The regulation of the cyclic GMP phosphodiesterase by the GDP-bound form of the alpha subunit of transducin.

The functional interactions of the retinal G protein, transducin, with the cyclic GMP phosphodiesterase (PDE) have been examined using the different purified subunit components of transducin and the native and trypsin-treated forms of the effector enzyme. The limited trypsin treatment of the PDE removes the low molecular weight gamma subunit (Mr approximately 14,000) of the enzyme, yielding a catalytic moiety comprised of the two larger molecular subunits (alpha, Mr approximately 85,000-90,000; beta, Mr approximately 85,000-90,000), which is insensitive to the addition of either the pure alpha T.GTP gamma S species or the pure beta gamma T subunit complex. However, the addition of the pure alpha T.GDP species to the trypsin-treated PDE (tPDE) results in a significant (90-100%) inhibition of the enzyme activity. This inhibition can be reversed by excess beta gamma T, suggesting that the holotransducin molecule does not (functionally) interact with the tPDE. However, the inhibition by alpha T.GDP is not reversed by the alpha T.GTP gamma S complex, over a range of [alpha T.GTP gamma S] which elicits a marked stimulation of the native enzyme activity, suggesting that the activated alpha T species does not effectively bind to the tPDE. The alpha T.GDP complex also is capable of inhibiting the alpha T.GTP gamma S-stimulated cyclic GMP hydrolysis by the native PDE. This inhibition can be reversed by excess alpha T.GTP gamma S, as well as by beta gamma T, indicating that the binding site for the activated alpha T species is in close proximity and/or overlaps the binding site for the alpha T.GDP complex on the enzyme. Overall, these results are consistent with a scheme where (a) both the small and larger molecular weight subunits of PDE participate in alpha T-PDE interactions, (b) the activation of PDE by the alpha T.GTP gamma S (or alpha T.GTP) species does not result in the complete dissociation of the gamma subunit from the enzyme, and (c) the deactivation of this signal transduction system results from a direct interaction between the alpha T.GDP species and the catalytic moiety of the effector enzyme.

The functional interactions of the retinal G protein, transducin, with the cyclic GMP phosphodiesterase (PDE) have been examined using the different purified subunit components of transducin and the native and trypsin-treated forms of the effector enzyme. The limited trypsin treatment of the PDE removes the low molecular weight y subunit (M, = 14,000) of the enzyme, yielding a catalytic moiety comprised of the two larger molecular subunits (a, M, = 85,000-90,000; 8, M. = 85,000-90,000), which is insensitive to the addition of either the pure aT. GTPyS species or the pure /~Y T subunit complex. However, the addition of the pure ~T -G D P species to the trypsin-treated PDE (tPDE) results in a significant (90-100%) inhibition of the enzyme activity. This inhibition can be reversed by excess &T, suggesting that the holotransducin molecule does not (functionally) interact with the tPDE. However, the inhibition by (YT. GDP is not reversed by the a~. GTPyS complex, over a range of [ a~. GTPyS] which elicits a marked stimulation of the native enzyme activity, suggesting that the activated aT species does not effectively bind to the tPDE. The aT. GDP complex also is capable of inhibiting the (YT. GTPyS-stimulated cyclic GMP hydrolysis by the native PDE. This inhibition can be reversed by excess (YT. GTPyS, as well as by f?yT, indicating that the binding site for the activated C~T species is in close proximity and/or overlaps the binding site for the a~. GDP complex on the enzyme. Overall, these results are consistent with a scheme where (a) both the small and larger molecular weight subunits of PDE participate in aT-PDE interactions, (b) the activation of PDE by the (YT. GTPyS (or C~T . GTP) species does not result in the complete dissociation of the 7 subunit from the enzyme, and (c) the deactivation of this signal transduction system results from a direct interaction between the a~. GDP species and the catalytic moiety of the effector enzyme.
