Eur. J . Biochem. 204, 1169-1181 (1992) 0FEBS 1992

Subunit interactions of GTP-binding proteins Helmuth HEITHIEK', Margareta FROHLICH Christian DEES', Martin BAUMANN', Martin HARING ', Peter GIERSCHIK 2 , Emile SCHILTZ3,Winchil L. C. VAZ4, Mirko HEKMAN' and Ernst J . M. HELMREICI-T' Department of Physiological Chemistry, University of Wiirzburg Medical School, Federal Republic of Germany Department of Pharmacology, University of Heidelberg Medical School, Federal Republic of Germany lnstitutc for Biochemistry and Organic Chemistry, University of Freiburg, Federal Republic of Germany Max Planck Institute for Biophysical Chemistry, Karl Friedrich Bonhoeffer Institute, Molecular Biology Department, Gottingen, Federal Republic of Germany (Received September 16/November 15,1991) - EJB 91 1228

Fluorescence energy transfer [cf. Forster, T. (1948) Ann. Phys. 6, 55-75] was tested for its suitability to study quantitative interactions of subunits of G o with each other and these subunits or trimeric Go with the P,-adrenoceptor in detergent micelles or after reconstitution into lipid vesicles [according to Feder, D., Im, M.-J., Klein, H. W., Hekman, M., Holzhofer, A, Dees, C., Levitzki, A., Helmreich, E. J. M. & Pfeuffer, T. (1986) EMBO J . 5, 1509-15141. For this purpose, ao- and Bysubunits and trimeric Go purified from bovine brain, the fly-subunits from bovine rod outer segment membranes and the PI-adrenoceptor from the turkey erythrocyte were all labelled with either tetramethylrhodaminmaleimide or fluorescein isothiocyanate under conditions which leave the labelled proteins functionally intact. In the case of ao- and By-interactions, specific high-affinity binding interactions (& z 10 nM) and nonspecific low-affinity binding interactions (Kd z 1 pM) could be readily distinguished by comparing fluorescence energy transfer before and after dissociation with 10 pM guanosine 5'-O-[y-thio]triphosphate and 10 mM MgClz where only low-affinity binding interactions remained. Interactions between ao- and by-subunits from bovine brain or from bovine retina1 transducin did not differ much. The By-subunits from bovine brain were found to bind with high transfer efficiency and comparable affinities to the hormone-activated and the nonactivated plreceptor reconstituted in lipid vesicles: Kd = 100 f 20 and 120 & 20 nM, respectively; however, Bysubunits from transducin appeared to bind more weakly to the P1-adrenoceptor than By-subunits from bovine brain. Separated purified homologous ao- and @subunits from bovine brain interfered mutually with each other in binding to the P1-adrenoceptor presumably because they had a greater affinity for each other than for the receptor. These findings attest to the suitability of fluorescence energy transfer for studying protein -protein interactions of G-proteins and G-protein-linked receptors. Moreover, they supported the previous finding [Kurstjens, N. P., Frohlich, M., Dees, C., Cantrill, R. C., Hekman, M. & Helmreich, E. J. M. (1991) Eur. J. Biochem. 197, 167- 1761 that /$subunits can bind to the nonactivated B1-adrenoceptor.

functions (cf. Birnbaumer et al., 1990). However, quantitative information on interactions of subunits with each other and with receptors is scarce. Until now, such interactions have been analyzed, with few exceptions (see Borochov-Neori and Montal, 1989), on the basis of changes in hydrodynamic parameters determined by sucrose gradient centrifugation or biotinylation (Codina et al., 1984; Kohnken and Hildebrandt, 1989). Association -dissociation of G-protein subunits were Correspondence ta E. J. M. Helmreich, Physiologisch-Chemisches also inferred indirectly from binding to matrix-bound subunits Institut der Universitit, Koellikcrstrasse 2, W-8700 Wiirzburg, Feder(Linder et al., 1991). Binding of guanosine 5'-0-[y-thiolal Republic of Germany Abbreviations. G-protcins,GTP-binding proteins; G,, stimulatory triphosphate (GTP[S]) and hydrolysis of GTP in the presence trirneric GTP-binding protein, cornposed of a, and y-subunits; Go, of Mg2+was also studied using the intrinsic tryptophan fluobrain-derived trimcric GTP-binding protein, composed of a, and yrescence of the a-subunits of G I and G o as signal (Higashijima subunits; rs.mo and or, denote the m subunits of different GTP-binding et al., 1987a). proteins; GTP[S], guanosinc 5'-0-[y-thioltriphosphate;CGP-12177, Previously, we have shown that isolated bovine brain piCiba Geigy product 12177. a fl-adrencrgic antagonist; 4-[3-(t-busubunits can associate with the p,-adrenoceptor purified from tylamino)-2-hydroxypropoxy]benz-imidazol-2-one (see also Heithier turkey erythrocyte membranes (Im et al., 1988; Kurstjens et et al., 1988); FITC, fluorescein isothiocyanate; TRM, tetramethylal., 1991). The j-adrenoceptor seems not to be the only horrhodamine maleimide. mone receptor which can couple to By-subunits. This was also Enzyme. Adenylatc cyclase (EC 4.6.1.1).

GTP-binding proteins (G-proteins), which function in signal transmission pathways, are heterotrimers composed of three different subunits (a, /3 and y) which are coded by different genes. All three subunits are heterogeneous, but the functional consequences of the considerable structural diversity are not yet clear (cf. Soege et al., 1991). Furthermore, a- and By-subunits have each been implicated in separate and distinct

1170 shown to be the case with somatostatin receptors from rat brain (cf. Law et al., 1991). This tempted us to speculate that pli-subunits might play a role in the selection of a-subunits for interaction with the receptor because the byreceptor complex was found to bind uo-subunits which are not likely to be physiological partners for the /I-adrenoceptor from turkey erythrocytes. However, it was reported recently by Rooney et al. (1991) that the turkey b-adrenoceptor is capable of binding to an unknown G-protein which is involved in the fl-receptormediated activation of phospholipase C in turkey erythrocytes and which is different from G, which is a cholera toxin substrate and participates in adenylate cyclase activation. A major aim of the present study was therefore to obtain quantitative data on the affinity of a0- and by-subunits to each other in the absence and presence of GTP[S] and Mg2+ and on the affinities of these subunits to the jl-adrenoceptor in the activated and nonactivated state. Fluorescence resonance energy transfer was chosen as method to follow these interactions in detergent micelles and in lipid vesicles.

MATERIALS AND METHODS Materials The materials used and their sources are given in the preceding paper (Kurstjens et al., 1991). Tetramethylrhodamine maleimide (TRM) and fluorescein isothiocyanate (FITC) were from Molecular Probes (Eugene, Oregon, USA). The basic buffer solutions used were buffer A (20 mM Hepes, 20 mM NaCl, 0.1 mM EDTA-Na2, pH 8.0) and buffer B (20 mM Tris/HCl, 1 mM EDTA-Na2, 1 mM dithiothreitol, pH 8.0). When necessary, these buffer solutions were supplemented with detergents and other additions as stated. All other chemicals and biochemicals were of the highest grade commercially available and were from the same sources as described before (Hekman et al., 1987: Kurstjens et al., 1991).

Preparations All preparations were carried out at 4 "C.

G,-protein G, was prepared from turkey erythrocyte membranes according to Hanski et al. (1981). The purity of the preparation is shown in Fig. 2 of Hekman et al. (1987). r o - arid

py-subunits

zo- and By-subunits were prepared by the procedure of Sternweis and Robishaw (1 984) from bovine brain membranes prepared according to Pfeuffer et al. (1985). The preparations are described in detail in the preceding publication where the purity of the proteins is also documented (Fig. 1 of Kurstjens et al., 1991). The cto-fractions were concentrated and equilibrated with buffer A containing 0.1 mM dithiothreitol and 0.1 YO Lubrol PX because, in contrast to the By-subunit, the r-subunit was more stable in Lubrol-PX than in cholate. The concentrated Ppsubunit (15 mg/ml) and ao-subunit preparations (10- 15 mg/ml) were stored in 20% glycerol at - 70 "C. Transducin z- and By-subunits were separated from heterotrimeric G, by chromatography on blue Sepharose CL6B (Pharmacia LKB). G,, the starting material, was purified from eluates of bovine rod outer segment membranes with hypotonic buffer containing 100 pM GTP (Gierschik et al.,

Fig. 1. SDS/PAGE of fluorescent bovine brain Go, separated ao- and By-subunits from bovine brain and transducin &subunits. 12% polyacrylamide gels were used according to Laemmli (1970). Proteins in lanes 1 - 5 were stained with Coomassie blue and proteins in lanes 6-10 were visualized by ultraviolet light. Lanes 1 and 6, FITClabelled cco-subunit; lanes 2 and 7, TRM-labelled q,-subunit; lanes 3 and 8, TRM-labelled trimeric Go; lanes 4 and 9, TRM-labelled bovine brain py-subunits and lanes 5 and 10, TRM-labelled transducin pi-

subunits.

