Shal (Kv4) potassium channel genes encode classical subthreshold A-currents, and their regulation may be a key factor in determining neuronal firing frequency. The inactivation rate of Shal channels is increased by a presently unidentified class of proteins in bothDrosophila and mammals. We have cloned a novel Shal channel subunit (jShalγ1) from the jellyfish Polyorchis penicillatus that alters Shal currents from both invertebrates and vertebrates. When co-expressed with the conserved jellyfish Shal homolog jShal1, jShalγ1 dramatically changes both the rate of inactivation and voltage range of activation and steady-state inactivation. jShalγ1 provides fast inactivation by a classic N-type mechanism, which is independent of its effects on voltage dependence. jShalγ1 forms functional channels only as a heteromultimer, and jShalγ1 + jShal1 heteromultimers are functional only in a 2:2 subunit stoichiometry.
Transient K+ currents that are active at subthreshold potentials (A-currents) (Connor and Stevens, 1971a) are a dominant K+ conductance during the interspike interval and have a major influence on the firing frequency of neurons (Connor and Stevens, 1971b). Of the four subfamilies of genes that encode the channel-forming α-subunits of voltage-gated K+ channels (Shaker, Shab, Shal, and Shaw), the Shal (Kv4) subfamily most closely matches the description of the classical subthreshold A-current. The importance of subthreshold A-currents is underscored by the fact that Shal is the most highly conserved voltage-gated K+ channel subfamily in higher triploblastic metazoans (Salkoff et al., 1992).
Recent genetic and molecular evidence supports the idea that Shal channels underlie classical subthreshold A-currents in neurons of both vertebrates and invertebrates. Serodio et al. (1994) have shown that the major subthreshold A-current expressed in the rat brain is likely to be encoded by Shal genes. Shal channels in the mammalian brain are located primarily on the dendrites and cell bodies of neurons (Sheng et al., 1992), where their placement might influence spiking behavior. Using mutant analysis, Tsunoda and Salkoff (1995a,b) also found Shal currents in cell bodies of neurons from the fruit flyDrosophila, suggesting that the intracellular location and function of these channels is conserved across species.
Shal inactivation rate is regulated by an unknown mechanism in both vertebrates and Drosophila. Shal currents vary almost 50-fold in inactivation rate in Drosophila embryonic neurons. Most in vivo Shal currents inactivate far more rapidly than currents expressed when Shal cRNA is injected intoXenopus oocytes (Pak et al., 1991). In mammals, the inactivation rates of Shal currents can be increased by co-expression in Xenopus oocytes with low molecular weight fractions of mRNA from rodent brain (Chabala et al., 1993; Serodio et al., 1994). Both results suggest a mechanism for increasing the rate of inactivation in Shal channels that is not intrinsic to Shal α-subunits. Because Shal currents influence the duration of the interspike interval (Connor and Stevens, 1971b), their inactivation rates may determine neuronal firing patterns. Thus, understanding the mechanism controlling Shal inactivation rates may be important for understanding how neuronal firing patterns are generated.
We have explored the evolutionary origins of Shal channels in a primitive metazoan and have discovered a molecular mechanism for regulating the inactivation rate of Shal currents. We show here that Shal is highly conserved in the jellyfish Polyorchis penicillatus, a diploblastic coelenterate. Because diploblasts are the most primitive metazoans to have nervous systems, our results show that Shal currents were present in the first neurons that evolved in the last common ancestors of diploblasts and triploblasts, ∼700 million to 1 billion years ago (Morris, 1993). OnePolyorchis Shal homolog, jShalγ1, appears to function only as a heteromultimer in concert with a Shal α-subunit. Compared with homomultimeric Shal channels consisting of Shal α-subunits alone, Shalα + Shalγ heteromultimers produced currents that inactivate far more rapidly. In contrast to the cytosolic β-subunits that provide rapid inactivation in mammalian Shaker K+ channels (Rettig et al., 1994), jShalγ1 is homologous to the α-subunits of voltage-gated K+ channels.
MATERIALS AND METHODS
Cloning. Amplification and isolation of fragments ofjShal1 and jShalγ1 from Polyorchis penicillatus genomic DNA was performed as described in Jegla and Salkoff (1995) and Jegla et al. (1995). Briefly, the degenerate primers 5′-TCGGAATTCTATGACTACTGTTGGNTAYGGNGA-3′ and 5′-ACCTCTAGAGGTAGTGCTATTRYNAGNACNCC-3′, which are derived from the consensus amino acid sequences of the P-domain (MTTVGYGD) and the S6 domain (GVL(T/V)TIAL) of voltage-gated K+ channels, were used to amplify initial fragments. These were size-selected, reamplified using overlapping nondegenerate primers, and then subcloned to allow for isolation of individual fragments.