The phototransduction system from vertebrate retina serves as an excellent model for studying receptor-coupled signal transduction. Each of the components participating in this system has been clearly identified and can be purified from rod outer segments in milligram quantities. These components include the photoreceptor, rhodopsin, the GTP-bind-which is responsible for the hydrolysis of cyclic GMP, while the smaller molecular weight subunit (7, MI = 14,000) (10,11) is suspected to represent the target for activated transducin (cf. . Over the past few years, a great deal of effort has been devoted toward understanding the general mechanisms which underlie receptor-G protein-coupled signal transduction (cf. Refs. 9,16,and 17). In all cases, the first step appears to be the activation of the receptor protein, i.e. by the binding of a hormone, or, in the case of the visual system, by the absorption of light by rhodopsin. This promotes the coupling of the receptor to the G protein, which in turn catalyzes the exchange of bound GDP or GTP. In the vision system, it is the binding of GTP which primes transducin for stimulating the activity of the cyclic GMP PDE. This stimulation proceeds until the bound GTP is hydrolyzed to GDP. The GTPase activity serves to deactivate the G protein and to return the system to its starting point.
Despite this general understanding, the details regarding the individual steps in the receptor-stimulated activationdeactivation cycle of the G protein, and the molecular mechanisms by which G proteins interact with, and regulate, effector proteins remain to be delineated. In the vertebrate phototransduction system, it has been commonly proposed that the transducin-mediated stimulation of the PDE activity is the outcome of a direct interaction between an activated a subunit of transducin and the y subunit of PDE (12)(13)(14)(15), However, it is still unclear whether the larger molecular The abbreviations used are: PDE, phosphodiesterase; DTT, dithiothreitol; G protein, guanine nucleotide-binding protein; GTPrS, weight subunits of the PDE participate in these interactions and whether the ( Y~. GDP species exerts any regulatory effect on PDE activity, i.e. does the GDP-bound form of (YT immediately dissociate from the enzyme or does it bind to the effector enzyme and impart some type of regulatory effect? The retinal visual transduction system is especially amenable to probing these types of mechanistic questions because the individual subunit components of transducin (i.e. both the GDP-and GTPyS-bound forms of (YT and the @YT complex), as well as the cyclic GMP PDE can be easily isolated. In addition, the y subunit of the effector enzyme can be removed by limited trypsin treatment yielding a catalytic moiety ( i e .
the (Y and 0 subunits of the enzyme) which is constitutively active (18,19). In the studies outlined below, we have examined the abilities of the different forms of the (YT subunit to functionally couple to both the native and trypsin-treated phosphodiesterase. The results of these studies suggest that the GDP-bound form of (YT, as well as the active, GTPySbound form, is capable of a direct interaction with the PDE. This interaction appears to involve the larger molecular weight subunits of the effector enzyme and may be directly responsible for the deactivation of the signaling system.
Purification, of Transducin and the cGMP Phosphodiesterase from Rod Outer Segments-Rod outer segments (ROS) were purified in the dark under dim red light from whole bovine retina essentially as described by Gierschik et al. (20). The purified ROS were suspended in an isotonic buffer containing 10 mM HEPES (pH = 7.5), 5 mM MgClZ, 1 mM DTT, 0.1 mM EDTA, 100 mM NaC1, 0.3 mM phenylmethylsulfonyl fluoride and washed several times by brief centrifugation. The ROS were then exposed to room light for 30 min, and the membranes were pelleted (46,000 X g for 10 min). The pellet was resuspended in a hypotonic buffer containing 10 mM HEPES (pH = 7.5), 1 mM DTT, 0.1 mM EDTA, 0.3 mM phenylmethylsulfonyl fluoride, and then several washes (at least five) were performed by repeated centrifugation and resuspension of the pellet. The supernatants which contain the cGMP PDE were pooled for further purification (see below). Holotransducin was eluted from the illuminated ROS membranes by resuspending the pellet in the hypotonic buffer supplemented with 100 p~ GTP and incubating the membranes (at 0 "C) for 30 min in room light. The membranes were pelleted and resuspended three times as above, and the supernatants were pooled. The transducin-containing pool was concentrated 10-fold in an Amicon ultrafiltration cell (YM 10 membrane). This procedure yields essentially pure holotransducin (6-10 mg per 300 retina).
