TIBS 17 - OCTOBER 1992 Second messengergeneration and destruction

ALL CELLS possess the capacity to

receive and process information from
their surroundings. External signals
such as light, odorants and dietary
chemicals stimulate target cells in
specialized sensory organs; circulating
The family of heterotrimeric guanine nucleotide-binding regulatory proteins
or locally released hormones, neuro-
transmitters and growth factors serve
(G proteins) serves an essential role in transducing receptor-generated
as chemical messengers between neigh- signals across the plasma membrane. Recent findings reveal unexpected
boring or distant cells. The interaction functional roles for individual G protein subunits. Thus, GTP-binding
of these messengers with specific o~-subunits and the [~,-subunit complex can influence the activity of ef-
receptors at the ceil surface represents fector molecules independently or simultaneously, either synergistically or
only the first step in a cascade of in opposition, to elicit a complex constellation of cellular events.
molecular events that underlies trans-
membrane signaling. In many cases,
stimulation of these receptors results in is inactive, whereas the GTP-bound receptors, G proteins and effectors is
activation of effector proteins (e.g. form of a dissociates from 137 and shown in Fig. 1 and details of the G
enzymes or ion channels), which mobil- serves as a regulator of effector pro- protein activation/deactivation cycle
ize chemical 'second messengers' that teins. Aluminium tetrafluoride (AIF~), are described in the legend (see also
initiate characteristic actions within the together with Mg2÷, can interact with a- Refs 6,7).
cell. In all eukaryotic organisms, a family bound GDP to mimic GTP and thereby
of heterotrimeric GTP-binding and hy- activate a. Increasing evidence now Relationships and functions of G protein
drolysing proteins (G proteins) plays an supports the hypothesis that [~', like a, subunits
essential transducing role in linking can also interact with and modulate the The family of G protein a-subunits
many cell-surface receptors to effector activity of at least some effector pro- can be subclassified according to func-
proteins at the plasma membrane. This teins (see below). All a-subunits are tional or structural relationships 4,8 and
review will provide a brief summary of themselves enzymes. That is, these pro- there is reasonable congruence be-
knowledge of G protein structure, func- teins possess intrinsic GTPase activity tween such schemes. The nomenclature
tion and mechanism of action. The G and will, at varying rates, hydrolyse the utilized to denote individual family
proteins are part of a larger superfamily terminal phosphate of bound GTP to members is unfortunately confusing,
of GTPases that includes factors in- yield bound GDP and free inorganic but knowledge of function is probably
volved in protein synthesis, for example phosphate (P~). In some cases, a-sub- still too limited to propose a lasting
elongation factor Tu (EF-Tu) and a large units possess specific residues that can alternative. G proteins were first identi-
number of monomeric 20-25 kDa pro- be covalently modified by bacterial tox- fied functionally and purified conven-
teins such as p21r~s; these will not be ins. Cholera toxin catalyses the transfer tionally; names were assigned with sub-
discussed here, but have been reviewed of the ADP-ribose moiety of NAD to a scripts chosen to evoke functional
previously TM. specific Arg residue in certain a-sub- roles. Since purification and cloning of
units. Similarly, pertussis toxin ADP- subsequently discovered members of
Structure and properties ribosylates those a-subunits that pos- the family were largely accomplished
G proteins are heterotrimers, com- sess a specific Cys residue near the with homology-based approaches, later
posed of three distinct subunits: carboxyl terminus. Modification of a by names were chosen according to the
a (molecular mass = 39-46 kDa), 13 (37 cholera toxin constitutively activates whim of the discoverer. One now
kDa) and ~/(8 kDa). The properties of these proteins (by inhibiting their resorts to numbers. To date, cDNAs
these polypeptides are presented in GTPase activity), whereas modification that encode 21 distinct G protein a-sub-
Tables I and II. The 13- and ~,-subunits by pertussis toxin prevents receptor- units (the products of 17 genes) have
exist as a tightly associated complex mediated activation of G proteins. been cloned; these can be divided into
that functions as a unit. Although the Although none of the G protein sub- four major subfamilies according to
same [3~,subunit complex can apparently units contains regions that might obvi- amino acid sequence relationships, i.e.
be shared among different a-subunits to ously associate with a lipid bilayer, the those represented by Gs, G~, Gq and G12.
form the heterotrimer, the identity of heterotrimer is bound to the plasma In addition, at least four distinct [3- and
the a-subunit is currently used to membrane. This is apparently due to the six ~,-subunits have been described; it is
define an individual G protein oligomer. fact that 7-subunits are prenylated and a safe bet that these numbers will
The a-subunits have a single, high- at least some a-subunits (those of the increase. The sequence relationships
affinity binding site for guanine G~ subfamily; see below) are myristoyl- between different a-subunits and family
nucleotides (GDP or GTP). The GDP- ated. These lipid modifications serve to groupings are shown in Fig. 2.
bound form of a binds tightly to [~, and anchor the subunits to the membrane The ubiquitous hormone-stimulated
(perhaps by increasing the affinity of adenylate cyclase system 6 and the
J. R. Hepler and A. G. Gilman are at the protein-protein interactions) and they specialized light-activated cGMP phos-
Department of Pharmacology, Universityof also increase the affinity of a for [3~,5 phodiesterase pathway in retinal rod
Texas Southwestern Medical Center, 5323 (J. Ifliguez-Lluhi et al., submitted). A outer segments 7 have served as models
Harry Hines Bird, Dallas, TX 75235, USA. working model of the interplay between for understanding G protein-receptor
© 1992.ElsevierSciencePublishers,(UK) 0376-5067/92/$05.00 383
 Second messenger generation and destruction TIBS 1 7 - OCTOBER 1992

