Guanine nucleotide regulatory proteins in insulin's action and in ...

619th MEETING, LONDON

627

Guanine nucleotide regulatory proteins in insulin's action and in diabetes MILES I>. HOUSLAY. NIGEL J. PYNE. R I C H A R D M. O'BRIEN. KENNETH SIDDLE, DEREK STRASSHEIM, TIMOTHY PALMER. S A N D R A SPENCE. M A R Y WOODS. A N D R E W WILSON, BRIAN LAVAN, G R E G O R Y J. MURPHY, M A R K SAVILLE. MARY-ANNE McGREGOR. E L A I N E KILGOUR. NEIL ANDERSON. J O H N T. KNOWLER, SUSANNE G R l F F l T H S and G R A E M E MlLLlGAN Molcwilur I'harrvitrc~ok)~?Grorip9Ikpartr~irrrtof' Niochemistn. Utiiivrsiry of Glusgou, G l r i s p t ~G I KQQ. S(~0tlrrrld.u.ti. T h e detailed pathways which link occupancy of the insulin receptor with actions o n target cells remain t o be defined. The insulin receptor expresses a tyrosyl kinase activity for which considerable evidence has accrued indicating that it plays a fundamental role in transducing at least certain o f insulin's actions. Nevertheless, the mechanism whcrcby tyrosyl phosphorylation o f proteins might lead to cellular responses is still as much of an enigma a s the physiological role o f the few substrates that have been identified 11.21. T h e functioning o f ii wide variety of cell surface receptors for hormones and neurotransmitters is transduced by their interaction with specific guanine nucleotide regulatory proteins (G-proteins) I3,4]. Not only that, but considerable evidence suggests that there is substantial 'cross-talk' bctwecn the machinery o f cellular signalling systems [ 51. This can lead to heterologous desensitization o r t o supersensitization depending upon the signalling system and cell type. Here we discuss evidence which supports the contention we made earlier 16, 71, that insulin can interact with the G-protein system. This shows that ( i ) insulin, acting through its receptor can interact with both individual G-proteins and specific G-protein-controlled systems, ( i i ) that a specific G-protein may be involved in transducing certain of insulin's actions and (iii) that in diabetes and in insulin-resistant states one sees the functional inactivation o f the inhibitory Gprotein G,in both liver 18.91 and white adipose tissue.

demonstrations that ( i ) treatment of isolated membranes or permeabilized cells with non-hydrolysable analogues o f G T P can, by activating G-proteins independently of receptor occupancy. regulate the signal generation system which is coupled to that G-protein; (ii) receptor occupancy can stimulate high-affinity GTPase activity. This cannot be seen in many membranes. however, due !o non-specific GTPase activites; (iii) guanine nucleotide-mediated shifts in receptor binding studies. T h e magnitude of such effects, however, varies widely: (iv) receptor occupancy can alter guanine nucleotide binding t o membranes. Again, the magnitude o f such effects can vary markedly, and (v)the ability of pertussis and cholera toxins t o block or mimic. respectively, the actions of certain hormones o r neurotransmitters. This is achieved by virtue of such toxins causing the N A D + dependent ADP-ribosylation of the a-subunits of G,, Gc> and G,. Thus cholera toxin activates G, and pertussis toxin inactivates G,, and forms o f G, [3.41.

Iiiitiul eiideric,iJthrrt irisriliri rviight ititeruct with the (;-protein .s~.ster?l In lY83 we suggested that insulin might exert certain o f its actions by interacting with the G-protein system and that, possibly, a particular G-protein might be involved in mediating at least certain o f insulin's actions 131. Such a contention arose from our studies which showed ( i ) that insulin could inhibit adenylate cyclase activity in a GTP-dependent fashion [ 101, (ii) that cholera toxin could both mimic the ability o f insulin to activate the peripheral plasma membrane cyclic A M P phosphodicsterasc 11 11 and block insulin's ability t o inhibit adenylate cyclase jlOj and (iii) that pertussis toxin could block the ability o f insulin to activate the 'densevesicle' cyclic A M P phosphodicsterasc I 121. Since that time we and others have presented evidence for the direct interaction of the insulin receptor with specific G-proteins and have shown that marked changes in G-protein function occur in diabetes and in insulin-resistant states.