The (YT subunit and the @ -y~ subunit complex of transducin were resolved by Blue Sepharose chromatography (21). The concentrated holotransducin (6-10 mg of protein in 5-10 ml of hypotonic buffer) was applied to a 50-ml Blue Sepharose column previously equilibrated with 10 mM HEPES (pH = 7.5), 6 mM MgC12, 1 mM DTT, 25% glycerol (buffer A). The column was washed with 200 ml of the above buffer, and the peak of unbound protein containing P-yT was pooled. The bound (YT was eluted from the Blue Sepharose column with the above buffer supplemented with 0.5 M KC1 (buffer B).
It has been reported that the UT subunit (22), as well as other G protein-a subunits (cf. Ref. 23), still contain bound GDP following their purification. We have quantitated the amount of bound GDP on (YT by eluting holotransducin from purified rod outer segments with [(Y-~'P]GTP. The eluted, 32P-labeled transducin was subjected to Blue Sepharose chromatography, as outlined above, with the 32Plabeled (YT subunit being eluted from the resin in high salt (500 mM KCl). This CYT subunit was subjected to an additional desalting step (Bio-Gel P6 DG) to ensure the removal of any unbound (free) labeled guanine nucleotide from the protein fraction. The stoichiometry of bound [oI-~'P]GDP (where the GDP is formed by the intrinsic GTPase activity of CYT (cf. Ref. 22)) per mol of aT (as determined by the method of Bradford (cf. Ref. 24)) was measured to be 0.70. The fact that this stoichiometry is less than 1.0 may be due to some of the CXT being inactive, or having lost some of its bound GDP, or it may reflect an overestimate in the total protein present based on the Bradford protein determination. This stoichiometry does agree well with the percentage (50-80%) of purified (unlabeled) (YT.GDP which is able to bind [35S]GTP-yS or [U-~'P]GTP in a light-dependent manner? suggesting that the actual amount of functional (YT may be overestimated when measured by the Bradford method by at least 20%. Thus, we conclude that the majority of the purified CYT, which is initially extracted by GTP elution, still contains bound GDP, and it will be referred to as the aT. GDP species throughout the text? Both the (YT. GDP and @-yT complexes were concentrated to -0.5 mg/ml protein and were stored in their elution buffers at -70 "C. The active form of the (YT subunit (ie. CYT.GTP-~S) was prepared in an identical manner except that the transducin was eluted from the illuminated ROS by the addition of 100 p M GTP-yS (instead of GTP). The subunits were separated by Blue Sepharose chromatography as described above, with the purified OIT. GTP-yS complex being stored in buffer B at -70 "C.
The cGMP PDE was purified (>go%) from the hypotonic wash supernatants essentially as described by Baehr  Activation of cGMP PDE by Limited Trypsinization-Purified PDE in buffer C was treated with trypsin (1:l w/w) for 80 s at 4 "C (see "Results" for the time dependence of the proteolytic activation). The proteolysis was stopped by the addition of a 4-fold molar excess of soybean trypsin inhibitor. The resulting PDE was constitutively active and will be referred to below as the trypsin-treated PDE (tPDE). The tPDE was generally prepared in 1-nmol quantities and stored at -70 "C in buffer C. Trypsin pretreated with inhibitor had no effect on basal PDE activity at the levels used in these experiments.