Table I, Properties of mammalian G protein ~-subunits

Family/subunit Mass % Amino Toxinb Tissue distribution Representative receptors Effector/role

(kDa × 10 -3) acid identitya

GS
cCs(s)(2X)c 44.2 100 CTX Ubiquitous BARd, glucagon, ~ 1" Adenylate cyclase
(Zs(L)(2X)c 45.7 - CTX Ubiquitous TSH, others j" 1" Ca 2+ channels
$ Na+ channels
C~olf 44.7 88 CTX Olfactory neuro- Odorant 1" Adenylate cyclase
epithelium
Gi
0~il 40.3 100 PTX Nearly ubiquitous
o¢i2 40.5 88 PTX Ubiquitous M2Ch°' (z2AR' / 1` K÷ channels
o¢i3 40.5 94 PTX Nearly ubiquitous others $ Ca2÷ channels
$ Adenylate cyclase (?)
O¢OAc 40.0 73 PTX Brain, others Met-Enk, c¢2AR, 1` Phospholipase C (?)
C(OBc 40.1 73 PTX Brain, others others 1" Phospholipase A2 (?)

C% 40 68 CTX,PTX Retinal rods Rhodopsin ] 1" cGMP-specific

0¢t2 40.1 68 CTX,PTX Retinal cones Cone opsin ~ phosphodiesterase

C(g 40.5 67 CTX (?), Taste buds Taste (?) ?

PTX

c~z 40.9 60 Brain, adrenal M2Cho (?), $ Adenylate cyclase (?)

platelets others (?) others (?)

Gq
C~q 42 100 Nearly ubiquitous MlCh°' c¢IAR, /
(](11 42 88 Nearly ubiquitous others J $ Phospholipase C-[31,
c~14 41.5 79 Lung, kidney, liver ? [~2, [~3 others (?)

c(15 43 57 B cells, myeloid cells ? ?

c~6 43.5 58 T cells, myeloid cells ? 1" Phospholipase C-131,

-132,-~3
612
c~12 44 100 Ubiquitous ? ?
o~13 44 67 Ubiquitous ? ?

