G Protein coupling of the rat A1-adenosine receptor—Partial purification of a protein which...

Pergamon

PII: s0028-3908(9'r)00135-4

Neuropharmacology, Vol. 36, No. 9, pp. 1211-1219, 1997 0 1997 Elsevier Science Ltd. All rights reserved

Printed in Great Britain 002%3908/97 $17.00 + 0.00

G Protein Coupling of the Rat Al-Adenosine Receptor- Partial Purification of a Protein Which

Stabilizes the Receptor-G Protein Association

CHRISTIAN NANOFF,* MARIA WALDHOER, FLORIAN ROKA and MICHAEL FREISSMUTH

Institute of Pharmacology, Vienna University, Wiihringer Strape 13a, A-1090 Vienna, Austria

(Accepted 27 June 1997)

Summary-A membrane protein identified in cortical brain membranes and termed ‘coupling cofactor’, modulates G protein-coupling of the Al-adenosine receptor by reducing the catalytic efficiency of the receptor. Coupling cofactor traps the A1 -adenosine receptor in the high affinity complex and, thus, is responsible for the resistance of high affinity Al -agonist binding to modulation by guanine nucleotides. In the present work, this effect was used for assaying the activity of coupling cofactor by reconstituting guanine-nucleotide resistant agonist binding to rat Al-adenosine receptors in detergent extracted brain membranes or in membranes from 293 cells after stable transfection with receptor cDNA. Coupling cofactor was partially purified from porcine brain membranes. The specific activity was modestly enriched (N 5-fold) after three chromatographic steps (DEAE-Sephacel, AcA34, MonoQ pH 8). Rechromatography of coupling cofactor over MonoQ at pH 7 resulted in a loss in specific activity if membranes of 293 cells but not if brain membranes were used as acceptor membranes. In addition, the molecular mass estimated by gel filtration decreased from >150 kDa in the initial stage of purification to 40-30 kDa after this fourth chromatographic step. These two observations suggest that coupling cofactor requires an additional component that is present in brain membranes and is lost in later stages of purification. The activity of partially purified preparations of coupling cofactor activity relied also on the abundance of G protein a-subunits in the membrane. The activity on reconstitution with brain membranes or pertussis toxin pretreated 293 membranes was supported by addition of Gi, (rank order of protency: ail > ai > aiz) but. not of Go,. The selectivity for G protein a-subunits suggests that coupling cofactor may provide for an additional level of specificity in organizing receptor-G protein coupling. 0 1997 Elsevier Science Ltd.

Keywords-Al -adenosine receptor, G proteins, agonist binding, purification.

The purine nucleoside adenosine is present in high nanomolar concentrations in the extracellular fluid of many tissues including the central nervous system. The adenosine levels in the brain rise rapidly during increased nerve activity, hypoglycaernia, hypoxia and ischaemia and extracellular adenosine is thought to be physiologi- cally important for neuroprotection (Rudolphi et al., 1992), for example, by antagonizing the excitotoxic effects of glutamate (Manzoni et al., 1994). Adenosine is a powerful inhibitor of neurotransmitter release from presynaptic nerve terminals. In addition, it modulates postsynaptic action potentials and causes hyperpolariza- tion of neuronal cells [for a review see Rudolphi et al.

(1992)]. Both pre- and postsynaptic actions are mediated

*To whom correspondence should be addressed. Tel.: 43-l-40 480/298; Fax: 43-l-402. 48 33; E-mail: christian.nanoff@ univie.ac.at.

via Al-adenosine receptors which are widely distributed in the brain with the highest densities in cortex, hippocampus and cerebellum. The signal produced by Al-adenosine receptor activation is transduced to a G protein of the Gi/, class leading to the modulation of effecters, for example, enzymes such as adenylyl cyclase and phospholipase Cj5isoforms. In neuronal cells, Al- adenosine receptor stimulation has long been known to inhibit Ca2+ currents and activate potassium currents (Fredholm et al., 1994).

Reconstitution studies with the Al-adenosine receptor purified from bovine brain or expressed in Escherichiu

coli membranes adequately reproduced the basic char- acteristics of receptor G protein coupling and indicated a preference of the Al-adenosine receptor to associate with Gi, over G,,-subunits (Freissmuth et al., 1991b; Munshi et al., 1991; Jockers et al., 1994). The coupling between Al-adenosine receptor and G protein in rat brain

1211

1212 C. Nanoff et al.

membranes, however, revealed a peculiar feature; that is, the Al-adenosine receptor in rat brain but not in human brain membranes is subject to a ‘tight’ G protein coupling mode (as opposed to the classical model of collision coupling) caused by the interaction with an additional membrane associated protein, the coupling cofactor (Nanoff et al., 1995). We have shown that coupling cofactor may be extracted from brain membranes and have characterized its effect on signal transduction by the Ar -adenosine receptor. Coupling cofactor impedes the guanine nucleotide induced dissociation of the high affinity complex (consisting of agonist, receptor and G protein, H/R/G). Instead the receptor is trapped in the agonist high affinity conformation whereas catalytic activation of G proteins by the agonist liganded receptor is markedly restrained. Thus, coupling cofactor down- tones signal amplification following stimulation of the Ar-adenosine receptor.

Previously, we have used reconstitution studies with detergent extracts from brain membranes and extracted brain acceptor membranes to assess the role of coupling cofactor in the tight R/G (receptor-G protein) coupling mode. Using this reconstitution assay, we have further purified coupling cofactor in the present work. In addition, we have expressed the rat brain Ai-adenosine receptor in a human cell line to obtain an alternative source of acceptor membranes for reconstitution. Since we had observed that the activity of coupling cofactor also relied on the abundance of G protein cr-subunits, an additional aim of the present work was to examine the role of coupling cofactor in organizing the specificity of G protein mediated signal transduction.

