The Extracellular Domain of the β1 Subunit Is Both Necessary and Sufficient for β1-like Modulation of Sodium Channel Gating*

The type IIA voltage-gated sodium Na+ channel from rat brain is composed of a large, pore-forming α subunit and the auxiliary subunits β1 and β2. When expressed in Xenopus oocytes, the β1 subunit modulates the gating properties of the type IIA α subunit, resulting in acceleration of both inactivation and recovery from inactivation and in a negative shift in the voltage dependence of fast inactivation. The β1 subunit is composed of an extracellular domain with a single immunoglobulin-like fold, a single transmembrane segment, and a small intracellular domain. A series of chimeras with exchanges of domains between the Na+ channel β1 and β2 subunits and between β1 and the structurally related protein myelin P0 were constructed and analyzed by two-microelectrode voltage clamp in Xenopusoocytes. Only chimeras containing the β1 extracellular domain were capable of β1-like modulation of Na+ channel gating. Neither the transmembrane segment nor the intracellular domain was required for modulation, although mutation of Glu158 within the transmembrane domain altered the voltage dependence of steady-state inactivation. A truncated β1 subunit was engineered in which the β1 extracellular domain was fused to a recognition sequence for attachment of a glycosylphosphatidylinositol membrane anchor. The β1ec-glycosylphosphatidylinositol protein fully reproduced modulation of Na+ channel inactivation and recovery from inactivation by wild-type β1. Our findings demonstrate that extracellular domain of the β1 subunit is both necessary and sufficient for the modulation of Na+ channel gating.

The voltage-gated Na ϩ channel is responsible for the increase in Na ϩ permeability during the initial rapidly rising phase of the action potential in neurons and other excitable cells. Na ϩ channels in rat brain contain an ␣ subunit of 260 kDa in association with auxiliary ␤1 and ␤2 subunits of 36 and 33 kDa, respectively (1,2). The ␣ subunit, which forms a functional voltage-gated channel by itself, is composed of four homologous domains, each containing six potential membranespanning segments and a pore-forming loop (3,4). The auxiliary subunits ␤1 and ␤2 enhance channel expression and modulate channel gating (5,6). Although they are not closely related in terms of amino acid sequence, the ␤1 and ␤2 subunits are predicted to be topologically similar (6). Both are heavily glycosylated intrinsic membrane proteins containing a large, N-terminal extracellular domain connected by a single transmembrane segment to a smaller C-terminal intracellular domain. In addition, both are predicted to have an Ig-like fold in their extracellular domains (6 -8).
When type IIA Na ϩ channel ␣ subunits are expressed in Xenopus oocytes, most of the expressed channels function in a slow gating mode, characterized by slow activation, slow and incomplete inactivation, and slow recovery from inactivation (9). Co-expression of the ␤1 subunit with ␣ IIA in Xenopus oocytes increases the proportion of channels that function in the normal fast gating mode (5,10). Na ϩ channel inactivation is accelerated 5-fold, the voltage dependence of fast inactivation is shifted in the negative direction, and a larger fraction of channels recovers rapidly from inactivation. In addition, the peak amplitude of Na ϩ current is increased, consistent with an increase in cell surface expression. The ␤2 subunit also increases Na ϩ channel expression and modulates channel gating but to a much lesser extent than ␤1 (6).
Previous studies of skeletal muscle and brain Na ϩ channels have shown that deletion of the intracellular domain of the ␤1 subunit has no effect on its modulation of ␣ subunit function, whereas deletions within the extracellular domain block modulation (8,11,12). Analysis of site-directed mutants revealed that the A/AЈ ␤ strand on the edge of the extracellular Ig-like fold is one point of interaction with the ␣ subunit (8). In contrast, another study reported that the transmembrane segment of the ␤1 subunit expressed alone was sufficient for Na ϩ channel modulation (13), and a human long QT syndrome mutation has implicated intracellular sequences in ␤1 function (14). In the experiments presented here, we have analyzed the functional properties of a series of chimeras between ␤1 subunits and either ␤2 subunits or the structurally related myelin protein zero (P0), as well as a construct in which the extracellular domain of ␤1 is attached to a glycophosphatidylinositol (GPI) anchor. Our results show that the extracellular domain of the ␤1 subunit is both necessary and sufficient for modulation of Na ϩ channel gating.

EXPERIMENTAL PROCEDURES
Plasmid Construction-cDNAs encoding the rat brain ␤1 subunit (5), the rat brain ␤2 subunit (6), and rat myelin P0 (15) were subcloned into plasmid pSP64T for analysis of Na ϩ channels by expression in Xenopus oocytes. Plasmid pSP64T-␤1 was modified such that extraneous restriction sites were removed from the polylinker, a silent ClaI site was introduced into the ␤1 cDNA at a position 14 nucleotides 5Ј from the proposed junction of the ␤1 extracellular (␤1 ec ) and transmembrane (␤1 tm ) domains (pSP64T-␤1ClaI), and a silent NotI site was introduced at a position 12 nucleotides 3Ј of the junction between the cDNA encoding the ␤1 tm and intracellular (␤ ic ) domains, to form plasmid pSP64T-␤1ClaI/NotI.
