereas G s can be trapped in the metastable 'open' state for a long time by GMP [23]. The 'open' conformation of G s can also be demonstrated by its increased affinity to the agonist-bound ~adrenoceptor. In the presence of [3-agonist and in the absence of GTP, one can separate an HR~G s (or G~C) complex [30] using molecular sieve chromatography. Upon addition of GTP, Gs decays to its 'closed' conformation, which displays a lower affinity towards HR s, and the complex dissociates.
The role of G-protein subunits in signaling GTPTS in the presence of high Mg 2÷ and lubrol PX induces the irreversible dissociation of the heterotrimeric protein complex: GsGTP7S ---) (~s(GTP7S) + ~7 This finding has led to a model [31] suggesting that hormonal stimulation within the native membrane, in the presence of the natural ligand GTP, also leads to the reversible dissociation of G s. The c~s*(GTP) then associates with the cyclase C and activates the catalytic unit C to form c%*(GTP)C*. Upon hydrolysis of GTP to GDP, %(GDP) dissociates from C and reassociates with 13T (Fig. 2). According to this model (~s'(GTP) binds strongly to C but weakly to 137, whereas the situation is reversed for %(GDP). Accordingly 137-subunits compete with C for %. Hence, elevation of the level of ~7-subunits causes c% to be less accessible to C and therefore
C
~
13"/°%( G D P ) ~ r m ? -
GTP
~'- 137
137(~i (GDP) 0%(GDP)'C
"X
~
"/
.,o' \
\
J o~'s(GTP)
./ " ~'i(GTP) "
/
i / ~ H20 Fig. 2:
The Gs-G i dissociation model.
o~,s(GVTP).C '
~
/
GDP
2098
Signal Transduction and Adenylyl Cyclase
Vol. 52, No. 26, 1993
induces inhibition. When an agonist binds to an inhibitory receptor (Ri), it induces the dissociation of G i to o~i(GTP) and 137, the latter scavenging c~ and thus inhibiting C (Fig. 2) [31, 33]. This mechanism implies that a s shuttles between G~ and the catalytic unit C and that G s and C are physically separate entities within the membrane bilayer. Any mechanism, however, which implies GsC dissociation is not compatible with the kinetic data in native membranes and in reconstituted systems as described above. Furthermore, direct biochemical evidence has demonstrated that the holotrimeric G s protein is permanently associated with C [29, 30] throughout the activation cycle of adenylyl cyclase. Subsequent to activation of the enzyme by hormone and the non-hydrolyzable GTP analog, one can isolate and purify the undissociated species ~s(GPPNHP)137. This has been demonstrated for both the bovine brain adenylyl cyclase system [29] and the turkey erythrocyte system [30]. We have pointed out [30] that other G-protein systems may indeed operate via G-protein dissociation. Indeed, G i type proteins tend to dissociate to their a and 137 subunits more readily. Even in the adenylyl cyclase system, it seems that G i can dissociate upon its activation with an inhibitory receptor (see also discussion below.) M e c h a n i s m of G i action
In spite of the compelling evidence against G s dissociation, the most commonly accepted model for the action of G i is still the dissociation mechanism (Fig. 2). The fact that [37 subunits cannot play the major role assigned to them has also gained support from recent studies [34]. Our findings and many other observations summarized by us [29, 30] have led us to suggest an alternative model for cyclase inhibition [29, 35]. The model suggests that Gi*GTP affects the cyclase only in its G~ activated state through the Goq subunit: HR s G~(GDP)C
G~(GTP) ,
G~(GTP)C* active cyclase
- > Gs(GTP)C*Gi(GTP) inhibited cyclase
For the inhibitory effect one requires, however, the oligomeric state of the G i. The site on C which interacts with G i is formed only when C is activated by G s. Since G i, in contrast to G s, tends to dissociate to its subunits, it is expected that the addition of 137 will have a strong inhibitory effect on the enzyme, by pushing the equilibrium to the associated state of G i, which is the inhibitory species. This type of model takes into account all the experimental observations made on the hormone-dependent adenylyl cyclase over the last 20 years. References
1. Levitzki A. From epinephrine to cyclic AMP. Science 1988; 241: 800-6. 2. Birnbaumer L, Abramowitz J, Brown AM. Receptor-effector coupling by G proteins. Biochim Biophys Acta Rev Cancer 1990; 103: 163-224. 3. Krupinski J, Coussen F, Bakalyar HA et al. Adenylyl cyclase amino acid sequence: possible channelor transporter-like structure. Science 1989; 244:1558~4. 4. Bourne HR, Sanders DA, McCormick F. The GTPase superfamily: a conserved switch for diverse cell functions. Nature 1990; 348: 125-32. 5. Lefkowitz RY, Caron MG. Adrenergic receptors - - models for the study of receptors coupled to guanine nucleotide regulatory proteins. J Biol Chem 1988; 263: 4993-6. 6. Hekman M, Feder D, Keenan AK et al. Reconstitution of ~-adrenergic receptor with components of adenylate cyclase. Embo J 1984; 3: 3339-45.
