Dynamic Modification of a Mutant Cytoplasmic Cysteine Residue Modulates the Conductance of the Human 5-HT3A Receptor*

Structural models suggest that Arg436 lies within five cytoplasmic portals of the 5-HT3A receptor. We tested both the accessibility of residue 436 and the influence of its charge on single channel conductance (γ) by substituting Arg436 with Cys and examining the effect of methanethiosulfonate (MTS) reagents on γ. Inclusion of positively charged 2-aminoethyl-MTS (MTSEA) within the electrode solution reduced γ of 5-HT3A(R436C) receptors in outside-out patches from 7.8 ± 0.5 to 5.0 ± 0.5 picosiemens (pS). To increase γ, we substituted Arg436 by Cys in the 5-HT3A(R432Q,R440A) mutant, yielding the 5-HT3A(QCA) construct with a γ of 17.7 ± 0.4 pS. Modification of 5-HT3A(QCA) receptors by MTSEA or 2-(trimethylammonium) ethyl-MTS reduced γ to 8.7 ± 0.5 and 6.7 ± 0.4 pS, respectively, both significantly below that of channels exposed to nonpolar propyl-MTS. Extracellular MTSEA, but not 2-(trimethylammonium) ethyl-MTS, crossed the membrane and in so doing slowly (τ = 94 s) reduced γ. MTSEA more rapidly (τ = 15 s) reduced the γ of 5-HT3A(QCA) receptors in inside-out patches, an effect reversed by the reducing agent dithiothreitol. Cys436 modification by negatively charged 2-carboxyethyl-MTS and 2-sulfonatoethyl-MTS increasedγ to 23 ± 1.0 and 26 ± 0.7 pS, respectively. MTS reagents did not affect γ values for 5-HT3A(QDA) constructs with Cys substituted for Lys431 predicted to be outside the entrance to the portals. Collectively, the data demonstrate that the dynamic modification of the charge of a cytoplasmic residue regulates γ, consistent with the existence of cytoplasmic portals that impose a rate-limiting barrier to ion conduction in Cys loop receptors.

5-hydroxytryptamine type-3 (5-HT 3 ) receptors, and Zn 2ϩ -activated channels conduct cations, whereas ␥-aminobutyric acid type A and glycine receptors conduct anions (1). Cys loop receptor subunits combine as pentamers forming a central ion pore that traverses the plasma membrane. Studies examining the role of residues comprising the pore-lining second transmembrane (TM2) helices of the nACh receptor led to the traditional view that this region of Cys loop receptors serves as the rate-limiting barrier to ion flux. Among the cationselective Cys loop receptor subunits, conserved acidic residues influence unitary conductance (␥) by forming concentric rings of negative charge positioned along the vertical axis of the central pore (2).
Although the role of residues within the TM2 helix of Cys loop receptors in controlling ␥ is well established, recent studies demonstrate a similar function for residues located in the large cytoplasmic TM3-4 loop (1,3,4). Based on a 4.6 Å resolution cryoelectron microscopy image of the Torpedo marmorata nACh receptor, Miyazawa et al. (5) first speculated that amino acids in the TM3-4 loop form "transverse tunnels" that may contribute to the ion conduction pathway. Subsequent electrophysiological studies demonstrated that cytoplasmic residues within membrane-associated (MA) helices of the 5-HT 3 and ␣ 4 ␤ 2 nACh receptors are determinants of ␥ (3,4). Arg 436 in the 5-HT 3A subunit at a position termed MA 0Ј is critical for maintaining the anomalously low ␥ (ϳ900 femtosiemens) that is a hallmark of the homomeric 5-HT 3A receptor. Furthermore, introduction of MA 0Ј Arg into ␣ 4 ␤ 2 nACh receptors halved ␥ (4).
A 4 Å resolution model of the T. marmorata nACh receptor indicates that adjacent MA helices frame portals with apertures only slightly larger than partially hydrated permeant cations (6). Using this structure as a template, homology models of the 5-HT 3A and ␣ 4 ␤ 2 nACh receptors indicate that charged amino acids line their cytoplasmic portals (4).
Although it is now clear that specific MA helix residues influence the ␥ of nACh and 5-HT 3 receptors, it remains to be determined whether their charge plays a role. We used substituted cysteine modification to address this question, a method fre-quently used to determine whether residues are located within water-accessible regions of proteins (7). Methanethiosulfonate (MTS) reagents form disulfide bonds rapidly with accessible cysteine residues. Instead of examining accessibility per se, we used a series of MTS compounds with differing charges to examine how the properties of the modified cysteine residue, substituted at the 5-HT 3A receptor MA 0Ј-position, influence ␥. This approach is advantageous compared with systematic mutagenesis; basic amino acids are large compared with acidic residues; thus, charge and volume cannot be treated as separate variables. Cysteine modification by contrast enables alterations of charge while minimizing changes in side chain volume. Furthermore, the effect of sulfhydryl modification on ␥ can be observed in real time and reversed by administration of the reducing agent, dithiothreitol, thereby diminishing the likelihood that altered function is caused by gross changes in assembly and/or tertiary structure.
Our results demonstrate that the dynamic modification of the charge of the MA 0Ј Cys residue modulates the ␥ of the 5-HT 3A receptor.

