The TRPM1 channel in ON-bipolar cells is gated by both the α and the βγ subunits of the G-protein Go

Transmission from photoreceptors to ON bipolar cells in mammalian retina is mediated by a sign-inverting cascade. Upon binding glutamate, the metabotropic glutamate receptor mGluR6 activates the heterotrimeric G-protein Gαoβ3γ13, and this leads to closure of the TRPM1 channel (melastatin). TRPM1 is thought to be constitutively open, but the mechanism that leads to its closure is unclear. We investigated this question in mouse rod bipolar cells by dialyzing reagents that modify the activity of either Gαo or Gβγ and then observing their effects on the basal holding current. After opening the TRPM1 channels with light, a constitutively active mutant of Gαo closed the channel, but wild-type Gαo did not. After closing the channels by dark adaptation, phosducin or inactive Gαo (both sequester Gβγ) opened the channel while the active mutant of Gαo did not. Co-immunoprecipitation showed that TRPM1 interacts with Gβ3 and with the active and inactive forms of Gαo. Furthermore, bioluminescent energy transfer assays indicated that while Gαo interacts with both the N- and the C- termini of TRPM1, Gβγ interacts only with the N-terminus. Our physiological and biochemical results suggest that both Gαo and Gβγ bind TRPM1 channels and cooperate to close them.

In mammalian retina, an increase in light intensity hyperpolarizes the photoreceptor and initiates two opposing signals: sign-preserving synaptic transmission to the OFF bipolar cells and sign-inverting transmission to the ON bipolar cells. In darkness, the depolarized photoreceptors tonically release glutamate into the synaptic cleft, hyperpolarizing the ON bipolar cells. Light hyperpolarizes the photoreceptors, reducing glutamate in the cleft and causing the ON bipolar cells to depolarize. The key steps in this 'sign inverting' cascade are: glutamate activates the ON bipolar cell's mGluR6 receptor [1][2][3] , and this activates the heterotrimeric G-protein G o that comprises α o β 3γ 13 [4][5][6][7][8][9][10] . Active G o closes the non-selective cation channel TRPM1 (melastatin), thought to be constitutively active [11][12][13][14] . In the retina, TRPM1 is required for night vision as mutations in its gene or autoimmune targeting of the protein lead to lack of the ERG b-wave and to night blindness [15][16][17][18] . Outside the retina, two splice variants of TRPM1 regulate pigmentation in melanocytes, and loss of this gene is correlated with tumor aggressiveness in human melanoma [19][20][21] .
While there is strong evidence that active G o closes the TRPM1 channel, it is not clear if this closure is caused by an active Gα o , or a free Gβ γ dimer. Evidence indicating that Gα o induces this closure is based on studies that transfected TRPM1 into CHO cells and found that applying activated Gα o purified from the brain to an excised patch closed the channel, but applying Gβ γ did not. These studies also found that co-transfecting CHO cells with TRPM1 and constitutively active Gα o rendered the TRPM1 channels inactive 12 . Evidence indicating that Gβ γ causes TRPM1 closure is based on results from several cell types, including bipolar cells, where dialyzing Gβ γ reduced the mGluR6-initiated response, but dialyzing an activated form of Gα o did not. Further support comes from transfected HEK cells and human melanocytes where Gβ γ rather than active Gα o reduced a Ca 2+ signal triggered with high extracellular Ca 2+ 22 . These contradicting data could result from the use of different cell lines that express different endogenous molecules that impact the channel. Indeed, activating endogenous mGluR6 in melanocytes opens the channel instead of closing it; but after expressing Gα o by transfection, mGluR6 activation closes the channel 23 . To better understand how the TRPM1 channel operates in retinal ON bipolar cells, we carried out a battery of experiments in which we dialyzed reagents that modify the status of the endogenous G-protein in rod bipolar cells and observed their effects on the basal current and on the light response. We further examined the interaction of G-protein subunits with TRPM1 using co-immunoprecipitation and energy transfer assays. Our results suggest that both Gα o and Gβ γ bind the TRPM1 channels and together cooperate to close them.

