QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

HOME  |   SEARCH  |   ARCHIVE  |   SUBSCRIBE  |   CONTACT  |   HELP

This Article Abstract Full Text (PDF) Submit an eLetter Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (110) Citing Articles via Google Scholar Google Scholar Articles by Xia, X.-M. Articles by Lingle, C. J. Search for Related Content PubMed PubMed Citation Articles by Xia, X.-M. Articles by Lingle, C. J.

 Previous Article  |  Next Article 

The Journal of Neuroscience, July 1, 1999, 19(13):5255-5264

Molecular Basis for the Inactivation of Ca²⁺- and Voltage-Dependent BK Channels in Adrenal Chromaffin Cells and Rat Insulinoma Tumor Cells

Xiao-Ming Xia ¹, Jiu Ping Ding ¹, and Christopher J. Lingle ^{1, 2}

Departments of ¹ Anesthesiology and ² Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri 63110

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Large-conductance Ca²⁺- and voltage-dependent potassium (BK) channels exhibit functional diversity not explained by known splice variants of the single Slo -subunit. Here we describe an accessory subunit (3) with homology to other -subunits of BK channels that confers inactivation when it is coexpressed with Slo. Message encoding the 3 subunit is found in rat insulinoma tumor (RINm5f) cells and adrenal chromaffin cells, both of which express inactivating BK channels. Channels resulting from coexpression of Slo  and 3 subunits exhibit properties characteristic of native inactivating BK channels. Inactivation involves multiple cytosolic, trypsin-sensitive domains. The time constant of inactivation reaches a limiting value ~25-30 msec at Ca²⁺ of 10 µM and positive activation potentials. Unlike Shaker N-terminal inactivation, but like native inactivating BK channels, a cytosolic channel blocker does not compete with the native inactivation process. Finally, the 3 subunit confers a reduced sensitivity to charybdotoxin, as seen with native inactivating BK channels. Inactivation arises from the N terminal of the 3 subunit. Removal of the 3 N terminal (33 amino acids) abolishes inactivation, whereas the addition of the 3 N terminal onto the 1 subunit confers inactivation. The 3 subunit shares with the 1 subunit an ability to shift the range of voltages over which channels are activated at a given Ca²⁺. Thus, the -subunit family of BK channels regulates a number of critical aspects of BK channel phenotype, including inactivation and apparent Ca²⁺ sensitivity.

Key words: accessory subunits; K⁺ channels; BK channels; Ca²⁺- and voltage-gated K⁺ channels; mSlo channels; inactivation

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Large-conductance calcium (Ca²⁺) and voltage-gated K⁺ channels are widely expressed channels (McManus, 1991) that couple changes in submembrane Ca²⁺ concentration ([Ca²⁺]_i) to the regulation of electrical excitability. BK channels are formed from four -subunits (Shen et al., 1994) arising from the Slo gene product (Atkinson et al., 1991; Adelman et al., 1992; Butler et al., 1993). In addition, in some tissues, including smooth muscle (Knaus et al., 1994a,b) and cochlea (Ramanathan et al., 1999), an accessory -subunit can regulate BK channel gating profoundly. -Subunits appear to be responsible for the increased sensitivity of smooth muscle BK channels to Ca²⁺ (McManus et al., 1995) and may play a role in the frequency tuning of hair cells (Ramanathan et al., 1999). Although -subunits may account for some of the diversity of BK channels, the BK phenotype in many cells remains unexplained (McManus, 1991). For example, in adrenal chromaffin cells (Solaro and Lingle, 1992), pancreatic -cell lines (Li et al., 1999), and in the hippocampus (Hicks and Marrion, 1998) some BK channels exhibit inactivation. Furthermore, in Drosophila flight muscles (Salkoff, 1983) and perhaps in hair cells (Armstrong and Roberts, 1999), there are inactivating Ca²⁺-dependent currents probably arising from the Slo gene product. Thus, under conditions of sustained [Ca²⁺]_i and depolarization, such BK-type channels initially activate and then become silent. The inactivation of BK channels may play a role in the regulation of electrical excitability in these cells by altering the availability of BK channels for rapid repolarization after an action potential.

The most studied form of inactivation of BK channels has been in rat adrenal chromaffin cells (Solaro and Lingle, 1992; Solaro et al., 1995, 1997; Ding et al., 1998). The inactivation of chromaffin cell BK channels (BK_i channels) shares some features with N-terminal type inactivation of voltage-dependent K⁺ channels (Hoshi et al., 1990). As with the ShakerB channel (Hoshi et al., 1990; Gomez-Lagunas and Armstrong, 1995), BK_i inactivation arises from multiple trypsin-sensitive cytosolic domains (Ding et al., 1998). However, in contrast to the open channel-like blockade produced by the ShakerB N-terminal peptide (Choi et al., 1991; Demo and Yellen, 1991), cytosolic blockers of the BK channel pore do not compete with the native inactivation domain (Solaro et al., 1997). Previous work has failed to reveal a Slo -subunit splice variant that might account for the inactivating phenotype (Saito et al., 1997).