5-HT 3A Constructs and Transfection of Subunit cDNAs-
cDNAs encoding human wild type 5-HT 3A and mutant constructs were cloned into pGW1. Point mutations were introduced into the 5-HT 3A construct using standard molecular biological techniques, and all constructs were sequenced to confirm fidelity. Transient transfection of HEK293 or tsA-201 cells with cDNA was performed using the calcium phosphate precipitation method or electroporation, respectively, as described previously (3,8). Cells were subcultured twice weekly and incubated in a medium composed of Dulbecco's modified Eagle's medium and 10% calf serum, supplemented with 100 g/ml streptomycin and 100 units/ml penicillin. Cells were maintained at 37°C in an atmosphere of 5% CO 2 (100% relative humidity). HEK293 cells were incubated at 37°C overnight during the transfection and were washed after 16 h with medium. Cell culture reagents were purchased from Invitrogen.
Electrophysiology-Outside-out and inside-out patch configurations were used to record single channel currents from patches excised from transfected cells. The bath solution contained 140 mM NaCl, 2.8 mM KCl, 2.0 mM MgCl 2 , 1.0 mM CaCl 2 , 10 mM glucose, and 10 mM HEPES (pH 7.2 adjusted with NaOH). Patch electrodes in the outside-out configuration were filled with a solution comprising 130 mM potassium gluconate, 5 mM NaCl, 2 mM MgCl 2 , 5 mM EGTA, 0.1 mM ATP (Mg 2ϩ salt), and 10 mM HEPES (pH 7.2 adjusted with KOH). The same bath solution was used for whole-cell current recordings, and electrodes contained 140 mM CsCl, 2.0 mM MgCl 2 , 0.1 mM CaCl 2 , 1.1 mM EGTA, 0.1 mM ATP (Mg 2ϩ salt), 10 mM HEPES, pH 7.2. 5-HT was dissolved in the bath solution and applied locally by pressure ejection to whole cells and outside-out patches held at Ϫ60 and Ϫ74 mV, respectively (the latter includes correction for a 14-mV liquid junction potential associated with the use of potassium gluconate). Inside-out patch electrodes were filled with the bath solution containing 5-HT (1 M), and the electrode potential was clamped at ϩ80 mV, corresponding to a holding potential of Ϫ80 mV. Stock solutions of MTS reagents (200 mM) obtained from Toronto Research Chemicals (Ontario, Canada) were stored at Ϫ20°C. Prior to each experiment, MTS reagents were diluted in the electrode solution (or in the extracellular solution in some experiments using MTSEA and 2-(trimethylammonium)ethyl-MTS (MTSET)) for outside-out patch experiments and in the bath solution for insideout patch experiments to yield a final concentration of 200 M. Single channel currents were recorded using an Axopatch 200A (Axon Instruments) and low pass-filtered at 1 kHz. Data were digitized (Digidata 1322A; Axon Instruments) at 10 kHz onto the hard drive of a PC for subsequent offline analysis. Sections of data recorded from outside-out patches ϳ10 s in length, in which unitary events predominated were selected for analysis and leak-subtracted using Clampfit (pCLAMP 8.0; Axon Instruments) for the generation of all-points amplitude histograms using Fetchan (pCLAMP 8.0). Multiple Gaussian distributions were fitted (least squares minimization) to all-points amplitude histograms using the Simplex method within pSTAT (pCLAMP 8.0) and used to determine single channel current amplitude as described previously (4). Single channel conductance (␥) values are reported as the chord conductance determined from the relationship ␥ ϭ i/(V m Ϫ E rev ), in which i is the current amplitude of single channel events, V m is the holding potential, and E rev is the experimentally determined reversal potential. The onset of reduced 5-HT-activated single channel amplitude following local pressure application of MTSEA (200 M) to inside-out patches was monitored by measuring the amplitudes of all unitary events individually using cursors in Clampfit. The recovery of single channel amplitude following the local pressure application of 100 mM dithiothreitol (DTT) was similarly monitored. The time courses of channel amplitude attenuation by MTSEA and its reversal by DTT were determined by fitting each with an exponential function. To assess the influence of extracellularly applied MTSEA (200 M) or MTSET (200 M), single channel events recorded over 1-min epochs were used to construct amplitude histograms from which ␥ was derived. The rates at which extracellularly applied MTSEA modified the amplitudes of single channel and macroscopic currents, recorded from outside-out patches, and whole cells were determined by fitting single exponential functions to plots of mean current amplitude (recorded every 60 and 30 s, respectively) versus time.
Statistics-Data are presented as the mean Ϯ either S.E. or S.D., as indicated. Data sets were routinely compared using one-way analysis of variance with a post hoc Tukey's test. However, the paired t test was used to compare ␥ values obtained from inside-out and outside-out patch recordings before and after the addition of MTS reagents.
