Common Determinants of Single Channel Conductance within the Large Cytoplasmic Loop of 5-Hydroxytryptamine Type 3 and α4β2 Nicotinic Acetylcholine Receptors*

Homomeric 5-hydroxytryptamine type 3A receptors (5-HT3ARs) have a single channel conductance (γ) below the resolution of single channel recording (966 ± 75 fS, estimated by variance analysis). By contrast, heteromeric 5-HT3A/B and nicotinic acetylcholine receptors (nAChRs) have picosiemen range γ values. In this study, single channel recordings revealed that replacement of cytoplasmic membrane-associated (MA) helix arginine 432 (-4′), 436 (0′), and 440 (4′) residues by 5-HT3B (-4′Gln, 0′Asp, and 4′Ala) residues increases γ to 36.5 ± 1.0 pS. The 0′ residue makes the most substantial contribution to γ of the 5-HT3AR. Replacement of 0′Arg by aspartate, glutamate (α7 nAChR subunit MA 0′), or glutamine (β2 subunit MA 0′) increases γ to the resolvable range (>6 pS). By contrast, replacement of 0′Arg by phenylalanine (α4 subunit MA 0′) reduced γ to 416 ± 107 fS. In reciprocal experiments with α4β2 nAChRs (γ = 31.3 ± 0.8 pS), replacement of MA 0′ residues by arginine in α4β2(Q443R) and α4(F588R)β2 reduced γ slightly. By contrast, the γ of double mutant α4(F588R)β2(Q443R) was halved. The MA -4′ and 4′ residues also influenced γ of 5-HT3ARs. Replacement of nAChR α4 or β2 MA 4′ residues by arginine made current density negligible. By contrast, replacement of both -4′ residues by arginine produced functional nAChRs with substantially reduced γ (11.4 ± 0.5 pS). Homology models of the 5-HT3A and α4β2 nAChRs against Torpedo nAChR revealed MA -4′, 0′, and 4′ residues within five intracellular portals. This locus may be a common determinant of ion conduction throughout the Cys loop receptor family.

There are five families of vertebrate Cys loop receptors as follows: the nicotinic acetylcholine receptor (nAChR), 3 the 5-hydroxytryptamine type 3 receptor (5-HT 3 R), the zinc-activated ion channel, the ␥-aminobutyric acid type A receptor, and the strychnine-sensitive glycine receptor (1,2). Structural analysis by cryo-EM of tubular crystals prepared from the Torpedo marmorata electric organ revealed that five subunits combine in nAChRs, forming a rosette around the central ion channel (3). The second transmembrane (M2) domain of each subunit participates in lining the channel pore, and collectively they present a hydrophobic constriction adjacent to what is traditionally believed to be the rate-limiting portion of the ion conduction pathway that controls single channel conductance (␥) and ionic selectivity (2,4).
The homomeric 5-HT 3A receptor is unique among Cys loop receptors, having a ␥ below the resolution of single channel recording, estimated by variance analysis to be in the femtosiemen range. The incorporation of the 5-HT 3B subunit into human heteromeric 5-HT 3A/B receptors increases ␥ to 16 pS, enabling direct observation of events by single channel recording from outside-out patches (5). The use of chimeric 5-HT 3A -5-HT 3B constructs and site-directed mutagenesis revealed a critical role of three arginine residues within the MA helix of the M3-M4 cytoplasmic loop in determining ␥ (6,7). This has prompted speculation that the MA helix may participate in the control of ␥ in other Cys loop receptors (4,7,8). In support of this interpretation, cryo-EM analysis revealed portals within the Torpedo nAChR formed, in part, by the MA helices of adjacent subunits that may participate in the ion conduction pathway (9).
In this study we investigated the influence of the 5-HT 3A subunit's MA helix Arg-432 (Ϫ4Ј), Arg-436 (0Ј), and Arg-440 (4Ј) residues in the control of ␥. We investigated the effect of introducing arginine into the equivalent locations within the nAChR ␣ 4 and ␤ 2 subunits. Our data confirm the critical role of MA Ϫ4Ј, 0Ј, and 4Ј residues in controlling ␥ of 5-HT 3A receptors and support the hypothesis that the MA helix also forms part of the ion conduction pathway of nAChRs. Our functional data provide support for the existence of cytoplasmic portals depicted in the 4 Å structural model of the nAChR (10).