METHODS

Materials

[35S]GTPyS [guanosine S-(3-O-thio)triphosphate] and [lz51] were purchased from NEN (Boston, MA, U.S.A.). [‘251]HPIA [N6-3-(iodo-4-hydroxyphenylisopropyl)aden- osine] was synthesized and purified according to Linden (1984). Guanine nucleotides and adenosine deaminase were from Boehringer Mannheim (Germany). XAC (xanthine amine congener) was obtained from RBI (Natick, MA, U.S.A.). Bovine serum albumin was from Serva (Heidelberg, Germany). The sources of the materials required for G protein purification have been described (Casey et al., 1989; Mumby and Linder, 1994). Buffers and salts were from Merck (Darmstadt, Germany). Hepes [4(2-hydroxyethyl)l-piperazineethane- sulfonic acid] and CHAPS (3[3-(cholamidopropyl)dia- methylammonio]-1-propanesulfonate) were from Biomol (Munich, Germany). The materials required for sodium dodecylsulphate-polyacrylamide gel electrophoresis were from BioRad (Richmond, CA, U.S.A.). Fetal calf serum (FCS), Dulbecco’s modified Eagle medium (DMEM) was obtained from GIBCO-BRL (Grand Island, NY, U.S.A.) The following reagents were from Sigma (St. Louis, MO): glutamine, penicillin G and streptomy-

tin, Lubrol PX. Pertussis toxin was obtained from Sigma-Aldrich, dissolved in 50% glycerol, 50 mM Tris pH 7.5, 10 mM glycine, 0.5 M NaCl at a concentration of 50 pg/n-ll.

HEK 293 cells (CRL 1573, American Type Culture Collection, Rockville, MD, U.S.A.) were a generous gift from Dr Ch. Pill, Institute of Biochemical Pharmacology, Vienna University. The cDNA coding for the rat Al- adenosine receptor in the plasmid vector pBC-AlR (Freund et al., 1994) were kindly provided by Dr M. J. Lohse (University Wtirzburg, Germany).

Stable transfection of HEK 293 cells with rat brain AI- adenosine receptor cDNA

HEK 293 cells were transfected with the rat brain Al- adenosine receptor cDNA as follows: cells were grown in 100 mm Petri dishes to a density of N 2-2.5 x lo6 cells/ dish using DMEM complete medium supplemented with 10% FCS, 2 mM gutamine, 100 U/ml penicillin G and 100 pg/ml streptomycin. Cells were maintained at a temperature of 37°C and 5% CO2. The plasmid vector cDNA [pBC-AlR, 7.5 pg see Freund et al. (1994)] together with 0.75 pg of plasmid containing the geneticin resistance gene under the control of a CMV promoter (pRc-CMV) were transfected into the cells by calcium phosphate precipitation (Chen and Okayama, 1988). Sixteen hours following transfection, the medium was removed and the cells were subjected to an osmotic shock by adding 15% glycerol for a few seconds. Cells were grown for another 24 hr in non-selective medium and replated in three dishes dispensing 50%, 25% or 17.5% of the total amount per dish. Positive clones were selected by cultivating the cells in the presence of G418 (0.8 mg/ml) for 4-6 days. Cells clones were picked and replated into a 48-well dish for further selection. About one-eighth of the selected cells yielded G418 resistant colonies containing the receptor. These colonies were replated in 100 mm dishes, grown to confluency and harvested for the preparation of plasma membranes. For treatment with pertussis toxin, cells were grown to 80% confluency in 100 mm dishes and were exposed to pertussis toxin at a concentration in 0.1 ,&ml diluted in DMEM complete medium supplemented with 10% FCS. After 24 hr, the cells were harvested, lysed and membranes were prepared.

Preparation of plasma membranes

At confluency stably transfected cells were rinsed free of culture medium and scraped off the dish using a plastic policeman. Cells were pelleted by centrifugation at 9OOOg, taken up in HME (HEPES 25 mM, pH 7.5, MgCl2 2 mM, ethylenediaminetetra-acetic acid (EDTA) 1 mM plus phenylmethylsulfonyl fluoride (PMSF) 0.1 mM, and aprotinin, l:lOOO, leupeptine 10 pg/ml) and were frozen in liquid nitrogen. After thawing, the cells were lysed by sonication. Membranes were pelleted by high-speed centrifugation (38 OOOg for 12 min) and washed once. The pellet was resuspended in HME at a

A cofactor for receptor-G protein coupling 1213

protein concentration of - 10 mg/ml and stored in aliquots at -80°C. The yield of membrane protein amounted to - 150 pg/dish of confluent transfected 293 cells (5 x lo6 cells).

Preparation of plasma membranes from rat brain cortex was carried out as outlined (Striiher et al., 1989), the preparation of memb’ranes from pig brain cortex was analogous to the preparation of bovine brain membranes (Freissmuth et al., 1991a). For the preparation of detergent-extracted acceptor membranes, rat brain mem- branes were solubilized as described (Nanoff et al., 1995). The membrane ;suspension was centrifuged for 12 min at 37 OOOg and resuspended at a protein-detergent ratio of (w/w) 4:l in solubilization buffer (10 mM CHAPS in HME). The membranes were resuspended using a glass pestle an,d a Dounce homogenizer (five strokes) and stirred on ice for 60 min. The suspension was centrifuged at 100 OOOg for 60 min. The pellet was washed once and resuspended in HME at a protein concentration of - 5 mg/ml and subsequently used as detergent-extracted acceptor membranes.

