L-3,3’,5-triiodothyronine and pregnenolone sulfate inhibit Torpedo nicotinic acetylcholine receptors

The nicotinic acetylcholine receptor (nAChR) is an excitatory pentameric ligand-gated ion channel (pLGIC), homologous to the inhibitory γ-aminobutyric acid (GABA) type A receptor targeted by pharmaceuticals and endogenous sedatives. Activation of the GABAA receptor by the neurosteroid allopregnanolone can be inhibited competitively by thyroid hormone (L-3,3’,5-triiodothyronine, or T3), but modulation of nAChR by T3 or neurosteroids has not been investigated. Here we show that allopregnanolone inhibits the nAChR from Torpedo californica at micromolar concentrations, as do T3 and the anionic neurosteroid pregnenolone sulfate (PS). We test for the role of protein and ligand charge in mediated receptor inhibition by varying pH in a narrow range around physiological pH. We find that both T3 and PS become less potent with increasing pH, with remarkably similar trends in IC50 when T3 is neutral at pH < 7.3. After deprotonation of T3 (but no additional deprotonation of PS) at pH 7.3, T3 loses potency more slowly with increasing pH than PS. We interpret this result as indicating the negative charge is not required for inhibition but does increase activity. Finally, we show that both T3 and PS affect nAChR channel desensitization, which may implicate a binding site homologous to one that was recently indicated for accelerated desensitization of the GABAA receptor by PS.

The neurosteroid 3α-hydroxy-5α-pregnan-20-one (allopregnanolone) activates the GABA A receptor [29], as does the related 5α-pregnane-3α,21-diol-20-one (THDOC) [30][31][32]. The thyroid hormone L-3,3',5-L-triiodothyronine (T3) (Fig 1A) was originally theorized to have neurosteroid-like effects due to its similarity to neurosteroids' size, shape, and lipophilicity [33]. In particular, the results of computational analysis of the overall volumes and shapes of pregnanolone sulfate (PS) and T3 are compared in Figs 5 and 6 of our previous study [33}. Superpositioning of the two molecules showed that not only the molecular volumes, but also the shapes, are nearly identical, and no unmatched functional groups extend from the main axis. Additionally, we have previously published molecular dynamics simulations of single molecules of T3 and allopregnanolone (see S1 Movie and S2 Movie of [29]), illustrating the rigidity of these two molecules in an aqueous solution. The rigidity of the steroid scaffold can be compared to the rigidity of T3 due to the steric hindrance due to the two iodines on the inner aromatic ring of T3." Despite structural analogy to positive allosteric modulators, T3 inhibits activity of the GABA A receptor at low concentrations [29,[34][35][36] and competitively inhibits GABA A receptor activation due to ivermectin or allopregnanolone [29]. The mechanism underlying this surprising result is unknown, and here we test the hypothesis that a population of anionic T3 at physiological pH interacts with pLGICs through a similar mechanism as the anionic neurosteroid pregnanolone sulfate (PS) (Fig 1B). PS is sulfated at C3 but is otherwise similar to positively-modulating neurosteroids, and also inhibits GABA A receptor function [37].
Structural biology has recently provided insight into the binding mode of PS with GABA A receptors. In a recent crystal structure of a homopentameric receptor chimera comprised of a Gleobacter ligand-gated ion channel (GLIC) ECD and a mouse GABA A R α 1 TMD, Laverty et al. [38] showed that PS binds alongside a groove between the M3 and M4 helices near the intracellular side of the transmembrane domain of GABA A receptors. This site is involved with the "desensitization gate" recently classified by Gielen et al. [39], indicated by mutations of amino acid residues at the M1-M2 linker and M3 helix of GABA A receptors affected receptor desensitization.
While lipid mixtures used for X-ray crystallography of membrane proteins have become more complex, there are still significant discrepancies with native membranes. pLGICs are extremely lipid sensitive; in particular, they may have specific interactions with anionic lipids [40,41], polyunsaturated fatty acids [42], and cholesterol [43,44]. The most feasible pLGIC for study in native membranes is the nicotinic acetylcholine receptor, due to its high concentration in the electric organ of the torpedo electric ray.
Changes in extracellular pH also affect nAChR ion conductance, with more acidic environments yielding lower ion flux and basic environments yielding higher ion flux relative to the receptor's acetylcholine response at a neutral pH [54,55]. Different pH environments also affect mean open channel time and desensitization of the receptor [55]. Palma et al. [54] and Li et al. [55] showed that pH changes in the surrounding fluid correlate with changes in nAChR channel conductivity, suggesting that charged residues in the nAChR vestibular and channel pore region affect allosteric transitions to active and inactive states, and that changes in the extracellular pH would titrate these residues. Here we control for the effect of pH on receptor function by comparison of pH effects on modulation by T3 with those of PS, which is not expected to change charge states over the pH range we consider.
We tested the hypothesis that a population of anionic T3 transduces inhibition of nAChR at physiological pH. by extracting functional nAChRs from Torpedo californica and used twoelectrode voltage clamp (TEVC) to demonstrate functional effects of PS and T3 on nAChRs. We found that both PS and T3 inhibit nAChR activity due to acetylcholine, representing a case in which the same effect on channel function is demonstrated in both nAChRs and GABA A receptors. We also performed TEVC in differing extracellular pH environments to demonstrate the differential effects pH has on ligand charge and resultant receptor activity. Further, we show that PS and T3 modulate the effects of different pH environments on current amplitude when binding acetylcholine. We also show that T3 and PS change nAChR kinetics, increasing desensitization rate at low pHs but minimally affecting desensitization rate at high pHs.