Construction of ␤1-␤2 and ␤1-P0 Chimeras-In general, chimeras in which entire domains were exchanged either between the ␤1 and ␤2 subunits or between ␤1 and P0 were constructed by polymerase chain reaction (PCR) 1 amplification of cDNA encoding the replacement do-* This work was supported by National Institutes of Health Research Grant NS25704 (to W. A. C.) and National Institutes of Health Postdoctoral National Research Service Award Grant NS09842-03 (to K. A. M). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. main, followed by subcloning into the appropriate pSP64T-based receiving plasmid. The 5Ј forward and 3Ј reverse PCR primers were designed with "tails" that contained small segments of nucleotides derived from the receiving plasmid that flank the region being replaced, resulting in in-frame fusions between replacement cDNAs and the cDNA of the receiving plasmid. Restriction endonuclease sites, either intrinsic or engineered silent sites, were incorporated into the PCR primer tails for use in the subsequent subcloning step.
Chimera ␤1P0␤1 was constructed by replacing the cDNA encoding ␤1 tm with that of P0 tm . The P0 tm PCR product contained 5Ј and 3Јflanking sequences derived from the ␤1 ec domain with the silent ClaI site and ␤1 ic with a silent NotI site, respectively. The PCR product was digested with ClaI and NotI and then ligated to ClaI/NotI-digested pSP64T-␤1ClaI/NotI to form plasmid pSP64T-␤1P0␤1.
To construct chimera P0␤1␤1, P0 ec cDNA was PCR amplified using a 5Ј PCR primer containing a 5Ј NcoI site, which mimicked an intrinsic NcoI site located at the start codon of ␤1. Nucleotides encoding the ␤1 tm domain, including a NdeI site intrinsic to ␤1, were added in frame to the 3Ј reverse primer used to make the P0 ec PCR product. The P0 ec PCR product was then subcloned into NcoI/NdeI-digested pSP64T-␤1 to form plasmid pSP64T-P0␤1␤1. Chimera P0␤1P0 was made by replacing the cDNA encoding ␤1 ic in pSP64T-P0␤1␤1 with cDNA encoding the P0 ic . The 5Ј end of the P0 ic PCR product contained cDNA encoding a portion of the ␤1 tm , including an intrinsic NdeI site, fused in frame to the P0 ic cDNA. A KpnI site was incorporated at the 3Јend of the P0 ic PCR product. The P0 ic PCR product was then subcloned into NdeI/KpnI-digested pSP64T-P0␤1␤1.
To facilitate the construction of chimera ␤1P0P0, myelin P0 cDNA was subcloned into plasmid pSP64T in a manner such that all extraneous AvaI sites were deleted. In this subcloning, a single AvaI located in the P0 tm domain was retained, and a ClaI site 5Ј to the P0 cDNA was carried over from the pSK ϩ polylinker. The ␤1 ec domain was then amplified with a ClaI site added 5Ј to the ␤1 ec cDNA, and the 3Ј end was fused in frame to a portion of the P0 tm cDNA with the intrinsic AvaI site. The ␤1 ec PCR product was subcloned into ClaI/AvaI-digested pSP64T-P0 to form pSP64T-␤1P0P0. All ␤1-␤2 and ␤1-P0 chimeras were confirmed by DNA sequence analysis.
Mutagenesis of ␤1 E158 -Oligonucleotide-directed mutagenesis was used to create the ␤1 E158Q and ␤1 E158K point mutations. Singlestranded M13Mp18-␤1 (8) served as the mutagenesis template. Mutagenesis was performed according to Kunkel (16). The mutations were identified by DNA sequence analysis. The corresponding cDNAs were then subcloned into pSK ϩ for further analysis.
Construction of ␤1 ec -GPI-Plasmid pSP64T-␤1 ec -GPI contains the cDNA encoding the ␤1 ec domain fused in-frame with cDNA encoding the GPI anchor recognition sequence from human placental alkaline phosphatase (HPAP) (17). cDNA encoding the HPAP GPI anchor recognition sequence was amplified from plasmid pBS-HPAP (18). The 5Ј primer tail contained a portion of the ␤1 ec domain, including a silent ClaI site, fused in-frame with nucleotides encoding the beginning of the HPAP GPI recognition sequence. A stop codon and KpnI site were incorporated into the 3Ј primer. The 186-base pair PCR product was subcloned into ClaI/KpnI-digested pSP64T-␤1ClaI/NotI. In this subcloning procedure, ␤1 tm and ␤1 ic were deleted.
Electrophysiological Recording-Following incubation at 20°C for at least 48 h, expressed Na ϩ channels were examined at room temperature by two-electrode voltage clamp (Dagan CA1, Minneapolis, MN) (8,20). Linear capacity currents were canceled electronically, and residual linear currents were subtracted using the P/4 procedure (21). The bath was perfused continuously with Frog Ringer (8). Na ϩ currents were elicited by step depolarizations to 0 mV from a holding potential of Ϫ90 mV. Inactivation curves were generated using prepulses of 100-ms duration to various voltages, followed by a test pulse to 0 mV. Recovery from inactivation was examined by applying a 15-ms conditioning pulse to 0 mV from a holding potential of Ϫ90 mV, followed by a recovery interval of variable duration and a test pulse to 0 mV.