Results
Our experiments involved blocking K + and Cl − channels, then dialyzing reagents into rod bipolar cells clamped at − 60 mV, and then testing over time the reagents' effects on the holding inward current. Given that most channels are blocked and only TRPM1 is likely to be gated by G o , the change in holding current indicates whether the TRPM1 channels are opening or closing in response to the dialyzed agent. We chose to monitor the holding current as the read out of the reagents' effects rather than the size of the mGluR6-initiated response (as done by Shen et al.) because the mGluR6 response is likely to decrease regardless of whether channels are opened or closed by the reagent.
In light-exposed retinas, dialyzing active Gα o closes the TRPM1 channel. In order to test if Gα o closes the channel, we had to first open the channels. Therefore, as soon as we established a seal, we provided a light step and then tested if the dialyzed solution closed the channel (see example in Fig. 1A). Every 35 sec we provided a voltage ramp to test slope conductance, and then briefly turned off the light to monitor OFF responses. Each of these 35-sec-protocol repetitions is referred to as a sweep, and a typical experiment had 5 sweeps. In general, the light step produced a large transient increase of the inward current followed by a smaller sustained current (Fig. 1A). Because our experiments required recording periods longer than the diffusion time (time scale of sec), and because BAPTA does not prevent the adapting fall off of the light response (as it only affects the response in the msec range), it was necessary to compare the sustained current that was achieved 13 sec after break-in. To compare this light-evoked sustained current before and after dialysis, we averaged the currents over 0.5 sec at the 1 st and 5 th sweeps (Fig. 1B). To determine if the dialyzed solution caused a statistically significant change, we applied the paired Students t-test. This test computes the probability that the average difference between the two time points is equal to 0 (i.e., it computes the probability that the dialyzed agent has no effect). We also provided voltage ramps to monitor changes in the slope conductance; this was computed in the linear range between − 95 mV and − 65 mV (to avoid contributions from the voltage-activated L-type calcium channel) (Fig. 1C). In control experiments (with only basic pipette solution), the inward sustained current remained relatively stable at around − 34 pA (n = 14 cells; Table 1). Then, to test our method, we confirmed established observations that dialyzing GTPγ S closes TRPM1 24,25 . When 50 μ M GTPγ S was perfused, only 5 out of 13 cells gave a light response. Since cells without responses could not be tested for channel closure, we increased the yield by switching to 25 μ M GTPγ S which evoked light responses in 6 out of 7 cells. Combining results from both concentrations, we found that the sustained current started at − 37.3 ± 7.1 pA, and as expected, greatly decreased during dialysis. At the 5 th time point, the holding sustained current was − 16.1 ± 4.1 pA, significantly lower than that at the 1 st time point (p < 0.01 paired Student's t-test; Table 1). This indicates that GTPγ S closes the channels that were opened by light. To illustrate the net effect of each dialyzed agent, we plotted the average difference in the sustained currents between the 5 th and the 1 st time points. Positive values indicate channel closure because the inward current becomes less negative (Fig. 1D).
Next we tested the effect of dialyzing a constitutively active mutant of Gα o . Previous experiments testing this effect activated the Gα o subunit with GTPγ S or with GMP-P(NH)P 12,14,22 . These approaches may have led to an excess of the non-hydrolysable GTP analogue in the solution, making it unclear whether the observed effect was caused only by active Gα o or also by activation of Gβ γ . Therefore we took the approach of producing a constitutively active mutant of myristoylated Gα o (myrGα o -QL) and incubating it with GTP prior to introducing it into the cells. This QL mutation prevents GTP from being hydrolyzed, thus profoundly shifting its conformation to the active state in the presence of GTP 26,27 . We found that after dialyzing 40 nM of myrGα o -QL (with GTP as in the control solution), the light-evoked sustained current significantly dropped from − 34.3 ± 6.3 to − 26.1 ± 4.4 pA (n = 17, p = 0.029) ( Fig. 1D and Table 1).