Modeling the Structures of WT 5-HT 3A and Mutant Receptors Using the 4 Å Resolution Model of the T. marmorata nACh Receptor-Homology models were generated as previously described (4). Briefly, amino acid sequences for the human WT 5-HT 3A , the 5-HT 3A (R432Q,R436C,R440A), and the 5-HT 3A (K431C,R432Q,R436D,R440A) subunits were aligned against the T. marmorata ␣, ␤, ␦, and ␥ subunits using MultiAlign (available on the World Wide Web at prodes.toulouse.inra.fr/multalin/multalin.html). Gaps between the template and query sequences in the extracellular domain were filled in with corresponding amino acids from the template sequence. The C termini of the query sequences were shorter than the templates and were filled in with amino acids from the template subunits. The structure of the T. marmorata nACh receptor was downloaded from the RCSB Protein Data Bank (Protein Data Bank code 2BG9) and loaded into a Deepview Swiss-PdbViewer for imaging and modeling. Wild type 5-HT 3A and 5-HT 3A (QCA) query sequences were threaded onto the backbone of the nAChR model using the "Fit Raw Sequence" and "Alignment" features in Deepview. Models were then submitted to Swiss-Model (available on the World Wide Web at swissmodel.expasy. org) for optimization. Following energy minimization (Gromos96, SwissModel), structures were uploaded into Deepview for imaging. The subunit interface illustrated ( Fig. 6) is equivalent to the T. marmorata 1Ј␣-5Ј␥ portal. The Protein Data Bank model of cysteine-MTSEA was imported from Spartan'04 (Wavefunction, Inc., Irvine, CA) and for illustrative purposes was manually aligned and positioned onto the cysteine residues of the 5-HT 3A (QCA) and 5-HT 3A (K431C,QDA) homology models.
Amino Acid Volumes-We used Spartan'04 (Wavefunction Inc., Irvine, CA) to estimate all volumes to ensure consistency when comparing amino acids and cysteine residues modified by MTS reagents. Structures were energy-minimized in the extended chain conformation. The volumes of amino acids determined by Spartan'04 correlated well with those previously published by Zamyatnin (9). A linear regression fitted to a plot of the amino acid volumes determined by Spartan'04 versus those published previously (9) yielded a coefficient of determination (r 2 ) of 0.97 (data not shown).

Substituted Cysteine Modification of Mutant 5-HT 3A Receptors by
Methanethiosulfonate Reagents-Wild type (WT) 5-HT 3A subunits combine to form homopentameric receptors that have a ␥ below the resolution of single channel recording (3,4,8). The anomalously low ␥ (estimated Right, an alignment of amino acids within the MA helices of WT 5-HT 3A and 5-HT 3B subunits and the mutant 5-HT 3A constructs used in this study, indicating the positions of Ϫ4Ј, 0Ј, and 4Ј residues (boxed). 5-HT 3A (QCA) and 5-HT 3A (QDA) represent the triple mutant 5-HT 3A (R432Q,R436C,R440A) and 5-HT 3A (R432Q,R436D,R440A) receptors, respectively. Substituted residues within MA helices of mutant constructs are indicated in boldface type. Amino acids are numbered according to the human 5-HT 3A amino acid sequence (20). Note that the sequences shown do not illustrate the entire MA helices. B, a general scheme for an MTS compound reacting with the sulfhydryl group on the side chain of a cysteine residue. R represents the chemical group that varies among MTS compounds (see Table 1 for structures). C, 5-HT (10 M)-activated current recorded at Ϫ74 mV from an excised outside-out patch containing the single point mutant 5-HT 3A (R436C) receptor under control conditions. The all points amplitude histogram derived from single channel events during the deactivation of the exemplar current was fitted with two Gaussian distributions representing the open and closed states. The single channel current amplitude was Ϫ0.72 pA, corresponding to a ␥ of 9.7 pS. D, 5-HT (10 M)-activated current recorded with the addition of MTSEA (200 M) to the electrode solution. Gaussian distributions fitted to the all points amplitude histogram provide an estimated single channel amplitude of Ϫ0.28 pA, corresponding to a ␥ in this case of 3.8 pS. In C and D, expanded sections of data are shown below each trace. In both cases, after acquisition, digitized currents were low pass-filtered at 500 Hz with a Gaussian filter to improve the signal/noise ratio. See Table 1 for mean ␥ values for control and MTSEA-modified receptors.
by variance analysis to be ϳ900 femtosiemens) derives from residues located in the MA stretch that form putative intracellular portals depicted in the ribbon rendering of the 5-HT 3A receptor homology model (Fig. 1A). The ␥ of 5-HT 3 receptor channels is enhanced to the resolvable range either by incorporation of the 5-HT 3B subunit into heteropentameric 5-HT 3A/B receptors (8,10) or by replacing the 5-HT 3A subunit MA helix 0Ј Arg by Asp, the MA 0Ј residue of the 5-HT 3B subunit (4) (Fig.  1A). The receptor assembled from human 5-HT 3A subunits harboring the single point mutation R436D has a ␥ ϳ 10 times that of the WT receptor (4). The R436D substitution introduces both opposite charge and reduced side chain volume at the MA 0Ј-position, and it is unclear whether a favorable change in local electrostatic potential, reduced steric hindrance, both, or other factors contribute to the enhancement of ␥.