A complete jShal1 genomic clone was obtained using the initial jShal1 fragment to probe a Polyorchisgenomic library (provided by Dr. Warren Gallin, University of Alberta) under high-stringency conditions (Butler et al., 1989) and sequenced. The coding region of jShal1 consists of three exons, as indicated in Figure 1. Two exons encoding the N-terminal, S1–S6, and neighboring C-terminal cytoplasmic regions were predicted based on their high homology to dShal and mShal. The third exon encoding the poorly conserved distal C-terminal region was identified by amplification from an oligo dT-primedPolyorchis cDNA library using a jShal1-specific sense primer in the S6 region (5′-CCTGGTAAACTA-GTTGGTAGTATTTGCTCA-3′) and an antisense primer corresponding to the library vector sequence (5′-TCCGGTCGACGTAGAGGG-GAATAAATCGCCATA-3′). Construction of thePolyorchis genomic library and of cDNA libraries from neuronally enriched Polyorchis penicillatus tissue samples have been described previously (Gallin, 1991).
The jShalγ1 genomic sequence was obtained in two sequential rounds of inverse PCR (Ochman et al., 1988). The genomic DNA used for inverse PCR was prepared by first cutting with a desired restriction enzyme and then ligating at concentrations of 2 ng/μl or less to promote self-circularization. In the first round of inverse PCR, a 1.4 kb fragment of jShalγ1 was amplified from DNA prepared with BglII using sense (5′-CAAGTCTAGATGATAGGCTCTATGTGTTGCTTGAT-3′) and antisense (5′-AACTAAGCTTGGATGGTAACAGGAACAACATC-3′) primers derived from the original P-domain–S6 jShalγ1 fragment. Sequencing revealed that the fragment included a BglII site at the beginning of S2 and extended through the 3′ end of the open reading frame. Inverse PCR was also performed on genomic DNA prepared withHindII using jShalγ1-specific primers (5′-CAGTTGTTTAGTTATAACCCTCTCC-3′) and (5′-GCCAAAGAAAAGGTGGGGTCTTCAC-3′) located 5′ of a HindII site in S3. A 900 bp fragment obtained in this screen was found to extend through the 5′ end of jShalγ1 coding sequence. The 3′ coding region of jShalγ1was confirmed by PCR amplification from a Polyorchis cDNA bank by the same method as for jShal1, but with thejShalγ1-specific sense primer used in the first round of inverse PCR. A stop codon was found just 3′ of S6 in two cDNAs and is also present in the genomic sequence. The 5′ coding region was not found in any cDNA clones but was instead determined from genomic sequence. Our analysis, based on several observations, indicated that no introns interrupt coding in this region. First, no consensus sequences for acceptor or donor splice junctions were found in this region. Second, the codons used in this predicted 5′ region ofjShalγ1 match the coding bias we have observed for sixPolyorchis K+ channel genes (data not shown). Finally, this region has an A/T content of 64%, which falls within the range we have observed for Polyorchis coding sequence (60–65%), but well below the range we have observed forPolyorchis introns (72–76%).
Alignments and phylogenetic trees. The amino acid alignment (see Fig. 1) was generated using Microgenie (Beckman, Palo Alto, CA) and optimized by eye. Phylogenetic trees were constructed from these alignments by maximum parsimony, as implemented in the PAUP computer program (Swofford, 1993). Only sections of the T1 region and membrane spanning core (S1–S6) that have relatively conserved lengths (and thus certain alignment) among the Shaker, Shal, Shab, and Shaw subfamilies were used for tree building. Tree lengths were calculated using a step matrix that weighted changes between amino acids according to the minimum number of nucleotide changes that were necessary to achieve the change. A heuristic search for optimal trees was performed using random addition to generate initial trees. Tree bisection and reconnection were then used to optimize the trees. The consensus maximum parsimony tree (see Fig. 2) was constructed from the 18 shortest trees found in this search.
Expression vector construction. The complete open reading frame of jShal1 was united by removing the two introns from the genomic clone. The 3′ intron was removed from jShal1 in two subcloning steps. First, the C-terminal cDNA fragment (which spanned from S6 to the C terminal) was subcloned into Bluescript II SK+ (Stratagene, La Jolla, CA) using an SpeI site in S6 and an EcoRI site from the cDNA library vector. Second, a genomic fragment spanning from an XbaI site 5′ of the initiator methionine to the SpeI site in S6 was cloned into the SpeI site of this cDNA subclone. The P-domain intron was removed using overlap extension PCR off this template (Ho et al., 1989). The final overlap extension product was produced using a sense primer (5′-TTACGAATTCG-CCACCATGAAT GGTGACATAGGCGCTT-3′) that adds a consensus translation initiation sequence (underlined) (Kozak, 1987) surrounding the jShal1initiator methionine and an antisense primer from the Bluescript vector. This product was cut with EcoRI and cloned into theEcoRI site of the Xenopus oocyte expression vector pBSMXT (Wei et al., 1994). The coding region was then sequenced and compared with the jShal1 genomic clone to confirm that no PCR-introduced mutations existed.