Radioligand binding experiments

Binding assays were c.arried out as described in Strijher et al. (1989). Saturation isotherms were generated on rat brain membranes and on membranes from stably transfected 293 cells. The assay was performed in a volume of 50 ~1 containing HEPESNaOH 25 mM (pH 7.5), EDTA 0.5 mM, MgC12 5 mM, adenosine deaminase 0.2 U/ml, plasma membranes (- 10 pg pro- tein), and [‘251]HPIA at the indicated concentrations. The binding reaction proceeded for 90 min at 25°C and was terminated by filtration over glassfiber filters which were thoroughly washed using a cell harvester (Skatron, Lier, Norway). Radioactivity bound to the membranes was determined in a Minaxi I!-counter (Canberra-Packard, IL, U.S.A.). Non-specific binding was determined with the addition of 5 PM CPA (N6-cyclopentyladenosine) or 1 PM XAC and amounted to 5% of total binding in the KD concentration range. Inhibition of IHPIA binding by GTPyS or by suramin was performed similarly with the radioligand concentration being in the KD range, that is, - 1.7 nM in 293 membranes and -0.7 nM in rat brain membranes.

The effect of soluble protein on IHPIA binding was evaluated as described (Nanoff et al., 1995). Soluble protein was preincubaLted with detergent extracted acceptor membranes or with membranes prepared from 293 cells at a detergent concentration of 6 mM CHAPS (60 min on ice). Before initiating the binding reaction, the mixture was diluted in HME [HEPESNaOH 25 mM (pH 7.5), EDTA 0.5 mM, MgC12 1 mM] resulting in a final detergent concentration of 1 mM CHAPS, and < 50 mM NaCl.

Protein purijcation

The G protein py-dimer was chromatographically resolved from the Go,i oligomer according to Casey et

al. (1989). Myristoylated recombinant Gia_1-3 and G,, was purified as described by Mumby and Linder (1994). The protein used was kindly donated by Dr Maureen Linder, Washington University, MO, U.S.A. The amount of G protein a-subunits used in individual experiments was determined by [35S]GTPyS binding in a medium containing (mM) 50 HEPESNaOH, pH 8, 1 EDTA, 10 MgS04, 0.1% Lubrol PX, 1 PM [35S]GTPyS (sp. act. -20 c.p.m./nmol). The incubation (2 hr at 30°C) was terminated by filtration over nitrocellulose BA85 filters. The concentration of G protein &subunit employed was calculated from the protein concentration assuming a molecular mass of 45 kDa.

Previously, we have used rat brain membranes as a source for the partial purification of coupling cofactor; for quantitative reasons, we have now employed pig brain membranes where coupling cofactor is also present to regulate the Al-adenosine receptor and have verified that solubilization of the protein component can be performed in an analogous manner. During the following chromato- graphic procedures coupling cofactor activity was tracked by reconstitution of sample aliquots with acceptor membranes and subsequent radioligand binding. Before the assay, samples were concentrated over YM30 membranes (Amicon) to a protein concentration of - 5- 10 mg/ml.

The following chromatographic steps were performed in the presence of 10 mM CHAPS.

I. DEAE-Sephacel batch chromatography (bed volu- me = 150 ml), equilibrated in solubilization buffer (HME, CHAPS 10 mM). Bulk elution was performed with increasing NaCl concentrations, the activity peak eluted at 150-330 mM NaCl in a volume of 150 ml. The volume was reduced in a pressure cell (Amicon) to one-fifth of the original volume using a porous polycarbonate membrane with a molecular mass cut off of N 30 000.

II. AcA34 gel filtration (- 350 ml bed volume), equili- brated and eluted with HME, 0.15 M NaCl, CHAPS 10 mM at a flow rate of 0.5 ml/min. The activity peak started to elute from the column at a volume of 180 ml and was contained in a volume of - 80 ml.

III. MonoQ anion exchange column chromatography (Pharmacia MonoQ column, bed volume - 8 ml) equilibrated in HME pH 8.0, CHAPS 10 mM and subsequently re-chromatographed on a 1 ml MonoQ column equilibrated at pH 7.0 (IV). Samples were applied in HMEKHAPS buffer with a NaCl concentration of d 10 mM. Elution was performed by applying a NaCl gradient from 150-350 mM where the activity peak elutes between 250 and 300 mM NaCl.

Superose gel filtration (bed volume N 24 ml). Equili- bration and filtration buffer: HME, pH 7.5; NaCl 0.15 M; CHAPS 10 mM. The concentrated activity peak (0.3 ml) from the second Mono-Q column was applied to a Pharmacia Superose column (bed volume - 24 ml). The

1214 C. Nanoff et al.

eluate was fractionated in 1.25 ml samples, the fractions were concentrated tenfold using Centricon micorconcen- trators (Amicon) and appropriate aliquots (corresponding to 10 ,ul original fraction volume) were added to acceptor membranes (detergent extracted rat brain membranes).

RESULTS

Characterization of the rat Al -adenosine receptor expressed in 293 cells

The rat brain Al-adenosine receptor cDNA was stably transfected into 293 cells using the pBC-AlR transfec- tion vector (Freund et al., 1994) and the geneticin resistance gene pRc-CMV. Expression of the Al- adenosine receptor was assayed by radioligand binding on membranes prepared from cells that have been selected and propagated in the presence of G418. Experiments with the selective agonist radioligand IHPIA demonstrated specific and saturable high affinity binding [Ko - 1.8 nM, Fig. l(B)]. Specific binding was not detectable in non-transfected 293 cells. In the following binding experiments cell membranes were with a receptor density that was similar to the density in rat brain cortical membranes ( - 500-600 fmol/mg mem- brane protein).