To address this issue, we used substituted cysteine modification to examine the ␥ of mutant receptors containing varying chemical entities at the MA 0Ј-position. We created 5-HT 3A constructs in which Cys replaced Arg at the 436position ( Fig. 1A) and subsequently modified the physicochemical properties of the substituted residue with MTS compounds that react rapidly and specifically with thiolate groups ( (Table 1). Thus, such agents differ predominantly in the substituent attached to an ethanethiol chain (i.e. positively charged ammonium, neutral methyl, and negative carboxylate and sulfonate groups) ( Table 1). We also utilized the larger quaternary ammonium compound MTSET. The estimated volume of cysteine modified by MTSET was 231 Å 3 (Table 1).
MTSEA Reduces ␥ of the 5-HT 3A (R436C) but Not the 5-HT 3A (R436D) Mutant Receptor-The single amino acid substitution of the MA 0Ј Arg by Cys in the mutant homomeric 5-HT 3A (R436C) receptor increased ␥ to a level that could be directly observed in recordings from outside-out patches. The transient application of 5-HT (10 M) to outside-out mem-brane patches clamped at Ϫ74 mV elicited rapidly rising inward currents from which unitary events with a mean ␥ of 7.8 Ϯ 0.5 pS (Table 1) were discernable during current deactivation (  Table 1). Since the ␥ of WT 5-HT 3A receptors is too low to enable direct observation of unitary events (3,4,8), we used homomeric mutant 5-HT 3A (R436D) receptors as a control to exclude potential effects of MTSEA that may occur independently of the substituted cysteine residue. MTSEA had no significant effect on the ␥ of events mediated by homomeric 5-HT 3A (R436D) receptors, which was 9.1 Ϯ 0.6 and 9.8 Ϯ 0.9 pS in the absence and presence of the reagent, respectively (Table 1). Therefore, the reduction in ␥ of the 5-HT 3A (R436C) receptor was caused specifically by modification of the substituted MA 0Ј Cys.
MTSEA is sufficiently membrane-permeant to access residues engineered within a cytoplasmic region of the Shaker B K ϩ channel when applied to the extracellular aspect of the membrane (11). Therefore, we investigated whether extracellular MTSEA could access the MA 0Ј residue of the 5-HT 3A (QCA)  construct in excised outside-out patches. In such experiments, single channel events evoked by pressure-applied 5-HT (10 M) were recorded prior to MTSEA application to obtain a control data set (Fig. 3, A and D). The subsequent addition of MTSEA (200 M) to the perfusate caused an ϳ50% reduction in ␥ (to 8.5 Ϯ 0.9 pS, n ϭ 6) within 2 min (Fig. 3, A and D). A single exponential function fitted to the onset of the reduction of the mean conductance by MTSEA yielded a ϭ 94 s (fit not shown). The extent of the depression of ␥ was essentially identical to that caused by intracellular application of MTSEA at the same concentration ( Fig. 3A and Table 1). A washout period of up to 20 min did not reverse the inhibition of ␥ by extracellularly applied MTSEA (data not shown). Furthermore, extracel-lular application of MTSEA did not decrease the ␥ of 5-HT 3A (QDA) receptors in outside-out patches (␥ ϭ 31.8 Ϯ 1.2, n ϭ 6), confirming that the inhibition of ␥ observed with 5-HT 3A (QCA) receptors is due to the specific reaction of MTSEA with MA 0Ј Cys, rather than a rapid "flickery" nonselective pore block from the extracellular environment that could, in principle, yield the impression of a reduction in ␥ (Fig.  3, B and D).
In parallel experiments, bath application of the membrane-impermeant compound MTSET (200 M) (11) to 5-HT 3A (QCA) receptors in excised outside-out patches had no influence upon ␥ (17.2 Ϯ 0.8 pS, n ϭ 6). However, in common with MTSEA, MTSET was clearly effective when applied to the intracellular aspect of the membrane, reducing ␥ to 6.7 Ϯ 0.4 pS (Fig. 3, C and D, and Table 1).
Extracellular MTSEA Crosses the Cell Membrane to Reduce 5-HT 3A (QCA)-mediated Currents-MTSEA has a smaller volume than MTSET (Table 1); thus, it is possible that the former, but not the latter, is able to pass through the open channel of the 5-HT 3A (QCA) receptor and modify Cys 436 when applied to the extracellular aspect of an outside-out patch. Alternatively MTSEA, which exists in both charged and uncharged species (11), may access the cytoplasmic Cys 436 residue by permeating the membrane, a route denied to MTSET, which is permanently charged and therefore membrane-impermeant. We performed whole-cell experiments to determine whether the inhibition of current by extracellularly applied MTSEA required activation of 5-HT 3A (QCA) receptors, which would indicate a necessity for the reagent to pass through the open channel (Fig. 4). First, we applied 5-HT (30 M) by local pressure ejection every 30 s throughout the course of MTSEA (200 M) application to the recording chamber (Fig. 4A). Using this approach, MTSEA irreversibly inhibited 5-HT-evoked macroscopic currents to an extent (ϳ50%) and with kinetics ( ϭ 112 s; Fig. 4C), similar to the inhibition of ␥ observed when MTSEA was applied to excised membrane patches (Figs. 2 and 3D and Table 1). In the second series of experiments, we established the control 5-HT (30 M)-evoked current amplitude and terminated receptor activation prior to MTSEA application to the recording chamber. After a 240-s exposure to MTSEA, in the absence of channel activation, we resumed 5-HT application every 30 s (Fig. 4B). Using this approach, MTSEA applied in the absence of repetitive 5-HT 3A (QCA) receptor activation inhibited currents subsequently activated by 5-HT by ϳ50%, similar to the level achieved over the same time course with repetitive channel activation (Fig. 4C). Such data suggest that, under the conditions of these experiments, MTSEA gained access to Cys 436 primarily by permeating the cell membrane.