Measurement of cGMP PDE Activity-The hydrolysis of cGMP by the retinal PDE was measured as the rate of proton release using a pH microelectrode as originally described by Yee and Liebman (30). All assays were carried out at room temperature in a final volume of 150 p1 and in a buffer containing 10 mM HEPES (pH = 8.0), 3 mM MgCIz, 60 mM KCl, 1 mM DTT. Generally, all PDE assays contained 4-5 pmol of either intact or trypsin-treated enzyme. The other protein components (or buffer controls) were added as indicated in the figure legends. When examining the effects of a particular subunit-component of transducin on PDE activity, the component typically was incubated with the enzyme for -5-10 min on ice prior to their addition to an assay incubation. The components and the PDE were then incubated with the assay mixture for an additional 1-5 min at room temperature to establish a baseline. The assay was always initiated by the addition of cGMP (final concentration = 5 mM) and the pH (in millivolts) was measured at one determination per s. At the end of the assay period (-200 s), the buffering capacity (mV/nmol) was titrated by the addition of 0.5 pmol of KOH. The rate of hydrolysis of cGMP (nmol/s) was determined from the ratio of the slope of the pH record (mV/s) and the buffering capacity of the assay buffer (mV/ nmol). P. Guy and R. Cerione, unpublished observations. The addition of AlCL and NaF to the purified a~ elicits the activation of this subunit, via the formation of an (A1FI)GDP complex (251, as reflected by a stimulation of PDE activity, and by an increase in the intrinsic tryptophan fluorescence which mimics the fluorescence enhancement induced by active guanine nucleotides, i.e. GTP or GTP-yS (26). These A1F;-induced effects provide further support for the notion that the pure CYT species, prepared from GTP-eluted holotransducin, contains tightly bound GDP.  inhibitor (150 pmol). The values on the ordinate were determined by titrating the pH changes occurring during the assays (at room temDerature) with KOH (where it is assumed that 1 mol of effector enzyme. As an initial approach toward addressing these issues, we examined the effects of the various purified subunit components of transducin on the trypsin-treated PDE (tPDE). The2limited tryptic digestion of the PDE results in a constitutively active enzyme (cf. Refs. 18 and 19), as monitored by measuring the proton release which accompanies the hydrolysis of cyclic GMP (Fig. L4). The results in Fig. 1B show that only a brief exposure of the purified PDE to trypsin (60-80 s) is necessary to yield an enzyme activity which is typically greater than the activity achieved via the stimulation of native PDE by the QT GTPyS complex (see inset to Fig. 2, below). Since these conditions of trypsin digestion appear to result in the specific removal of the y subunit of the PDE (cj. Ref. la), it has been generally proposed that the stimulation of PDE activity by an activated transducin molecule reflects the removal of the inhibitory y subunit of the enzyme by the G protein (cf. Refs. 12-15).  (Fig. 2). We also have found that the addition of excess BYT to mixtures of native PDE and the pure QT-GTPyS complex has no effect on the QT. GTPyS-stimulated hydrolysis of cyclic GMP (i.e. where By, = 10 x aT.GTPyS; data not shown). Thus, these results suggest that the @YT complex has little affinity for either the active form of the QT subunit or for the effector enzyme itself. This is in contrast to the adenylate cyclase system where the (37 complex appears to be capable of inhibiting an as t' 1mulated enzyme activity (cj. Refs. 31-33) as well as eliciting a direct inhibition of the enzyme (34).