1984). Fractions containing a-or by-subunits were pooled and concentrated about 10-fold by pressure filtration using a cell where the content could be stirred and which was equipped with an Amicon PM 10 membrane. The purity of the a- and By-transducin preparations was greater than 95% as judged by SDS/PAGE (see Fig. 1 of Grunwald et al., 1986).

Reaction of Goo,with FITC Purified a. (500 pg) was reacted with 1 mM FITC after fluoride activation in buffer A containing 0.1 mM dithiothreitol, 5 mM NaF, 50 pM AlC13, 10 mM MgC12 and 0.4% Lubrol PX, pH 8.5. For some experiments (see Fig. 5), aoFITC was purified once more by affinity chromatography on a jy-agarose column following the procedure of Pang and Sternweis (1989). The column-bound ao-FITC fraction was eluted with 5 pM GDP, 30 pM AIC13, 50 mM MgC12, 10 mM NaF in buffer A, pH 8.0, containing 0.1% Lubrol PX. Whereas the starting material had a molar ratio of fluorophore to a. of 1.5: 1 based on protein concentration and measurements of absorbance at 490 nm after precipitation with acetone (loo%, by vol.), the material specifically eluted contained only 0.8 mol FITC/mol ao. Moreover, the capacity of the eluted material to bind GTP[S] increased and reached > 80% of the expected maximal binding capacity.

Reaction of do- and fly-subunits of Go with TRM

For labelling of ao- or By-subunits with TRM, 150 nmol of the dye in N,N-dimethylformamide (15 pl) were added to 5 - 10 nmol a. or by in 285 p120 mM Tris/HCl, 10 mM NaCl and 0.1% sodium cholate, pH 7.8. In order to minimize denaturation, the ao-subunit was activated first with AlF; and the samples were kept on ice and in the dark for 2 h which was a sufficiently long time to bring the reaction to completion. The reaction proceeded optimally with a 5 - 30fold molar excess of dye over protein. The unreacted dye was removed nearly completely by hydroxyapatite chromatography. The [35S]GTP[S]binding activity of the labelled a0 was decreased only slightly when compared with the non-labelled starting material. The molar ratio of TRM to the ao-subunit was 1 : 1. The purity of the preparations is documented in Fig. I . Whether the heterogeneity in the case of TRM-labelled P-subunits (see lanes 4, 9, 5 and 10 in Fig. 1) represents a

1171 mixture of p-isoforms (see Discussion) or is due to a mixture of differently labelled p-subunits, is not clear. In order to determine the distribution of TRM between pand y-subunits, they were separated into their components by preparative SDS/PAGE. The protein bands were cut out and eluted, the absorbance of the eluted p- and y-subunits was measured and the protein concentration determined (Schaffner and Weissmann, 1973). The molar ratio of dye/ protein ranged for bovine brain and transducin P-preparation over 2.0-2.3 for the p- and 0.7- 1.0 for the y-subunit. The spectrum of Pl-TRM exhibits maxima at 550nm and at 520 nm. The latter absorbance, however, was shown not to be a property of the free or unbound dye or a consequence of decomposition. The fact that it nearly completely disappeared on addition of SDS (or detergents) suggested that the 520nm absorbance was due to a solvent (detergent) effect. The contribution of free dye to the 550-nm absorbance was determined in the following way. The By-TRM preparation was precipitated with acetone and the precipitate and supernatant were separated by centrifugation, dried and taken up separately in 300 p1 buffer A containing 0.5 mM dithiothreitol and 0.2% sodium cholate. Based on the absorbance of the supernatant, the contamination of fly-TRM by free dye was estimated to be less than 5%. Labelling of fly was also carried out with tetramethylrhodamine 5(6)-iodoacetamide, which has similar spectral properties as tetramethylrhodamine maleimide, but the latter was preferred because it proved to be more uniformly reactive.

acid containing 0.5 M cyanogen bromide. The solution was diluted with 150 pl water and incubated at 20°C for 12 h. The fragments were separated by SDSjPAGE according to Schlgger and von Jagow (1987) and electrophoretically transfererd to polyvinyldifluoride membranes (Immobilon IPVH 20200, Millipore) (Western blotting). The fluorescent protein bands were localized by ultraviolet light, cut out and stored at - 70 "C. The blotted peptides were sequenced in an Applied Biosystem 477 sequencer with on-line identification of the derivatives. The electrophoretic properties of marker proteins labelled with tetramethylrhodamine 5(6)-isothiocyanate were unchanged. Functional tests of the fluorescent protein derivatives

The function of modified py-subunits was tested based on their effects on deactivation of AIF, -activated G, according to Northup et al. (1983). The activity of G, was followed by adenylate cyclase activation for 20 min at 30°C (Hekman et al., 1984) using a crude rabbit heart muscle preparation as source of adenylate cyclase (Pfeuffer and Metzger, 1982). CAMPwas isolated and determined according to Salomon et al. (1974). In addition, the protection of a. by &subunits against heat denaturation was tested, whereby the activity of the cco-subunitswas followed by GTP[S]-binding usually with 1 pM [35S]GTP[S] (20 Ci/mmol and 25 mM MgCI2) (cf. Sternweis and Robishaw, 1984). Unspecific binding was determined in the presence of 1 mM GTP. Pertussis-toxin-catalyzed ADP-ribosylation of ao-subunits, which is dependent on flysubunits (Neer et al., 1984), provided another test. Reaction of the fll-adrenoceptor with FITC and purification FITC-labelled /?,-adrenoceptor preparations were tested of the labelled receptor in detergent solutions and after reconstitution in lipid vesicles Turkey erythrocyte plasma membranes were prepared as by binding of antagonists, such as [3H]dihydroalprenolo1, described by Puchwein et al. (1974). The ,&receptor was react- [3H]CGP12177 and ['251]iodocyanopindolol, and by compeed with FITC in the membrane-bound state. The unreacted tition for binding between the agonist I( - )-isoproterenol and FITC was removed effectively first through exhaustive washes [3H]dihydroalprenolo1. In addition, coupling efficiencies of of the membranes and subsequently through washes of the FITC-labelled and unlabelled P1-receptor with purified G, solubilized receptor attached to the affinity column (cf. were compared in reconstituted phospholipid vesicles (cf. Hekman et al., 1984). The molar ratio of fluorophore to the Feder et al., 1986). The NaCl concentration was raised to PI-adrenoceptor was estimated to be 1 : 1. The purity of the 150 mM, and hormone-dependent activation was followed on affinity-chromatography-purified P,-adrenoceptor prepa- additon of 10 pM I( -)-isoproterenol, 100 nM GTP[S] and ration is documented (see Fig. 1 in Kurstjens et al., 1991). 0.5 mM MgClz at 30°C (cf. Hekman et al., 1987). The actiThe purified PI-adrenoceptor preparation was an equimolar vated G, was determined by activation of adenylate cyclase as mixture of the 50-kDa and 40-kDa receptor forms (cf. Jurss described above. Finally, subunit association -dissociation of TRMet al., 1985; Boege et al., 1987). labelled py-subunits and FITC-labelled ao-subunits in Lubrol PX was checked in isopycnic sucrose density gradients accordLocalization of the fluorophores ing to Codina et al. (1984) and as described in the legend to Fig. 6 . Trypsin cleuvuge The ,+subunits tagged with TRM were cleaved with trypsin in the presence and absence of SDS (0.05%). The reaction was stopped by addition of 0.1Oh (by vol.) trifluoroacetic acid and the reaction mixture was stored at - 70" C. Cyunogen bromide cleuvuge

Reconstitution of y-, fly-subunits and fl,-adrenoceptor in lipid vesicles

Reconstitution was achieved by addition of purified a-and py-subunits (micromolar concentrations) and the p-receptor (nanomolar concentrations) to lipid mixtures in buffer A containing 0.1 mM dithiothreitol and 0.2% sodium cholate or 0.01% lauroylsucrose (see Feder et al., 1986). The mixture was incubated for 5 min at 20°C and the detergent was removed with Extractigel D (Pierce).

