Programmable messengers: a new theory of hormone action M. Rodbell Many hormone receptors are linked to GTP-regulatory proteins in membranes. When these proteins are activated by hormones and GTP, the a-subunits are released from the membrane as soluble proteins. It is proposed that these «-subunits are modified by kinases, proteases and other protein-modifying enzymes to give new forms with differing functions. This provides a way of explaining the multiple actions of a hormone on its target cell, and the released a-subunits of GTP-regulatory proteins can be called ‘programmable messengers’. Two ideas have dominated the field of signal transduction over the past 25 years. One is that hormone/neuro- transmitter receptors interact with vari- ous effector enzymes in the plasma membrane to generate signals in the form of small molecules. The classical example is the receptor-controlled adenylate cyclase system in eukaryotic cells. The other is that receptors exist either in membranes or in the cytosol as ‘mobile’ elements which, when com- bined with the activating hormone, induce the receptor to collide with or move to the site(s) of the effector sys- tems. Examples of theories that have evolved from the mobile-receptor theory are the ‘collision-coupling’! and ‘two-step? theories proposed for the coupling of B-adrenergic receptors to the adenylate cyclase system. Another example is the estrogen receptor; it has been thought that the receptor first reacts with the steroid in a cytosolic compartment, and that the activated receptor then enters the nucleus where it regulates gene expression. There is ample evidence that cyclic AMP and other small molecules (cyclic GMP and inositol trisphosphates are recent examples) mediate some of the effects of hormones. The question is M. Rodbell is at the National Institute of Environ- mental Health Sciences, Research Triangle Park, NC 27709, USA. whether the pleiotypical responses induced by a hormone are due solely to any of these molecules. If not, what type of molecule might be more closely linked to receptors that could serve as primary messengers of hormone action” As for the concept of receptor mobility, there is evidence that membrane recep- tors can be induced by agonists to move about in the plane of the membrane. However, there is no compelling evi- dence that mobility is necessary or causal for signal transduction to take place. Indeed, there is a report that increasing the fluid environment to enhance receptor mobility in mem- branes is detrimental to hormone action’, For the estrogen receptor, recent studies indicate that most of the receptors are bound to the nuclear matrix prior to their occupation by hor- mone; receptor release into the cytosolic compartment is an artifact of the methods used for isolating the nucleus’. This article proposes an alternative view of the function of membrane recep- tors and develops a logical framework ' for a theory that the primary messengers of hormones acting on membrane recep- tors are proteins that bind and degrade GTP. These are the so-called GTP-reg- ulatory proteins (G) that are linked to numerous receptor types in eukaryotic cells. The fundamental aspects were presented five years ago in a theory called ‘Disaggregation Theory of Hor- mone Action’*. This theory is now extended and modified in the light of information acquired recently. The disaggregation theory Briefly, this theory suggests that vari- ous classes of receptors are complexed with a family of oligomeric GTP-regula- tory proteins. When the receptors are occupied by agonists and the G units by GTP, the oligomers dissociate into monomers. In the process, the receptors are transformed from a high affinity state when they can bind physiological concentrations of hormones, into a low affinity state in which they are no longer active. At the same time, the G units are transformed to a ‘monomeric’ structure that reacts specifically with an effector unit (E) such as adenylate cyclase. The theory is thermodynamically sound®; it explains the apparent paradox of recep- tors undergoing transitions from high to low affinity states during concerted activation of G by hormone and GTP; it explains the findings of target analysis that the ground-state structure of recep- tors coupled to G exhibits a much higher molecular weight than the activated adenylate cyclase. This theory predicts that the putative monomeric form of G is the primary messenger of hormone action, whereas the product of the effec- tor unit(s) is a secondary signal. G units are oligomeric proteins In recent years, G units have been purified and structurally analysed’. It is now clear that G units coupled to rho- dopsin (termed transducin) and those coupled to receptors (R) that stimulate or inhibit adenylate cyclase (termed G, and G,, respectively), and a newly dis- covered G unit of unknown action (termed G,) are composed of three dis- tinct protein subunits, only