EXPERIMENTAL PROCEDURES
DNA Constructs and Transient Transfection of Subunit cDNAs-cDNAs encoding rat wild-type (WT) nAChR ␣ 4 and ␤ 2 subunits (Dr. J. M. Boulter, Department of Psychiatry and Biobehavioral Sciences, UCLA), human WT 5-HT 3A subunits, and mutant nAChR and 5-HT 3A subunits were cloned into pGW1 (11). Point mutations were introduced using standard molecular biological techniques (7). All cDNAs were sequenced to confirm fidelity. Transfection of tsA-201, or HEK-293 cells, with subunit cDNAs, at equimolar ratios when appropriate, was performed by either the calcium phosphate precipitation method or electroporation (400 V, infinite resistance, 125 microfarads) using a Bio-Rad gene electropulser II. Transfected cells were routinely cultured at 37°C for 24 -72 h before use, although in the case of nicotinic receptor constructs incubation temperature was reduced to 29°C overnight prior to recording in order to enhance cell surface expression (12).
Electrophysiology-Whole-cell and outside-out patch configurations were used to record macroscopic and single channel currents, respectively, from transfected cells. The recording chamber was perfused (5 ml min Ϫ1 ) with a solution of (in mM) NaCl 140, KCl 2.8, MgCl 2 2.0, CaCl 2 1.0, glucose 10, and HEPES 10; pH 7.3. For macroscopic current recordings, patch-clamp electrodes contained (in mM) KCl 140, MgCl 2 2.0, CaCl 2 0.1, EGTA 1.1, HEPES 10; pH 7.3. Acetylcholine (1 mM) was applied by pressure ejection to whole cells to obtain estimates of peak current, which were then normalized to cell capacitance in order to determine current densities. Slowly rising and decaying whole-cell currents, suitable for fluctuation analysis, were evoked by agonists applied to voltage-clamped cells (Ϫ60 mV) via diffusion pipettes. Single channel conductances were estimated from the macroscopic current (1-1000-Hz bandwidth) response to agonist as described previously (13). Variance analysis was performed using the electrophysiology data recorder (Dr. J. Dempster, University of Strathclyde). For single channel recordings from outside-out patches, or variance analysis recorded from the latter, electrodes were filled with (in mM) potassium gluconate 130, NaCl 5, MgCl 2 2, EGTA 5, HEPES 10; pH 7.2. Single channel currents recorded from outside-out patches were either low pass filtered at 2 kHz in nACh receptor experiments or 1 kHz in 5-HT 3 receptor experiments (digitized at 10 kHz in both cases). Data were acquired as described previously (14). Agonists were applied by pressure ejection to patches voltage-clamped at either Ϫ74 mV in the case of WT and mutant 5-HT 3A constructs or Ϫ94 mV in the case of nAChR constructs (values include correction for liquid junction potential compensation). Sections of digitized data (ϳ10-s duration), in which unitary events predominated, were selected for analysis and were leak current-subtracted (using Clampfit) for the creation of all-points amplitude histograms using Fetchan (pCLAMP 8.0, Axon Instruments, CA). Multiple gaussians were fitted (least squares minimization) to amplitude histograms using the Simplex method within pSTAT (pCLAMP 8.0). The amplitude of the single channel current recorded from each patch was determined from the difference between the mean current amplitudes determined from the gaussians fitted to the closed and open-state currents. Single channel conductances are reported as the chord conductance derived as ␥ ϭ i/(V m Ϫ E rev ), where i is unitary current amplitude; V m is the holding potential, and E rev is the reversal potential of the agonist-evoked macroscopic current response. In several outside-out patch recordings from 5-HT 3 and nACh receptors, two conductances were evident. Our analysis of single channels was restricted to the main state, which in all cases corresponded to the larger amplitude events readily detected in all-points amplitude histograms.
Statistics-Data are presented as mean Ϯ S.E. Data sets were compared using one-way analysis of variance (ANOVA) with the post hoc Tukey's test.
Modeling the 5-HT 3A and nAChR ␣ 4 and ␤ 2 Subunits onto the 4 Å Model of the Torpedo nAChR-Amino acid sequences of 5-HT 3A (WT and 5-HT 3A (R432Q,R436D,R440A) mutant) and WT ␣ 4 and ␤ 2 subunits were aligned against Torpedo ␣, ␤, ␦, and ␥ subunits using Mul-tiAlign (prodes.toulouse.inra.fr/multalin/). Gaps between N-terminal amino acids (predominantly in loop 9) introduced into the template sequence (for modeling against the AChBP) were also introduced in the appropriate locations of each aligned subunit. The 5-HT 3A sequence lacks several N-terminal residues found in nAChR subunits, and these were filled with corresponding residues from the equivalent template subunit. The lengths of the C termini of query sequences differ from template sequences. To allow successful homology modeling, residues were either introduced from the corresponding template subunit or removed from the query sequence.