A similar overlap extension protocol was used to remove both introns from the open reading frame of jShalγ1. The final product was made with a sense primer (5′-ATATGGATCCACCATGTATTCGGTTACTTC-CACTGCAAC-3′) that introduced a consensus translation initiation sequence (underlined) to the jShalγ1 initiator methionine and an antisense primer (5′-CTTATCTAGATCAATCTTCTTCGCTAGCCTTCA- TTTGAATTATTGGGACAGG-3′) that includes the jShalγ1 stop codon and spans coding sequence on both sides of the 3′ intron. It was cut at flanking BamHI andXbaI sites introduced in the primers and subcloned into the pOX expression vector. Four individually amplified clones were sequenced and compared with each other as well as with the original inverse PCR-generated jShalγ1 clones, allowing PCR-introduced mutations to be identified. A jShalγ1 expression vector clone containing two silent mutations was used in all physiological experiments. The pOX vector was constructed by insertingXenopus β-globin 5′ and 3′ untranslated sequences contained in pBSMXT (Wei et al., 1994) into new restriction sites in pBluescript II KS+ (Stratagene). Briefly, the Xenopusβ-globin 5′ untranslated and an NheI site was inserted between the KpnI and SalI sites of the vector, while an XhoI site and the Xenopus β-globin 3′ untranslated were inserted between the XbaI andNotI sites.
Expression and electrophysiology. Capped cRNAs were prepared by run-off transcription with T3 RNA polymerase using the mMessage mMachine kit (Ambion, Austin, TX) and diluted in RNase-free ddH2O to desired concentrations before injection. Mature stage IV Xenopus oocytes were prepared for injection as described in Wei et al., 1990. Oocytes were injected with 50 nl of cRNA and incubated at 18°C for 1–5 d in ND96 containing (in mm): 96 NaCl, 2 KCl, 1.8 CaCl2, 1 MgCl2, 5 HEPES–NaOH, pH 7.5, supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, and 2.5 mm sodium pyruvate. Recording methods used were as published previously inCovarrubias et al., 1991. Briefly, whole-cell recordings were made 1–5 d after injection at room temperature (∼22°C) by conventional two-microelectrode voltage-clamp techniques. Electrodes ranged from 0.2 to 0.5 MΩ in resistance and were filled with 3 m KCl. The standard recording solution consisted of ND96 with 1 mm4,4′-diisothiocyanatostilbene-2,2′ disulfonic acid to block native oocyte chloride currents. Currents were digitally acquired with CCURRENT using linear leak subtraction, filtered at 1 kHz with an eight-pole Bessel filter and analyzed with CQUANT. Capacitative transients were removed by leak subtraction or clipped out.
Stoichiometry calculations. Time constants (see Fig. 7) were calculated by least-squares fits of double-exponential functions to traces like those shown in the insets. Because the slowly inactivating jShal1 current could be inactivated by a prepulse to −90 mV, such a prepulse was often used to help isolate the fast components produced by jShal1 + jShalγ1 heteromultimers in oocytes expressing both currents. The free-mixing curve shown in Figure 7 assumes a binomial distribution of channels containing zero to four jShalγ1 inactivation balls. Ifp equals the fraction of jShal1 subunits and qequals the fraction of jShalγ1-subunits, then the distribution of channel types is as follows: p4 (0 balls) +p3q (1 ball) +p2q2 (2 balls) +pq3 (3 balls) + q4 (4 balls). The fractional size of the slowly inactivating current (jShal1 homomultimers, 0 balls) represents p4 and could thus be used to predict the fraction of each type of channel. The predicted fast-inactivation time constants were obtained by adjusting the inactivation rate constant calculated for these currents to the average number of inactivation balls per channel predicted from the binomial distribution of channel types. Only those channels containing one to three balls were included in the calculations of the fast-inactivation time constants, because channels with four balls (jShalγ1 homomultimers) are nonfunctional. Thus, if free mixing occurs, the fastest inactivation time constants observed in jShalγ1-biased currents with no jShal1 homomultimeric fraction (2.69 msec) should be produced primarily by channels with three inactivation balls. Thus, the time constant of fast N-type inactivation should approach three times this value as the proportion of jShal1 is increased. The equations for calculating inactivation rate constants from the time constants of inactivation and recovery and for calculating the free mixing curve are from MacKinnon et al., 1993.
