The double mutant channel D413C_R297C can be locked in the open state

R297 is highly conserved in all known CNG channels, whereas D413 is conserved in all CNGs with the exception of CNGA4, a subunit that fails to form functional channels when expressed in a heterologous system (Fig. 2A, Supplementary Table ¹)³³. Analysis of the Cryo-EM structure of TAX-4 channel captured in the open state (PDB ID:5H3O) located R297 in the intracellular portion of S5 (Fig. 1a)²². In order to verify the involvement of D413 and R297 in the gating process, we constructed the double mutant D413C_R297C. The mutant channel D413C_R297C inactivated, following the application of 1mM cGMP. Within 1–2minutes the cGMP-activated current was reduced by about 50% (Fig. 2B,C), but it recovered its original amplitude after cGMP removal (Fig. 2D, for details see figure legend), as in mutant channels D413A and R297A (Fig. 1c). Remarkably, we found that a longer exposure (about 10minutes) to cGMP irreversibly locked the channel in the open state (Fig. 2E). The double mutant channel D413C_R297C was quickly and irreversibly locked in the open state by the simultaneous exposure to 1mM cGMP and the cross-linkers MTS-2-MTS (Fig. 2F,G) or MTS-4-MTS (Fig. 2H,I); in this case, the current flowing through locked-open channels had an amplitude similar to and often larger than the cGMP-activated current before inactivation. Locking -in the open state - is associated to a potentiation and a strong reduction of the current fluctuations (Fig. 2J,K). In the locked open state, collected data indicate a reduction of the variance of current fluctuations σ² of approximately 2 folds (please compare 2 and 3 in panel Fig. 2K). When channels fluctuate between an open and closed state σ² is proportional to the closed probability P_c which is halved in the locked open state. In CNGA1 channels and in the presence of saturating cGMP concentration the P_o is close to 0.9¹⁵ and, therefore, in the locked open state P_o becomes close to 0.95. Analysis of the power spectrum of these fluctuations shows that the Lorentzian component with a cut-off frequency of around 1600 Hz is reduced when the electrostatic interactions between D413 and R297 are replaced by a S-S bond between residues in the same locations (Supplementary Fig. ³ and Fig. 2K), suggesting that fluctuations of the interactions between the C-linker and S5 are a source for the channel noise observed in saturating cGMP concentrations. SMFS experiments with the double mutant channel D413C_R297C in the presence of 1mM cGMP indicated that an intrasubunit bond between the two exogenous cysteines can occur (Supplementary Fig. ⁴).

Open in a separate window

Figure 2

The double mutant channel D413C_R297C can be locked in the open state. (A) Web logo analysis of the sequence surrounding the conserved R297 and D413 residues (marked by an arrow). (B–E) Inactivation and locking in the open state of the double mutant channel D413C_R297C. in symmetrical Na⁺ solution in the presence of 1mM cGMP at±60mV immediately after the membrane patch excision (B) and after 1minute (C). The recovery (D) of the current after 1minute in the closed state. Irreversible lock of the current in the open state (E) after a prolonged exposure to cGMP (10minutes). (F–G) MTS-2-MTS locking effect in the open state. cGMP-activated currents immediately after the membrane patch excision (F) and after 2minutes of 100 μM MTS-2-MTS application in the open state (G). (H–I) As in (F) and (G), but in the presence of 100 μM MTS-4-MTS. Red/black lines represent the current evoked by 0/1mM cGMP. (J) Enlargements of boxes in (H) and (I). (K) Box plot summarize the quantitative noise analysis in the absence (1: n=6 at +60mV and n=6 at −60mV) and in the presence of 1mM cGMP, before (2: n=6 at +60mV and n=6 at −60mV) and after (3: n=6 at +60mV and n=6 at −60mV) application of MTS-4-MTS. The horizontal line within each box indicates the median of the data; boxes show the twenty-fifth and seventy-fifth percentiles of the data; whiskers show the fifth and ninety-fifth percentiles of the data.