Labelled purified P,-adrenoceptor, uo and separated pand y-subunits and a 22-amino-acid-long tryptic fragment of the P-subunit were subjected to cyanogen bromide cleavage. The /Iand g-subunits were separated by SDS/PAGE and recovered by clectroelution. The polypeptides were incubated first in buffer A with 0.5 mM dithiothreitol, 0.2% sodium Fluorescence resonance energy transfer cholate and 1 % (by vol.) 2-mercaptoethanol at 65OC for Fluorescence resonance energy transfer between FITC15 min. The proteins were then precipitated with acetone and the precipitate was taken up in 3 5 0 ~ 1concentrated formic . and TRM-labelled components- was measured using a

1172 Schoeffel RRS-1000 fluorescence spectrometer. The samples were in temperature-controlled microcuvettes and measurements were started after temperature equilibration. Incubations were for 10-20 min at 20°C. FITC-labelled probes were excited at 468 nm. For measurements with ao-FITC and P;-TRM in buffer A, pH 7.6, containing 0.25% Lubrol PX, the same sample without by-TRM served as reference. Fluorescence of detergent- and lipid-buffer solutions was negligible. The residual concentration of free dye was up to about 10% of the xo-bound dye. This variable contribution of free dye fluorescence to the nonspecific low-affinity transfer to the acceptor was not corrected for. The contribution of by-TRM to the fluorescein donor fluorescence at 520 nm was also determined and subtracted. It varied with the concentration of piTRM, but did not exceed 10% at the highest concentrations. Experiments with p-adrenoceptors-FITC were carried out in buffer A containing 0.02% lauroylsucrose, otherwise the conditions were the same as for the experiments with ao-FITC and Bp-TRM without receptor. Energy transfer measurements were also carried out in lipid vesicles in buffer A, pH 8.0 (cf. Feder et al., 1986; Hekman et al., 1987). FITC-labelled proteins have an emission maximum at 520 nm which allows for efficient fluorescence resonance energy transfer to TRM-labelled a- or &-subunits which absorb maximally at 550 nm. The slit widths of the monochromators used made possible a large enough excitation at sufficiently narrow band width giving good resolution and high yield. The molar absorption coefficients of the fluorophoreswere determined in 0.1 M NaOH with excitation at 468 nm. Maximal values were, in the case of FITC, &490 = 8 x lo4 M-' cm-' and for TRM E~~~ = 6 x lo4 M-' cm-'. Quantum yields were determined according to Gennis et al. (1972). For that purpose, absorption spectra were recorded in the same solutions which were used for the actual measurements. Fluorescein in 0.1 M NaOH which has a quantum yield 4 = 0.92 (Weber and Teale, 1957) served as standard. Overlap integrals were calculated according to the equation: J

1

450 470

L90 510 530

550 570 590 610 630 650

Wavelength (nml Fig. 2. Calculated overlap integral for the pair ao-FITC//ly-TRM. The measurements were carried out in bufler A, pH 7.6, with 0.25% Lubrol PX; fluorescence emission of fluorescein is maximal at pH 8.5; see Materials and Methods. No scales are given for absorption and

fluorescence spectra.

transferred fluorescence energy (see legend to Fig. 5). Curves were fitted by computer using a least-squares fitting program (Enzfitter by R. J. Leatherbarrow) and the Kd values were derived from Scatchard plots. Other methods Protein was determined with the method of Bradford (1976) or according to Schaffner and Weissmann (1973) modified by Kaplan and Pedersen (1985). SDSjPAGE was carried out by the procedure of Laemmli (1970). Silver-staining of proteins was as described by Oakley et al. (1980).

IF(>.) x c ( i ) x l?d;l

RESULTS J F ( L ) x dA

Labelling of ao- and by-subunits and of the j1-adrenoceptor

0

The purity of FITC-labelled and unlabelled a,)- and Pjsubunits and of the P1-receptor is shown in Fig. 1 in the preceding paper of Kurstjens et al. (1991) and of FITC- and TRM-labelled ao- and of by-subunits and trimeric Go labelled with TRM in Fig. 1 of this paper. Fluorescent ao, separate /Iand y-subunits, the By-complex and the P,-adrenoceptor were cleaved with trypsin and cyanogen bromide. After tryptic cleavage of TRM-labelled by-complex under non-denaturing conditions, only two fragments were recovered, one with M , 22000 (p22),which was labelled with TRM, and a fragment with M , 13000 which was not labelled (Fig. 3A, lanes 2 and 4). Two trypsin fragments were also found by Pines et al. (1985) and it was shown that the larger fragment contains the C-terminal sequence (cf. Zaremba et al., 1988). The TRMreactive cysteines were localized in the C-terminal fragment consisting of amino acids 130 - 340. The cyanogen bromide cleavage pattern of the P-subunit is given in Fig. 3B. Three strongly and two weakly fluorescent bands with M , between 3000 and 17000 are apparent. The fragments CB-1 ( M , 3000), CB-2 ( M , 4000), CB-3 ( M , 7000)

x i4 were derived from the fluoThe values for F ( i ) x ~(i) rescence emission spectrum of a,-FITC and the absorption spectrum of Py-TRM. Wavelengths (2) were measured in cm. The integral in the denominator served for normalisation, since the data were expressed in arbitrary scale units (see Fig. 2). The overlap integral for the pair ao-FITC and p1,TRM was J = 6.1 x 10- l o x cm' x mol-'. Fluorescence energy transfer decreases with increasing distance betwccn donor and acceptor by a factor of lo6. The critical transfer radius, Ro. where the efficiency of energy transfer is 50"/0, was calculated according to Forster (1948): for the donor/acceptor pair Eo-FITC and by-TRM, Ro was estimated to be 4.9 nm. A practical limit for fluorescence encrgy transfer is set by a distance between donor and acceptor of 6.5 nm (Stryer, 1968). The efficiency of fluorescence energy transfer ( E ) was quantitated as fractional decrease of the donor fluorescence at 520 nm due to binding increasing concentrations of the TRM-labelled partner and was expressed as percentage of the

1173

(Gautam et al., 1989, 1990; Robishaw et al., 1989). Only the C-terminal fragments of the three cyanogen bromide cleavage products of the isolated y-subunit from bovine brain were expected to contain the cysteine reactive with TRM. But sequencing of the cyanogen bromide fragments of separate ysubunits purified to homogeneity indicated additional heterogeneity of bovine brain y-subunits suggesting the existence of another y-isoform. This evidence will be reported soon (Helmreich, E. J. M., Lohse, M. J., unpublished results). We have not determined the position of TRM-labelled cysteine in the fl- and y-subunits of transducin. However, from the known sequences of transducin PI (Amatruda et al., 1988),it is apparent that it is identical with the bovine brain p,-subunit. Moreover, the COOH-terminal regions where the cysteines reactive with TRM are located are rather similar in transducin yl (Gautam et al., 1990) and bovine brain y s and y6-subunits (Gautam et al., 1989; Robishaw et al., 1989). It may therefore be assumed that distribution of TRM in fly-transducin subunits was not much different from that in the bovine brain Pysubunits. In the case of the FITC-labelled ao-subunit, all five theoretically predicted fluorescent fragments were obtained (Fig. 3B). Thus the lysines which had reacted with FITC were distributed randomly. The same seems to be the case with the FITC-labelled /j’,-adrenoceptor where the fluorophore was also distributed randomly (see Fig. 3 B).

