Nicotinic acetylcholine receptors from Torpedo marmorata electroplax were purified by bromoacetylcholine-affinity chromatography in the presence of asolectin lipids using cholate as a detergent. After elution with carbamylcholine, purified receptors were dialyzed and reconstituted in asolectin lipids at a final protein concentration of 0.3–1.2 mg ml^-1. Samples were aliquoted and stored in liquid nitrogen (Ivorra et al., 2002).

Adult female Xenopus laevis (purchased from Harlan Interfauna Ibérica S.L., Barcelona, Spain; and Centre National de la Recherche Scientifique, Montpellier, France) were immersed in cold 0.17% MS-222 for 20 min and a piece of ovary was drawn out aseptically. Animal handling was carried out in accordance with the guidelines for the care and use of experimental animals adopted by the E.U. and the animal protocol was approved by the Ethic Committee of Universidad de Alicante. Stage V and VI oocytes were isolated and their surrounding layers removed manually. Cells were kept at 15–16°C in a modified Barth’s solution [88 mM NaCl, 1 mM KCl, 2.40 mM NaHCO₃, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, 0.82 mM MgSO₄, 10 mM HEPES (pH 7.4), 100 U ml^-1 penicillin and 0.1 mg ml^-1 streptomycin] until used. Oocytes were microinjected with 100 nl of an aliquot of reconstituted nAChRs (Morales et al., 1995).

Membrane current recordings were performed at 21–25°C, 16–72 h after proteoliposome injection, using a high compliance two-microelectrode voltage-clamp system (TurboTEC-10CD, npi Tamm, Germany). The recording methodology has been described previously (Morales et al., 1995; Alberola-Die et al., 2016). Briefly, oocytes were placed in a 150 μl recording chamber and continuously superfused with normal frog Ringer’s solution (NR: 115 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 5 mM HEPES, pH 7.0) supplemented with 0.5 μM atropine sulfate (ANR) to block any muscarinic response (Kusano et al., 1982). The membrane potential was held at -60 mV, unless otherwise stated. ACh and other tested drugs were diluted in ANR solution and oocytes were superfused with them at a flow rate of 13–17 ml min^-1. Membrane currents elicited by ACh (I_ACh) either alone or co-applied with DMA, were low-pass filtered at 30–1000 Hz and, after sampling at fivefold the filter frequency (Digidata series 1200 and 1440A; Axon Instruments, Foster City, CA, USA), recorded on two PC-computers, using the WCP v. 3.2.8 package developed by J. Dempster (Strathclyde Electrophysiology Software, University of Strathclyde, Scotland, UK) and AxoScope v. 10.0.0.60 (Molecular Devices Corporation, Sunnyvale, CA, USA).

Experimental procedures were similar to those used to study the modulating effects of lidocaine (Alberola-Die et al., 2011) and DEA (Alberola-Die et al., 2016) on nAChRs. Briefly, DMA concentration-I_ACh inhibition relationship was determined by measuring I_AChs evoked by 10 μM ACh alone or together with different DMA concentrations. For competition assays, ACh concentration-I_ACh amplitude curves were obtained by exposing injected oocytes to increasing ACh concentrations, either alone or together with 2 mM DMA. I_AChs were recorded in the presence or absence of DMA, normalized to the I_ACh evoked by 1 mM ACh alone, and the values fitted to a sigmoid curve (see below Eq. (3)). To allow nAChRs to recover from desensitization, the interval between consecutive ACh applications was at least 5 min. To assess the blockade of resting nAChRs by DMA, we compared the I_AChs elicited by ACh (from 1 μM to 1 mM) alone or co-applied with 2 mM DMA either directly or after pre-application of DMA (same concentration) for 12 s.

The voltage dependence of the I_ACh blockade by DMA was determined by applying to the oocyte series of 800 ms voltage pulses (from -120 to +60 mV, in 20 mV steps) before ligand superfusion and during the I_ACh plateau elicited by 10 μM ACh, either alone or co-applied with DMA at different concentrations; the -120 mV pulse duration was extended up to 1500 ms to allow a more complete current relaxation.

Inhibition curves were determined by measuring I_ACh evoked by 10 μM ACh in the presence of different DMA concentrations. I_AChs elicited in the presence of DMA were normalized to the I_ACh evoked by ACh alone. Data were fitted to a single-site inhibition curve using the Origin 6.1 software (OriginLab, Corp. Northampton, MA, USA).