Homology Modeling-The Torpedo template structure was downloaded from the RCSB Protein Data Bank (code 2BG9) into the Deep-View Swiss-Pdb Viewer (swissmodel.expasy.org/). Aligned sequences of 5-HT 3A (WT and 5-HT 3A (R432Q,R436D,R440A) mutant) and WT nAChR ␣ 4 and ␤ 2 subunits were threaded onto the Torpedo template in DeepView. The nAChR subunits were threaded onto the template with the 2␣ 4 :3␤ 2 stoichiometry in the ␤␤␣␤␣ arrangement. The models were then submitted to SwissModel (swissmodel.expasy.org/) for optimization. Following energy minimization (Gromos96, SwissModel), models were returned to DeepView for viewing and imaging. Excess C-terminal residues were removed from query sequences for imaging. The subunit interface illustrated (Fig. 7) represents the homology model corresponding to the Torpedo 1Ј␣-5Ј␥ interface (10).

RESULTS
Residues in the MA Helix of the 5-HT 3A Subunit Control ␥-Application of 5-HT (10 M) to outside-out membrane patches excised from HEK cells transiently transfected with cDNA encoding the human 5-HT 3A subunit activated currents without resolvable single channel events (Fig. 1A), consistent with our previous observations (5). Variance analysis of whole-cell currents activated by 5-HT (10 M) reveals WT 5-HT 3A receptors to have a single channel conductance (␥) of 966 Ϯ 75 fS (Table 1).
We aligned the human 5-HT 3A subunit amino acid sequence with those of the human 5-HT 3B subunit and various nAChR subunits, including the T. marmorata ␣, ␤, ␥, and ␦ subunits. The 5-HT 3A sequence was then threaded on to the 4 Å structural model of the Torpedo nAChR (10), and the homology model was energy-minimized (see "Experimental Procedures"). The 5-HT 3A subunit contains a series of regularly spaced arginine residues located within the MA helix (Fig. 1B).
Here we denote these residues MA Ϫ4Ј, 0Ј, and 4Ј for ease of comparison with the homologous residues in the structurally related nAChR subunits (Fig. 1B).
By using variance analysis of whole-cell currents, we previously suggested that arginine residues at positions MA Ϫ4Ј (Arg-432) and 4Ј (Arg-440) participate with MA 0Ј (Arg-436) in the control of ␥. In agreement, single channel recordings of 5-HT-activated currents from outside-out patches containing triple mutant homomeric 5-HT 3A (R432Q,R436D,R440A) receptors (in which MA Ϫ4Ј, 0Ј, and 4Ј residues were replaced by corresponding 5-HT 3B residues (hereafter referred to as the 5-HT 3A (QDA) mutant)) revealed robust unitary current events with a ␥ of 36.5 Ϯ 1.0 pS (Fig. 1, C and D; Table 1), i.e. an ϳ40-fold increase in ␥ in comparison to WT 5-HT 3A receptors. The determination of ␥ by variance analysis is indirect and can result in an underestimation of this parameter. Indeed, the value of ␥ reported here from direct observation of single channel events mediated by the 5-HT 3A (QDA) receptor is greater than that which we reported previously, using a different electrode solution (7), but is in agreement with values recently published by others using a similar recording configuration (15). Therefore, we re-investigated the contribution of the MA Ϫ4Ј, 0Ј, and 4Ј 5-HT 3 residues in the control of ␥ individually and in combination by single channel analysis, where resolution permitted. Fig. 2A illustrates the effect of replacing the MA Ϫ4Ј, 0Ј, and 4Ј residues in 5-HT 3A with equivalent residues from the 5-HT 3B subunit. Direct measurements of ␥ using single channel analysis are qualitatively similar to estimates of ␥ that we reported previously using variance analysis. There is a good correlation between ␥ values determined using these two approaches (r 2 ϭ 0.98; Fig. 2B). The systematic underestimation of ␥ using variance analysis could be caused by the presence of subconduc-tance events below the resolution of single channel recording or alternatively the use of different recording solutions.