Here we describe an EST sequence that encodes a homolog of the previously identified BK -subunits [1, Knaus et al. (1994a); 2, Oberst et al. (1997)]. This new subunit (3) confers inactivation on BK channels, and message-encoding 3 is found in both chromaffin cells and rat insulinoma tumor (RINm5F) cells. We also show that, in terms of inactivation rates, dependency on voltage and Ca²⁺, lack of effect of cytosolic blockers, sensitivity to trypsin, and relative insensitivity to charybdotoxin (CTX), the channels arising from the coexpression of Slo and 3 subunits share key properties with native BK_i channels in chromaffin cells and RIN cells.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Isolation of cDNA clones. A novel member of the BK -subunit family, 3, was identified in the human EST database on the basis of homology with the known human 1 (Knaus et al., 1994b) and quail 2 (Oberst et al., 1997) subunits. The EST clone (accession number AA904191) containing the human 3 cDNA (accession number pending) then was obtained from a human lung neuroendocrine tumor library (Genome Systems, St. Louis, MO). The rat 3 homolog (accession number pending) was isolated from a RINm5F cDNA library (gift of Dr. Graeme Bell, University of Chicago, IL). The hybridization was performed in 1 M NaCl, 1% SDS, and 50% formamide at 37°C overnight and then washed with 2× SSC and 0.5% SDS. The screening membranes were exposed to x-ray film for 2 d. Initially, eight positive clones were identified in the RINm5F library. Of these, two purified positive cDNA clones contained full-length coding sequences of 1360 and 1042 bp, respectively. The DNA sequencing was performed with a BigDye Terminator Sequencing Kit (Applied Biosystems, Foster City, CA) on both strands. The -clones were all subcloned into pBF vector for in vitro RNA synthesis (Xia et al., 1998b).

Tissue distribution of BK 3 subunits. Northern blots were performed on membranes obtained from Clontech (Palo Alto, CA). Northern blots contained ~2 µg of poly(A⁺) RNA per lane from the indicated tissues. The loading of each lane was adjusted to contain an equal amount of human -actin per lane, based on a previous blot that used human -actin cDNA as a probe. The experimental procedure was as previously described (Xia et al., 1998a). Expression of r3 in adrenal and pancreatic tissue was determined by PCR and confirmed by sequence analysis. Total RNA from adrenal and pancreas was extracted and reversed-transcribed to cDNA (Promega, Madison, WI). Two primers (5' r3 5'-ACCATGACCTCCTGGACAAAAGG-3' and 3' r3 5'-TTCTGTGTGGTAGAGGAGGAGCT-3') were used for the PCR reaction (Taq polymerase, 32 cycles: 94°C for 30 sec, 52°C for 30 sec, and 72°C for 30 sec). Then the PCR products were subjected to automatic sequencing.

Expression constructs. The mSlo -subunit (accession number L16912; Butler et al., 1993) used in these experiments was identical to that used in previous work from this lab (mbr5; Wei et al., 1994; Saito et al., 1997). Specifically, we used a construct containing no insert at the mammalian splice site 2 (Tseng-Crank et al., 1994). RNA from rat adrenal chromaffin cells also contains other splice site 2 variants (Saito et al., 1997), including a 59 or 61 amino acid insert (Saito et al., 1997; Xie and McCobb, 1998; Li et al., 1999). The longer forms confer an ~20 mV leftward shift in the voltage dependence of gating at a given Ca²⁺. In preliminary experiments in which the 3 subunit was coexpressed with the Slo construct containing the 59 amino acid insert, the resulting channels exhibited inactivation. The mSlo construct was subcloned in pBScMXT.

To make the h1 construct for expression, we used two h1 specific oligonucleotides each with a XbaI or SalI tail (5'-ACTATCTAGACCCAGTGAATATGGTGAAGAAGCT-3' and 5'-TACTAGTCGACTGGCTCTACTTCTGGGCCGC-3') in a PCR reaction (pfu polymerase, Stratagene), using h1 (accession number U38907; the generous gift of the Merck Research Labs, West Point, PA) as the DNA template. Then the product was digested and subcloned in pBF vector (Xia et al., 1998b).

The N-terminal deletion [h3(1-33)] construct also was produced via PCR reaction (5'-GGAATTCTCTAAGATGAGGAAAACAGTCACAGCAC-3' and 5'-TACTAGTCGACAAAAATTATTTTATCCATTTTTG-3') and subcloned in pBF. To generate C43A189 [h3(1-43):h1], we applied two rounds of PCR. In the first-round PCR, reaction A used h1 as the DNA template with two primers: one was a specific primer of 3' h1 (5'-TACTAGTCGACTGGCTCTACTTCTGGGCCGC-3'), and the second one was a bipartisan primer of sense strand of h3 and h1 (5'-ACAGCACTGAAGGCAGGAGAGACACGAGCCCTTTG-3'). Reaction B used h3 as a template with two primers: one was a specific primer of 5' h3 (5'-AAGGAATCTAGACCCTGGACCAACATTCTCTAAG-3'), and the other one was another bipartisan primer of antisense of h1 and h3 (5'-CAAAGGGCTCGTGTCTCTCCTGCCTTCAGTGCTGT-3'). This latter primer is the complement of the previous bipartisan primer (pfu polymerase, 32 cycles, 96°C for 30 sec, 51°C for 45 sec, and 72°C for 1 min). In the second-round PCR, the products of first-round PCR were used as a template (equal volume mixture of reaction A and B), and the two specific primers of h1 and h3 (5' h3 primer and 3' h3 primer) were used as primers (pfu polymerase, 32 cycles, 96°C for 30 sec, 51°C for 45 sec, and 72°C for 1.5 min) (Tucker et al., 1996). Then the final product was digested with XbaI and SalI and subcloned in pBF. All of the constructs were verified by DNA sequencing.

