Mutational analysis to explore long-range allosteric coupling and decoupling in a pentameric channel receptor

Pentameric ligand-gated ion channels (pLGICs) mediate chemical signaling through a succession of allosteric transitions that are yet not completely understood. On the prototypic bacterial channel GLIC, we explored the conformational landscape of the protein during pH-gating. To this aim, we introduced a series of allosteric mutations, and characterized the protein conformation over a broad pH range. We combined electrophysiological recordings, fluorescence quenching experiments monitoring key quaternary reorganizations, and simulations by normal mode analysis. Moderate loss-of-function mutations and the allosteric modulator propofol displace allosteric equilibria involved in pre-activation and pore opening processes, highlighting long-range allosteric coupling between distant regions of the protein. In contrast, total loss-of-function mutations stabilize the protein in unique intermediate conformations where motions are decoupled. Altogether, our data show that the protein can access a wide conformational landscape, raising the possibility of multiple conformational pathways during gating.


Introduction 1
Pentameric ligand gated ion channels (pLGICs) mediate fast synaptic 2 communication in the brain. In mammals, this family includes the excitatory nicotinic 3 acetylcholine and serotonin receptors (nAChRs and 5-HT3Rs) as well as the 4 inhibitory γ-aminobutyric acid (GABA) and glycine receptors (GABAARs and GlyRs) 5 (Jaiteh et al., 2016). pLGICs are also present in bacteria, notably with the pH-gated 6 channels GLIC (Bocquet et al., 2007) and sTeLIC (Hu, Nemecz, et al., 2018), the 7 GABA-gated channel ELIC (Zimmermann and Dutzler, 2011), and the calcium-8 modulated DeCLIC (Hu et al., 2020). 9 pLGICs physiological function is mediated by alternating between different 10 allosteric conformations in response to neurotransmitter binding. Seminal work in 11 the 80s showed that a minimal four-state model describes the main allosteric 12 properties of the muscle-type nAChR (Heidmann and Changeux, 1980; Sakmann 13 et al., 1980). The ability of ACh binding to activate the nAChR involves a resting-to 14 active-state transition. In addition, prolonged ACh occupancy promotes a biphasic 15 desensitization process underlying the transition to so-called fast-and slow-16  Those events are at the heart of the protein's function, allowing the coupling 29 between the neurotransmitter site and the ion channel gate which are separated by 30 a distance of 5 nm in pLGIC structures. 31 The past decade has seen great advances by structural biology to seek 32 understanding about the molecular mechanisms involved in gating (Nemecz et al., 33 and Farrens, 2014). Bimane/quencher pairs on GLIC combined with kinetic analysis 23 allowed us to characterize pre-activation motions occurring early in the 24 conformational pathway of activation (Menny et al., 2017). Indeed, they occur at 25 lower proton concentrations than pore opening, and are complete in less than a 26 millisecond, much faster that pore opening that occurs in the 30-150 millisecond 27 range in electrophysiology (Laha et al., 2013). 28 Here, we exploit the TrIQ approach to explore the conformational landscape 29 of GLIC during pH-gating, in combination with allosteric ligands and mutations. To 30 help interpreting the fluorescence quenching data into structural terms, we first built 31 atomistic models of the various bimane-labeled proteins, and simulated their gating 32 transition pathways using coarse grain modeling and normal mode analysis. We  In the A trajectory, the twist motion occurs in the first half of the trajectory 10 ( Figure 1B). This motion describes opposite rotations between ECD and TMD 11  Pore opening is also associated with, at the bottom of the ECD, a contraction of the 21 β-sandwich (evaluated by a decrease in distance between Cα of residues Asp32 22 and Gly159; Figure 3A & 3C), as well as quaternary reorganization around loop 2 23 (measured by a decrease in inter-subunit distance Cβ Lys33/Trp160, Figure 3-24 Supplement 1). In addition to these two consecutive global motions, the progressive 25 quaternary compaction of the ECD, another crucial landmark of GLIC 26 reorganization, occurs throughout the trajectory. This compaction is quantified 27 through measurement of inter-subunit distances (between Cβ Asp136/Gln101 and 28 Arg133/Leu103; Figure 3A, 3B & 3-Supplement 2), indicating a progressive 29 decrease in distance throughout the frames. It is noteworthy that these inter-subunit 1 distances are highly variable, due to the asymmetric nature of the ECDs of the GLIC-2 pH7 structure, where each subunit β-sandwich presents a unique orientation as well 3 as relatively high B-factors (Sauguet et al., 2014). This variability decreases over 4 the frames to reach the structure of GLIC-pH4 which is compact and essentially 5 symmetric. 6 Trajectory B shows substantially the same components but with an inverted 7 sequence of events, pore opening and associated motions start first, followed by 8 the twist motion in the last frames, the compaction of the ECD being spread over 9 the whole trajectory. In conclusion, iMODfit generate two distinct trajectories that 10 are equally plausible to describe a gating transition of activation of GLIC.