The time constants (see Fig. 8) were determined by triple exponential fits to current traces like those shown in the insets, because two time constants of slow inactivation (from jShal1 + jShalγ1(T) heteromultimers) contributed to the accuracy of the fit. The curve for the prediction ofchannels containing three jShalγ1-subunits assumes that four types exist, all heteromultimers with three jShalγ1-based-subunits and one jShal1-subunit, but with zero to three inactivation balls. Because of the high ratio of jShalγ1 and jShalγ1(T) to jShal1 used in these experiments, few jShal1 homomultimers are predicted to exist, allowing these channels to be excluded from the calculations. If p equals the fraction of jShalγ1(T)-subunits and q equals the fraction of jShalγ1-subunits, then the distribution of channel types is:p 3 (0 balls) +p 2 q (1 ball) +pq 2 (2 balls) + q 3 (3 balls). The fraction of slowly inactivating current (p 3) was used to calculate the fraction of each channel type. For the prediction of two jShalγ1-subunits in each channel, the distribution is simplified to three possible channel types: p 2 (0 balls) + pq (1 ball) +q 2 (2 balls). In this case, the slowly inactivating current fraction equals p 2. For the prediction of a single jShalγ1-subunit, just two channel types will exist, p and q, and the slowly inactivating current fraction equals p. The curves were again calculated by the method of MacKinnon et al. (1993) but with the binomial distributions explained above.
Cloning and conservation
A previous study of the evolutionary origins of voltage-gated K+ channels showed that homologs of mammalian channels are present in the jellyfish Polyorchis penicillatus, which is among the simplest extant metazoans to have an organized nervous system (Jegla et al., 1995). We infer from this result that a similar set of voltage-gated K+ channels play an indispensable role in the nervous systems of all metazoans. Shal is the most conserved among these channels in the higher triploblastic metazoa, and here we describe highly conserved Polyorchis Shal homologs,jShal1 and jShalγ1. Because regulation of Shal channels may be a key element in the generation of patterned neuronal output, we have now focused attention on the mechanism of this regulation, which may be conserved as well.
jShal1 and jShalγ1 gene fragments were initially isolated by a PCR screen of Polyorchis penicillatus genomic DNA. Degenerate primers for this screen were based on regions of the P-domain (MTTVGYGD) and S6 transmembrane domain (GVL(T/V)IAL) that are highly conserved among voltage-gated K+ channels. The P-domain–S6 fragments of jShal1 and jShalγ1 were used to isolate complete coding sequences for these genes both by hybridization and PCR techniques, as described in Materials and Methods. Figure1 shows the deduced amino acid sequences of the jShal1 and jShalγ1 proteins compared with the sequences of Shal proteins from the fruit fly Drosophila (dShal) and mouse (mShal, Kv4.1). Each contains the characteristic structural features of voltage-gated K+ channel α-subunits.
jShal1 is very clearly a direct homolog of triploblastic Shal genes, because the jShal1 protein shares high conservation to triploblastic Shal α-subunits over virtually its entire length. It is ∼65% identical to dShal and mShal across the membrane-spanning channel core (Table 1). Interspecies homologs of specific voltage-gated K+ channel genes typically share at least 50% amino acid identity over this region, and jellyfish and triploblastic Shaker homologs share only this 50% amino acid identity (Jegla et al., 1995). The higher level of conservation in jShal1 is consistent with previous observations that Shal is the most highly conserved subfamily of voltage-gated K+ channels (Salkoff, 1992). Conservation is nearly as high in the T1 domain, which mediates subfamily-specific channel assembly (Li et al., 1992; Shen et al., 1993; Shen and Pfaffinger, 1995). jShal1 also has a conserved genomic structure; the positions of two introns, one in the P-domain motif and one in the C-terminal cytoplasmic domain, are perfectly conserved among jShal1, dShal, andmShal (Fig. 1, arrows).
In contrast, jShalγ1 is less well conserved and shares barely >40% amino acid identity from S1 to S6 (Table 1). Despite this lower conservation, several lines of evidence lead us to putjShalγ1 in the Shal K+ channel subfamily. As with all Shal α-subunits, jShalγ1 contains an intron at the “Shal-specific” site in the P domain. Introns are not found at this position in genes from the Shaker, Shab, or Shaw subfamilies. Secondly, Shal-specific conservation is high (>50%) in the T1 subfamily-specific assembly domain. Finally, phylogenetic analysis of jShal1, jShalγ1, and 15 other voltage-gated K+ channel proteins representing the Shaker, Shab, Shal, and Shaw subfamilies unequivocally places jShalγ1 within the Shal subfamily. A consensus of the 18 shortest trees found in a heuristic search using maximum parsimony (PAUP, Swofford, 1993) is shown in Figure 2. jShalγ1 is placed within the Shal subfamily in all of these trees. The order of branching between channel subfamilies shown in Figure 2 is not consistent in these 18 trees and should not be taken to indicate the evolutionary relationships of these subfamilies.
The lower conservation shared between jShalγ1 and Shal α-subunits is especially evident in the S4 voltage sensor (Papazian et al., 1991), which is perfectly conserved between dShal and mShal and has only two substitutions of 22 residues in jShal1. In contrast, there are 12 substitutions in the same 22 residues of jShalγ1 (Fig. 1). Interestingly, these S4 differences are found in the hydrophobic residues, whereas the string of positively charged residues characteristic of S4 is identical.