Expression in Xenopus oocytes. In vitro transcription was performed to prepare ^M7GppGp-capped cRNA for oocyte injection. First the plasmids were linearized with an appropriate enzyme (SalI for mSlo and MluI for -constructs). T3 (mSlo construct; Wei et al., 1994) or Sp6 (all -related constructs) RNA polymerases were used for run-off transcription (Xia et al., 1998b). Total cRNA (0.1-0.5 ng) was injected into stage IV Xenopus oocytes that were harvested the day before. By weight, the ratio of - to -subunit RNA was 1:1 or 1:2, ensuring a large molar excess of -subunit RNA.

Electrophysiological recordings were performed after 1-7 d incubation of oocytes in ND-96 [containing (in mM) 96 NaCl, 2.0 KCl, 1.8 CaCl₂, 1.0 MgCl₂, and 5.0 HEPES, pH 7.5] supplemented with sodium pyruvate (2.5 mM), penicillin (100 U/ml), streptomycin (100 µg/ml), and gentamicin (50 µg/ml).

Electrophysiology. Currents were recorded in either inside-out or outside-out patches (Hamill et al., 1981). The generation of ensemble averages of detected channel openings (1-10 channels in patch) was done with patches from oocytes maintained for 1-3 d before recording. Patches used for G-V relations contained larger numbers of channels and were typically from oocytes maintained for 3-7 d. Currents typically were digitized at 10-50 kHz. Macroscopic records were filtered at 5 kHz (Bessel low-pass filter; 3 dB) during digitization. Single-channel records were filtered at 10 kHz (Bessel low-pass filter; 3 dB).

During seal formation the oocytes were bathed either in ND-96 or a frog Ringer's solution [containing (in mM) 115 NaCl, 2.5 KCl, 1.8 CaCl₂, and 10 HEPES, pH 7.4]. For inside-out patches the patches were excised and moved quickly into a flowing 0 Ca²⁺ solution. The pipette extracellular solution was (in mM) 140 potassium methanesulfonate, 20 KOH, 10 HEPES, and 2 MgCl₂, pH 7.0. Test solutions bathing the cytoplasmic face of the patch membrane contained (in mM) 140 potassium methanesulfonate, 20 KOH, 10 HEPES, pH 7.0, and one of the following: 5 mM EGTA (for nominally 0 Ca²⁺ and 1 µM Ca²⁺ solutions), 5 mM HEDTA (for 4 and 10 µM Ca²⁺ solutions, or no added Ca²⁺ buffer (for 60, 100, and 300 µM and 10 mM Ca²⁺ solutions). The methanesulfonate solutions were calibrated to be identical to a set of chloride-containing solutions, with free Ca²⁺ determined from a computer program (EGTAETC; E. McCleskey, Vollum Institute, Portland, OR). To accomplish this, we obtained the desired free Ca²⁺ by titrating the solution with calcium methanesulfonate until the electrode measurement of the methanesulfonate-based solution matched that of a chloride-based solution. The local perfusion of membrane patches was as described previously (Solaro and Lingle, 1992; Solaro et al., 1997).

Voltage commands and the acquisition of currents were accomplished with pClamp 7.0 for Windows (Axon Instruments, Foster City, CA). Currents were converted to conductances by using the measurement capabilities of the ClampFit program. For noninactivating currents the conductances were determined both from tail currents and from the peak current at the activation potential. Both estimates typically agreed within 5 mV. Ensemble averages were generated with custom software. Either null traces or traces obtained in 0 Ca²⁺ were used to generate an idealized leakage trace in the absence of channel openings. Then the idealized record was subtracted from records with channel openings before the detection and idealization of channel activity, using a 50% threshold detection procedure.

G-V curves for noninactivating variants were generated from both peak currents and tail currents with no significant difference in resulting fit values. G-V curves for inactivating variants were constructed from peak currents alone. Current values were measured by ClampFit (Axon Instruments), converted to conductances, and then fit with a nonlinear least-squares fitting program. Normalized G-V curves were fit with a Boltzmann equation with the form G(V) = {1 + exp[ (V  V_0.5)]/k}¹. G-V curves in Figure 8 display the mean and SEM of the values at a given voltage for a set of patches. Because of variability in the G-V curves from individual patches, this flattens the G-V curve for the populations. Therefore, values for the means of each set of individual G-V curves for each patch are provided also.

The onset and recovery from blockade by CTX were fit as described previously (Saito et al., 1997) assuming a simple, first-order bimolecular blocking reaction involving two parameters: f, the forward rate of block (in units of M/sec) and b, the unblocking rate in seconds. The K_D was calculated from b/f.

