J. Biochem. 114, 535-540 (1993)

Recoverin Alters Its Surface Properties Depending on Both Calcium-Binding and N-Terminal Myristoylation1

Mikio Kataoka, Ken'ichi Mihara, and Fumio Tokunaga Department of Biology, Faculty of Science, Osaka University, Toyonaka, Osaka 560

Received for publication, June 16, 1993

The solution structure and calcium-dependent structural changes of recoverin, a 23kDa calcium binding protein of vertebrate photoreceptors, have been studied by small-angle X-ray scattering and CD, as well as the effect of N-terminal myristoylation. The CD spectrum is not affected by N-terminal myristoylation, but strongly affected by Ca", indicating that N-terminal myristoylation alone does not cause a conformational change. The major conformational change in recoverin induced by Ca2+ is characterized as a decrease in the a-helical content of the protein and an increase in global size upon removal of Ca". In the presence of Ca", unmyristoylated recoverin is monomeric and globular in solution, while N-terminal myristoylation brings about aggregation. In the absence of Ca", unmyristoylated recoverin tends to aggregate, while myristoylated recoverin becomes monomeric and globular. These observations indicate that recoverin changes its surface properties depending on both calcium binding and N-terminal myristoylation. Melittin interacts non-specifically only with the myristoylated recoverin in the absence of Ca2+. This may be indicative of the properties of the interaction between recoverin and its normal physiological target enzyme.

Many cellular functions are regulated by intracellular free duction cascade, depending on the free Ca" concentration, calcium ions. Calcium receptor proteins acting as molecular rather than mediate direct regulation of guanylate cyclase switches mediate these regulatory processes. The calcium or phosphodiesterase activity (8, 9). In fact, the function receptor proteins detect changes in the intracellular free and mechanism of recoverin as a Ca2+-dependent guanylate calcium ion concentration which are produced by primary cyclase activator were denied by the discoverers them- stimuli, and alter their interaction with their target en selves (10). Although uncertainty remains as to the iden zymes. A common structural motif present in many cal tity of the physiological target enzyme of recoverin (8, 9, cium receptor proteins is known as the EF-hand (1). To 11), recoverin is an EF-hand calcium regulatory protein. understand the mechanism of action of these EF-hand A further important and interesting feature of recoverin proteins, it is essential to characterize calcium induced is that its N-terminus is heterogeneously acylated with one changes in their structures. of four types of fatty acyl residues including myristoyl Recoverin is a 23kDa EF-hand calcium binding protein (C14:0), C14:1, C14:2, and C12:0 (12). N-terminal acyl isolated from bovine retinal rod (2, 3). The protein was residues would be expected to play a role in protein- firstly discovered as a calcium-dependent modulator of membrane interactions (11, 13), rather than in protein- retinal guanylate cyclase (2, 3). The enzyme cascade protein interactions (14). Ray et al. (15) demonstrated that following the absorption of light by rhodopsin causes the the fluorescence emission spectra of recoverin are respon closure of a cyclic GMP dependent membrane channel, sive to both myristoylation and the binding of Ca2+, sugges which hyperpolarizes the plasma membrane, and lowers ting that both the N-terminal fatty acyl residues and the cytosolic Ca2+ concentration (4-6). Reopening of the calcium binding alter the conformation of recoverin. channel, which is induced by binding of cyclic GMP, is We investigated the conformational change in recoverin required for recovery to the dark state. Since cyclic GMP is upon binding of calcium, as well as the effect of myristoyla tion on its solution structure by small-angle X-ray scatter- hydrolyzed by phosphodiesterase during photoexcitation, the synthesis of cyclic GMP by guanylate cyclase and ing (SAXS) and CD. Although SAXS is a lower resolution inhibition of phosphodiesterase are necessary for the technique compared with X-ray crystallography, it is useful for detecting conformational changes under various solvent recovery. It was suggested that recoverin activated guan conditions, which is frequently impossible to achieve with ylate cyclase at lower Ca2+ concentrations (2). On the other hand, a homologous protein, S-modulin, was isolated from protein crystals. In fact, calcium-dependent conformational changes in calmodulin, a typical EF-hand calcium modula frog retinal rod as a calcium-dependent modulator of tor, have been demonstrated with this technique, including phosphodiesterase (7). Recent studies suggested that recoverin-like proteins modulate termination of the trans the effect of binding of target peptides (16-21). The conformational change in calmodulin on the binding of both 1A part of this study was supportedby Grants-in-Aidfrom the calcium and target peptides has been confirmed subse Ministryof Education, Science and Cultureof Japan to M.K.and F.T. quently by both NMR solution structure analysis (22) and Abbreviations: DTT, dithiothreitol; MW, molecular weight; Rg, X-ray crystallography (23). radius of gyration;SAXS, small-angle X-ray scattering.