one of which, the a-unit, binds GTP. The type of a-subunit coupled depends on the type of G unit (and associated R) to which it is attached. The other two sub- TIBS -— November 1985 462 HE* + cyc” HE HE cyc™ + cyc” Adenylate cyclase 92 0.9 0.1 0.8 (pmol/min) Cholera toxin + - + + + + %5 NAD Pertussis toxin + + - - - + pb NAD Fig. 1. Effects of pretreatment of human erythrocyte ghosts (HE) with pertussis toxin + NAD (HE*) on levels of adenylate cyclase activity and levels of «, subunit transferred to S49 lymphoma cyc— membranes. HE and cyc— membranes were co-incubated for 15 min at 30°C in presence of 0.1 mma Gpp(NH)p + 10 mm MgCl, The mixtures were layered over 33% sucrose and centrifuged for 20 min at 30000 = g. The upper layer containing only cyc— membranes was assayed for adenylate cyclase activity (with 10 um Gpp(NH)p, 5 mm MgCl, 50 um ATP). Cyc— membranes were also treated with either cholera toxin or pertussis toxin, or both in presence of [2P] NAD. Membranes were extracted and extracts electrophoresed (PAGE) for separation of a, (43 kDa) and a, (39 kDa) subunits, followed by autoradiography. units, designated B and y, are highly conserved proteins — they are found in many cell types and species and have similar if not identical structures ir- respective of the type of attached a-sub- unit. Coupling to receptors In reconstitution studies with purified components, G units interact with receptors when incorporated into lipid vesicles. The complexes formed exhibit the properties of R-G complexes in native membranes, i.e. hormones induce binding and degradation of GTP; R can take different affinity states, the higher affinity presumably linked to G; and GTP decreases the affinity of R for agonists®*®, Kinetically, the process of activation of G by agonists does not require hormone-induced associations between R and G, suggesting that the pre-formed complexes are the active species. Thus, there is no need to invoke the theories suggesting that: hormones act by promoting such associations. Reconstitution studies with rhodopsin and f-adrenergic receptors indicate that all three subunits of G are required for coupling between receptors and G. It follows that factors that disrupt the G unit must functionally uncouple R from this unit. Disaggregation of G oligomers In their purified, detergent-soluble form, G units dissociate when incubated with non-hydrolysable analogs of GTP (e.g. Gpp(NH)p or GTP-y-S) or with aluminum fluoride in the presence of high concentrations of Mg?+ (Ref. 10). GTP is probably ineffective because GTP is hydrolysed to GDP as soon as the a unit dissociates, and the subunits re-aggregate to form the holoprotein. This cyclical behaviour of the trimer may explain why GTP is relatively inef- fective in the receptor-coupled systems within native membranes in the absence of hormones. The observation most relevant to the ‘disaggregation’ theory is that G units are oligomers which, in the absence of activating ligands, cannot dissociate to release the ‘active’ GTP binding a-sub- unit. In this sense, the postulated mono- mer of G is equivalent to the activated a-subunit(s). Theoretically, activation of the R-G complex by concerted actions of hormone and GTP should lead to two interrelated phenomena: release of acti- vated free asubunits and conversion of receptors to a lower affinity, inactive form of R. Until a re-associates with the B/y subunits, R is de-sensitized, even if it is still linked to the B/y subunits. o-Subunits are released from membranes Proof that a-subunits are released from R-G complexes in membranes by actions of hormones and GTP has been lacking. A possible means of testing release from native membranes arose from an apparently peculiar finding: co-incubation of membranes containing G, units (rat liver, RL, and human erythrocyte ghosts, HE), with mem- branes lacking this unit (isolated from a variant termed cyc— of S49 mouse lym- phoma cells) rendered the cyc— membrane able to be activated by Gpp(NH)p or fluoride!'. For this activa- tion to occur, the cyc~ membranes must be co-incubated with HE membranes which lack R units, or with RL mem- branes in the presence of glucagon plus GTP, or with donor membranes pre- treated with cholera toxin and NAD (a procedure that ADP-ribosylates the a-subunit and which renders the G, unit susceptible to activation by GTP). Recently, we succeeded in separating donor and recipient cyc— membranes after co-incubation under various activating conditions!?. Separation was achieved because the cyc— membranes have a lower density than either HE or RL membranes; layering the mixture of membranes over a sucrose gradient fol- lowed by centrifugation resulted in a layer of cyc— membranes free of donor membranes, as indicated by assays of various enzymes present in donor but not in cyc— membranes. When isolated after co-incubation with donor mem- branes under appropriate activating con- ditions, cyc— membranes acquired an active a-, subunit (a of G,) donated by HE or RL membranes. This was indi- cated by (1) the levels of Gpp(NH)p- stimulatable adenylate cyclase activity induced in cyc— membranes and (2) by the quantity of a-, transferred to cyc—. The latter was monitored by labelling a-, with [”P] ADP-ribose catalysed by chol- era toxin. A typical example of the rela- tionship between transfer of a-, and the degree of activation of cyclase is illus- trated in Fig. 1 using HE membranes co-incubated with cyc—. This experiment also revealed, in- directly, that when G, in HE membranes is activated by Gpp(NH)p and Mg?