Preparation of crude Torpedo nicotinic acetylcholine receptors
All steps before microinjection of resuspended receptor proteins were performed at 0 to 4˚C and were modified from the preparation method described previously [56]. Briefly, flash-frozen Torpedo electric organ was purchased from Aquatic Research Consultants (San Pedro, CA). The frozen tissue (600 g) was homogenized in homogenization buffer (10 mM sodium phosphate, 5 mM ethylenediaminetetraacetic acid (EDTA), 5 mM ethylene glycol-bis(β-aminoethyl ether)-N,N,N 0 ,N 0 -tetraacetic acid (EGTA), 10 mM iodoacetamide, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), 0.02% NaN 3 , pH 7.5) in 30-second bursts at liquefaction setting (Oster blender) and centrifuged for 10 min at 4,080 x g. The supernatant was then filtered through 4 layers of cheesecloth and reserved. The pellets were then resuspended in 200 mL homogenization buffer and centrifugation was repeated. All of the supernatants were then pooled and centrifuged at 134,000 x g for 45 minutes. This pellet was resuspended in 32 mL 28% (w/v) sucrose buffer (10 mM sodium phosphate, 0.1 mM EDTA, 0.02% NaN 3 , pH 7.0). In order to isolate membrane proteins, the aliquots of resuspended pellet (8 mL each) were layered on top of a discontinuous sucrose gradient (10 mL 30% sucrose, 12 mL 35% sucrose, 7 mL 41% sucrose) and centrifuged at 121,800 x g for 4 hours. The membrane band at the interface of the 30-35% sucrose layer was collected using a thin glass transfer pipet, diluted 1:1 with deionized water, and centrifuged at 142,400 x g for 35 minutes. The membranes in the resulting pellet were base-extracted by resuspending them to approximately 0.5 mg/mL protein in water, and the pH was carefully titrated to 11.0 with NaOH, followed by stirring for 45 minutes. The base-extracted membranes were then centrifuged at 142,400 x g for 45 minutes, resuspended in buffer A (100 mM NaCl, 10 mM Tris-HCl, 0.1 mM EDTA, 0.02% NaN 3 , pH 7.4), and flash-frozen in liquid nitrogen.

