Functional modifications of transducin induced by cholera or pertussis-toxin-catalyzed ADP-ribosylation

Eur. J. Biochem. 210, 33-44 (1992) (CC FEBS 1992

Functional modifications of transducin induced by cholera or pertussis-toxin-catalyzed ADP-ribosylation FrCdkric BORNANCIN, Michel FRANCO, Joelle BIGAY and Marc CHABRE Centre National de la Recherche Scientifique, Institut de Pharmacologie Moleculaire et Cellulaire, Valbonne, France

(Received June lO/July 7, 1992) - EJB 92 0816

Transducin (TaPy), the heterotrimeric GTP-binding protein that interacts with photoexcited rhodopsin (Rh*) and the cGMP-phosphodiesterase (PDE) in retinal rod cells, is sensitive to cholera (CTx) and pertussis toxins (PTx), which catalyze the binding of an ADP-ribose to the a subunit at Argl74 and Cys347, respectively. These two types of ADP-ribosylations are investigated with transducin in vitro or with reconstituted retinal rod outer-segment membranes. Several functional pertur- bations inflicted on Ta by the resulting covalent modifications are studied such as: the binding of Ta to TPy to the membrane and to Rh*; the spontaneous or Rh*-catalysed exchange of GDP for GTP or guanosine 5-[y-thio]triphosphate (GTP[yS]), the conformational switch and activation undergone by transducin upon this exchange, the activation of TaGDP by fluoride complexes and the activation of the PDE by TaGTP.

ADP-ribosylation of transducin by CTx requires the GTP-dependent activation of ADP-ribosylation factors (ARF), takes place only on the high-affinity, nucleotide-free complex, Rh*-Tae,,,,-T/3y and does not activate Ta. Subsequent to CTx-catalyzed ADP-ribosylation the following occurs : (a) addition of GDP induces the release from Rh* of inactive CTxTaGDP (CTXTa, ADP-ribosylated a subunit of transducin) which remains associated to TPy ; (b) CT"TaGDP-TPy exhibits the usual slow kinetics of spontaneous exchange of GDP for GTP[yS] in the absence of Rh*, but the association and dissociation of fluoride complexes, which act as y-phosphate analogs, are kinetically modified, suggesting that the ADP-ribose on Arg174 specifically perturbs binding of the y-phosphate in the nucleotide site; (c) CTxTaGDP-Tfly can still couple to Rh* and undergo fast nucleotide exchange; (d) CTxTaGTP[yS] and crxTaGDP-AIF, (AlF,, Aluminofluoride complex) activate retinal cGMP- phosphodiesterase (PDE) with the same efficiency as their unmodified counterparts, but the kinetics and affinities of fluoride activation are changed; (e) CTxTaGTP hydrolyses GTP more slowly than unmodified TaGTP, which entirely accounts for the prolonged action of CTxTaGTP on the PDE; (f) after GTP hydrolysis, CTxTaCDP reassociates to T/3y and becomes inactive. Thus, CTx catalyzed ADP-ribosylation only perturbs in Ta the GTP-binding domain, but not the conformational switch nor the domains of contact with the TPy subunit, with Rh* and with the PDE.

PTx is active on TaGDP in the absence of membrane and of ARF, but the cooperation of TPy is needed. Subsequent to PTx catalyzed ADP-ribosylation (a) PTxTaGDP remains associated to TPy and is inactive, (b) PTxTaGDP-TPy displays the usual slow kinetics of spontaneous exchange of GDP for GTP[yS] in the absence of Rh* and unmodified association and dissociation kinetics for fluoride complexes, suggesting that the nucleotide-binding domain is not perturbed, (c) PTXTaGDP-Tfly does not bind to Rh* and thus does not undergo the fast-receptor-catalyzed nucleotide exchange, but (d) PTxTaGDP-AIF, activates PDE with the same efficiency as its unmodified counterpart. Thus, PTx- catalyzed ADP-ribosylation only perturbs the receptor contact domain of Ta but not the GTP binding domain, the conformational switch, or the effector and the TBy contact domains.

These two modifications are analogous to point mutations at Arg174 or Cys347, which would affect two structurally and functionally independent domains of Ta. Sequential ADP-ribosylations by CTx and PTx are, however, much hindered.

The heterotrimeric GTP-binding proteins (G proteins) that transmit signals from cell-membrane receptors to intra- cellular effectors are sensitive to the action of bacterial toxins such as cholera-toxin (CTx) from Vibrio cholerue, and pertussis-toxin (PTx) from Bordetellu pertussis, which perturb their functions by catalyzing the ADP-ribosylation of their

guanine-nucleotide-binding a subunits [l , 21. Among the nu- merous heterotrimeric G proteins identified so far, a few, like the stimulatory G protein (GJ, seem to be sensitive only to CTx [3], while others, like the inhibitory G protein (Gi), only to PTx [4]. Originally, only transducin (Tap?), the G protein of retinal rods, was sensitive to both toxins [5 - 71. CTx trans-

Correspondence to M. Chabre, Centre National de la Recherche Scientifique, Institut de Pharmacologie, 660, route des Lucioles Sophia Antipolis, F-06560 Valbonne, France

Abbreviations. ROS membranes, retinal rod outer segment membranes; Rh, rhodopsin; Rh*, photoactivated rhodopsin; Tcc, Tpy, transducin subunits; CTx, cholera toxin; PTx, pertussis toxin; C T x T ~ and PTxTz, ADP-ribosylated CI subunit of transducin; PDE, retinal

cGMP-phosphodiesterase; GPP[NH]P, guanosine 5'-[p,y-imido] triphosphate; GPP[CH,]P, guanosine 5'-[/3,y-methylene]triphosphate; GTP[yS], guanosine 5'-[y-thioltriphosphate; G protein, GTP binding protein; G,, stimulatory G protein; Gi, inhibitory G protein; AlF,, aluminofluoride complex.

Note. Both F. Bornancin and M. Franco contributed equally to this work.

34

fers an ADP-ribose from NAD’ onto an arginine located close to the GTP-binding site, most probably Arg174 in Ta [S]. We will denote the modified a-subunit of transducin as CTxTa. PTx ADP-ribosylates a cysteine located four residues away from the carboxy terminus, which is Cys347 in transducin [9]. We will denote the modified protein as PTxTa. The two modifications seem to both reduce the GTPase turnover rate of G proteins in situ, but have opposite effects on the activation by G proteins of their effectors. CTx-catalyzed ADP-ribosylation of Argl74 which is close to the nucleotide- binding domain in Ta, prolongs the action of the G protein on its effector, the retinal cGMP-phosphodiesterase (PDE), by reducing the GTP hydrolysis rate in CTxTaGTP. This main- tains the G protein in its PDE-activating TaGTP state for a longer time [7]. PTx-catalyzed ADP-ribosylation of Cys347, which is most probably located in a receptor-contact domain of TaGDP, uncouples the G protein from its specific agonist- liganded receptor, i.e. photoactivated rhodopsin (Rh*) [6], and prevents the Rh*-catalyzed GDP/GTP exchange and G- protein activation. AS a consequence, the GTP-dependent action of transducin on the PDE appears as quenched by PTx.

While these basic schemes of CTx and PTx action are commonly accepted, many issues remain confused or contro- versial. The lack of sensitivity of a G protein, like G, to PTx correlates with the absence of a subterminal cysteine, but the lack of sensitivity to CTx by other G proteins is not related to the absence of a target arginine, as the peptide segment which contains Arg174 is very well conserved in all heterotrimeric G proteins. Gierschik and Jakobs [lo] first observed that, in HL 60 cells, a PTx-sensitive Gi, which couples a chemotactic receptor to a phospholipase C, can also be a CTx substrate provided the receptor is activated by its agonist [ l l , 121. Recently Iiri et al. [13] showed that Gi can even undergo, albeit with low efficiency, the two modifications induced sequentially by CTx and PTx. We have recently shown [14] that transducin is a substrate for CTx only when it is coupled to photoexcited rhodopsin and depleted of its nucleotide.

The action of CTx on a heterotrimeric G protein requires the presence of ADP-ribosylating factors (ARF) which are also G proteins but of the small monomeric ‘ras-like’ superfamily [15, 161. The requirement for activation of ARF by GTP or analogs may account for the strange dependences of the sensitivity of G proteins to CTx on the presence of GTP or of non-hydrolyzable analogs [14]. The effect of CTx- catalyzed ADP-ribosylation on the rate of endogenous GTP hydrolysis by G protein a subunits is well established in the case of transducin [7] and can entirely account for the increased action of the modified G protein on its effector. How- ever, it is often assumed that besides the decreased rate of endogenous GTP hydrolysis, CTx-catalyzed ADP-ribosylation has other effects on G proteins, namely, that the modification ‘activates’ the Ga subunit, regardless of what nucleotide is bound to it. Kahn and Gilman [17] have also claimed that CTx-catalyzed ADP-ribosylation can inhibit the reassociation of Ga to GBy after the hydrolysis of GTP, and suggested that this lasting dissociation of ADP-ribosylated GaGDP from GPy results in the increased action of CTxGaGDP on the effector.