These experiments suggest that, when the electrostatic interactions between R297 and D413 are disrupted, CNGA1 channels inactivate, but when residues in position 297 and 413 are covalently linked by a S-S bond, CNGA1 channels do not inactivate and are locked in the open state. The next question is how do electrostatic interactions between R297 and D413 lead to the widening of the pore lumen? Two mechanisms could be in action: similarly to what occurs in usual voltage gated Na⁺ and K⁺ channels^22,23 rearrangements of the S4-S5 linker and of the lower portion of S5 are transmitted to S6, and its motion and conformational changes are an essential component of the gating of CNGA1 channels. An alternative view is that the electrostatic interactions between R297 and D413 induce significant rearrangements of S5, which play a major role in the gating of CNGA1 channels, as opposed to what happens in Na⁺ and K⁺ channels.

SMFS experiments show that S5 moves during gating and that loops change their folding

By using methods of SMFS, we unfolded CNGA1 channels from their C-terminal end³⁴: the resulting F-D curves are composed of a series of force peaks, indicating that an unfolded region or a poorly structured polypeptide was in the process of being unfolded. During the unfolding of the transmembrane domain, a.a. were unfolded sequentially so that it was possible to identify the location of the unfolded region. The location of a force peak was obtained by fitting the experimental F-D curve with the Worm-Like Chain (WLC) model^35,36, providing the values of the contour lengths (Lc). From the value of Lc it was possible to estimate the number of a.a. of the polypeptide unfolded between the occurrence of two consecutive force peaks (as ΔLc/0.4nm)³⁶. In CNGA1 channels the unfolding of the transmembrane domain occurs − both in the open and closed states - following the force peaks with an Lc of about 116nm − corresponding approximately to N400 located at the base of S6 (Fig. 3A). In the closed state (red traces in Fig. 3A) there is a force peak corresponding to residues ranging from R297 to F280 which, in the open state (blue traces in Fig. 3A), splits into two force peaks corresponding approximately to W330 and K260. These changes indicate that in the closed state S5 is mechanically coupled to S6, but in the open state S5 is mechanically coupled to S4³⁴. These data, in addition, also show major conformational changes during gating at the N- and C-terminal end of S5 as well as at the loops connecting S5 to the P-helix and S5 to the S4 helix. These changes can be deduced from the computation of the histograms of the increase of ΔLc from the force peak corresponding to N400 to the following force peak in the closed (Fig. 3B) and open states (Fig. 3C). In the closed state, the values of ΔLc ranged from 90 to 120 a.a. indicating that residues in the loop connecting S5 to the S4 - from approximately F280 to R297 - have a loose folding (Fig. 3D). In the open state, in contrast, the values of ΔLc ranged from approximately 60 to 80 a.a. so that the polypeptide chain from S320 to G340 in the extracellular loop, connecting S5 to the P-helix are poorly structured and can be unfolded with a low force (Fig. 3C). Therefore, during gating, major conformational changes occur primarily in S5 and in the connecting loops: in the open state the loop connecting S5 to the P-helix is loosely folded but the loop connecting S5 to S4 provides mechanical coupling (Fig. 3D,E).

Open in a separate window

Figure 3

Flexibility in the loops between S4-S5 and S5-P-Helix during CNGA1 channels gating. (A) Superposition of F-D curves from unfolding of CNGA1 in the closed (n=10) (CNGA1-CS, light and dark red) and open (n=10) (CNGA1-OS, blue) states. The solid line represents the WLC fitting to the corresponding unfolded peaks, and the numbers represents the approximate peak position in the a.a. chain. Inset represents the ∑Lc histogram for the peak at N400 (290 a.a.), implying a conserved unfolding step for both open and closed states. (B) Distribution of unfolded length for the domains S6–S5 for CNGA1-CS (n=110). (C) As in (B) but for CNGA1-OS (n=71). (D–E) Cartoon representation of the S5-S6 domains, showing the change in flexibility between the intracellular and extracellular loops of S5 for closed and open state, respectively.