One particular feature of receptor-G protein coupling was different between A1 -adenosine receptors in rat brain

0

I I I I

A

::::

500

400

300

200

100

0

0 2 4 6 0 2 4 6 8

IHPIA (nM)

Fig. 1. IHPIA binding to (A) rat brain membranes and (B) to the rat Al-adenosine receptor expressed in 293 cells. Saturation isotherms were generated on membranes (- 10 pg membrane protein) prepared from rat brain cortex or from 293 cells stably transfected with rat Al -adenosine receptor cDNA. The binding reaction was carried out for 90 min at 25°C and terminated by filtration over glassfibre filters. IHPIA binding was performed in the absence (circles) or presence (squares) of GTPyS (3 PM). Non-linear least square analysis of the rat brain experiments gave the following parameter estimates: KD values were 0.47 + 0.07 nM in the absence, 0.34 f 0.06 nM in the presence of 3 PM GTPyS; B,, values were 465 f 20 fmol/ mg in the absence and 345 f 80 fmol/mg in the presence of

3 PM GTPyS (means f SD from three experiments).

membranes and the recombinant receptor expressed in 293 cells. This discrepancy is illustrated in Fig. 1. In rat brain membranes, the guanine nucleotide analogue GTPyS (3 PM) reduced high affinity agonist binding by only 25%. Analysis of saturation binding isotherms revealed that the binding curve was slightly but significantly shifted to the left indicating an increase in the receptor affinity in the presence of the guanine nucleotide [Fig. l(A)]. In contrast, in 293 membranes, the inclusion of GTPyS greatly reduced IHPIA high-affinity binding such that the residual binding could not be fitted reliably to a saturation isotherm by a non-linear least square algorithm [Fig. l(B)].

Previously, we have demonstrated that an ancillary membrane protein component (termed ‘coupling cofac- tor’) controls the signal transduction from the Al- adenosine receptor to the G protein in rodent brain (Nanoff et al., 1995). This component stabilizes the high affinity agonist/receptor G protein complex (R/G com- plex) even in the presence of high guanine nucleotide concentrations and downtones the catalytic amplification of the signal passed on by the activated receptor. As reported previously, the ability of GTPyS to dissociate the binding of IHPIA was weak in native rat brain membranes [Fig. 2(A), n ]. In 293 membranes, however, the inhibitory potency of GTPyS (1~50 - 50-150 nM) was roughly lOO-fold higher than in rat brain membranes

[Fig. 2(A), @I. In addition, we employed suramin, a polysulfonated

benzamidonaphtalene compound, to assess the Al- adenosine receptor-<; protein coupling in brain and 293 membranes. Suramin binds to G protein a-subunits and inhibits the interaction between various types of receptor and their cognate G protein in membranes (Butler et al., 1988; Huang et al., 1990; Freissmuth et al., 1996; Beindl et al., 1996). For the rat Al-adenosine receptor, uncoupling of the receptor-G protein complex by suramin was readily observed in 293 membranes [Fig. 2(B), A] and in detergent extracted rat brain membranes [Fig. 2(B), 01. However, the Al-adenosine receptor population in native rat brain membranes was resistant to the effect of suramin up to concentrations of 30 PM [Fig. 2(B), n ]. Thus, in 293 cell membranes as opposed to brain membranes, the rat A1 -adenosine receptor-G protein complex is readily dissociated by GTPyS and suramin. This suggests that a tight G protein coupling mode is not an intrinsic feature of the rat Al-adenosine receptor but that an additional component, that is, the coupling cofactor, controls receptor G protein coupling in rat brain.

Partial purijication and assessment of the molecular size of the coupling cofactor protein

While 293 membranes carry a high density of Al- adenosine receptors they harbor only moderate amounts of endogenous G protein g-subunits when compared to brain membranes (data not shown). G proteins markedly enhance coupling cofactor activity and some remainder

A cofactor for receptor-G protein coupling 1215

w rat brain l 293 cells

0 9 8 7 6 5 4

GTPyS (-log M)

0 01 1 10 100

suramin (PM)

Fig. 2. Inhibition of IHPLA binding to the rat Al-adenosine receptor by GTPyS and by suramin. (A) II-WA (1.5 nM) equilibrium binding was performed in the absence and in the presence of increasing concentrations of GTPyS, on rat brain membranes or on 293 membranes carrying the rat Al- adenosine receptor. (B) IHPIA equilibrium binding was carried out in native rat brain membranes (N 15 pg membrane protein, n ), in rat brain membranes subjected to detergent extraction ( N 25 pg membrane protein, 0) and to the recombinant Al- adenosine receptor expressed in 293 cell membranes (- 15 pg membrane protein, A). Inhibition of IHPIA binding by suramin at the indicated concentrations was assayed at IHPIA concentrations in the KD range, that is, 0.7 nM in rat brain

and 1.5 nM in 293 membranes.

of endogenous coupling cofactor activity appears to be

present in brain membranes even after detergent extrac-

tion. Therefore, 293 membranes were substituted for

brain membranes in assessing coupling cofactor activity

during purification.