Solubilization of crude nicotinic acetylcholine receptor
The protein product from extraction was diluted to 2 mg protein/mL in buffer A. Sodium cholate (10% w/v) dissolved in buffer A was added to give a final cholate concentration of 1%. The mixture was then stirred for 30 minutes, centrifuged at 142,400 x g for 30 minutes, and the supernatant was collected.

Preparation of asolectin liposomes for resuspension
Solid asolectin lipid was suspended in 4.16% cholate solution in buffer A to form a 65 mM solution. The mixture was vortexed and sonicated at 20˚C under argon gas for 45 minutes, and then stored at 0-4˚C until reconstitution.

Reconstitution of Torpedo nicotinic acetylcholine receptors in asolectin liposomes
Solubilized nAChRs (1-1.65 mg/mL) were mixed with 0.3 mL of liposome/cholate mixture to yield a final volume of 1 mL, and a final concentration of 2% cholate. The mixture was dialyzed for 48 hours against 1,000 volumes buffer A, with a change of buffer every 12 hours. Before use in dialysis, buffer A was bubbled with argon gas for 15 minutes.

Oocyte microinjection
Ready-to-inject, defolliculated Xenopus oocytes were purchased from Ecocyte Bioscience (Austin, TX). The glass injectors were 1.6 to 2 mm o.d., 1.2 to 1.6 mm i.d., and were pulled using a Sutter Instrument Co. Model P-97 puller. The oocytes were injected with 46 nL asolectin-resuspended nAChRs using a digital microdispenser (Drummond Nanoject II). The

Two-electrode voltage clamping of oocytes
Sixteen hours after injection, the oocytes were clamped using a TEVC system. All ligands were dissolved in modified Barth's solution at the indicated pH and were perfused into the chamber using a gravity-flow system. Thyroid hormones were initially dissolved in 0.1 M NaOH. The oocytes were impaled with two 3 M KCl-filled glass microelectrodes (1-2 MO each) and were clamped at -60 mV with an OC-725C Oocyte Clamp (Warner Instruments). Acetylcholine was applied to the bath surrounding the oocyte using a gravity-flow system at 10-13 mL/min. T3 or PS were co-applied with acetylcholine, also using a gravity-flow system. Each ligand perfusion lasted 60-70 seconds; after each perfusion, the bath solution was exchanged with ligand-free buffer until the current response signal had returned to baseline for at least 120 seconds before the next application of ligand(s). Data were recorded using iWorx LabScribe v1.959.

Data analysis
The current flux signal for each administration of ligand was exported from iWorx into Matlab version 2012b. Due to significant batch-to-batch variation of ion conductance after receptor microinjection, maximal current response (peak) values for inhibition data were normalized to each oocyte's control maximal peak. For all data related to inhibitor concentration curves, the data were fit according to the conditions specified in Table 1. The equation for the fit was Y = Bottom + (Top-Bottom)/(1+10^((LogIC50-X) � HillSlope)).
The data were then trimmed to begin with the response as it reached 0.96 of the response peak (to avoid false inflection points by signal noise just after the response's maximum point), and end 35 seconds post-peak. The data were then fit to a two-degree exponential function (i.e., slow and fast response; ). The slower response of the twodegree exponential function was used in each calculation for analysis of the decay rate, as small irregularities in the raw signal yielded extremely high values for the fast component. For each ligand/environmental condition (n = 3, per condition), decay response and amplitude were calculated and averaged.