With PTx, ADP-ribosylation of the subterminal cysteine is observable only on the GDP-bound form of the sensitive Ga and does not require the presence of membranes or of any GTP-dependent ARF, but GBy is needed as a catalyst [18,19]. It is not clear whether PTx-induced ADP-ribosylation only uncouples the receptor-contact domain of GaGDP or whether

it also modifies the rate of receptor-independent GDP/GTP exchange or the rate of binding of fluoride complexes which act as y-phosphate analogs in the GDP binding site [20-221. Once nucleotide exchange (preferably with a non-hydrolysable GTP analog) or fluoride activation is complete, does the ADP-ribosylated Ga switch to the active conformation and activate the effector as efficiently as an unmodified Ga? In other words, does PTx-catalyzed ADP-ribosylation act exclusively on the receptor-contact domain of Ga, or does it interfere with any other of the Ga functional domains [23, 241 such as the GBy contact domain the nucleotide-binding domain and the effector-contact domain?

To answer these questions, we investigate in detail the pertubations imparted to the functional domains of the transducin a subunit by both of these toxin-catalyzed ADP- ribosylations. ADP-ribosylation is usually monitored by [32P] NAD labelling, but it can also be quantified by the detection on SDS/polyacrylamide gels of a band with a slightly modified electrophoretic mobility. It has been recently confirmed [25] that a small decrease in the electrophoretic mobility of Ta, detectable upon PTx action, parallels the incorporation of ADP-ribose. As for CTx-catalyzed ADP-ribosylation, we have observed a small increase in the electrophoretic mobility of modified Ta. With these shifts, we can assess the effective- ness of and distinguish between these two ADP-ribosylation processes. Conditions under which total ADP-ribosylation is achieved can thus be devised. Any modification in the coupling of the modified G protein to its activated receptor is studied by taking advantage of the original observation by Kiihn [26] that Rh*-transducin coupling suppresses the extractibility of transducin from the retinal rod membrane in the absence of detergent. The association to TBy of CTXTa or PTxTa in their various nucleotide binding states, i.e. with a bound GDP, GDP-AIF, or guanosine 5’-[y-thio]triphosphate (GTP[yS]) is ascertained by determining whether they coelute from retinal rod membranes with buffers of appropriate ionic strengths and further analyzed by gel-filtration chromatography. The activatory conformational switch associated with the GDP/ GTP exchange or the binding of fluoride complexes to GDP in the Ta nucleotide site, is monitored through the changes in intrinsic fluorescence of a tryptophan in Ta, which is characteristic of the inactive or active conformation of Ta [21, 27, 281. The functionality of the effector-contact domain is tested by determining the efficiency with which the ADP-ribosylated Ta activates PDE in reconstituted membranes.

MATERIALS and METHODS

Buffers

Solution A (isotonic buffer)

20 mM Tris/HC1, pH 7.5, 120 mM NaCI, 0.1 mM MgC12.

Solution B (hypotonic buffer)

5 mM Tris/HCl, pH 7.5, 0.1 mM MgC12. For experiments with PTx, these buffers were complement-

ed with 0.1 mM phenylmethylsulfonyl fluoride and 1 mM 2-mercaptoethanol.

CTx solution

4 mM MgCI2, 10 mM dithiothreitol, pH 7.0. 200 mM NaH2P04, 12 mM arginine, 1 mM thymidine,

35

PTx solution

15 pM ATP, 3 mM dithiothreitol. 90 mM Tris/HCl, pH 8.0, 2.6 mM MgCI2, 1 mM EDTA,

Retinal rod outer-segment membrane preparations

Bovine retinal rod outer-segment membranes (ROS) were prepared under dim red light as described [29] and stored at -80°C. Thawed pellets were resuspended in buffer and homogenized in a teflon-glass homogenizer. Unless specified otherwise, all procedures were conducted at 4 “C, in total darkness or under dim red light. Photoactivation was achieved by a I-min illumination with orange light (A > 540 nm, Kodak Wratten 21 filter) and photoregeneration of Rh* (during incubations with CTx) by continuous illumination with blue light (A,,, = 460 nm, Kodak Wratten 48 filter). Washed ROS membranes, depleted of transducin and other peripheral proteins, were obtained by three successive steps of suspension ( 5 pM rhodopsin) and sedimentation, once in solution A and twice in solution B supplemented with 3 mM EDTA, the final pellet was resuspended in solution A. Urea-washed ROS membranes were obtained by introducing a supplementary step of resuspension in 20 mM Tris/HCl and 4 M urea and sedimentation.

ADP-ribosylation of transducin on ROS membranes with CTx

CTx from Sigma (0.5 mg/ml) was first activated for 15 min at 37°C in 50 mM Tris/HCl, pH 7.5, 50 mM dithiothreitol. A ROS suspension (70 pM Rh) in CTx solution was photoactivated, then incubated at 30°C under photoregenerating blue light, with the following additions: 75 pg/ml CTx, 500 pM NAD or [32P]NAD (20- 50 Ci/mol), 100 pM guanosine 5’[P,y-methylene]triphosphate (GPP[CH,]P). Unless specified otherwise, incubation lasted 2 h and resulted in total ADP- ribosylation. After incubation, the suspension was sedimented to remove soluble ROS proteins and toxin, the pellet was resuspended in solution B (70 pM Rh) to extract PDE and sedimented. The ADP-ribosylated transducin could then be solubilised from the membrane by resuspending the pellet in solution B (70 pM Rh) and adding GDP (200 pM), GTP or GTP analogs.

Isolation of holotransducin and ADP-ribosylation with PTx

Illuminated ROS membranes were suspended (25 pM Rh) and sedimented twice in solution A (with 2 mM MgC12 in the first washing) and twice in solution B (100 pM Rh). Trans- ducin was solubilized by resuspending the washed membranes (100 pM Rh) in solution B with 200 pM GDP. PTx from Sigma (10 pg/ml) was first activated by a 30-min incubation at 37°C in 20 mM dithiothreitol. The last ROS supernatant, containing almost pure transducin (1.5 pM), was then incubated for 60 min at 30°C in PTx solution with 1 pg/ml activated PTx and 1 mM NAD, after which the sample was kept at 4°C.

Gel-filtration analysis of the association between transducin subunits

Gel filtration of various ADP-ribosylated or control samples of extracted transducin was performed by running 0.2-ml samples on a Superose 12 HR 16/30 column (Pharmacia-LKB). Elution at room temperature (approxi-

mately 23°C) was at a flow rate of 0.5 ml/min, with 20 mM Tris/HCl, pH 7.5, 120 mM NaC1, 1 mM MgC12, 1 mM 2- mercaptoethanol, 100 pM phenylmethylsulfonyl fluoride and 0.1 % Lubrol PX. The presence of Lubrol PX (or an equivalent detergent) is necessary to preserve the association of TaGDP to TPy in solution and at room temperature. This is most probably due to the hydrophobic farnesyl residue on the TPy subunit (J. Bigay, unpublished results). In the absence of Lubrol, the same gel-filtration procedure always yields well- separated Ta and TBy subunits and is used for preparative purification of the subunits of ADP-ribosylated or control transducin.

Fluorescence monitoring of transducin activation Activation by GDP/GTP exchange or by fluoride com-

plexes correlates with a large change in the intrinsic tryptophan fluorescence of the Ta subunit. As in previous work from our laboratory [21], fluorescence measurements were performed with a Shimadzu RF 5000 fluorimeter with exci- tation at 284 nm (Fig. 6) or 292 nm (Fig. 7) and emission at 340 nm (bandwidths 3 nm and 30 nm, respectively). The samples, stirred in a 10-mm x 10-mm cuvette, contained about 50 nM transducin.