Application of 5-HT (10 M) to outside-out membrane patches excised from cells expressing the 5-HT 3A (R432Q) receptor produced small inward current responses, but individual channel events were not discernible (data not shown). Variance analysis of 5-HT-induced current responses recorded from outside-out patches suggests a sub-pS unitary conductance ( Fig. 2A; Table 1), similar to that reported previously (7) using analysis of whole-cell variance for WT 5-HT 3A and mutant 5-HT 3A (R432Q) receptors. This demonstrates that estimates of ␥ using variance analysis are not influenced by the whole-cell or outsideout patch recording configuration. A role for the Ϫ4Ј arginine in the control of ␥ can be directly observed by comparing the amplitude of channels mediated by the 5-HT 3A (R436D,R440A), in which the Ϫ4Ј arginine is preserved, to the amplitude of channels mediated by 5-HT 3A (QDA) mutation in which the Ϫ4ЈArg was replaced by Gln. Outside-out membrane patches expressing 5-HT 3A (R436D,R440A) receptors exhibited clearly discernible 5-HT-activated single channel events of a modestly reduced ␥ in comparison to 5-HT 3A (QDA) receptors ( Fig. 2A; Table 1). Collectively, these data reveal Arg-432 to have a limited influence on ␥. In contrast to the R432Q exchange, replacement of Arg-436 of the 5-HT 3A subunit by the equivalent 5-HT 3B subunit residue (Asp) caused a substantial increase in ␥ such that single channel events were evident in outside-out patch recordings (Fig. 2C). From the all points amplitude histograms (Fig. 2D), we determined ␥ for channels mediated by the 5-HT 3A (R436D) mutant receptor to be 9.1 Ϯ 0.6 pS ( Table 1). The importance of Arg-436 is further emphasized by the 5-HT 3A (R432Q,R440A) receptor, which mediates 5-HT-activated single channels of a greatly reduced (p Ͻ 0.001) conductance compared with the 5-HT 3 (QDA) receptor ( Fig. 2; Table 1). The FIGURE 1. MA helix ؊4, 0, and 4 residues of the 5-HT 3 receptor are determinants of single channel conductance. A, current recording from an outside-out patch excised from an HEK cell expressing human WT homomeric 5-HT 3A receptors. The patch was voltage-clamped at Ϫ74 mV. The current was activated by 5-HT (10 M) applied by pressure ejection. No unitary events are evident even in the expanded section of data because the ␥ of WT 5-HT 3A receptors is below the resolution of recording. B, left, the 5-HT 3A receptor homology modeled onto the Torpedo nAChR indicating the position of cytoplasmic MA helices. Right, alignment of amino acids within the MA helices of human and rat 5-HT 3 receptor A and B subunits and human and rat nAChR ␣ 4 , ␣ 7 , and ␤ 2 subunits. Totally conserved residues are identified by shaded boxes, and MA helix Ϫ10Ј, Ϫ4Ј, 0Ј, and 4Ј residues are identified by open boxes. Numbers refer to the location of arginine residues found in the human 5-HT 3A subunit. C, 5-HT (10 M)-activated currents recorded from an outside-out patch containing homomeric triple mutant 5-HT 3A (R432Q,R436D,R440A) receptors in which native MA Ϫ4Ј, 0Ј, and 4Ј arginine residues were replaced by corresponding glutamine, aspartate, and alanine residues located in the MA helices of the 5-HT 3B subunit. D, the all-points amplitude histogram of the current trace reveals that there are at least four channels active in this patch. The amplitudes of the events, as revealed by the sum of five gaussians were Ϫ2.4, Ϫ4.6, Ϫ6.9, and Ϫ9.3 pA at a holding potential of Ϫ74 mV. The unitary amplitude corresponds to a chord conductance of 32.7 pS. No subconductance events were detected.

nACh constructs
The ␥ values were determined from the amplitudes of single channels in recordings from outside-out patches activated by 5-HT (10 M) or ACh (100 nM) unless indicated otherwise. Variance analysis was used to determine ␥ when resolvable channels were not observed in outside-out patch recordings.
Introduction of MA 0Ј Residues of nAChR into the 5-HT 3A Subunit Influences ␥-The ␥ of neuronal nicotinic receptors (e.g. ␣ 7 and ␣ 4 ␤ 2 ) is much greater than that of the 5-HT 3A receptor. Given that the much higher ␥ of the heteromeric 5-HT 3A /5-HT 3B receptor versus the homomeric 5-HT 3A receptor is due, at least in part, to the nature of MA residues and the MA 0Ј residue in particular, we investigated the influence of the residues that occupy homologous MA 0Ј positions in the nAChR ␣ 7 (Glu), ␤ 2 (Gln), and ␣ 4 (Phe) subunits on the ␥ of the 5-HT 3A receptor (see Fig. 1B).