Recovery from I_ACh blockade by DMA was determined by giving 32 s pulses of ACh either alone or co-applied with DMA, for solely the first 12 s or during the whole pulse; I_ACh recovery was measured 20 s and 7 min after DMA washout. The percentage of recovery from blockade (% Recovery) was obtained using the Eq. (1):

where I_ACh is the current amplitude evoked by 10 μM ACh alone; I_ACh+DMA, is the current elicited by co-application of 10 μM ACh with 2 mM DMA; and I_AChafterDMA is the current obtained 20 s or 7 min after DMA removal.

The rate of desensitization (I_ACh decay) was determined by measuring the I_ACh amplitude elicited by 100 μM ACh, either alone or co-applied with different concentrations of DMA (100 μM–2 mM), at different times after I_ACh peak. Desensitization rates were obtained using the Eq. (2):

where D_ti is the desensitization value at 2, 10, or 20 s after the peak; I_peak the I_ACh amplitude at the peak; and I_ti the current amplitudes remaining after the specified times (Olivera-Bravo et al., 2007). The apparent time-to-peak was determined as the time elapsed from I_ACh onset to the I_ACh peak, from currents elicited by ACh either alone or with DMA. We have called this parameter as “apparent” time-to-peak, just to indicate that these values do not necessarily reflect “real” time-to-peak values of nAChR activation but those observed in our experimental conditions.

To characterize the pharmacological profile of DMA, nAChRs were activated by different ACh concentrations either alone or co-applied with DMA (at roughly its IC₅₀, unless otherwise stated), just directly or after its pre-application for 12 s. Dose-response data were fitted to the following form of the Hill Eq. (3):

where I is the I_ACh peak elicited at a given ACh concentration ([ACh]; applied either alone or together with DMA); I_max is the maximum I_ACh recorded; EC₅₀ is the agonist concentration required to obtain one-half the maximum I_ACh; and n_H is the Hill coefficient.

Net i/v curves for I_ACh were obtained by subtracting, for each voltage, the steady-state currents attained in ANR (measured during the last 100 ms of the pulse) from the corresponding ones recorded in presence of 10 μM ACh. These net I_ACh values were normalized, for each oocyte, to the ACh response at -60 mV.

To explore whether DEA and DMA molecules bind at the same loci of nAChRs, we applied both molecules, at their corresponding IC₅₀s, together with ACh, to assess if their co-application causes additive inhibiting effects on I_ACh. The extent of I_ACh inhibition was later compared to that predicted by allotopic and syntopic models (Jarvis and Thompson, 2013). The inhibition values for the allotopic model were computed with Eq. (4):

where In_DEA,DMA is the predicted I_ACh inhibition when DEA and DMA are co-applied; In_DEA and In_DMA are the I_ACh inhibitions caused by DEA and DMA, respectively. The inhibition values for the syntopic model were computed with Eq. (5):

Unless otherwise specified, values given are the mean ± SEM; “n” indicates the number of oocytes and “N” is the number of oocyte-donor frogs from which data were obtained. When comparing two-group means of normally distributed values, the Student’s t-test was used; otherwise, Mann-Whitney rank-sum test was applied. Among-group differences were determined by the analysis of variance (ANOVA) and mean differences for each pair of groups were determined with the Bonferroni t-test. The one-sample t-test was used to compare the mean of an experimental group with a specified value. For the comparison of IC₅₀ or EC₅₀ values we used the confidence intervals computed by the curve-fitting function of Origin 6.1 software, using a percentage of confidence of 95%. The criterion of “non-overlapping 95% confidence intervals” was used to determine significant difference between EC₅₀s. A significance level of p < 0.05 was considered for all cases.