Experiments were done at room temperature (21-24°C). All salts and chemicals were obtained from Sigma-Aldrich (St. Louis, MO). Trypsin was applied at a concentration of ~0.25 mg/ml.

Removal of inactivation by trypsin. Predictions for the relationship between the apparent inactivation time constant (_i) and the fraction of total BK current that is noninactivating current (f_ss) were determined on the basis of the following assumptions (Ding et al., 1998). (1) Channels can have 0-4 inactivation domains. All channels in the population start with some average number, n, of inactivation domains per channel. (2) For the total BK channel population the number of inactivation domains per channel follows a binomial distribution. (3) A single inactivation domain is sufficient to produce inactivation, and each inactivation domain acts independently to produce channel inactivation, with the microscopic rate of inactivation of one inactivation domain being k_f (see MacKinnon et al., 1993; Ding et al., 1998). The value of f_ss allows for the use of the binomial distribution to calculate the fraction of channels of other stoichiometries. Then, for a given f_ss, a predicted macroscopic _i can be predicted for the population of channels. Specifically, to determine the predicted _i for a set of channels, we determined the contribution of channels of each stoichiometry from I_m(t) = A_m · exp(m · k_f · t), where m = 0-4 (the number of inactivation domains per channel) and A_m is the fraction of channels containing m inactivation domains. For the set of channels:

The resulting I(t) then was fit with a single exponential function to obtain the predicted _i for a given f_ss. Although the inactivation time course for a population of channels with different numbers of inactivation domains should exhibit multiple exponential components, in practice the predicted components are difficult to distinguish even with simulated currents. The solid lines in Figure 4 were determined from this procedure with the additional assumption that, for the cases of n = 3.2 and 3.6, the initial f_ss before the initiation of the digestion was indistinguishable from 0.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Identification of a new BK -subunit that confers inactivation on BK channels

In a search for novel genes that might contribute to the BK -subunit family, a gene database search was performed and a human EST was identified that shares partial homology with previously described BK -subunits from smooth muscle (1, Knaus et al., 1994a,b) and quail (2, Oberst et al., 1997) (Fig. 1). A cDNA clone containing the full-length coding region of human 3 (h3) was obtained from a human lung neuroendocrine tumor source (Genome Systems, St. Louis, MO). Subsequently, a rat homolog (r3) of h3 was isolated from a RINm5F cDNA library. The deduced amino acid sequences of both human and rat 3 contain 235 residues that are identical except for 10 amino acids. There are 43% identities and 63% similarities between h1 and human h3. At the N terminal 3 has no apparent signal peptide, and a hydrophilicity plot shows that it has two hydrophobic regions toward the N and C terminals, which appear to correspond to the proposed transmembrane segments for the 1 subunit (Knaus et al., 1994a) (Fig. 1B). There are three potential N-glycosylation sites located between the two hydrophobic regions of 3. The predicted topology of 3 is similar to 1, having two transmembrane segments with both N and C terminals inside the cell and a large extracellular loop that has several N-glycosylation sites. 3 shares with 1 and 2 the presence of four conserved cysteines thought to contribute to the formation of disulfide bridges in the extracellular loop of the subunit.

View larger version (88K): [in this window] [in a new window]   Figure 1.   A family of -subunits for the Slo Ca2+-dependent voltage-gated K+ channel. Top, Amino acid alignment of human 1 (h1), quail 2 (c2), human 3 (h3), and rat 3(r3) subunits. Darkly shaded residues are those with identity to aligned residues in at least one other -subunit family member. Lightly shaded residues are those in which rat and human 3 subunits show identity. The two predicted transmembrane segments, TM1 and TM2 (Knaus et al., 1994a), are marked with lines both above and below. Residues proposed to be involved in the CTX binding of h1 (amino acids 90-94; Hanner et al., 1997) are marked with a row of asterisks. Bottom, Hydrophilicity plot (Kyte-Doolittle algorithm) of h3 with a window of nine amino acids.

The expression pattern of 3 in rat and human tissues

Northern blot analysis shows that h3 is highly expressed in human kidney, heart, and brain. In kidney and heart two major mRNA products, 1.6 and 2.9 kb, were observed, whereas in brain only a 2.9 kb product was detected (Fig. 2B). Low-level expression of h3 mRNA products also was detected in human lung, small intestine, spleen, and colon. The relative expression level of 3 in rat tissues differed somewhat from human tissues. As in human tissue, 3 was abundant in brain and heart. However, 3 was expressed more strongly in rat lung and less so in rat kidney, in comparison to human tissue.

View larger version (68K): [in this window] [in a new window]   Figure 2.   Northern blot analysis of 3 subunit distribution. Membranes were probed with radiolabeled 3 sequence (h3 for human tissues and r3 for rat tissues). A, Rat tissues: 1, testis; 2, kidney; 3, skeletal muscle; 4, liver; 5, lung; 6, spleen; 7, brain; 8, heart. B, Human tissues: 1, peripheral blood leukocytes; 2, lung; 3, placenta; 4, small intestine; 5, liver; 6, kidney; 7, spleen; 8, thymus; 9, colon; 10, skeletal muscle; 11, heart; 12, brain. C, Human brain: 1, thalamus; 2, substantia nigra; 3, whole brain; 4, hippocampus; 5, corpus callosum; 6, caudate nucleus; 7, amygdala; 8, putamen; 9, temporal lobe; 10, frontal lobe; 11, occipital pole; 12, spinal cord; 13, medulla; 14, cerebral cortex; 15, cerebellum.