Negative Charge of the MA 0Ј Residue Overcomes Steric Hindrance to Increase ␥-Neutral PMTS also caused a significant reduction in ␥ (to 11.9 Ϯ 0.4 pS) of 5-HT 3 (QCA) receptors relative to control (Fig. 5A), but the effect was not as pronounced as that found for the positively charged MTSEA or for MTSET (p Ͻ 0.001; Table 1). These data suggest that for cysteines modified by MTS reagents of comparable volume (Table 1), the introduction of a positive charge causes an additional attenuation of ␥. The impor-  tance of the local potential was confirmed by the inclusion of the negatively charged MTSCE in the electrode solution, which caused a significant increase in ␥ to 22.7 Ϯ 1.0 pS ( Fig.  5B and Table 1). Similarly, the introduction of negative charge by reaction of the 5-HT 3A (QCA) receptor with MTSES (Fig. 5C) also increased ␥ to 26.2 Ϯ 0.7 pS (Table 1). Interestingly, the ␥ determined in the presence of MTSES was statistically greater than that observed for MTSCE (p Ͻ 0.001; Table 1). The ␥ values for the 5-HT 3A (QDA) receptor in the absence and presence of all MTS compounds are summarized in Table 1. Importantly, none of the MTS reagents caused a significant change in ␥ of the 5-HT 3A (QDA) receptor. Collectively, these data demonstrate that MTS reagents modulate ␥ of 5-HT 3A (QCA) receptors by reacting specifically with the thiolate group present on the cysteine residue substituted into the MA 0Ј-position and emphasize the importance of charge at this position.
MTSEA Reduces ␥ of the Mutant 5-HT 3A (QCA) Receptor in Excised Inside-out Patches-In experiments performed on outside-out membrane patches with the reagent in the electrode, it was not possible to observe the onset of changes in ␥, presumably due to the rapid reaction rate of MTS reagents with the substituted cysteine. At the commencement of recording, a relatively homogeneous population of single channel amplitudes was observed, suggesting that thiolate modification had reached equilibrium. A time-dependent reduction in ␥ of the 5-HT 3A (QCA) receptor was observed when MTSEA was added to the extracellular aspect of the excised outside-out patch (Fig. 3B), which was paralleled by a quantitatively similar decrement in the peak amplitude of the macroscopic current response recorded from whole cells (Fig. 4B). However, the kinetics of the development of thiolate modification probably does not represent the rate of cysteine modification, because the time over which the deprotonated species of MTSEA (ϳ5% assuming a pK a of 8.5 (11)) diffuses across the membrane is unknown, as is the concentration of the reagent in the "intracellular" environment. Therefore, we used the inside-out patch configuration to demonstrate the effects of cysteine modification on the ␥ of the 5-HT 3A (QCA) receptor in real time. In such experiments, 5-HT (1 M) was included within the electrode solution, and MTSEA (200 M) was applied to the cytoplasmic aspect of the excised patch by pressure ejection. Patches were bathed in symmetrical NaCl-based solutions with an electrode potential of ϩ80 mV, corresponding to a holding potential (Ϫ80 mV) close to that used in the outside-out patch recordings (Ϫ74 mV). 5-HT-activated single channel events were first recorded for several minutes in the absence of MTSEA to generate a control data set. Channel openings mediated by 5-HT 3A (QCA) receptors under such conditions occurred sporadically with a ␥ of 18.2 Ϯ 1.7 pS (Fig. 6A and  Table 1. Table 1), their low frequency presumably resulting from the low 5-HT concentration employed to minimize receptor desensitization. The subsequent application of MTSEA reduced the amplitude of single channel currents recorded from inside-out patches by ϳ50% (Fig. 6B). MTSEA-modified unitary events had a mean ␥ of 8.7 Ϯ 0.4 pS ( Table 1). Cysteine modification by MTS reagents can be reversed by high concentrations of reducing agents, such as DTT (11). Pressure application of DTT (100 mM) to an inside-out patch containing 5-HT 3A (QCA) receptors modified by MTSEA caused a slow recovery of the amplitude of 5-HT-activated single channels toward values observed prior to the application of MTSEA (Fig. 6C). We investigated the time course of MTSEA-modification by measuring the amplitude of 5-HTactivated single channels before, during, and after MTSEA application. The successful completion of this experiment requires sustained channel opening events over a prolonged period of time despite the confounding influence of desensitization caused by constant exposure of the receptors to 5-HT within the electrode solution. Unfortunately, the majority of patches exhibited sporadic channel activity with insufficient events, and thus temporal resolution, to determine the rate at which MTSEA reduced ␥. However, in the exemplar patch (Fig. 6D), channel activity was relatively frequent, and the reduction in ␥ had a ϭ 15 s, determined by fitting an exponential function to the data points during the onset of the action of MTSEA. Reversal of the effect of MTSEA by DTT had a ϭ 77 s (Fig. 6D). This experiment demonstrates that real time modification of the physicochemical properties of MA 0Ј residue reduces ␥, an effect that can be reversed by application of a reducing reagent.