Two other key features distinguish jShalγ1 from jShal1 and other Shal homologs. One is that jShalγ1 has a group of seven positively charged residues at near its N terminal that is not found in other Shal homologs (Fig. 1). This motif is reminiscent of an inactivation “‘ball,” similar to those responsible for rapid inactivation in both triplobastic and diploblastic Shaker channels (Hoshi et al., 1990;Zagotta et al., 1990; Jegla et al., 1995). Later, we will show that these charges are indeed part of a functional inactivation ball in channels containing jShalγ1. The second feature is that after S6, jShalγ1 has a segment of just seven amino acids and thus lacks a long conserved C-terminal cytoplasmic segment found even in jShal1 (Fig. 1).
Shal1 expresses a rapidly activating transient current inXenopus oocytes that resembles Shal currents fromDrosophila and mammals. Figure 3 shows a comparison of the jShal1 current and the Drosophila Shal current dShal. The biophysical properties of these two currents are summarized in Table 2. Both currents share features of classic subthreshold A-currents that distinguish Shal currents from other voltage-dependent K+ currents. These include a hyperpolarized steady-state inactivation curve and an inactivation time course that is relatively insensitive to voltage. However, the jShal1 current differs in two important ways. First, its inactivation rate is severalfold slower than that of dShal. Whereas the fastest inactivation time constant is ∼140 msec for jShal1 at +50 mV, it is ∼40 msec for dShal. Second, jShal1’s activation and steady-state inactivation curves are shifted to even more hyperpolarized voltages. TheV 50 for activation of jShal1 is approximately −35 mV more hyperpolarized than that of dShal (Fig. 3 C). The V 50 for steady-state inactivation for jShal1 is approximately −106 mV compared with −62 mV for dShal (Fig. 3 D).
jShalγ1 is unlike an α-subunit in that it does not form functional voltage-dependent channels when expressed as a homomultimer in Xenopus oocytes (Fig.4 A). Instead, jShalγ1 functions only in combination with Shal α-subunits. Figure 4 Bshows currents resulting from its co-expression with jShal1 inXenopus oocytes. Because this current is distinct from jShal1 currents, it is assumed that it results from the heteromultimeric assembly of jShalγ1 and jShal1. Relative to jShal1 homomultimers, jShal1 + jShalγ1 heteromultimers inactivate much more rapidly with a time constant of <3 msec at +50 mV. This represents a several hundredfold increase in the inactivation rate from jShal1 currents (Table 2). These heteromultimeric jellyfish Shal currents resemble the majority of Shal currents expressed inDrosophila embryonic neurons, which inactivate with time constants near 5 msec (Tsunoda and Salkoff, 1995a). A second major change produced by the co-expression of jShalγ1 with jShal1 is a large depolarizing shift in the activation and steady-state inactivation curves compared with jShal1 (∼+30 mV for activation and ∼+35 mV for steady-state inactivation) (Fig.4 C,D, Table 2). Unlike jShal1 homomultimers, these heteromeric channels are much more typical of Shal channels from triploblasts with regard to their voltage range of activation.
An N-type inactivation mechanism in jShalγ1
Because jShalγ1 is unique among Shal homologs in containing a positively charged inactivation ball-like motif at its N terminal, we tested the possibility that jShalγ1 causes rapid inactivation in jShal1 currents by contributing an N-type inactivation mechanism. To investigate this, we removed the positively charged N terminal from jShalγ1; this truncated construct is referred to here as jShalγ1(T). This N-terminal truncation does indeed result in removal of fast inactivation from the heteromultimeric current (Fig.5 A, Table 2), providing the first clear example of rapid N- type inactivation in Shal channels. The mechanism of N-type inactivation conferred by jShalγ1 is independent of the shift in the voltage range of activation. Thus, the activation and steady-state inactivation curves of jShal1 + jShalγ1(T) heteromultimers resemble those of jShal1 + jShalγ1 heteromultimers; both have a depolarizing shift relative to that of jShal1 homomultimers (Fig. 5 B,C). Hence, this depolarized voltage range is likely to be conferred by regions other than the N terminal such as the poorly conserved regions of the jShalγ1 core. For instance, it was demonstrated that changes in the hydrophobic residues of the S4 region of Shaker can cause large shifts in the activation range (Lopez et al., 1991). Although fast N-type inactivation is absent in jShal1 + jShalγ1(T) heteromultimers, slower inactivation remains that is accelerated with respect to the inactivation of jShal1 homomultimers (Fig. 5 A, Table 2). The residues responsible for this increased rate of residual inactivation are not known. Because N-type inactivation has been removed from jShalγ1(T), the remaining inactivation may occur by a mechanism similar to C-type inactivation in Shaker channels, which depends on residues in the P-domain and S6 (Hoshi et al., 1991, Lopez-Barneo et al., 1993).