Transducin Interactions with the Trypsin-treated Cyclic GMP Phosphodiesterase-
While these results were consistent with previous suggestions regarding transducin interactions with the cyclic GMP PDE, an unexpected finding was obtained when the QT. GDP complex was added to assay incubations containing the tPDE. Specifically, unlike the case for the native PDE, where the addition of aT. GDP alone or in the presence of stoichiometric amounts of @yT elicits no change in the basal levels of activity (cj. Ref. 28), the addition of excess QT'GDP to the trypsintreated enzyme results in a significant decrease in the levels of cyclic GMP hydrolysis (i.e. 80 f 5% S.E., N = 5, for conditions shown in Fig. 2). This inhibition is specific for the aT. GDP complex, i.e. no changes in the levels of cyclic GMP hydrolysis are observed when the trypsin-treated PDE is incubated with buffer B alone. The extent of the inhibition is unaffected by the levels of cyclic GMP, over a range of 0.5-5 mM (data not shown), indicating that the inhibition by QT-GDP is not due to its occupancy of the substrate binding site on the enzyme. As shown by the dose-response profile in Fig.  3, the inhibition of the tPDE activity approaches 100% and typically occurs over a range of levels of QT. GDP which is similar to the levels of QT. GTPyS necessary to elicit a stimulation of the native PDE activity (see Fig. 3, inset)  or the ByT subunit complex (330 pmol) was incubated for 5 min at 0 "C prior to adding these components to an assay mixture and measuring the proton release which accompanies cyclic GMP hydrolysis at room temperature. The assay was initiated with 5 mM cyclic GMP. The data shown represent the initial rate for the proton release following the addition of cyclic GMP and was determined by titrating the pH changes occurring during the assays with KOH as described under "Experimental Procedures." The activities shown were subtracted from the (basal) activities measured in the absence of added tPDE (-0.3 nmol of cGMP hydrolyzed per s) and are the average of two determinations (from an experiment that was repeated twice with essentially identical results) with the error bars indicating the range of the determinations. Controls, which were performed with buffers A and B, indicate that these buffers have no effect on tPDE activity.
Inset, the stimulation of the cyclic GMP phosphodiesterase by the GTPyS-activated (YT subunit. The native cyclic GMP phosphodiesterase (PDE) (4 pmol) was incubated with the (YT.GTP~S complex (128 pmol) for 5 min at 0 "C prior to adding these components to an an assay mixture and measuring proton release at room temperature. The data shown represent the initial rate for the proton release as determined by titrating the pH changes occurring during the assays with KOH. The data bars represent the average of two experiments with the error bars indicating the range of the results from the individual experiments.
suggests that the relative affinities of the (YT GDP species for the tPDE and the (YT. GTPyS complex for the native enzyme are comparable. Overall, the results presented in Figs. 2 and  3 indicate that the GDP-bound form of the CYT subunit, as well as the GTP-(or GTPyS-) bound form of this subunit, must be capable of a direct interaction with the effector enzyme, and that a site on the effector enzyme other than the y subunit must be involved in the interactions with the (YT. GDP species. The trypsin-treated cyclic GMP phosphodiesterase (tPDE) (4 pmol) was incubated with the different levels of the (YT. GDP complex that are indicated in the figure for 5 min at 0 "C. The proton release which accompanies cyclic GMP hydrolysis was measured at room temperature following the addition of 5 mM cyclic GMP. The results shown represent the initial rate for proton release following the addition of cyclic GMP as determined by titrating the pH changes occurring during the assays with KOH as described under "Experimental Procedures." This dose-response profile was performed twice with essentially identical results. Inset, dose response for the stimulation of the native cyclic GMP phosphodiesterase by the (YT.GTP~S complex. The native cyclic GMP phosphodiesterase (4 pmol) was incubated with the different levels of the (YT. GTPyS complex that are indicated on the abscissa for 5 min at 0 "C. Cyclic GMP hydrolysis was measured as described above. The activities shown were obtained by subtracting the total activities from the activities measured with PDE alone (~0 . 6 nmol of cyclic GMP hydrolyzed per s).