Functional tests of labelled ao-and fly-subunits and of the labelled jl-adrenoceptor Fig. 3. Localisation of fluorophores in bovine brain do- and fly-subunits and in the pl-adrenoceptor. (A) SDSjPAGE of trypsin-treated TRMlabelled By-subunits. 11 O h polyacrylamide gel containing 0.2% SDS was used. Lanes 1 and 2 were stained with Coomassie blue, lanes 3 and 4 were visualized by ultraviolet light. Lanes 1 and 3, Pj-TRM; lanes 2 and 4,trypsin-treated by-TRM. (B) SDSjPAGE of cyanogenbromide-treated fll-receptor, a,,-, fl- and y-subunits. 16.5% polyacrylamide gel was used (Schagger and von Jagow, 1987). The subunits were treated with cyanogen bromide as described and approximately 0.5 nmol cleaved peptide was applied to SDS/PAGE. Lanes 1-4 were stained with Coomasie bluc and lanes 5-8 are fluorescence tracings with ultraviolet light. Lanes 1 and 5, cyanogenbromide-cleaved FITC-labelled b-receptor ; lanes 2 and 6 , cyanogenbromide-cleaved ct,-FITC; lanes 3 and 7, cyanogen-bromide-cleaved TRM-P-subunit ; lanes 4 and 8, cyanogcn-bromide-cleaved TRM-ysubunit .

and CB-4 ( M , 10000) were sequenced. CB-1 contains mainly the sequence Arg46 - Met61 and about 10% of a sequence starting with Ser189. TRM was assigned to Cys204 in the smaller CB-1 fragment with Ser189 -217. Both CB-2 and CB3 contain sequences starting with Tyr264 and Arg46. The sequence starting with Tyr264 could be assigned to the f12subunit. The blot of CB-4 contained the sequence 218-300 and the TRM could be assigned to Cys271 and Cys294. Three forms of fl-subunits encoded by distinct genes have been described (Gao et al., 1987; Levine et al., 1990), but this fragment could not have originated from f13 which also contains Cys in positions 242 and 263 because B3 does not have a Met at 217. From comparison of the sequences, it became apparent that the TRM-labelled fragments were derived from both PI- and P,-subunits. In bovine brain at least two y-isoforms, y 5 and y 6 , are present, but a complete sequence is available only for the latter

From Fig. 4A, it is apparent that modified and unmodified bovine brain and rod outer segment transducin By-subunits cannot be distinguished on the basis of their capacity to protect mo against heat denaturation. Purified fly-preparations showed consistently a small amount of GTP[S] binding which amounted to 3-5% of the total bound GTP[S] in the c$y mixture. Whether this is due to a contamination with aosubunits is not certain. The data in Fig. 4A were corrected for this amount of GTP[S] binding carried over from the Bysubunits. In Fig. 4B is shown that native and TRM-labelled fly-subunits were equally effective in deactivation of AlF, activated G,; Fig. 4C indicates that labelled &subunits were even somewhat more effective than their native counterparts in stimulation of pertussis-toxin-catalysed ADP-ribosylation of Go a-subunits. The binding properties of purified and FITC-labelled adrenoceptor preparations for several antagonists alone and in competition with the agonist I( -)-isoproterenol were found to be indistinguishable from that of the unlabelled receptor (not shown). Moreover, stimulation of GTPase activity of G, by B-adrenoceptor-FITC activated with /( - )-isoproterenol after reconstitution in lipid vesicles was the same as that of the native nonlabelled receptor: the pseudo-first-order rate constant, k , was 0.16 min-’ in both cases (see also Feder et al., 1986).

Fluorescence resonance energy transfer as a probe to examine a,- and fly-interactions Data in Fig. 5 are representative examples of fluorescence energy transfer measurements carried out with bovine brain Go a-FITC and by-subunits tagged with TRM. In Fig. 5 A are data without and with 50 pM GTP[S] and 10 mM MgC12. The experiments were carried out in detergent solution with 0.2%

1174 h

6-?

I

1

I

A

\

A 0 a,-FITC+ a. -FITC+ a* + BY

B7-TRM BY

0 a,-FITC+ 0 a,-FITC

&-TRM

9 0

a0

I

, 1

I

2

3

5

4

6

0

TIME (MIN)

1

7

10

L

Molar Ratio

100

-

80

-

20

OylGS

A

0

I 800

fly-subunits ( n g / h b e )

Fig. 4. Functional integrity of modified subunits. (A) Protection against heat denaturation of ao-subunits by py-subunits. Modified and unmodified a,-subunits (7.5 pmol) were incubated at 30°C in 60 p1 buffer A, pH 8.0,l mM dithiothreitol and 0.2% Lubrol PX with or without @subunits (15.0 pmol). At the times indicated, 10-pl aliquots were withdrawn and [35S]GTP[S]binding was measured and plotted as a percentage of the initial value (100% at zero time). (B) Deactivation of AIF; -activated G,-protein by By-subunits. Purified AIF; -activated turkey erythrocyte G, (6.3 pmol) was mixed with increasing amounts (0- 135 pmol) of either native or TRM-modified pi-subunits in 30 p1 buffer A, 0.1 m M dithiothreitol and 0.2% sodium cholate pH 7.8. The mixtures wcre incubated for 5 min at 4°C. Subsequently, 10-pl aliquots were removed and G, activity was determined. (C) Promotion of pertussis-toxin-catalyzed ADP-ribosylation of a0 by fly-subunits. A solution of zo-subunits (7.5 pmol) in 12.5 p1 buffer A, 0.1 mM dithiothreitol and 0.05% Lubrol PX was supplemented with increasing amounts (018 pmol) of either unmodified or TRM-labelled fly-subunits. ADP-ribosylation was initiated by addition of 37.5 pI of a solution containing 1 pM [32P]NAD(1 pCi) and 6.4 pmol preactivated pertussis toxin in 100 mM Tris/HCI, 2.7 mM ATP and 32 mM dithiothreitol, pH 8.0. ADP-ribosylation in the absence of By-subunits was 3 - 5% of maximal incorporation in their presence. [32P]ADP-riboseincorporated into the cr,-subunit was determined after separation on SDS/PAGE. In each case a representative experiment is shown which was repeated at least twice. The variations between different preparations were within experimental error ( 55%).

Lubrol PX. It should be noted that in Fig. 5A in the presence of 10-50 pM GTP[S] and 10 mM MgClz at 20°C the highaffinity binding interaction between CI- and by-subunits is nearly completely abolished. An apparent Kd value for the specific GTP[S]-sensitive high-affinity interaction was calculated from Scatchard plots as 10 nM. The experiments in Fig. 5B were carried out with a,-subunits additionally purified by affinity chromatography with a By-matrix (Pang and Sternweis, 1989). The specific transfer in terms of changes of the donor fluorescence ranged from 35% for the associated (without GTP[S]) to 15% for the dissociated states (with GTP[S] and MgCl2), respectively. Hence, in this case the fluorescence energy transfer with homologous By-subunits was not completely quenched on addition of GTP[S] and M g Z f , but the Kd value for the high-affinity energy transfer remaining in the presence ofGTP[S] and Mgz+was about 10 nM, like the high-affinity portion of the transfer measured in the absence of

GTP[S] and Mg2+ (compare Fig. 5 with the experimental series I in Table 1). The specific GTP[S]-sensitivehigh-affinity energy transfer was quenched completely by addition of 1.5% SDS in detergent micelles (not shown). The nonspecific, lowaffinity transfer varied from about 40% up to 80% of the total (100%) transfer (see Tables 1 and 2). The extent of change allowed quantitative measurements over a 1 : 10- 100fold range of the reaction partners. The reproducibility with a given preparation was within experimental error I( 5%). The data could therefore be transformed in the form of Scatchard plots based on which Kd values were determined. A representative example of a Scatchard plot is shown as an insert in Fig. 5B. The variations using different preparations of labelled subunits could be as high as 30%, making comparisons between bovine brain and transducin by-subunits equivocal. This is expressed in the SEM values in Tables 1 and 2.

1175

0 00

0 20

0 LO

0 60

0 80

1 00

1 20

1 60

1 LO

2 00

1 80

n y -TRM.10-6#

YO

0

s

30.C

U

U \

20.0

--7 I1

w

10.0

x

".0.001

S

0.0

I

5 r-

I

0.~0

0.80

1.20

1.60

ny-TRM

, 10.0

. . .