a% Amino acid identity; comparison is with the first-listed member of each family.
bCholera toxin (CTX) and pertussis toxin (PTX) catalyse the ADP-ribosylation of an Arg residue (CTX) and a Cys residue (PTX), respectively, of the indicated
~¢-subunits.
CSplice variants, c~s(s),short forms of C(s; c(s(L),long forms of c¢s.
~Receptor abbreviations: ~AR, ~-adrenergic; M2Cho, M2-muscarinic cholinergic; c~2AR, c¢2-adrenergic; met-enk, met-enkephalin; M1Cho, Ml-muscarinic
cholinergic; c¢IAR, ~¢l-adrenergic.

and G protein-effector interactions. In
both cases, definitive evidence for the
involvement of a particular G protein
was achieved by reconstitution of puri-
fied components (activated a-subunit
and effector protein). Such studies still
provide the most rigorous criterion for
the involvement of a G protein in a
specific signaling pathway.
Hormone and odorant receptors
interact with members of the Gs family
(G~ and GoIL)to stimulate adenylate cy-
clase and thus enhance the rate of
cyclic AMP (cAMP) synthesis. Because
of alternative splicing of a single pre-
cursor mRNA, G~ is expressed as four
distinct polypeptides with predicted
molecular masses ranging from 44 200
to 45700 [although these proteins
migrate on SDS gels as two distinct
bands with apparent molecular weights
384
of 45000 (Gs~.s) and 52000 (Gs~-L)].
These splice variants have not been
well-distinguished functionally. Five dis-
tinct isoforms of adenylate cyclase have
been described to date; all of these are
activated by Gs~. The Gotf protein sub-
unit is expressed exclusively in olfac-
tory neuroepithelium and presumably
serves to link odorant receptors with
an olfactory-specific form of adenylate
cyclase. In addition to activation of
adenylate cyclase, purified Gs~ regulates
at least two ion channels in reconstituted
systems, stimulating dihydropyridine-
sensitive voltage-gated Ca2+ channels in
excised patches from skeletal muscle 9
and inhibiting cardiac Na+channels 1°.
In photoreceptor rod outer seg-
ments, light-activated rhodopsin acti-
vates transducin (Gt]) to stimulate a
cGMP-specific phosphodiesterase; cyto-
plasmic concentrations of cGMP are
thereby decreased 7. Although Gtl is
expressed exclusively in retinal rods, a
second form of transducin, Gt2, is
expressed in cones and presumably
links cone opsins to activation of a dis-
tinct phosphodiesterase. A novel trans-
ducin-like G protein named gusducin
(Gg) has been described recently and is
apparently expressed only in taste
buds 11. The amino acid sequences of Gg
and the transducins are remarkably
similar (80%), particularly in those
regions of the G protein c¢-subunit
known to interact with receptors and
effectors. No information is yet avail-
able about the actions of Gg, although
one might predict the involvement of a
phosphodiesterase.
The physiological actions of a broad
class of hormones, neurotransmitters
TIBS 1 7 - OCTOBER 1992
and growth factors can be explained by
their capacity to activate phosphoino-
sitide-specific phospholipase C (PLC),
an effector enzyme that catalyses
hydrolysis of the minor lipid phospha-
tidylinositol 4,5-bisphosphate [Ptdlns-
(4,5)P2] to form two second mess-
engers, inositol 1,4,5-trisphosphate
[Ins(1,4,5)P3] and diacylglycerol. Com-
pelling evidence for the involvement of
pertussis toxin- and cholera toxin-insen-
sitive G proteins in receptor-mediated
activation of PLC has been obtained
from many laboratories. At least eight
c~-subunits have been described that
are not modified by these bacterial tox-
ins (Table I) and, by default, all are can-
didates. Recent reports have identified
members of the Gq family as regulators
of this pathway. A mixture of Gq and GH
has been purified from bovine brain ~2
and rat liver~3; a closely related protein
has also been isolated from turkey
erythrocytes 14. Reconstitution of these
proteins with purified PLC-[3115,16 or
turkey PLCTM results in specific and
marked stimulation of the enzyme.
Reconstitution of purified recombinant
forms of either Gq~ or Gll ~ with purified
PLC-[3~, PLC-[32and PLC-[33confirm these
findings (J. R. Hepler et al., unpub-
lished); recombinant G~4~has the same
action. Although Gq, G~ and G~4 are
found in many tissues, the other two
members of the Gq family, G~5and G~6,are
expressed only in cells of hematopoietic
lineage (Table I; Ref. 8). GI5~and G~6~are
also more distantly related to Gq~ (57%
and 58% identity). Such limited tissue
distribution suggests a specialized sig-
nalling role for these proteins.
However, reconstitution studies reveal
that purified recombinant G16a also acti-
vates PLC-[31, PLC-[~2 and PLC-[~ (T.
Kozasa et al., unpublished). Further