Porcine brain membranes were solubilized and the

soluble extract was sequentially passed over the follow-

ing chromatographic supports, DEAF+Sephacel (bulk

elution), AcA 34 gel filtration, Mono Q anion exchange

chromatography at pH i3.0 and subsequently at pH 7.0

(see ). Coupling cofactor activity was tracked following

concentration of the fraction volume and reconstitution of an aliquot with acceptor membranes. Agonist receptor

012345678 01234567

soluble protein added (pg)

Fig. 3. Chromatographic enrichment of coupling cofactor activity extracted from pig brain membranes. A detergent extract was prepared from pig brain membranes and was batch adsorbed to DEAE-Sephacel (0) as outlined in the Methods section; the eluate was applied onto a gel filtration column (AcA 34, n ) followed by two rounds of anion exchange column chromatography (Mono Q, pH 8.0 A, pH 7.0 v). Peak activity fractions were tracked by reconstitution with acceptor membranes, were combined and increasing amounts were assayed on reconstitution with membranes from 293 cells expressing (A) the rat brain Al -adenosine receptor and (B) with rat brain acceptor membranes. Following preincubation of membranes with soluble protein IHPIA binding was assayed (in the presence of 0.1 PM recombinant cli) with and without the additon of GTPyS. The relative amount of specific IHPIA binding resistant to dissociation by 3 PM GTPyS (% resistance)

is given.

binding and its dissociation by GTPyS was determined. Peak activity fractions were combined and increasing amounts were assayed by reconstitution with 293 membranes. Protein recovery and specific activity was measured in the peak activity fraction after each

Table 1

Protein (mg)

Volume (ml)

Sp. act. Total activity (%‘olj@ recovered

DEAE! 804 150 2.2 35 376 AcA 34 106 75 5.3 11236 MQ, pH 8 12.5 40 12.8 3200 MO. DH 7 -2.0 5 4.5 180

Pig brain extract was applied to various chromatography matrices and the eluate was fractionated. Coupling cofactor activity was tracked by reconstitution with 293 membranes; fractions contain- ing the activity peak were pooled, protein content was determined by dye binding. Specific activity was assessed using serial dilutions of the pooled material (three concentrations) and estimated from the slope of the linear regression [see Fig. 3(A)]. Recovery of coupling cofactor activity during chromato- graphy is given in arbitrary units where the amount of protein required to reverse GTPyS-dependent inhibition of IHPIA binding by 50% was set one unit.

1216 C. Nanoff et al.

chromatographic step. The results are given in Fig. 3 and Table 1. A moderate enrichment of activity was achieved through steps I to III whereas on the subsequent column, that is, rechromatography on Mono Q at pH 7.0 a drastic loss in activity occurred apparently. If we used brain membranes for reconstitution, however, the specific activity was similar to that recovered from the preceding chromatography column. This is illustrated in Fig. 3(B). This finding indicates that the coupling cofactor activity was unlikely to be lost by degradation or inactivation of the protein; rather the reduced ability of coupling cofactor to associate with the receptor in 293 membranes may have resulted from a loss of a component required for interaction with the recombinant At -adenosine receptor. In order to substantiate this interpretation, we have determined the apparent molecular weight of coupling cofactor by gel filtration. If the material eluted from the first chromatographic step (DEAE-Sephacel) was subjected to gel filtration, coupling cofactor activity migrated with a molecular weight of - 150 kDa [not shown, see Nanoff et cd. (1995)]. However, if the coupling cofactor activity eluted from the second Mono-Q chromatography was applied to a Pharmacia Superose column (bed volume -24 ml), the peak of activity was recovered in fraction 10 (i.e. at an elution volume of 12.5-13.75 ml) and cu 30% of the activity was recovered in fraction 11. Calibration of the Superose column with four molecular weight markers indicated that proteins with a molecular mass of -50-25 kDa are expected to elute from the column at an elution volume of 12.5-14 ml. A predominant protein band could not be identified as a candidate for coupling cofactor at this stage of purification. Visualization of protein bands on a polyacrylamide gel, however, revealed several protein bands migrating in the appropriate molecular weight range. Among these were the a-subunits of the pre- dominant brain G protein, G,,, and fiy-subunits (data not shown) (Fig. 4).

G protein cc-subunit specijicity of coupling cofactor

Gi,, which is capable of a productive interaction with the Al-adenosine receptor enhances coupling cofactor activity and was used in the standard reconstitution assay at a fixed concentration (0.1 PM). Although G,, was present in partially purified prepartions of coupling cofactor obviously it would not substitute for Gi, in supporting reconstitution of coupling cofactor with the At-adenosine receptor. We therefore tested whether the coupling cofactor plays a role in enhancing the specificity of receptor G protein coupling. Thus, a fixed amount of coupling cofactor plus increasing concentrations of Gi,_t or G,, were co-reconstituted with two types of acceptor membranes, either detergent extracted brain membranes [Fig. 5(A and B)] or Al-receptor carrying membranes from 293 cells that had been pretreated with pertussis toxin to uncouple endogenous G proteins from the receptor [Fig. 5(C and D)]. In brain membranes, co- reconstitution with Gia_i but not with G,, supported

6 8 10 12 14 16 16

elution volume (ml)

Fig. 4. Molecular mass of the coupling cofactor protein. Following the chromatographic steps I-IV a preparation of coupling cofactor was applied to a Pharmacia Superose gelfiltration column (bed volume 23.6 ml) and collected in 1.25 ml fractions. Each fraction along the elution profile was concentrated and assayed on reconstitution with brain acceptor membranes. The Superose column was calibrated using the following molecular weight standards (relative molecular mass in kDa): bovine serum albumin (67), /$lactoglobulin (35), cyctochrom C (12.4) and cyanocobalamin (1.35). Given are the elution volurnina of the molecular weight standards (a), fitted by linear regression in a semi-logarithmic diagram and the activity of coupling cofactor as assessed by reversing the inhibition of MPIA binding in the presence of GTPyS (V). The solid line is a rendering of the UV trace of the elution profile.

