Introduction

Top Summary Introduction Materials & Methods Results Discussion References

The voltage-gated sodium channel is responsible for the initiation and propagation of action potentials in neurons (Hille, 1992). Sodium channel function is regulated by voltage-dependent transitions between three sets of functionally distinct conformational states. At resting membrane potentials, most sodium channels are in closed resting states. In response to membrane depolarization, channels first open within a few hundred microseconds, resulting in inward sodium flux, and then convert within a few milliseconds to nonconducting inactivated states. Inactivated channels will not open until they are converted back to resting states by repolarizing the membrane.

In the mammalian brain, the sodium channel consists of a central  subunit (260 kDa) and two smaller auxiliary subunits, designated 1 (36 kDa) and 2 (33 kDa) (Hartshorne and Catterall, 1984). The  subunit is the main structural component of the sodium channel and forms the ion-conducting pore, the activation and inactivation gates, and the binding sites for various neurotoxins and therapeutic drugs (Catterall, 1992). The  subunit consists of four quasi-homologous domains (I-IV), each composed of six hydrophobic segments (S1-S6) that are believed to traverse the membrane as -helices. The four domains are thought to form a square array in the membrane, with the ion-conducting pore located in the center.

Ionic currents through sodium channels are inhibited by a number of different types of therapeutically important drugs including local anesthetics, such as lidocaine, and anticonvulsants, such as phenytoin (Catterall, 1987). Local anesthetics and related drugs are poor blockers of sodium channels at hyperpolarized membrane potentials, but inhibition is greatly enhanced by prolonged membrane depolarization or high frequency channel activity. According to the modulated receptor hypothesis (Hille, 1977; Hondeghem and Katzung, 1977), inhibition of sodium channels by local anesthetics is voltage- and frequency-dependent because the affinity of the drug receptor on the channel protein depends on whether the channel is resting, open, or inactivated. Resting states, which predominate at hyperpolarized membrane potentials, are thought to bind local anesthetics with low affinity, whereas open and inactivated states, which are more prevalent at depolarized membrane potentials and during trains of channel activity, are proposed to bind local anesthetics with higher affinity.

The voltage- and frequency-dependent action of local anesthetics has been extensively studied by electrophysiological recording; however, the precise location of the drug receptor and the molecular basis of drug action are less well understood. Biophysical studies suggest that the local anesthetic receptor is located within an inner vestibule of the ion-conducting pore (Strichartz, 1973; Cahalan and Almers, 1979; Gingrich et al., 1993; Zamponi and French, 1994). Site-directed mutagenesis studies (Lopez et al., 1994; Taglialatela et al., 1994; Liu et al., 1997) and recent crystallographic data from a structurally related bacterial potassium channel (Doyle et al., 1998) indicate that S6 transmembrane segments line the inner vestibule of voltage-gated ion channel pores. Thus, it is a plausible hypothesis that the local anesthetic receptor is formed by pore-lining residues within the S6 segments.

A previous study used alanine-scanning mutagenesis to demonstrate that two residues in the S6 segment in domain IV of the sodium channel  subunit are critical determinants of local anesthetic action (Ragsdale et al., 1994). One residue is a phenylalanine located approximately half-way through the transmembrane segment. The other residue is a tyrosine located near the cytoplasmic end of the segment. Mutation of either of these residues to alanine reduced the blocking action of local anesthetics by 1 to 2 orders of magnitude. These two residues were proposed to face toward the channel pore, where they form part of the local anesthetic receptor. In the present study, we have more rigorously examined how these residues influence local anesthetic binding by using site-directed mutagenesis to make a series of amino acid substitutions that vary the size, hydrophobicity, and aromaticity at these two sites. The results of this study support the hypothesis that the critical phenylalanine residue at position 1710 contributes to local anesthetic binding through interactions that depend on hydrophobicity for resting channels or aromaticity for open and inactivated channels.