What are G-proteirisY

I'kosphor3,brioti of G, arid G,, by thc hriinuii irisrilin receptor

I t is now well established that many hormones act to alter the functioning of target cells because their receptors interact with specific G-proteins. Such regulatory proteins act as 'gobctweens' to couple receptors to their effcctor, signal generator systems in the plasma membrane. Thcsc are proteins embedded on the inner surface o f the ccll plasma membrane which, upon interaction with an appropriate receptor. bind G T P and become activated. T h e subsequent hydrolysis reaction elicits deactivation. A family of such proteins exist, with probably as many as a dozen types occurring. They have a similar structure, consisting of a-. p- and y-subunits. It is thc a-subunit which interacts with both the receptor and the signal generator on the inner surface o f the ccll membranc 13.41. We here discuss four particular G-proteins. These are G, and G , which, respectively, transduce the activation and inhibition o f adenylate cyclase; G,, which is, like G,, a substrate for inactivation by pertussis toxin and is thought t o be involved in regulating certain types o f Ca2+ channels and p2l '"', which is a 21 kDa G-protein found as a normal cellular protein and as an activated oncogenic product. Evidence that G-proteins are involved in the action o f hormones has come from a number of routes. These include

We have shown 11 31 that a pure preparation of insulin receptors, solubilized from human placcntal tissue, can cause the phosphorylation o f the a - and p-subunits of pure preparations of G, and G,, from bovine brain. For these G-proteins t o become phosphorylated, it was necessary for them t o be in their holomeric state, where they exhibited a ti,,,of approx. 2 0 p ~This . would thus appear to be a high-affinity reaction when such a value is compared with those found for various exogenous substrates that can be phosphorylated by the pure insulin receptor. T h e rate of phosphorylation of G,, and G, was found to be comparable to that using casein as a substratc. but rather less than that seen with histones. which, t o date, have provided some of the best exogenous substrates. T h e stoichiometry and thc functional significance of the phosphorylation of G,, and G, has yet to be determined. It may be, however, that the phosphorylation of G,, and G, acts t o stabilize the inactive, holomeric state of these species as activation of these G-proteins, using non-hydrolysable GTPanalogues, obliterated the ability of the insulin receptor t o phosphorylatc them. Such phosphorylation was alkali stable and shown to be on tyrosine residues rather than serine. T h e presence of G,, and G,, however, did not affect the extent o r rate o f the insulin-stimulated autophosphorylation of the D-subunit of the insulin receptor. T h u s neither G,, nor G, regulate autophosphorylation o f the insulin receptor.