PS and T3 inhibit nAChRs at concentrations similar to GABA A receptor IC 50 s
We co-administered several neurosteroid and neurosteroid-like ligands with acetylcholine during TEVC to determine their effects on nAChR function. Fig 2A shows the inhibitory effect of T3 on nAChR stimulation by 30 μM acetylcholine, as well as a representative trace of the reduction in response due to acetylcholine when co-applied with T3 (Fig 2A, inset). The apparent maximal effect of T3 (seen at � 100 μM) reduced the nAChR control response by 80 ± 6%, with an IC 50 of 5.4 ± 1 μM T3. This is very similar to the IC 50 of T3 for the GABA A receptor (8 ± 2 μM) [29]. Fig 2B shows the inhibitory effect of PS on nAChR stimulation by 30 μM acetylcholine, as well as a representative trace of the reduction in response due to acetylcholine when co-applied with PS (Fig 2B, inset). The apparent maximal effect of PS (seen at � 100 μM) reduced the nAChR control response by 81 ± 4%, with an IC 50 of 4.9 ± 3 μM PS. Triiodothyroacetic acid  https://doi.org/10.1371/journal.pone.0223272.g002 L-3,3',5-triiodothyronine and pregnenolone sulfate inhibit Torpedo nicotinic acetylcholine receptors (triac), which lacks the amine group of T3, also inhibits nAChR, with an IC 50 of 7 ± 3.1 μM triac (Fig 3). Further, allopregnanolone inhibits activity of nAChRs due to carbachol, an ACh agonist (IC 50 : 20 ± 13 μM) (Fig 4).

Both negatively-charged and neutral forms of T3 inhibit nAChR
Environmental pH affects molecular charge. According to the titration curve of T3 (Fig 5), T3's pK2 value is slightly lower than 7.3, in which the hydroxyl group becomes anionic while the amine group remains neutral. However, over this pH interval, PS is expected to retain a negative charge. At physiological pH, there are both charged PS and a proportion of charged  https://doi.org/10.1371/journal.pone.0223272.g004 L-3,3',5-triiodothyronine and pregnenolone sulfate inhibit Torpedo nicotinic acetylcholine receptors T3 molecules. We performed TEVC tests of nAChR inhibition due to T3 at different pHs; changing the surrounding pH allowed observation of effects due to T3's amino acid (headgroup) in different charge states, as well as effects due to different charge states of the amino acids on the receptor, on nAChR binding. These effects could be compared to the effect due to the unchanging PS sulfate group (headgroup) charge. The pH environment affects inhibition of nAChRs by both T3 and PS. Dose response curves show the effect of pH (6-9, in 0.5 pH increments) on the IC 50 of T3 and PS (Fig 6A-6G). A two-way ANOVA (for hormone and dose) indicates that pH has a significant effect on nAChR flux for experiments using T3 (p < 0.0001, F(6, 70) = 74.25) or PS (p < 0.0001, F(6, 70) = 24.02) and that the effect due to the pH environment and due to T3 (p < 0.0001, F(6,70) = 11.69) or PS (p < 0.0001, F(6,70) = 6.214) dose have interactive effects on nAChR flux.
The effects of pH on inhibition of nAChRs by T3 and PS diverge at the T3 pK2 value. From pH 6-7, the IC 50 values of T3 and PS (Fig 6H) are similar. However, above the pK2 value of T3 (~7.3), the IC 50 values of T3 and PS are distinct, with the T3 IC 50 s shifting downward relative to the IC 50 values of PS. This divergence is coincident with the increase in extracellular pH above the pK2 value of T3. The PS IC 50 value decreases below T3's at pH 9, where the net charge of T3 becomes -1, and becomes more prominent than the neutral form dominant at lower pHs. Figs 8 and 9 show representative traces of inhibition due to T3 and PS over 60-second administrations. Raw traces of 0-10 μM T3 or PS at pH 6 and 7.5 (Fig 8) illustrate the effect of dose on  Fig 7). The dotted vertical line denotes the pK2 value of T3. the decay rate of ion conductance. According to a two-way ANOVA, dose of inhibitor affects decay rate for both T3 (P<0.0001) and PS (P<0.0001; n = 3, per dose, per inhibitor) ( Table 2, Table 3). The representative traces of a 10 μM dose of T3 or PS at each pH (6-9, in 0.5 pH increments) (Fig 9) demonstrate the effect due to ligand dose and pH on desensitization of the channel. According to a two-way ANOVA, at corresponding pHs, PS-inhibited channels exhibit a significantly different decay rate of ion conductance from T3 (P<0.0001) ( Table 2). The effect of each ligand on nAChR desensitization may be involved with a region https://doi.org/10.1371/journal.pone.0223272.g009 L-3,3',5-triiodothyronine and pregnenolone sulfate inhibit Torpedo nicotinic acetylcholine receptors homologous to the desensitization gate structure of GABA A receptors, where PS has been shown to bind [38,39]. This binding is theorized to cause constriction at the base of the channel pore, inducing a desensitized state [38]. Differences in binding site, orientation or binding affinity may cause differential effects on desensitization.