An inspection of the Cryo-EM structure of TAX-4 channel (Supplementary Fig. ⁵) reveals that P293 is in the short loop connecting S4 to S5, where it acts as a helix initiator. Therefore, we constructed several mutant channels in position 293 with different intracellular helical-initiating inclination. Electrophysiological experiments indicate that all these mutant channels show an unusual gating and permeation which is more pronounced for the P293A mutant (Supplementary Fig. ⁶ and Supplementary Fig. ⁷). An altered gating and permeation caused by mutations at the intracellular side of S5 indicate that in the open state S5 is coupled to the pore walls and its conformational changes influence the filter structure. All together, these observations clearly suggest that gating is mediated by the coupling of molecular events at the N-terminal end of S5 to the selectivity filter.

Cysteine scanning mutagenesis shows a state dependent accessibility of S5 and of the P-helix

In order to obtain further evidence of the pivotal role of S5 in the gating of CNGA1 channels, we then analyzed the conformational changes during gating of S5 and of the P-helix by Cysteine Scanning Mutagenesis (CSM). We scanned the residues in the intracellular (from R297 to I308) and extracellular portions (from S320 to D328) of S5 and in the P-helix (from R345 to S359) comparing the effect of MTSET in the closed and in the open states. According to TAX-4 structure residues R297, L301 and Y304 are located in the intracellular portion of S5 (Fig. 4A) and residues A322 and I323 are located at the extracellular end of S5 (Fig. 4B). As suggested by our SMFS experiments (Fig. 3) during gating, these residues are expected to undergo conformational changes. The cGMP-activated current for mutant channels R297C was abolished within 1–3minutes, by MTSET in the closed state but not in the open state (Fig. 4C). Blockage by 0.5 and 2.5mM MTSET had similar effects both in amplitude and time course (Fig. 4D), indicating that the residue in position 297 is easily accessible by intracellular MTSET. The effect of intracellular MTSET on mutant channels L301C was rather variable and collected data (n=8 closed state; n=8 open state) did not show a clear and significant state-dependent blockage by MTSET (Fig. 4I, Supplementary Table ^2c). Mutant channels Y304C were potentiated by intracellular 2.5mM MTSET in the open state, but not in the closed state (Fig. 4E). Potentiation caused by 0.5mM MTSET was significantly lower than that observed with 2.5mM (Fig. 4F), suggesting that residues in position 304 are less accessible than residues in position 297. Mutant channels A322C and I323C (Fig. 4B) had different reactivity to MTSET: mutant channel A322C was significantly potentiated by extracellular MTSET in the open but not in the closed state (Fig. 4G,J). Mutant channels I323C were blocked both in the open and closed states by extracellular but not by intracellular MTSET (Fig. 4I,J and Supplementary Table ^2c,d). The effect of MTSET on the other tested mutant channels was either not state-dependent or could not be distinguished from what observed in the WT (Fig. 4I,J, - here named as C314 WT CNGA1 channels - Supplementary Fig. ^1b–e, Supplementary Table ^2c,d). The state-dependent effect of MTSET on mutant channels R297C, Y304C and A322C supports the notion that during gating there are conformational rearrangements − possibly associated to a motion of S5 related to the P-helix - involving the whole S5-helix. Indeed, according to the TAX-4 structure, R297, L301 and A322 span the entire S5 helix: these residues are located near the intracellular end, around the center, and at the C-terminal end of S5, respectively (Fig. 3A,B). The high density of hydrophobic residues in the extracellular portion of S5 and in the P-helix suggests that hydrophobic interactions could transmit the conformational rearrangements of S5 to the P-helix. Therefore, we performed a CSM also in the P-helix and we found 2 mutant channels V348C and L351C, which were blocked by extracellular MTSET in the closed, but not in the open state (Fig. 4J)¹⁰. Residues V348 and L351 are coupled to the pore walls and their motion is likely to cause the widening of the pore lumen.