The ByT complex is able to prevent the inhibition by the aT. GDP complex of the trypsin-treated enzyme. Specifically, the results presented in Fig. 4 illustrate that when increasing amounts of the pure /?YT complex are preincubated with (YT. GDP, prior to the addition of tPDE, an essentially complete (75-100%) restoration of the levels of cyclic GMP hydrolysis occurs ( i e . 88 f 7% S.E., N = 8, recovery of the activity measured in the absence of (YT. GDP). Thus, the complexation of the PyT complex with (YT. GDP to re-form the intact holotransducin most likely eliminates (YT GDP-tPDE interactions. The ability of the LYT. GTPyS complex to reverse the inhibition by LYT. GDP of the trypsin-treated enzyme activity also was examined. The rationale here was that if under normal physiological conditions, the GDP-bound (YT complex stayed associated with the effector enzyme following GTP hydrolysis ( i e . following the conversion of (YT.GTP to (YT-GDP), then the GTP-(or GTPyS-) and GDP-bound forms of (YT were likely to share a common binding domain on the effector enzyme. As noted above (cf. Fig. 2), the ~T . G T P~S complex does not elicit a stimulation of cyclic GMP hydrolysis by the trypsin-treated enzyme (Fig. 5, O), which is unlike the case for the addition of (YT. GTPyS to the native enzyme (cf. Fig.  3, inset). Likewise, under these conditions, there is little effect by LYT. GTPyS on the extent of the inhibition by (YT-GDP of the trypsin-treated enzyme activity (Fig. 5, D). Thus, the active (GTPyS-bound) form of the (YT subunit is not able to effectively compete with the (YT. GDP species for a site on the tPDE molecule. In some cases, at levels of a~. GTPyS > 50 pmol, a slight inhibition (typically -10%) of the activity of the trypsin-treated enzyme is observed (cf. Fig. 5 , also Fig. 2). This either can reflect the ability of the LYT. GTPyS species to weakly mimic the inhibitory effects elicited by the (YT. GDP complex or an inhibitory effect elicited by a small amount of free aT (i.e. GTPyS-free) molecules which are present in the CYT. GTPyS preparation.
Transducin Interactions with the Native Cyclic GMP Phosphodiesterase-In order to determine whether the CYT. GDP complex is able to interact with the native PDE, we examined the ability of this LYT species to inhibit the a~.GTPyS-stimulated enzyme activity. The results presented in Fig. 6A illustrate that the addition of excess LYT. GDP (65 pmol) to a mixture of ~T . G T P~S (32 pmol) and native PDE (4 pmol) results in about a 50% decrease in the levels of cyclic GMP hydrolysis (i.e. 51 & 2% S.E., N = 4). As is the case with the trypsin-treated enzyme, the PYT complex is capable of revers-  The inhibition by (YT. GDP can be reversed by increasing the amounts of ( Y T -G T P~S present in the assay incubation. For example, when (YT. GDP = 2 X (YT. GTPyS, there is a 40-60% inhibition of the CUT. GTPyS-stimulated enzyme activity (e.g. compare the open and closed circles corresponding to an abscissa value of 32 pmol of CYT. GTPyS in Fig. 6B: also cf. Fig.  6A), while when (YT. GTPyS = 2-3 X CYT.GDP, the inhibition by CXT-GDP is essentially eliminated ( 5 5 % , N = 2, e.g. compare the open and closed circles corresponding to an abscissa value of 192 pmol of CYT-GTP~S in Fig. 6B). Although maximal stimulation by (YT. GTPyS of the native PDE activity can be achieved for the conditions described in Fig. 6B, the levels of (YT. GTPyS necessary for half-maximal stimulation are typically higher than those obtained in the absence of CYT. GDP, by about a factor of 2. Overall, these results suggest that the (YT. GDP and CUT. GTPyS complexes cannot be bound, simultaneously, to an activated PDE molecule (see below).

DISCUSSION
Currently, little is known regarding the molecular details characterizing the coupling of GTP-binding proteins to biological effectors (i.e. enzymes or ion channels). The retinal phototransduction system offers an excellent model for probing these mechanisms since most of the key components of this signal pathway can be purified rather easily and in relatively high abundance. Previous schemes depicting the interactions between transducin and the cyclic GMP PDE  were based, for the most part, on the findings that a free UT subunit which contains a tightly bound, active guanine nucleotide (such as GTPyS or GppNHp) is sufficient to stimulate cyclic GMP hydrolysis by the effector enzyme (cf. Refs. 15 and 35). The fact that the ByT subunit complex, either alone or in the presence of activated (YT subunits, has no effect on PDE activity is consistent with the suggestions that G protein-effector enzyme coupling is specific for the (YT subunit.