Na ϩ Channel Modulation by Chimeras of the ␤1 and ␤2
Subunits-The Na ϩ channel auxiliary subunits ␤1 and ␤2 have little contiguous primary sequence similarity, but they share similar predicted conformation and topology (6). Inactivation of Na ϩ currents mediated by rat brain type IIA ␣ subunits alone is relatively slow (Fig. 1B, trace a). Both ␤1 and ␤2 modulate Na ϩ channel gating mode, but the modulatory effect of the ␤1 subunit is much greater than ␤2 (Ref. 6; Fig. 1B, traces b and c). Chimeric ␤1-␤2 cDNAs were constructed to investigate which domains of ␤1 are important for the modulation of Na ϩ channel gating (Fig. 1A). Chimera ␤1␤1␤2, in which the intracellular domain of ␤1 was substituted with the intracellular domain of ␤2, modulated Na ϩ channels like wild-type ␤1 when co-expressed with ␣ IIA (Fig. 1B, compare traces b and d). Co-expression of chimera ␤1␤2␤2, in which only the extracellular domain was derived from ␤1, also resulted in rapidly inactivating Na ϩ currents, similar to those seen for wild-type ␤1 (Fig. 1B, compare traces b and e). These results show that the extracellular domain of ␤1, exclusive of any amino acids in the ␤1 transmembrane segment, contains the necessary molecular determinants for modulation of Na ϩ channel kinetics. Na ϩ Channel Modulation by Chimeras Exchanging the Transmembrane Segments of Myelin P0 and the ␤1 Subunit-Although the results with the ␤1␤2␤2 chimera clearly implicated the extracellular domain in modulation of gating mode, we also investigated a different family of chimeras because ␤2 subunits physically interact with ␣ IIA and have a significant modulatory effect on sodium channel gating (6). Like ␤1, myelin P0 is a type-I intrinsic membrane protein, containing a large N-terminal extracellular domain, connected by a single transmembrane segment to a C-terminal intracellular domain (15). The extracellular domain of P0 contains an Ig fold of known three-dimensional structure that has previously been used as a model for the extracellular domain of ␤1 (8,22). Co-expression of the ␣ IIA subunit with myelin P0 in Xenopus oocytes at a 1:10 w/w ratio of RNAs yielded Na ϩ channels that were functionally similar to channels composed of only the ␣ IIA subunit ( Fig. 2A, compare traces a and c; Fig. 2, B and C). Channels comprised of ␣ IIA ϩ P0 were slower to inactivate and showed a slight positive shift in the voltage dependence of inactivation relative to ␣ IIA alone (Fig. 2, A and B). Recovery from inactivation was slower for ␣ IIA ϩ P0 channels than for ␣ IIA alone at a 1:10 w/w ratio of injected RNAs, resulting from an increase in both fast and slow time constants for recovery from inactivation ( Fig. 2C and Table I). When a 1:30 w/w ratio was tested, the time constants were similar, but the fraction of current recovering with the fast time constant was increased (Fig. 2C). The change in time constants in comparison with control suggests that all of the channels were associated with P0, even at the 1:10 ratio. The increase in the fraction of channels with rapid recovery from inactivation observed at the 1:30 ratio may be due to formation of homo-oligomers of P0 associating with the channel (22). Neither of these effects resembles the effect of the ␤1 subunit on Na ϩ channel function.
Because of the structural similarity and functional difference between ␤1 and myelin P0, we constructed chimeras that exchanged domains between these two proteins and analyzed their functional effects on ␣ subunits. The functional characteristics of chimera P0␤1P0 (Fig. 3A) closely resembled those of P0. The kinetics of inactivation and the voltage dependence of inactivation were similar to ␣ subunits alone (Fig. 3B, traces a and c; Fig. 3C). The recovery from inactivation of ␣ IIA ϩ P0␤1P0 channels was accelerated slightly compared with that for ␣ IIA alone because of an increase in the fraction of current recovering with the fast time constant (Fig. 3D and Table I).
Co-expression of the inverse chimera, ␤1P0␤1 (Fig. 3A), yielded fast inactivating Na ϩ channels that were similar to those observed in oocytes injected with ␣ IIA ϩ wild-type ␤1 transcripts (Fig. 3B, compare traces b and d). The voltage dependence of steady-state fast inactivation was shifted to more negative potentials, like that of channels containing the wild-type ␤1 subunit (Fig. 3C). Recovery from inactivation was fast, with no apparent slow component ( Fig. 3D and Table I). The most obvious difference between channels containing the wild-type ␤1 and those containing the ␤1P0␤1 chimera was the rate at which the channels were expressed. Channels composed of ␣ IIA and ␤1P0␤1 subunits expressed more slowly than their wild-type counterparts, requiring 3-4 days post-injection to reach peak currents comparable with those of ␣ IIA ϩ wild-type ␤1 channels (data not shown). The results of these studies demonstrate that although cell surface expression is delayed, the ␤1P0␤1 chimera associates with the pore-forming ␣ IIA sub-FIG. 1. Modulation of inactivation by chimeras between the Na ؉ channel ␤1 and ␤2 subunits. A, schematic representations of the domain structures of the Na ϩ channel ␤1 and ␤2 subunits and chimeras ␤1␤1␤2 and ␤1␤2␤2. Vertical divisions in the drawing represent the predicted boundaries of the extracellular (ec), transmembrane (tm), and intracellular (ic) domains (5,6). The large loop in the extracellular domains represents the Ig fold. ␤1-derived portions of the chimeras are gray, whereas the ␤2-derived portions are cross-hatched. The amino acid sequences of the chimera junctions