We then dialyzed wild type myristoylated Gα o (100 nM), also after incubating it with GTP. Unlike the QL mutant, WT myrGα o hydrolyzes all bound GTP within about 1-2 minutes 28 and hence is expected to be largely inactive in its GDP-bound form when delivered into the cells. For WT myrGα o , the sustained current remained stable throughout the experiment (ranging between − 29 pA and − 31 pA; n = 17), as it did with the control solution (Fig. 1D). If a decrease in sustained current indicates that channels are closing, these current changes should correlate with changes in the slope conductance. Indeed, increases in sustained current were highly correlated with increases in slope conductance (R = − 0.92; the minus sign results from inward currents being assigned negative values).
To confirm that these conductance changes were due to TRPM1 modulation, in several experiments we extended the light OFF period to 4 sec and provided a voltage ramp during this period. We then subtracted for each sweep the I-V curve during light OFF from that during light ON, thus isolating the contribution of TRPM1 to the measured current at different voltages. At the first sweep, TRPM1 contribution was significant (supp. Fig. 1), but at the 5 th sweep it was practically null, indicating that TRPM1 was closed and that the main difference seen during light ON was due to TRPM1. Thus we conclude that the closure of TRPM1 observed when dialyzing myrGα o -QL is due to Gα o 's active state. While we cannot compare the effect of Gα o -QL to that of GTPγ S because of the difference in concentrations, diffusion properties, and nature of the activity, it appears that the effect of  active myrGα o is smaller than that of GTPγ S. If so, this smaller effect may be due to an additional effect by Gβ γ (which is also activated by GTPγ S).
Sequestering Gβγ opens the TRPM1 channel in the dark. To test the effect of Gβ γ on TRPM1, Shen et al. dialyzed recombinant Gβ 1γ 2 or native Gβ γ subunits purified from the brain into rod bipolar cells 22 . They found that Gβ γ reduced responses to light and to mGluR6 antagonists, suggesting that Gβ γ closes the channel and prevents it from opening. In apparent disagreement, applying Gβ γ to an inside-out excised patch of TRPM1-transfected CHO cells did not change the probability of channel opening 12 . To address these opposite findings, we tested the contribution of Gβ γ in rod bipolar cells using the approach of inhibiting the endogenous Gβ γ by sequestering it with Gα o -GDP or phosducin, a 28 kDa phospho-protein that binds Gβ γ and thus inhibits activity of the free dimer [29][30][31][32] . These experiments were performed on dark-adapted cells to induce a state in which Gβ γ is dissociated from the endogenous Gα o . The cells were clamped at − 60 mV, and the dark holding current (which we term basal current) was measured at different time points after break-in (see example in Fig. 2A). Every 35 sec during the recording period, we provided a voltage ramp in darkness to test slope conductance followed by a strong 10 msec light flash to test the cell's ability to produce a light response ( Fig Table 2 and Fig. 2D). We wondered if longer dialysis would make a difference, so for some experiments we measured the current at the 10 th time point. For myrGα o -QL, the holding current remained similar (from − 38.4 ± 6.6 to − 37.7 ± 5.3 pA; n = 16; p = 0.88) (sup Fig. 2), but for 40 nM WT myrGα o , the current continued to increase (− 67.4 ± 10.9 at the 10 th time point; p = 0.03) (supp Fig. 2). These findings indicate that the opening of the TRPM1 channels by wild type Gα o is due to its GDP-bound state.
Next, we tested the effect of dialyzing phosducin, a reagent that offers the advantage of preventing Gβ γ from interacting with effectors without affecting the activation state of Gα and without changing its concentration 33 . When 9 μ M phosducin was added to the pipette solution, the basal inward current progressively increased from − 29.9 ± 5.9 pA to − 45.3 ± 7.3 pA ; n = 14; p = 0.01) (Fig. 2D). This change in basal current highly correlated with the change in slope conductance: while this conductance decreased a little for control and myrGα o -QL, it increased for both WT myrGα o and phosducin. The correlation between basal current and slope conductance was − 0.91 (Fig. 2C,E), supporting the notion that an increase in basal current indicates channel opening. Thus, our results show that dialyzing reagents that sequester Gβ γ opens TRPM1 channels, suggesting that Gβ γ closes the channel.