V,,! '_14, No. 4, 1993 535 536 M. Kataoka et al.

SAXS measurements indicate that recoverin is globular tained at five or more protein concentrations (17). The final in solution, in contrast to the dumbbell shape of calmodulin extrapolated data were analyzed by Guinier analysis (26) and troponin C. Calcium binding brings about a confor and the indirect Fourier transform method of Moore (27). mational change in recoverin. N-terminal myristoylation Data analyses were performed with a personal computer, alone does not bring about a major conformational change in PC9801, or an ACOS 2020 of the Computer Center of recoverin, but appears to alter the surface properties of Osaka University. The details of the measurements and the recoverin. The effect of the binding of recoverin to its target analysis techniques have been presented previously (17- protein on its solution structure is physiologically impor 19). tant. Since we have no information about the true target of Measurement of CD Spectra-CD spectra were mea recoverin, we have tried to examine the effect of binding to sured with a JASCO spectropolarimeter, model J-500A melittin, a bee venom peptide, as a model. An effect of (JASCO, Tokyo), controlled with a personal computer, melittin was observed only for the myristoylated recoverin PC9801, using a quartz cell of 1mm path length. The in the absence of Ca 2+. sample concentration was 0.1 or 0.2mg/ml. Each spectrum was collected 4 times. The temperature was maintained at

MATERIALS AND METHODS 20°C by circulating thermostated water through the cuvette holder. Preparation of Recoverin•\Both myristoylated and unmyristoylated recoverins were expressed in Escherichia RESULTS coli, and purified as described by Ray et al. (15). The purified recoverins were generous gifts from Drs. S. CD spectra of unmyristoylated and myristoylated recover Zozulya and L. Stryer at Stanford University. ins show a conformational change in the protein on the Preparation of Samples for X-Ray Scattering-Stock binding of calcium, although the myristoyl modification solutions of recoverins were concentrated with a Centricon- alone does not bring about a major conformational change 10 (Amicon, Danvers, MA) to ca. 30 mg/ml. The final (Fig. 1). The change in the spectrum on binding Ca2+ is concentrations were determined by measuring the dry essentially the same for both unmyristoylated and myris weight. Melittin was purchased from Sigma and purified by toylated recoverins. The spectral change is characterized as HPLC before use (24). Samples of recoverin alone or of recoverin mixed with equimolar amounts of melittin were dialyzed against one of the solutions described below with a Spectra/por 4 dialysis membrane. The dialysis of 0.2 to 0.5ml samples was performed, at 4`C, against three changes of 100ml of the described solutions, the first two changes being after 24 h each, and the final dialysis being for 72 h to ensure complete equilibration. At the end of the final dialysis, the samples were removed from the dialysis tubing, and aliquots were quantitatively diluted with the final dialysis fluid to produce a concentration series ranging from 3 to 30mg/ml. The dialysis solutions all contained 100mM KCl, 20mM Tris, 1mM DTT, 1mM MgCl2, and 5% glycerol (pH8.0). For samples in the presence of calcium, the initial dialysis fluid included an additional 1mM CaCl2; the subsequent two changes of dialysis fluid each contained 0.1mM CaC12. For samples in the absence of calcium, the initial dialysis fluid contained 5mM EGTA, and subsequent changes contained 1mM EGTA. The final dialysis fluids were used for the measurements of the backgrounds. Small-Angle X-Ray Scattering Measurements -SAXS measurements were performed with a SAXES camera installed at BL10C of the Photon Factory at Tsukuba using synchrotron radiation (19, 25). The samples were measured in a specially designed cell with quartz windows that were 15 ,u m thick, 10mm wide, and 3mm high. The path length was 1mm. Samples were maintained at 20°C by circulating thermostated water through the cell holder. The exposure times were 15min for samples with lower concentrations than 8 g/ml, and 10min for the others. Scattering profiles were recorded with a 20-cm-long position-sensitive proportional counter

