TIBS 1 8 - SEPTEMBER 1 9 9 3

G proteins

and . ARK: a

new twist for the coiled

coil

Activation of heterotrimeric G proteins by hormone or sensory receptors may propagate bifurcating signals through the parallel action of the GTP-liganded a subunit and the free ~ complex. The effector targets of the various G-protein a subunits undergo activation or inhibition until the biological effect is terminated by GTP hydrolysis. Evidence that free [57 complexes can exert independent effects on key regulat- ory enzymes and ion channels is accumulating at a dizzying tempo. Apart from the role of [57 homologs in Saccharomyces cereuisiae in mediating the complex effects of pheromones in the mating response, it is now clear that multiple effectors in higher organ- isms respond to free [57 complexes, including specific subtypes of adenylate cyclase and phospholipase C, and receptor kinases such as the [5-adrener- gic receptor kinase ([SARIC).

Current structural models of G- protein subunit interactions emphasize domains on the a subunit that interact with the [5 7 complex and effectors I. A conceptual framework for the inter- actions of [5 and 7 subunits with each other, with Ga and with [3y-responsive effector molecules has been lacking. (a) Recently we proposed a novel struc- tural model for the interaction of [5 and 7 subunits through a two.stranded a- helical coiled coil 2,3. The model was based in part on computer analysis with a predictive algorithm that scores the probability of coiled-coil formation by comparing the primary protein sequence with sequences in a database of known coiled-coil domains 4, and was tested by site-directed mutagenesis of the [5 subunit 3.

Coiled coils are stabilized protein structures resulting from the inter- action of two or more right-handed a- helices that wind around each other in a left-handed super~:oil s-7. The con- stituent a-helices exhibit a specific seven amino acid periodicity ('heptad repeat') which favorably positions cer- tain amino acid sidechains for inter- helical contacts. Residues at positions a and d of the heptad are generally non- polar, which creates a hydrophobic rib- bon along one side of each helix, pro- viding an interface for interhelical interactions, whereas ionic bonds between residues in the e and g pos- itions of neighboring helices are often present, which reinforce the inter- molecular assembly. Extended two- or © 1993. Elsevier Science Publishers, (ILK) 0968-0004/93/$06.00

three-stranded coiled coils are common in fibrous proteins such as tropomyosin and fibrinogen, and shorter coiled coils (four or five heptads) are found as dimerization domains in transcriptional regulatory proteins such as Jun, Fos and GCN4 ('leucine zippers') ~-~.

We tested our model of [5 and 7 inter- action by mutagenesis of a region of predicted coiled-coil structure at the amino terminus of the [5 subunit, which is postulated to interact with a similar domain in the 7 subunit2'3. Deletion of this coiled-coil domain or mutation of a

single residue (EIOK) predicted to form an interhelical ionic bond was sufficient to prevent [5-7 association3- Further- more, insertion of two aianines within the 15 coiled-coil region - a mutation designed to disrupt the heptad period- icity but maintain the propensity for a- helical secondary structure - blocked interaction with 7, even though similar alanine insertions in two other regio~ls did not 3. Beyond the proposed role of a coiled-coil structure for [3 and 7 (Refs 2, 3), which are strongly interacting sub- units that require denaturation for dis- sociation, it was recognized that the a subunit may also contain a coiled.coil domain in its a-helical amino terminus z. Thus the G-protein heterotrimer, with in its GDP-bound form, might contain a three-stranded coiled coil involving the a, [5 and 7 subunitsZ- In this model the dissociation of c( from the [~7 complex would occur due to a GTP-induced con- formational change in the a subunit amino terminus diminishing its coiled coil propensity thereby disrupting the three-stranded coil 2.