Docking assays were carried out as previously described (Alberola-Die et al., 2016). Briefly, Torpedo nAChR structures in the closed (4 Å resolution, code 2BG9; Unwin, 2005) and open (6.2 Å resolution, code 4AQ9; Unwin and Fujiyoshi, 2012), were taken from RCSB Protein Data Bank. The edition of the protein was made using DeepView v4.1 (Guex and Peitsch, 1997) and Yasara (Krieger et al., 2002, 2004) software without further optimization. DMA, lidocaine and propofol structures (CIDs, 6896, 3676, and 4943, respectively) were taken from NCBI Pubchem database¹. A global docking procedure was accomplished with AutoDock 4 (Morris et al., 2008) implemented in Yasara, where a total of 500 flexible docking runs were set and clustered around the putative binding sites. The program then performed a simulated annealing minimization of the complexes, which moved the structure to a nearby stable energy minimum, by using the implemented AMBER 99 force field (Duan et al., 2003). The best binding energy complex in each cluster was stored, analyzed and used to select the best orientation of the interacting partners. Figures were drawn with open source Pymol (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC²). Yasara pH command was set to 7.0, ensuring that molecules preserve their pH dependency of bond orders and protonation patterns. In this way, DMA molecules remained during the docking procedure uncharged, but 86% of the lidocaine molecules were protonated.

Acetylcholine, atropine sulfate, DEA, DMA, DMSO, MS-222, penicillin and streptomycin were from Sigma (St. Louis, MO, USA). HEPES was obtained from Acros Organics (Morris County, NJ, USA). Reagents of general use were purchased from Scharlau Chemie SA (Barcelona, Spain). DMA solutions were prepared from a 1M stock solution in DMSO. All solutions were made in ANR just before each application.

Either in uninjected cells or in oocytes bearing nAChRs, with the membrane potential held at -60 mV, DMA application did not appreciably modify the cell membrane conductance at concentrations lower than 5 mM, indicating both a lack of unspecific effect of DMA on native ion channels opened at rest and that DMA did not act as a partial agonist of nAChRs. Nevertheless, at higher DMA concentrations some oocytes showed a slight decrease in their membrane conductance (not shown), although we have not pursued the basis of this effect.

In oocytes that had incorporated nAChRs, co-application of 10 μM ACh with DMA, at concentrations ranging from 100 μM to 50 mM, inhibited peak I_ACh amplitude in a concentration-dependent manner (Figure 1). The half-inhibitory DMA concentration (IC₅₀), obtained by fitting the data to the Hill equation, was 2.1 mM (confidence interval, 1.7–2.7 mM), and the Hill coefficient (n_H) 1.2 ± 0.2 (Figure 1C), indicating that a single DMA molecule is sufficient to block the nAChR.

Nicotinic acetylcholine receptor blockade by DMA outlasted the drug application, as also occurred for lidocaine and DEA (Alberola-Die et al., 2011, 2016). Thereby, 20 s after rinsing out DMA (2 mM), the percentage of I_ACh recovery (see Materials and Methods, Eq. (1)) was only 52.0 ± 2.2% (Figures 2A,C), increasing to 85.1 ± 5.1% when the elapsed time was 7 min (Figures 2B,C). Thus, the I_AChs after both DMA washout times were significantly smaller than in control conditions, indicating a slow nAChR recovery from blockade.

We measured I_AChs at different membrane potentials by applying voltage jumps (from -120 to +60 mV, in 20 mV steps) in absence (not shown) or presence of 10 μM ACh applied either alone or together with 2 mM DMA (Figure 3A) to determine if I_ACh inhibition by DMA has any voltage-dependence, which would suggest its binding into the channel pore.

The i/v curves of net I_AChs (see Materials and Methods) elicited by ACh either alone or co-applied with 2 mM DMA are shown in Figure 3B. The presence of DMA did not affect the I_ACh reversal potential, thus the channel ion selectivity was unaffected. However, 2 mM DMA reduced I_ACh amplitude in a similar percentage at all tested potentials (Figures 3A,B), indicating that DMA caused a voltage-independent blockade of nAChRs. This lack of voltage-dependence of I_ACh blockade by DMA is more evident when plotting the percentage of I_ACh remaining after co-applying ACh with DMA, normalized to its control I_ACh, against membrane potential (Figure 3C). Notice that when measured at the I_ACh plateau, the current left upon 2 mM DMA was roughly 30% of the control values at any potential tested (Figure 3C). This percentage of I_ACh remaining is fairly smaller than that found when considering the I_ACh peak (40–50%, Figure 1C) and this discrepancy is most likely due to the enhancement of I_ACh decay by DMA (see below).