PDE-activation assay The assays were performed by the pH-metric technique

[30] with partially purified PDE reconstituted on ROS membranes that had been depleted of photoexcited rhodopsin and of transducin. Illuminated ROS membranes were first suspended and sedimented in solution A to remove soluble proteins, then in solution B to release PDE which was kept for subsequent use. The transducin was solubilized upon adding 200 pM GDP in solution B. The membranes were then incubated with 100 mM hydroxylamine for 30 min at 30 “C to eliminate all the photoactivated rhodopsin. After incubation, the hydroxylamine was removed and the PDE added back under isotonic conditions; the suspensions were pelleted and kept frozen at -20°C until use. Aliquots were thawed and resuspended (3 pM Rh, approximately 15 nM PDE) in 3 ml 20mM Hepes, pH 7.5, 2 m M MgCI2, 120mM KCl, with GDP (50 pM) and when required, NaF (5 mM) and A1C13 (30 pM). The transducin to be tested was added (approximately 60 nM) and the assay was initiated upon the addition of 300 pM cGMP. The pH variation due to the cGMP hydrolysis was recorded with a combined pH electrode (Ingold, type Xerolyt). The PDE activity was deduced directly from the slope of the pH recordings.

Quantification of proteins, SDS/polyacrylamide gels, densitometry

Protein concentrations in the various extracts were mea- sufed according to Bradford [31]. Extracts were analyzed on mini or large SDSjPAGE (6 cm or 13 cm long) using 10% acrylamide and 0.32% bisacrylamide. Autoradiography was performed by exposing Dupont (cronex 4NIF 100) films. In- tensities of Coomassie-blue-stained SDS/polyacrylamide gel bands and those of developed autoradiogram bands were measured with a Hoefer GSI 300 scanner.

RESULTS Choleratoxin ADP-ribosylates transducin only when it is Rh*-bound and nucleotide-free

CTx-catalyzed ADP-ribosylation of transducin is maximal when the ROS membranes are incubated under continuous

36

illumination with 10 - 100 FM guanosine 5'-[P,y-imido] triphosphate GPP[NH]P, or better GPP[CH,]P, but not with GTP or GTP[yS] [5 - 71. We recently elucidated this curious requirement for the less efficient of the non-hydrolysable GTP analogs [14]: These nucleotide analogs have an activatory effect on the ADP-ribosylating factor(s), ARF, to which they bind with high affinity, but an inhibitory effect on the ADP- ribosylation substrate, transducin, to which they bind with lower affinities. By reconstitution experiments we demonstrated that transducin is the substrate of CTx and activated ARF, only when it is bound to photoexcited rhodopsin and depleted of nucleotide. This accounts for the requirement for illumination and for the inhibitory effect of GTP and GTP[yS], which activate transducin and dissociate it from Rh* too rapidly. In our previous work, maximal ADP-ribosylation was obtained upon activating ARF with small amounts of GTP[yS] in transducin-depleted membranes. GTP[yS] was then removed before transducin was added back and ADP- ribosylated by CTx (and NAD) in the presence of low concentrations of GPP[NH]P or GPP[CH,]P. This sequential procedure is informative but complex and results inevitably in significant loss of proteins in the repeated washings. In the present work, high ADP-ribosylation yields were obtained upon a single incubation of the illuminated native membranes with CTx, NAD and 100 pM GPP[CH2]P. The membranes were then sedimented to eliminate the GPP[CH2]P without losing transducin, since this GTP analog has a very low affinity for Ta and does not significantly dissociate the G protein from its high-affinity coupling to Rh*. As discussed below, the ADP-ribosylated transducin could then be eluted from the membrane, upon addition of micromolar amounts of GTP[yS], which releases CTxTaGTP[yS] as permanently dissociated from Tpy, or upon addition of GTP which releases CTxTaGTP which decays slowly to CTxTaGDP and reassociates with TPy or even, upon addition of high concentrations of GDP which directly gives inactive and undissociated CT"TaGDP-Tjly.

CTx-catalyzed ADP-ribosylation induces an electrophoretic-mobility shift in Ta

ADP-ribosylation is usually monitored by radioactivity measurements of the radiolabelled Ta bands on SDS/ polyacrylamide electrophoretic gels after using 32P-labelled NAD. However, this technique does not permit assessment of the ADP-ribosylation yield, as the unlabelled transducin pool is not visualised. We observed, on Coomassie-blue-stained gels, that incubation of the membranes with CTx, GPP[CH2]P and unlabelled NAD induces the progressive conversion of the subsequently extracted Ta to a form that shows a slightly increased electrophoretic mobility on SDS/polyacrylamide gels (Fig. 1). The conversion was practically complete after 1 h of incubation. Upon more prolonged incubation, another band with a still higher electrophoretic mobility could be detected. The appearance of the shifted bands was strictly dependent on the presence of both toxin and NAD in the incubation mixture; with radioactive NAD, only the shifted bands were radiolabelled. The specific radioactivity of the first shifted band, as estimated from scans of the autoradiogram and of the Coomassie-blue-stained gel, was independent of the incubation time, suggesting that this shift correlates with a single ADP-ribosylation of Ta. For the second shifted band, which was observed only after lengthy incubations with radioactive NAD, the specific radioactivity was 2.2 f 0.4-fold higher, suggesting that this band contained doubly ADP- ribosylated Ta. To further characterize this second band,

A B

c *Ta +**TO:

TP - c *Ta --**TlX

t *Ta t * * T a

+ *Ta --**TlX

a b c a b c d

Fig. 1. Modifications of the electrophoretic mobility of Tcc upon ADP- ribosylation by CTx. (A) Aliquot samples of ROS membranes were incubated, under blue light, at 30°C in the CTx ADP-ribosylation medium with "P-NAD (see Materials and Methods) for periods of (a) 0 min, (b) 15 min, (c) 60 min and (d) 150 min. The samples were cooled down to 4"C, sedimented and transducin was eluted in solution B supplemented with GTP[yS] (see Materials and Methods) and sub- mitted to SDS/PAGE. The gels were Coomassie-blue stained (upper panel) and autoradiographed (lower panel). One sees, in the upper panel, the progressive shift of the position of the Ta band to that of *Ta, and on the lower panel the corresponding progressive radioactive labelling of the ADP-ribosylated *Ta. In the last lane, a second shifted band, **Tcc, appears, which has a significantly higher specific radioactivity, suggesting a higher degree of ADP-ribosylation. (B) Three samples were incubated in the CTx ADP-ribosylation medium under the same conditions, each for a total time of 150 min, but with (a) unlabelled NAD for 150 min, (b) unlabelled NAD for the first 60 min and [3ZP]NAD for the following 90min and (c) [3ZP]NAD for 150 min. After the incubation period, transducin was eluted in solution B at 4°C with GTP[yS] (see Materials and Methods) and submit- ted to SDS/PAGE. In the upper panel, all three samples are fully ADP-ribosylated, but in lane b only the **Ta band shows radioactive labelling. Thus, only **Ta and no *Tcc was produced in the second incubation period, after the first incubation of 60 min with unlabelled NAD, which was sufficient to fully mono ADP-ribosylatc *Ta (see A, lane c). This further suggests that the **Ta band corresponds to a multiple ADP-ribosylation, most probably a double ADP-ribosylation, as on lane c the specific radioactivity of **Ta was 2.2 & 0.4-times that of *Ta.

ADP-ribosylation by CTx was performed in two successive incubations, starting with unlabelled NAD, and adding the 32P-NAD only in a second phase (Fig. lB, lane b). Radioac- tivity was then found exclusively in the second band, confirming that it resulted from a secondary ADP-ribosylation, following the first modification with unlabelled NAD. These experiments allowed us to control the yield of ADP-ribosylation by CTx, and to define conditions under which the quasi totality of the transducin pool is mono ADP-ribosylated, as seen for example in Fig. IA, lane c.