are provided in single-letter codes. ␤2-derived sequences are shown in bold type, and predicted transmembrane segments are underlined. Letters in parentheses following the construct nomenclature refer to electrophysiological traces shown in B. B, Xenopus oocytes were injected with in vitro transcribed RNA encoding the Na ϩ channel ␣ IIA subunit alone or in combination with RNA derived from the ␤ subunit constructs depicted in A. All co-injections were at a 1:10 w/w ratio of ␣ to auxiliary protein, except for ␣ ϩ ␤2, which was injected at a 1:5 w/w ratio, to avoid the increase in cell capacitance conferred by high concentrations of the ␤2 subunit (6). Na ϩ currents were generated from 30-ms pulses to 0 mV, from a holding potential of Ϫ90 mV. Traces a (␣ IIA alone) and b (␣ IIA ϩ ␤1) are normalized averages of 9 and 12 cells, respectively. Dotted lines represent Ϯ 1 S.D. for these control traces. Normalized representative traces are given for FIG. 2. Electrophysiological properties of Na ؉ channels containing ␤1 subunit or myelin P0. A, Xenopus oocytes were injected with in vitro transcribed RNA encoding the Na ϩ channel ␣ IIA subunit alone or in combination with RNA encoding either the ␤1 subunit or myelin P0. Na ϩ currents were generated as described in the legend to  Table I. C, normalized, averaged recovery from inactivation time courses for channels without auxiliary subunits or containing myelin protein P0 or ␤1 subunits. Recovery from inactivation was examined as described under "Experimental Procedures." Symbols are as defined for B, with the addition of open circles representing ␣ IIA ϩ P0 (1:30 w/w, n ϭ 4). Solid lines represent fits to the averaged recovery values. Recovery data for ␣ IIA ϩ ␤1 channels were well fit with one exponential, whereas all other data were fit with sums of two exponentials. The derived recovery fractions, time constants, and number of cells examined for each condition are given in Table I. unit and modulates channel function normally, supporting the conclusion that the transmembrane segment of the ␤1 subunit is not required for modulation of channel gating.
Effects of ␤1 E158 in the Transmembrane Segment on the Voltage Dependence of Steady-state Inactivation-Chimera ␤1P0␤1, when co-expressed with the ␣ IIA subunit, caused a negative shift in the voltage dependence of fast inactivation of approximately 10.2 mV, significantly greater than the 7.1 mV shift observed for ␣ IIA ϩ wild-type ␤1 (Table I). Comparison of the proposed transmembrane segments of the Na ϩ channel ␤1 subunit and myelin P0 shows that the ␤1 transmembrane domain contains a negatively charged Glu at position 158 of ␤1, whereas the corresponding residue in the P0 membrane-spanning segment is hydrophobic (Fig. 4A, underlined letter). We  The P0 domain boundaries are as described previously (15). Letters in parentheses following the construct nomenclature refer to electrophysiological traces shown in B. Note that the predicted transmembrane domain of P0 (26 amino acids) is larger than that of ␤1 (22 amino acids). B, Xenopus oocytes were injected with in vitro transcribed RNA encoding the Na ϩ channel ␣ IIA subunit alone or in combination with RNA encoding the ␤1 subunit, myelin P0 or the ␤1-P0 tm chimeras depicted in panel A. Na ϩ currents were recorded as described in the legend to Fig. 1B. Traces a (␣ IIA alone) and b (␣ IIA ϩ ␤1) are as described in the legend to Fig. 1B. Normalized representative traces are given for ␣ IIA ϩ chimera P0␤1P0 (trace c), and ␣ IIA ϩ chimera ␤1P0␤1 (trace d). C, voltage dependence of steady-state fast inactivation for channels containing ␤1-P0 chimeras, relative to that for control channels. Steady-state fast inactivation of Na ϩ channels expressed in Xenopus oocytes was measured as described under "Experimental Procedures." Averaged normalized inactivation curves ␣ IIA alone (filled circles; n ϭ 8), ␣ IIA ϩ ␤1 (filled square; n ϭ 6), ␣ IIA ϩ P0␤1P0 (open triangle; n ϭ 4), and ␣ IIA ϩ ␤1P0␤1 (open inverted triangle; n ϭ 5) are shown. Fits and error bars are as described in the legend to Fig. 2B. D, normalized, averaged recovery from inactivation profiles for channels containing ␤1-P0 transmembrane chimeras and control channels. Recovery from inactivation was examined as described under "Experimental Procedures." Symbols are as defined for C. Solid lines represent curve fits of the averaged recovery values. Recovery curves for ␣ IIA ϩ ␤1 and ␣ IIA ϩ ␤1P0␤1 channels were well fit with one exponential, whereas all remaining curves were fit with the sum of two exponentials. The derived recovery fractions, time constants, and number of cells examined for each condition are given in Table I. examined the influence of ␤1 E158 on the voltage dependence of steady-state fast inactivation by neutralizing the negative charge by conversion to Gln (␤1 E158Q ) and by replacing the negative Glu with a positively charged Lys (␤1 E158K ). As shown in Fig. 4B, the replacement of ␤1 E158 with a Gln or Lys had no detectable effect on the kinetics of inactivation. Recovery of channels from inactivation was only slightly slowed as a result of the ␤1 E158Q and ␤1 E158K mutations ( Fig. 4C and Table I). However, both of the mutant ␤1 subunits conferred an enhanced negative shift in the voltage dependence of steady-state inactivation, compared with the shift conferred by the wildtype ␤1 subunit ( Fig. 4D and Table I). Evidently, neutralization or replacement of the negatively charged ␤1 E158 mimics the effects of the replacement of the ␤1 transmembrane domain with that of myelin