Since DMA is a non-charged molecule, the lack of voltage-dependence of nAChRs blockade by DMA does not fully exclude that this molecule can bind into the channel pore. Therefore, to ascertain if DMA actually binds into the channel pore we analyzed the “rebound” currents elicited by ACh either alone or in the presence of 2 mM DMA (Figure 4). It is well-established that high doses of ACh elicit open-channel blockade of nAChRs, evidenced by an I_ACh rebound just when rinsing out the agonist. This current arises during the agonist washout because then the ACh leaves the channel, unplugging the pore, when it can be still bound to the high affinity orthosteric sites (Legendre et al., 2000; Liu et al., 2008). This open-channel blockade of nAChRs by high ACh concentrations is only found at negative membrane potentials, because at positive voltages the positively charged ACh is electrostatically repelled from the channel pore (compare control I_AChs, black recordings, of Figures 4A₁,A₂). However, when 1 mM ACh was co-applied with 2 mM DMA, rebound currents were elicited both at positive and negative potentials (Figures 4A₁,A₂, green traces), indicating that the uncharged DMA is binding into the channel pore with low affinity, and thereby eliciting an open-channel blockade of nAChRs. Furthermore, this I_ACh rebound was also elicited when 2 mM DMA was co-applied with a low ACh concentration (10 μM) at negative potentials, in spite of the fact that ACh, at this concentration, cannot block by its own the channel pore (Figure 4B). Noticeably, when DMA concentration decreased below 500 μM, this rebound currents were not elicited (see in the recordings of Figure 1B that the rebound current appears at 2 mM DMA) and they were of larger amplitude when the cell was challenged with a relatively high ACh concentration (100 μM or higher; compare recordings of panels A₂,B of Figure 4).

When either 10 or 100 μM ACh were co-applied with DMA, at roughly its IC₅₀, I_ACh decays were significantly accelerated with respect to those evoked by ACh alone, suggesting an enhancement of nAChR desensitization by DMA. D_ti values at 2 and 20 s (see Eq. (2) in Materials and Methods) were: 36 ± 5% and 92 ± 1%, for 100 μM ACh alone versus 50 ± 5% and 99 ± 1% for 100 μM ACh plus 2 mM DMA, respectively (same cells in both groups; n = 18, N = 14; p < 0.05, ANOVA; see Figures 5A,B). This effect was fully reverted 7 min after DMA rinsing with ANR (40 ± 5% and 93 ± 2%; p > 0.05, ANOVA, Figure 5B). Additionally, DMA diminished the apparent time-to-peak, i.e., the time elapsed from I_ACh onset to I_ACh peak, from 1.6 ± 0.2 s for 100 μM ACh alone to 1.1 ± 0.2 s for 100 μM ACh plus 2 mM DMA (same cells that I_ACh decay measurements; p < 0.05, ANOVA; Figures 5A,C). Noteworthy, the time-to-peak reverted to control values 7 min after DMA washout (1.6 ± 0.3 s; see Postcontrol of Figures 5A,C), as the I_ACh decay rate did. Interestingly, DMA hastening of I_ACh decay was dose-dependent, starting the increase of desensitization at concentrations as low as 100 μM DMA (Figure 5D; p < 0.05, one sample t-test).

Since co-application of ACh with 2 mM DMA elicits rebound-currents during the washout, suggesting that DMA can plug the channel pore, the acceleration of I_ACh decay could be due either to an enhancement of nAChR desensitization or to a slow binding of DMA to the channel pore. In order to differentiate between both mechanisms, we co-applied 2 mM DMA with two different concentrations of ACh (10 μM and 1 mM) at +40 mV (Figures 6A₁,A₂), because at this membrane potential ACh does not contribute to the open-channel blockade. Co-application of 10 μM ACh with 2 mM DMA inhibited I_ACh by 50.0 ± 9.3% (n = 4; N = 3), as it would be expected from the dose-inhibition curve (Figure 1C), and there was a pronounced acceleration of I_ACh decay, which followed a single exponential function (Figure 6A₁, red discontinuous line), with a time-constant (τ) of 3.34 ± 1.44 s (Figure 6B). When the same concentration of DMA was co-applied with 1 mM ACh, the I_ACh decreased by only 35.4 ± 4.3% (see Figures 6A₂ and 7A₁) and the I_ACh decayed following a double exponential function (Figure 6A₂, red discontinuous line), with a τ value for the fast component of 0.40 ± 0.05 s (Figure 6B). Given the large differences in the τ values for the I_ACh decay caused by the same DMA concentration when co-applied with two different ACh concentrations, it follows that the acceleration of I_ACh decay by DMA cannot only be explained by its binding into the channel pore, acting as an open-channel blocker, but rather it points out that DMA actually enhances nAChR desensitization.