Linoleic and myristic acids do not modulate TRPM1. In the experiments above, we used myristoylated forms of Gα o because native Gα o harbors this post-translational modification that is important for its normal association with the membrane 34 . However, because lipids are well known to mediate or modulate gating of TRP channels [35][36][37] , we tested if the myristoyl group present on myrGα o may contribute directly to channel opening in our experiments by using non-myristoylated Gα o where the myristoylation signal at the N-terminus was replaced with a His 6 affinity tag (His-Gα o ). Similar to myrGα o , 150 or 300 nM His 6 -Gα o increased the basal inward current (from − 7.9 ± 2.7 pA to − 28.7 ± 7.0 pA; n = 9; p = 0.013) as well as the slope conductance (Fig. 2D,E).
Next, since puffing certain lipid modifiers on the extracellular face of the plasma membrane can modulate certain TRP channels [38][39][40] , we further tested TRPM1 modulation by puffing alpha linoleic acid (LNA, 20-100 μ M; 5 cells) or myristic acid (MA, 100-250 μ M; 8 cells). We found that although these cells responded to light, they  When channels are opening, the inward current increases (has a more negative value) and the difference between the 5 th and the 1 st time points is negative. (E) Quantitative analysis of changes in slope conductance for each of the above conditions. *indicates a significant difference (p < 0.05) between the 1 st and 5 th time points, and **indicates a highly significant difference (p < 0.01).
did not respond to the lipid modifiers (Fig. 3), indicating that the main mechanism for channel opening is solely through the G-protein.   Fig. 4). This binding was equivalent across all conditions, suggesting that the association of TRPM1 with Gα o is independent of the activity state of Gα o . No binding was observed in the absence of TRPM1, indicating that this interaction is specific. We further studied the association of G o subunits with the TRPM1 channel using a Bioluminescence Resonance Energy Transfer (BRET) assay. In this approach, cytoplasmic N-terminal and C-terminal domains of TRPM1 were fused with a highly efficient energy donor (Nluc) and paired with the Gβ γ or Gα o subunits fused with Venus, the fluorescent acceptor (Fig. 5A,B). In these experiments we used a prototypic Gβ γ pair, Gβ 1γ 2, for its functional equivalence and effectiveness in gating the TRPM1 Channel 22 . To direct the TRPM1 fragments to the plasma membrane where the G-protein subunits are naturally found, the constructs were further appended with an engineered membrane localization sequence. When Gβ γ was co-transfected with the N-terminus of TRPM1, the acceptor/donor titration experiments revealed a hyperbolic profile of the BRET signal that saturated at an acceptor/donor ratio of about 1 (Fig. 5C). In contrast, when Gβ 1γ 2 was combined with the C-terminus of TRPM1, the BRET signal was not different from the shallow linear signal observed with membrane-targeted Nluc luciferase (Fig. 5C,D). This suggests that under our experimental conditions only the N-terminus of TRPM1 specifically interacts with Gβ γ .
When Gα o was used as an energy acceptor, both the N-terminus and the C-terminus fragments produced significant BRET signals, but the N-terminus gave a stronger signal (Fig. 5E,F). These results indicate that both Gα ο and Gβ γ interact with the TRPM1 channel, and they further localize the site of interaction: while Gα o interacts with both ends of TRPM1, Gβ γ appears to interact only with the N-terminus.

Discussion
We present evidence that both Gα o and free Gβ γ play a role in modulating the TRPM1 channel open-state. Furthermore, the close association of these subunits with TRPM1 suggests that the actions of the G-protein subunits are direct rather than acting via a second messenger. To our knowledge, this is the first example of a TRP channel that is directly gated by both arms of a G-protein. TRPC4 has been shown to interact with Gα i2 , but not with Gβ γ 41 .