We also explicitly tested for the presence of 3 message in tissues known to express inactivating BK channels (Solaro and Lingle, 1992; Li et al., 1999). PCR reactions using substrate cDNA that was reverse-transcribed for rat adrenal and pancreatic tissue generated a single band of the expected size. DNA sequence analysis confirmed that the product in rat adrenals and pancreas is identical to r3 (data not shown).

Coexpression of human or rat 3 with mSlo -subunits confers inactivation on BK channels

Expression of either the human or rat 3 subunits with Slo -subunits in Xenopus oocytes results in BK-type channels that exhibit rapid inactivation after depolarizing voltage steps in the presence of constant submembrane Ca²⁺ (Fig. 3A). Channels exhibited the characteristic large single-channel conductance (~250-260 pS in symmetrical K⁺). Ensembles of channel openings arising from the coexpression of the Slo and 3 constructs exhibited an apparent inactivation time constant (_i) of ~25 msec at +100 mV and 10 µM Ca²⁺. Figure 3B plots ensemble current averages for the patch shown in A for 0, 1, 4, and 10 µM Ca²⁺. As Ca²⁺ is increased, the apparent _i becomes faster. Patches with a larger number of channels exhibited similar inactivation behavior (Fig. 3C for Slo with h3). Figure 3D displays normalized currents activated at different command potentials but with 10 µM Ca²⁺. Stronger membrane depolarization increases the apparent _i. Figure 3E plots the measured _i as a function of Ca²⁺ and voltage for currents generated by channels in patches from oocytes injected with Slo and human 3. The onset of inactivation appears to reach both a Ca²⁺ and voltage-independent limiting rate, as has been observed for native BK_i channels in chromaffin cells and RIN cells. Estimates of _i from ensembles of detected channel openings were indistinguishable from those shown in Figure 3E (data not shown). These properties of _i closely mirror the behavior of BK_i channels in RIN and chromaffin cells.

View larger version (32K): [in this window] [in a new window]   Figure 3.   Coexpression of the 3 subunit with the Slo -subunit produces inactivating BK channels. A, Traces show channel openings from an inside-out patch from a Xenopus oocyte injected with cRNA encoding both rat 3 and mouse Slo -subunits. Channels were activated by voltage steps to +100 mV, after a prepulse to 140 mV to remove inactivation, in the presence of 0, 1, or 10 µM Ca2+. Averages of openings for each condition are shown below the traces. B, Left, Ensemble-averaged currents are overlaid to emphasize differences in the amplitude and time course of current activation and inactivation for 0, 1, 4, and 10 µM Ca2+. B, Right, Currents obtained at 1, 4, and 10 µM were normalized to the peak current values to allow for a comparison of the time course of inactivation. Increases in Ca2+ result in a faster apparent i. C, Traces show macroscopic currents recorded from excised inside-out patches, with a larger number of channels resulting from the h3 construct coexpressed with mSlo. Activation was elicited at potentials from 100 through +100 mV with 0, 1, and 10 µM Ca2+. D, Sloh3 currents activated with 10 µM Ca2+ at the indicated voltages were normalized to the same amplitude to show that the apparent i becomes faster with increased depolarization. E, The i is plotted as a function of command potential for 1, 4, and 10 µM Ca2+. With sufficient depolarization a similar limiting i of ~25-30 msec is achieved for each [Ca2+].

Inactivation properties of Slo3 channels share key similarities with native BK_i channels found in RIN and chromaffin cells

We also wished to determine whether other properties of currents arising from Slo3 subunits are consistent with the known properties of BK_i channels in RIN and chromaffin cells.

Inactivation of BK_i channels in native cells involves multiple trypsin-sensitive inactivation domains (Ding et al., 1998). To examine this possibility, we applied trypsin (0.5 mg/ml) for 3-5 sec periods to inside-out patches, and current averages were generated from channels activated between trypsin applications. In Figure 4A, trypsin was applied to a patch containing a single inactivating channel. Inactivation was removed gradually by trypsin and eventually was abolished completely. Ensemble averages from patches with one or a small number of channels indicated that trypsin slowed _i, although stochastic fluctuations resulted in large variation in such estimates. When patches with larger numbers of channels were used (n = 4 patches), trypsin clearly resulted in the gradual removal of inactivation, in a manner consistent with the inactivation of Slo3 channels arising from multiple independent cytosolic inactivation domains per channel (Fig. 4B). For a given number of inactivation domains per channel, _i is predicted to change in accordance with the fraction of current that is noninactivating (see Materials and Methods) (MacKinnon et al., 1993). Such a relationship is plotted for one patch in Figure 4C, suggesting that the inactivation of Slo3 currents is consistent with four independent inactivation domains. As in previous work with native BK_i channels (Ding et al., 1998) and earlier work with ShakerB K⁺ channels (Gomez-Lagunas and Armstrong, 1995), with >50% removal of inactivation the estimates of prolongation are vulnerable to additional slow components of inactivation, resulting in deviation from the predicted relationships. The results in Figure 4 support the idea that multiple independently acting inactivation domains participate in the Slo3 inactivation process. The results also suggest a stoichiometry of four inactivation domains per channel. However, because the experimentally determined relationship between the fractional prolongation of _i and the fraction of noninactivating current (f_ss) is quite sensitive to rather small variability in the measured estimates of the apparent _i before the removal of inactivation, we feel that any conclusion concerning the stoichiometry of inactivation is only tentative at present.

