Transducin activates cGMP phosphodiesterase by trapping inhibitory γ subunit freed reversibly from the catalytic subunit in solution

Activation of cGMP phosphodiesterase (PDE) by activated transducin α subunit (Tα*) is a necessary step to generate a light response in vertebrate photoreceptors. PDE in rods is a heterotetramer composed of two catalytic subunits, PDEα and PDEβ, and two inhibitory PDEγ subunits, each binding to PDEα or PDEβ. Activation of PDE is achieved by relief of the inhibitory constraint of PDEγ on the catalytic subunit. In this activation mechanism, it is widely believed that Tα* binds to PDEγ still bound to the catalytic subunit, and removes or displaces PDEγ from the catalytic subunit. However, recent structural analysis showed that the binding of Tα* to PDEγ still bound to PDEα or PDEβ seems to be difficult because the binding site of PDEγ to PDEα or PDEβ overlaps with the binding site to Tα*. In the present study, we propose a novel activation mechanism of PDE, the trapping mechanism, in which Tα* activates PDE by trapping PDEγ released reversibly and spontaneously from the catalytic subunit. This mechanism well explains PDE activation by Tα* in solution. Our further analysis with this mechanism suggests that more effective PDE activation in disk membranes is highly dependent on the membrane environment.

. Possible PDE activation mechanisms. (a) Conventional mechanism. In the inactive state of PDE (purple), PDEγ (Pγ) binds to the PDE catalytic subunit (PDEα or β, indicated as Pcat) at the binding site on the catalytic subunit (yellow oval). Activated Tα (Tα*) binds to PDEγ to displace (a 1 ) and/or remove PDEγ (a 2 ) from the catalytic subunit to activate PDE (pale red). (b) Trapping mechanism. PDEγ is bound to the catalytic site of PDE (yellow oval) with the binding site in PDEγ (pink oval), but PDEγ is freed reversibly from the catalytic subunit according to the dissociation constant, K D1 (upper). This freed PDEγ is trapped by Tα* with the dissociation constant, K D2 , at the binding site of PDEγ (pink oval) to Tα* (yellow rectangular) to inhibit rebinding of PDEγ to the catalytic subunit (lower).
Formulation of the trapping mechanism. According to the reaction scheme shown in Fig. 1b, we formulated Eq. (1), same as equation (s10) in SI Methods, which expresses the PDE activity as a function of Tα* concentration (Tα-S* concentration in this study):  are the only unknown parameters in the above equation. In the followings, we tried to determine these values experimentally, and examined whether we can explain PDE activities elicited by addition of Tα-S* of known concentration. Because the effectiveness of Tα-S* on PDE activation was different in the measurement using purified PDE and that using PDE in rod outer segment (ROS) membranes, we examined both cases in the followings.
Determination of K D1 of the complex of pDeγ and pDecat. To determine K D1 of the PDEγ·PDEcat complex, first we measured the PDE activity using purified PDE at diluted low concentrations (Fig. 3a, filled circles). As dilution increases, the concentration of freed PDEγ, and therefore, relative PDE activity increases depending on K D1 . The relation between the concentration of PDE and the measured relative PDE activity was fitted with an equation formulated for a simple binding-dissociation reaction of PDEγ and PDEcat (equation (s5) in SI Methods) to determine K D1 . The best-fitted K D1 of the PDEγ·PDEcat complex in purified PDE was 10 pM. However, the data points scattered slightly in Fig. 3 so that we could only determine the range of K D1 : it was approximately 5-20 pM (broken curves for K D1 of 5 and 20 pM) and close to the reported value of <10 pM obtained with purified bovine PDE previously 4 . It should be mentioned here that freed PDEγ is completely freed from PDEcat. If freed PDEγ is removed from the active site but is still attached to PDEcat, dilution will not induce the increase in the relative PDE activity. It is because dilution does not affect re-binding of PDEγ to PDEcat in this case. Similar dilution study was made using ROS membranes (filled circles in Fig. 3b). The best-fitted K D1 of the PDEγ·PDEcat complex in ROS membranes was 54 pM, and the range was 40-60 pM (broken curves for K D1 of 40 and 60 pM). Postulated elution of freed PDEγ in the dark in ROS membrane suspensions was examined with washing membranes, and the resultant increase in PDE activity was observed with repetitive washes (Fig. S1).