P0. Na ϩ Channel Modulation by Chimeras Exchanging the Extracellular Segments of Myelin Protein P0 and the ␤1 Subunit-To further examine the role of the ␤1 extracellular domain, chimera ␤1P0P0 was constructed in which only the extracellular domain was derived from ␤1 (Fig. 5A). Currents expressed from cells co-injected with transcripts encoding the ␣ IIA subunit and ␤1P0P0 were smaller on average than those obtained from cells injected with only the ␣ IIA transcript, reaching 0.75-1.0 A 5 days post-injection compared with 1.5-8.0 A for ␣ IIA . Although the peak currents were small, channels from ␣ IIA ϩ ␤1P0P0-expressing oocytes were fast inactivating and resembled channels containing the wild-type ␤1 subunit in all parameters tested (Fig. 5, B-D, and Table I). As observed for the ␤1P0␤1 chimera and ␤1 E158 mutations, the voltage dependence of steady-state inactivation was shifted to more negative potentials than for wild-type ␤1 (Fig. 5C and Table I). Recovery from inactivation was fast for these channels, indicating that the ␤1P0P0 chimera associated well with the ␣ IIA subunit and fully modulated its gating (Fig. 5D). In contrast, chimera P0␤1␤1 (Fig. 5A), in which the extracellular domain of ␤1 was replaced with P0, did not significantly influence the kinetics of Na ϩ channel gating (Fig. 5B). In addition, the voltage dependence of inactivation for ␣ IIA ϩ P0␤1␤1 channels was shifted to more positive potentials than that of ␣ IIA alone, opposite to the negative shift conferred by the wild-type ␤1 subunit (Fig. 5C and Table I). Recovery from inactivation for channels expressed in ␣ IIA ϩ P0␤1␤1-injected oocytes resembled that observed for ␣ IIA alone ( Fig. 5D and Table I). However, recovery from inactivation was significantly accelerated when the P0␤1␤1 chimera RNA was injected at 40-fold weight excess relative to ␣ IIA transcript ( Fig. 5D and Table I). Aside from this small effect on recovery at high expression levels, the data from ␤1-P0 chimeras support the conclusion that the functional effects of ␤1 are conferred entirely by its extracellular domain. Na ϩ Channel Modulation by the Extracellular Domain of the ␤1 Subunit Attached to a Glycophospholipid Anchor-Taken together, the results of our studies of ␤1-␤2 and ␤1-P0 chimeras demonstrate that the ␤1 extracellular domain is critical for the modulation of channel function. One potential concern about the results with these chimeras is that the foreign transmembrane domains of the ␤2 subunit or myelin protein P0 could participate in association with the ␣ subunit or modulation of its function. To test the possible role of these foreign transmembrane domains, we designed a construct in which the extracellular domain of ␤1 is fused to the GPI anchor recognition FIG. 4. Electrophysiological properties of Na ؉ channels containing the ␤1 subunits with E158Q and E158K mutations. A, comparison of the predicted transmembrane segments from the Na ϩ channel ␤1 subunit and myelin protein P0. The amino acid sequences are given in single-letter codes and are shown with the putative membrane stop transfer positions aligned. The Glu at position 158 of the ␤1 subunit is underlined. B, Xenopus oocytes were injected with in vitro transcribed RNA encoding the Na ϩ channel ␣ IIA subunit alone or in combination with RNA encoding the wild-type ␤1 subunit, ␤1 E158Q or ␤1 E158K . Na ϩ currents were measured as described in the legend to Fig. 1B. Traces a (␣ IIA alone) and b (␣ IIA ϩ ␤1) are as described in in the legend to Fig. 1B. Normalized representative traces are given for ␣ IIA ϩ ␤1 E158Q (trace c) and ␣ IIA ϩ ␤1 E158K (trace d). C, normalized, averaged recovery from inactivation profiles for channels containing mutant ␤1 subunits and control channels. Recovery from inactivation was measured as described under "Experimental Procedures." Filled circle, ␣ IIA alone; filled square, ␣ IIA ϩ ␤1; open triangle, ␣ IIA ϩ ␤1 E158K ; open inverted triangle, ␣ IIA ϩ ␤1 E158Q . Solid lines represent curve fits of the averaged recovery values. The recovery curves for ␣ IIA ϩ ␤1, ␣ IIA ϩ ␤1 E158K , and ␣ IIA ϩ ␤1 E158Q channels were well fit with one exponential, whereas the ␣ IIA curve was fit with the sum of two exponentials. The derived recovery fractions, time constants, and number of cells examined for each condition are given in Table I. D, the voltage dependence of steady-state fast inactivation is plotted for channels containing mutant ␤1 subunits, relative to that of control channels. Symbols are as defined for C. The voltage dependence of steady-state fast inactivation of Na ϩ channels was measured as described under "Experimental Procedures." The inactivation curves shown are normalized averages. Fits and error bars are as described in the legend to Fig. 2B. sequence from HPAP (17,18) (Fig. 6A). GPI-anchored fusion proteins have been successfully used to study a variety of cell surface proteins (18,23,24). In GPI-anchored proteins, the recognition site for GPI attachment is large, usually over 30 amino acids. The GPI moiety is attached to an Asp residue within the recognition sequence, and the following 30 C-terminal residues are removed from the site (Refs. 25 and 26; Fig.  6A). cDNA encoding the entire 37-amino acid GPI recognition sequence of HPAP was inserted in frame into the expression plasmid pSP64T␤1 to form pSP64T␤1 ec -GPI. In the pSP64T␤1 ec -GPI construct, the GPI recognition sequence immediately follows the ␤1 extracellular domain, and the transmembrane and intracellular domains of ␤1 are deleted (see "Experimental Procedures").