Activation of the 5-HT 3A (R436E) receptor elicited single channel events that were readily resolved in outside-out patch recordings ( Fig.  3A; Table 1). Furthermore, the ␥ of the 5-HT 3A (R432Q,R436E,R440A) receptor was similar to that of the 5-HT 3A (QDA) receptor, demonstrating that negatively charged Glu (the MA 0Ј residue in the ␣ 7 subunit) and Asp (the MA 0Ј residue in the 5-HT 3B subunit) residues are approximately equally effective in facilitating ␥ at this location ( Table 1). The 5-HT 3A (R436Q) receptor also mediated discernible 5-HT-activated single channels ( Fig. 3B; Table 1). The ␥ of the 5-HT 3A (R432Q, R436Q,R440A) receptor was modestly reduced in comparison to the 5-HT 3A (QDA) receptor (Table 1). Hence, for the 5-HT 3A receptor, these two constructs reveal that Gln (the MA 0Ј residue of the nACh ␤ 2 subunit) facilitates ␥, although it is not as effective in this regard as Asp or Glu (Table 1). No single channel events were resolvable in outsideout patch recordings of 5-HT 3A (R436F) receptors, although inward current responses were clearly present (Fig. 3C). Variance analysis of whole-cell currents confirmed that this receptor has a ␥ below the resolution of single channel recording (Table 1). Similarly, the 5-HT 3A (R432Q,R436F,R440A) receptor exhibited a low ␥ estimated by fluctuation analysis of the 5-HT-induced patch current (Table 1). These experiments demonstrate that the introduction into the 5-HT 3A subunit of the MA 0Ј residues found in the nAChR ␣ 7, ␤ 2 , but not ␣ 4 subunits substantially increases ␥.
To test for a possible involvement of poorly resolved unitary events in the effect of the 5-HT 3A mutations on ␥, we attempted to correlate the mean variance of single channel currents with ␥. There was no correlation between the open channel current variance and ␥ values of mutant 5-HT 3A (R436D), 5-HT 3A (R436E), and 5-HT 3A (R436Q) receptors (r 2 ϭ 0.29 from linear regression; data not shown). A lack of an inverse correlation between open channel current variance and ␥ values suggests that changes in ␥ are not an aberration because of incompletely resolved unitary events.
Replacement of MA 4Ј Residues of ␣ 4 ␤ 2 nAChRs by Arginine Reduces Current Density-In contrast to the WT homomeric 5-HT 3A receptor, single channel events mediated by both homomeric and heteromeric FIGURE 3. Replacement of the MA 0 residue of the 5-HT 3A subunit by equivalent residues in nAChR subunits alters single channel conductance. A, 5-HT (10 M)activated currents recorded at Ϫ74 mV from an excised outside-out patch containing homomeric mutant 5-HT 3A (R436E) receptors in which the 5-HT 3A subunit MA 0Ј arginine residue has been replaced by glutamate (the MA 0Ј residue of the nAChR ␣ 7 subunit). Clearly resolvable channels were activated by 5-HT (10 M) applied by pressure ejection. The all-points amplitude histogram was fitted with the sum of two gaussians to determine the single channel amplitude of Ϫ1.0 pA, in this case corre sponding to a ␥ of 13.6 pS. B, single channel currents mediated by homomeric mutant 5-HT 3A (R436Q) receptors in which the MA 0Ј residue has been replaced by glutamine (the MA 0Ј residue of the nAChR ␤ 2 subunit). The all-points amplitude histogram was fitted with the sum of two gaussians yielding a single channel amplitude of Ϫ0.57 pA, corresponding to a ␥ of 7.7 pS. C, current recording from an outside-out patch excised from a cell expressing homomeric mutant 5-HT 3A (R436F) receptors in which MA 0Ј arginine residue has been replaced by phenylalanine (the MA 0Ј residue of the nAChR ␣ 4 subunit). The patch was voltage-clamped at Ϫ74 mV. No unitary events are evident even in the expanded section of data because the ␥ of 5-HT 3A (R436F) receptor is below the resolution of recording.
nAChRs have ␥ values in the picosiemen range (12, 16 -19). Cell-attached patch recordings from Xenopus oocytes reveal that the recombinant ␣ 7 nAChR has a ␥ of ϳ80 pS (18). Given the influence of the nicotinic MA stretch Ϫ4Ј, 0Ј, and 4Ј residues on the ␥ of the 5-HT 3A receptor, we examined whether the identities of the equivalent residues in nAChR subunits contribute to the ␥ of the nACh receptors. Functional homomeric ␣ 7 receptors express poorly in HEK cells; therefore, we chose to study recombinant ␣ 4 ␤ 2 receptors. This receptor provides the additional advantage that MA residues can be introduced into all, or just some, of the subunits within heteromeric nAChRs by expressing mutant ␣ 4 and ␤ 2 subunits together or in combination with WT subunits.