J3IURK (639-670) ( a b c d e f g

E

n: IA Io OILIv 0 R . . . . . . . IV . I , ~I~IK , K

~1 11-311 a b o d e f g

n:ii °H Q K D RKA D TL S

y2 (6-36| abc de f g

TAS 0 RKL Q KME I RI K K AAD

Rgum 1 A coiled-coil domain in the [lARK carboxyl terminus mediates binding to the 97 complex through the reversible formation of a triple-stranded coiled coil. (a) Heptad repeats in [~ARK-1, [31 and 72. Shown in upper case are the sequences of the predicted coiled-coil structure in ~ARK-1 (639-667), [It (1-31), and I'2 (6-36) (Refs 2,4,11). The hallmark hydrophobic residues in heptad pos tions a and d are boxed. Ti~e se~,uenc~ of 6ARK-1 cor- responding to 'Peptide G' of Koch et al. it, which inhibits the [IARK-[:~7 interaction, is under- lined, including three residues (668-670) that are not part of the predicted coiled coil. (b) Schematized helical view of the proposed triple-stranded coiled-coil structure involving hep- tad repeats in pARK and [ly. Potential stabilizing interactions between neighboring helices, including ionic bonds between positions e and g and hydrophobic interactions involving positions a and d, are shown by dashed lines. Asterisks indicate positions of the ~sub;rlit helix containing potential effector-contact residues in the yeast homoiog STE4 (heptad position c, K55, D62; position d, A56; position g, Q52, K59; see Ref. 12).

315

TIBS 18 - SEPTEMBER 1993

t ~ G T P

I

F

a-GDP

pARK

I

Rgure 2 The J3~, cycle of G proteins. A simplified view of the cyclic interaction of the ~7 complex with G= and [3ARK through formation of reversible triple-stranded coiled coils. The heterotrimer with triple-stranded coiled¢oil structure (G. in its GDP-bound form, antiparallel with respect to J~7) is anchored to the plasma membrane by lipid modifications including myristoylation and/or palmitoylation of cx, and isopreny- lation of 7 (top). Upon interaction with agonist-activated receptor, GTP binding to ~x causes allosteric disruption of its amino-terminal coiled-coil domain (shown as a break in the helical region) resulting In dissociation from ~¥, exposing the two-stranded ~Y docking site (right), I3ARK is translocated from the cytosol to the membrane through binding of a carboxy.terminal coiled-coil domain to the 13'/ docking site, creating a new triple.stranded coiled coil (J3ARK parallel with respect to 137), and Is poised to phosphorylate the egonist-occupied receptor (bottom), The cycle is completed as I~ARK dissociates from ~7, through an unknown mechanism perhaps involving a 'hinge' at P662 (shown as black line), The process is favored by reassociation of G.-GDP with l~7 to reform a triple-stranded coiled coil, Possible interaction between other regions of ~ARK and ~7 is of course not excluded by this model.

With respect to other interactions of the [57 complex, the recent appreciation that receptor kinases such as [SARK serve as novel downstream targets of [57 provides the first detailed clue as to how [57 subunits may interact with effectors. Such receptor kinases under- go ~y-dependent activation and trans- location to the membrane, promoting phosphorylation and desensitization of agonist-occupied receptors s-L0. In an elegant series of experiments, Lefkowitz and co-workers showed that the inter. actions of G-protein a subunlts and [SARK with [57 are mutually exclusive l°. They also localized the {57-binding domain of [3ARK to its carboxy-terminal 222 amino acids ~° and have now mapped the [57-binding site to a smaller region, approximately 28 amino acids in length u.

316

The role of three.stranded coiled¢oll formation

Our studies on the assembly of the [57 complex 2,3, combined with sequence analysis of this [57-binding domain, now suggest a novel interpretation of the 15ARK findings, which offers a mech- anism for the way in which [57 interacts exclusively with either the ~x subunit or effectors such as receptor kinases: the reversible formation of three-stranded coiled coils (Figs 1,2). Although not appreciated in the recent mapping study a, the region of ISARK comprising the [s7-binding domain OV643-$670) is nearly superimposable on a 29 amino acid region predicted to contain a coiled<off domain (E639-K667)* (Fig. la). The presence of coiled coils in both the [SARK 157-binding domain and in the [57 heterodimer is unlikely to be