Figure 2.
Comparison of the bindings of PDEγ and PDE to immobilized Tα-S* with the SPR method. Recombinant PDEγ or purified PDE was injected at the concentrations indicated. Injections were made as indicated (horizontal bars), and bound proteins were washed out after each injection. Immobilization level of Tα-S* was ~2300 resonance unit (RU).
www.nature.com/scientificreports www.nature.com/scientificreports/ Determination of K D2 of the complex of pDeγ and tα-s*. To measure K D2 of the PDEγ·Tα-S* complex, we measured it in two configurations (Fig. 4) using the SPR method. One configuration was similar to that shown in Fig. 2: Tα-S* was immobilized. In Fig. 4a, 2-16 nM PDEγ was perfused until the signal reached to a steady level and bound PDEγ was washed out almost completely at each PDEγ concentration. All of the measured time courses were then globally fitted with a program provided by the manufacturer (black broken traces in Fig. 4a, see Methods) to determine K D2 of the PDEγ·Tα-S* complex. In a total of three different measurements using two different sensor chips, we obtained K D2 of 0.73 ± 0.13 nM (mean ± SE, n = 3) for the PDEγ·Tα-S* complex.
The value of K D2 was determined in the reversed configuration: PDEγ was immobilized and Tα-S* was perfused (Fig. 4b). In this case, Tα-S* of increasing concentration was perfused in a less-time consuming way: Tα-S* Figure 3. Determination of K D1 of the PDEγ·PDEcat complex. PDE activity was measured using purified PDE (a) and ROS membranes (b) at the concentrations of PDE shown in the horizontal axis. Each data point shows the result of a single activity measurement. The activity is shown as the relative activity (%) against the full PDE activity that was determined after treatment with trypsin (trypsin-treated). (a) The concentration of purified PDE was calibrated with SDS-PAGE. The relation between relative PDE activity and the concentration of PDE was fitted with equation (s5) to determine K D1 of the PDEγ·PDEcat complex. The best-fitted K D1 in solutions of purified PDE was 10 pM (solid curve), and the expected curve for K D1 of 5 pM and that of 20 pM are also shown (broken curves). (b) Similar as in (a), but PDE content in a ROS membrane suspension was estimated by assuming that the molar ratio of PDE to rhodopsin is 1/270 in ROS membranes 14 . The best-fitted K D1 was 54 pM (solid curve). Expected curve for K D1 of 40 pM and that of 60 pM are also shown (broken curves). Recombinant PDEγ was injected at various concentrations indicated, and perfused until the binding signal was almost saturated. The bound proteins were washed out almost completely after each of the injections. Immobilization level of Tα-S* was ~400 RU. The binding signals (solid traces) were fitted globally using a Heterogeneous Ligand with MTL program to calculate K D2 , and it was 0.73 ± 0.13 nM (mean ± SE, n = 3). Flow rate was 10 μl/min. (b) SPR measurements of the binding of Tα-S* to immobilized PDEγ. Tα*-S was injected at indicated concentrations and perfused for 125 sec (horizontal bars) for the binding and then washed out for 175 sec each time. The binding signal (pink solid trace) was globally fitted using a 1:1 binding with MTL program (black broken trace) to calculate K D2 . The best-fitted K D2 was 5.6 ± 1.3 nM (mean ± SE, n = 5). Immobilization level of PDEγ was ~100 RU. Flow rate was 30 μl/min. www.nature.com/scientificreports www.nature.com/scientificreports/ was added before bound Tα-S* was washed out completely (pink trace). Measured time course was fitted with the other program provided by the manufacturer (black broken trace in Fig. 4b, see Methods). From the fitting results, K D2 was estimated to be 5.6 ± 1.3 nM (mean ± SE, n = 5). (According to the manufacturer's protocol, the same dissociation constant can be obtained no matter whether bound protein is completely washed out as in Fig. 4a or not as in Fig. 4b.) Obtained values of K D2 in two configurations (Tα-S* immobilized or PDEγ immobilized) were ~8 times different (0.73 nM/5.6 nM = 1/7.7). Although immobilizations of Tα-S* and PDEγ were designed not to affect the binding site seriously (see Methods), immobilization seemed to affect K D2 slightly. We, therefore, concluded that K D2 is 0.73-5.6 nM, which is consistent with the values of 0.1-33 nM reported previously utilizing various methods for the measurement 4,8-10 .