We transfected HEK cells with triple mutant ␣ 4 (E584R, F588R,E592R) and ␤ 2 (E439R,Q443R,E447R) constructs (i.e. the Ϫ4Ј, 0Ј, and 4Ј residues of both subunits mutated to Arg, the homologous residue occupying these positions in the 5-HT 3A subunit). ACh (100 nM) failed to activate single channels in outside-out patches excised from transfected GFP-positive HEK cells (six patches tested). By contrast ACh (100 nM) readily activated channels recorded from patches expressing WT ␣ 4 ␤ 2 receptors (see Fig. 5 and below). Furthermore, the application of ACh (1 mM) to GFP-positive cells did not induce inward currents in any of the 16 cells tested, although 67% (14 cells of 21 cells tested) of GFP-positive cells transfected with WT ␣ 4 ␤ 2 cDNAs gave robust whole-cell inward current responses. By normalizing the peak amplitudes of ACh-activated current to cell capacitance, we calculated the mean current density mediated by WT ␣ 4 ␤ 2 receptors (Fig. 4). ACh (1 mM) was also relatively ineffective on cells expressing WT ␣ 4 and triple mutant ␤ 2 or vice versa (Fig. 4). Therefore, we investigated the influence on the whole-cell current response to ACh (1) of MA Ϫ4Ј, 0Ј, and 4Ј mutations within one, or both, nAChR subunits. Robust responses were obtained for the Ϫ4Ј and 0Ј mutants, expressed either with the corresponding mutant or a WT subunit. Strikingly, no wholecell response (14 cells tested) was obtained for the ␣ 4 (E592R)␤ 2 (E447R) receptor (Fig. 4). Indeed, even when the 4Ј mutation was only carried by one subunit, either the ␣ 4 or the ␤ 2 , the whole-cell current was dramatically reduced (Fig. 4). Unsurprisingly, single channels were not detected from outside-out patches excised from such cells. Clearly, the exchange of the negatively charged glutamate residue by the positively charged arginine residue at the 4Ј position of either the ␣ 4 or the ␤ 2 subunit has a dramatic effect on the functional expression of the receptor. We are currently investigating the nature of this deficit, and we note that in both the ␣ 4 and ␤ 2 subunits the mutation disrupts a putative site for phosphorylation by casein kinase II (i.e. SXX(D/E)).
Replacement of MA Ϫ4Ј or 0Ј Residues of ␣ 4 ␤ 2 nAChRs by Arginine Reduces ␥-In contrast to the MA 4Ј mutants, resolvable single channels could readily be detected for receptors carrying the Ϫ4Ј or 0Ј mutations. We first examined the role of the MA 0Ј residue in nAChR. In comparison to WT ␣ 4 ␤ 2 nAChRs, the ␥ of receptors assembled from ␣ 4 (F588R) and ␤ 2 (Q443R) subunits was approximately halved, when assessed both by fluctuation analysis of whole-cell currents (Fig. 5, A and  B) and by direct observation of single channel events (Table 1; Fig. 5, D and G). However, when the ␣ 4 (F588R) mutant was co-assembled with WT ␤ 2 subunits, there was only a small reduction of ␥ (Table 1) determined from outside-out patch recordings (Fig. 5, E and G). Similarly, when the ␤ 2 (Q443R) mutant was expressed with the WT ␣ 4 subunit, ␥ was only modestly reduced (Table 1; Fig. 5, F and G).
Gating kinetics can influence estimates of ␥. For example, rapid open channel block can cause channel "flickering" resulting in failure to resolve full openings (19). To investigate the possibility that such a mechanism could account for the apparent reduction in ␥ caused by mutation of MA 0Ј residues in ␣ 4 and ␤ 2 subunits, we attempted to correlate the mean open channel current variance (derived from the gaussian fits to all-points amplitude histograms) to the mean conductances of WT and mutant ␣ 4 ␤ 2 receptors (Fig. 5H). If channel flickering caused the reduction in ␥, an inverse correlation between ␥ and variance is anticipated. However, a linear regression fitted to the individual data points of ␥ versus current variance revealed a coefficient of determination (r 2 ) of 0.10 (fit not shown). A lack of an inverse correlation between these two parameters suggests that changes in ␥ cannot be explained by a failure to resolve unitary events (19). Taken together, these data demonstrate that the introduction of arginine residues into the MA 0Ј positions of ␣ 4 and ␤ 2 subunits causes a significant reduction in ␥.
We next explored the contribution of the MA Ϫ4ЈGlu residue to the ␥ of ␣ 4 ␤ 2 nAChRs. The Ϫ4Ј location had a relatively modest effect on the single channel conductance of 5-HT 3A receptors when compared with that of the 0Ј residue ( Fig. 2 and Table 1). However, introduction of an arginine at this location in both the ␣ 4 and ␤ 2 subunits causes a charge reversal and therefore may be anticipated to have a greater effect on conductance in this nAChR. Indeed, the ␥ of receptors assembled from ␣ 4 (E584R) and ␤ 2 (E439R) subunits was substantially reduced (␥ ϭ  11.4 Ϯ 0.5 pS; Fig. 6, A and D) when compared with that of WT ␣ 4 ␤2 nAChRs (Fig. 5C). In common with the nAChR 0Ј residue, when the Ϫ4Ј mutation was introduced only into one of the subunits, either the ␣ 4 (Fig. 6B) or the ␤ 2 (Fig. 6C), there was a less dramatic but nevertheless significant reduction of ␥. In fact, the ␣ 4 ␤ 2 (E439R) receptor had the lowest ␥ of any of the nAChRs tested carrying a single point mutation on one of its subunits (␥ ϭ 16.3 Ϯ 0.9 pS; Table 1).