The highly charged N-terminal domain of jShalγ1 had a profound effect on recovery from inactivation as well as on the inactivation rate. Recovery from inactivation in jShal1 + jShalγ1 heteromultimers was measured and compared with recovery of jShal1 + jShalγ1(T) heteromultimers and jShal1 homomultimers (Fig. 5 D). jShal1 + jShalγ1 heteromultimers recovered slowest, with a time constant of several seconds at −100 mV. However, in jShal1 + jShalγ1(T) heteromultimers, recovery is significantly more rapid and is complete within a few hundred milliseconds (Fig. 5 D). Thus, the very slow phase of recovery from inactivation is attributable to fast N-type inactivation. On the other hand, recovery from slow “C-type” inactivation is actually faster in jShal1 + jShalγ1(T) heteromultimers than in jShal1 homomultimers (Fig. 5 D). jShal1 recovery was measured at −120 mV instead of −100 mV because, as Figure 5 C illustrates, most jShal1 channels are inactivated at −100 mV. jShal1 + jShalγ1 heteromultimers have a small but faster component to recovery, which may be caused by a few channels entering an alternative “C-type” inactivated state before N-type inactivation can occur. The very slow recovery from N-type inactivation indicates that the association between the jShalγ1 inactivation ball and its blocking site on jShal1 + jShalγ1 heteromultimers is very strong.
jShalγ1 heteromultimerizes with dShal and mShal
jShalγ1 does not appear to form functional heteromultimers with α-subunits from subfamilies of voltage-gated K+ channels other than Shal. In co-expression experiments, we found no evidence that jShalγ1 alters any biophysical properties of channels formed by the Polyorchis Shaker homologs jShak1 and jShak2 (data not shown). However, the ability of jShalγ1 to form heteromultimers with Shal α-subunits from other species is conserved. Evidence for this is presented in Figure 6, which shows currents resulting from the co-expression of jShalγ1 with dShal or mShal. Inactivation rates are increased when dShal and mShal are co-expressed with jShalγ1 in Xenopus oocytes. However, co-expression of jShalγ1 with dShal and mShal has only minor effects on the activation range, probably because homomultimeric dShal and mShal channels already operate in the voltage range in which jShalγ1 biases activation (data not shown).
Several observations suggest that the inactivation ball binding site on heteromultimers containing dShal or mShal has lower affinity for the jShalγ1 inactivation ball than the binding site on heteromultimers containing jShal1. The increase in inactivation rate in the interspecies heteromultimers is clearly caused by the N-terminal inactivation ball of jShalγ1, because inactivation is slowed if this N-terminal ball is removed (Fig.6 A,B). Nevertheless, N-type inactivation in these heteromultimers is not nearly as fast or complete as when jShalγ1 is mixed with jShal1 (Fig. 6 C, Table3). This shows that the Shal α-subunits, which do not have a high affinity inactivation ball themselves, contribute to the binding affinity of the jShalγ1 inactivation ball. This result is consistent with observations of the Drosophila Shaker channel showing that only one inactivation ball at a time is able to block the channel (MacKinnon et al., 1993). Taken together, these results suggest that N-type inactivation particles may bind to a single site formed by all four subunits (Murrell-Lagnado and Aldrich, 1993).
Stoichiometry of jShal1 + jShalγ1 heteromultimers
We have exploited the presence of the N-terminal inactivation ball on jShalγ1 to estimate the subunit stoichiometry of jShal1 + jShalγ1 heteromultimers. Our strategy was based on two assumptions. (1) The functional mechanism of N-terminal inactivation provided by jShalγ1 is the same as in Shaker N-type inactivation (MacKinnon et al., 1993). (2) jShal1 and jShalγ1 form tetrameric channels. This latter assumption is based on the structural similarity of Shal α-subunits to Shaker α-subunits, which are known to form tetramers (MacKinnon, 1991). In Shaker, the inactivation ball contributed by each of the four subunits functions independently (MacKinnon et al., 1993). This independence results in a simple additive effect of each ball on the inactivation rate. Thus, the inactivation rate of channels that have only one inactivation ball would be roughly four times slower than the inactivation rate for channels containing four inactivation balls. This result was shown by mixing increasing proportions of Shaker subunits lacking an inactivation ball with Shaker subunits having an inactivation ball. In these experiments, the inactivation rate is progressively slowed, because the number of inactivation balls per channel is progressively reduced.