(Rigaku, Tokyo). The data were transferred to a personal computer, PC9801 (NEC, Tokyo). Fig. 1. CD spectra of unmyristoylated recoverin (a) and Scattering curves at infinite dilution were obtained by myristoylated recoverin (b). Curve 1: in the presence of Ca2+; linear extrapolation of a series of scattering curves ob curve 2: in the absence of Ca2+.

J. Biochem. Ca2+ Dependent Conformational Change of Recoverin 537

a change of negative ellipticity around 222nm , suggesting th tially the same (15). These facts indicate that myristoyl at the a-helical content decreases upon the release of modification does not bring about a global conformational Ca2+. The helical content estimated from [ƒÆ]222nm (28) is change in recoverin. Therefore, we conclude that it is the 65% in the presence of Ca 2+ and 52% in its absence .F myristoyl moiety itself which is responsible for the aggre igure 2 shows Guinier plots of the scattering profiles for gation. both unmyristoylated (a) and myristoylated (b) recoverins . Unmyristoylated recoverin in the absence of Ca2+ shows The Guinier plot of unmyristoylated recoverin in the aggregation, which is clearly seen as an upward curvature of presence of Ca2+ is well-approximated by a straight line. In the Guinier plot. On the other hand, myristoylated recover- such a case, the X-ray scattering intensity is given as I (q) = in in the absence of Ca2+ gives a Guinier plot with a single I(0)exp(-Rg2Q2/3), where 1(0) and Rg are the intensity at straight line, the slope of which gives an Rg value of 21 A. zero angle and the radius of gyration, respectively. Q , The 1(0) value indicates that the protein is monomeric. momentum transfer is defined as 4ƒÎsinƒÆ/ƒÉ , where 28 and Again, the CD spectra of these two samples are identical, .i are the scattering angle and the wavelength of X-rays indicating that the myristoyl modification in this case (1.488 A), respectively (26). The molecular weight was prevents aggregation. estimated to be 26kDa from 1(0) with cytochrome c (MW The aggregation and the monomerization were revers 12.4kDa) as a standard, indicating that unmyristoylated ible. Myristoylated recoverin became aggregated when it recoverin with Ca2+ exists as a monomer in solution. The was dialyzed in EGTA buffer against calcium buffer, while slope of the Guinier plot indicates Rg to be 20 A, which is a it became monomerized when it was dialyzed in calcium normal value for a globular protein with a molecular weight buffer against EGTA buffer. Similar reversibility was of 23kDa. On the other hand, the Guinier plot of myris observed for unmyristoylated recoverin. toylated recoverin in the presence of Ca2+ shows a steep Figure 3 shows the distance distribution functions, P(r), upward curvature toward zero angle. This indicates that the of unmyristoylated recoverin with Ca2+ and myristoylated

protein in this state is highly aggregated. As shown above, recoverin without Ca2+. P(r) for the other states cannot be these two samples give identical CD spectra. The fluores calculated because of aggregation. The functions show only cence emission spectra of these two states are also essen- a single peak, indicating clearly that the recoverins in these states are globular in solution, contrary to the dumbbell shape of calmodulin and troponin C. A small increase in the maximum dimension is observed for myristoylated re coverin, which agrees with the change of Rg. Since the myristoyl moiety does not contribute to X-ray scattering significantly, we can regard the change in the scattering pattern as indicative of a change in the protein itself. Thus, we conclude that the release of calcium brings about slight expansion of recoverin. The increase in the size is accom panied by a loss of helical conformation, according to the CD measurements. The observed structural properties are summarized in Table I. A physiologically interesting problem is the effect of the binding of the target enzyme on the recoverin structure. Since the target enzyme should have a recognition site(s)