mere coincidence. An edu- cated guess would be that, analogous to the proposed l~7-a interaction 2, the two- stranded coiled coil in [~¥ provides a docking site for the compatible region in the carboxyl terminus of I~ARK to form a three- stranded coiled coil (Fig. lb). Such a mode of inter- action would account for the observation that the interactions of [~ARK and G~ with [~7 are mutually ex- clusive l°. This interpret- ation would also provide a timing mechanism to limit the duration of the J~7-- [SARK interaction, as the G~ helix replaces the I3ARK helix on the [~y docking site in a proposed functional cycle (Fig. 2). In this regard the unusual occurrence of a proline (P662) in the putative [3ARK coiled<off region* may reflect the need for a 'hinge' structure to impart reversibility to the [SARK-[~ interaction by allowing variable participa- tion of the fourth heptad. Such a mechanism would be functionally analogous to the GTP-induced confor- mationai change proposed for the G a coiled<oil region =.

Competition between (x-GDP and effector molecules for binding to a two-stranded ~7 coiled coil could provide a general

mechanism for termination of [S7--effec- tor signalling. Do the ~ARK findings have any pertinence to other [S7-regulated effectors? Recent evidence in yeast independently implicates the [57 coiled- coil region in effector signalling 12. Whiteway and co-workers localized the effector interaction regions of STE4, the yeast [5-subunit homolog, to two dis- crete clusters of residues, one of which maps to amino acids 52--62 (Ref. 12).

*The carboxy-terminal 125 residues of IIARK-1 were analysed by the coiled-coil predictive algorithm as previously described 2,4. A single region of elevated score from E639 to V661 was identified (probability -62%). The presence of an additional heptad was suggested by in-frame hydrophobic residues at positions a and d, but was scored lower because of the proline at position b (P662). Replacement of this proline with alanine elevated the overall coiled-coil probability from E639 to K667 to 98%.

TIBS 18 - SEPTEMBER 1 9 9 3

Although unrecognized in this study 12, this group of amino acids falls within the series of predicted coiled<oil hep- tads homologous to those found in mammalian ~ subunits 2,4. Furthermore, these residues segregate to heptad positions c, d and g of the proposed []- subunit (STE4) helix, providing a clue as to which 'face' of the [57 coiled coil might serve as the active site for effec- tor or G,, binding (Fig. It)) n. The pres- ence of an essential effector contact site within the coiled<oil region of [37 might allow 'switching off' by G,, reas- sociation, whether or not the target molecule interacts with [57 via a three- stranded coiled coil.

Future prespe©ts While confirmation of proposed

coiled<oil structures awaits definitive biophysical characterization, further support may be forthcoming from additional mutagenesis experiments directed selectively at coiled<oil struc- tures. Mutation of the putative 'hinge'

proline (P662) in [3ARK, for example, might increase its affinity for J57 by enhancing coiled<oil formation. The design of inhibitors of ISARK-[5¥ interac- tion in the context of a coiled<oil model might facilitate the identification of new tools to diminish receptor desensitization u. The view presented here unifies biochemical, genetic and modelling data and offers a new way to envision certain interactions among G-protein subunits and effector molecules.

Acknowledgements The authors are grateful to Andrew

Shenker for his critical review of the manuscript and to Allen Spiegel for his support and encouragement.

WILLIAM F. SIMONDS, HUSSEINI K. MANJI AND ANJA GARRITSEN

Molecular Pathophysiology Branch, National Institutes of Diabetes, Digestive and Kidney Diseases, Bethesda, MD 20892, USA.

Straightening out the dihedral ' angles

ANDREI N. LUPAS

Max Planck Institut for Biochemie, Am Klopferspitz 18a, W-8033 Martinsried, Germany.

References 1 Conklin, B. R. and Bourne, H~ R. (1993) Cell 73,

631-641 2 Lupas, A. N., Lupas, J. M. and Stock, J. B.