Modification of a Cysteine Residue at MA Ϫ5Ј outside the Putative Portal Did Not Affect ␥-We constructed homology models of wildtype and mutant 5-HT 3A receptors based on the structural model of the T. marmorata (Fig. 7). One of the five putative cytoplasmic portals framed by adjacent MA helices is shown in each case. Three arginine residues lie at the entrance to portals of the wild-type 5-HT 3A receptors (Fig. 7A), located at the MA Ϫ4Ј-, 0Ј-, and 4Ј-positions (Fig. 1A), each of which contribute to the receptor's characteristically low ␥ (4). The portal of the 5-HT 3A (QCA) receptor by comparison contains smaller uncharged residues (Fig. 7B). Modification of the MA 0Ј Cys by MTSEA adds substantial bulk and a positive charge within the portal (Fig. 7C). Consistent with the model, this modification substantially reduces the ␥ of the 5-HT 3A (QCA) receptor (Figs. 2, 3, and 6 and Table 1). To further test the validity of the structural model, we substituted a Cys for Lys 431 in the 5-HT 3A (QDA) mutant receptor background. The model (Fig. 7D) predicts that Cys 431 at the MA Ϫ5Ј-position (Fig. 1A) resides outside the portal at a locus where modification of its charge and volume by MTS reagents would be anticipated to have a minimal effect upon ␥. We specifically selected MA Ϫ5Ј Lys for modification, because the replacement of this basic residue by Cys is broadly analogous to the Arg to Cys mutation at the MA 0Ј-position. Outside-out patch recordings revealed that 5-HT-evoked single channels mediated by the 5-HT 3A (K431C,QDA) receptor had a small but significantly (p Ͻ 0.01) increased ␥ compared with those mediated by the 5-HT 3A (QDA) receptor (Table 1). However, consistent with the predicted location of Cys 431 outside the portal (Fig. 7D), FIGURE 6. DTT reverses the MTSEA-induced reduction of ␥. A, 5-HT (1 M)-activated currents recorded under control conditions from an excised inside-out patch expressing 5-HT 3A (QCA) mutant receptors. 5-HT was applied to the inside-out patch by inclusion within the recording electrode. The patch was held at ϩ80 mV and bathed with a symmetrical NaCl-based solution. The unitary current amplitude of sporadic 5-HT-activated currents in control was determined to be 1.5 pA from the all points amplitude histogram fitted with the sum of two Gaussians. B, the effect of MTSEA (200 M) applied by pressure ejection for 40 s directly onto the exposed "cytoplasmic" aspect of the same patch used in A. During the application of MTSEA, the amplitude of single channel currents declined to 0.66 pA (determined from the all points amplitude histogram). Mean ␥ values of channels recorded from inside-out patches containing 5-HT 3A (QCA) mutant receptors before and after MTSEA application are provided in Table 1. C, application of the reducing agent DTT (100 mM) caused an increase in the amplitude of single channel currents compared with those recorded from the same patch following modification by MTSEA (B). The corresponding all points amplitude histogram reveals that the amplitude of unitary currents after treatment with DTT was 1.3 pA. D, the time course of the effects of MTSEA and DTT on the amplitudes of single channels recorded from the inside-out patch from which data were obtained for A-C. The amplitudes of all unitary events were measured to construct the graph of current amplitude versus time illustrated in D. Data points before and after the abscissa break represent the average single channel current amplitudes for events occurring in the preceding 4 and 10 s, respectively. The vertical lines indicate ϮS.D., and data points were fitted with single exponential functions, from which the time constants of onset and reversal of the actions of MTSEA were determined. modification of the residue by intracellular application of either MTSEA or MTSES to outside-out patches did not significantly affect ␥ (Table 1).

DISCUSSION
We used cysteine modification by a series of MTS reagents to investigate the mechanism by which the substituted 0Ј residue within the cytoplasmic MA helix influences ␥ of the human 5-HT 3A receptor. Substitution of the MA 0Ј Arg by Cys within either the WT 5-HT 3A receptor background or the 5-HT 3A (R432Q,R440A) construct in which Ϫ4Ј and 4Ј arginines were replaced by glutamine and alanine, respectively, substantially increased ␥ compared with equivalent values observed previously in the absence of the 0Ј Cys residue (4). Furthermore, modification of substituted MA 0Ј Cys residues by MTS reagents either further increased or reduced ␥, depending on the physicochemical properties of the reagent. This study lends further support to the cryoelectron microscopy-derived atomic scale model of Cys loop receptors based on the T. marmorata nACh receptor in which adjacent MA helices frame cytoplasmic portals (Fig. 7), forming an obligate pathway through which ions pass to traverse the cell membrane (6).
The homology model of the human homomeric WT 5-HT 3A receptor based on the 4 Å resolution structure of the T. marmorata nACh receptor (6) suggests that MA Ϫ4Ј, 0Ј, and 4Ј arginine residues lie near the mouth of each putative intracellular portal (Fig.  7A). Indeed, there is a preponderance of basic residues in the MA helices of the 5-HT 3A receptor (Fig. 1A), and it is likely that this concentration of positive charge contributes to the anomalously low ␥ of the channel (4,8). Consistent with this hypothesis, structural models of the T. marmorata and ␣ 4 ␤ 2 nACh receptors, which both have vastly higher ␥ values than that of the WT 5-HT 3A receptor, indicate that acidic residues predominate within the portals and their immediate vicinity (4,6).