20.0

E (%I

W

0.0

0

.

2.00

2.LO

2.80

30.0 1

220

M

Fig. 5. Fluorescence resonance energy transfer between ao-FITC and Py-TRM. (A) ao-FITC (3.3 nM) was mixed at 20°C with increasing concentration of Py-TRM (0-2.3 pM) in 50 mM NaCI, 1 mM EDTA, 1 mM dithiothreitol,0.25% Lubrol PX solution, pH 7.6. Measurements were made 20 min after mixing the components at 20°C. The decrease in the fluorescence of the donor a-FITC at 520 nm was recorded without GTP[S] (m) and with 50 pM GTP[S] and 10 mM MgCI2 (0). Ordinate: Energy transfer ( E ) is expressed as a percentage of the total. Fdais fluorescence of donor in the presence of acceptor; Fd is donor fluorescence without acceptor. A decrease of Fdato zero is taken as 100% (total) energy transfer. In the insert the lower range of concentration was amplified. (B) Fluorescence resonance energy transfer was measured between a,-FITC, further purified by affinity chromatography on a by-agarose column according to Pang and Sternweis (1989), and by-TRM. The experimental protocol was identical with that described in A. Decrease is shown in fluorescence of the donor ao-FITC at 520 nm with (0) and without ( W ) 50 pM GTP[S] and 10 mM MgCI2.A Scatchard plot of the data is in the insert.

When unlabelled fly-subunits were added in 0.2% Lubrol PX up to a 100-fold molar excess over by-TRM, resonance transfer was effectively quenched (not shown). This was also the case for the interactions with FITC-labelled receptor and by subunits. The same was true when unlabelled a, was added to the FITC-labelled L X ~(not shown). In other experiments (not shown) we noted some nonspecific energy transfer between free FITC and by-TRM with Kd values in the range 130 pM, that is in the same range as the GTP[S]-insensitive low-affinity transfer, whereas no such low-affinity transfer was observed between free unbound TRM and a,-FITC. Additional factors may contribute to the low-affinity transfer, but this was not studied in more detail. Although the experiments reported above provide convincing evidence that the high-affinity part of the energy resonance transfer between uo- and by-subunits and with the fl-

adrenoceptor is a valid measure of the specific binding interactions between the coupling partners, we wished nevertheless to obtain additional proof by an independent method under conditions comparable to those of fluorescence energy transfer measurements. Codina et al. (1984) found that the associated a- and by-subunits in trimeric G i sedimented at about 4 S and dissociated to a 2-S form at 32°C in the presence of GTP[S], representing separated a- and by-subunits. In Fig. 6, a representative example of the behaviour of FITC-a, and byTRM on isopycnic sucrose density gradient centrifugation at 20°C is shown. The Szo,w coefficients were 2.4 S for separate a. and fly-subunits (Fig. 6A). By means of densitometric and GTP[S] binding measurements, it was ascertained that the 2.8S form in Fig. 6 B represents ao-subunits. The higher than expected sedimentation coefficient most likely is due to association (Sternweis, 1986;Wessling-Resnick and Johnson, 1989).

1176 Table 1. Fluorescence resonance energy transfer measurements with a- and by-subunits of Go and G,. The method is described in Materials and Methods and in the legend to Fig. 5. The data are presented with the standard errors of mean (SEM). The temperature was 2O'C, the Lubrol concentration 0.2%, GTP[S] was 10 pM and MgC12 10 mM. Series

I

Number of cxpts

partners

Reaction

6

zO-FITC, PJ-TRM

Conditions

Extent of transfer

Kd .~

Specific

Nonspecific

Specific

Nonspecific

Lubrol

11 f 2

83 f 22

10k3

1 & 0.3

GTP[S] MgClz Lubrol lipid vesicles

none

70 f 5

none

1 f 0.2

11 & 2

n.d."

30_+10

I f 0.7 7 & 0.2

brain

I1

111

6

MO-FITC,Pj-TRM brain

3

aO-FITC, Py-TRM brain

3

xO-FITC, By-TRM brain

GTP[S] MgC1* lipid vesicles

none

76 f 5

none

2

@o-FITC,By-TRM transducin

Lubrol

23

58

%

Q-FITC, By-TRM

GTP[S] MgCL Lubrol

none

72 f 2

none

3

transducin

a

20

2

4 & 0.7

This value could not be determined by Scatchard transformation because nonspecific energy transfer did not reach its limIt value.

Non-associated By-subunits were also found in the absence of GTP[S] (see Fig. 6C) when fly was at a molar excess over M,,. At equimolar E,, and f l y concentrations and at 4"C, clo and by were completely associated in the absence of GTP[S] as shown in Fig. 6F. Therefore, taking into account the higher sedimentation coefficients of the separated cro-FITC and By-TRM (Fig. 6A) which are probably due to the introduction of the fluorophores, different temperatures and conditions, the agreement with the results of Codina et al. (1984) is reasonable. In any case, these experiments have proved that the introduction of the fluorophores did not markedly alter the association - dissociation behaviour of G0a and fly-subunits. Experiments were also carried out after reconstitution in lipid vesicles (see series I1 in Table 1 and I1 and 111 in Table 2) with comparable results. Also the Kd values calculated for the high-affinity fluorescence resonance energy transfer were somewhat lower than in detergent solution: 10 & 3 n M cf. 3 0 A 1 0 n M for a,,/&interactions (compare I and I1 in Table 1). So far as the affinity is concerned, the interactions between bovine brain mo and transducin By-subunits with and without GTP[S] did not appear to differ much from the homologous a,/p;-interactions; the actual fluorescence energy transferred was too small in the case of the transducin subunits for reliable quantitative interpretations. In Table 2, representative examples of a number of measurements of the interactions of the P1-adrenoceptor with bovine brain and transducin fly-subunits are shown, pointing to a higher affinity of the receptor for the former than for the latter fly-subunits (compare I and I1 in Table 2). As a whole, the affinities of the /l-receptor for &subunits were smaller than the affinities of a- and pi-subunits for each other (compare 1-111 in Table 1 with I-V in Table 2). I n addition to the representative experiment shown in series I of Table 2, we

have also compared interactions of transducin and bovine brain fly-subunits with nonactivated fl,-adrenoceptor in liposomes, but without removal of detergent. The rationale was to form more uniformly mixed lipid detergent micelles to trap the interacting partners and to equalize the fluorescence energy transfer between them. In these experiments (not shown), the Kd value for the high-affinity P-receptorltransducin-By interactions was again at least twice as large as that for the corresponding interaction with bovine brain ,&)-subunits: > 100 n M cf. = 50 nM. The extent of specific energy transfer was comparable ranging over 7-11% of the total. Although the differences were small, the results were consistent, suggesting strongly that transducin By-subunits compared to bovine brain fly-subunits have lower affinity to the PIadrenoceptor. The lower affinity in the case of interactions of transducin By-subunits with the fll-adrenoceptor (see I in Table 2) is probably a reflection of structural differences and of the greater hydrophilicity of transducin fly-subunits compared to the bovine brain fly-subunits. Figure 7 is a representative example for the interaction between bovine brain By-subunits tagged with T R M and the ,8,-adrenoceptor labelled with FITC in lipid vesicles. The interactions of fly-subunits with the isoproterenol-stimulated receptor and the nonactivated propranolol-occupied receptor were compared and found to be nearly the same (Kd z 120 nM and 100 nM, respectively). It is of interest to note that when the P-receptor was co-reconstituted with fly-subunits in lipid vesicles, the fluorescence energy transferred was much higher but was about the same, regardless of whether the receptor was activated o r not (compare I and I1 in Table 2). For the interaction between the fl-adrenoceptor and E,), one calculates a Kd value of 100 f 30 n M and for the interaction with &subunits a Kd value of 50 f 10 nM under compar-

1177 able conditions (compare I and V in Table 2). Howevcr, the difficulties inherent in the estimation of a Kd value when specific energy transfer is small (6 3 % in the experiment with ria)make the small difference doubtful. We have therefore tried to obtain additional information using a trimeric Go preparation labelled only in the ao-subunit with T R M (see series VI in Table 2). Here again, Kd was higher, 120 nM, and like that of zo (series V in Table 2), in support of the suggestion that a. as well as r/+ptrimer, like the separate ao-subunit, at large enough high-affinity energy transfer (24%) may bind

more weakly to the [I,-adrenoceptor than By-subunits alone. As was expected, addition of GTP[S] and MgClz increased the Kd values for the high-affinity interaction of the receptor with a. (V and VI in Table 2). Finally, the results of the experimental series IV and V in Table 2 show that addition of either unlabelled a"- or unlabelled by-subunits a t about 100fold molar excess over the respective Kd values for their highaffinity interactions completely quenched high-affinity fluorescence resonance energy transfer between FITC-labelled ,8receptor and TRM-labelled ,!Pi- and ao-subunits.