Since the limited trypsin treatment of the PDE results in an enzyme which lacks the low molecular weight y subunit and which is constitutively active, it has been proposed that the interaction of the active form of (YT with PDE occurs at the y subunit of the effector enzyme (12)(13)(14)(15). This interaction was thought to release the inhibitory subunit from the catalytic subunit(s) of the PDE and thus enhance the hydrolysis of the substrate, cyclic GMP. Recently, it has been suggested that there are at least two forms of y subunits (15). The concentrations of transducin typically used for in uitro assays ( 5 1 p M ) were proposed to be sufficient to remove one, but not both, of the inhibitory y subunits from the PDE molecules (cf. Ref. 15). However, trypsin treatment would effectively eliminate both inhibitory subunits and generate a fully active enzyme. Taken together, these various notions are seemingly consistent with the results presented here which illustrate that the active ( Y T -G T P~S species does not enhance the activity of the tPDE and that the trypsin-treated enzyme is typically more active than the ( Y~. GTPyS-stimulated enzyme. However, they would not predict the finding that the GDPbound form of the (YT subunit markedly inhibits the activity of the trypsin-treated enzyme. Rather, the ability of the (YT-GDP complex to inhibit the tPDE suggests some type of a direct interaction between this ( Y~ subunit and the catalytic moiety of PDE (i.e. with the (Y and/or fl subunits of this enzyme). Overall, the results presented here can be incorporated into a working model for the interactions of the GTP-(or GTPyS-) and GDP-bound forms of ( Y~ with the effector enzyme (Fig. 7). For simplicity, the PDE molecule is shown to contain only a single lower molecular weight, y subunit. However, the presence of two types of y subunits on the PDE would not change the general features of this model. We The cyclic GMP phosphodiesterase (PDE) is suggested to be comprised of a single type of y subunit, although recent reports suggest that more than one type of y subunit may exist (15). RHO = light-activated rhodopsin. HT = holotransducin. The initial interaction of (YT. GTP with the cyclic GMP PDE is suggested to occur exclusively on the y subunit of PDE. The interaction of (YT. GDP with PDE is suggested to occur on a larger molecular weight subunit(s); the specific location of the ~T -G D P binding site on the enzyme has not been identified.
suggest that the initial interaction between an active aT. GTP species, as well as an ( Y~. GTPyS complex, and native PDE would occur at the y subunit of the enzyme (step I in Fig. 7 ) .
This interaction is not suggested to elicit the complete release of the y subunit from the enzyme, for reasons that will be described below, but rather shifts its position in a manner that increases the accessibility of the catalytic site and thereby stimulates the production of GMP via cyclic GMP hydrolysis. In the case of the (YT.GTP complex, the hydrolysis of the bound GTP to GDP would alter the conformation of LYT and increase its affinity for a binding domain on a larger molecular weight subunit(s) of the enzyme (step 2). This binding interaction between the (YT.GDP complex and the enzyme does not block substrate (cyclic GMP) binding but rather must induce a change in the conformation of the enzyme, resulting in an immediate inhibition of its catalytic activity. The binding site for the (YT.GDP complex on the larger molecular weight subunit(s) is likely to be immediately adjacent to the binding site of the (YT. GTP or (YT. GTPyS species on the y subunit of the enzyme. Such a location may explain the recent observations that an activated (YT. GppNHp complex can be cross-linked to the larger molecular weight subunits of the native PDE (cf. Ref. 36). Thus, the initial binding of an ( Y~. GTPyS complex to the PDE would prevent the binding of the L Y T . GDP species to the catalytic moiety, such that at high concentrations of LYT. GTPyS the inhibition by LYT-GDP is eliminated (cf. Fig. 6B). This also requires that the (YT. GTPyS species, following its initial interaction with the y subunit of the effector enzyme, does not become physically separated from the larger molecular weight subunits of the PDE. Otherwise, the bound LYT . GTPyS complex could not prevent the binding of LYT-GDP to the catalytic moiety. The results of steady state kinetic studies, monitoring the inhibition of the native PDE by the purified y subunit, also suggest that the stimulation of the effector enzyme by activated transducin does not result in the complete removal of the y subunit from the enzyme complex (cf. Ref. 37). However, in the trypsintreated enzyme, the binding domain for the LYT. GTPyS species (i.e. the y subunit) is proteolytically removed, and thus the aT.GTPyS complex does not effectively interfere with the inhibitory effects of the LYT. GDP species (cf. Fig. 5).