The pharmacological profile of nAChR inhibition by DMA was studied by superfusing ACh at different concentrations (1, 3, 10, 100, and 1000 μM) alone or co-applied with 2 mM DMA either directly (Figure 7A₁) or after 12 s pre-application of the same DMA concentration (Figure 7A₂). Figure 7B shows the relationship between ACh-concentration and I_ACh amplitude in absence and presence of DMA. The sigmoid curve fitted for I_AChs elicited by ACh alone gave an EC₅₀ of 29 μM (confidence interval, 22–44 μM) and an n_H of 2.0 ± 0. 1, which are similar values to those previously reported for this receptor (Morales et al., 1995; Alberola-Die et al., 2016). When 2 mM DMA was co-applied with the different ACh concentrations, the I_ACh amplitude decreased, even with the highest (almost saturating) ACh concentration (Figures 7A₁,B), suggesting a non-competitive blockade of nAChRs by DMA. Furthermore, the dose-response curve shifted to the right (Figure 7B), increasing significantly the EC₅₀ up to 63 μM (confidence interval, 50–96 μM) and decreasing the slope to 1.3 ± 0.1. A similar reduction in the slope of the ACh concentration-I_ACh amplitude relationship was found when co-applying ACh with the quaternary-ammonium BW284c51 (Olivera-Bravo et al., 2005) or lidocaine (Alberola-Die et al., 2011), though the mechanisms underlying this effect remains unclear, since it is only partially dependent on the increase of nAChR desensitization caused by these drugs (Alberola-Die et al., 2011). Nonetheless, given that the percentage of I_ACh inhibition was also dependent on ACh concentration, nAChR blockade by DMA was not exclusively a non-competitive antagonism. Thus, at low (10 μM) ACh concentration, 2 mM DMA blocked roughly half the control I_ACh (Figures 7A₁,B,C), as expected from its estimated IC₅₀ (see Figure 1). However, at very high ACh concentration (1 mM), the percentage of I_ACh blockade decreased significantly (Figures 7A₁,B,C; p < 0.05, ANOVA). Although, we cannot fully discard some competitive interactions of DMA and ACh on the orthosteric binding sites (see below), this apparent competitive mechanism of blockade could be explained by the binding of DMA to the nAChR in its closed state, in a similar way as we have previously proposed for lidocaine and also for DEA (Alberola-Die et al., 2011, 2016). To test this hypothesis, we determined the percentages of I_ACh blockade induced by DMA (2 mM) when it was pre-applied to the oocyte for 12 s before being co-applied with ACh at increasing concentrations (1 μM–1 mM; Figures 7A₂,B,C). As shown in panels 7A_2,B, the percentage of I_ACh remaining when ACh was co-applied with 2 mM DMA, after being pre-applied for 12 s, was significantly reduced with 10 μM or higher ACh concentrations, as compared with those corresponding to solely ACh and DMA co-application. The sigmoid curve fitting the I_ACh values obtained at the different concentrations tested had an EC₅₀ of 25 μM (confidence interval, 8–78 μM) and an n_H of 1.0 ± 0.3. Thus, pre-application of DMA followed by its co-application with 10 μM ACh increased only modestly the percentage of I_ACh inhibition obtained by barely DMA and ACh co-application (53.9 ± 1.9%, n = 28, N = 13, for direct co-application, versus 65.1 ± 2.4%, n = 15, N = 8, for pre-application followed by co-application; p < 0.05, t-test; Figures 7A₁,A₂,C). Noteworthy, the enhancement of I_ACh inhibition by DMA pre-application was stronger when it was later co-applied with high ACh concentrations, as it would be expected if DMA pre-application blocked resting nAChRs. Thereby, the percentage of I_ACh inhibition by 2 mM DMA and 1 mM ACh co-application was 28.7 ± 5.7% (n = 4, N = 2) and increased to 90.1 ± 3.5% (n = 5, N = 3; p < 0.05, t-test) when the same DMA concentration was pre-applied and then co-applied with 1 mM ACh (compare recordings of panels Figures 7A_1,A_2,C).