DISCUSSION Critique of method

201

t

70

I-

ny-TRM f 0 ) plus a-FlTC (0) with GTP yS

C

60

-

f 9

50

-

f

I

tly-TRMf 0 ) plus OL- FlTC f D) GTPyS 2 LS

r ul

1

3

4

k 0 -

01

f5

z -$

30

The experimental protocol details conditions which allow labelling without impairment of function of purified components of G-protein-mediated signal transmission chains, e.g. trimeric Goand ria- and /+subunits from bovine brain, transducin py-subunits from rod outer segment membranes and turkey erythrocyte p1-adrenoceptor with FITC or T R M , respectively. Although these conditions were satisfied, other factors limit the application of fluorescence resonance energy transfer to the quantification of interactions between the proteins in question. In our experiments, the lower limit to which specific high-affinity interactions could be measured was given by signal detectability. The lower limit for thc donor fluorescence was at 1 nmol xo and a t a Kd value of about 10 nM, the limit of reliable measurements was at about a \:I00 ratio of a,/py. Another factor which must be borne in mind is the heterogeneity of the interacting proteins (Simon et al., 1991). The a,-subunit with MI 39000 is probably a mixture of four closely related isoforms (Birnbaumer et al., 1989; Inanobe et al., 1990), whereas the bovine brain /hubunits were probably mixtures of the MI 37000, 36000 and 35000 polypeptides referred to as /13, P I - and B2-subunits (Gao et al., 1987; Levine et al., 1990). The y-subunits used in our

-

20-

(7i

t-@

l o t @

'

7

9

11

13 15 17 19 21 23 25

Froction ny -TRM

9

11

13 15 17 19 21 23 25

Fraction Number

Number

(0)

I)Y-TRMfO)pluS a-FITC (0)

without GTPyS

Fraction

Number

I

Fraction Number

rt-q==;

Fig. 6. Association/dissociation of modified a- and by-subunits in isopycnic sucrose density gradients. The samples in a final volumc of 500 p1 buffer A, 5% sucrose, 0.1 mM dithiothreitol and 0.2'/0 Lubrd PX, pH 7.8, were placed 011 top of 5-20% sucrose density gradients made in buffer A, 0.1 mM dithiothreitoland 0.2% Lubrol PX, pH 7.8. (A) (0)By-TRM (750 pmol) and (0) ao-FITC (250 pmol) wcre each loaded on a separate gradient. (B) One gradient was loaded simultaneously with 750 pmol by-TRM together with 250 pmol GTP[S]Mg'+-activated ao-FITC. ao-FITC was prcactivated with 10 pM GTP[S] in buffer A, 0.1 mM dithiothreitol, 10 mM MgCI2. 5% sucrosc and 0.2% Lubrol PX, pH 7.8. (C) With nonactivated ao-FITC and By-TRM. In E and F the samc experiments arc shown as in B and C but with equimolar concentrations (250 pmol) of ao-FITC and By-TRM. Tn A - D, the experiments were carried out at 20 C, and in E and F at 4°C. Centrifugation was at 65000 rpm for 90 min in a VTi-65.1 Beckman vertical rotor. The gradient was collected and aliquots of each fraction were analysed for GTP[S]binding according to Sternweis and Robishaw (1984) or concentratcd and subjected to SDS/PAGE. At the bottom of F is an example of where the protein bands in the gels are. They were stained with silver according to Oaklcy et al. (1980) and the protein concentrations were estimated densitometrically. The results were expressed as sedimentation coefficients. (D) Proteins for calibration: lactate dehydrogenase 7.3 S, bovine serum albumin 4.2 S and myoglobin 2.0 S were used for calibration (Sober, 1968). The data are representative of four similar experiments.

1178 Table 2. Fluorescence resonance energy transfer measurements of interactions between a B,-adrenoceptor and ao- and &subunits of Go and trimeric Go.A purified turkey erythrocyte p1-receptor (PAR) was used. For details see text and legends to Fig. 5. The nonactivated receptor was liganded with propranolol. The receptor was activated with I( -)-isoproterenol. Temperature was 20 -C; the lauroylsucrose and sodium cholate concentrations were 0.02% and 0.1YO,respectively. GTP[S] was 10 pM and MgClz 10 mM.

Series

Number of expts

Reaction partners

Conditions

Extent of transfer

Kd

-~

~-

Specific

Nonspecific

Y" I

3

1

BAR-FIX, Py-TRM brain PAR-FITC, By-TRM transducin

3

I1

PAR-FITC, hj-TRM brain

lauroylsucrose cholate lauroylsucrose cholate

18&3

nonactivated

6.3

61 f 6

Specific

Nonspecific

nM

PM

51 f 10

2

0.5

77

205

2.1

38 & 6

33 & 8

120 & 20

2

38f6

100 f 20

1.8

80+9

none

2 f 0.5

6+3

34 f 6

100 f 30

3 i 2

5.5

66

150

5.0

sucrosc cholate lauroylsucrosc cholate

none

79

none

6.2

lauroyl-

24

12

120

4.5

16

81

31 5

3.3

0.5

PAR in lipid

vesicles I11

3

PAR-FITC, Py-TRM brain

activated PAR in lipid vesicles

35

IV

3

BAR-FITC, By-TRM brain unlabelled ro (4 pmou

lauroyIsucrose cholatc

none

V

4

PAR-FITC, zO-TRM brain

1

BAR-FITC, CXO-TRM brain

lauroylsucrose cholatc GTP[S] MgCIz

6

0.5

lauroyl-

3

PAR-FITC, zO-TRM brain

unlabellcd

6

(6 POI)

BAR-FITC, Go-TRM brain

sucrose

cholate 7

PA R- FI TC, G 0-TRM

GTP[S]

brain

MgCL lauroylsucrose cholate

experiments were a mixture of M , x 8000 isoforms referred to as and ?,-subunits (Gautam et al., 1989; Robishaw et al., 1989) and a n additional polypeptide, yx, of which the sequence differs from that reported for y s and 76 (Baumann, 1990, unpublished experiments from this laboratory). Transducin from rod outer segments of the bovine eye is thought to have only one form of /?-subunit (PI) (Amatruda et al., 1988) and only one form of 7-subunit (yl) (Gautam et al., 1990). but additional /?-subunit isoforms may exist. Because of the heterogeneity of the preparations it remains uncertain what contribution each component makes and to what extent it may be rcsponsible for differences in affinities. The random labelling of the ro-subunit and of the P1-adrenoceptor preclude estimates of the distance between sites to which the ;is

fluorophores are attached, but this is not necessarily disadvantageous when one is primarily interested in the overall affinities of proteins for each other, as was the case here. Although proteins to be compared were selected not only on the basis of functional integrity but also with respect to a comparable degree of labelling with fluorophore, good matches between partners were not always achieved. In such instances, comparisons between different coupling partners did not give unequivocal information, as was pointed out. Interactions between ao- and /+subunits and these subunits with the P,-adrenoceptor Kohnken and Hildebrandt (1989) have recently quantified to Bg-subunits. The binding of bovine brain Gomj9and