Y197 as reporter of key allosteric motions of GLIC 14
To directly relate the conformational reorganizations generated by iMODfit to  (and quencher when necessary) were performed and the bimane moiety was 24 docked into each frame while keeping its carbon at a covalent-bond compatible 25 distance to the sulfur atom of the cysteine. The distance between the centers of 1 mass of the bimane and indole/phenol moieties in each frame was then measured 2 to follow the evolution of the distance between the bimane and quencher. 3 For the ECD quenching pair Bim136-Q101W, the simulations show that 4 Bim136 and the indole ring of Trp101 are separated in the resting-like state, and are 5 in close contact in the active-like state ( Figure 3A). These observations are in good 6 agreement with the previously published fluorescence data that show a decrease in 7 fluorescence intensity upon pH drop indicating a decrease in distance between the 8 bimane and quencher. The A trajectory shows a progressive decrease in distance 9 that parallels the ECD quaternary compaction movement described previously (Cβ 10 Asp136/Gln101), whereas the B trajectory shows a more variable pattern, and a 11 sharper distance decrease only in the last frames ( Figure 3B). 12 For the ECD-TMD interface quenching pair Bim250-Y197, our simulations 13 show that Bim250 is in close contact with the phenol ring of Tyr197 in the resting-14 like state, while both moieties are separated in the active-like state, the bimane 15 moiety moving on the other side of loop 2 ( Figure 2A). This is also in agreement with 16 the fluorescence recordings that showed an increase in fluorescence upon pH-drop 17 indicating that the Bim250 is moving away from its quencher Tyr197. Both A and B 18 trajectories show an abrupt change in Bim250/Y197 distances, corresponding 19 respectively to a late versus early separation, and these changes occur during the 20 outward motion of the M2-M3 loop (P250Cα/Y197O distance; Figure 2D). 21 In contrast, for Bim135-W72 ( Figure 3D), the distances between the centers 22 of mass of the bimane and indole moieties are not correlated with the distances 23 between the residue's backbone, while the pH-dependent changes in fluorescence 24 show a bell shape curve suggesting important but complex changes in distances at 25 this level. Since at this position bimane occupies a rather buried location within the 26 protein structure, we suggest that these discrepancies come from local 27 reorganization of surrounding residues that are not directly correlated with the 28 movement of the backbone. Interestingly, we previously solved the X-ray structure 29 of Bim135-W72 at pH4 by crystallography, which shows a similar location of the 30 bimane moiety with that of our docking in GLIC-pH4 ( Figure 3-Supplementary 3). 31 In conclusion, for Bim136-Q101W and Bim250-Y197 quenching pairs, the 32 simulations of the end states show good agreement with steady-state fluorescence 33 data. Interestingly, kinetic analysis show that the conformational motions monitored 34 by both pairs report on a pre-activation transition that occurs well before channel 1 opening ( Figure 4A; Menny et al., 2017). This observation is compatible with 2 trajectory A, for which the quenching of Bim136-Q101W occurs throughout the 3 trajectory from frame 1 to 10, the unquenching of Bim250-Y197 occurs between 4 frame 10 and 11, while the full reorganization of the pore and associated motions 5 are completed only in the very last frame of the trajectory. The pre-activation 6 scheme is clearly not compatible with trajectory B, that shows an inverted order of 7 events. 8 To explore the conformational landscape of GLIC and of its mutants, we thus 9 monitored in parallel the ECD quaternary compaction with the Bim136-Q101W 10 fluorescent sensor and reorganizations at the TMD level using electrophysiology or 11 the fluorescent sensor Bim250-Y197 reporting on M2-M3 loop motions. In specific 12 cases, we additionally used Bim135-W72 to monitor tertiary motions in the ECD. 13 The quaternary compaction of the top of the ECD is strongly allosterically 14