Both the volume and charge of residues within the TM2 helices lining the central channel pore play a role in determining ␥. The identity of the amino acid at the TM2 Ϫ1Јposition near the cytoplasmic face of the membrane is a crucial determinant of ␥. Substitution of the Ϫ1Ј Glu by polar uncharged Gln in the ␣ subunit reduced the ␥ of the Torpedo californica nACh receptor ϳ4-fold (2,12). Even more profound reductions in ␥ ensue from the mutation of the Ϫ1ЈGlu, or Ϫ2Ј Gly, of the ␦-subunit of adult muscle nACh receptor to lysine (13). Charge inversion substitutions at the extracellular 20Ј and intracellular Ϫ4Ј rings of the T. californica ␣-subunit resulted in a 50% reduction in ␥. Introduction of the bulky Tyr residue in place of the smaller Thr or Ser at the 2Ј central ring position also reduced ␥ by 50% (14), suggesting that the pore narrows considerably in this region. Recent data indicate the most constricted region of the adult muscle nACh receptor to be at positions Ϫ1Ј and Ϫ2Ј (13).
Consistent with the strategic location of the cytoplasmic MA 0Ј arginine in the 5-HT 3A subunit, wherein the side chain gua- nidinium group is predicted to protrude into the mouth of the portal (Fig. 7A), its replacement by various other amino acids has a large impact on ␥. Single point substitutions to Asp, Glu, Gln (4), or Cys cause increases in ␥ values of mutant 5-HT 3A (R436X) receptors, such that single channels can be resolved in outside-out patch recordings. All of these residues differ from the native MA 0Ј arginine in both their smaller side chain volume and lack of positive charge. Either or both of these physicochemical properties may contribute to an increased ␥. The 4 Å model of the T. marmorata nACh receptor reveals that cytoplasmic portals have a maximum width of only ϳ8 Å, which is approximately equal to the diameter of a sodium or potassium ion that retains its first hydration shell. A comparison of homology models of the ␣ 4 ␤ 2 nACh receptor and the 5-HT 3A receptor suggests that the portals of the latter are further constricted (4). Consistent with the idea that portals of the 5-HT 3A receptor hinder ion flow due to their extreme narrowness, substitution of the MA 0Ј arginine (173 Å 3 ) by the large aromatic residue phenylalanine (174 Å 3 ) yielded a 5-HT 3A (R436F) mutant receptor with a low ␥, estimated by variance analysis to be ϳ400 femtosiemens (4).
The availability of a variety of MTS reagents that differ by charge and/or volume provides a useful tool for correlating modifications in the structure of the cysteine-modified MA 0Ј residue of the 5-HT 3A subunit with function, in this case ␥. Single channels recorded from outside-out patches containing mutant 5-HT 3A (R436C) receptors were readily resolvable, although the ␥ of this mutant was smaller than that observed when the MA 0Ј Arg was replaced by either Glu or Asp (4). In bulk water, cysteine is largely uncharged at physiological pH. Therefore, the smaller ␥ of mutant 5-HT 3A (R436C) receptors compared with either 5-HT 3A (R436E) or 5-HT 3A (R436D) receptors might be explained by a relative lack of a negative charge. However, it is not clear what the pK a of the cysteine side chain is within the environment of the MA helix of the 5-HT 3A receptor. The recent demonstration that the pK a values of amino acid residues within the TM2 of the ␦-subunit of the adult muscle nAChR can deviate markedly from their bulk water values highlights the need for caution in this respect (13).
Modification of 5-HT 3A (R436C) receptors by inclusion of MTSEA in the electrode solution caused an ϳ50% reduction in ␥ (to 5 pS), making single channel events difficult to resolve. By contrast, MTSEA had no effect on ␥ of 5-HT 3A (R436D) receptors. We previously demonstrated that replacement of MA Ϫ4Ј and 4Ј arginines by the equivalent residues in the 5-HT 3B subunit (Gln and Ala, respectively) elevates the ␥ of mutant 5-HT 3A receptors (3,4). The simultaneous replacement of all three arginines at MA Ϫ4Ј-, 0Ј-, and 4Ј-positions in the 5-HT 3A (QDA) triple mutant increased ␥ by ϳ4-fold in comparison with the ␥ of the single mutant 5-HT 3A (R436D) receptor (4). Similarly, the triple mutant 5-HT 3A (QCA) receptor had a ␥ more than double that of the single point mutant 5-HT 3A (R436C) receptor ( Table 1). The homology model of the 5-HT 3A (QCA) construct predicts that the relatively small size of the amino acid side chains at MA Ϫ4Ј, 0, and 4Ј increases the dimensions of the portals compared with the WT 5-HT 3A receptor (Fig. 7). We took advantage of the increased signal/ noise ratio of the 5-HT 3A (QCA) receptor in experiments investigating the modification of the MA 0Ј Cys.