studies are in progress to determine
the specificity of interactions between
members of the Gq family and the
plethora of isoforms of PLC. Pertussis
toxin inhibits hormonal activation of
PLC in a limited number of tissues, sug-
gesting the involvement of a G~-like pro-
tein rather than a member of the Gq fam-
ily. To date, however, conclusive evi-
dence implicating an individual G~or Go
protein in this pathway is lacking.
Less is known of direct interactions
between effectors and several members
of the G~ subfamily. Only the trans-
ducins have been unambiguously
assigned to a signaling pathway. The
fact that pertussis toxin blocks many
cellular responses implies the involve-
ment of G, family members in a variety
Second messengergeneration and destruction
Table II. Properties of mammalian G protein ~- and ~subunits
Subunit Mass %Amino acid Tissue distribution Effector/role
(kDa x 10 -3) identitya
131 37.3 100 Ubiquitous "~ Required for %-receptor interaction
132 37.3 90 Nearly ubiquitous
[33 37.2 83 Nearly ubiquitous Inhibition of % activation
134 37.2 89 Nearly ubiquitous
Modulate activation of certain
adenylate cyclases by Gs~ or
calmodulin
Support of agonist-induced receptor
phosphorylation and desensitization
T
T1 8.4 100 Retina, other (?) 1" Phospholipase C
T2 7.9 38 Brain, adrenal,
other (?) 1" K÷ channels (?)
T3 8.5 36 Brain, testis
Y4 (?partial) (34) [Kidney,retina (?)] 1` Phospholipase A2 (?)
T5 7.3 25 Liver, other (?)
7.5 35 Brain, other(?) j
a% Amino acid identity: comparison is with the first-listed member of each family.
of signaling pathways. The most widely fled Go~. However, demonstration of
studied of these include inhibition of direct interactions between the G protein
adenylate cyclase and activation or a-subunits and K+ channel proteins in
inhibition of several ion channels - lipid bilayers has not yet been reported
roles that are attributed to the three G~ (the relevant channels have not been
subunits and to the two splice variants purified) and some believe that G pro-
of Go~. Stimulation of many hormone tein [3y-subunits may also play an
receptors results in inhibition of aden- important role 23.
ylate cyclase activity. This response is Cellular concentrations of the G~ and
blocked by treatment with pertussis G proteins are much higher than are
toxin, and there is ample evidence for those of Gs and Gq. In brain, Go~ is 1-2%
involvement of the G~ proteins. Never- of membrane protein. It seems unlikely
theless, attempts to inhibit adenylate that the protein is used only to regulate
cyclase activity using purified G~ K~ and Ca2+channels. If so, why is Go so
polypeptides have met with limited suc- concentrated in neural growth cones24?
cess at best 17,18.These observations led Recent findings suggest that members
to the idea that [3y derived from the G~ of the G~ family may regulate pathways
oligomer might inhibit adenylate deep within the cell. In a kidney cell
cyclase indirectly by interaction with line, G~3 has been localized to the Golgi
its activator, Gs~19. Alternative expla- complex. Overexpression of Gi~3 in
nations include the ability of [3~'to inhibit these cells slows secretion of packaged
certain types of adenylate cyclase di- proteins and associated vesicular trans-
rectly2° and the possibility that effects port, and this effect is reversed by
of G~ proteins on adenylate cyclase treatment with pertussis toxin 25. Fur-
activity are mediated indirectly21. Per- ther evidence comes from studies with
tussis toxin-sensitive G proteins also the fungal toxin brefeldin A, which
regulate ion channels in muscle, neur- stimulates release of the Golgi mem-
ons and elsewhere. For example, mus- brane coat protein [3-COP to cause a
carinic cholinergic receptors activate K+ pathological fusion of Golgi membranes
channels in cardiac myocytes in a mem- with those of the endoplasmic reticu-
brane-delimited fashion. The pathway lum. Treatment of perforated cells with
can be blocked by pertussis toxin, and GTPyS or AIF4 antagonizes the actions
channel activity is responsive to guan- of brefeldin A26,27.Furthermore, addition
ine nucleotides. Addition of activated of G protein [3y-subunits to a cell-free
G~ proteins to membrane patches ex- system prevents GTPTS-mediated bind-
cised from these cells stimulates K+ ing of B-COP to Golgi membranes 28.
channel activity; G~I, Gia2 and Gia3 work Taken together, these findings indicate
equally well 22. Similarly, certain neur- that a heterotrimeric G protein of the
onal K+ channels are activated by purl- G~ subclass serves an important
385
 Second messenger generation and destruction TIBS 17 - OCTOBER 1992