Open in a separate window

Figure 4

Cysteine scanning mutagenesis shows a state dependent accessibility of S5. (A) Mapping CNGA1 residues on TAX-4 structure (PDB ID: 5H3O) shows the location of key residues in the intracellular portion of S5. (B) as in (A) but for extracellular portion of S5. Key residues are highlighted with red spheres. Different regions (S4–S5 and P-helix) are illustrated in different colours. (C) Comparison between the cGMP-activated currents at±60mV for the mutant channels R297C before (black) and after (grey) 3minutes application of 2.5mM MTSET to the intracellular side (MTSET _i) in the closed (c.s.) and open (o.s.) states. (D) Plot representing the comparison of MTSET _i at different concentration (n=4 for 0.5mM and n=5 for 2.5mM) in the closed state for mutant channels R297C; histograms are the average±s.e.m. and circles indicate data from single experiments. (E) as in (C) but for the mutant channels Y304C. (F) as in (D) but for the mutant channels Y304C in the open state (n=4 for 0.5mM and n=7 for 2.5mM). (G,H) as in (E) and (F) but for the mutant channels A322C with the addition of 2.5mM MTSET to the extracellular side (MTSET _o) and with different concentration of MTSET (n=3 for 0.5mM, n=4 for 1.5mM and n=7 for 2.5mM). (I,J) Histograms representing the average±s.e.m. for many mutant channels, circles indicate data from single experiments after the application of 2.5mM MTSET in the open (red) and closed (black) states. The number (n) of experiments for different mutant channels in different states varies between 3 and 8 (see also Supplementary Table ²). (n.e.) represents mutant channels without expression. Significativity is shown only for mutants where the MTSET effect was >30%. Unpaired two–tail T-test in i for R297C and Y304C: t₇=−7,529, P<0.001; t₈=−2,649; P<0.05. Unpaired two–tail T-test in j for A322C, V348C and L351C: t₈=−3,668, P<0.01; t₁₂=−4,544, P<0.001; t₆=−3,109, P<0.05. Further statistical tests for these mutant channels and for all other tested mutants are reported in Supplementary Table ^2c,d. 1-I/Imax represents the normalized current where I is the residual current measured in the presence of 1mM cGMP after application of MTSET (in different conditions) and Imax is the cGMP-activated current at the beginning of the experiment.

The blockage and potentiation of MTSET reached a steady and complete effect within 5 minutes^28–30 but with some consistent variability: in the closed state, blockage of mutant channels R297C occurred in less than 3minutes and often was completed in 1–2minutes but potentiation of mutant channels A322C, in the open state, required 3–5minutes. These different action times are likely to be associated to the accessibility of mutated residues to MTSET^28–30.

The insertion of S-S bonds between S5 and the P-helix alters channel gating

We also constructed mutant channels with one exogenous cysteine in S5 and another cysteine in the P-helix, with the aim of establishing an S-S bond between S5 and the P-helix, possibly locking these mutant channels in the open state. Based on the high-resolution Cryo-EM structure of TAX-4 we built a model of the CNGA1 channel in the open state which shows that S5 and the P-helix are knitted together by a combination of an hydrogen-bonding network, with hydrophobic interactions involving Y316 (S5) and W353 (P-helix) (Fig. 5A–C). I309, H310, A313, Y316, F317 in S5 form a hydrophobic patch with L343, K346, Y347, L351, L358 in the P-helix that, together with the H-bonds, could be responsible for transmitting S5 conformational changes to the P-helix during ligand gating. These residues are highly conserved in all CNG channels, consistent with their key role in the coupling between S5 and the P-helix (Supplementary Table ¹).