1179

30.0

20.0

10.0

0.0 0.00 0.60

0.80

1-20 1.60 2.00

2.M

2.80

fly- TRM . l o - 6 M

3.20

3.60

L.00

L.10 6.80

I

Fig. 7. Fluorescence resonance energy transfer between turkey B1-adrenoceptorlabelled with FITC and By-TRM incorporated in lipid vesicles. For details see legend to Fig. 5 and Materials and Methods. fl,-Adrenoceptor labelled with FITC (3 nM) was mixed with increasing amounts of Py-TRM (0-5.2 pM) in buffer A with 0.1 mM dithiothreitol, 100 mM NaCI, 1 mM EDTA, pH 7.6. The lipid concentration was 0.1 mg/ ml. The receptor was activated with 0.1 JLMI(-)-isoproterenol (U) in the presence of 0.1 pM ascorbate or occupied by 0.1 pM /(-)propranolol (B). Incubation was for 20 min at 2 0 ' T The decrease in the donor fluorescence of the PI-adrenoceptor-FITC was recorded. In the insert is the Scatchard plot of the data.

by-subunits were biotinylated and it was shown by separation on sucrose gradients that GTP[S]-Mg2+ or AIF, in Lubrol PX buffer dissociated the ufly-holocomplexes. Affinity of [ly for a*, and uS9and binding capacity was estimated by binding the a-subunits to biotinyl-fly immobilized on streptavidinagarose. A Kd for cx41 binding to biotinyl-by of 19-24nM was estimated, whereas with cxo39 the Kdwas more than 10-fold larger, i.e. 0.34 - 0.39 pM. Also, the experimental conditions were not comparable, the apparent Kd values for the ao-py interaction reported by Kohnken and Hildebrandt (1989) are about ten times larger than the Kdvalues reported here (Fig. 5 ; Table 1 ) and at least three times larger than the EC,o value derived indirectly from the By-requirement for ADP-ribosylation of by pertussis toxin (Huff and Neer, 1986; Katada et al., 1986). Qualitatively however, these results agree with the data reported here and it should be noted that in both kinds of experimcnts the effect of GTP[S] was very much dependent on Mg2+ concentrations. In the experiments reported here, GTP[S] was no longer able to dissociate the r/3pholocomplex at Mg2' concentrations below 0.1 mM (not shown). This is also in agreement with the experiments of Higashijima et al. (1987b) who have shown that the release of GDP, which is assumed to be the rate-limiting step of the GTPase reaction, is strongly modulated by Mg2+ and Bysubunits. The residual high-affinity fluorescence energy transfer in the presencc of GTP[S]-Mg2+ between ao-subunits purified by affinity chromatography with a By-matrix according to Pang and Sternwcis (I 989) and bovine brain By-subunits (Fig. 5B) might be due to the greater hydrophobicity of the affinity-purified ro-preparation which may be myristoylated (Jones et al., 1990; Mumby et al., 1990a), since according to Pang and Sternweis (1989) and Linder et al. (1991) nonmyristoylated a. does not bind well to By-subunits coupled to agarose. Incomplete separation might also be interpreted to mean that complete dissociation of trimeric Go does not necessarily accompany activation by GTP[S]-Mg2+. Such a conclusion was recently reached by Yi et al. (1991) on the basis or experiments with Go subunits crosslinked with 1,6bismaleiinidohexane and activated by GTP[S].

A Kd value for the interaction between a,,- and transducin fly-subunits could only be estimated because the amount of energy transfer was too small for measurement but it was comparable to the corresponding interaction of a. with brain by-subunits (compare I and I l l in Table 1). This is reasonable considering that bovine brain and transducin fly-subunits could be exchanged in the case of Go,G iand G, without great functional changes (Cerione et al., 1987; Hekman et al., 1987). On transfer from detergent micelles into lipid vesicles (compare 1 with I1 and 111 in Table 2), an increase in the Kd values and an increase in the fluorescence energy transferred due to the specific high-affinity interactions was noted. This might reflect immobilization of the fluorescent receptor in liposomes. Borochov-Neori and Montal (1989), who have measured energy transfer between N-(l-pyreny1)maleimidelabelled rhodopsin (or transducin) and monobromobimanetransducin or 7-(diethylamino)-3-(4-maleimidylphenyl)-4methylcoumarin-labelled transducin (or rhodopsin) in rod outer segment membrane vesicles, have assigned an important role in energy transfer to trapping and immobilization of rhodopsin by lipids. In our preceding paper (Kurstjens et al., 1991),we presented qualitative evidence that bovine brain fly-subunitscan bind to the nonactivated b I-adrenoceptor from turkey erythrocyte membranes. The fluorescence resonance energy transfer measurements carried out with -adrenoceptor labelled with FITC and By-TRM in detergent micelles and in lipid vesicles substantiate these findings (Table 2). Moreover, binding of bovine brain by-subunits to isoproterenol-activated p-receptor was not much different from binding to the nonactivated receptor. This was actually expected in the light of the previous evidence (Kurstjens et al., 1991). Moreover, transducin bysubunits were consistently bound to a smaller extent and with lower affinity to the p,-receptor (see Table 2). This difference is actually not surprising because, as was pointed out by Gautam et al. (1989, 1990), carboxymethylation or geranylgeranylation of a COOH-terminal cysteine of bovine brain y-subunits (Yamane et al., 1990; Mumby et al., 1990b) should make brain by more hydrophobic than transducin-py.

1180 The affinity of the P-receptor for fly was considerably less than the affinity of a0 for By: Kd values were 50 nM compared to 10 nM in detergent solution and 100- 120 nM compared to 30 nM in lipid vesicles (compare I and I1 in Table 1 with I and I1 in Table 2). Accordingly, it was to be expected that unlabelled fly-subunits effectively compete with the P-receptor for ao-binding and, conversely, that a high enough molar excess of unlabelled xO-subunitscan interfere with binding of fly-subunits to the receptor (see IV and V in Table2). The partition between free and receptor-bound subunits is, however, drastically changed in the presence of GTP[S] when the Kd value for the c ( -fly ~ interaction is increased and is now in the micromolar range. The work reported here did not allow us to study the catalytic action of the activated P-receptor, since this would have required functionally intact active separate labelled E,subunits which are not available (see Kurstjens et al., 1991). Until such experiments are feasible, it remains a matter of speculation whether and to what extent the data reported here bear on the important question of the partitioning of Gprotein subunits between themselves, receptors and other partners (see also Law et al., 1991). The experiments do, however, demonstrate that fluorescence energy transfer under carefully controlled conditions may be used for studying selectivity and affinity of protein - protein interactions. The method may help mapping specific domains involved in binding of receptor and G-protein subunits (see Palm et al., 1989, 1990). It is likely that the method detailed here will find its most fruitful application in such studies. We are much indcbted to M s P. Fischer for expert technical assistance. This work was supported by grants from Deutsche For.sc11ungsgemeinschaft.He 22136-4, SFB 176, projects A1 and A2, University of Wiirpburg, the Fonds der Chemischen Industrie e.V. and by a pcrsonal award to E.J.M.H. by the VW-Stiftung, Hannover. E. J . M. H. gratefully acknowledges the hospitality offered to him by Drs Manfred Eigen and Tom and Donna Jovin during his stay at the Max Planck Institute for Biophysical Chemistry in Gottingen. Some of the work was carried out by M.F. and M.B. in partial fulfillment of the requirements for a Ph.D. degree in Biology at the University of Wiirzburg and by M.H. in partial fulfillment of the requirements for a Diploma in Chemistry at the University of Wurzburg.

REFERENCES Amatruda, T.. Gautam, N., Fong, H. K. W., Northup, J. K. &Simon, M. I. (1 988) J . Biol. Chem. 263, 5008 - 5011. Baumann. M. (1990) Doctoral thesis, University of Wiirzburg. Birnbaumcr, L.. Yatani, A,, Codina, J., Van Dongen, A., Graf, R., Mattera, R., Sanford, J. & Brown, A. M. (1989) Colloq. Ges. Biol. Chem. Moshuch 40, 147- 177. Birnbaumer, L., Abramowitz, J. & Brown, A. M. (1990) Biochim. Biophys. Acta 1031, 163-224. Bocge, F., Jiirss, R., Cooney, D., Hekman, M., Keenan, A. K. & Helmreich, E. J. M. (1987) Biochemistry 26, 2418-2425. Boege, F., Neumann. E. & Helmreich, E. J. M. (1991) Eur. J . Biochrm. 199, 1-15. Borochov-Neori, H. & Montal, M. (1989) Biochemistry 28, 1711 1718. Bradford, M. M. (1976) Anal. Biochem. 72,248-254. Cerione, R. A., Gierschik, P., Staniszewski, C., Benovic, J. L., Codina, J., Somers, R., Birnbaumer, L., Spiegel, A. M., Lefkowitz, R. J. & Caron, M. G. (1987) Biochemistry 26, 1485- 1491. Codina, J., Hildebrandt, J. D., Birnbaumer, L. & Sekura, R. D. (1984) J . Biol. Chem. 259,11408-11418. Feder. D.. Im, M.-J., Klein, H. W., Hekman, M., Holzhofer, A., Dees, C., Lcvitzki. A., Helmreich, E. J. M. & Pfeuffer, T. (1986) EMBU J . 5, 1509-1514.