CTx-catalyzed ADP-ribosylation does not activate Rh*-bound transducin and upon GDP addition, inactive CTxTaGDP-Tpy is released

We have previously shown [I41 that transducin which has been ADP-ribosylated on illuminated membranes and without nucleotides does not dissociate from Rh* unless some guanosine nucleotide is subsequently added. After completion of the ADP-ribosylation process in the presence of activated ADP-ribosylating factors but in the absence of free nucleotides, the addition of GDP at concentrations higher than 20 pM induces the dissociation of the ADP-ribosylated transducin from Rh*, as it does for unmodified transducin bound to Rh*. However, in solution A, the ADP-ribosylated and GDP-bearing transducin remains membrane bound, as

37

Native membranes

extracts

m - +

Ta + TP -

Reconstituted membranes extracts

ISO * Hypo - HYPO + GTPF

--m - + - + - +

Ta +

a b c d e f g h Fig. 2. Nucleotide and illumination dependence of solubilisation of CTx- ADP-ribosylated transducin. Transducin in ROS membranes was ADP-ribosylated by a 120-min incubation at 3 0 T , under blue light in the CTx-ADP-ribosylation medium (see Materials and Methods), with [3ZP]NAD and CTx (without CTx for control). The ADP- ribosylated, but still Rh*-bound transducin, could be eluted from the membrane pellet, as well as transducin in the control sample, by sedimenting in solution B after addition of 200 pM GDP (lanes a and b). Both supernatants were readjusted to isotonic ionic strength and incubated with transducin-depleted non-illuminated ROS membranes (see Materials and Methods). cTxTd3DP-TBy bound to the reconstituted membranes as well as the control native TctGDP-Tjy, as checked by sedimentation (transducin-free supernatants c and d). After illumination of the membrane pellet and elimination of the GDP, the ADP- ribosylated (as well as control) transducin became resistant to elution in low ionic-strength buffer (transducin-free supernatants e and f). This is characteristic of a high-affinity binding to Rh* in the membrane. Upon addition of GTPbS] the ADP-ribosylated as well as control transducin was released from the membrane, that is from Rh* (supernatants g, and h). This demonstrates the Rh*-catalysed GTP[yS] binding and activation of the ADP-ribosylated transducin. Upper panel, Coomassie-blue-stained gels; lower panel, corresponding autoradiographs.

would unmodified TaGDP [29]. The dissociation of CTxTctGDP-T/ly from Rh* can be revealed by sedimenting the membrane in solution B in which, as for unmodified TaGDP- Tpy, both the CTxTaGDP and TPy subunits elute together, as shown in Fig. 2. The association of these subunits in the solubilised extract was analyzed by gel filtration in the presence of 0.1 % Lubrol (see Materials and Methods). It has been found that the presence of a small amount of Lubrol, or any equivalent detergent, is required to preserve the association of TaGDP with Tpy in non ADP-ribosylated, control samples. This effect of a detergent on the association of the native, unmodified but solubilised subunits of transducin has been systematically studied (J. Bigay, unpublished results) and seems to be related to the presence on the TPy subunit of a farnesyl residue which, when this TPy subunit is detached from the membrane, interferes with its association to Ta [32, 331. In the presence of Lubrol, the two subunits of the ADP- ribosylated transducin eluted together, as a single CTxT~GDP- TPy complex of the same apparent size as unmodified TctGDP- TPy (Fig. 3). The activity of this ADP-ribosylated holotransducin was checked by PDE activation assays on reconstituted ROS membrane (see Materials and Methods); CTxTaGDP-TPy is as incapable of effector activation as the unmodified TaGDP-Tby (Fig. 4).

Inactive CTxT~GDP-TBy couples to Rh* and undergoes Rh*-catalysed nucleotide exchange, like unmodified transducin

In its interactions with the membrane and with the receptor, CT"TaGDP-Tjly does not seem to differ from unmodified TaGDP-TPy ; in solution A the modified protein binds weakly to non illuminated ROS membranes, as native TaGDP-TPy. If the membranes are illuminated, and no nucleotides are added, the binding of CTxTctGDP-TPy becomes resistant to washing in solution B, a characteristics of high-affinity interaction with Rh* (Fig. 2). Upon addition of GTP or GTP[yS], the modified transducin is instantly released (Fig. 2), which indicates that, like unmodified transducin, it has undergone a rapid Rh*-catalyzed nucleotide exchange.

Addition of GTP[iS] to CTx-ADP-ribosylated and Rh*-bound transducin releases permanently active CTxT~GTP[yS]

After completion of the ADP-ribosylation of transducin by CTx on illuminated membranes, the addition of a few micromoles of GTP[yS] induces the rapid dissociation of CTxTaGTP[yS] from Rh* and from TPy. We showed previously [14] that, upon sedimentation of the membranes in solution A, whose ionic strength and composition approximate those in the cytoplasm, only CTxTaGTP[yS] is solubilised from the membrane while TPy remains membrane-bound. This is a characteristics of the activated state of Ta [26]. The conformational change responsible for the dissociation and solubilisation of CTxTaGTP[yS] thus depends on the occu- pancy of the nucleotide site by the GTP analog rather than on the CTx-catalyzed ADP-ribosylation. If the sedimentation medium is the low-ionic-strength solution B, both transducin subunits are eluted from the membrane. Gel filtration studies demonstrated that the solubilized CTxTaGTP[yS] is permanently dissociated from the solubilised TPy, as observed under the same conditions for unmodified TaGTP[yS] (Figs 3 and 5). cGMP hydrolysis assays with PDE on reconstituted membranes (see Materials and Methods) showed that CTXTaGTP[yS] activates the PDE with the same efficiency as does unmodified TaGTP[yS] (data not shown).

Active CTxT~GTP, released from Rh* by GTP addition after ADP-ribosylation, slowly hydrolyses its GTP

The addition of GTP after ADP-ribosylation of transducin by CTx on illuminated ROS membranes in the absence of nucleotide also induces the dissociation and activation of the modified transducin [14]. If the membranes are sedimented in solution B before the exhaustion of GTP by the GTPase activity, the ADP-ribosylated Ta subunits are specifically eluted as with GTP[yS] [14]. With a special stopped-flow technique to be described elsewhere (B. Antonny, M. Chabre and T. M. Vuong, unpublished results), CTxTaGTP was rapidly eluted from ROS membranes retained on a poly(viny1idene difluoride) filter and injected into a fluorescence cuvette. In- trinsic GTP hydrolysis by the thus isolated CTxTaGTP was time-resolved by recording the decrease in intrinsic tryptophan fluorecence associated with the GTP - GDP conformational change. A value k,,, of 5.10-3 s-' was measured for GTP hydrolysis in CTxTaGTP, as compared to k,,, of 4.8 l op2 s-' for isolated TaGTP under the same conditions (25°C) . Thus CTxTaGTP hydrolyses its GTP 10-times slower than does TaGTP.

2 r A

30 35 40 45 50

3.0 r

2.5

2.0

1.5

1 .o

0.5

0 30 35 40 45 50

Fraction number Fig. 3. Gel-filtration-chromatography analysis of the association of CTxTa with TPy subunits after ADP-ribosylation and release from Rh* by GDP, GTP or GTP[yS]. Aliquots of ROS membranes were submit- ted to the CTx ADP-ribosylation procedure as described in Materials and Methods, with [32P]NAD and CTx, or no CTx for controls. The membrane pellets were eluted in low-ionic-strength buffer supplemented with 300 pM GDP (-), 300pM GTP (-) or 100 pM GTP[yS] (- - - - -). The extracts were concentrated/diluted three times on Centricon 30 (Amicon) in the gel-filtration buffer containing 0.1% Lubrol PX (see Materials and Methods). This removed the free nucleotide excess and raised the final concentration of transducin to 2 pM. 150 p1 samples were loaded on the gel-filtration column at room temperature, eluted at the rate of 0.5 ml/min. (A) Elution profile as monitored by ultraviolet absorbance at 280 nM. (B) [32P]NAD labelling profile obtained by Cerenkov counting of 0.3-ml fractions. The elution profiles are identical for the samples extracted with GDP (m) or with GTP (+) and correspond to associated heterotrimeric complexes (see Fig. 5). With GTP[yS] ( O ) , the elution profile corresponds to dissociated TGC and TJy subunits (see Fig. 5).

After GTP hydrolysis, CTxTaGDP reassociates with TPy and is inactive

After ADP-ribosylation by CTx on illuminated ROS membranes and activation by GTP in solution B, both CTxTaGTP and TPy elute from the membrane but most likely as separate entities. They stay apart until CTxT~GTP hydrolyses its bound GTP. When enough time was allowed for the hydrolysis (at room temperature) of the bound GTP in the modified Ta to reach completion, the two subunits co-elute from the gel- filtration column as a complex of the same apparent size as that of unmodified TaGDP-TPy (Fig. 5) . This indicates that, after the hydrolysis of its bound GTP, CTxTaGDP reassociates

100

50

+ 5 O 0

x t ,-

- AlFx + AlFx

Fig.4. Dependence on fluoride complexes of PDE activation by CT"TaGDP-TBy or by PT"TaGDP-T/?y. The measurements were performed by the pH-metric technique (see Materials and Methods), with partially purified PDE reconstituted on ROS membranes that had been depleted of photoexcited rhodopsin and endogenous transducin (see Materials and Methods). Transducin was ADP-ribosylated by CTx on illuminated ROS membranes and eluted with 200 pM GDP, to obtain CTxTd3DP-TJy, or was eluted by GDP from washed and illuminated ROS membranes, and ADP-ribosylated by PTx in solution, to obtain PT"T~GDP-TJy, or incubated in the absence of toxin for control T?wCDP-TJy. The ADP-ribosylated or control transducin was added to a suspension of reconstituted membranes (3 pM Rh, 15 nM PDE, 60 nM trdnsducin) in 20 mM Hepes, pH 7.5,2 mM MgC12, 120 mM KCI, 50 WM GDP. For fluoride activation, 5 mM NaF and 30pM AICI3 were added. The pH-variation rates that monitored the PDE activity were recorded upon the addition of 300 pM cGMP.

immediately with TPy. PDE-activity measurements confirmed that this reassociated CTxTaGDP-TPy is inactive (Fig. 4).