We find that the addition of the &T subunit complex can prevent the inhibition of PDE activity by the (YT. GDP species. This then provides a potential mechanism for the signaling system to return to its resting state. Specifically, the reassociation of the LYT. GDP species with PYT would form holotransducin (step 3 in Fig. 7 ) which does not inhibit the enzyme activity. Most likely, this reflects the inability of the holotransducin to bind to the enzyme, thereby enabling the G protein to recouple to light-activated rhodopsin (step 4 ) and returning the PDE to its initial basal state to await another stimulatory signal from an active LYT subunit.
It should be noted that many of the results reported in this study also would be consistent with a model where both the (YT. GTP (or (YT-GTPyS) species and the LYT. GDP complex bind to a single, common site on a larger molecular weight subunit(s) of PDE. In this case, the LYT~GTP (or L Y T ' G T P~S ) would bind with higher affinity to the native (heterotrimeric) PDE, stimulating enzyme activity, while the LYT. GDP would bind more tightly to the activated form of PDE and inhibit activity. However, our results, when viewed within the context of this scheme, would imply that a significant percentage (if not all) of the LYT-GTP (or LYT. GTPyS) species dissociates from the activated effector enzyme under conditions where the L Y~ species can initially bind to the native enzyme. Such a scheme seems unsatisfactory since the dissociation of the (YT. GTP (or (YT'GTP~S) from the activated enzyme would then enable the PDE to return to its basal, inactive state, even before the occurrence of the GTPase activity on the L Y~ subunit. In addition, if the LYT. GTP (or (YT. GTPyS) species dissociates from the activated, native PDE, it is difficult to reconcile how the (YT. GTPyS species effectively protects against the inhibition by the LYT. GDP complex of the native enzyme activity.
Clearly, the key feature of the model shown in Fig. 7 (and an apparent requirement of any scheme proposed to explain these findings) is that the LYT. GDP species is directly responsible for the deactivation step in the visual transduction pathway through its ability to interact with the effector enzyme. Future studies will be directed toward characterizing the specific domains involved in this interaction, as well as the nature of the putative inhibitory-conformational change which is induced in the PDE by this LYT species. It also will be of interest to re-examine the molecular mechanisms underlying the interaction of active LYT complexes with the effector enzyme. Particularly relevant to this issue is whether the photoreceptor, by remaining associated with the active LYT subunit at least until it couples to the PDE, might enhance the stimulation of the effector enzyme activity. We have, in fact, recently reported that an antibody which binds to the rhodopsin-binding domain on the (YT subunit (i.e. at the COOH-terminal) accentuates the stimulation of cyclic GMP hydrolysis by the LYT. GTPyS complex (28). It therefore will be interesting to examine whether the photoreceptor might function in a similar role and if either the photoreceptor or the COOH-specific antibody against LYT will alter the affinities of the active (Y subunit for the effector enzyme or in any way affect the interplay between the GTP-and GDP-bound forms of L Y~ in regulating PDE activity.