Open in a separate window

Figure 5

The double mutant channel F317C_Y347C is locked in the open state. (A,B) Mapping CNGA1 residues on TAX-4 structure reveals the presence of hydrophobic interactions. (C) Enlargement of panels a and b illustrating the network of H-bonds and hydrophobic π-interactions. (D) left: current recordings obtained in the absence of cGMP (0 cGMP) for mutant channels F317C_Y347C, elicited by voltage steps from −200 to +200mV (ΔV=20mV); right: after the addition of 1mM cGMP. (E) left: current recordings obtained in the absence of cGMP; right: 1minute after the addition of 1µM dequalinium. (F,G) Current recordings obtained in the absence of cGMP following subtraction of those currents recorded in the presence of dequalinium (F) or after P/-4 procedure for leak and capacitive artefact subtraction (G). (H) Dependence of G/Gmax on V with 1µM dequalinium and in the absence of cGMP (n=4, black) and in the presence of 1mM cGMP (n=3, red). (I) single channel recordings obtained in the absence of cGMP for mutant channels F317C_Y347C, elicited at different voltages. Amplitude histograms are shown at the right of each trace. isc, single-channel current amplitude. (K) current recordings obtained from the WT CNGA1 channels in the presence of 1mM cGMP, elicited by the same voltage steps as in (D). (L) Dependence of G/Gmax on V obtained from the recordings of WT CNGA1 channels (K) in the presence (n=1) of DMA.

As F317 and Y347 could be two of the residues mediating the coupling between S5 and the P-helix, we constructed the double mutant channel F317C_Y347C in which a S-S bond was inserted between these two domains. When the mRNA for these double mutant channels was injected in some oocytes, no cGMP-activated current was recorded but the I/V relations measured in excised patches - in the absence of cGMP - were outwardly rectifying (Fig. 5D,E). This result is in contrast with what is usually recorded in uninjected oocytes (Supplementary Fig. ^8a,b) suggesting that the double mutant channels F317C_Y347C could have some unusual properties and could be locked in the open state.

To verify whether these mutant CNGA1 channels were locked in the open state, we added 1µM of Dequalinium to the intracellular medium, a powerful and specific blocker of CNGA1 channels³⁷: after addition of Dequalinium, the I-V relations became linear within 60seconds (Fig. 5E) indicating that the F317C_Y347C mutant channels were locked in the open state, as also demonstrated by the application of DTT (Supplementary Fig. ^8f). The Dequalinium sensitive current, obtained as the difference of the current initially measured either in the presence (Fig. 5E, left panel) or absence of cGMP minus the current measured in the presence of Dequalinium (Fig. 5E, right panel), was highly nonlinear (Fig. 5F). In some experiments (Fig. 5G) we recorded a rectifying current in which at positive voltages the current clearly increased with a time constant of about 1ms. The relation between G/Gmax (Fig. 5H) of the Dequalinium sensitive current showed a significant voltage dependency. The voltage dependence of current recordings (Fig. 5G,H) obtained from the mutant channel F317C_Y347C in the absence of cGMP is reminiscent of what observed in some kind of K2P K⁺ channels, such as the TRAAK and TREK-1 channels and could originate, partially, from similar mechanisms¹⁸. In TRAAK and TREK-1 channels the voltage dependent gating is attributed to the simultaneous presence of three or four ions inside the selectivity filter, whereas, in CNG channels − because of the shorter size − only two ions are likely to be simultaneously present in the pore¹⁸.

We recorded also single channel openings at different voltages (Fig. 5I) and the values of the single channel current (isc) were 8 and 4.8 pA and the Po was 0.54 and 0.21 at the membrane voltages of +140 and +100mV, respectively. The dependence of G/Gmax of the double mutant channel F317C_Y347C (Fig. 5H) locked in the open state was very similar to what seen in the WT CNGA1 channels in the presence of large organic cations such as dymethylammonium (Fig. 5K,L) or etylammonium¹⁴. The similarity suggests a common molecular mechanism, i.e. the motion of the voltage sensor S4 and its coupling to S5. Comparable results were also obtained with the double mutant channels W330C_Y347C, but in this case the I/V relations of the Dequalinium sensitive current were less rectifying (Supplementary Fig. ^8c–e).