Validation of the trapping mechanism for PDE activation of purified PDE in solution.
In Figs 3 and 4, we determined the ranges of K D1 of the PDEγ·PDEcat complex for purified PDE (Fig. 3a) and PDE in ROS membranes (Fig. 3b), and the range of K D2 of the PDEγ·Tα-S* complex (Fig. 4) using the SPR method. To validate the trapping mechanism, we then examined whether this mechanism can explain the activation of PDE by Tα-S* of known concentrations with use of Eq. (1) formulated for this mechanism. Figure 5a shows the measurement of activation of purified PDE by purified Tα-S* in solution at indicated concentrations with the pH assay method 11,12 . The pH decrease accompanied by hydrolysis of cGMP was calibrated, and the PDE activity was determined from the slope. Full PDE activity was determined after treatment with trypsin (trypsin-treated). PDE activity at a given Tα-S* concentration is expressed as the % of the full PDE activity, and the summarized result is shown in Fig. 5b (filled circles and bars showing mean ± SE). Then, the relation between the relative PDE activity and the Tα-S* concentration was fitted with Eq. (1). As shown above, we determined the range of K D1 in solution (5-20 pM, Fig. 3a) and that of K D2 (0.73-5.6 nM, Fig. 4), and for this reason, we tried to examine whether we can explain PDE activation by Tα-S* in Fig. 5a with these dissociation constants in those ranges. First, we used K D1 of 10 pM, but could not obtain a best-fitted value of K D2 within the range of K D2 we determined in Fig. 4. For this, we set K D1 at 5 pM, for example, and then determined K D2 that provides the best fit to the PDE activation curve. The value of K D1 was increased by 1 pM step and the best-fitted K D2 was determined each time. We then found that at each K D1 value from 2 pM to 6 pM, we can find a K D2 value that gives a reasonable fit to the PDE activation curve in Fig. 5b. Interestingly, each pair of K D1 and K D2 we determined showed similar goodness of fit (χ 2 , Table 1), and we show the result of K D1 = 5 pM and K D2 = 4.5 nM in Fig. 5b (solid curve). We tried to estimate the activation curve of PDE by Tα-S* with the conventional binding mechanism shown in Fig. 1a under the condition of K D = 4.5 nM. The expected curve deviated greatly from the measured result (broken curve in Fig. 5b). From this result, we concluded that Tα-S* activates PDE by trapping PDEγ freed reversibly from PDE and by inhibiting its re-binding to PDEcat. An alternative possibility is that Tα-S* binds to PDE (PDEαβ or PDEγ still bound to PDEαβ). However, this possibility can be excluded because the binding signal of PDE to Tα-S* is low (Fig. 2). PDE activation with use of purified PDE in solution was measured only at 15 nM PDE. It was because at higher concentrations of purified PDE, the measurement was not possible because of protein aggregation.