Inclusion of the uncharged PMTS in the electrode solution significantly decreased the ␥ of 5-HT 3A (QCA) receptors in outside-out patches. Such data support the hypothesis that increased volume at MA 0Ј hinders ion conduction (Table 1). However, whereas PMTS and the positively charged reagent MTSEA modify cysteines to achieve similar estimated volumes (Table 1), the latter caused a significantly greater reduction in ␥ to less than half that recorded in the absence of the reagent.
MTSEA was equally effective as an inhibitor of ␥ when applied to the intracellular or extracellular aspect of the membrane. When applied to whole cells, the degree of inhibition of 5-HT-activated macroscopic currents was similar to the reduction in ␥ observed in excised patch recordings. Furthermore, the inhibition by MTSEA was independent of channel activation. Thus, under the conditions of these experiments, MTSEA primarily accesses the MA 0Ј residue by diffusion through the membrane as the uncharged species (11,15). Consistent with this interpretation, MTSEA reduced ␥ more rapidly when applied directly onto the intracellular aspect of an inside-out patch. Using this approach, the rate of reduction of the amplitude of unitary events was still substantially slower than the rate of MTS modification of simple thiol compounds and easily accessible substituted Cys residues in the outer mouth of the Shaker B K ϩ channel (16,17). Such data suggest that MTSEA encounters a rate-limiting barrier en route to the MA 0Ј residue within the portal. However, the estimated rate of accessibility derived from MTSEA application to the inside-out patch needs to be treated with caution due to the sporadic nature of channel gating with 5-HT included in the recording electrode.
The membrane-impermeant quaternary ammonium compound, MTSET, caused a reduction in ␥ only when applied to the intracellular aspect of the membrane. MTSET caused a marginally larger reduction in ␥ than was observed with MTSEA, in keeping with the larger estimated volume of the MTSET-modified cysteine (Table 1). Such observations are consistent with the hypothesis that the positive charge of the modified MA 0Ј Cys causes electrostatic repulsion of permeating cations. Interestingly the ␥ of the 5-HT 3A (QCA) receptor modified by MTSEA (or MTSET) ( Table 1) is comparable with that of the 5-HT 3A (QRA) construct (ϳ6.5 pS), in which the MA 0Ј arginine remained while the Ϫ4Ј and 4Ј arginines were replaced by Gln and Ala, respectively (4). This is perhaps not surprising, given that both the estimated volume and charge of the MTSEA-modified Cys resemble those of arginine (Fig. 6).
Based on the effects of MTSEA, MTSET, and PMTS, it would have been unsurprising if additional volume contributed by any MTS reagent reduced ␥ of modified 5-HT 3A (QCA) receptors. However, negatively charged MTSES and MTSCE reagents increased ␥ to values greater than that of the unmodified 5-HT 3A (QCA) receptor (Table 1). Such data suggest that the electrostatic attraction exerted by the negatively charged reagents more than compensates for the steric hindrance imposed by the increased volume of the modified MA 0Ј Cys. MTSCE contains a carboxyethyl group that is identical in structure to a glutamic acid side chain, yet the ␥ of 5-HT 3A (QCA)-MTSCE is less than that of the 5-HT 3A (QEA) receptor by Ͼ10 pS (4). This is consistent with greater steric hindrance of the MA 0Ј Cys plus a carboxyethyl group compared with a MA 0Ј Glu (volumes estimated to be 185 and 136 Å 3 , respectively). Interestingly, although MTSCE and MTSES both carry one net negative charge, the latter, despite its larger volume (Table 1), caused a greater increase in the ␥ of the 5-HT 3A (QCA) receptor. This is potentially due to the more strongly acidic nature of the sulfonate group compared with the carboxylate group. However, the pK a values for the sulfonate and carboxylate side groups of MTSES and MTSCE, respectively, in the environment of the 5-HT 3A portal are unknown.
The reducing agent DTT reversed the reduction in ␥ of the 5-HT 3A (QCA) receptor observed in the inside-out patch configuration following exposure to MTSEA. These data demonstrate that the effects of MTS reagents were caused specifically by cysteine modification. Furthermore, MTS-dependent changes in ␥ were seen only in channels containing a substituted MA 0Ј Cys. MTS reagents were without effect on the mutant 5-HT 3A (QDA) receptor even upon substitution of Cys for the MA Ϫ5Ј Lys. These data support the structural model of the 5-HT 3A receptor, which predicts that the MA Ϫ5Ј residue lies outside the portal (Fig. 7D). In future studies, it should be feasible to employ the substituted cysteine accessibility method to identify all residues within the MA stretch that impinge upon the permeation pathway in a manner analogous to that adopted to evaluate the channel-lining TM2 residues of, for example, the 5-HT 3A receptor (18,19).
Our results suggest that the MA 0Ј residue affects the ␥ of the 5-HT 3A receptor both by steric and electrostatic influences. Determination of whether the MA helix residues within the anionic subgroup of Cys loop receptors (i.e. ␥-aminobutyric acid type A and glycine receptors) influence ␥ by similar mechanisms awaits additional studies. The existence of residues accessible to the cytoplasmic milieu that dictate ␥ raises the possibility of dynamic regulation of fundamental properties previously thought to be obdurate in Cys loop receptors.