Similarly, if jShal1 and jShalγ1 mix freely to produce functional heteromultimers of several stoichiometries, then increasing the proportion of jShal1 in an oocyte will progressively reduce the average number of inactivation balls on each heteromultimeric channel. This should result in slower N-type inactivation. If, on the other hand, there is only a single functional stoichiometry of jShal1 + jShalγ1 heteromultimer, then the rate of N-type inactivation should remain constant as the proportion of jShal1 is increased (while the relative amount of slowly inactivating current increases). Figure7 illustrates this unchanging rate of fast inactivation and suggests that there is only a single functional stoichiometry in jShal1 + jShalγ1 heteromultimers. As predicted, when the amount of jShal1 is increased relative to the amount of jShalγ1, the fast-inactivating component of the current is progressively reduced, leaving a larger slow component (corresponding to the jShal1 homomultimeric current). Significantly, the time constant of fast inactivation remains unchanged.
To further investigate the stoichiometry of jShal1 + jShalγ1 heteromultimers, we co-injected jShalγ1(T), which has the inactivation ball removed, with jShal1 and jShalγ1. Unlike the previous experiment, increasing proportions of jShalγ1(T) should progressively reduce the average number of inactivation balls per heteromultimeric channel, even if the functional stoichiometry of jShalγ1 + jShal1 heteromultimers is fixed. In contrast to the experiments shown in Figure 7, the fast-inactivation rate should progressively slow. Oocytes injected with jShalγ1(T), jShalγ1, and jShal1 again had both fast- and slow-inactivating current fractions (Fig. 8). Most importantly, the rate of fast N-type inactivation is slowed almost twofold by the addition of increasing amounts of jShalγ1(T). By analogy to the Shaker results, the twofold slowing of N-type inactivation suggests that the number of inactivation balls is reduced by half when jShalγ1(T) is added. These results can be explained by postulating three significant current components: a very fast component produced by channels with two jShal1 and two jShalγ1-subunits, a slower “fast” component produced by channels with two jShal1, one jShalγ1, and one jShalγ1(T)-subunits, and the slow component produced by channels with two jShal1 and two jShalγ1(T)-subunits. A second slow component produced by jShal1 homomultimers is of insignificant size in these mixes, because of the high ratio of jShalγ1 and jShalγ1(T) to jShal1. Thus, it appears that heteromultimers are limited to a single stoichiometry of two jShal1-subunits and two jShalγ1-subunits (Fig. 8). The fact that we can progressively reduce the number of inactivation balls without completely removing N-type inactivation strongly supports our initial assumption that the jShalγ1 inactivation balls function independently of each other, exactly like Shaker inactivation balls.
Significance of the conservation of Shal in diploblasts
Connor and Stevens (1971b) first recognized that subthreshold A-currents have the capacity to influence neuronal firing frequency because of their active role during interspike intervals. Patterned neural output from single neurons appears to be highly conserved among animals. Neurons of diploblastic coelenterates such as Polyorchis are capable of firing trains of action potentials or rhythmic bursts of action potentials that are virtually indistinguishable from those of vertebrates (Anderson, 1979; Przysiezniak and Spencer, 1989). The high conservation of Shal channels is likely to be part of the reason. The conservation of several additional classes of voltage-gated ion channels also contributes to this conservation of intrinsic electrical properties between diploblastic and triploblastic neurons (Anderson and McKay, 1987; Dunlap et al., 1987; Holman and Anderson, 1991;Przysiezniak and Spencer, 1992, 1994; Anderson et al., 1993; Meech and Mackie, 1993; Jegla et al., 1995). This growing set of channels known to be shared by diploblasts and triploblasts may represent an essential set for the patterning of signals in all nervous systems and suggests that the intrinsic electrical properties of neurons were optimized early in the evolution of the nervous system.
Regulation of Shal inactivation by jShalγ1
Our finding that jShalγ1, a functionally novel Shal-subunit, regulates the inactivation rate of Shal currents parallels the discovery that β-subunits can induce rapid inactivation in Shaker currents (Rettig et al., 1994). In contrast to jShalγ1, these β-subunits are cytosolic proteins homologous to members of the NAD(P)H-dependent oxidoreductase superfamily (McCormack and McCormack, 1994) and are not structurally homologous to voltage-gated K+ channel α-subunits. Although both jShalγ1 and Shaker β-subunits cause rapid inactivation by N-type mechanisms, jShalγ1 has additional effects on the voltage range of channel activation. This is, perhaps, attributable to the fact that jShalγ1 forms an integral part of the voltage-sensing mechanism of the channel.
Although jShalγ1 is similar in structure to the α-subunits of Shal and other voltage-gated K+ channels, it is unique in that it does not form functional homomultimeric channels. Instead, it modifies the gating properties of Shal α-subunits, functioning only as a heteromultimer. This distinct functional role has led to an interesting pattern of conservation between Shal α-subunits and jShalγ1. Regions involved in subfamily-specific assembly (T1) and ion selectivity (pore region) are highly conserved, whereas regions involved in determining gating properties (N terminal, S1–S4) differ substantially.