Overall our CSM analysis clearly indicates that the S5-helix undergoes significant conformational changes during gating. When an S-S bond between S5 and the P-helix is inserted, the mutant channel is locked in the open state and becomes voltage gated, similarly to what observed in WT CNGA1 channels in the presence of large organic cations^13,14.

Molecular biology

The CNGA1 channel from bovine rods consisting of 690 a.a. was used²¹. Selected residues were replaced as previously described¹⁰ using the Quick Change Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA, USA) or ordered from Genscript, 860 Centennial Ave., Piscataway, NJ 08854, USA. Point mutations were confirmed by sequencing, using a LI-COR sequencer (4000l; LI-COR Biosciences, Lincoln, NE, USA). cDNAs were linearized and were transcribed to cRNA in vitro using the mMessage mMachine kit (Ambion, Austin, TX, USA).

Oocyte preparation and chemicals

Mutant channel cRNAs were injected into Xenopus laevis oocytes. Oocytes were prepared as previously described¹⁰. Injected eggs were maintained at 18°C in a Barth solution supplemented with 50μg/ml gentamycin sulfate and containing (in mM): 88 NaCl, 1 KCl, 0.82 MgSO₄, 0.33 Ca(NO₃)₂, 0.41 CaCl₂, 2.4 NaHCO₃ and 5 Tris-HCl, pH 7.4 (buffered with NaOH). During the experiments, oocytes were kept in a Ringer solution containing (in mM): 110 NaCl, 2.5 KCl, 1 CaCl₂, 1.6MgCl₂ and 10 Hepes, pH 7.4 (buffered with NaOH). Usual salts and reagents were purchased from Sigma Chemicals (St Louis, MO, USA).

Application of sulfhydryl-specific reagents

Sulfhydryl-specific reagents effect was tested using a concentration of 0.5, 1.5 and 2.5mM. MTSET links covalently to only one cysteine. Cross-linker compounds such as MTS–2–MTS (1,2-Ethanediyl bismethanethiosulfonate) and MTS–4–MTS (1,4-Butanediyl bismethanethiosulfonate) had different maximum cross-linking spans — i.e. the longest distance between the S atoms of the cross-linker reacting with the S atoms of cysteine and forming S–S bonds⁴⁷. These compounds, in contrast, could link to two cysteines. The cross-linker MTS–2–MTS had a cross-linking span of 5.2Å while MTS–4–MTS had a cross-linking span of 7.8Å. The different effect of reagents able to link to one cysteine - such as MTSET, MTSPT, MTSPtrEA - and those able to link to two cysteines - M-X-M cross-linkers - with the same size and volume was used to discriminate whether blockage was caused by bridging two cysteines by the binding of the compound to one end only. MTS compounds were purchased from TRC (Toronto Research Chemicals, Canada). In some selected experiments, the application, in the bath or in patch pipette solution, of 2mM dithiothreitol (DTT) was used to reduce the formation of disulfide bonds²⁹.

Recording apparatus

cGMP-gated currents from excised patches were recorded with a patch-clamp amplifier (Axopatch 200; Axon Instruments Inc., Foster City, CA, USA), 2–6 days after RNA injection, at room temperature (20–24°C)³⁰. The perfusion system allowed a complete solution change in less than 0.1 seconds^10,27. Macroscopic and single-channel current recordings were obtained with borosilicate glass pipettes which had resistances of 2–5 MOhm in symmetrical standard solution. The standard solution on both sides of the membrane consisted of (in mM) 110 NaCl, 10 Hepes and 0.2 EDTA (pH 7.4). Solutions were buffered with tetramethylammonium hydroxide at the desired pH. When the cation X⁺ was used as the charge carrier, NaCl in the standard solution on both sides of the membrane patch was replaced by an equimolar amount of the cation X⁺. We used Clampex version 10.0 for data acquisition. Recordings to perform noise analysis were low-pass filtered at 10kHz. All other macroscopic current recordings were low-pass filtered between 1 to 10kHz.