Role of Gα o . The critical piece of evidence that supports the role of Gα o is our finding that dialyzing a constitutively active mutant of this subunit during light exposure leads to channel closure. Under a prolonged strong light stimulus, the majority of the G-protein must be in its inactive form where Gβ γ is bound to Gα o .GDP. Interaction between TRPM1 and G-protein subunits was studied upon co-transfection into HEK293T cells. Cells were lysed and TRPM1 was precipitated by specific antibodies against TRPM1. Proteins present in the lysates and IP eluates were detected by Western blotting. Approximately 6-fold more material was loaded for the eluates relative to the amount of material present in the lysates. Quantification of protein content revealed 70-100% IP efficiency for TRPM1, 10-12% for Gα o and 7-14% for Gβ 3 across samples with no significant differences between nucleotide states. The shown Western blots were cropped at the expected molecular weight; both Gα o and Gβ 3γ 13 were precipitated regardless of added nucleotides. Because the dialyzed Gα o -QL does not affect Gβ γ , the observed channel closure is attributed to the dialyzed reagent and not to Gβ γ . Since dialyzing wild type Gα o did not change the channel open-state, we conclude that Gα o. GTP contributes to channel closure. This conclusion agrees with experiments showing that application of GMP-P(NH)P-activated Gα o to an excised patch of TRPM1-transfected CHO cells closes the channel, and that transfection with Gα o Q205L renders the channel closed 14 . The idea that active Gα o contributes to channel modulation is further supported by our findings that in HEK cells, Gα o .GTP physically associates with TRPM1. Based on these experiments and the excise patch experiment 14 , we suggest that the action of Gα o is direct.

TRPM1 open-state requires associated proteins. TRPM1 is thought to be constitutively open because
in ON bipolar cells its closure requires activation of G o . If so, in situations when G o is naturally inactive, such as in rod bipolar cells lacking mGluR6, TRPM1 channels are expected to stay open. Contrary to this expectation, we previously found that the resting membrane potential in rod bipolar cells lacking mGluR6 is more hyperpolarized than in WT cells by about − 15 mV, and the holding current is similar to that of TRPM1-KO 14,42 . This suggests that TRPM1 requires additional components to stay open. This requirement for an associated protein has also been proposed in two other studies. In the first, it was suggested because TRPM1 was shown to lack 4-fold symmetry 43 , characteristic of channel oligomerization seen for TRPV1 and other TRP channels, and in the second becasue capsaisin could not stimulate heterogeneously expressed TRPM1 44 . One possibility is that Gα o is this auxiliary protein, and four lines of evidence support this idea. (1) In melanocytes, which natively do not express Gα o , TRPM1 appears to be constitutively closed and stimulation of mGluR6 opens the channel. When the cells are transfected with Gα o , activation of mGluR6 closes the channel 23 . (2) In rod bipolar cells lacking Gα o1 , TRPM1 seems closed 8 even though the channel is expressed in the dendritic tips as in WT cells, and Gβ γ is practically absent 45

. (3) Gα o .GDP interacts with TRPM1 (this study). (4) Dialyzing Gα o .GDP is extremely efficient in open-
ing the channel (this study), suggesting that some of the observed effect can be due to direct interaction and not only to sequestering Gβ γ . While Gα o may be an auxiliary protein, it is unlikely the only one that support channel opening since ON bipolar cells lacking mGluR6 still express Gα o 42,46 yet the TRPM1 channels are closed. At least two proteins are required for stable expression of TRPM1 in the membrane, nyctalopin and LRIT3, and in their absence rod bipolar cells are unresponsive to light and the channel is closed or absent from the dendritic tips 47,48 . Thus it is possible that these two proteins, mGluR6, and/or other unknown components contribute not only to trafficking and stable expression, but also to maintaining open TRPM1 conformations.

Cooperation between Gα and Gβγ.