Fig. 3. P(r) functions for myristoylated recoverin in the absence of Ca2+ (i) and unmyristoylated recoverin in the Fig. 2. Guinier plots of SAXS curves of recoverin. (a) Unmyris toylated recoverin in the presence of Ca2+ (;) and in the presence of presence of Ca2+ (>). Both P(r) curves were calculated with the use of the extrapolated scattering curve to zero protein concentration. The EGTA The protein concentration was 6.7mg/ml for each curve. P(r) functions for myristoylated recoverin in the presence of Ca2+ and (b) Myristoylated recoverin in the presence of Ca2+ (G) and in the unmyristoylated recoverin in the absence of Ca2+ could not be presence of EGTA (O). The protein concentration was 5.5mg/ml for obtained because of the severe aggregation, as is clear in Fig. 2. each curve.

vi,I 114, No. 4, 1993 538 M. Kataoka et al. for recoverin as well as an active site(s), it is unlikely to be in the binding of the target enzyme with recoverin, as in the smaller than recoverin. In order to investigate the struc case of calmodulin. To examine this possibility, we ex tural change in recoverin, we need to use a target peptide amined the effect of melittin on recoverin. Melittin, a bee which mimics the recoverin binding site of the target venom peptide, is an amphiphylic peptide of 26 amino acid enzyme. In the case of calmodulin, one could use model residues, 5 of which are basic and the rest largely hydro- peptides such as melittin and mastoparan (17, 20), or a phobic (29). It binds to calmodulin in a calcium-dependent synthetic peptide of the calmodulin binding region of its manner (17). Figure 4 shows the effect of melittin on target enzyme (19, 21) to reveal a large conformational myristoylated recoverin. In the presence of Ca2+, the change in calmodulin. However, the direct target enzyme of scattering curve of recoverin in the presence of melittin is recoverin remains unknown, although a recent study sug identical with that in the absence of melittin. On the other gested that recoverin regulates termination of the trans hand, in the absence of Ca2+, the scattering curve changes duction cascade, such as at the rhodopsin phosphorylation drastically. As can be seen in Fig. 4, melittin induces a large step (8, 9). Further, we have no information on the extent of aggregation, suggesting that it interacts with recoverin binding region of its target enzyme. myristoylated recoverin in the absence of Ca2+. The aggre We expected that a hydrophobic interaction participated gation, however, suggests that the interaction is non- specific. Melittin showed little effect on unmyristoylated recoverin, although the extent of aggregation appeared to be increased by melittin in the absence of Ca2+ (data not shown). The effect of melittin is summarized in Table I.

DISCUSSION

We have reported here changes in recoverin induced by calcium binding and myristoylation, as measured by CD and SAXS. Myristoyl modification alone does not appear to alter the conformation of recoverin. However, this modi fication has a significant effect. Firstly, myristoylated recoverin becomes aggregated in the presence of Ca2+, while unmyristoylated recoverin does not, behaving as a monomeric globular protein. This indicates that the myris toyl moiety is responsible for the aggregation. Secondly, in the absence of Ca2+, unmyristoylated recoverin tends to aggregate. However, the myristoyl modification prevents this aggregation. Myristoylated recoverin without Ca2+ remains a monomeric globular protein. The effect appears to be aggregation, which is an indica tion of a non-specific interaction between proteins. In terms of SAXS, aggregation generally makes detailed analysis impossible. However, the fact that the tendency of recov erin to reversibly aggregate and disaggregate depends on both N-terminal myristoylation and Ca2+ binding indicates that both myristoylation and Ca2+ affect the surface prop erties of recoverin. Such changes in its surface properties would be closely related to the physiological function of recoverin. Both recoverin and S-modulin, a protein homologous to recoverin, are bound to disk membranes in the presence of Ca2+ , but only little is bound in the absence of Ca2+ (11, 13). Fig. 4. Effect of melittin on the scattering curve of myris It is suggested that the myristoyl group protrudes from the toylated recoverin. (a) In the presence of EGTA; (b) in the presence of Cap*. For both panels, (n) denotes the scattering curve with protein surface in the presence of Ca2+ and acts as an anchor melittin and (O) the curve without melittin. The concentration of to the membrane (11, 30). These observations reasonably recoverin for each curve was 5.5mg/ml. explain the aggregation of the myristoylated recoverin in