Discussion
Here we present the first direct observation of T3's inhibitory effect on nAChRs (Figs 2 and 6). The effect due to T3 on nAChRs is quantitatively similar to the effect due to T3 on GABA A receptors. For the first time, we also demonstrate an inhibitory effect due to PS on nAChR activity (Fig 2B). This effect is comparable to the inhibitory effect due to PS on GABA A receptors [57], and to T3's effect on nAChRs (Fig 2A) and GABA A receptors [29]. It has been hypothesized that lipophilic ligands have opposing functional effects on GABA A receptors and nAChRs, [45] but here we see this trend does not extend to PS and thyroid hormones. T3 may act as a neurosteroid-like inhibitor of nAChRs. Thyroid hormone can access the brain via the blood-brain barrier [58][59][60]; nerve terminal fractions show T3 concentrations of 13.0-65 nM [61,62], and synaptosomal T3 content is elevated 9.5 times in hypothyroid rats compared when compared to euthyroid rats [63]. Clinical presentations are associated with longer and less restful sleep in hypothyroid patients [64][65][66], and with insomnia and anxiety in hyperthyroid patients [67,68]. When injected into the median preoptic nucleus of freely-moving adult rats, T3 increases REM sleep and wakefulness [69,70], demonstrating short-term, nongenomic effects on behavior. T3 also directly inhibits GABA A receptor activity in vitro [29,34]. The similarity of effects due to T3 and due to neurosteroids, which also modulate sleep [71][72][73][74] and anxiety [75,76], may indicate similar mode of action due to T3 and neurosteroids.
Further, T3 structure resembles the structure of neurosteroids. While it lacks a steroid's fundamental four-ring molecular group, T3 has a nearly identical volume and shape to PS [33]. Based on molecular dynamics simulations in which the two molecules favorably occupy TM-localized regions, in conjunction with GABA A R binding data, Westergard et al. posit that T3 and the PS-like molecule allopregnanolone have a shared binding site in GABA A receptors [29]. This would indicate a possible similar mode of T3 action to that of neurosteroids.
T3 appears to be an exception to established neurosteroid structure-activity relationships. Here we may reject the hypothesis that a population of anionic T3 transduces the inhibitory effect of T3 in nAChR. Further study using analogues of T3 with singular changes in functional groups (e.g., the presence or absence of an amino group, or of bulky and non-bulky groups) may be required to isolate the molecular origin of its unexpected inhibitory effects.
We show here that T3 and PS both affect channel desensitization. Further, changing environmental pH levels appear to affect T3 and PS-induced nAChR desensitization. This may be a result of pH influence on channel residues at the TMD, as has been suggested by previous investigations [54,55].
The present work adds to the mounting evidence of T3's nongenomic effect on adult brain tissue [34, 69,70,[77][78][79][80]. These effects may be mediated through interactions with the nAChRs, with GABA A receptors, or with both. This work also indicates, through the inhibitory action of PS and the inhibitory, neurosteroid-like action of T3 on the nAChR, a complex structure-function relationship. The actions of both ligands relative to one another, and to the nearly identical molecules allopregnanolone and triac, can be a tool to investigate binding and inhibition of the channel.  Martin.