View larger version (36K): [in this window] [in a new window]   Figure 4.   Selective digestion of the inactivation mechanism by trypsin applied to the cytosolic membrane face. A, Trypsin (0.5 mg/ml) was applied briefly to a patch containing one BKi channel resulting from the expression of h3 with mSlo -subunits. The cytosolic face of the patch was bathed with 10 µM Ca2+. The left column shows traces before trypsin exposure, the middle column shows traces after partial trypsin digestion, and the right column shows traces after removal of inactivation. Averages expressed as channel open probability (PO) from each set of sweeps are shown below the individual traces. B, Ensemble currents from a patch with a larger number of channels are displayed. Channels were activated with 10 µM Ca2+, with voltage steps to +100 mV. Trypsin was applied for 3-5 sec between the generation of each ensemble. As inactivation is removed, there is also a slowing in the apparent i. C, The fractional prolongation of i is plotted as a function of the fraction of peak current that does not inactivate by the end of each voltage step. When a limiting i of 28 msec is used, the experimental points closely follow the predictions for a population of channels, each of which initially contained four inactivation domains per channel.

Inactivation of native BK_i channels also exhibits one feature that distinguishes it from inactivation of the Shaker family of voltage-dependent K⁺ channels (Choi et al., 1991) and from inactivation of BK channels observed in hippocampal neurons (Hicks and Marrion, 1998). Specifically, cytosolic blockers of BK channels do not slow the native BK_i inactivation process (Solaro et al., 1997), indicative that occupancy of a site within the mouth of the ion permeation pathway does not hinder movement of the inactivation domains to their blocking sites. This contrasts with inactivation of BK channels in hippocampal neurons in which TEA and the ShakerB ball peptide have been shown to compete with the inactivation process (Hicks and Marrion, 1998). We therefore tested whether the rather bulky quaternary ammonium cytosolic blocker of BK channels, the lidocaine derivative QX-314, might hinder the Slo3 inactivation process. QX-314 produces a voltage-dependent reduction in BK single-channel current (Oda et al., 1992; Solaro et al., 1997) consistent with a rapid open-channel block mechanism in which QX-314 occupies a site within the ion permeation pathway. In Figure 5A, the cytosolic application of 2 mM QX-314 reduced the averaged peak current to ~24% of the control amplitude, having only small effects on _i. For a simple model in which QX-314 is proposed to compete with the native inactivation process, a reduction of current to 24% is predicted to produce a 4.17-fold prolongation of _i (Choi et al., 1991; Ding et al., 1998). For four patches, 2 mM QX-314 produced a 1.18 ± 0.07-fold prolongation of _i. Thus, the effect of QX-314 is inconsistent with a simple competitive scheme, indicative that inactivation conferred by the 3 subunit does not involve interaction with the site occupied by QX-314.

View larger version (15K): [in this window] [in a new window]   Figure 5.   A cytosolic blocker does not slow the inactivation of Sloh3 currents. A, Ensemble current averages obtained before, during, and after the application of 2 mM QX-314 to the cytosolic face of an inside-out patch expressing Sloh3 channels are plotted. Peak current was reduced to ~24% by QX-314. A simple competitive interaction between QX-314 and movement of the inactivation domain to its blocking site predicts an ~4.17-fold prolongation in current decay, indicated by the dotted line. B, Normalized currents are plotted to show that QX-314 appears to produce a slight prolongation of current decay. However, the magnitude of the prolongation is inconsistent with that expected for a simple competitive interaction with the inactivation domain and site occupied by QX-314.

The 3 subunit confers a reduced sensitivity to CTX

BK_i channels in chromaffin cells (Ding et al., 1998) and RIN cells (Li et al., 1999) exhibit a reduced sensitivity to the scorpion venom, CTX (Ding et al., 1998). Specifically, blockade of BK_i current by CTX exhibits a slower onset and slower recovery than were observed for noninactivating BK current in chromaffin cells (Ding et al., 1998). If incorporation of the 3 subunit accounts for the inactivating BK_i phenotype in these cells, it also must confer a reduced sensitivity to CTX. In Figure 6, the sensitivity of Slo, Slo1, and Slo3 channels is compared by using outside-out patches. When 5 mM TEA was applied in the same experiments, complete block and recovery from TEA blockade occurred within the first voltage step after the solution change, an interval of 10-20 sec. Thus, the time course of onset and recovery from block by CTX should reflect the kinetics of the CTX blocking reaction. With Slo expression alone, 20 nM CTX produces a relatively complete and rapid block of current. With coexpression of h1 (Fig. 6B2), blockade by 20 nM CTX was somewhat less effective at blocking in current, although 100 nM CTX produced an almost complete block. In contrast, an application of 100 nM CTX of duration comparable to that in Figure 6B2 produced only a slowly developing and incomplete block of the Sloh3 current (Fig. 6B3). Qualitatively, these results indicate that the 3 subunit appears to decrease the amount of block by CTX at a given concentration and to slow the rate of onset of block.