CTxTaGDP-TPy is activated by fluoride complexes, but the apparent affinity and activation kinetics differ from those for unmodified transducin

Native holotransducin TaGDP-Tjy can be activated by fluoride complexes of aluminium or beryllium [20] or even magnesium [21], which bind into the nucleotide site of Ta next to the GDP and simulate the presence and action of the y phosphate of a bound GTP. The fluoride-activated species, denoted TaGDP-AlF, without further specifying the nature of the fluoride complex, has all the structural and functional characteristics of TaGTP[yS]; it dissociates from Rh* and from TPy, is solubilised from the membrane in solution A and can activate the PDE. After ADP-ribosylation by CTx on illuminated ROS membranes, transducin that was eluted as inactive CTxTaGDP-TPy and incubated in solution with NaF and A1C13, activated the PDE with an efficiency comparable to that observed under the same conditions for TaGDP-Tby (Fig. 4). Fluoride dose/response curves and the K , values for these PDE activations are, however, difficult to determine, due to the multiplicity of the metallofluoride complexes, and

39

the difficulty in accurately evaluating the fraction of activatory complexes when varying the concentration of one of the component ions, F-, A13' or Mg2+ [22]. When fluoride activation was attempted on ADP-ribosylated transducin still bound to Rh* and to TPy on illuminated ROS membrane, and tested by the solubilization assay, it seemed much less efficient than on unmodified transducin under the same conditions; while unmodified Ta was almost quantitatively released from illuminated ROS membrane in solution A upon incubation with 100 pM GDP and the standard NaF plus AlC13 mixture, CTxTa remained completely membrane bound under the same conditions. This suggests that the affinity of the fluoride complexes for the nucleotide site in Ta is modified by ADP-ribosylation. We therefore studied the kinetics of Ta activation by fluoride complexes in solubilized and purified CTxTctGDP. Activation of transducin in solution can be monitored with time through a tryptophan fluorescence change that correlates with the conformational switch between the inactive (GDP-bound) and active (GTP-bound) states. This fluorescence change is observed upon metallofluoride binding, as well as with GDPIGTP exchange [21, 27, 281. Addition of fluoride complexes to CT"TaGDP elicited fluorescence changes of maximal amplitude close to that observed for unmodified TaGDP, confirming that full activation can be obtained. The rate constant for fluoride-complex binding and ensuing activation can be deduced directly from the initial slope of the tryptophan fluorescence change (Fig. 6). This on- rate for the activation by aluminofluoride complexes is barely modified by CTx-catalyzed ADP-ribosylation. The deactivation rate constant was measured upon chelation of the aluminium once activation had reached saturation; as compared with TaGDP, the off-rate is found to increase by more than 10-fold in CT"TaGDP (Fig. 6). Similarly, for an activation by fluoride and magnesium in the absence of aluminium [21], the on-rate is not modified by ADP-ribosylation while the off- rate is increased by a factor of four (data not shown). In both cases, the affinity of the GDP-containing site for the

Fig. 5. Gel-filtration chromatography and SDSlPAGE analysis of the Ta and Tjy subunit association after ADP-ribosylation by CTx and release by GTP or GTP[yS]. For the gel-filtration procedures, see Fig. 3 . After Cerenkov counting, the 0.3-ml fractions were concentrated under vacuum and complemented with denaturing buffer for SDSjPAGE analysis. The Coomassie-blue-stained gels were scanned to quantify Ta (0) and TP (0). (A) [32P]-labelled Ta and the TP subunits of the ADP-ribosylated transducin extracted with GTP are seen to elute together as single complex with an apparent size of 80 kDa, as, in (B) for the control of unlabelled transducin extracted with GTP. Identical patterns are obtained when the ADP-ribosylated transducin has been released by adding GDP instead of GTP (see Fig. 3 , SDS/polyacryhmide gel not shown). (C) [32P]-labelled Ta and T/3 subunits of the ADP-ribosylated transducin extracted with GTP[yS] elute separately, with apparent sizes of 40 kDa and 50 kDa, respectively, as in the control (D) of unlabelled transducin extracted with GTP[yS]. Thus, upon ADP-ribosylation and binding of GTP[yS], CT"TaGTP[yS] dissociates permanently from TPy (C), as does unmodified TaGTP[yS] (D), but with GTP, at room temperature, the dissociated CT"TaGTP, hydrolyses (albeit slowly) its GTP, and reassociates with TPy, forming a CTxTaGDP-T/3y complex (A), as does unmodified TctGTP, which has rapidly hydrolyzed its bound GTP (B). BSA, bovine serum albumin; OVA, ovalbumin; C. Anh., carbonic anhydrase.

y-phosphate analog is thus considerably decreased in the ADP-ribosylated protein.

CTxT~GDP-Tfly can be activated by receptor-independent exchange of GDP for GTP(yS], with the same kinetics as for unmodified transducin

Like other G proteins, in the presence of magnesium at near physiological concentration (i.e. around 1 mM) and in the absence of activated receptor, transducin exchanges spon- taneously, albeit very slowly, its bound GDP for another guanine nucleotide. The exchange rate is limited by the off- rate of the bound GDP. In the presence of GTP[yS] and absence of a large excess of free GDP, the rate of GDPi

(kDa) 60 45 31 I I I

BSA OVA C.Anh.

T a + a-

0

1

0.5

0

T a - TP-, 1

3 o 0.5 0 7

x

0

TP -D

1

TCL +

1

C 0 .- c

0 : a E

2

0.5 1

0 O 35 40 4s

Fraction number

B

EDTA-Mg 5 mM

control

u 200 s

Fig. 6. Fluorescence analysis of the kinetics of binding and dissociation of fluoride complexes to CT"TaGDP. Transducin was ADP-ribosylated on ROS membranes with unlabelled NAD and CTx, or no CTx for control, and was eluted from the membrane in solution B with 200 FM GDP. Ta was resolved from Tjy by gel filtration (see Materials and Methods), and diluted to 30 nM in 1.2 ml buffer (20 mM Tris, pH 7.5, 1 mM MgCI2, 120 mM NaC1, 1 mM dithiothreitol) in a fluorescence cuvette. The tryptophan fluorescence changes induced by fluoride complexes wcre monitored as described in Materials and Methods. For the activation kinetics (A) 2 mM NaF and 2 pM A1CI3 were injected when indicated after starting the measurement on the inactive Ta. For the deactivation kinetics (B), 2 mM NaF and 27 pM AIClj had been added to the Ta solution before starting the measurement to saturate Ta activation, and 5 mM MgEDTA was injected as indicated to initiate deactivation, by chelating the aluminium as performed in previous works [21]. The kinetic constants can be obtained from exponential fits of the activation and deactivation curves, but can also be evaluated from the initial slope of thc curves, provided one knows the maximal amplitude of the fluorescence changes, not measurable on the early part of the curve which is only shown here. In (A), the initial slopes (related to k,, rates) are comparable in the CTx (CTX) and control samples, but the fast saturation at a low level of the fluorescence of the CTx sample suggests that the corrcsponding kOff rate is much higher than in the control. This is confirmed in (B), where the initial slope of the CTx sample is 10-times higher than that of the control (both samples will rcach the same final low fluorescence level; the small step visible for the control is due to the sample dilution upon the injection). From the analysis of the complete data, we obtained k,, = 4 x lo3 M-'s-' , koff = 3 x s - ' for the control TaGDP and k,, = 3 x lo3 M-ls-' ,

kOff = 40 x s - l for CT"TaGDP. Thus, the apparent affinity of the aluminofluoride complex for the y-phosphate site of the nucleotide in TaGDP is reduced about 10 times by the ADP-ribosylation.