The ␤1 ec -GPI RNA was synthesized in vitro and co-injected with ␣ IIA RNA into Xenopus oocytes at ␣ IIA : ␤1 ec -GPI weight ratios of 1:10 or 1:20, and Na ϩ currents were measured by two-electrode voltage clamp. The ␤1 ec -GPI protein expressed well as assessed by modulation of Na ϩ channel function (Fig. 6,  B-D). Peak currents for channels containing ␤1 ec -GPI were similar to those recorded for channels containing the wild-type ␤1 subunit, ranging from 2.5 to 8 A over a recording period of 3 days (␣ ϩ wild-type ␤1 (1:10), 3.07 Ϯ 0.44 A, n ϭ 6; ␣ ϩ ␤1 ec -GPI (1:10), 3.14 Ϯ 0.61 A, n ϭ 7; ␣ ϩ ␤1 ec -GPI (1:20), 3.54 Ϯ 0.39 A, n ϭ 5). The ␣ IIA ϩ ␤1 ec -GPI channels inactivated rapidly like wild-type ␣ IIA ϩ ␤1 channels (Fig. 6B). The voltage dependence of steady-state fast inactivation for channels containing the GPI-anchored ␤1 extracellular domain was similar to the intact ␤1 subunit, with V h1/2 shifted in the negative direction relative to ␣ IIA alone (Fig. 6C). In fact, the negative shift in the voltage dependence of steady-state inactivation was more pronounced than that observed for ␣ IIA ϩ wild-type ␤1 and more closely resembled the values determined for channels containing the ␤1P0␤1 or ␤1P0P0 chimeras or the ␤1 E158 mutations (Table I). These results are consistent with neutralization of the negative charge of ␤1 E158 causing a negative shift in the voltage dependence of inactivation. Analysis of the recovery from fast inactivation for channels expressed in ␣ IIA ϩ ␤1 ec -GPI injected oocytes showed no evidence of the slowly recovering channels typical of the ␣ IIA subunit alone (Fig. 6D). Moreover, the time constant for recovery from inactivation was statistically indistinguishable from that of ␣ IIA ϩ wild-type ␤1 channels (2.56 Ϯ 0.18 ms versus 2.68 Ϯ 0.15 ms; Table I). Taken together, the electrophysiological data show that the GPI-anchored ␤1 extracellular domain is efficiently expressed, is capable of association with the ␣ IIA subunit, and fully modulates the gating properties of the ␣ subunit. Evidently, ␤1 function was not compromised by the deletion of its transmembrane segment and intracellular domain. DISCUSSION Previous work has highlighted the importance of the ␤1 extracellular domain in modulation of Na ϩ channel function (8,11,12) and led to the working hypothesis that the Ig fold motif in the ␤1 extracellular domain serves as a scaffold to present molecular determinants of ␤1 for interaction with the ␣ subunit. This interaction causes an increase in the fraction of channels that gate in a fast mode (8). Negatively charged residues predicted to lie in the A/AЈ ␤ strand on one edge of the Ig fold were shown to be important for modulation of gating mode (8). The extracellular loop IVSS2-S6 in the ␣ subunit has also been identified as a point of attachment of the ␤1 subunit (12,27). In contrast with these experiments, studies in which the transmembrane domain of the ␤1 subunit was expressed as Components of the schematics are as described in the legend to Fig. 3A. Letters in parentheses following the construct name refer to electrophysiological traces shown in B. Note that extracellular domain of ␤1 (141 amino acids) is larger than that of P0 (123 amino acids). B, Xenopus oocytes were injected with in vitro transcribed RNA encoding the Na ϩ channel ␣ IIA subunit alone or in combination with RNA encoding the wild-type ␤1 subunit, chimera P0␤1␤1, or chimera ␤1P0P0. Na ϩ currents were measured as described in the legend to Fig. 1B. Traces a (␣ IIA alone) and b (␣ IIA ϩ ␤1) are as described in the legend to Fig. 1B. Normalized representative traces are given for ␣ IIA ϩ P0␤1␤1 (trace c) and ␣ IIA ϩ ␤1P0P0 (trace d). C, normalized, averaged steady-state fast inactivation measured as described under "Experimental Procedures." Filled circle, ␣ IIA alone (n ϭ 8); filled square, ␣ IIA ϩ ␤1 (n ϭ 6); open triangle, ␣ IIA ϩ P0␤1␤1 (n ϭ 5); open inverted triangle, ␣ IIA ϩ ␤1P0P0 (n ϭ 3). Fits and error bars are as described in the legend to Fig. 2B. D, normalized, averaged recovery from inactivation time courses are given for channels containing ␤1-P0 extracellular chimeras and control channels. Recovery from inactivation was examined as described under "Experimental Procedures." Symbols are defined as in panel C. Solid lines represent curve fits of the averaged recovery values. Recovery curves for ␣ IIA ϩ ␤1 and ␣ IIA ϩ ␤1P0P0 channels were well fit with one exponential, whereas all remaining curves were fit with the sum of two exponentials. The derived recovery fractions, time constants, and number of cells examined for each condition are given in Table I. a separate protein suggested that the transmembrane domain could modulate sodium channel gating by itself (13). Moreover, human heart ␣ subunits containing a mutation in the intracellular C-terminal tail are linked to long QT syndrome and are functionally abnormal only when co-expressed with the ␤1 subunit (14). In the experiments described here, we have further examined the possibility that the modulation of Na ϩ channel gating conferred by the ␤1 subunit is indeed due to extracellular interactions using chimeric and lipid-anchored ␤1 subunits.