(1992) FEBS Lett. 314, 105-108 3 Garritsen, A., van Galen, P. J. M. and Simonds,

W. F. (1993) Proc. Natl Acad. ScL USA 90, 7706-7710

4 Lupas, A,, van Dyke, M. and Stock, J. (1991) Science 252, 1162-1164

5 Alber, 7. (1992) Curt. Opin. Genet. Dev. 2, 205-210

60'Shea, E. K,, Klemm, J. D., Kim, P. S. and Alber, T. (1991) Science 254, 539-544

7 Cohen, C. and Parry, D. A. (1990) Proteins 7, 1-15

8 Haga, K. and Haga, T. (1990) FEBS Lett. 268, 43-47

9 Haga, K. and Haga, T. (1992) J. BioL Chem. 267, 2222-2227

10 Pitcher, J. A. et al. (1992) Science 257, 1264-1267

11 Koch, W. J., Ingleae, J., Stone, W. C. and Lefkowitz, R. J. (1993) J. Biol. Chem. 268, 8256-8260

12 Leberer, E. eta/. (1992) EMBOJ. 11, 4805-4813

Ramachandran was the first to express the need for an analytical description of the polypeptide chain configurationL He defined the position of the two planar peptlde bonds around the C a atom by two dihedral angles; to and to' (which was later changed to ¥), which defined the sense of orientation and zero point. But he was somewhat unlucky in his definition; Indeed his proposals we':e not In line with conventions in use In or- ganic and polymer chemistry, In order to avoid confusion and controversy, a change in standard angle definition was first proposed in a concerted action by a group of leading protein biochemists headed by Edsall who published their proposal in the Journal of Biological Chemistry 2 and simultaneously in Bio- polymers and the Journal of Molecular Biology. A final proposal was intro- duced by the IUPAC-IUB Commission on Biochemical Nomenclature 3 and is illustrated in Fig. I and Table I.

While the IUPAC-IUB Nomenclature was in preparation, Dickerson and Gels published their book, The Structure and Action of Proteins 4, which contains superb drawings on protein structure. The suc- cess of these drawings is responsible for the present-day confusion about the description of the dihedral angles. © 1993, Elsevier Science Publishers, (UK) 0968-0004/931506.00

Dickerson and Gels took a quite inde- pendent view and defined the (to, ¥) angles in their own way: they used the Ramachandran proposed sense of orien- tation of the (to, ¥) angles but defined the zero ¥ in another way. They prob- ably felt unhappy about the forthcoming IUPAC-IUB Nomenclature because in a footnote in their book they make the following remark:

Apparently yet another shift of definitions is being considered by the IUPAC-IUB Commission on Biochemical Nomenclature. This will achieve consistency with the usage of organic chemistry, but will make the literature virtually unreadable. The new (~0, ¥) values can be obtained from the ones in this book by subtracting 180 o from both angles. According to tra- ditional sources, there are two types o! sin: sins of omission and sins of commission. This is most definitely a sin of commission.

To the best of our knowledge, only Schulz and Schirmer explicitly explain and illustrate in their book, Principles of Protein Structure, how the dihedral angles of the peptide bond are defined 5.

Table I. Dihedral angles for various conformations of a polypeptlde chain of L-amino acids

to(o) Rotation around N,-C", bond ~(o) Rotation around C ~,-C', bond

0 (=360 °) C~,-C ', bond (trans to N,--H, bond) 0 (=360 °) C",-N, bond (trans to C'-O, bond)

+60 C',-H a, bond (c/s) +60 C",-R, bond (c/s)

+120 C~-R, bond (trans) +120 C",-H =, bond (trans)

+180 C'~--C', bond (cis) +180 C",-N, bond (cis)

+240 +240 (=-120) C=,-H=~ bond (trans) (=-120) C",-R, bond (trans)

+300 +300 (---60) C=,-R, bond (cis) (=-60) C~,-H Q, bond (cis)

317