Validation of the trapping mechanism for activation of pDe in Ros membrane suspension. As shown above, it is highly possible that purified PDE is activated by the trapping mechanism in solution. Then we examined whether this mechanism is applied to PDE activation in ROS membranes. In the measurement of PDE www.nature.com/scientificreports www.nature.com/scientificreports/ activity in an illuminated ROS membrane suspension, we added GTPγS at a concentration lower than that of Tα, of which concentration was estimated on the assumption that the molar ratio of Tα to rhodopsin 13 is 1/10. In this way, we limited the amount of Tα-S* by the amount of added GTPγS 12 . In our previous study 12 , PDE activation by addition of GTPγS is dependent on the ROS membrane concentration: the lower the concentration, the lower the maximum PDE activation. For this reason, ROS membranes containing rhodopsin of 1.5, 10 and 20 μM (abbreviated as 20 μM rhodopsin membranes, for example) were used to measure the PDE activation at various concentrations of Tα-S*. The activity was measured similarly as in Fig. 5a, and the results are shown in Fig. 6a-c (circles and bars showing mean ± SE). Note that the horizontal axis is different in each panel, which is because the maximum Tα-S* concentration should be equal to the concentration of Tα at different membrane concentrations (0.15 μM Tα in 1.5 μM rhodopsin membranes, for example). As reported previously 12 , in 20 μM rhodopsin membranes, we obtained almost a full PDE activity that is observed in trypsin-treated ROS membranes (Fig. 6c). Then, we fitted the results in Fig. 6 with Eq. (1) to estimate K D1 and K D2 in ROS membranes at each membrane concentration. As shown in Fig. 3b, we found that the range of K D1 in ROS membranes is in the range of 40-60 pM. We therefore arbitrary set K D1 at 40-60 pM with 5 pM step, and determined K D2 each time. The results are summarized in Table 2. Each pair of K D1 and K D2 in Table 2 gave reasonable fit to the PDE activation curve in ROS membranes without significant differences at each ROS membrane concentration (see χ 2 for each membrane concentration in Table 2). Because we obtained the value of 54 pM as K D1 in ROS membranes (Fig. 3b), fitting result with K D1 of 55 pM is shown at each membrane concentration (Fig. 6). We assume the presence of freed PDEγ in the dark or in the absence of Tα-S*, and our fitting showed that its population is low and 1.9% in 20 μM rhodopsin membranes (Fig. 6c).
In Table 2, best-fitted K D2 values varied significantly depending on the ROS membrane concentration, and they decrease as membrane concentration increases: at a constant K D1 value of 55 pM, the best-fitted K D2 values are 23.6, 0.201 and 0.0565 nM in 1.5, 10 and 20 μM rhodopsin membranes, respectively. Apparent ROS membrane concentration-dependent changes in K D2 suggest the loss of intrinsic Tα-S* from membranes (see Discussion). Although K D2 values in 1.5 and 20 μM rhodopsin membranes (23 and 0.0565 nM, respectively) were not in the K D2 range we observed in Fig. 4 in solution (SPR study, 0.73-5.6 nM), our analysis seemed to explain PDE activation by Tα-S* in ROS membranes with the trapping mechanism as well (but, see Discussion). We estimated the PDEcat activation at pseudo-physiological concentrations of Tα and PDEcat with the trapping mechanism. For this, we assumed that the rhodopsin concentration is 3 mM and that the transducin 13   www.nature.com/scientificreports www.nature.com/scientificreports/ PDEcat, and with K D values in membranes (K D1 = 55 pM and K D2 = 56.5 pM), we obtained the result that 93% of PDEcat is activated in membranes. This high activation of PDE at these Tα* and PDEcat concentrations was not obtained with use of K D values we obtained in solution (K D1 = 5 pM and K D2 = 4.5 nM), and the PDEcat activation in this case was only ~11%. The result showed that PDE activation is different between in solution and in membranes, and that one cannot apply the K D values obtained in solution directly to the study on PDE activation in membranes.

Discussion
In the activation mechanism of photoreceptor PDE, it has been generally believed that Tα* binds to PDE, and removes or displaces PDEγ from the catalytic site of PDEcat (conventional mechanism, Fig. 1a). However, our SPR analysis showed that Tα* (actually Tα-S*) binds much more effectively to PDEγ than to PDE (Fig. 2), which is not consistent with the mechanism a 1 in Fig. 1a. The result in Fig. 2 also revealed that Tα* hardly releases PDEγ from PDE, which shows that the mechanism a 2 shown in Fig. 1a is unlikely (Fig. 2). Furthermore, with the conventional activation mechanism, Tα*-dependent PDE activation could not be explained quantitatively in solution (Fig. 5b). Instead, PDE activation in solution is reasonably explained by the trapping mechanism ( Fig. 5 and Table 1) using experimentally estimated dissociation constants of the PDEγ·PDEcat complex (K D1 , Fig. 3a) and the PDEγ·Tα* complex (K D2 , Fig. 4).