The Transmembrane Domain of the ␤1 Subunit Is Not Sufficient for Na ϩ Channel Modulation-If the transmembrane domain were necessary for ␤1 function, one would expect to observe loss of function for the ␤1P0␤1 chimera and for ␤1 ec -GPI. Alternatively, if the transmembrane domain were sufficient for ␤1 function, one would expect to observe ␤1-like function for the P0␤1P0 chimera. Our results show the opposite. The ␤1P0␤1 chimera and ␤1 ec -GPI conferred wild-type levels of ␤1 modulation of Na ϩ channel gating, whereas the P0␤1P0 chimera lacked ␤1 function. The expression of channels containing the ␤1P0␤1 chimeras was delayed, probably because of protein folding problems caused by the longer, P0-derived transmembrane segment. We did observe a small but significant acceleration of recovery from inactivation in oocytes expressing the ␣ IIA subunit in combination with high levels of either the P0␤1P0 or P0␤1␤1 chimeras. In general, the Na ϩ currents mediated by these chimeras in oocytes having fast recovery kinetics were very small (Ͻ1 A; data not shown).
Because both of these chimeras contain the P0 extracellular domain, it is likely that residues within the P0 Ig fold motif can accelerate the recovery process but with much lower efficacy than the wild-type ␤1 subunit. This idea is plausible, because there are several short regions of sequence identity within the extracellular domains of ␤1 and P0 (8). The results of our experiments with chimeras demonstrate that the transmembrane domain of the ␤1 subunit cannot support modulation of sodium channel gating and therefore that its role in interactions with and modulation of the ␣ subunit is secondary to that of the extracellular domain. This conclusion is further supported by our results showing that ␤1 ec -GPI is fully active in channel modulation.
Removal of a Negatively Charged Residue in the ␤1 Transmembrane Domain Causes a Larger Hyperpolarizing Shift in the Voltage Dependence of Steady-state Inactivation-We observed that the replacement of the ␤1 transmembrane segment with that of myelin protein P0, as well as the complete removal of the transmembrane domain, resulted in a more pronounced negative shift in the voltage dependence of steady-state fast inactivation than that measured for channels containing the wild-type ␤1 subunit. These effects could be mimicked by neutralizing or charge replacement mutations at position ␤1 E158 , located within the transmembrane segment. Therefore, although it is unusual for a charged residue to be found in a transmembrane segment, ␤1 E158 does appear to play a significant role in determining the voltage dependence of inactivation. This suggests that ␤1 E158 is situated where it can alter the FIG. 6. Electrophysiological properties of Na ؉ channels containing a GPI-anchored ␤1 extracellular domain. A, schematic representations of the ␤1 subunit and the precursor and mature ␤1 ec -GPI proteins. Na ϩ Ch ␤1, amino acid sequences of the ␤1 extracellular (ec) and transmembrane (tm) shown in single-letter codes. The transmembrane domain is underlined. The last residue of the ␤1 extracellular domain and site of the fusion to the GPI recognition sequence, E141, are labeled. Pre-␤1 ec -GPI, amino acid sequence of the fusion region between ␤1 ec and the GPI recognition sequence is given. The residues in italics are enzymatically cleaved in the endoplasmic reticulum during the attachment of the GPI moiety. ␤1 ec -GPI, mature form of the GPI anchor showing the amino acid sequence at that attachment site in single letter code. B, Xenopus oocytes were injected with in vitro transcribed RNA encoding the Na ϩ channel ␣ IIA subunit alone or in combination with RNA encoding the wild-type ␤1 subunit or ␤1 ec -GPI. Co-injections were at a 1:10 w/w ratio of ␣ to auxiliary subunit RNA. An additional injection of ␣ IIA ϩ ␤1 ec -GPI was done at a 1:20 w/w ratio of ␣ IIA to ␤1 ec -GPI RNA. Na ϩ currents were measured as described in the legend to Fig. 1B. Traces a (␣ IIA alone) and b (␣ IIA ϩ ␤1) are as described in the legend to Fig. 1B. Normalized representative traces are given for ␣ IIA ϩ ␤1 ec -GPI (1:10) (trace c) and ␣ IIA ϩ ␤1 ec -GPI (1:20) (trace d). Traces b, c, and d were indistinguishable from each other. C, voltage dependence of steady-state fast inactivation for ␣ IIA ϩ ␤1 ec -GPI (1:20) channels relative to controls. Steady-state fast inactivation of Na ϩ channels was measured as described under "Experimental Procedures." The inactivation curves shown are normalized averages. Filled circle, ␣ IIA alone (n ϭ 8); filled square, ␣ IIA ϩ ␤1 (n ϭ 6); open diamonds, ␣ IIA ϩ ␤1 ec -GPI (n ϭ 7). Fits and error bars are as described in the legend to Fig. 2B. D, normalized and averaged recovery from inactivation profiles for ␣ IIA ϩ ␤1 ec -GPI (1:20) and control channels. Recovery from inactivation was examined as described under "Experimental Procedures." Symbols are as defined for C. Solid lines represent fits of the averaged recovery values. Recovery curves for ␣ IIA ϩ ␤1 and ␣ IIA ϩ ␤1 ec -GPI channels were well fit with one exponential, whereas the control ␣ IIA alone curve was fit with the sum of two exponentials. The derived recovery fractions, time constants, and number of cells examined for each condition are given in Table I. electric field sensed by the gating charges that control inactivation or can interact with them directly or indirectly through the protein structure.