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Site-Directed Mutagenesis. Mutations were introduced into the type III subtype of the sodium channel  subunit (III) (Kayano et al., 1988) using either the Altered States mutagenesis kit (Promega, Madison WI) or polymerase chain reaction (PCR)-based mutagenesis. For mutations using the Altered States kit, a 4.4-kb XbaI fragment of III cDNA containing the region of interest was subcloned into the pAlter vector, and mutations were introduced into this construct following the procedure described in the kit. A 3.3-kb fragment was excised from mutant constructs using ClaI and SpeI and subcloned back into the full length III cDNA in the vector pSP64t; the mutated region was then sequenced to confirm the presence of the mutation. PCR-based mutagenesis was performed by the megaprimer method (Barek, 1993). Using III-pSP64t as a template, a 2.1-kb fragment containing 1.1 kb of III coding sequence was amplified by two rounds of PCR with Pfu Turbo polymerase (Stratagene, La Jolla, CA). The PCR product was cut with BstEII, subcloned into III-pSP64t, and sequenced to confirm the presence of the mutation.

RNA Preparation. RNA was transcribed from wild type and mutant III-pSP64t constructs using the mMessage mMachine RNA synthesis kit (Ambion, Austin TX). RNA was resuspended in 0.1 mM EDTA, 5 mM HEPES (pH 7.5); samples of each preparation were analyzed by agarose gel electrophoresis. Total RNA yields for each preparation were estimated by comparing the intensity of ethidium bromide-stained bands on agarose gels with the intensity of bands corresponding to RNA standards of known concentration.

Isolation and Injection of Xenopus Oocytes. Pieces of ovary were surgically removed from female Xenopus frogs (Xenopus I, Ann Arbor, MI or Boreal, St. Catherine, Ontario), which were anesthetized with 3-aminobenzoic acid ethyl ester. Oocytes were separated and defolliculated by shaking in 1.5 mg/ml collagenase in OR2 (82.5 mM NaCl, 2 mM KCl, 1 mM MgCl₂, and 5 mM HEPES pH 7.5). Healthy stage V-VI oocytes were selected and incubated overnight at 18°C in Barth's medium (88 mM NaCl, 1 mM KCl, 0.82 mM MgSO₄, 0.33 mM Ca(NO₃)₂, 0.41 mM CaCl₂, 2.4 mM NaHCO₃, and 10 mM HEPES pH 7.4) supplemented with 50 µg/ml gentamicin. On the day after isolation, oocytes were microinjected with 50 nl of wild type or mutant III RNA. The concentration of each RNA was adjusted to give whole cell sodium currents in Xenopus oocytes of <10 µA. Typical sodium currents were between 1 and 5 µA. Coexpression in Xenopus oocytes of III with auxiliary  subunits results in whole cell sodium currents with highly complex kinetic properties, reflecting multiple gating modes with rates of inactivation and recovery from inactivation that vary over more than three orders of magnitude (Patton et al., 1994; Meadows et al., 1997). In contrast, expression of III alone results in a functionally more uniform population of channels with intermediate kinetic properties; this allows a more straightforward interpretation of the effects of site-directed mutations. Therefore, in all experiments in this study, wild type and mutant III subunits were expressed alone, without auxiliary  subunits.

Electrophysiological Recording. After injection, oocytes were incubated for 2 to 3 days at 18°C. They were then examined by two-electrode voltage clamp recording using a Turbo TEC 10C amplifier (Adams & List, Westbury, NY) and pCLAMP software (Axon Instruments, Foster City, CA). Electrodes were filled with 3 M KCl, and had resistances of <0.5 M. Data were filtered at 2.5 kHz and sampled at 20 kHz. Capacity transients were partially compensated using the internal clamp circuitry. Remaining transients, as well as leak currents, were subtracted using the P/4 procedure (Armstrong and Bezanilla, 1974). Oocytes were continuously superfused with frog Ringer's solution (115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl₂, and 10 mM HEPES pH 7.2). One hundred millimolar stock solutions of tetracaine were prepared in dimethylsulfoxide. Stock solutions were diluted to the appropriate concentration with Ringer's solution, and applied by superfusion. Data were analyzed with pCLAMP. Graphing and curve fitting were performed with SigmaPlot (Jandel Scientific, San Rafael, CA).