coupled with the lower part of the ECD interface 15
To allow an accurate comparison between mutants, we first measured 16 is functional in electrophysiology, but does not undergo pH-dependent quenching 30 (figure 4C). We also confirm that pH-dependent quenching curves for Bim136-31 Q101W and Bim250-Y197 display higher sensitivity (especially for Bim250-Y197) 32 and lower cooperativity than the pH-dependent activation curves recorded by 1 electrophysiology ( Figure 4C).

12
We first investigated allosteric mutants located at the inter-subunit interface 13 in the lower part of the ECD ( Figure 5A). We previously showed that E26Q produces to the Bim136-Q101W (F0 = 0.71). This suggests that substantial quenching is 1 present at neutral pH and E26Q not only alters the allosteric transition, but also 2 modifies the conformation of the resting state itself which appears to be more 3 compact when the E26Q mutation is present.

15
Another mutation, Y28F two residues apart, was reported to produce a 16 moderate gain of function (Nemecz et al., 2017). Surprisingly, mutating Y28F in the 17 Bim136-Q101W-C27S background yields a drastic loss of function characterized by 18 a slow activating receptor and a marked decrease in pH50 ( Figure 5B and 5C). We 19 verified that this loss of function is due to the combination of C27S and Y28F 20 mutations since, without the endogenous cysteine mutation to serine, Bim136-21 Q101W-Y28F electrophysiological response is similar that of Bim136-Q101W 22 (Table 1). In fluorescence, the quenching curve of Bim136-Q101W-Y28F-C27S also 1 shows a large decrease in pH50, associated with an apparent higher cooperativity. 2 Again, in this case, the ∆pH50 are in the same range in fluorescence quenching (-3 2.2) and in electrophysiology (higher than -1.5, the plateau could not be reached 4 with this mutant preventing accurate measurement of the pH50). 5 We thus identify here the lower part of the b-sandwich as a region of the 6 protein that controls the quaternary structure of the ECD at pH7 and following pH 7 decrease. The quaternary compaction of the top of the ECD, monitored with the 8 Bim136-Q101W sensor, thus appears strongly coupled in an allosteric manner with 9 the lower part of the ECD interface.    Performing these mutations on the Bim136-Q101W-C27S background 7 shows overall a conservation of their phenotype, with a 10-fold (D32E and E222Q) 8 and more than 30-fold (H235Q) decrease in the pH50 of activation as compared to 9 Bim136-Q101W-C27S ( Figure 6B and 6C). The fluorescence quenching curves are 10 also shifted to lower pH50s, with ∆pH50s of 3-fold (D32E and E222Q) and 10-fold 11 (H235Q) ( Figure 6D).