TABLE I. Structural properties of recoverin.

'The values could not be determined because of aggregation . bMelittin possibly stimulates aggregation in this state.

J. Biochem. Ca2+ Dependent Conformational Cha nge of Recoverin 539

phenyl-Sepharose, a hydrophobic support. Their model seems to be rather contradictory to the present observa tion, because the appearance of a hydrophobic surface is expected to cause the aggregation of unmyristoylated Ca2+-recoverin. However , recoverin in this state is globular and monomeric. Even if their model is correct, the model still does not explain why unmyristoylated Ca2+-free recoverin becomes aggregated. We assume that a hydro- phobic area is exposed on the surface of Ca2+ recoverin. The surface would facilitate the interaction between the myris toyl tail and the protein moiety, while it leads to aggrega tion of unmyristoylated recoverin. Further experiments are now underway to characterize the surface properties of recoverin, as well as the effect of a hydrophobic reagent. The interaction of recoverin with its target enzyme is a physiologically interesting problem. Since the physiological target enzyme of recoverin is unknown, we examined the effect of melittin on the solution structure of recoverin. In selecting melittin as a model peptide, we assumed that a hydrophobic interaction participated in the interaction between recoverin and its target enzyme, as in the case of calmodulin. Melittin has a large hydrophobic patch (29) and Fig. 5. Schematic illustration of the solution structure of is known to bind to calmodulin in a calcium-dependent recoverin under the conditions examined in the present study. manner (17). Melittin does not necessarily have a specific Upon the release of Ca2+, the a-helical content decreases and the effect on recoverin, but causes non-specific aggregation of global size increases. A myristoyl tail may protrude from the protein myristoylated recoverin in the absence of Ca2+. No effect of in the presence of Ca2+ (30) and may cause aggregation in this state . melittin was seen on the other states (Table I). At the time The tail would be enfolded with the protein moiety (11, 30). The reason why unmyristoylated recoverin becomes aggregated in the when the experiments were carried out, we believed that absence of Ca2+ is unknown. recoverin was a Ca2+-dependent activator of guanylate cyclase (2, 3), and that recoverin was an active or activating form in the absence of Ca2+, rather than in the Ca2+ bound the presence of Ca2+, i.e., tthe protruding hydrophobic tails state (2, 3). Therefore, we concluded that the effect was on interact with each other to bring about aggregation. the expected active form, and that a hydrophobic interac Upon the release of Ca2+, recoverin undergoes a confor tion was partly responsible for the interaction between mational change, which is characterized by a decrease in the recoverin and its target enzyme. a-helical content and an increase in global size. In a retinal However, our conclusion should be reconsidered, be cell, recoverin is transferred from the membrane to the cause, recently, a conclusion concerning the function and cytosol in accordance with a decrease in the Ca2+ concentra mechanism of recoverin as a Ca2+-dependent activator of tion (11, 13). Thus, the conformational change must be guanylate cyclase has been withdrawn by the authors (10). related to the translocation of recoverin. Myristoylated Recent observations suggest that recoverin plays an active recoverin becomes monomerized upon the release of Ca2+, role in termination of the transduction cascade, such as at while unmyristoylated recoverin becomes aggregated. the rhodopsin phosphorylation step, in the presence of Ca2+ Myristoylated and unmyristoylated recoverins show the (8, 9). Since the action mechanism of recoverin remains same CD in the absence of Ca2+, suggesting that the protein unknown, the effect of melittin is difficult to explain. moiety is similar for both states. Thus, a physiological role Nevertheless, the behavior of melittin maybe indicative of of the myristoyl tail in the absence of Ca2+ would be the properties which the true target enzyme of recoverin prevention of aggregation. Dizhoor et al. demonstrated that possesses. the N-terminal of unmyristoylated, Ca2+-free recoverin is Calmodulin, a multifunctional protein, interacts with susceptible to trypsin digestion, while myristoylated, Ca2+- various target enzymes and plays a regulatory role in free recoverin is resistant to trypsin (11). Fluorescence various cellular functions (31). Therefore, the binding to emission spectra of recoverin suggest that myristoylation target peptides is not sequence specific. The target peptides makes the environment of one or more tryptophans less of calmodulin generally are basic amphiphylic helices (32), and melittin satisfies this criterion. On the other hand, polar (15). These observations strongly suggest the inter- action between the myristoyl tail and the protein moiety in recoverin affects the termination of the visual transduction the absence of Ca2+. The myristoyl tail would be enfolded cascade (8, 9). Although proteins homologous to recoverin with the protein moiety to prevent aggregation in the have been found in various cells (2, 7, 33, 34), their absence of Ca2+. The present observations are illustrated functions are expected to be closely related. In fact, it was schematically in Fig. 5. proved that S-modulin and recoverin are identical proteins One point remaining unexplained is why unmyristoylat structurally and functionally (35). It is therefore likely that ed Ca2+-free recoverin becomes aggregated. Zozulya and the target enzyme of recoverin will be specific and more limited than calmodulin. We need to identify the true Stryer (30) suggested that a hydrophobic patch is exposed on the surface of both myristoylated and unmyristoylated target enzyme and its recoverin binding region. A synthetic recoverins in the presence of Ca2+, based on the binding to peptide with the sequence of the binding region will be