While the classical view of G-protein function is that upon GTP/ GDP exchange, Gα .GTP activates its effector, there are several examples where the effector is activated by Gβ γ 49,50 . Interestingly, most known Gβ γ effectors, including adenylyl cyclase, PLCβ , and the G-protein-gated inwardly rectifying potassium channel (GIRK), are also effectors for Gα ; i.e., Gα cooperates with Gβ γ to modulate the downstream activity [51][52][53][54][55] . The GIRK channel provides an especially interesting example because being a channel, the function and interaction of the G-protein subunits with it can readily be compared to their function and interaction with TRPM1. It is well known that GIRK is directly gated by Gβ γ 56,57 , but the function of Gα i/o in the GIRK-Gα β γ complex is emerging more slowly. It is now understood that the non-activated Gα i3 has 3 independent functions: it reduces the basal current of GIRK, enhances the evoked current, and regulates its kinetics [58][59][60] . Both Gα i3 .GDP and Gα i3 .GTP interact and regulate GIRK1/2 61 . Thus, the analogy of the interaction of the G-protein with the GIRK channel to that with TRPM1 holds on several levels, but the effects of the G-protein subunits are opposite. While Gβ γ opens GIRK, it closes TRPM1. While Gα i/o .GDP reduces the basal GIRK current, Gα o .GDP seems to increase the basal TRPM1 current. In either case, both Gα .GDP and Gα .GTP are retained in the complex, and Gα .GTP works synergistically with Gβ γ 61 . The simplest model that can explain the TRPM1-G-protein interaction (summarized in Fig. 6) is that Gα o .GDPβ γ binds TRPM1 and endows it with an open conformation; when GTP replaces GDP, Gβ 3γ 13 dissociates from Gα o .GTP and both arms twist TRPM1 and change its conformation to the close state. Gβ γ binds to TRPM1 probably via its N-terminus, while Gα o may bind both ends of TRPM1. We speculate that Gα o swings from one terminus to the other upon nucleotide exchange; when GDP-bound, it joins Gβ γ at the N-terminus, and when GTP-bound, it binds the C-terminus. It is important to remember that the macromolecular complex must contain other proteins as well since Gα o .GDPβ γ must also bind mGluR6 to be activated, and Gα o .GTP must also bind the GAP complex to be deactivated.

Materials and Methods
Ethical approval. Procedures involving animals were performed in accordance with National Institute of Health guidelines and the protocol was reviewed and approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania and the competent ethics committees at Jinan University. C57BL/6J wild type mice (WT) were purchased from Charles River laboratories. A mouse was deeply anesthetized by intraperitoneal injection of a mixture of 100 μ g/gm ketamine and 10 μ g/gm xylazine; the eyes were enucleated and the mouse was euthanized by anesthetic overdose.
Whole cell recording experiments. Recording. Retinal slices were prepared as described previously 62 .
Briefly, retinas were isolated under red light and cut into 200 μ m thick slices with a tissue slicer (Narishige, Japan). The slices were transferred to a recording chamber, secured with vacuum grease and then moved to the microscope stage of an Olympus microscope equipped with a 60x water immersion objective. The chamber was perfused at a rate of 0.5-1 ml/min with oxygenated (95% O 2 , 5%CO 2 ) Ames medium (Sigma, St. Louis, MO) containing sodium bicarbonate (1.9 g/l) and 2 μ M Strychnine and 100 μ M picrotoxin (to block GABA A/C receptors) at 32-34°C.
Patch pipettes with resistances of 7-9 MΩ were fabricated from borosilicate glass using an electrode puller (Sutter, Novato, CA). Pipettes were filled with the following solutions (in mM): 108 Cs-gluconate, 10 BAPTA, 10 Scientific RepoRts | 6:20940 | DOI: 10.1038/srep20940 HEPES, 10 NaCl, 4 MgATP and 1 LiGTP. The pH was adjusted to 7.4 with KOH and the osmolality was 290 mOsmol. All chemicals were obtained from Sigma. The solution was aliquoted, stored at -20 °C and thawed before each experiment. For each set of experiments, we patched several cells with electrodes that contained the above control solution, and several with electrodes that also contained a modifier reagent as explained in Results.
Source of modifier reagents. Myristoylated Gα o1 (hence referred to as myrGα o ) was obtained from Calbiochem and used for most experiments; myrGα o -Q205L and its control myrGα o were prepared as described below; GTPγ S was obtained from Sigma (Sigma-Aldrich, St Louis, MO); phosducin was a gift from Dr. Vadim Arshavsky (Duke University); and His 6 -Gα o was a gift from Dr. Richard Neubig (University of Michigan).