22
The quenching data thus reveal an allosteric coupling between both ends of 23 the protein, since the structural perturbations performed around the TMD are 24 transmitted to the top of the ECD, impairing its compaction. However, as opposed 1 to the ECD mutations E26Q and Y28F/C27S, these mutations have a stronger effect 2 on the pH50 of the electrophysiological response as compared to fluorescence 3 quenching. It thus suggests that both processes are not fully coupled for mutations 4 further away from the sensor site. 5

Total loss of function mutations decouple ECD and TMD allosteric motions 6
To further explore the conformational landscape accessible to GLIC, we 7 extended the analysis to mutations known to totally prohibit channel opening (  L157A nor L246A significantly alter the movement at Bim250, which occurs with a 20 complete amplitude and no change in pH50. Thus, unlike the moderate loss of 21 function mutants investigated above, these mutations decouple the protein motions, 22 partially impairing quaternary compaction of the ECD but allowing full motion of the 23

M2-M3 loop. 24
The mutation H235F leads to a phenotype opposite to that of L157A or 25 L246A. It allows for a full compaction of the ECD, since its Bim136-Q101W pH-26 dependent curve shows a full quenching amplitude and a decrease in pH50.   increases that of Bim136-Q101W-H235Q by more than half a unit ( Figure 8D). Data 5 thus show that propofol do not act locally by altering the conformation of the TMD, 6 but rather acts on the global allosteric transitions by displacing the equilibrium 7 between resting and active conformation and preserving coupling between ECD and 8 TMD.

18
We also investigated propofol in the Bim135-W72 H235Q mutant.  Therefore, the fluorescence data related to the Bim135-W72 pair cannot be 3 interpreted in structural terms, however they further document the strong allosteric 4 coupling with the upper part of the ECD. 5

Discussion 6
Long-range allosteric coupling associated with pre-activation and pore-7 opening processes. 8 In this study, the fluorescence quenching and electrophysiological data are 9 recorded on an ensemble of GLIC proteins, on which pH-dependent changes lead 10 to a cascade of allosteric reorganizations from the resting to the active and 11 potentially desensitized conformations. 12 We found that a series of five loss-of-function mutations, which shift the pH-13 dependent electrophysiological curves to higher concentrations, also shift the pH-  Our previous kinetic analysis showed that the conformational motions 19 followed by fluorescence occur early in the pathway of activation ( Figure 4A). 20 Following a rapid pH-drop using a stopped-flow device, these "pre-activation" 21 motions are complete in less than a millisecond, much faster than pore opening that  Since pre-activation precedes pore-opening, it is likely that a shift in pre-activation 24 (followed here by fluorescence) will be reflected as a parallel shift in activation 25 (followed by electrophysiology). ECD mutations E26Q and Y28F/C27S present such 26 a phenotype of paralleled variations in electrophysiology and fluorescence 27 quenching pH50 which indicates that those mutations mainly impact the pre-28 activation transition. In contrast, ECD-TMD interface and TMD mutations D32E, 29 E222Q and H235Q lead to a stronger pH50 shift in electrophysiology than in 30 fluorescence quenching. This phenotype suggests that these mutations alter not 31 only the pre-activation, but also the downstream pore-opening transitions leading to 1 an additive effect on the function. 2 In the same line, the quenching data related to the allosteric modulation of 3 propofol suggest its major effect on the pre-activation transition. The emerging 4 picture is thus that pre-activation, that is associated with complete motions at the 5 various quenching pairs, involves a global reorganization of the protein. The low 6 cooperativity of the fluorescence curves potentially suggests that more than one 7 state might be involved in this process. In contrast, the downstream pore-opening 8 process would involve a more local reorganization that would be restricted to the 9 TMD (E222Q, H235Q, pore opening) and the lower part of the ECD (D32E). We 10 speculate that the slowness of this process may arise from the wetting of the pore, 11 particularly its upper region, which is highly hydrophobic and for which hydration 12 should be energetically costly. 13 To investigate the structural dynamics of the protein, we generated two in 14 silico conformational trajectories between the resting-like and active-like X-ray  Unlike a mutation to glutamine, mutation of His235 to phenylalanine prevents this 30 critical interaction leading to a non-functional receptor (Prevost et al., 2012). We 31 show here that following a pH decrease H235F undergoes a complete "quenching 32 motion" of the ECD, but only a partial "unquenching motion" of the M2-M3 loop. The 33 mutation thus seems to block the protein in an intermediate state in the A trajectory, 1 where the pH-elicited motions of the ECD are complete but not transmitted to the 2 TMD ( Figure 9). Interestingly, the structure of the H235F mutant was previously 3 solved by crystallography at pH 4 in a "locally closed" (LC, 3TLT) conformation 4 (Prevost et al., 2012). This conformation is characterized by a fully active-like 5 structure of the ECD, but a resting-like structure of the TMD including the M2-M3 6 loop and a closed channel. This structure fits well with the quenching data of H235F 7 at Bim136-Q101W, but not with that of Bim250-Y197 where the fluorescence 8 quenching data indicate a partial movement of M2-M3 that is not seen in the X-ray 9 structure. Inspection of the structure of H235F shows however a movement of the 10 quencher Y197 which side chain is re-oriented toward the transmembrane domain, 11 away from Bim250, a feature that could plausibly account for this partial 12 Another important conclusion from our data is that the protein has access to 33 an unanticipated repertoire of conformations, and NMA analysis suggests that these 34 conformations might contribute to different allosteric pathways of pre-activation. The 1 idea that GLIC can follow different trajectories during the gating process was already 2 proposed. For instance, through the use of a hybrid elastic-network Brownian 3 dynamics simulation predicting two possible pathways for GLIC gating, that are 4 characterized by different compactions of the ECD (Orellana et al., 2016). Here, we 5 extend this concept by proposing two pathways involving either an early motion of 6 the ECD or an early motion of the M2-M3 loop (Figure 9). 7 8 Figure 9. Speculative summary scheme. Bimane-quencher centroid distances are shown for 9 the trajectories A and B. Non-functional mutants H235F and L157A/L246A are tentatively placed 10 on the trajectories A and B respectively, according to their quenching pattern.