	Results

Top Summary Introduction Materials & Methods Results Discussion References

Inhibition of Type III Channels by Tetracaine Is Voltage- and Frequency-Dependent. Neurons in the mammalian nervous system express multiple isoforms of the sodium channel  subunit. These isoforms differ in their developmental, regional, and subcellular expression patterns, as well as in their functional properties (Reviewed in Ragsdale and Avoli, 1998). In this study, we examined the type III  subtype (Kayano et al., 1988), which is expressed primarily during embryogenesis (Beckh et al., 1989; Brysch et al., 1991; Black et al., 1994). III forms functional sodium channels in Xenopus oocytes; however, these channels inactivate more than an order of magnitude more slowly than other sodium channel subtypes (Joho et al., 1990; Patton et al., 1994; Meadows et al., 1997). It is not understood why type III channels inactivate so slowly, nor is it known whether this channel inactivates by the same molecular mechanism as other more rapidly inactivating sodium channel isoforms. Because voltage- and frequency-dependent inhibition of sodium channels by local anesthetics depends strongly on channel inactivation, we began our study by examining how these drugs affect type III channels.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   Voltage- and frequency-dependent block of sodium channels by tetracaine. All four panels show data from the same oocyte. In this and subsequent figures, the concentration of tetracaine was 100 µM unless otherwise indicated, and all currents were evoked by test pulses to +10 mV. A, inhibition of whole cell sodium currents by tetracaine. Currents were elicited from holding potentials of either 100 mV (top two traces) or 40 mV (bottom two traces). In each set of traces, the larger current was in control conditions and the smaller current was after application of tetracaine. Scale bars are 1 µA and 10 ms. B, inactivation curves in control () and in the presence of tetracaine (). Current amplitudes were normalized with respect to the largest control currents and plotted as a function of holding potential. The smooth lines are best fits of the equation a/(1+exp(Vhold  V1/2)/k), where Vhold is the holding voltage, V1/2 is the midpoint of the curve, k is a slope factor, and a is a scaling factor. The dashed line shows the fit of the drug data rescaled to have the same maximum value as the control curve. C, frequency-dependent block by tetracaine. The traces shown were evoked by the 1st, 2nd, 5th, 10th, and 20th pulses in a 1-Hz train, applied from a holding potential of 90 mV. D, amplitudes of currents elicited by 1-Hz pulse trains given in control () and in the presence of tetracaine (). Each set of data points was normalized with respect to the amplitude of the first pulse in the train and plotted as a function of pulse number.

Mutations of Residues F1710 and Y1717 Alter Tetracaine Inhibition of Type III Channels. It was previously shown that mutation of either a phenylalanine residue or a tyrosine residue in the IVS6 transmembrane segment to alanine reduced sodium channel sensitivity to local anesthetics by 1 to 2 orders of magnitude (Ragsdale et al., 1994, 1996). These studies were performed on the type IIA  subunit, which is a major isoform in the adult brain. The critical phenylalanine and tyrosine residues in IVS6 are conserved in other  subtypes; however, it is not known whether the importance of these residues in local anesthetic action is also conserved. To investigate this question, we substituted alanine for the homologous phenylalanine (mutation F1710A) or tyrosine (mutation Y1717A) residues of III, expressed the mutant channels in oocytes, and examined their sensitivity to tetracaine.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Mutants F1710A and Y1717A attenuate block of type III channels by tetracaine. A, B, typical currents for mutants F1710A (A) and Y1717A (B) in control and after application of 100 µM tetracaine. Currents were evoked from the holding potentials shown. Scale bars are 1 µA and 10 ms. C, D, inactivation curves for F1710A (C) and Y1717 (D) in control () and after application of tetracaine (). The data points and smooth lines were plotted in the same manner as in Fig. 1. E, F, amplitudes of currents for F1710A (E) an Y1717A (F) evoked by 1-Hz pulses given in control () and with tetracaine (). Data were plotted in the same manner as in Fig. 1.