In the fitting of PDE activation in ROS membrane suspension, best-fitted K D2 decreased as the membrane concentration increased: at a constant value of K D1 of 55 pM, K D2 was 23.6 nM in 1.5 μM rhodopsin membranes, and it decreased significantly to 0.0565 nM in 20 μM rhodopsin membranes (Table 2). Apparently, K D2 that can be determined in a ROS membrane suspension is dependent on the membrane concentration. The reason for this is not known. However, we previously found that ~65% Tα-S* is eluted from 0.75 μM rhodopsin membranes, but ~50% from 15 μM rhodopsin membranes in carp 12 . This 15% excess in the amount of Tα-S* remaining in 15 μM rhodopsin membranes seems to be sufficient to activate all of PDEcat molecules 12 , which could be deduced by the Tα/rhodopsin molar ratio 13 of 1/10 and the PDE/rhodopsin (i.e., 1/2PDEcat/rhodopsin) molar ratio 14 of 1/270: the molar ratio of 15% of Tα-S* to PDEcat in ROS membranes is ~2:1. It is possible that there could be two types of Tα*. One type binds to membranes tightly and the other loosely, and the latter re-binds to the membrane effectively when the membrane concentration is high. We speculate that the loosely-bound Tα* becomes soluble rather easily at low membrane concentrations to increase K D2 and to reduce the maximum PDE activation by reducing the effective Tα* concentration. At high membrane concentrations, this type of Tα* re-binds effectively to the membranes to contribute significantly to activate PDEcat and to lower K D2 . In fact, Tα has been known to be differentially lipidated with 65% of unsaturated and 30% of saturated C12 or C14 fatty acids 15 .
The trapping mechanism explains PDE activation in solution with K D1 and K D2 , both determined experimentally (Fig. 5). It also explains the activation of PDE in ROS membranes with K D1 obtained experimentally in membranes and K D2 estimated by a fitting (Fig. 6). According to the measured K D values in membranes, the affinity of PDEγ to PDEcat (K D1 = 55 pM, Fig. 3b) is almost the same as that to Tα* (K D2 = 56.5 pM, Fig. 6c), which is consistent with a very effective activation of PDE by Tα-S* in membranes. In contrast, K D1 is lower (5 pM) and K D2 is higher (4.2 nM) in solution than those in membranes, which would be the reason why PDE activation by Tα-S* is not so efficient in solution (Fig. 5b). One possible reason for these differences in K D values in membranes and in solution would be the difference in protein conformation in membranes and in solution.
We further examined whether we can expect sufficiently large PDE activation at a level of a single photon response with the trapping mechanism. Using Eq. (1) and expected Tα* concentration necessary for generation of the response together with K D1 and K D2 values obtained in membranes, we found that 6.32% of PDEcat, is active in a single surface of a disk membrane. In contrast, in case no Tα* is present, i. e., in the dark, 0.16% of  www.nature.com/scientificreports www.nature.com/scientificreports/ PDEcat is active (for details, see SI methods). This result seems to indicate that the trapping mechanism can be applied also to PDE activation in membranes.
However, the values of K D2 in ROS membranes were not determined experimentally and further we are not sure how we can apply K D1 and K D2 values to the activation of PDE in membranes. Tα* and PDE in a disk membrane are undoubtedly situated at certain orientations on the disk membrane, which probably increases the chance of encounter of Tα* to PDEγ. Additionally, membrane proteins (for example, PDEcat) are localized only in disk membranes while soluble proteins (Tα-S* and freed PDEγ) are in an aqueous phase in our measurement in ROS membrane suspensions. It is not certain whether we can appropriately apply Eq. (1) to PDE activation in these cases. Further complication seems to be present when we want to extend our analysis to PDE activation in intact ROS. Freed PDEγ is supposedly present in the inter-diskal space of which volume is of the order of <1 fL, where re-binding of freed PDEγ to PDEcat would be much more effective compared with the re-binding in a test tube. For these reasons, further studies seem to be necessary to find how we can apply the trapping mechanism to PDE activation in intact ROS from kinetical and mechanistic view point. Nonetheless, because PDEγ binds to PDEcat or Tα* using the same region, Tα* should bind to PDEγ after the dissociation or displacement of PDEγ from PDEcat even when the dissociation or displacement is induced after multistep interaction between Tα* and PDEγ as suggested 16,17 . Methods preparation of rod outer segment (Ros) membranes from frog. All experiments with frogs (Rana catesbeiana) in this study were performed in accordance with the institutional guidelines and all experimental protocols were approved by Osaka University Graduate School of Frontier Biosciences (approval number FBS-15-003). ROS membranes were prepared as described previously using a stepwise sucrose density gradient 18 . Obtained ROS membranes were frozen in liquid nitrogen and stored at −80 °C until use. To calibrate the concentration of ROS membranes, the amount of rhodopsin in an aliquot of the membranes was quantified spectrophotometrically with assuming that the molar absorption coefficient of frog rhodopsin is 40,000 M −1 cm −1 at 500 nm. All of these manipulations were carried out in complete darkness with the aid of an infrared image converter (NVR 2015; NEC, Tokyo, Japan).