+ there is simultaneous activation of G,. Activation of cyclase and transfer of «-, to cyc— membranes in HE membranes was slight unless the donor membranes were pre-treated with pertussin toxin plus NAD. This toxin ADP-ribosylates a-, and renders G, inactive?. As shown in Fig. 1, toxin-treatment of HE causes cyc— membranes to acquire high levels of Gpp(NH)p-stimulatable — cyclase TIBS - November 1985 463 actitivy with concomitant transfer of a-, (labelled with cholera toxin and 2-P NAD on re-isolated cyc-). We interpret these findings as evi- dence that a-; released from G, in the donor membranes influences the ability of a-, to interact with cyc— adenylate cyclase. We are investigating whether this is due to competition between released a-, and a-, for sites on adenyl- ate cyclase or to some other process, such as the ‘scavenging’ of released a-, by exposed (/y subunits of G, (Ref. 10). Irrespective of the mechanism, the thrust of these findings is that simul- taneous activation of G, and G,, with consequent release of their respective a-subunits from the membrane, can dra- matically affect the amount of a-, trans- ferred to cyc—. Similar results are seen with liver membranes using combina- tions of hormones and GTP to induce activation. Obviously, results obtained with a test system of two isolated membranes do not necessarily simulate what happens in an intact cell. Nonetheless, it is reason- able to speculate that release of a-sub- units is the primary step leading to the pleiotropic effects of hormones on their target cells. If it is true that these pro- teins are primary messengers of hor- mone action, the present concepts of hormone action will have to be altered radically. Programmable messengers Perhaps the most significant dif- ference between proteins and small molecules such as cyclic AMP is that a protein messenger is pluripotent in its capacity to react as a regulatory signal. Proteins can be phosphorylated, methyl- ated, sulfated, oxidized, appended to other proteins via disulfide groups, and degraded to smaller forms by proteases, to name a few well known covalent modifications. Such modifications yield different structures with different func- tions. If a-subunits are modified after their release into the cytosolic compart- ments of the cell, and if some of the modifications lead to a different regula- tory structure, then the a-subunit, the initial primary messenger, can be con- sidered programmable. This concept of ‘programmable messengers’ is illustrated | in Fig. 2. In this scheme, each type of a-subunit released from the plasma membrane as a consequence of actions by hormone and GTP becomes exposed to different modifiers (M) that alter the structure and function of that unit. Each new form of a reacts selectively with an effector (E) which emits a signal (S) that can bring forth one or more responses. Possible examples of M include protein kinase C, insulin-receptor tyrosine kinase and calcium-activated protease. Examples of E are adenylate cyclase, guanylate cyclase, calcium transporters, phospholipases and glucose transpor- ters. The central point of this thesis is that a single primary signal can give rise to an array of new signals which, in wave-like fashion, can propagate vast changes in the structure and metabolism of target cells. Specificity of response will depend on the types of receptor and G (or a-) units, and on the modifiers and effector of the cell phenotype. Given that there are various classes of receptors linked to G units and many potential signal-generating effector sys- tems, a variety of cell responses can be envisaged. The idea of programmable a-subunits as messengers provides an explanation for the frequently cited lack of correla- tion between hormone-stimulated AMP levels, for example, and other responses given by a hormone. Levels of activated cAMP-dependent protein kinase are also not correlated in all responses; a recent example is the discordance in the actions of ®-adrenergic agonists on lipolysis and glucose transport in rat adipocytes'*, There are examples of cer- a? —____» Fo _____» Se ————> Response Gte =< a' ———» E' —____» S' —_____» Response a? —__» £2 ____» S?