NEWS AND VIEWS

involves a putative 'killer-GAP' that is Turning off GTP signals not associated with an effector target, THE discovery of GAP activities in occurs when it binds simultaneously to Although they normally function as targets of the heterotrimeric G proteins its target molecule and to its GAP. GAPs in association with specific brings them into step with their Indeed PLC-f:l1, a single molecule, acts target molecules, PDE-y and the ribo• cousins in the GTPase superfamily1, as both the effector target and the GAP some show that it is possible to deacti• the Ras-like monomeric GTPases and for £1'q·GTP. vate the GTP-bound form of a protein in the initiation and elongation factors of Two other models merit considera• the absence of its target. At least two ribosomal protein synthesis. Of these, tion. First (as in c in the figure), the proteins, Ras-GAP and neurofibromin, bacterial elongation factor EF-Tu has can trigger the GTPase activity of fas been subjected to the most inten• p21 , and GAPs have been identified sive scrutiny. Aminoacyl transfer RNA a for many other members of the ex• (aa-tRNA) binds selectively to TU'GTP, panding family of monomeric small but does not accelerate its very slow GTPases (for review, see ref. 22). Are intrinsic GTPase activity. Through bind• these killer-GAPs or do they also act as ing of its anticodon to the mRNA• (or in association with) effector programmed ribosome, however, targets? Despite generating a number aa-tRNA does position TU'GTP for inter• of ingenious experiments, the debate action with a GAP located in the ribo• over these alternatives22- 24 has come some's 50S subunit; very quickly (with• to no firm conclusion, for a very simple in a few milliseconds), the ribosomal reason: we are in no position to deter• GAP triggers hydrolysis of Tu-bound mine the relation between GAPs and GTP. effectors, because no example of the It is instructive to compare protein latter is in hand. Despite the many complexes that regulate GTPase activi• well-studied cellular consequences of fas ties of different members of the super• activating p21 , we cannot point to a family. Each GTPase in its unadorned Single molecule (or even a biochemical G·GTP state (a in the figure) hydrolyses activity) that is directly regulated by GTP relatively slowly (tens of seconds the GTP-bound form of this GTPase (or for £1"GTP, many minutes for Tu·GTP). of other Ras-like proteins). To speak In most cases (b in the figure) the plainly, we do not even know whether physiologically relevant and much fas• £1'·GTP or EF-Tu is the appropriate mod• fas ter turn-off switch is thrown when el for the action of p21 - that is, G·GTP forms a complex with other com• whether the protein works by regulat• ponents - £1'q + phospholipase C ing the biochemical activity of a target (PLC)-f:l1, £1'(GTP + PDE-£1'f:l + PDE-y, or protein or by regulating the assembly TU'GTP + aa-tRNA + ribosome. Just as of separate components into a macro• the y-subunit of the PDE holoenzyme molecular complex. can suffice to stimulate the GTPase of target is not associated with a GAP Despite our ignorance, the discovery £1't, ribosomes by themselves (at high activity. The example of aa-tRNA and of GAP activities associated with effec• concentrations) can stimulate GTP TU'GTP shows that a target molecule tor targets of G-protein £1'-subunits fur• hydrolysis by EF-Tu. Although neither may bind to a GTPase without nishes a persuasive analogy. The new aa-tRNA nor PDE-£1'f:l stimulates GTP accelerating hydrolysis of bound GTP. observations strongly support the view hydrolysis, both of these non-GAP Some of the many isozymes of that some of the GAPs for p21fas and elements serve to regulate the GAP hormone-sensitive adenylyl cyclase its siblings are effectors (by analogy - by inhibiting GAP in the presence of may belong to this category. In fact. with PLC-f:l1 and £1'q), whereas others cGMP (PDE-£1'f:l) or by guiding TU'GTP the rates at which adenylyl cyclase (by analogy with PDE-y and £1't) act in to the GAP (aa-tRNA). activities decay after hormonal association with effector target pro• Despite these and other differences, stimulation19,2o agree rather well with teins. The likelihood that p21fas or the prevailing theme is the same: measured rates of GTP hydrolysis by other monomeric GTPases associate although the GAP by itself can act on pure £1'521 . only with killer-GAPs now appears to be G·GTP, physiological turn-off of G·GTP The second model (d in the figure) vanishingly small. H.R.B. 8& L.S. stimulated PDE actIVIty turning off at activity, seemed to account for the abil• recently discovered adenylyl cyclase the same rapid rate, even in the presence ity of G proteins to amplify hormone isozymes (see ref. 14 for summary) differ of a high concentration of cGMP. Be• signals. Now we realize that fixing the primarily in their abilities to act as GAPs cause both sets of results are convincing spotlight on the prima ballerina gave for the £1'-subunit of Gs , the stimulatory and internally consistent, it is not clear predicted turn-off rates that were far too regulator of adenylyl cyclase. why cGMP should slow GTPase activity slow for physiology - a difficulty that is Finally, we should recognize that in one case7 but fail to prolong PDE neatly solved by converting the turn-off sometimes extinction of a G-protein• 13 activity in the other .) into a pas de deux. Association of GAP mediated response may require more The exciting discovery that effectors of activities with effectors preserves ampli• than the GAP activity of an effector. G proteins also act as GAPs for G fication of signals, but to a degree deter• Until now we have assumed that only proteins carries broad implications for mined by the effector rather than by the one £1'·GTP can activate each effector, so understanding other proteins in the G-protein £1'-subunit. Furthermore, the that the effector's GAP activity effec• GTPase superfamily, including the ever GAP activity can itself be regulated and tively turns off the response. What if an enigmatic p21'as (see box). This discov• it may vary among different effectors, activated receptor generates a number of ery also forces us to modify the prevail• thus further enhancing the versatility and £1'·GTP molecules greater than the num• ing model of how signalling G proteins flexibility of signals mediated by G pro• ber of effectors available to turn them work. In that model, the slow turn-off of teins. It would not be surprising to learn, off? Indeed, an excess of G proteins £1"GTP, mediated by its intrinsic GTPase for example, that some of the many over effectors may be in some cases a 542 NATURE· VOL358 . 13 AUGUST 1992 © 1992 Nature Publishing Group