Extraction and Purification of PDE and Tα-s* from Ros membranes. Crude PDE and crude
Tα-S* were extracted basically as described previously 12,19 . Crude PDE was then loaded on a Mono Q PC 1.6/5 column (ÄKTAmicro system, GE Healthcare), and a 0-1 M NaCl gradient in an elution buffer (10 mM HEPES-NaOH, 2 mM MgCl 2 , 1 mM DTT, pH7.5) containing 0.005% (v/v) Tween 20 was applied. Eluted fractions at 0.47-1 M NaCl containing purified PDE were concentrated using a Spin-X UF column (Mr 30,000 cutoff, Corning). Then, the buffer containing purified PDE was changed to a potassium gluconate buffer (K-gluc buffer; 115 mM potassium gluconate, 10 mM HEPES, 2.5 mM KCl, 2 mM MgCl 2 , 0.2 mM EGTA, 0.1 mM CaCl 2 , and 1 mM dithiothreitol (DTT), pH 7.5) containing 0.005% (v/v) Tween 20 using a Superdex 200 PC 10/300 GL column (ÄKTAmicro system, GE Healthcare). The resultant purified PDE solution was concentrated using a Spin-X UF column, and stored at −80 °C until use. An aliquot of purified PDE was subjected to SDS-PAGE and the gels were stained with Oriole Fluorescent Gel Stain Kit (Bio-Rad) to assess the purity of PDE and also to quantify its amount using bovine serum albumin as a molar standard. Purity of PDE was almost 100% and the molar ratio of 2PDEγ/PDEαβ was 1.01 ± 0.01 (mean ± SE, n = 3).
Tα-S* was purified from crude Tα-S* according to the method reported previously 19 . Briefly, a Blue Sepharose 6 Fast Flow column (GE Healthcare) and a DEAE Sepharose Fast Flow column (GE Healthcare) were connected in tandem in this order for the purification. Then, a solution of crude Tα-S* supplemented with 2 mM MgCl 2 was loaded on the column equilibrated with the elution buffer. The column was washed with the elution buffer sufficiently to remove unbound proteins, and then the Blue Sepharose column and the DEAE Sepharose column were separated: frog Tβγ bound to the Blue Sepharose column, and most of Tα-S* passed through this column and bound to the DEAE Sepharose column. Thus, Tα-S* bound to the DEAE Sepharose column was eluted using a 0-1 M NaCl gradient in the elution buffer. Tα-S* was then concentrated using a Spin-X UF column (Mr 10,000 cutoff, Corning). The buffer was changed to K-gluc buffer containing 0.005% (v/v) Tween 20 using a Superdex 75 PC 3.2/30 column (ÄKTAmicro system, GE Healthcare). Purified Tα-S* was stored at −80 °C until use. Purity and the concentration of Tα-S* were assessed with SDS-PAGE, and the purity was almost 100%. All of the manipulations for extraction and purification were performed at 4 °C.
Expression and purification of recombinant PDEγ. DNA sequence of frog PDEγ (GenBank Accession Number AB578858.1) was inserted into NdeI/BamHI sites of expression vector, pET-3a (Novagen). PDEγ was expressed in E. coli BL21(DE3) pLysS strain (Novagen) after induction with IPTG for 3 hr at 30 °C. Purification of expressed PDEγ was carried out based on the method described previously 20,21 . Purified recombinant PDEγ was lyophilized, dissolved in K-gluc buffer and stored at −80 °C until use.
Measurement of binding of pDeγ or pDe to immobilized tα-s* and that of tα-s* to immobilized pDeγ. Proteins were immobilized on the SPR sensor tip according to the protocol described by Biacore.