The reason that jShalγ1 does not form functional homomultimers is not clear, but it is not exclusively attributable to constitutive inactivation resulting from the presence of four high-affinity inactivation balls. This is demonstrated by the fact that the N-terminal truncated construct jShalγ1(T) also fails to express a voltage-dependent current as a homomultimer. The unusually short C-terminal cytoplasmic domain is also unlikely to be responsible for the failure of homomultimer formation. This was shown by substituting the longer C-terminal domain of jShal1 for the shorter C-terminal domain of jShalγ1; expression of voltage-dependent currents was not recovered (data not shown). These experiments strongly imply that jShalγ1 is specifically designed to function only in a heteromultimeric configuration.
The restriction of functional jShal1 + jShalγ1 heteromultimers to a single 2:2 functional stoichiometry may be necessary to precisely fix the voltage range and inactivation rate of the current rather than to allow gradations of these properties. Because the jShalγ1 inactivation ball binding site appears to involve both the jShal1 and jShalγ1-subunits, a precise arrangement of these two types of subunits may be necessary to produce a high-affinity binding site. If so, only one of two possible arrangements of the two jShalγ1-subunits within the 2:2 heteromultimers (side by side or opposite) may produce functional fast-inactivating channels. It may be possible to address this question in the future by using tandem constructs of jShalγ1 and jShal1. If both arrangements are functional, fast-inactivating heteromultimers should be obtained by expression of jShalγ1 + jShal1 dimers (opposite arrangement) as well as by co-expression of jShalγ1 + jShalγ1 and jShal1 + jShal1 dimers (side by side arrangement). We do not know whether 3:1 and 1:3 stoichiometries of jShal1 + jShalγ1 heteromultimers assemble but are nonfunctional or simply do not assemble.
In addition to jShalγ1, three other K+ channel subunits have been isolated that appear to belong to a specific voltage-gated K+ channel subfamily but do not express currents as homomultimers: IK8 and K13 from rat (Drewe et al., 1992) and nShaw1 from the nematode Caenorhabditis elegans (Dr. Aguan Wei, personal communication). IK8 and K13 are more closely related to the Shab subfamily, whereas nShaw1 is a Shaw homolog. Although no functional link between these subunits and any Shab or Shaw α-subunits has yet been established, it is conceivable that these could be similar γ-subunits for the Shab and Shaw subfamilies. Similar subunits have been found for cyclic nucleotide-gated channels (Bradley et al., 1994; Liman and Buck, 1994) and for inward rectifier K+ channels (Krapivinsky et al., 1995). Such “regulatory” subunits may turn out to be common but have simply been difficult to clone and characterize by traditional methods because of their sequence divergence and inability to function independently.
Because the functional expression of jShalγ1 appears to depend on co-expression with jShal1 α-subunits, both must be expressed in the same cells for either to have functional significance. [Most homomultimeric jShal1 channels would be inactivated at the resting potentials of all Polyorchis cells described so far (Przyiezniak and Spencer, 1989).] Because of this apparent functional interdependence, it seems likely that both are expressed in the same cells, but an additional possibility is that jShalγ1 co-assembles with an as yet unknown Polyorchis Shal α-subunit. Although we have not verified that jShalγ1 and jShal1 are expressed in the same cells in Polyorchis, we do know that both are present in the same neuronally enriched cDNA libraries (Gallin, 1991).
Are jShalγ1 homologs present in triploblasts?
Because Shal subfamily K+ channel genes are so highly conserved, it is plausible that triploblasts will be found to have homologs of jShalγ1. Indeed a physiological role for such homologs has been tentatively identified. The fast-inactivating Shal currents found in Drosophila neurons (Tsunoda and Salkoff, 1995a) are much more similar to jShal1 + jShalγ1 heteromultimeric currents than to the homomultimeric currents expressed by dShal in Xenopus oocytes. Similar γ-subunits could also explain the faster inactivation and altered activation range of Shal currents produced by co-expression of mammalian Shal channels with the 2–4 kb fraction of brain mRNA (Chabala et al., 1993; Serodio et al., 1994). Significantly, jShalγ1 forms functional heteromultimers with both dShal and mShal. This shows that the ability of Shal α-subunits to co-assemble with Shal γ-subunits has been conserved throughout metazoan evolution. Thus, jShalγ1 may represent a conserved molecular mechanism for regulating neuronal firing rate.
This research was supported by a grant from National Institutes of Health and the MDA to L.B.S. We especially thank Dr. Warren Gallin (University of Alberta) and Dr. Andrew Spencer (Bamfield Marine Station and University of Alberta) for providing the Polyorchislibraries and genomic DNA. We also thank Alice Butler, Michael Pak, and Jim Ray for their work in determining intron positions in dShal and mShal.
Correspondence should be addressed to Prof. Lawrence B. Salkoff, Department of Anatomy and Neurobiology, Washington University School of Medicine, 660 South Euclid Avenue, P.O. Box 8108, St. Louis, MO 63110.
Dr. Jegla’s current address: Department of Molecular and Cellular Physiology, Stanford University School of Medicine, Palo Atlo, CA 94305.