This experiment was performed twice.

coupling cofactor activity [cf 0, Fig. 5(A and B)]. The G protein Gia-2, and Gicy_-3 produced intermediate enhance- ment of coupling cofactor activity (not shown). In pertussis toxin treated 293 membranes, reconstitution with a partially purified preparation fully restored receptor agonist binding presumably due to the contam- ination with Go, and by [Fig. 5(C and D), 0 at zero added a-subunit]. This binding, however, was fully sensitive to GTPyS. Similarly, sole addition of Giar reconstituted high-affinity agonist binding [Fig. 5(C), 01; this high-affinity agonist binding reconstituted by Gia_t was sensitive to GTPyS over the entire concentration range of Gia-1 [Fig. 5(C), n ] The effect of Gist on agonist binding was near maximal at a concentration of 25 nM reflecting the high affinity of the At-adenosine receptor for a-subunits of the Gi-class (Jockers et al., 1994). If a fixed amount of coupling cofactor was combined with increasing concentrations of Gi,i, the amount of agonist binding did not increase further [Fig. 5(C), 01; however, increasing concentrations of Giui progressively suppressed the ability of GTPyS to dissociate agonist binding and the maximum effect was reached at a concentration of - 200 nM [Fig. 5(C), q !]. As predicted from previous findings (Jockers et al., 1994)

A cofactor for receptor-G protein coupling 1217

5.0

2.5

o- 0 i

0.0

a

alpha i-l

-

alpha o

0 100 200 300

a-subunit added (nM)

Q-purified preparations of coupling cofactor [ 1 fig in (A and B), Fig. 5. Selective interaction with G protein c( subtypes

4 ftg in (C and D)] were rec,onstituted with acceptor membranes

Mono

together with increasing amounts of recombinant Gist (A and C) or G,, (B and D). Acceptor membranes are detergent extracted rat brain membranes in A and B or membranes prepared from 293 cells expressing the recombinant Ar- adenosine receptors that had been pretreated with pertussis toxin in C and D (which lea’ds to a complete loss of high-affinity agonist binding). Open symbols (0, q ) represent experiments carried out with the addition of coupling cofactor; filled symbols (0, n ) represent control experiments performed in the absence of coupling cofactor. After preincubation in the presence of 6 mM CHAPS, the protein mixture was diluted to give the indicated concentrations of cc-subunit (final detergent concentration - 1 mM CHAPS). IHPIA binding was carried out to equilibrium in the absence (circles) or presence (squares) of

7.5

5.0

2.5

o-

0.0 S i

B -E

6.0 2

g

4.0 z

2.0

0.0

3 AIM GTPyS.

recombinant G,, was less active in reconstituting high affinity binding to the Ai-adenosine receptor than Gia-1 [Fig. 5(B and D), 01. More importantly, a, failed to support coupling cofactor activity in both, rat brain membranes and pertussls toxin-treated 293 membranes [Fig. 5(B and D), n ].

DISCUSSION

In the present work we expressed the rat brain Al- adenosine receptor in a cell line (293) derived from human embryonic tissue and showed that G protein coupling of the Ar-adenosine receptor is divergent from that in native rat cortical membranes. The differences suggest that the coupling cofactor which regulates the

Ar -adenosine receptor-G protein interaction in rat brain membranes is absent in 293 cell membranes and the pertinent observations may be summarized as follows.

The sensitivity of the A,-adenosine receptor to modulation by GTPyS and suramin is markedly (>lOO-fold) higher in 293 membranes than in rat brain membranes. In the presence of 3 PM GTP@, IHPIA binds to a large proportion of the receptor population in rat brain membranes with a slightly increased binding affinity. This is consistent with findings previously obtained in reconstitution experiments (Nanoff et al., 1995). In 293 membranes, however, GTPyS reduces the affinity of almost the entire receptor population and precludes labelling of Ai -adenosine receptors with high affinity. These results suggest that G protein coupling of the rat receptor expressed in 293 membranes is similar to that of the Al-adenosine receptor in human brain rather than in rat brain. The human brain Al-adenosine receptor is not under the control of coupling cofactor but reveals the features of the classical ternary complex model of R-G protein coupling.

coupling cofactor could be conferred to the recombinant Ar -adenosine receptor. Most importantly, selective interaction with G protein a-subunits was observed on

In addition, we demonstrated that the activity of brain

reconstitution with the native Al-adenosine receptor in brain membranes and with the recombinant receptor in 293 membranes that had been uncoupled from endogen- ous G proteins by pertussis toxin. Under the latter experimental conditions, high-affinity agonist binding is dependent on the addition of exogenous G proteins and the selectivity of partially purified coupling cofactor for Gi,-1 vs G,, was faithfully reproduced. This finding can only be explained by assuming that coupling cofactor activity is conferred by a specific component.

While the reconstitution assay used to monitor coupling cofactor activity makes it difficult to calculate specific activities precisely, enrichment of activity was clearly obtained in the initial three steps of purification, irrespective of whether rat brain membranes or 293 cell membranes were used as acceptor substrate. However, rechromatography over MonoQ at pH 7 resulted in an apparent loss in specific activity upon reconstitution in 293 membranes; we also failed to obtain substantial enrichment in specific activity in additional chromato- graphic steps (i.e. phenylsuperose and hydroxyapatite). In addition, at this stage of purification, the apparent molecular weight of coupling cofactor was reduced substantially. This suggests that coupling cofactor initially migrated in a complex and that one component of coupling cofactor had been lost which is required for efficient reconstitution with 293 membranes. Detergent extracted rat brain membranes may still contain this component at levels sufficient to support coupling cofactor activity and hence the specific activity is less

1218 C. Nanoff et al.

affected if reconstitution is carried out with rat brain acceptor membranes. Furthermore, these reconstitution experiments with partially purified coupling cofactor preparations revealed an additional similarity between the Ai-adenosine receptor in 293 membranes and in native human brain membranes. Both types of mem- branes are poorer acceptor substrates for the reconstitu- tion of coupling cofactor activity than are rat brain membranes. In human brain membranes the presence of coupling cofactor activity was detected only after reconstitution of human membrane extracts with rat acceptor membranes (Nanoff et al., 1995).