The Extracellular Domain of the Na ϩ Channel ␤1 Subunit Is Sufficient for Modulation of Gating Mode-We found that chimera ␤1␤2␤2, in which the only extracellular domain is derived from ␤1, modulates channel function in a manner indistinguishable from that of wild-type ␤1. This agrees with previous experiments with skeletal muscle Na ϩ channels and a ␤1-␤2 chimera that contained the extracellular domain plus six adjacent transmembrane amino acid residues derived from ␤1 (12). Our data narrow the region required for ␤1 function in ␤1-␤2 chimeras to residues in the extracellular domain. In addition, the ␤1 extracellular domain is functional when spliced to the transmembrane and intracellular domains of myelin protein P0. Chimera ␤1P0P0 yielded channels with fast inactivation, negatively shifted voltage dependence of steady-state inactivation, and rapid recovery from inactivation. On the other hand, channels containing chimera P0␤1P0 were slow to inactivate, showed no negative shift in the voltage dependence of inactivation, and had slow recovery from inactivation. Taken together, the results of the ␤1-␤2 and ␤1-P0 chimeras strongly implicate the ␤1 extracellular domain as the region involved in the modulation of Na ϩ channel gating.
In further support of the conclusion that only the extracellular domain of the ␤1 subunit is necessary, we found that the GPI-anchored ␤1 extracellular domain is able to mimic the effects of the intact ␤1 subunit for all parameters tested. This is a clear demonstration that the action of the ␤1 subunit on channel modulation is primarily or exclusively via extracellular contacts. Our data from experiments assessing recovery from fast inactivation show that the ␤1 ec -GPI protein fully shifts the ␣IIA subunit to the fast gating mode. Moreover, the association of ␤1 ec with the ␣ subunit is stable for several days, consistent with stable, high affinity interaction like the native ␤1 subunit.
In addition to causing changes in the gating kinetics of the Na ϩ channel, the ␤1 subunit has also been shown to enhance Na ϩ channel expression in both Xenopus oocytes and mammalian cells (6,28). It would be interesting to determine which domain of the ␤1 subunit is responsible for this increase in channel cell surface expression. Chimeras between the ␤1 subunit and myelin P0 proved difficult to analyze in this respect. In general, the chimeras between ␤1 and P0 expressed poorly, yielding Na ϩ currents that were smaller on average than those observed for channels composed of only the ␣ IIA subunit. This dilemma was circumvented by the successful expression of Na ϩ channels containing the GPI-anchored ␤1 extracellular domain. These channels expressed well, with current amplitudes comparable with those observed for ␣ ϩ wild-type ␤1 subunits. The data suggest that the ␤1 extracellular domain is sufficient for increasing channel cell surface expression as well as modulation of gating.
Mechanism of Modulation of Gating by ␤1 Subunits-It is surprising that the extracellular domain of ␤1, independent of the transmembrane and intracellular segments, can alter Na ϩ channel gating. Activation of Na ϩ channels involves outward movement of the S4 transmembrane segments, which serve as voltage sensors (29,30). Fast inactivation is largely an intracellular event involving closure of an inactivation gate formed by the intracellular loop between domains III and IV (31)(32)(33)(34). Because the ␤1 subunit is unlikely to interact directly with either of these gating structures, it seems reasonable to propose that the ␤1 extracellular domain influences or evokes a conformational change in the ␣ subunit and thereby affects channel gating indirectly through the protein structure.
Polypeptide toxins from scorpions and sea anemones are also able to modulate Na ϩ channel gating by interaction with a site on the extracellular face of the channel, and, in that case, direct interaction of the bound toxins with the S3-S4 loops at the extracellular end of the S4 gating segments is proposed to mediate the actions of the toxins via a voltage-sensor trapping model (35,36). Likewise, small mutations in the S3-S4 loops of potassium channels have profound effects on voltage-dependent gating (37). The ␤1 subunit may also influence the movement of one or more S4 gating segments, perhaps by interacting with the S3-S4 loop of one or more homologous domains and, in that way, alter the kinetics and voltage dependence of channel gating.
Significance of the Extracellular Domain of the ␤1 Subunit in Vivo-It was recently discovered that a mutation of a Cys residue predicted to be disulfide-linked in the Ig fold in the extracellular domain of ␤1 is responsible for an inherited form of febrile seizure (38). This finding demonstrates the physiological importance of the Na ϩ channel ␤1 subunit and again emphasizes the role of the Ig fold in the ␤1 extracellular domain in regulation of Na ϩ channel function. Further molecular and structural analysis may reveal both the mechanism of action of the ␤1 subunits and the molecular basis for this inherited seizure syndrome.