FBrster, T. (1948) Ann. Phys. 6, 55-75. Gao, B.. Gilman, A. G. & Robishaw, J . D. (1987) Proc. Null Acrid. Sci. USA 84,6122-6125. Gautam, N., Baetscher, M., Aebcrsold, R. & Simon, M. 1. (1989) Science 244,971 -974. Gautam, N., Northup, J . K., Tamir, H. & Simon. M. I. (1990) Proc. Natl Acad. Sci. USA 87,7973 - 7977. Gennis, L. S., Gennis. R. B. & Cantor, C. R . (1972) Biochemistry 11, 2517 -2524. Gierschik, P., Simons. C.. Woodard. C., Somers, R. & Spiegel. A. (1984) FEBS Letr. 172, 321 - 325. Grunwald, G. B., Gierschik, P., Nirenberg, M. & Spicgel, A. (1986) Science 23I, 856-859. Sober, H. A. (ed.) (1968) Handbook cf’hiochemistry Chemical Rubber Publishing Co., Cleveland, Ohio. Hanski, E., Sternweis, P. C.. Northup, J. K.. Dromcrick. A. W. & Gilman, A. G. (1981)J. Biol. Chem. 256, 12911-12919. Heithier, H., Jaeggi, K. A . , Ward, L. D., Cantrill, R. C. & Helmreich. E. J. M. (1988) Biochimie 70, 687-694. Hekman, M., Feder, D., Keenan, A. K., Gal, A,, Klein, H. W.. Pfeuffer, T., Levitzki, A. & Helmrcich, E. J. M. (1984) EMBO J . 3, 3339 - 3345. Hekman, M., Holzhofer, A,, Gierschik, P., Im, M.-J.. Jakobs, K.-H.. Pfeuffer, T. & Helmreich, E. J. M. (1987) Eur. J. Biochem. 169, 431 -439. Higashijima, T., Ferguson, K. M., Sternweis, P. C., Smigel, M. D. & Gilman, A. G. (1987a) J . Biol. Chem. 262,762-766. Higashijima, T., Ferguson, K. M., Sternweis, P. C., Ross. E. M.. Smigel, M. D. &Gilman, A. G. (1987b) J . B i d . Chem. 262.752756. Huff,R. M.&Neer,E. J. (1986)J. Biol. Chem.261, 1105-1110. Im, M.-J., Holzhofer, A., Bottinger, H., Pfeuffer, T. & Helmrcich, E. J. M. (1988) FEBS Lett. 227, 225-229. Inanobe, A., Shibasaki, H., Takahashi, K., Kobayashi, I., Tomita, U., Ui, M. & Katada, T. (1990) FEBS Lett. 263, 369- 372. Jones, T. L. Z., Simonds, W. F., Merendino, J. J. Jr, Brann, M. R . & Spiegel, A. M. (1990) Proc. Natl Acad. Sci. USA 87, 568 - 572. Jiirss, R., Hekman, M. & Helmreich, E. J. M. (1985) Biochemi.s/ry24, 3349 - 3354. Kaplan, R. S. & Pedersen, P. L. (1985) Anal. Biochem. I50, 97- 104. Katada, T., Oinuma, M. & Ui, M. (1986) J. Biol. Chem. 261, 52155221. Kohnken, R. E. & Hildebrandt, J. D. (1989) J . Biol. Chem. 264, 20688 -20696. Kurstjens, N . P., Frohlich, M., Dees, C., Cantrill, R. C., Hekman, M. & Helmreich, E. J. M. (1991) Eur. J . Biochern. 197, 167-176. Laemmli, U. K. (1970) Nature (Lond.) 227, 680-685. Law, S. F., Manning, D. & Reisine, T. (1991) J . Biol. Chem. 266. 17885- 17 897. Levine, M. A., Smallwood, P. M., Moen, P. T. Jr, Helman, L. J. & A h , T. G. (1990) Proc. Natl Acad. Sci. USA 87,2329-2333. Linder, M. E., Pang, I.-H., Duronio, R. J., Gordon, J. I., Sternweis, P. C. & Gilman, A. G. (1991) J . Biol. Chem. 266,4654-4659. Mumby, S . M., Heukeroth, R. O., Gordon, J. I. & Gilman, A. G. (1990a) Proc. Natl Acad. Sci. USA 87, 728- 732. Mumby, S. M., Casey, P. J., Gilman, A. G., Gutowski, S. & Sternweis, P. C. (1990b) Proc. Nut1 Acad. Sci. USA 87, 5873 - 5877. Neer, E. J., Lok, J. M. & Wolf, L. G . (1984) J . Biol. Chem. 259, 14222-14229. Northup, J. K., Sternweis, P. C . & Gilman, A. G. (1983) J . Biol. Chem. 258,11361 - 11368. Oakley, B. R., Kirsch, D. R. & Morris, N. R. (1980) Anal. Biochem. 105, 361 -368. Palm, D., Munch, G., Dees, C. & Hekman, M. (1989) FEBS Lett. 254,89-93. Palm, D., Munch, G., Malek, D., Dees, C. & Hekman, M . (1990) FEBS Lett. 261,294-298. Pang, I.-H. & Sternweis, P. C. (1989) Proc. Natl Acad. Sci. U S A 86, 7814-7818. Pfeuffer, T. & Metzger, H. (1982) FEBS Lett. 146, 369 - 375. Pfeuffer, E., Mollner, S. & Pfeuffer, T. (1985) EMBU J . 4. 36753619.

1181 Pines, M., Gierschik, P. & Spiegel, A. (1985) FEBS Lett. 182, 355359. Puchwein, G., Pfeuffer, T. & Helmreich. E. J. M. (1974) J . Bid. Chern. 249,3232-3240. Robishaw, J. D., Kalman, V. K., Moomaw, C. R. Slaughter, C. A. (1989) J . B i d . Chem. 264, 15758-15761. Rooney, T. A., Elagcr, R. & Thomas, A. P. (1991) J . B i d . 266, 15068 - 15074. Salomon, Y.,Londos, C. & Rodbell, M. (1974) Anal. Biochem. 58, 541 - 548. SchPgger, H. & von Jagow, G. (1987) Anal. Biochem. 166,368-379. Schaffner, W. & Weissmann, C . (1973) Anal. Biochem. 56, 502-514. simon,M. I,, Strathmann, M. p. & Gautam, N.(1991) science 252, 802 - 808. Sternweis, P. C. (1986) J . B i d . Chem. 261, 631 -637.

Stcrnweis, P. C. & Robishaw, J. D. (1984)J. Biol. Chem. 259,1380613813. Stryer, L. (1968) Science 162, 526-533. Weber, G. & Teak, F. W. J. (1957) Trans. Faraday Soc. 53.646-655. Wessling-Resnick, M. & Johnson, G. L. (1989) Biochem. Biophys. Res. Comun. 159,651 -657. Yamane, H. K., Farnsworth, C. C., Xie, H., Howald, W., Fung, B. K. K., Clarke, S., Gelb, M. H. & Glomset, J. A. (1990) Proc. Natl Acad. sci, USA 87, 5868-5872. Yi, F., Denker, B. M. & Neer, E. J. (1991) J . Biol. Chem. 266,39003906. Zaremba, T., Gierschik, P., Pines, M., Bray, P., Carter, A., Kahn, K., Simons, C., Vinitsky, R., Goldsmith, P. & Spiegel, A. (1988) Mol. Pharmacol. 33,257 - 264.