Vol. 52, No. 26, 1993
Signal Transductlon and Adenylyl Cyclase
2099
7. Feder D, Im M-J, Klein HW et al. Reconstitution of 131-adrenoceptor-dependent adenylate cyclase from purified components. Embo J 1986; 5: 1509-14. 8. Feder D, Im M-J, Pfeuffer T et al. The hormonal regulation of adenylate cyclase. Biochem Soc Symp 1986; 52:145-51. 9. May DC, Ross EM, Gilman AG, Smigel MD. Reconstitution of catecholamine-stimulated adenylate cyclase activity using three purified proteins. J Biol Chem 1985; 260: 15829-33. 10. Casey PJ, Gilman AG. G protein involvement in receptor-effector coupling. J Biol Chem 1988; 263: 2577-80. l l . Gilman AG. G proteins: Transducers of receptor-generated signals. Ann Rev Biochem 1987; 56: 615-49. 12. Schramm M. Transfer of glucagon receptor from liver membranes to a foreign adenylate cyclase by a membrane fusion procedure. Proc Natl Acad Sci USA 1979; 76:1174-8. 13. Tolkovsky AM, Levitzki A. Coupling of a single adenylate cyclase to two receptors: Adenosine and catchecholamine. Biochemistry 1978; 17:3811-7. 14. Levitzki A. [3-adrenergic receptors and their mode of coupling to adenylate cyclase. Physiol Rev 1986; 66:819--42. 15. Cassel D, Selinger Z. Activation of turkey erythrocyte adenylate cyclase and blocking of the catecholamine stimulated GTPase by guanosine 5'-(y-thio) triphosphate. Biochem Biophys Res Commun 1977; 7: 868-73. 16. Arad H, Rimon G, Levitzki A. The reversal of the Gpp (NH)p-activated state of adenylate cyclase by GTP and hormone is by the 'collision coupling' mechanism. J Biol Chem 1981; 25: 1593-7. 17. Levitzki A. The role of GTP in the activation of adenylate cyclase. Biochem Biophys Res Commun 1977; 74:1154-9. 18. Arad H, Levitzki A. The mechanism of partial agonism in the II-receptor dependent adenylate cyclase of turkey erythrocytes. Mol Pharmacol 1979; 16: 748-56. 19. Tolkovsky AM, Levitzki A. Mode of coupling between the II-adrenergic receptor and adenylate cyclase in turkey erythrocytes. Biochemistry 1978; 17:3795-810. 20. Atlas D, Steer ML, Levitzki A. Affinity label for the 13-adrenergic receptor in turkey erythrocytes. Proc Natl Acad Sci USA 1976; 73: 1921-5. 21. Atlas D, Levitzki A. Tentative identification of the ~-adrenoreceptor subunits. Nature 1978; 272: 5651-2. 22. Tolkovsky AM, Levitzki A. Theories and predictions of models describing sequential interactions between the receptor, the GTP regulatory unit and the catalytic unit of hormone-dependent adenylate cyclase. J Cyclic Nucl Res 1991; 1: 139-50. 23. Braun S, Tolkovsky AM, Levitzki A. Mechanism of control of the turkey erythrocyte II-adrenoceptor dependent adenylate cyclase by guanyl nucleotides: A minimum model. J Cyclic Nucl Res 1982; 8: 133-47. 24. Levitzki A, Sevilla N, Atlas D, Steer ML. Ligand specificity and characteristics of the I~-adrenergic receptor in turkey erythrocyte plasma membrane. J Mol Biol 1975; 91: 35-53. 25. Tolkovsky AM, Braun S, Levitzki A. Kinetics of interaction between 13-receptors, GTP protein, and the catalytic unit of turkey erythrocyte adenylate cyclase. Proc Natl Acad Sci USA 1982; 79: 213-7. 26. Citri Y, Schramm M. Resolution, reconstitution and kinetics of the primary action of a hormone receptor. Nature 1980; 287: 297-300. 27. De Lean A, Stadel JM, Lefkowitz RJ. A ternary complex model explains the agonist-specific binding properties of the adenylate cyclase-coupled I~-adrenergic receptor. J Biol Chem 1980; 255:7108-17. 28. Arad H, Rosenbusch J, Levitzki A. Stimulatory GTP regulatory unit Ns and the catalytic unit of adenylate cyclase are tightly associated. Mechanistic consequences. Proc Natl Acad Sci USA 1984; 81 : 6579-83. 29. Bar-Sinai A, Aldouby Y, Chorev M, Levitzki A. Association of the I~-adrenoceptor with a specific lipid component. Embo J 1986; 5:1175-80. 30. Bar-Sinai A, Marbach I, Short RGL, Levitzki A. The GppNHp-activated adenylyl cyclase complex from turkey erythrocyte membranes can be isolated with its [37 subunits. Eur J Biochem. In press. 31. Gilman AG. G-proteins and dual control of adenylate cyclase. Cell 1984; 36: 577-9.
2100
Signal Transductlon and Adenylyl Cyclase
Vol. 52, No. 26, 1993
32. Katada T, Bokoch G, Smigel MD et al. The inhibitory guanine nucleotide-bindingregulatory component of adenylate cyclase - - subunit dissociation and the inhibition of adenylate cyclase in $49 lymphoma cyc- and wild type membranes. J Biol Chem 1984; 259: 3586-95. 33. Katada T, Northup JK, Bokoch GM et al. The inhibitory guanine nucleotide-bindingregulatory component of adenylate cyclase - - subunit dissociation and guanine nucleotide-dependent hormonal inhibition. J Biol Chem 1984; 259: 3578-85. 34. Hildebrandt JD, Kohnken RE. Hormone inhibition of adenylyl cyclase - - differences in the mechanisms for inhibition by hormones and G protein 137.J Biol Chem 1990; 265: 9825-30. 35. Levitzki A, Bar-Sinai A. The regulation of adenylyl cyclase by receptor operated G proteins. Pharmacol Ther 1991; 50: 271-83.