Since DEA and DMA are structurally quite different molecules and both cause inhibitory effects on nAChRs (see Alberola-Die et al., 2016), we assessed the effect of co-application of DMA with DEA in the presence of 10 μM ACh, aiming to unravel whether or not their inhibitory actions on nAChRs are additive. Co-application of 10 μM ACh with 70 μM DEA (close to its IC₅₀) decreased I_ACh by 47.6 ± 2.0% (Figures 8A₁,B). A similar percentage of I_ACh blockade was obtained when ACh (10 μM) was co-applied with 2 mM DMA (53.9 ± 1.9%; Figures 8A₂,B). The I_ACh inhibition increased significantly when these same doses of DMA and DEA were co-applied together with 10 μM ACh (77.7 ± 1.3%; Figures 8A₃,B; p < 0.05, t-test). This enhancement of I_ACh inhibition by co-application of DEA and DMA could be due to either syntopic (both molecules sharing a single binding site) or allotopic (binding to different loci) interaction of these molecules on the nAChR. Using the theoretical approach proposed by Jarvis and Thompson (2013) to discriminate between both interaction models (see Materials and Methods Eqs. (4) and (5)), we found that I_ACh inhibition caused by DEA and DMA co-application properly fitted to the values predicted by the allotopic model, but were significantly different to those estimated by the syntopic one (Figure 8B).

A further experimental evidence for allotopic interaction of DMA and DEA was attained by determining the nAChR pharmacological profile in the presence of both DMA and DEA. Thus, nAChR bearing oocytes were challenged with different ACh concentrations (1 μM–1 mM) in the presence of 1 mM DMA and 30 μM DEA, which are the concentrations corresponding to their IC₃₀ for nAChR blockade (see Figure 1 and Alberola-Die et al., 2016, respectively). Then, if DMA and DEA interact with nAChRs at different loci, it should be expected roughly a 50% decrease in the I_ACh when applied together at these doses. Noticeably, the dose-response curve obtained in the presence of DMA and DEA fairly well-matched the pharmacological profile of nAChRs in the presence of the whole lidocaine molecule at its IC₅₀ (Figure 9). Actually, we found non-significant differences between the values found for the dose-response curve in the presence of DEA and DMA and those previously attained in the presence of lidocaine (p > 0.05, t-test; Alberola-Die et al., 2011). Furthermore, the estimated EC₅₀ value (65 μM) of nAChRs in the presence of DEA and DMA was within the confidence interval of the EC₅₀ in the presence of lidocaine (64–147 μM). Therefore, as it could be expected from their differences in molecular structure, DEA and DMA act by different mechanisms and bind to different sites on nAChRs.

We have explored the interactions between DMA and the nAChR, using as template the full structure of Torpedo nAChR in both closed and open conformations (see Materials and Methods). We carried out 500 runs for DMA-nAChR interactions for both the closed and the open states. For the closed state, we found 60 clusters of interaction sites that differ in less than 5 Å of root-mean-square-deviation. DMA clusters on the nAChR were mainly located at the TM (52%) and EC (46%) domains, with only 1 cluster sited at the intracellular (IC) domain, sited adjacent to the TM region (Figure 10A). On the TM domain, DMA interacted both at intrasubunit crevices and at intersubunit interfaces, being these latter ones the more numerous, involving each single pair of nAChR subunits (Figures 10B₁,B₂). Interestingly, we found discernable changes on DMA binding to the TM domains in the closed and open states. Thus, in closed nAChRs DMA interacted preferentially with residues located at intra- and intersubunit spots but not into the channel pore (Figures 10B₁,B₂), whereas in the open state the hotspots for DMA were less numerous at intersubunit crevices (compare panels B_2,C₂ of Figure 10) and some appeared inside the channel pore (Figure 10C₁, red circle). Noticeably, DMA binding sites on nAChR at the TM domain follow a pattern similar to that found for the entire lidocaine molecule on this receptor (see Supplementary Figure S1; Alberola-Die et al., 2016). At the EC domain, DMA bound at several intrasubunit crevices, mainly located on α₁, α₂, and β subunits (Figure 10A) and at the interface of α₁-γ, γ-α₂, δ-β, and β-α₁ subunits. DMA was not found occupying the orthosteric sites, though there was a hotspot for DMA at the α₁-γ interface relatively close to the ligand-binding site (Figure 10A).