Taken together these results demonstrate that in common with the 5-HT 3A receptor, the nature of the MA 0Ј residue greatly influences the ␥ of the ␣ 4 ␤ 2 nAChR. However, for the 5-HT 3A receptor, the impact of the Ϫ4Ј residue on ␥ was modest in comparison to that of the 0Ј residue, and for the ␣ 4 ␤ 2 nAChR the influence of the Ϫ4Ј residue is greater than that of the 0Ј residue. Furthermore, for the ␣ 4 ␤ 2 nAChR the Ϫ4Ј or the 0Ј arginine residue must be introduced on both the ␣ 4 and the ␤ 2 subunit to achieve the maximum reduction of ␥.

DISCUSSION
This study demonstrates that elements of the cytoplasmic loop within the region referred to as either the MA helix (10) or helical amphipathic stretch (6) influence the single channel conductance (␥) of both the 5-HT 3 and nACh receptors; a property traditionally assigned to the pore-lining M2 domains (2,4). A mechanistic explanation is derived from ultrastructural studies of the Torpedo nAChR showing rods of density, corresponding to MA helices, projecting into the cytoplasm to form an "inverted pentagonal cone" that constitutes the inner vestibule  . Replacement of MA ؊4 residues in the nACh ␣ 4 ␤ 2 receptor by arginine reduces single channel conductance. A-C, single channel currents activated by acetylcholine (100 nM) applied to outside-out patches (clamped at Ϫ94 mV) containing mutant ␣ 4 ␤ 2 nAChRs. All-points amplitude histograms were fitted with the sum of two gaussians. A-C, unitary events were mediated by double mutant ␣ 4 (E584R)␤ 2 (E439R), single mutant ␣ 4 (E584R)␤ 2 , and single mutant ␣ 4 ␤ 2 (E439R) receptors, respectively. D, a bar graph of mean chord conductance (ϮS.E.) for WT, single, and double mutant nAChRs. Mutant ␣ 4 (E584R)␤ 2 , ␣ 4 ␤ 2 (E439R), and ␣ 4 (E584R)␤ 2 (E439R) receptors exhibited chord conductances that were significantly below those of WT receptors, determined by ANOVA with post hoc Tukey's test. All values were significantly different from one another at p Ͻ 0.001. of the ion channel (9,10). Unlike conventional depictions of the vestibule, the images of the nAChR suggest that permeating ions must pass through "portals" formed between adjacent subunits. The dimensions of the portals are estimated to be comparable with those of the permeant cations and would thus be anticipated to influence the rate of ion flux (9,10). Thus, we hypothesize that amino acids within the MA helices of the homomeric 5-HT 3A receptor line five portals (Fig. 7A). In each portal three repetitive arginine residues (MA Ϫ4Ј, 0Ј, and 4Ј) form a ratelimiting barrier to ion conduction (7). These residues are responsible for the femtosiemen conductance that is a unique hallmark of homomeric 5-HT 3A receptors (5). In contrast, the 5-HT 3B subunit cannot form homomeric receptors but increases the ␥ of heteromeric 5-HT 3A/B receptors to ϳ16 pS.
We produced homology models of the 5-HT 3A and ␣ 4 ␤ 2 nACh receptors by threading their amino acid sequences onto the 4 Å resolution structure of the ␣␤␥␦ Torpedo nAChR. The energy-minimized homomeric 5-HT 3A receptor model is asymmetric by virtue of the fact that the original ␣-carbon coordinates were derived from a heteromeric nAChR. Nevertheless, this qualitative approach suggests that MA Ϫ4Ј, 0Ј, and 4Ј arginine residues are located at the cytoplasmic mouths of the 5-HT 3A portals (Fig. 7A). Introduction of 5-HT 3B subunit MA Ϫ4ЈGln, 0ЈAsp, and 4ЈAla residues into the model predicts that these more compact residues lessen a steric impediment to ion flux providing a potential explanation for the increased ␥ seen in recordings from the 5-HT 3A (R432Q,R436D,R440A) receptors (Fig. 7B). The triple mutant 5-HT 3A (QDA) receptor has a Ͼ36-fold higher ␥ compared with that of the WT 5-HT 3A receptor. It is likely that the increased density of acidic residues in 5-HT 3A (QDA) also participates in their large ␥.