t 4, No. 4, 1993 540 M. Kataoka et al. valuable for understanding the structural change in recov Dizhoor, A.M., McKay, D.B., Hurley, J., & Stryer, L. (1992) erin on binding to the target enzyme, especially for under- Proc. Natl. Acad. Sci. USA 89, 5705-5709 16. Seaton, B.A., Head, J.F., Engelman, D.M., & Richards, F.M. standing the role of myristoyl modification in the interac tion. (1985) Biochemistry 24, 6740-6743 17. Kataoka, M., Head, J. F., Seaton, B. A., & Engelman, D. M. (1989) Proc. Natl. Acad. Sci. USA 86, 6944-6948 The authors express their sincere thanks to Professor Lubert Stryer 18. Kataoka, M., Head, J.F., Persechini, A., Kretsinger, R.H., & and Dr. Sergay Zozulya for the gifts of recoverin as well as for the Engelman, D.M. (1991) Biochemistry 30, 1188-1192 discussion. They also express their gratitude to Professor James Head 19. Kataoka, M., Head, J.F., Vorherr, T., Krebs, J., & Carafoli, E. (Boston University School of Medicine) for his critical reading of the (1991) Biochemistry 30, 6247-6251 manuscript. SAXS measurements were performed with the approval 20. Matsushima, N., Izumi, Y., Matsuo, T., Yoshino, Y., Ueki, T., & of the Program Advisory Committee of the Photon Factory (Proposal Miyake, Y. (1989) J. Biochem. 105, 883-887 No. 92-066). 21. Heidorn, D.B., Seeger, P.A., Rokop, S.E., Blumenthal, D.K., Means, A.R., Crespi, H., &Trewhella, J. (1989) Biochemistry 28, REFERENCES 6757-6764 22. Ikura, M., Clore, G.M., Gronenborn, A.M., Zhu, G., Klee, C.B., 1. Moncrief, N.D., Kretsinger, R.H., & Goodman, M. (1990) J. Mol. & Bax, A. (1992) Science 256, 632-638 Evol. 30, 522-562 23. Meador, W.E., Means, A.R., & Quiocho, F.A. (1992) Science257, 2. Dizhoor, A.M., Ray, S., Kumar, S., Niemi, G., Spencer, M., 1251-1255 Brolley, D., Walsh, K.A., Phillipov, P.P., Hurley, J.B., & Stryer, 24. Goto, Y. & Hagihara, Y. (1992) Biochemistry 31, 732-738 L. (1991) Science 251, 915-918 25. Ueki, T., Hiragi, Y., Kataoka, M., Inoko, Y., Amemiya, Y., 3. Lambrecht, H.G. & Koch, K.W. (1991) EMBO J. 10, 793-798 Izumi, Y., Tagawa, H., & Muroga, Y. (1985) Biophys. Chem. 23, 4. Stryer, L. (1991) J. Biol. Chem. 266, 10711-10714 115-124 5. Liebman, P.A., Parker, K.R., & Dratz, E.A. (1987) Anna. Rev. 26. Guinier, A. & Fournet, G. (1955) Small-Angle Scattering of Physiol. 