Current recordings in the US were obtained with an Axopatch 1D amplifier (Molecular Devices) and in China with an EPC− 10 patch clamp amplifier (HEKA, Lambrecht, Germany). Membrane potentials were corrected for liquid junction potential calculated to be ~15 mV. Cells were discarded if the baseline current in the dark exceeded − 100 pA at a holding potential of − 60 mV. Voltage command generation and data acquisition were accomplished with Clampex (Molecular Devices) or PatchMaster (HEKA). Cells were voltage clamped at − 60 mV and the holding current and light-evoked current responses were compared over time for control and different dialyzed reagents. It has been reported that clamping a cell at + 50 mV may extend the recording time because less calcium-dependent desensitization occurs 63 . However, in our hands, such clamping did not prove beneficial, and we preferred to make the measurements under more physiologically-relevant voltages.
Light stimuli. The retina was stimulated using a green full-field light generated by a light emitting diode with a peak wavelength of 565 nm (or 500 nm in China). During dark adaptation, the light response was tested with a 10 ms flash (3.8 × 10 4 photons/μ m 2 /s) that was turned on every 35 s. To measure changes in slope conductance, a voltage ramp from − 95 mV to + 45 mV was applied every 35 s. During light adaptation, a background illumination with an intensity of 1.1 × 10 4 photons/μ m 2 /s was applied immediately after break-in. To test the OFF response, the light was turned off for 1 or 4 s every 35 s. Slope conductance was measured every 35 s while the light was ON. In some cases another ramp was applied during light OFF.
Analysis. For each cell, the baseline current was calculated every 35 s as the average of current for 0.5 s before light ON for dark-adapted retinas or light OFF for light-adapted retinas. The light response was measured as the peak response to a flash of light (ON response). The slope conductance was measured from the linear range between − 95 mV to − 65 mV. Waveform analysis of the response was done off-line with Clampfit (Molecular Devices). All data are reported as mean ± SEM (Standard Error of the Mean). The values of the holding current and conductance at different time points after break-in were compared to those at the first sweep using paired Student's t-test using Excell. Differences were considered significant when p ≤ 0.05. For p values above 0.01, we report the actual value, and for values below 0.01, we simply state p < 0.01.
Expression and purification of the myristoylated forms of rat Gα o1 and Gα o1 −Q205L subunits (hence referred to as myrGα o and myrGα o −QL). For Gα o bacterial expression vector, the full-length rat Gα o1 (or Gα o1 -Q205L mutant) cDNA was fused to a C-terminal His 6 tag and cloned into pET21 (Novagen). pHV738 (a gift from Dr.Richard Kahn, Emory University), a vector which contains the human N-myristoyltransferase 1 (hNMT1) and E. coli methionine aminopeptidase (map) genes, was used to express In the dark, mGluR6 is bound to glutamate and G o is active (depicted in green). Both active Gα o and free Gβ γ contribute to channel closure, and both can interact with TRPM1. We speculate that Gβ γ is bound to the N-terminus and Gα o.GTP to the C-terminus; both twisting the channel to close it. When light decreases glutamate, and hence deactivates G o (depicted as red), the heterotrimer G o is reformed and allows the channel to open. Not shown in the model are other proteins that must interact with this complex to allow these activities to cycle; in particular, mGluR6 must bind the inactive G o , and the GTPaseactivating complex must bind Gα o .GTP. the eukaryotic N-myristoylation machinery 64 . To increase N-myristoylation efficiency, the E. coli formylmethionine deformylase (def) gene additionally was inserted in pHV738, and the resulting plasmid was called pHV738/def. The E. coli strain BL21-CodonPlus(DE3)-RIPL (Stratagene) was simultaneously transformed with pET21-Gα o WT/His 6 (or