GTP[yS] exchange in TaGDP can also be monitored by the associated change in intrinsic tryptophan fluorescence. Purified native TaGDP and modified CTxTaGDP were incubated in the fluorimeter in the presence of 10 pM GTP[yS] at 25 "C. The fluorescence increase that monitors the exchange of GDP for GTP[yS] was recorded for 1 h, after which saturating concentrations ofNaF and A1Cl3 were added to rapidly obtain a saturating fluorescence level which is identical for TaGTP[yS] and TaGDP-AlF,. The absolute rate of GDP/ GTP[yS] exchange could then be deduced from the normalised slope of the fluorescence increase (Fig. 7) and compared to that observed under the same conditions for native TaGDP. The rate of 2.4 x s-l at 25°C in the control sample is not significantly increased (by less than 10%) in the ADP- ribosylated sample. Thus CTx-catalyzed ADP-ribosylation does not perturb the spontaneous GDP off-rate which governs the GDP/GTP[yS] exchange rate.

t GTPyS 10 pM

Fig. 7. Fluorescence analysis of the kinetics of receptor-independent GDP/GTP[yS] exchange in CT"TaGDP. The samples and the fluorescence-measurement procedure are identical to that in Fig. 6. As the fluorescence increase seen after the GTP[yS] injection was extremely slow, NaF and A1C13 were injected 1 h later to rapidly reach the saturation level, which is identical for TaGTP[yS] and TaGDP-AIF,. The fluorescence increase that monitors the GDP/GTP[yS] exchange had reached only about 10% of the saturated amplitude over the first hour. The GDP/GTP[yS] exchange rates were then deduced directly from the normalised initial slope of the fluorescence increase observed during this first hour. This rate was of 2.4 x s-' at 25°C in the control sample (thin line), and not significantly increased (by less than 10%) in the ADP-ribosylated sample (thick line).

PTx-catalyzed ADP-ribosylation of TaGDP requires the presence of T/?y; PTxTaGDP does not dissociate from T/?y

ADP-ribosylation of Ta by PTx can be performed on solubilised transducin in the absence of membrane. It does not require any ADP-ribosylating factor, but Ta is substrate only when in the inactive TaGDP form and associated to TPy [18]. The active and dissociated form TaGTP[yS] is not a substrate, even in the presence of excess TBy [6]. The strict dependence of the PTx action on the presence of TPy was verified (not shown). We confirmed that, as a catalytic cofactor, TPy is not required in stoichiometric amounts with respect to the Ta pool to allow for its complete ADP-ribosylation [19]. However, we observed that when TPy is present only in sub-stoichiometric amounts, the ADP-ribosylation kinetics are drastically slowed down and depend linearly on the TPy concentration. This suggests that TPy is a slow ADP-ribosylation catalyst which does not rapidly dissociate from TrCDP once ADP-ribosylation is completed : By the gel-filtration technique, PTxTaGDP and TPy co-eluted from the column as a complex of the same apparent size as that of unmodified TaGDP-TPy (Fig. 8).

PTx-catalyzed ADP-ribosylation induces an electrophoretic mobility shift in Ta

After some controversy [35], it has been recently confirmed [25] that ADP-ribosylation by PTx induces a shift in the electrophoretic mobility of Ta on SDS/polyacrylamide gels. PTxTa migrates slower than native Ta. This provides for a direct assay of the ADP-ribosylation efficiency and helped us define the conditions under which Ta can be fully ADP- ribosylated by PTx on native ROS membranes (Fig. 9).

41

(kDa) 60 45 31 I I I

E ~ A O ~ A C.Anh.

" A

7 4 x F A

0 35 40 45

Fraction number

Fig. 8. Gel-filtration-chromatography analysis of the association of PT"T~GDP with Tby and of its dissociation by fluoride complexes. Transducin was ADP-ribosylated by PTx with [3ZP]NAD as described in Materials and Methods. The procedures for gel-filtration chromatography were identical to that described for CTx ADP- ribosylated transducin in Fig. 3 . The gel-filtration column was run in the presence (B) of 5 mM NaF and 30 pM AlC13. The fractions were counted in a Cerenkov counter, then concentrated under vacuum and complemented with denaturing buffer for SDSjPAGE analysis, as described in Fig. 3, but only the ["PI counts are shown here (0) . In (A), the [32P]NAD is seen to elute in the fractions that correspond to an 80-kDa Ta-Tpy complex (see Fig. 5A), confirming the PT"TrGop- TPy association; in (B) the ["PI elutes in fractions that correspond to a 40-kDa, isolated Ta subunit, (see Fig. 5C), confirming the dissociation of PT"T~GDP-AIF, from Tpy. BSA, bovine serum albumin; OVA, ovalbumin; C. Anh., carbonic anhydrase.

PTxTaGDP-Tfiy binds to the membrane, but not to Rh" which cannot catalyze its activation by nucleotide exchange

The PTx-catalyzed ADP-ribosylation of a G protein uncouples it from the activated receptor. The site of ADP- ribosylation by PTx must therefore be in the receptor-contact domain. Since binding to the activated receptor is prevented, receptor-catalyzed GDP/GTP exchange is no longer possible. The receptor - G-protein interaction can be most easily demonstrated in the retinal system by the tight binding (i.e. resistant to washing in solution B) to illuminated ROS membranes of transducin as opposed to its total solubilisation from non- illuminated membranes under the same conditions [26]. ROS membranes, depleted of endogenous transducin, were illuminated and incubated with an equimolar mixture of PTxTaGDP- TPy and of control TaGDP-TPy (Fig. 9). Upon sedimentation in solution B, control transducin was totally retained in the membrane pellet, while all the ADP-ribosylated transducin, identified by its altered electrophoretic mobility, remained in the supernatant. The binding of TaGDP-Tpy to the membrane is strictly dependent on the illumination of the ROS and results from the formation of high affinity Rh*-TaCmpty -TPy complexes which are stabilized when Rh* has induced the release of GDP from Ta [36]. These complexes can be fully dissociated upon addition of GTP[yS]. 'The fact that PTXTaGDP-Tfly was not retained on the illuminated ROS membrane confirms that it cannot form a high-affinity complex with Rh*. This suggests that Rh* cannot open the nucleotide site in PTXTctGDP and hence cannot catalyse its

I I I I I I I

a b c d e f g h i

Fig. 9. Evidence for the lack of high-affinity coupling of PT"TaGDP- Tby to Rh* and for the persistence of its ionic-dependent binding to the ROS membrane. Lane b, clectrophoretic mobility of PTxTa in fully ADP-ribosylated transducin, compared to that of Ta (lane a) in a control aliquot incubated under the same conditions with NAD in the absence of toxin. Lane c, equimolecular mixtures of the ADP- ribosylated and control transducin samples. An aliquot of the mixed sample was incubated with washed ROS membranes in solution A (50 pM Rh). The pellet was illuminated, resuspended in solution B and sedimented; the supernatant (lane d) contains only the ADP- ribosylated transducin which was thus not bound lo Rh*. The pellet was resuspended in solution B supplemented with 100 pM GTP[yS] and sedimented; the supernatant (lane e) contains the non-ADP- ribosylated transducin which had bound to Rh* and was quantitatively released in solution B upon binding GTP[yS], as controlled after a second sedimentation with GTP[yS] (lane f, in which no more transducin is released. A second aliquot of the mixed sample was added to a new aliquot of washed ROS membranes in solution A and sedimented. The pellet was illuminated, resuspended in solution A and sedimented. No transducin is found in the supernatant (lane g). Thus, in solution A, the ADP-ribosylated transducin is also retained on the membrane (compare lane g to lane d), but this binding is not released upon the addition of GTPyS, which releases only (but not totally in solution A) the non ADP-ribosylated transducin (lane h). The ADP-ribosylated transducin is finally released with the residual control transducin upon a last sedimentation in solution B (lane i). Thus, the low ionic strength of solution B, rather than the GTP[yS], caused the release of PT"Ta from the membrane, rather than from a nucleotide-dependent coupling to Rh*.

dissociation and activation by GTP or GTP[yS]. A second experiment further proves that the lack of coupling of the ADP-ribosylated transducin to Rh* was not due to a lack of rhodopsin-independent binding of PTxTaGDP-Tpy to the membrane, that would cause its early release from the membrane before it could couple to Rh". The mixture of ADP- ribosylated and control transducin was incubated with transducin-depleted and non-illuminated ROS membrane in solution A. Under such ionic conditions, TaGDP-Tpy is known to bind even to non-illuminated membrane, from which activated and dissociated TaGTP[yS] is released [29]. Upon sedimentation, the ADP-ribosylated transducin was bound to the unilluminated membrane as much as the control transducin. However, after illumination, addition of GTP[yS] and sedimentation in solution A, only TaGTP[yS] from the control transducin was released into the supernatant. The ADP- ribosylated transducin could be only eluted by a subsequent sedimentation in solution B and appeared in the supernatant as a still-associated and inactive PTXTaGDP-Tpy complex.