To immobilize Tα-S* on the sensor chip, Tα-S* was first biotinylated at its thiol groups. For this purpose, 3.3 μl of 2 mM EZ-Link Maleimide-PEG2-Biotin (Thermo Fisher Scientific) was added to 100 μl of 13 μM of purified Tα-S* in K-gluc buffer without DTT, and the mixture was incubated for 1 hr on ice. After the incubation, 0.35 μl of 1 M DTT was added to reduce and deactivate the non-reacted maleimide group of Maleimide-PEG2-Biotin. Then, the buffer was changed to K-gluc buffer to remove the deactivated Maleimide-PEG2-Biotin using a Zeba Spin Desalting Column (Thermo Fisher Scientific). Purified biotinylated Tα-S* was immobilized on a streptavidin (SA) sensor chip (GE Healthcare) through the streptavidin-biotin interaction. There are 8 thiol groups in www.nature.com/scientificreports www.nature.com/scientificreports/ bovine Tα (NCB Accession # NM_181022.2). However, it is known that only one group is chemically modified by N-ethylmaleimide in the GTP-bound form of Tα and that this modification does not affect PDE activation 22 , and therefore the binding to PDEγ. Frog Tα also contains 8 thiol groups (NM_001090561.1), and our Tα-S* showed almost a single component in the binding to PDEγ (see below).
Recombinant PDEγ was immobilized at its lysine amino groups on a carboxymethylated dextran (CM5) sensor chip (GE Healthcare) using 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC) and N-hydroxysuccinimide (NHS). Unreacted NHS-ester was blocked with ethanolamine after the immobilization. There are 8 lysine residues in PDEγ, and all of them are at the region outside of the major binding site of PDEγ to Tα-S* and to PDEcat, but four of them are near the possible binding site of PDEγ to Tα*and PDEcat. However, these four did not seem to affect the binding (see below).
The binding of a protein to the immobilized protein was measured using Biacore X100 (GE Healthcare) at 25 °C. The buffer used was K-gluc buffer containing 0.005% (v/v) Tween 20. Binding signals were stored and processed in Biacore X100. We used two ways to record the binding, one with binding and dissociation both terminated before their completion (Figs 2 and 4b) and the other after their completion (Fig. 4a).
When necessary, the binding data were analyzed by BIAevaluation software (GE Healthcare) to determine K D2 . The programs used are designed to include one of the crucial effects, mass transport effect (MTL) 23 . For the analysis of binding signals of PDEγ to immobilized Tα-S*, we used Heterogeneous Ligand with MTL program with assuming that there are at least two populations of Tα-S* immobilized differently depending on which thiol site was immobilized. However, our analysis indicated that the binding of PDEγ to immobilized Tα-S* consisted of only one major component (>98%). Unfortunately, the binding signal of Tα-S* to immobilized PDEγ was not analyzed with this program, and instead, 1:1 binding with MTL program was used. However, as shown in Fig. 4b, 1:1 binding with MTL program gave a very good fit to the measured binding signals, which indicated that the binding consisted of one major component.
Determination of K D1 of the pDeγ·pDecat complex with dilution. To determine K D1 , PDE activity was measured using purified PDE or ROS membranes at various concentrations of PDE (≤40 nM), both in the light without GTP. The activity was measured with the pH assay method using a combination glass microelectrode (MI-410, Microelectrodes, Inc.) as described previously 11,12,24 . At time 0, 5 mM cGMP was added to initiate the hydrolysis. All measurements were performed at room temperature. To measure the full PDE activity, PDEγ bound to PDEcat was digested with trypsin (final concentration, 0.1 mg/ml) for 5 min at room temperature, and the digestion was terminated by adding trypsin inhibitor at a final concentration of 0.5 mg/ml. Then, the full PDE activity measurement was initiated with adding 5 mM cGMP.

pDe activation with tα-s* of various concentrations. PDE activities at various concentrations of
Tα-S* were measured with the pH assay method in a solution and in a ROS membrane suspension. In a solution, Tα-S* of known concentration was added to 15 nM purified PDE in K-gluc buffer in the light. At time 0, 5 mM cGMP (final concentration) was added to initiate cGMP hydrolysis. In a ROS membrane suspension, first