Ahhreviation uscd: plNH]ppG, guonosine-l/j, y-imido]triphosphate. Vol. 17

628

BIOCHEMICAL SOCIETY TRANSACTIONS

is not present [8].This contrasted with the situation in liver membranes from obese Zucker rats where we have shown that GI was still present, but in an inactive form [91. Furthermore, G, is still present in adipocyte membranes from diabetic animals and obese Zucker rats, but again in an inactive form. Indeed, we have some preliminary evidence which sugp2I gests that GI is even expressed in hepatocytes from diabetic animals. However, it appears that it may easily be displaced It has been suggested [14] that the proto-oncogene product of this G-protein plays a regulatory role in the cell, from the plasma membranes of hepatocytes from such perhaps by coupling certain receptors to the stimulation of animals, which would account for our failure t o find it in inositol phospholipid metabolism in systems which are purified membranes. Nevertheless. G, is clearly inactive in hepatocytes as P,-purinergic inhibition of the adenylate insensitive to the action of pertussis toxin. Using a pure preparation of p21"', from an Escherichia cyclase activity of intact hepatocytes is lost in diabetic coli expression system, incubated with the pure human animals. Using synthetic 34-mer oligonucleotide probes to the insulin receptor, we failed to elicit the phosphorylation of p2 1rfzy, irrespective of whether p2 1 was in its inactive or various forms of GI, we have been able to show that liver activated states [15]. However, if ~ 2 1 " ' was purified by a expresses G,-2 and G,-3 only. The induction of diabetes, method involving a denaturation/renaturation cycle with however, has little effect upon mRNA concentrations in urea then p2 1'"' did act as a substrate for the insulin receptor hepatocytes from such animals compared with those from tyrosyl kinase. Under such circumstances p2 1"' was phos- normal animals. Again this suggests that expression of G, is occurring in hepatocytes from diabetic animals. phorylated upon tyrosyl residues. In our opinion, a unifying feature of the insulin-resistant However, we found that in its GDP-bound state, ~ 2 1 " " states that we have investigated is the inactivation of G,. acted to attenuate autophosphorylation of the insulin Whether G, is indeed involved in mediating certain of receptor. It remains to be seen whether this has any physioinsulin's actions, or whether this inactivation occurs as a conlogical role. However, if p2 1"" in its GDP-bound form exerts a tonic inhibitory effect upon the insulin receptor then this sequence remains to be seen. We would like to suggest that the molecular mechanism of the inactivation of G I is due to may be abolished by conversion of p21"' to its GTP-bound its phosphorylation, possibly via the action of protein kinase form during receptor activation. Indeed, there is evidence which suggests that p2 1"' is involved in insulin signalling C. Certainly we have been able t o show that G, can be phosphorylated and inactivated in intact hepatocytes by virtue of reactions (see [ 21 for review). the activation of protein kinase C by hormones which stimulate inositol phospholipid metabolism [ 161. Loss offi~nctionulG, in insulin-resistant states We, however, have so far been unable to show that the phosphorylation of GI occurs in situ in the intact hepatocyte, despite providing evidence to show that the insulin receptor can interact with GIin membranes from such cells.

We have shown that functional GI in liver plasma membranes is lost in rats made diabetic with either alloxan or streptozotocin, thus providing the first evidence of a lesion in G, in a disease state [S]. This was assessed by virtue of the ability of low concentrations of guanosine-[@,y-imidoltriphosphate (p[NH]ppG) to inhibit forskolin-stimulated adenylate cyclase activity in liver plasma membranes. A functional consequence of this was the observation that glucagon expressed a markedly enhanced ability to stimulate adenylate cyclase. This is, indeed, what one might expect if G, function was abolished, as it was also observed by treating membranes from normal animals with pertussis toxin to inactivate G,. Consistent with diabetes causing the inactivation of G,, we found that treatment of hepatocytes from diabetic animals with pertussis toxin failed to augment glucagon's activation of adenylate cyclase. Loss of functional G, can be readily reversed by insulin therapy, demonstrating that its loss was a consequence of the diabetic condition rather than the drug used. Recently, we have made similar observations in plasma membranes from adipocytes, where again G, function is lost upon induction of diabetes. Liver and adipose tissue from chemically induced diabetic animals exhibit tissue insulin resistance. Another model of insulin resistance is the obese Zucker rat at about 8-10 weeks of age. Using such animals we showed that little functional Gi was evident in membranes from either hepatocytes [9] or adipocytes. In contrast, membranes from their lean littermates exhibited GI activity which was comparable to that seen in normal rats [S]. We have made similar observations with db/db mice, where hepatocytes from normal animals exhibit functional G,, whereas those from diabetic animals have no detectable G, activity. Is GI inactivated or just not expressed in such tissues? An effective way of detecting the a-subunit of G, is by using specific antipeptide antibodies. In liver membranes from diabetic animals we have used this approach t o show that a-G,

Metformin redirces G, furictiori and reserisitizes heputocyte adenylate cyclase activity to irihihiriori by irisirliri in diabetic rats Insulin causes of the order of 25-30%, inhibition of adenylate cyclase activity in either isolated membrane systems [ l o ]or in intact hepatocytes [ 121. This effect is GTPdependent and can be blocked by zither pertussis toxin or by cholera toxin. Thus insulin does not appear to inhibit adenylate cyclase through G I , as cholera toxin docs not block G, functioning. We have provided support for this contention using the hypoglycaemic agent metformin. This biguanide drug is used to treat insulin-resistant states. Its molecular mechanism of action is unknown, although it is believed to rectify a post-receptor defect in the insulin signalling system. When normal rats were treated with metformin, they exhibited a loss of functional G, which was not accompanied by any alteration in the ability of insulin to inhibit adenylate cyclase activity. Furthermore, treatment of diabetic rats with metformin completely restored the ability of insulin to inhibit adenylate cyclase activity, despite the fact that functional G, was not apparent [ 171. This demonstrates that G, cannot be transducing the inhibitory action of insulin upon adenylate cyclase activity.