'*"TaGDP is activated by fluoride, with the same kinetics as for TaGDP

Activation by fluoride complexes was demonstrated by the dissociation of PTxT~GDP-AIF, from TPy and its elution from non illuminated ROS membrane in solution A upon the addition of NaF and AIC13 (Fig. 10). The non-illuminated ROS membrane was depleted of native transducin beforehand and supplemented with PTxTaGDP-T/ly. The elution pattern of PTXTaGDP-AIF, is identical to that of TaGDP-AIF,. The kinetics of activation by NaF and AIC13 of PTxTaGDP in solution

42

m T a T a =k

TP - Fig. 10. Evidences for the activation of membrane-bound PTxTaGDP- TPy by aluminofluoride. Aliquot samples of solubilised PT”TctGDP- T/3y and control unmodified TaGDP-Tby were incubated with urea washed non illuminated ROS membranes in solution A. After sedimentation, the pellets were resuspended in solution A and incubated for 5 min with or without the fluoride mixture ( 5 mM NaF, 30 FM A1C13). All samples were sedimented simultaneously and the supernatants were analyzed on SDS/polyacrylamide gels. In the absence of fluoride, PT”TctGDP-T/3y remains mostly membrane bound in solution A, as does the unmodified transducin, and upon incubation with fluoride, PTxTaGDP-AIF, is specifically released from the non-illuminated ROS membrane, as is unmodified TaGDP-AlF,.

were studied by monitoring the tryptophan fluorescence as discussed above (Fig. 6) for CTxTcrGDP. The activation and deactivation kinetics of PTxTaGDP do not differ from those of TaGDP (data not shown). This suggests that, unlike ADP- ribosylation by CTx, PTx-catalysed ADP-ribosylation of Tcc does not perturb the binding of a y-phosphate analog to the GDP site. Gel-filtration assays confirmed (Fig. 8) that fluoride activation induces the dissociation of PT”TaGDP-AIF, from TBy, as for an unmodified transducin. By the pH-metric technique, PTxTaGDP-AIF, activates the PDE as efficiently as fluoride-activated, unmodified TaGDP (Fig. 4).

PTx-catalyzed ADP-ribosylation does not modify the receptor- independent exchange rate of GDP for GTP[yS]

The GDP/GTP exchange rate was monitored by fluorescence exactly as with CTx-catalyzed ADP-ribosylation (see Fig. 7). No measurable difference on the spontaneous GDPj GTP[yS] exchange rate was detected upon PTx-catalyzed ADP-ribosylation (data not shown). This confirms the recent finding of Ramdas et al. [25].

FTx-catalyzed ADP-ribosylation of CTxT~GDP-T/?y is hindered

Transducin was first ADP-ribosylated by CTx and [32P] NAD in the absence of free nucleotide, on illuminated ROS membranes and released from Rh* by addition of GDP. The solubilised CTxTaGDP-TPy was subjected to PTx ADP- ribosylation, with unlabelled NAD. The PTx ADP-ribosylation procedure fully ADP-ribosylated a control sample of unmodified TaGDP-TBy, as checked from the complete shift of the T a band on an SDS/polyacrylamide gel (Fig. 9). How- ever, under the same conditions, only a small proportion of the 32P-labelled Ta band modified by CTx was shifted by the PTx activity (not shown). Inverting the unlabelled and radioactive NAD labelling in the successive CTx and PTx ADP-ribosylation steps further confirmed that once it has been ADP-ribosylated by CTx, transducin, even in the inactive GDP-bound conformation, and although still associated with TBy, is not a good substrate for PTx.

DISCUSSION Their sensitivities to PTx-catalyzed or CTx-catalyzed

ADP-ribosylation are often considered merely as differential

markers for the various heterotrimeric G-protein subtypes. Sensitivity to PTx is indeed directly conditioned by the presence, in a particular G protein, of the characteristic cysteine- bearing subterminal, but we have seen that it is also specific of the ‘inactive’ GDP-bound conformational state. It is not clear whether sensitivity to CTx can be related to the presence of a specific ‘consensus sequence’ in a G-protein subtype, although it seems specific of the transition-state conformation that the G protein takes when it couples to its activated receptor in order to release its GDP and switch from the inactive GaGDP to the active GaGTP conformation. This was already known to be the case for Gi [lo- 121; we have shown it to be true for transducin, and preliminary evidence (F. Bornancin, Y. Audigier and M. Chabre, unpublished results) suggests that G, also becomes a better substrate for CTx when coupled to its agonist-liganded receptor. This may therefore be a general property of all heterotrimeric G proteins. The different sensitivities to CTx observed in various cell preparations may reflect differences in the distribution of the activation state of the G protein, rather than differences in G-protein sequences. As markers, sensitivities to the ADP-ribosylating toxins relate more to the G-protein conformational state than to their primary structures. It is remarkable in this respect that the two toxin-induced modifications have opposite requirements with respect to the high-affinity coupling of the G protein to the agonist-activated receptor. High-affinity coupling is needed for an ADP-ribosylation by CTx, but is inhibited by ADP-ribosylation by PTx.

ADP-ribosylation by choleratoxin

This is often assumed to ‘activate’ G proteins, that is to trigger or increase the interaction with their natural effectors, e.g. adenylate cyclase for G, and PDE for transducin. Our results suggest that this is not strictly true; the ADP-ribosylation by CTx of transducin, most probably on Arg174, does not by itself confer an active conformation to the protein. We conclusively demonstrate that, as long as the Ta subunit keeps a GDP in its nucleotide site, this ADP-ribosylation has no observable consequence on its functional properties. The con- fusion arose from the fact that ADP-ribosylation by CTx requires the presence of GTP-activated ADP-ribosylating factors, hence it is usually performed in the presence of GTP (or non-hydrolyzable GTP analogs) and activated receptors. It is this combination of GTP and liganded receptor that is responsible for the activation of the G protein after ADP- ribosylation. If the ADP-ribosylating factors are pre-activated, ADP-ribosylation can be performed in the absence of free GTP and will not result in an activated Ta. The process depends on the formation of a ‘high-affinity’ complex between Rh* and transducin, which loses its GDP. However, once ADP-ribosylated in this complex, transducin can rebind a GDP and is released from Rh* with all of its functional characteristics remaining those of an inactive transducin. The modified subunit CTxTaGDP interacts as does native TaGDP with its three natural partners; the other G-protein subunit TBy, the receptor Rh* and the effector PDE. Exactly as with native transducin, CTxTaGDP does not bind to PDE and thus does not activate it, but it binds to TPy, and the modified heterotrimer CTxTaGDP-TBy binds to Rh* which then catalyses its activation by promoting GDPjGTP exchange. Only upon the binding of GTP do the effects of CTx-catalyzed ADP-ribosylation manifest themselves; CTxTaGTP dissociates from TPy and its activation of the PDE appears stronger than that of an unmodified TaGTP. Yet, with non-hydrolysable

43

GTP[yS], CTxTaGTP[yS] has the same activating power on the PDE as unmodified TaGTP[yS]. Thus, the increased action of CTxT~GTP, as compared to that of TaGTP, results entirely from the slower GTP hydrolysis by CTxTaGTP and does not imply any modifications in the effector site of the G protein. The activity of CT"TaGTP is prolonged rather than increased. The side chain of Arg174, which is ADP-ribosylated by CTx, most probably partakes in the binding of GTP, and the at- tached ADP-ribose must interfere with the GTP hydrolysis mechanism, reducing the k,,, for GTP hydrolysis and thus allowing a prolongation of the activatory interaction of the GTP-bound form of CTxTa with the PDE.

Our kinetic studies on the activation by fluoride complexes of CTxTaGDP further suggest that ADP-ribosylation by CTx only perturbs the nucleotide site. The fluoride complexes of aluminium or beryllium act as phosphate analogs that bind in the nucleotide site next to the P-phosphate of the resident GDP and activate the Ta subunit by simulating the presence of a GTP y-phosphate. The efficiency of activation of CTxTaGDP-TPy and of isolated CTxTaGDP by saturating concentrations of fluoride complexes is the same as with their unmodified counterparts, but the activation-rate constants (upon fluoride-complex binding) and to a larger extent that of deactivation (upon fluoride-complex dissociation), are modified in the ADP-ribosylated protein. The dissociation of a bound fluoride complex is phenomenologically similar to the release of the y-phosphate upon the hydrolysis of a bound GTP, thus this effect on the kOff for the fluoride complexes might be compared to the effect of ADP-ribosylation on the k,,, of the transducin GTPase. Both effects suggest that the modification on Arg174 perturbs the nucleotide site and more precisely one of the residues that contribute to the binding of the y-phosphate of GTP. In agreement with this conclusion, CTx-catalyzed ADP-ribosylation of TaGDP does not perturb its rate of spontaneous GDP/GTP[yS] exchange and activation in the absence of receptor. This rate is limited by the very slow dissociation of GDP from TaGDP when transducin is not coupled to an activated rhodopsin. Thus, the ADP- ribosylation by CTx of the Arg174 residue does not perturb the binding of GDP in the inactive transducin. This suggests that Arg174 does not contribute to the binding of GDP and interacts only with the y-phosphate of the bound GTP in activated transducin.