Data analysis

Single-channel currents (i) were estimated from patches containing only one channel and fitting normalized all-point histograms with two-component Gaussian functions. Data in this manuscript are presented as mean±SEM (V-I relationships), dot-plots (with means±SEM) or box-and-whisker when data sets need to be statistically compared. n indicates the number of patches.

Data analysis and figures were made with Clampfit version 10.1 (Molecular Devices, Sunnyvale, CA, USA) and Sigmaplot 12.5 (Systat Software, Chicago, IL, USA).

We excluded data when we lost the patch during the experiments, when the level of expression was too low and we could not distinguish the noise from random channel openings, from electrical noise due to an unstable patch and/or from spurious electrical noise. We kept only data obtained during experiments in which the amplitude of the seal (i.e. the current evoked by voltage pulses in the absence of cGMP) was stable. These criteria were not used for those experiments describing the locking of channels in the open state (see Fig. 2). All these criteria were established prior to data collection.

Statistical significance for parametric analysis was determined using unpaired two-tailed T-test or single-variable ANOVA, as specified in Supplementary table ². For pairwise comparisons, a Holm–Sidak test was used following ANOVA. P<0.001 (***), P<0.01 (**) and P<0.05 (*) were considered significant, as indicated. No statistical methods were used to predetermine sample sizes, but our sample sizes are similar to those reported in previous publications^28–30,32. In this study particular care was given to MTSET/MTS-2-MTS reactivity experiment interpretation, as “run-ups” attributed to de-phosphorylation of the channel by endogenous patch-associated phosphatases are known to occur in CNGA1 channels⁴⁸. Indeed, we often observed a slow time-dependent “run-up” or “run-down” of the cGMP-gated current (after 5minutes, up to 10% of the current initially recorded) in control conditions. Therefore, to avoid that such non-specific effect might be attributed to MTS reactivity, in this study only changes (either current potentiation or blockage) greater than 30% were considered. These effects are usually highly significant (P<0.001, see Figs 4, ​,6,6, ​,77).

SMFS experiments and data processing

The NanoWizard 3 AFM system (JPK) and an inverted optical microscope (Olympus IX71) were used under liquid conditions in Standard solution (in mM: 110 NaCl, 10 HEPES and 0.2 EDTA pH 7.4 buffered with NaOH) with or without 2mM cGMP. Rectangular silicon nitrite gold-coated cantilevers (HYDRA2R-50NGG from APPNANO) were used to localize plasma membrane patches and to perform SMFS experiments. SMFS experiments were performed using membrane extracted from injected oocytes expressing constructs (P293A and D413C-R297C) at a high level. Oocytes were attached to a mica substrate and clean fragments of membrane remained anchored to the substrate with the intracellular side exposed to the bathing medium and to the cantilever tip of the AFM. We used the worm-like chain (WLC) model where Lp is the persistence length (equal to 0.4nm), a parameter that represents the stiffness of the molecule. For each tip-sample separation (D or TSS) value, the WLC model was used to compute the corresponding value of Lc that was obtained by solving the third order polynomial: 4λ³+ωλ² −1=0, where λ=1 − D/Lc and ω=4F(D,Lc)/α − 3 and α=kbT/Lp. Each portion of the F-D curve was fitted perfectly by the WLC model³⁸. For details please see Maity et al.³⁶.

We acknowledge the financial support of the NEUROSCAFFOLDS Project n. 604263. A.M.’s work was supported by a Long Term EMBO Fellowship (ALTF 1427-2014) and Marie Curie Action (LTFCOFUND2013, GA-2013-609409). We thank S. Onesti for supervising the sequence alignment analysis. We thank F. Romagnoli and M. Lough for checking the English.