Insulin affects the ability of pertirssis roxiri to cuiise the ADPrihosylution of G, GI can be ADP-ribosylated, in its holomeric form. by pertussis toxin, but not when it is in its activated, dissociated state. The pretreatment of intact hepatocytes with insulin does not affect the subsequent ability of pertussis toxin to ADP-ribosylate G, in isolated membranes [ 161. However, if insulin is added to membrane ribosylation mixtures with GTP, then it will attenuate the ability of pertussis toxin t o ADP-ribosylate G,. However, when p[NH]ppG is added to dissociate G, and thus attenuate ADP-ribosylation. we found 198Y

629

62Yth MEETING. LONDON that the addition of insulin actually augmented ADP-ribosylation. This suggests that the insulin receptor might indeed interact with G, to elicit a conformational change. This, we suggest, might serve to stabilize the holomeric state o f G,. Itisitliti

utid cholera roxiti

Treatment of isolated hepatocyte plasma membranes with cholera toxin caused the ADP-ribosylation of a variety of proteins, including a - G , [ 1x1. T h e ribosylation o f one of these, a 25 kDa protein is attenuated by the addition of insulin [ 181. This led us to suggest that a cholera toxin-sensitive G-protein may b e involved in mediating certain of insulin's actions. In support of this, we have shown that insulin appears to stimulate a GTPase activity in human platelets that is modified by cholera toxin [ 191. However, further resolution of such a species demands a suitable reconstitution system. Insulin also appears to affect the ability of cholera toxin t o activate adenylate cyclasc in intact hepatocytes. Thus, although treatment of intact hepatocytes with cholera toxin caused the ADP-ribosylation o f a-G,, and thus the constitutive activation of adenylate cyclase, if insulin was also added then the ability of cholera toxin to activate adenylatc cyclase was dramatically reduced [20].This was not due to insulin affecting the ability of a-G, to be ADP-ribosylated by cholera toxin, but rather appeared to be an attenuating action exerted either at the function of adenylate cyclase itself o r the functioning of ADP-ribosylated a-G,. C'oricliisiori

A considerable amount o f evidence suggests that insulin interacts with the G-protein system at a number of points. However, the precise mechanisms and functional roles remain to be elucidiated. It is. however, becoming apparent that loss of G, function appears to be symptomatic of insulin resistance. T h e consequences o f this and whether it also occurs in human beings may be of considerable interest and potential importance.