In contrast with the previous suggestion [I71 that cTxG~ctGDP remains dissociated from TPy and may stay active even after the hydrolysis of GTP, our gel-filtration data suggest that, once GTP is hydrolyzed, CTxTaGDP reassociates with TPy and is thus inactivated. The absence of a reassociation of CTXGsa with GPy in the experiments of Kahn and Gilman [I71 might have been due to the low temperature (4°C) of their sedimentation gradients, which could have sufficiently delayed the already hindered GTP hydrolysis in ADP- ribosylated G, to maintain the physical separation of CTxGsaGTP from GPy in the gradient.

ADP-ribosylation by pertussis toxin

We confirm that the toxin is only active on the TclGDP form and that it requires association with TPy. After the modification, ""TxGDP remains associated with TPy, inactive and membrane-bound but loses its capacity to bind to photoactivated rhodopsin and to undergo Rh*-catalyzed GDP/GTP exchange and activation. The effect of ADP- ribosylation of the subterminal Cys347 appears strictly local- ised to the receptor-contact domain, since the modification

does not interfere with any other functions of Ta. We confirm the recent finding of Ramdas et al. [25] that the 'activation' conformational switch occurs in PTxTaGDP upon spontaneous GDP/GTP[yS] exchange, with the same kinetics as in unmodified TaGDP and we further observe that the kinetics of activation of PTXTctGDP by fluoride complexes are also unchanged. This asserts that the nucleotide-binding site which controls the switch mechanism, and more specifically the y - phosphate binding site, is not affected by the ADP-ribose on the subterminal cysteine. The 'active' state of Ta, reached upon incubation of PT"TaGDP with saturating concentrations of fluoride complexes, has the same efficiency on PDE activity as for fluoride-activated unmodified TaGDP. Thus, the effector-contact domain of transducin appears unaffected by this ADP-ribosylation. The TPy contact domain is also untouched since, just like unmodified transducin, the association and dissociation of the Ta and TPy subunits of PTx-ADP- ribosylated transducin depend on the presence of GDP and GTP (or its analogs) in the nucleotide site, repectively.

Conclusion

The effects of ADP-ribosylation by two different toxins, CTx and PTx can be compared to those of two point mutations on distant sites of Ta that would independently perturb two different functional domains of the protein. The lack of high-resolution structural data for heterotrimeric G proteins in general and transducin in particular, aside from analogies with the known structure of the small G protein ras p21 (23), prevents a more detailed analysis of the structural conse- quences of these modifications. The observation that double ADP-ribosylation, by the sequential action of the two toxins, is much hindered, does not necessarily imply that the two modified sites interfere functionally or overlap structurally. It became obvious that ADP-ribosylation by CTx cannot pro- ceed after an ADP-ribosylation by PTx, once we recognized that the CTx action requires the formation of a high-affinity Rh*-T complex, a process blocked by the previous PTx modification of the receptor-contact domain. However, one could not predict that a first ADP-ribosylation by CTx would hinder the action of PTx on inactive, fully associated CTxTaGDP- TPy. This observation does not necessarly imply that the ADP- ribose on Arg174 interferes with the PTx ADP-ribosylation site on Cys347. It suffices, for example, to assume that the modification of Argl74 hinders the binding of PTx to CTxTaGDP-TPy.

F. B. was supported by a fellowship from the Association pour la Recherche sur le Cancer (ARC). We wish to thank Dr T. M. Vuong for his helpful comments on the manuscript.

REFERENCES

419. 1. Stryer, L. & Bourne, H. R. (1986) Annu. Rev. Cell Biol. 2, 391 -

2. Gilman, A. G. (1987) Annu. Rev. Biochem. 56,615-649. 3. Cassel, D. & Pfeuffer, T. (1978) Proc. Nut1 Acad. Sci. U S A 74,

4. Katada, T. &Ui, M. (1982) J. Biol. Chem. 257,7210-7216. 5. Abood, M. E., Hurley, J. B., Pappone, M. C., Bourne, H. R. &

Stryer, L. (1982) J . Biol. Chem. 257, 10540-10541. 6. Van Dop, C., Yamanaka, G., Steinberg, F., Sekura, R. D.,

Manclark, C. R., Stryer, L. & Bourne, H. R. (1984) J . Biol. Chem. 259, 23 - 26.

7. Navon, S. E. & Fung, B. K. K. (1984) J . Biol. Chem. 259,6686- 6693.

3307- 331 1.

44

8. Van Dop, C., Tsubokawa, M., Bourne, H. R. & Ramachandran,

9. West, R. E. Jr., Moss, J., Vaughan, M., Liu, T. & Liu, T.-Y.

10. Gierschik, P. & Jakobs, K. H. (1987) FEBS Lett. 224, 219-223. 11. h i , T., Tokhin, M., Naoki, M., Ohoka, Y., Ui, M. & Katada, T.

12. Iiri, T., Ohoka, Y., Ui, M. & Katada, T. (1991) Eur. J . Biochem.

13. Iiri, T., Ohoka, Y., Ui, M. & Katada, T. (1992) J . Biol. Chem.

14. Bornancin, F. & Chabre, M. (1991) FEBS Lett. 291,273-276. 15. Kahn, R. A. (1990) in G-proteins (Birnbaumer, L. & Iyengar, R.

16. Moss, J., Tsai, S. C., Price, S. R., Boback, D. & Vaughan, M.

17. Kahn, R. A. & Gilman, A. G.(1984) J . Biol. Chem. 259, 6235-

18. Watkins, P. A,, Burns, D. L., Kanaho, Y., Liu, T.-Y., Hewlett,

19. Casey, P. S., Graziano, M. P. & Gilman, A. G. (1989) Biochemis-

20. Bigay, J., Deterre, Ph., Pfister, C. & Chabre, M.(1985) FEBS Lett.

21. Antonny, B., Bigay, J. & Chabre, M. (1990) FEBS Lett. 268,

22. Antonny, B. & Chabre, M.(1992) J . Biol. Chenz. 267,6710-6718.

S. (1984) J. Biol. Chem. 259, 696-698.

(1985) J . Biol. Chem. 260, 14428- 14430.

(1989) J . Biol. Chem. 264,21394-21400.

202, 635-641.

267,1020-1026.

eds) pp. 201 -214, Academic Press, New York.

(1991) Methods Enzymol. 195, 243-256.

6240.

E. L. & Moss, J. (1985) J. Biol. Chem. 260, 13478-13482.

try 28, 611 -616.

191, 181-185.

277 -280.

23. Bourne, H. R., Sanders, D. A. & Mac Cornick, F. (1991) Nature

24. Berlot, C. H. &Bourne, H. R.(1992) Cell 68,911 -922. 25. Ramdas, L., Disher, R. M. & Wensel, T. G. (1991) Biochemistry

26. Kiihn, H.(1980) Nature 283, 587- 589. 27. Higashijima, T., Fergusson, K. M., Sternweis, P. C., Ross, E. M.,

Smigel, M. D. & Gilman, A. G. (1987) J . Bid. Chem. 266,

28. Philips, W. J. & Cerione, R. A. (1988) J . Biol. Chem. 263,15498- 15505.

29. Kiihn, H. (1981) in: Current topics in membrane and transport (Miller, W. H., ed.)vol. 15,pp. 171 -201, Academicpress, New York.

30. Yee, R. & Liebman, P. A. (1978) J . Biol. Chem. 252,8902-8909. 31. Bradford, M. M. (1976) Anal. Biochem. 72,248-254. 32. Fukada Y., Takao T., Ohguro H., Yoshizawa, T., Akino, T. &

Shimonishi, Y. (1990) Nature 346, 658 -660. 33. Ohguro, H., Fukada, Y., Yoshizawa, T., Saito, T. & Akino, T.

(1990) Biophys. Biochem. Res. Commun. 167,1235 - 1241. 34. Bigay, J., Deterre, Ph., Pfister, C. & Chabre, M.(1987) EMBO J.

35. Ribeiro-Neto, F. A. P. & Rodbell, M. (1989) Proc. Natl Acad.

36. Bornancin, F., Pfister, C. & Chabre, M. (1989) Eur. J . Biochem.

349,117-127.

30,11637-11645.

3396-3401.

6,2907 - 291 3.

Sci. USA 86,2577 - 2581.

184,687 - 698.

Documents

Functional modifications of transducin induced by cholera or pertussis-toxin-catalyzed ADP-ribosylation