Effects of Different Amino Acid Substitutions at Position 1710. To investigate in more detail how residues at positions 1710 and 1717 determine sodium channel sensitivity to local anesthetics, we introduced a series of amino acid substitutions that systematically altered the size, hydrophobicity, and aromaticity at these two positions. The effects of substitutions at position 1710 on resting block, V1/2, and frequency-dependent block are summarized in Fig. 3. The amino acids in each panel are arranged in order from the most hydrophilic (serine) to the most hydrophobic (tryptophan) (Hopp and Woods, 1981). Two distinct patterns emerge from this analysis. First, resting block increased steadily with increasing hydrophobicity (Fig. 3A) as if the binding of tetracaine to resting sodium channels involves hydrophobic interactions with the residue at this position. In contrast, V1/2 and frequency-dependent block were large for tyrosine and tryptophan as well as for the wild type phenylalanine, but greatly attenuated for serine, cysteine, alanine, and isoleucine (Fig. 3, B and C). This observation suggests that high-affinity drug binding to open and inactivated channels requires the presence of an aromatic residue at position 1710. Hydrophobicity did not appear to be important, as substitution with the hydrophobic nonaromatic residue isoleucine at this position resulted in virtually no voltage-dependent or frequency-dependent block by tetracaine.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Mutations at position 1710 alter resting, voltage-dependent, and frequency-dependent block. In this and subsequent figures, columns show means ± S.E.M.. The number of experiments for each mean are shown above the columns. A, resting block for wild type and 1710 mutants. Columns show fraction of current inhibition at a holding voltage of 100 mV. The residues at position 1710 are arranged in order from the most hydrophilic (serine) to the most hydrophobic (tryptophan). The wild type residue, phenylalanine, is shown as a white column. B, V1/2 values for wild type and 1710 mutants determined from the difference of V1/2 values obtained from fits of tetracaine and control inactivation curves. C, fraction of frequency-dependent current inhibition at the end of a train of 20 pulses applied at 1 Hz. Inhibition was determined from the ratio of the currents evoked by the last pulse/first pulse in the train. For wild type and mutant channels, current attenuation in control was  5%.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   KI and KR for wild type and mutants F1710A and F1710I. A, typical dose-effect relationships for a wild type experiment at holding voltages of 100 mV () or 40 mV (). The smooth lines are according to 1/(1+ [tetracaine]/B1/2), where B1/2 is the concentration giving 50% block. B, KI (open symbols) and KR (filled symbols) for wild type, F1710I, and F1710A, determined as described in the text. Within each row, each pair of symbols (e.g.,, ) show KI and KR values from a single experiment. The drop lines correspond to mean values.

(1)

Effects of Different Amino Acid Substitutions at Position 1717. Unlike substitutions at position 1710 described in the previous section, the effects of mutations at position 1717 did not follow a pattern that could be well correlated with parameters such as hydrophobicity or aromaticity. Mutant Y1717A greatly attenuated V1/2 and frequency-dependent block (Fig. 5B and C) without altering resting block (Fig. 5A). However, other substitutions at position 1717 had a completely different effect: enhancement of resting block (Fig. 5A) accompanied by a modest to large decrease in V1/2 and frequency-dependent block (Fig. 5, B and C). To more directly assess the effects of mutations at position 1717 on resting state and inactivated state affinity, we determined K_R and K_I values for the representative mutants Y1717C, Y1717F, and Y1717A (Fig. 6). K_R was 34 µM for Y1717C and 27 µM for Y1717F; these values represent increases in sensitivity of 4- to 5-fold compared to wt. In contrast, K_I values were 7.0 µM for Y1717C and 3.1 µM for Y1717F, similar to wt. Apparently, these mutations selectively increased resting state affinity without altering inactivated state affinity. Conversely, K_R for mutant Y1717A was indistinguishable from wild type but K_I was 17 times larger; this result is virtually the mirror image of the effects of other mutations at this site.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Mutations at position 1717 alter resting block (A), V1/2 (B), and frequency-dependent block (C). Data presentation is the same as in Fig. 3.

View larger version (12K): [in this window] [in a new window]   Fig. 6.   KI and KR values for Y1717C, Y1717F, and Y1717A. Data presentation is the same as in Fig. 4B.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

Hydrophobicity and Aromaticity Are Critical Determinants at Position 1710. Most therapeutically useful local anesthetic molecules have a characteristic molecular structure with a tertiary amine at one end and an aromatic ring at the other, separated by a linking alkyl chain. Structure-activity studies suggest that both the tertiary amine and the aromatic moieties are involved in interactions with sodium channels (Courtney, 1980; Bokesch et al., 1986; Sheldon et al., 1991). The correlations observed in this study between local anesthetic block and hydrophobicity or aromaticity at position 1710 suggest several possible mechanisms by which molecular interactions between sodium channels and these drug moieties could occur. In the case of block of resting channels, hydrophobic interactions with F1710 may be involved. Structure-activity studies have shown that increasing the hydrophobicity of the local anesthetic molecule increases resting block (Bokesch et al., 1986). Our results provide the first complementary molecular evidence for an important hydrophobic determinant on the channel protein. F1710 may contribute to a hydrophobic pocket that stabilizes drug binding to resting sodium channels.