cal evidence for interactions of [37 with

Basal state jr Receptor activation effectors has come from study of dif-
H ferent isoforms of adenylate cyclase.
(a) (b) Although the adenylate cyclases ex-
GTP pressed in most peripheral tissues
appear to be largely immune to regu-
GDP lation by 137, at least some of the en-

1 zymes found largely in brain are regu-

lated by [~7 in a type-specific fashion2°.
Type I (Gs~- and calmodulin-activated)
(e) (e) adenylate cyclase is inhibited by 97,
apparently directly. Type I! and type IV
adenylate cyclases are activated con-
eo~.N ditionally by ~7; that is, ~7 is stimulat-
GTPase
/ Subunit dissociation
ory only when Gs~ is also present. This
appears to represent a mechanism for
crosstalk between signaling pathways.
(d)
Concentration requirements suggest
that 137would arise by dissociation of G~
or Go, while Gs~ would obviously be lib-
S P erated by activation of G~-linked recep-
Effector activation tors. Other effectors also appear to be
subject to regulation by ~7. An un-
defined form of PLC in HL-60 granulo-
Figure 1
G protein-mediated transmembrane signaling. In the basal state (a), G proteins exist as
cytes is markedly activated by 137~°. In
heterotrimers with GDP bound tightly to the c(-subunit; the hormone receptor (R) is un- addition, ~7 appears to have effects on
occupied and the effector (E) is inactive. Upon hormone binding and receptor activation K÷ channel activity, although it is un-
(b), the receptor interacts with the heterotrimer to promote a conformational change and known if these effects are mediated
dissociation of GDP from the guanine nucleotide-binding site; at normal cellular concen- directly or indirectly23. Recent findings
trations of guanine nucleotides, GTP fills the site immediately. (Under experimental con- also indicate that 137plays an obligatory
ditions where GTP is absent, the hormone has high affinity for the receptor and the H-R-G role in agonist-induced receptor phos-
protein complex is stable.) Binding of GTP to c~ induces a conformational change with two
consequences (c). The G protein dissociates from the H-R complex, reducing the affinity
phorylation and desensitization3L Taken
of hormone for receptor and, in turn, freeing the receptor for another liaison with a neigh- together, these observations suggest
boring quiescent G protein. GTP binding also reduces the affinity of (z for 13)', and subunit that dissociation of G protein subunits
dissociation occurs. This frees (z-GTP to fulfill its primary role as a regulator of effectors in the membrane can generate parallel
(d). At least in some systems, the free J3ysubunit complex may also interact directly with and/or interactive signals via both 0(-
an effector (El) and modulate the activity of the active complex, or it may act independently and [37-subunits. Investigators should
at a distinct effector (E2). The c(-subunits possess an intrinsic GTPase activity (e). The rate
have an open and cautious attitude
of this GTPase determines the lifetime of the active species and the associated physiologi-
cal response. The c(-catalysed hydrolysis of GTP leaves GDP in the binding site and causes about which G protein subunit (and
dissociation and deactivation of the active complex. The GTPase activity of (z is, in whether one or both subunits) is me-
essence, an internal clock that controls an on/off switch. The GDP bound form of (z has diating the effect in question.
high affinity for [37; subsequent reassociation of c~GDPwith I~Y returns the system to the It is clear that many more G protein-
basal state (a). regulated pathways exist, and several
poorly understood cellular responses
appear to be dependent on guanine
regulatory role in membrane trafficking. Traditionally, the 0(-subunit has been nucleotides, activated by A1F;, and/or
Information regarding a receptor-like viewed as the 'business end' of a G pro- sensitive to pertussis toxin. Likely ef-
analog in this pathway is lacking, tein because it binds and hydrolyses fectors include phospholipases A 2 and
although mastoparans (peptides that GTP and interacts with effectors. In D and a plethora of ion channels, trans-
mimic the actions of G protein-coupled contrast, the ]3~'-subunit complex has porters and exchangers. Cellular re-
receptors) stimulate the binding of [~- been viewed as a regulatory component sponses under consideration include
COP to Golgi membranes and block the for 0( which stabilizes the GDP-bound growth control, effects of tyrosine-
effects of brefeldin A in a pertussis form of 0(, 'presenting' 0( to receptors, kinase growth factor receptors, exo-
toxin-sensitive manner 27. These findings serving as a membrane anchor for the cytotic secretory events and protein
provide the most compelling evidence oligomer. Yet the activation of G pro- translocation. The receptors, G pro-
to date that the actions of heterotrimeric teins yields two subunits (0(/137) that teins and effectors involved in many of
G proteins are not limited to transmem- could act on downstream targets. A these pathways remain unknown.
brane signaling at the cell surface. Yet, growing body of evidence now supports Conversely, there is little knowledge of
one is left uncomfortable with the the idea that free ~Y can itself interact the actions of some 'orphan' G proteins,
assignment of the same G protein to functionally with effector proteins. such as 612 , 613, 615 and G~. Further-
play seemingly unrelated regulatory Genetic studies first demonstrated that more, we have discussed examples
roles in presumably different cellular responses to mating factors in budding where a single G protein 0(-subunit (e.g.
pathways. It seems likely that an unsus- yeast are dictated by [~yrather than 0(29. G~) can regulate more than one ef-
pected common theme will emerge. In mammalian systems, direct biochemi- lector; thus, the assignment of a given G
386
TIBS 17 - OCTOBER 1992 Second messengergeneration and destruction