View larger version (34K): [in this window] [in a new window]   Figure 6.   The 3 subunit confers a reduced sensitivity to CTX on the Slo subunit. Currents were elicited in outside-out patches with 10 µM pipette Ca2+. Each patch was perfused with the concentrations of CTX indicated by the horizontal bars. For each patch 1 or 5 mM TEA also was applied separately to show that the onset and recovery from TEA block were rapid relative to the time course of CTX block. In A1-D1, raw currents obtained for control, CTX, and recovery salines are shown for mSlo alone (A), mSlo plus h1 (B), mSlo plus h3 (C), and mSlo plus 3(1-33) (D). In A2-D2, normalized peak current amplitudes from the patches shown above are plotted as a function of elapsed experimental time. The lines indicate the optimal fit to a first-order blocking reaction (see Materials and Methods) in which the forward rate of block (f, M/sec) and the unblocking rate (b, per sec) are the only free parameters. The KD values calculated from the fit rates are given on the graph, with mean values given in Results.

Blockade by CTX for BK channels has been described by a relatively simple bimolecular reaction (MacKinnon and Miller, 1988). With the assumption of this type of reaction, an estimate of the approximate K_D for block by CTX can be obtained from a simultaneous fit of the onset and recovery from block (see Materials and Methods), assuming that the time course of CTX concentration changes is sufficiently rapid relative to the time course of the blocking reactions. This procedure takes advantage not only of information available from the magnitude of blockade produced by one or two concentrations of CTX but also of the rates of the blocking reaction. With the use of this procedure, CTX blocks Slo currents with an IC₅₀ of 3.5 ± 0.6 nM (mean ± STD; n = 4 patches) and blocks Slo1 currents with an IC₅₀ of 13.9 ± 2.4 nM (n = 3). Blockade of Slo3 currents was more difficult to assess quantitatively, because washout from CTX blockade was difficult to achieve (Fig. 6C). This difficulty in achieving reversibility from CTX block of inactivating BK currents also was observed in chromaffin cells (Ding et al., 1998) and RIN cells (Li et al., 1999). Assuming a simple block model, in which recovery from block is expected to be complete, the example in Figure 6C yields an IC₅₀ of ~29 nM, although the data are not fully described by this model. With the use of this procedure for three cells with at least partial recovery from block, the IC₅₀ was 49.8 ± 26.9 nM. Because of the difficulty in achieving adequate recovery, this value certainly overestimates the sensitivity of the Slo3 currents to CTX. The difficulty in obtaining adequate recovery from CTX block of the Slo3 construct may indicate that a simple block scheme may not be fully adequate to describe the blocking mechanism. However, qualitatively, the slower onset of block and weaker blocking effect of CTX on the Slo3 currents relative to Slo alone or Slo1 are consistent with the weaker effects of CTX on BK_i currents in RIN and chromaffin cells (Ding et al., 1998). This suggests that the 3 subunit may account for the reduced CTX sensitivity of BK_i channels in these cells.

Two other studies also have reported the existence of BK channels with reduced sensitivity to CTX, but these other channels were noninactivating in nature. First, some BK-type channels studied in bilayers appear to lack sensitivity to CTX (Reinhart et al., 1989). Second, BK channels in peptide-secreting terminals of the posterior pituitary are CTX-resistant (Bielefeldt et al., 1992). Because these other CTX-resistant BK channels are noninactivating, it raises the possibility that additional, unidentified -subunits may contribute to the properties of these other channels.

The 3 N terminal is necessary and sufficient for inactivation

The additional N-terminal sequence of the 3 peptide suggested that this region was responsible for the inactivation behavior. To address this issue, we created a construct [3(1-33)] lacking the first 33 amino acids of the 3 sequence. Expression of 3(1-33) with Slo completely abolished any rapid inactivation of current (Fig. 7A). To verify that the 3(1-33) subunit actually was expressed and associated with Slo subunits, we examined the IC₅₀ for blockade of Slo3(1-33) by CTX and found it to be 44.5 ± 16.9 nM (n = 3) (see Fig. 6D), comparable to results with the intact 3 subunit but distinct from Slo alone. Thus, removal of the N terminal from the 3 subunit abolishes inactivation but maintains a change in CTX sensitivity similar to that produced by the intact 3 subunit.

View larger version (29K): [in this window] [in a new window]   Figure 7.   The N terminal of the 3 subunit is necessary and sufficient for inactivation. A, Left, Currents are shown resulting from the coexpression of mSlo with the h3(1-33) construct in which 33 amino acids were removed from the h3 N terminal. A, Right, Shown are currents resulting from the coexpression of mSlo with the h3(1-43):h1 construct in which the initial 43 amino acids of h3 replaced the first 12 amino acids at the N terminal of h1. In both cases the currents were activated by the indicated voltage steps with 10 µM Ca2+. No inactivation is observed with h3(1-33), whereas inactivation is observed with h3(1-43):h1.