Crystallographic Data Suggest Residue 1710 Could Lie Within the Pore. Biophysical evidence suggests that the binding site for local anesthetics is within an inner vestibule of the channel pore (Strichartz, 1973; Cahalan and Almers, 1979; Gingrich et al., 1993; Zamponi and French, 1994). An important recent breakthrough in understanding pore structure has been the X-ray crystallographic analysis of a potassium channel from Streptomyces lividans (Doyle et al., 1998). This bacterial ion channel contains transmembrane segments that correspond to the S5 and S6 segments of voltage-gated ion channels. These transmembrane segments are separated by a pore loop that dips into the membrane from the extracellular side to form the selectivity filter of the pore. The remainder of the pore is lined by the transmembrane segment corresponding to S6. One unexpected finding of this analysis is a cavity located approximately half-way through the pore. This cavity, which is proposed to be important for ion permeation, contains the intracellular binding site for TEA. (The internal site is distinct from the external TEA site described above.) It is plausible that local anesthetics also bind within this internal cavity. Residue F1710, which is located approximately half-way through the S6 segment, is well situated to contribute to the cavity, although at present it is not known whether this residue is oriented toward the pore, where it could interact with local anesthetics, or away from the pore where it could interact with other residues on the  subunit.

The Residue at 1710 Could Affect Drug Action Without Contributing Directly to the Receptor. Although the results of this study are consistent with the hypothesis that the residue at 1710 contributes directly to the local anesthetic receptor, it is also possible that mutations at this site affect drug binding to the receptor through an indirect mechanism. For example, Hille (1977) proposed that when sodium channels are closed, local anesthetics reach their receptor through a hydrophobic access pathway. Based on dissociation rates of local anesthetics of different size, Courtney (1984) suggested that this hydrophobic access route was a narrow passage, approximately 3.6 Å in radius, through the channel protein. The strong correlation between resting block and hydrophobicity at residue 1710 could be related to the contribution of the residue to this hydrophobic access pathway. Alternatively, mutations at position 1710 could determine sodium channel affinity for local anesthetics through indirect allosteric effects on the conformation of the receptor. For example, the importance of aromaticity in high affinity drug block of open and inactivated sodium channels could reflect the dependence of the high-affinity receptor conformation on an interaction between the aromatic residue at position 1710 and another residue on the channel protein. Clearly, further work will be required to unequivocally determine whether the residue at position 1710 contributes directly to the local anesthetic receptor site.

Mutations at 1717 Increase Resting State Affinity or Decrease Open and Inactivated State Affinity. Although all mutations at position 1717 had strong effects on local anesthetic potency, they did not show a clear pattern that would reveal relationships between side chain structure and drug action. As has been shown previously (Ragsdale et al., 1994, 1996), substitution of alanine at this position greatly reduced open and inactivated state affinity without altering resting state affinity. In contrast, other mutations at this position, even the conservative mutation Y1717F, increased resting affinity without affecting open and inactivated affinity. The diametrically opposite effects of substitution of alanine on the one hand, or cysteine, serine, isoleucine, or phenylalanine on the other hand, are difficult to explain in terms of direct interaction between the local anesthetic molecule and the residue at position 1717, and suggest that the residue at this position does not contribute directly to the local anesthetic receptor. Although mutations at this position strongly alter drug block, the molecular mechanism for these effects is at present unknown.

Other Determinants of Local Anesthetic Action. The four domains of the sodium channel  subunit are thought to be arranged in a square array with the channel pore at the center. Thus, it is likely that the other S6 segments also contribute to the lining of the pore and may also contribute to the local anesthetic receptor. In addition, other regions of the channel could be involved. For example, a mutation that disables the channel inactivation gate has been shown to attenuate frequency-dependent block by lidocaine (Bennett et al., 1995), suggesting that the inactivation gate may contribute to stabilizing drug binding. Residues in the pore loops that form the pore selectivity filter are also likely candidates for determinants of drug action. Future work will address these various possibilities.