It is evident that there are several acidic residues lining the putative ␣ 4 ␤ 2 nAChR portals (Fig. 7C). Instead of the basic residues at MA Ϫ4Ј, 0Ј, and 4Ј locations within the 5-HT 3A subunit, ␣ 4 and ␤ 2 subunits have acidic (glutamate) residues at both the MA Ϫ4Ј and 4Ј locations. The portals of the nAChRs also appear less cluttered by voluminous residues lining their cytoplasmic mouths compared with those of the 5-HT 3A receptor (Fig. 7). These observations are in keeping with the substantially larger ␥ observed for ␣ 4 ␤ 2 nAChRs compared with 5-HT 3A receptors.
The MA 0Ј residue has the largest impact on ␥ of any single 5-HT 3A receptor residue tested. Replacement of the MA 0Ј arginine residue of the 5-HT 3A subunit by the equivalent residue in the 5-HT 3B subunit (aspartate) is sufficient to increase ␥ from ϳ900 fS (below the resolution of direct observation) to ϳ10 pS enabling direct observation of unitary events in outside-out patch recordings. Furthermore, introduction of MA 0Ј residues of either ␣ 7 or ␤ 2 subunits (glutamate and glutamine, respectively) into the 5-HT 3A receptor also substantially increased ␥. Substitution of the MA 0Ј arginine by phenylalanine, the MA 0Ј residue of the ␣4 subunit, by contrast, reduced the ␥ of 5-HT 3 (R436F) receptors. Like arginine, phenylalanine is a voluminous residue, and its ability to maintain ␥ within the femtosiemen range supports the hypothesis that the volume of the MA 0Ј residue is one determinant of ␥.
The critical role of the 5-HT 3A MA 0Ј residue in controlling ␥ is emphasized by replacing the 0Ј aspartate with arginine (5-HT 3A (QRA)) in the large ␥ 5-HT 3A (QDA) construct. This leads to an 83% reduction in ␥ ( Table 1). The MA Ϫ4Ј and 4Ј residues neighboring 0Ј in the MA helix (Fig. 7A) also make a significant contribution to ␥. This can be most directly observed when either the Ϫ4Ј or the 4Ј arginines are returned to the 5-HT 3A (QDA) construct. Under these conditions the single channel conductances of 5-HT 3A (RDA) and 5-HT 3A (QDR) are reduced by 32 and 51%, respectively, compared with 5-HT 3A (QDA) ( Table 1).
Introduction of arginine into either the MA Ϫ4Ј or 0Ј locations of the ␣ 4 ␤ 2 nAChR also caused significant reductions in ␥. This effect was most obvious when arginine was introduced into either the Ϫ4Ј or 0Ј locations of both the ␣ 4 and ␤ 2 subunits (i.e. ␣ 4 (E584R), ␤ 2 (E439R), and ␣ 4 (F588R) ␤ 2 (Q443R) receptors, respectively), when the ␥ values of the mutant receptors were reduced by 63 and 50%, respectively, from that of the WT receptor. Under these conditions all five portals of each of the mutant receptors contain basic residues at critical positions in the cytoplasmic conduction pathway.
Although the introduction of arginines into the MA Ϫ4Ј and 0Ј locations of the ␣ 4 and ␤ 2 subunits substantially reduced ␥, mutant nAChRs had conductances considerably larger than that of the WT 5-HT 3A receptor. It is likely that the higher density of basic residues seen in the 5-HT 3A receptor is required to achieve the femtosiemen level ␥. To test this hypothesis, we constructed triple mutant ␣ 4 (E584R,F588R,E592R) and ␤ 2 (E439R,Q443R,E447R) constructs. Whether expressed in combination with WT ␣ 4 and ␤ 2 or their respective triple mutant partners, these constructs failed to produce sufficient functional receptor expression to enable quantification of ␥. Furthermore, introduction of arginine into the MA 4Ј positions of either the ␣ 4 or ␤ 2 subunits caused a near abolition of functional expression. We note that this mutation disrupts a putative casein kinase II phosphorylation site; however further experiments will be required to determine the cause of the reduced current density.
A more extensive investigation of the relationship between the physicochemical properties of MA Ϫ4Ј, 0Ј, and 4Ј residues and ␥ of 5-HT 3 and nACh receptors is warranted to elucidate their precise roles in ion conduction. Nevertheless, this study demonstrates that the substitution of the MA Ϫ4Ј and 0Ј residues in both ␣ 4 and ␤ 2 subunits leads to a substantial reduction of ␥, suggesting that cytoplasmic residues influence ion conduction through heteromeric ␣ 4 ␤ 2 nAChRs in a manner consistent to that observed in homomeric 5-HT 3A receptors. These data support a role for cytoplasmic portals in the ion conduction pathways of Cys loop receptors.