______» Response ——___——» Endocytosis s B} , R Y Me | | GTP Mt Hormone = |R2:G|-———————————_ M2 Plasma membrane Fig. 2. Theory of ‘Programmable Messengers’. Hormones interact with receptoriGTP-regulatory complexes (R.G) causing, in presence of GTP, release of the subunit of G from the interior face of the plasma membrane. R and By subunits of G remain in the membrane and can either re-bind a (-GTP) to re-form R.G or they are taken into cell by endocytosis. The released « subunis is exposed to modifying enzymes (M) that transform a. into new structures having affinities for different effector (E) units which, activated, yield signals (S) the combination of which give the pleiotypical responses of the target cell. 464 TIBS — November 1985 tain hormones inducing simultaneous tises in adenylate cyclase and cAMP- phosphodiesterases by apparently inde- pendent processes. Hormones that induce activation of G,, while inhibiting the production of cAMP induced by a hormone operating through G,, exert effects which are clearly unrelated to regulation of cAMP production’. Receptor desensitization The programmable messenger theory can explain why receptor desensitization in intact cells can be reversed rapidly with low pulses of hormone and short times of exposure, but not with high concentrations and longer exposure to hormone. In the theory, only a fraction of the a-subunit would be released in a short pulse-type experiment with sub- maximal concentrations of hormone; during this brief interval, there may not be sufficient modification of a by modi- fiers to alter the equilibrium between bound and free a-subunit (Fig. 2) when the hormone is withdrawn. By the same reasoning, with higher concentrations of hormones and longer exposure times, more a-subunit is discharged from its union with @/y subunits and there is greater opportunity for modifiers to pre- vent a from re-associating with these B/y. If this is so, reconstitution of func- tional receptors coupled to G units may require internalization of receptors (with or without attached B/), resynthesis of new a, and recycling of the units back to the plasma membrane. Synergy Perhaps the most interesting possible consequence of the programmable mes- senger theory is an explanation for the long-known synergism with which two hormones, operating through com- pletely different mechanisms, exert effects on cells. A good example is the synergistic effect of insulin and ade- nosine on the metabolism of rat adipocytes'®. Neither insulin nor aden- osine alone have much effect at physio- logical concentrations. Combined at such concentrations, the hormones exert large effects on such metabolic processes as lipolysis and glucose transport. Since insulin activates a tyrosine kinase asso- ciated with the receptor’, it is possible that the activated kinase phosphorylates a liberated a-; subunit (adenosine oper- ates through a receptor linked to G, in these cells) converting it from a weak or inactive regulatory signal into one that is wary active. There are many examples of synergism between two hormones or neurotransmitters operating on the same cell through different primary mech- anisms. The point to be stressed is that the search for the usual small molecule messengers, such as cAMP, as the prim- ary agents of synergism has not yet been successful. Hopefully, the ideas put forth here will stimulate investigations along different, more productive lines of research. Acknowledgement The author thanks Professor Torben Clausen of Aarhus University for sug- gesting the expression ‘programmable messengers’. References 1 Tolkovsky, A. M. and Levitzki, A. (1978) Biochemistry 17, 3811-3817 2 Stadel, J., DeLean, A. D. and Lefkowitz, R. J. (1980) J. Biol. Chem. 255, 1436-1441 3 Salesse, R., Garnier, J., Leterrier, F.. Daveloose, D. and Viret, J. (1982) Biochemistry 21, 1581-1586 4 King, W. J. and Greene, G. L. (1984) Nature 307, 745-747 5 Rodbell, M. (1980) Nature 284. 17-22 6 Minton, A. P. in The Receptors (Conn. P. M., ed.), Vol. 2, Academic Press (in press) ‘ 7 Gilman, A. G. (1984) Cell 36, 77-579 8 Cerione. R. A., Staniszewski, C.. Benovic, J.L., Lefkowitz, R. J.. Caron, M. J. Gierschik. P.. Somers, R., Spiegel. A. M.. Codina. J. and Birnbaumer, L. (1985) J. Biol. Chem. 260, 1493-1500 9 Brandt, D. R.. Asano, T., Pedersen. S. E. and Ross, E. M. (1983) Biochemistry 22, 4357- 4362 10 Northup, J. K.. Smigel. M. D.. Sternweiss, P.C. and Gilman, A. G. (1983) J. Biol. Chem. 259, 3560-3567 11 Nielson, T. B., Lad, P. M., Preston, M. S. and Rodbell, M. (1980) Biochim. Biophys. Acta 629, 143-148 12 Zaremba, T., Ngo, D. and Rodbell, M. (1985) Fed. Proc. 44, 700 13 Okajima, F., Katada, T. and Ui. M. (1985) J. Biol. Chem. 260, 6761-6768 14 Honnor, R. C., Dhillon, G. S. and Londos. C. J. Biol. Chem. (in press) 15 Lad, P. M., Olson, C. V. and Smiley, P. A. (1985) Proc. Natl Acad. Sci. USA 82, 869-873 16 Schwabe, U., Ebert, R. and Erbler, H. C. (1973) Naunyn-Schmiedebergs Arch. Phar- makal. 276, 133-138 17 Haring, H-V., Kasuga. M. and Kahn, C. R. (1982) Biochim. Biophys. Res. Commun. 108, 1538-1540