Several lines of evidence have indicated that in the native membrane receptors and G proteins are organized in multimeric complexes by structural components forming microdomains [reviewed by Neubig (1994)]. The use of defined G protein cc-subunits offered the advantage to test the specific requirements for G proteins in supporting coupling cofactor activity. Thus, the Gicu(l_3) subunits supported coupling cofactor activity whereas G,, did not. Modest differences (up to lo-fold) have been observed in the affinity of At -adenosine receptors for G proteins of the Gi,/G,,-group (Freiss- muth et al., 1991b; Munshi et al., 1991; Jockers et al., 1994). However, the difference between Gia-1 and G,, in supporting coupling cofactor activity was clearly more pronounced, that is, all or none. Given that in brain membranes G,, is present in markedly higher concentra- tions than the isoforms of Gi,, a modest preference of the Ai-adenosine receptor for Gi, over G,, may not be biologically relevant. In contrast, the high degree of selectivity for coupling cofactor suggests an important role for coupling cofactor. We speculate that the latter may act as a scaffold and thereby provide an additional level of specificity in organizing R/G coupling.

If slices of rat cortex are incubated with GTPyS, the nucleotide is depleted while the levels of Gpp(NHp), a related analogue, do not decrease upon incubation with the tissue. The loss of GTPyS has been proposed to account for the inability of GTPyS to suppress Ai -agonist binding under these conditions (Parkinson and Fredholm, 1992). However, it is highly unlikely that the coupling cofactor activity is due to a hydrolytic enzyme that accepts GTPyS as a substrate.

GppNHp also fails to destablize high affinity agonist binding in the presence of coupling cofactor (Stroher et ul., 1989). The activity of coupling cofactor depends on the type of acceptor substrate for reconstitution (i.e. rat brain membranes, human brain membranes, 293 mem- branes, see Fig. 3). In the presence of high amounts of coupling cofactor, GTPyS stimulates agonist binding (due to increased agonist affinity of the R/G complex) such that the binding in the presence of GTPyS exceeds that measured in its absence [see Nanoff et al. (1995)]. Similarly, we rule out that the G protein fly-dimer

mediates cofactor activity as purified /?y-subunit complexes exerted only a modest effect on GTPyS resistance when reconstituted to acceptor membranes (Nanoff et al., 1995). In addition, coexpression of various combinations of /3y-dimers did not confer guanine nucleotide resistance to agonist binding in a recombinant expression system (Figler et al., 1996).

The molecular identity of the candidate protein for the coupling cofactor remains to be determined. At present, several proteins are known to interfere with R/G coupling and are stably localized on or translocate to the inner side of the cell membrane for action. Among them, a family of proteins (RGS for regulators of G protein signalling encoded for by the homologue of the SST2 gene from yeast) has been identified very recently in mammalian tissue; members of this family are negative regulators of G protein-mediated signal transduction (DeVries et al., 1995). Several RGS proteins (GAIP, RGS 1,4 and 10) were reported to act as GAPS (GTPase activating proteins) for c( subunits of the Gi/, class (Berman et al., 1996; Hunt et al., 1996; Watson et al., 1996). Since coupling cofactor also exerts a negative effect on receptor-induced signalling (Nanoff et al., 1995), there is an obvious analogy between the role of RGS proteins and that of coupling cofactor. However, RGS proteins do not inhibit guanine nucleotide exchange nor do they associate with the GTPyS bound form of CI subunits (Watson et al., 1996).

We have also previously ruled out several structural and regulatory proteins (e.g. tubulin, phosducin) as candidates for coupling cofactor (Nanoff et al., 1995). In cells of neuronal origin various non-receptor proteins have been identified that associate with G, and enhance guanine nucleotide exchange: neuromodulin [GAP43; Strittmatter et al. (1993)], P-amyloid precursor protein (Okamoto et al., 1995) and a membrane associated protein component from a neuroblastoma-glioma hybrid cell line (Sato et d., 1996). These proteins, however, exert an effect opposite to the effect of coupling cofactor. Thus, the function of coupling cofactor appears to be unparalleled in the repertoire of recently identified G protein modulators.

Acknowledgements-This work was supported by a grant from the Austrian Science Foundation (FWF-P12125GEN) to CN and the Biomed program of the European Community (ENBST).

REFERENCES

Beindl W., Mitterauer T., Hohenegger M., IJzerman A. P., Nanoff C. and Freissmuth M. (1996) Inhibition of receptor-G protein coupling by suramin analogues. Molecular Pharmacology 50: 415-423.

Berman D. M., Wilkie T. M. and Gilman A. G. (1996) GAIP and RGS4 are GTPase-activating proteins for the Gi subfamily of G protein tl subunits. Cell 86: 445-452.

Butler S., Kelly E. C. H., McKenzie F., Guild S., Wakelam M. 0. and Milligan G. (1988) Differential effects of suramin on

A cofactor for receptor-G protein coupling 1219

the coupling of receptors to individual species of pertussis toxin-sensitive guanine nucleotide binding proteins. Bio- chemical Journal 251: :!Ol-205.