Consequences on the gating mechanism within the pLGIC family 1
The conservation of the gating mechanism between bacterial and eukaryotic 2 pLGICs is well documented by the available structures with the common allosteric of the orthosteric site is predicted to be rather complete, but where the channel is 14 closed, would fit the functional requirement of a pre-active state (Lape et al., 2008). 15 Our observation that propofol specifically affects the pre-activation step might thus 16 tentatively be extended to eukaryotic receptors. 17 An unexpected finding here is that GLIC and its mutants have access to a slits from 420 to 530 nm. Other parameters were kept constant throughout the study. 30 On the sample at pH 7.4, an addition of SDS to reach 1 % final concentration was 31 done to obtain the FSDS value and a tryptophan emission spectrum was done before 1 and after SDS addition in order to monitor denaturation. where Fmax represents the maximal change in fluorescence amplitude; F0 the initial 7 fluorescence at pH 7.8; nH represents the hill number and EC50 the proton 8 concentration for which half of the maximal fluorescence change is measured. For 9 Bim136-Q101W and Bim250-Y197 and in some other mutants, we excluded the 10 data point below pH 3.5 that show a small but significant change in fluorescence 11 intensity in the opposite direction to the quenching curves. We did not fit the Bim135-12 W72 mutant that shows a bell shape curve. where Imax represents the maximal current in percentage of the response from the 2 reference solution. nH represents the hill number and EC50 the proton concentration 3 for which half of the maximal electrophysiological response is recorded. 4

Xenopus oocytes immunolabeling 5
Mutants generating currents smaller than 500 nA at high proton 6 concentrations were categorized as non-functional. For these non-functional 7 mutants, expression tests were performed by immunolabeling of oocytes as

Molecular Modeling 20
Each structure (4NPQ and 4HFI) was fitted, using iMODfit (Lopéz-Blanco and 21 Chacón, 2013), to the simulated electron-microscopy envelope of the other 22 structure. The EM density map resolution was set to 5 Å and the grid size to 0. 5