49, 765-791 X-Rays, J. Wiley, New York 6. Hurley, J.B. (1987) Annu. Rev. Physiol. 49, 793-812 27. Moore, P.B. (1980) J. Appl. Crystallogr. 13, 168-175 7. Kawamura, S. & Murakami, M. (1991) Nature 349, 420-423 28. Chen, Y.-H., Yang, J.T., & Martinez, H.M. (1972) Biochemistry 8. Gray-Keller, M.P., Polans, A.S., Palczewski, K., & Detwiler, 11,4120-4131 P.B. (1993) Neuron 10, 523-531 29. Habermann, H. (1972) Science 177, 314-322 9. Kawamura, S. (1993) Nature 362, 855-857 30. Zozulya, S. & Stryer, L. (1992) Proc. Natl. Acad. Sci. USA 89, 10. Hurley, J.B., Dizhoor, A.M., Ray, S., & Stryer, L. (1993) Science 11569-11573 260,740 31. Klee, C.B. & Vanaman, T.C. (1982) Adv. Protein Chem. 35,213- 11. Dizhoor, A.M., Chen, C.-K., Olshevskaya, E., Sinelnikova, V.V., 321 Phillipov, P., & Hurley, J.B. (1993) Science 259, 829-832 32. Blumenthal, D.K., Takio, K., Edelman, A.M., Charbonneau, H., 12. Dizhoor, A.M., Ericsson, L.H., Johnson, R.S., Kumar, S., Titani, K., Walsh, K.A., & Krebs, E.G. (1985) Proc. Natl. Acad. Olshevskaya, E., Zozulya, S., Neubert, T.A., Stryer, L., Hurley, Sci. USA 82, 3187-3191 J.B., & Walsh, K.A. (1992) J. Biol. Chem. 267, 16033-16036 33. Yamagata, K., Goto, K., Kuo, C.H., Kondo, H., & Miki, N. 13. Kawamura, S., Takamatsu, K., & Kitamura, K. (1992) Biochem. (1990) Neuron 4, 469-476 Biophys. Res. Commun. 186, 411-417 34. Bunt-Milam, A.H., Dacey, D., & Dizhoor, A.M. (1991) Invest. 14. Kokame, K., Fukada, Y., Yoshizawa, T., Takao, T., & Shimo Ophthalmol. Visual Sci. 32, 1264 nishi, Y. (1992) Nature 359, 749-752 35. Kawamura, S., Hisatomi, 0., Kayada, S., Tokunaga, F., & Kuo, 15. Ray, S., Zozulya, S., Niemi, G.A., Flaherty, K.M., Brolley, D., C.-H. (1993) J. Biol. Chem. 268,14579-14582

J. Biochem.