This work wa\ \upported by the M.R.C.. A.F.R.C.. Lipha Pharmaceutical$. C.M.K.F. and the B.D.A. I , Houslay. M. 0. ( 1986) In Kccrrrt i l r h w r c ~ e s in Dicibetes (Nattrass, M.. ed.1. vol. 3. pp. 35-53. Churchill-Livingstone. Edinburgh 2. Houslay. M. I>.di Siddle. K. ( 19x9) Hr. Mcd. H/iIi. 45, 264-284 3. Houslay. M. I). ( 1984) Trends Nioc~hem.Sci. 9, 30-40 4. Milligan. G. ( 1988)Biochem. J. 255, 1 - I3 5. Houslay, M. I),, Wakelam, M. J. 0..Murphy, G. J., Gawler, 0.J. di Payne. N. J. ( 19x7) Biochcwr. .So(,. Trcrm 15, 7 1-24 6 , Houslay. M. I). di Heyworth. C. M. ( 1983) Triwdc Hioc./wm. S c i . 8,449-457 7. Houslay. M. I).( 19x6) Rioclrc~rtr..So(.. li.cir1.s 14, 1 X3- I Y 3 X. Gawler. I>.. Milligan, G.. Spiegel. A . M.. Unson. C. G. di Houslay, M. I).( 1987) Niitiiro (1.ondon) 327, 229-232 9, Houslay. M. 11.. Gawler. I). J., Milligan. G. di Wilwn, A. ( 19x9) ('eliicltrr ~ S i p ~ i i l i It r, ~9-77. 10. Heyworth. C. M. Kr Houslay. M. D. ! 19x31 Nioc~heni.J. 214. 547-557 I I . Heyworth. C. M., Wallace. A. V. di Houslay. M. D. (1983) Hiochc,nl. J. 2 14, 99- 1 I 0 12. Heyworth. c'. M.. Grey. A. M.. Wilson, S. K.,Hanski, E. Kr Houslay. M. I). ! l 9 X 6 ) H i o ( ~ / / c ~ t ,J.r . 235, 145-llU 13. O'Brien. K. M.. Houslay. M. I>.. Milligan. Ci. Kr Siddle. K. j 19x7) F/:R.SIdc,t/.212. 381-2XX 14. Wakelam. M. J. 0..Davie\. S. A,. Houslay. M. I).. McKay. I., Marshall. C. J. C ! Hall. A . ! 19x6) "Vuriire ( [ m i d o n ) 323, 173- I 7 6 15. OBrien. K. M., Siddle, K.. Houslay, M. D. di Hall, A. [ 19x7) FEES Lc/I.2 17. 253-259 16. Pyne, N. J., Murphy, G. J., Millipan, G. di Houslay, M. I). (1989) F-EB.S Ltt.2 43. 7 7- X 2 7. Gawler. D. J., Milligan. G. & Houslay, M. D. ( 19x8) Rioclieni. J. 249.537-542 8. Heyworth, C. M., Whetton. A . I>., Wong, S., Martin. B. K. di Houslay. M. I). ( I 9 x 5 ) HiocAeni. J. 228, 593-603 9. Gawler. D. di Houslay, M. D. (1987) FEBSLett. 216, 94-98 I). Irvine. F. J. Kr Houslay. M. D. ( I 988) Hiochem. J. 251,447-452

Received 3 1 January 1989

Mitogenic signalling in murine 3T3 cells: the cyclic AMP pathway ENRIQUE ROZENGURTC and S H A R O N S. O B E R lrnperiul Cuticer Keseurch Fiitid?1'. 0. Box 12.t Liricdtl k I r l r l Fields, Lotidoti WC'ZA 3PX, U.ti. Ititrodiictiorl

T h e elucidation of the mechanisms by which growth factors regulate cellular mitogenesis has emerged as one of the fundamental problems in biology and may prove crucial for understanding the unrestrained proliferation o f cancer cclls. In this respect. cultured fibroblasts. such a s murine 3T3 cells, have emerged as a model system. These cclls cease t o proliferate when they deplete the medium o f its growth promoting activity and enter a quiescent o r non-dividing state. However, such cells remain viable and can be stimulated to reinitiate D N A synthesis and cell division either by replenishing the medium with fresh serum, o r by the addition of purified growth factors o r pharmacological agents in serum-free medium (Rozengurt. 1986). Studies performed using such growth-arrested cells and defined combinations of Abbreviations used: PKC. protein kinase C : VIP. vamintestinal peptide; EGF. epidermal growth Factor; F.p.l.c.. last protein liquid

chromatography. 'To whom correspondence should he addressed.

Vol. 17

growth factors have revealed the existence of potent and specific synergistic interactions involved in mitogenic stimulation (revicwcd in Rozengurt. 1986). An important step in elucidating the basis o f growth factor action is t o identify the signal-transduction pathwayis) involved in the gcncration of thc mitogenic response. For years many investigators have proposed and searched for a single key signal governing the initiation o f cell proliferation. Extensive analysis o f early signalling events, however, has led to the formulation of an alternative model: the existence of multiple growth-factor-activated signalling pathways that synergistically Icad to a mitogenic response (Rozengurt. 19x6; Rozengurt tv ul., I Y X X ) . In this paper we summarize some recent findings concerning the cyclic A M P signal-transduction pathway in thc mitogenic response of Swiss 3T3 cells. C j c l i c A MI'

(i I I

(1 t t I i foget i esis

A variety of agents that promote cyclic A M P accumulation in Swiss 3'13 cells, including prostaglandin E , ( P G E , ) , the adenosine agonist 5'- N-ethylcarboxamideadenosine ( N E C A ) . cholera toxin and permeable cyclic A M P analogues stimulate D N A synthesis by acting syncrgistically