Casey P. J., Graziano M. P. and Gilman A. G. (1989) G protein beta gamma subunits from bovine brain and retina: equivalent catalytic support of ADP-ribosylation of alpha subunits by pertussis toxin but differential interactions with G, alpha. Biochemistry 28: 611-616.

Chen C. A. and Okayama H. (1988) Calcium phosphate- mediated gene transfer: a higly efficient transfection system for stably transforming cells with plasmid DNA. Bio-

techniques 6: 632-638. DeVries L., Mousli M., Wurmser A. and Farquhar M. G. (1995)

GAIP, a protein that specifically interacts with the trimeric G protein Gict3 is a memb’er of a protein family with a highly conserved core domain. Proceedings of the National Academy of Sciences of the U.S.A. 92: 11916-l 1920.

Figler R. A., Graber S. G., Lindorfer M. A., Yasuda H., Linden J. and Garrison J. C. (1996) Reconstitution of recombinant bovine At-adenosine receptors in Sf9 cell membranes with recombinant G proteins of defined composition. Molecular Pharmacology 50: 1587-1595.

Fredholm B. B., Abbrachio M. P., Bumstock G., Daly J. W., Harden T. K., Jacobson K. A., Leff P. and Williams M. (1994) Nomenclature and classification of pruinoceptors. Pharmacological Reviews 46: 143-156.

Freissmuth M., Schtitz W. and Linder M. E. 1991b Interactions of the bovine brain At adenosine receptor with recombinant G protein a-subunits. Journal of Biological Chemistry 266: 17778-17783.

Freissmuth M., Boehm S., Beindl W., Nickel P., IJzerman A. P., Hohenegger M. and Nanoff C. (1996) Suramin analogues as subtype-selective G protein inhibitors. Molecular

Pharmacology 49: 602-6 11. Freund S., Ungerer M. and Lohse M. J. (1994) At-adenosine

receptors expressed in CHO-cells couple to adenylyl cyclase and to phospholipase C. Naunyn-Schmiedeberg ‘s Archives of Pharmacology 350: 49956.

Huang R.-R. C., Dehaven R. N., Cheung A. H., Diehl R. E., Dixon R. A. F. and Strader C. D. (1990) Identification of allosteric antagonists of receptor-guanine nucleotide binding protein interactions. McNlecular Pharmacology 37: 304-3 10.

Hunt T. W., Fields T. A., Casey P. A. and Peralta E. G. (1996) RGSlO is a selective activator of Gai GTPase. Nature 383: 175-177.

Jockers R., Linder M. E., Hohenegger M., Nanoff C., Bertin B., Strosberg A. D., Mamllo S. and Freissmuth M. (1994) Species difference in the G protein selectivity of the human and the bovine At -adenosine receptor. Journal of Biological Chemistry 269: 32077-32084.

Linden J. (1984) Purification and characterization of (-)[ 1251]hydroxyphenylisopropyladenosine, an adenosine R-site agonist radioligand and theoretical analysis of mixed stereoisomer radioligand binding. Molecular Pharmacology

26: 414423.

Manzoni 0. J., Manabe T. and Nicoll R. A. (1994) Release of adenosine by activation of NMDA receptors in the hippocampus. Science 265: 2098-2101.

Mumby S. M. and Linder M. E. (1994) Myristoylation of G protein cc-subunits. Methods in Enzymology 237: 254-267.

Munshi R., Pang I. H., Stemweis P. C. and Linden J. (1991) At - adenosine receptors of bovine brain couple to guanine nucleotide binding proteins Gi 1, Gi2, and Go. Journal of

Biological Chemistry 266: 22285-22289.

Nanoff C., Mitterauer T., Roka F., Hohenegger M. and Freissmuth M. (1995) Species differences in At -adenosine receptor-G protein coupling: identification of a membrane protein which stabilizes the association of the receptor G protein complex. Molecular Pharmacology 48: 806-817.

Neubig R. R. (1994) Membrane organization in G protein- mechanisms. FASEB Journal 8: 939-946.

Okamoto T., Takeda S., Murayama Y., Ogata E. and Nishimoto I. (1995) Ligand-dependent G protein coupling function of amyloid transmembrane precursor. Journal of Biological

Chemistry 270: 42054208.

Parkinson F. E. and Fredholm B. B. (1992) Differential effect of magnesium on receptor-G-protein coupling of adenosine At and A2 receptors: a quantitaive autoradiographical study. Molecular Neuropharmacology 1: 179-186.

Rudolphi K. A., Schubert P., Parkinson F. E. and Fredholm B. B. (1992) Adenosine and brain ischemia. Cerebrovascular

Brain Metabolism Reviews 4: 346-369.

Sato M., Ribas C., Hildebrandt J. and Lanier S. M. (1996) Characterization of a G protein activator in the neuroblas- toma glioma cell hybrid NG108-05. Journal of Biological

Chemistry 271: 30052-30060.

Strittmatter S. M., Cannon S. C., Ross E. M., Higashijima T. and Fishman M. C. (1993) GAP-43 augments G protein- coupled receptor transduction in Xenopus laevis oocytes. Proceedings of the National Academy of Sciences of the

U.S.A. 90: 5327-5331. Stroher M., Nanoff C. and Schtitz W. (1989) Differences in the

GTP-regulation of membrane-bound and solubilized At - adenosine receptors. Naunyn-Schmiedeberg’s Archives of

Pharmacology 340: 87-92.

Watson N., Linder M. E., Druey K. M., Kehrl J. H. and Blumer K. J. (1996) RGS family members: GTPase activating proteins for heterotrimeric G protein a-subunits. Nature 383:

172-174.