[37-subunits pre- References

40 60 80 100 sumably interact 1 Hall, A. (1990) Science 249, 635-639
¢,t I ' I ' I 2 Bourne, H. R., Sanders, D. A. and McCormick, F.
' I ' I ' I ' I to dictate the spe- (1990) Nature 348, 125-132

I-- s 1 Gs 0~ol f
cificities of these
choices and to
control, synergis-
3 Bourne, H. R., Sanders, D. A. and McCormick, F.
(1991) Nature 349, 117-127
4 Kaziro, Y. et al. (1991) Annu. Rev. Biochem. 60,
349-400

__
~ - -
Oql
~i3
0~i2
tically or in op-
position, the acti-
vities of effectors.
5 Linder, M. E. eta/. (1991) J. Biol. Chem. 266,
4654-4659
6 Gilman, A. G. (1987) Annu. Rev. Biochem. 56,
Thus, activation 615-649
7 Stryer, L. (1986) Annu. Rev. Neurosci. 9, 87-119
f I OoA Gi
- - O~oB
of a given recep-
tor in different
cells can produce
8 Simon, M. I., Strathmann, M. P. and Gautam, N.
(1991) Science 252,802-808
9 Mattera, R. eta/. (1989) Science 243, 804-807
F~ O~tl a highly varied 10 Schubert, B., VanDongen, A. M. J., Kirsch, G. E.
(Zt2 and Brown, A. M. (1989) Science 245,
O~g constellation of 516-519
activated G pro- 11 McLaughlin,S. K., McKinnon, P. J. and
~z teins and effec- Margolskee, R. F. (1992) Nature 357, 563-569
-// tors, and cellular 12 Pang, I-H. and Sternweis, P. C. (1990) J. Biol.
Chem. 265, 18707-18712
~ 1 O~q responses will 13 Taylor, S. J., Smith, J. A. and Exton, J. H. (1990)
0('11 change with deve- J. Biol. Chem. 265, 17150-17156
~14 lopment or with 14 Waldo, G. L., Boyer, J. L., Morris, A. J. and
Gq cellular history of
Harden, T. K. (1991) J. Biol. Chem. 266,