pET21-Gα o -QL/His 6 ) and pHV738/def. The transformed cells were grown in terrific broth (TB) supplemented with ampicillin (50 mg/ml), kanamycin (25 mg/ml), and chloramphenicol (34 mg/ml). Bacteria were inoculated and allowed to grow at 37 °C; when the culture reached OD 600 of 0.6, the temperature was reduced to 25 °C. At OD 600 of 0.8, 50 μ M sodium myristate (Sigma # M8005) was added, and 10 min later induction of gene expression was triggered with 1M isopropyl-β -D-thiogalactopyranoside (IPTG) (Invitrogen). Cells were incubated under gentle agitation and harvested 16 hours later by centrifugation. Cell pellets were resuspended and sonicated on ice for 4 min (5 sec on/15 sec off). The crude lysate was clarified by centrifugation and purified (at 4 °C) by chromatography over Ni-NTA. MyrGα o was eluted with 10 mM EDTA in 50 mM Tris-HCl, pH 7.5. The eluted protein fractions were collected and analyzed by SDS-PAGE. Fractions containing myrGα o were pooled, concentrated, and further purified on a Superdex 200 column (GE Healthcare). The purity of myrGα o was confirmed by SDS-PAGE gel analysis. After purification, the biological activity of recombinant Gα o subunits was tested by [ 35 S]GTPγ S binding. On average, 1 mole of myrGα o bound 0.4-0.6 moles of GTPγ S. N-myristoylation, determined by mass spectrometry, showed 80-85% of the protein to be myristoylated. The protein was stored at − 80 °C. Further details about purification and testing of the material will be published elsewhere.
Co-immunoprecipitation Assays. HEK293T cells were grown in six-well plates and transfected with Lipofectamine LTX (Invitrogen). Transfected plasmids encoded the following constructs (per well): 0.42 μ g Gα o , 0.42 μ g Gβ 3, 0.42 μ g Gγ 13 and 1.25 μ g TRPM1 or 1.25 μ g empty pcDNA3.1 vector. After 24 hours, cells were harvested and lysed by sonication in ice-cold PBS IP buffer (150 mM NaCl, 1% Triton X-100, 5 mM MgCl 2 and Complete protease inhibitor tablets) supplemented with three different compositions (GDP: 0.01 μ M GDP; GDP/AlF 4 : 0.01 μ M GDP, 10 mM NaF and 0.02 mM AlCl 3 ; GTPγ S: 0.01 mM GTPγ s). Lysates were cleared by centrifugation at 14,000 rpm for 10 minutes. The supernatant was incubated with 20 μ l of 50% protein G slurry (GE Healthcare) and 3 μ g sheep anti-TRPM1 antibody on a rocker at room temperature for 1 hour. After three washes with the indicated IP buffer, proteins were eluted from beads with 50 μ l of 2X SDS sample buffer. Proteins retained by the beads were analyzed with SDS-PAGE, followed by Western blotting using HRP conjugated secondary antibodies and an ECL West Pico (Thermo Scientific) detection system. Signals were captured on film and scanned by densitometer.
BRET experiments. HEK293T/17 cells were cultured at 37 °C and 5% CO 2 in DMEM supplemented with 10% fetal bovine serum, MEM non-essential amino acids and 1 mM sodium pyruvate. Cells were plated at a density of 50,000 cells/well in a white 96 well plate with clear bottom (Greiner Bio-One) and transfected using Lipofectamine LTX (Invitrogen) and PLUS TM Reagent (Invitrogen). Cells were co-transfected with a fixed concentration of Nanoluc-fused constructs (donors) and increasing concentrations of Venus-fused constructs (acceptors). Empty vector was used to balance the amount of transfected DNA. Readings were obtained 24 h after transfection, immediately following media exchange to PBS containing 0.5 mM MgCl 2 and 0.1% glucose and Nanoluc substrate (Nano-Glo, Promega) diluted 1:100. Fluorescence (Venus; 535 nm with 30 nm band path width) and luminescence (Nanoluc; 475 nm with 30 nm band path width) emissions were recorded simultaneously in real time with a microplate reader (POLARstar Omega, BMG Labtech) equipped with two photomultiplier tubes. The BRET signal was calculated as the ratio of the light emitted by acceptor over the light emitted by the donor. The ratio of emissions at acceptor and donor wavelengths from donor-only samples has been subtracted. All measurements were performed at room temperature.