4 0(,16
latory signals.
14217-14225
exposure to regu- 15 Smrcka, A. V., Hepler, J. R., Brown, K. O. and
Sternweis, P. C. (1991) Science 251, 804-807
Such vast 'com- 16 Taylor, S. J., Chae, H. Z., Rhee, S. G. and Exton,
0~12 J. H. (1991) Nature 350, 516-518
0~13
G12 binatorial power' 17 Katada, T. et al. (1984) J. Biol. Chem. 259,
endows cells and 3586-3595
I , I , I , I I I I I I I organisms with 18 Linder, M. E., Ewald, D. A., Miller, R. J. and
40 60 80 100 extraordinary ca- Gilman, A. G. (1990) J. Biol. Chem. 264,
8243-8251
% Amino Acid Identity pacity for fine 19 Gilman, A. G. (1984) Cell 36, 577-579
tuning both the 20 Tang, W-J. and Gilman, A. G. (1991) Science
magnitude and 254, 1500-1503

Figure 2 the nature of 21 Federman, A. D. eta/. (1992) Nature 356,

Sequence relationships between mammalian G. subunits and
family groupings(modified,with permission,from Refs8 and 4).
protein to a specific effector does not
end 'the game'.
Conclusions
As the number of identified G protein
subunits continues to grow, the task of
unraveling the complexity of the G
protein-regulated cellular switchboard
expands exponentially. The number of
identified (z-, ~- and 7-subunits indicates
a limit of nearly 1 000 possible oligo-
meric combinations; reasonable extra-
polation suggests that this number
could grow to roughly 5 000 (with fur-
ther discovery). Multiplication by the
number of G protein-linked receptors
and effectors indicates that an individ-
ual cell must sift through a very large
number of choices to complete its own
customized, G protein-controlled regu-
latory network. Further complexity at
each step in the signal transduction
cascade compounds the problem. Thus,
there is both divergence and conver-
gence of protein-protein interactions at
both the receptor-G protein level and
the G protein-effector level, and (z- and
their responses 22 Yatani, A. et al. (1988) Nature 336, 680-682
to the environ- 23 Logothetis, D. E. et al. (1987) Nature 325,
ment.
Acknowledgements
The authors would like to thank
M. Linder, T. Kozasa, J. l~iguez-Lluhi
and N. Ueda for hlepful comments.
Work from the authors' laboratory was
supported by NIH Grant GM34497,
American Cancer Society Grant BE30N,
The Perot Family Foundation, The
Lucille P. Markey Charitable Trust and
The Raymond and Ellen Willie Chair of
Molecular Neuropharmacology. J. R. H.
is the recipient of a National Research
Service Award F32 GM13569.
159-161
321-326
24 Strittmatter, S. M. et al. (1990) Nature 344,
836-841
25 Stow, J. L. et al. (1991) J. Cell Biol. 114,
1113-1124
26 Donaldson, J. G., Lippincott-Schwartz,J. and
Kiausner, R. D. (1991) J. Cell Biol. 112,
579-588
27 Ktistakis, N. T., Linder, M. E. and Roth, M. G.
(1992) Nature 356, 344-346
28 Donaldson, J. G., Kahn, R. A., Lippincott-
Schwartz, J. and Klausner, R. D. (1991) Science
254, 1197-1199
29 Blumer, K. J. and Thorner, J. (1991) Annu. Rev.
Physiol. 53, 37-57
30 Camps, M. et al. (1992) Eur. J. Biochem. 206,
821-831
31 Haga, K. and Haga, T. (1992) J. Biol. Chem.
267, 2222-2227
387