To assess further the role of the 3 N terminal, we used amino acids 1-43 of the 3 subunit to replace the N-terminal 1-12 amino acids of the 1 subunit [h3(1-43/h1)]. As shown in Figure 7B, this addition of the 3 N terminal to the 1 subunit was sufficient to confer inactivation behavior on the 1 subunit. The apparent _i for this construct was indistinguishable from that of 3 (data not shown).

The 3 subunit shifts gating of Slo subunits

A major functional role of the previously described 1 and 2 subunits is to produce a shift in activation by Ca²⁺ to more negative membrane potentials (McManus et al., 1995; Dworetzky et al., 1996; Wallner et al., 1996; Saito et al., 1997; Ramanathan et al., 1999). Thus, smooth muscle BK channels or other BK channels containing a 1 subunit can be activated by elevations of submembrane Ca²⁺ even at membrane potentials near rest. In contrast, other BK channels in native cells appear to exhibit sensitivities to Ca²⁺ more similar to the Slo -subunit in the absence of any -subunit (McManus, 1991; Saito et al., 1997).

Given the important role of the 1 subunit in the regulation of gating of the -subunit, we examined the activation of current by 10 µM Ca²⁺ at various voltages for six different constructs: Slo alone, Sloh3, Slo(1-33)3, Slo[3(1-43)]1, Sloh1, and Slor3. For constructs that exhibited inactivation (Sloh3 and Slo[3(1-43)]1), peak currents were used to generate conductance-voltage (G-V) curves. We expect that estimates of G-V curves for inactivating currents that use the peak current will reduce the steepness of the apparent voltage dependence, thereby producing a somewhat more positive estimate of the V_0.5. However, the threshold for current activation should be similar with inactivation intact or removed. For the other constructs the G-V curves were generated from both peak and tail currents, with only small differences in the results. Figure 8 plots normalized G-V curves for four constructs. Clearly, both the 1 and 3 subunits result in a substantial shift in the voltage range over which the channel effectively gates with 10 µM Ca²⁺.

View larger version (35K): [in this window] [in a new window]   Figure 8.   The 3 subunit shifts gating of the mSlo -subunit. Normalized conductance-voltage curves were constructed from each of the indicated constructs by using 10 µM Ca2+ to activate the currents. G-V curves from individual patches were normalized separately and then averaged. This results in a flattening of the G-V curve for the averages. The fit values for the V0.5 for the grouped data for each construct included the following: for Slo alone, 24.5 ± 0.5 mV (± 90% confidence limit of fit value); for Slo plus h3, 28.5 ± 1.4 mV; for Slo plus 3(1-33), 21.4 ± 0.6 mV; and for Slo plus h1, 28.1 ± 0.9 mV. For comparison, the means of the V0.5 values from each individual patch included the following: for Slo alone, 25.8 ± 11.8 mV (mean ± STD; n = 5); for Slo plus h3, 32.6 ± 11.5 mV (n = 11); for Slo plus 3(1-33), 22.2 ± 19.3 mV (n = 4); and for Slo plus h1, 27.2 ± 18.0 mV (n = 9). For six patches with the rat 3 coexpressed with Slo, the V0.5 was 19.5 ± 7.8 mV. For nine patches for the 3(1-43):1 construct expressed with Slo, the V0.5 was 55.5 ± 10.2 mV.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The present results demonstrate that the coassembly of Slo -subunits with the 3 subunit is responsible for expression of inactivating BK channels in RIN and chromaffin cells. Furthermore, the presence of the 3 subunit in a number of other tissues, including brain, suggests a more general role of inactivating BK channels than previously realized. The existence of native BK channels with phenotypes unaccounted for by the known -subunits and known Slo splice variants suggests that additional accessory subunits may remain to be identified. For example, the inactivation of hippocampal BK channels (Hicks and Marrion, 1998) cannot be accounted for by the Slo splice variants and -subunits so far described. Similarly, CTX-resistant, but noninactivating, BK channels found in brain (Reinhart et al., 1989; Bielefeldt et al., 1992) would appear to require components other than those so far described. If so, the tissue-specific expression of -subunits might help to explain the large phenotypic diversity among native BK channels (McManus, 1991).

Our results indicate that the 3 subunit is present in a number of both neuronal and non-neuronal tissues. Although the inactivation of BK-type channels has been observed in a number of cell types [Drosophila muscle, Salkoff (1983); chromaffin cells, Solaro and Lingle (1992); RIN cells, Li et al. (1999); hippocampal neurons, Ikemoto et al. (1989) and Hicks and Marrion (1998); skeletal muscle, Pallotta (1985); frog hair cells, Armstrong and Roberts (1999)], it is unclear to what extent the 3 subunit plays a role in many of these cases.

In chromaffin cells and RIN cells the apparent _i of inactivation reaches a relatively voltage-independent and Ca²⁺-independent value at more positive voltages and more elevated Ca²⁺. In chromaffin cells this value is in the range of 25-40 msec (Solaro and Lingle, 1992; Prakriya et al., 1996). The variability in the chromaffin cells probably arises from the possibility that the BK channels may, on average, contain less than a full complement of four inactivation domains per channel (Ding et al., 1998). In fact, an implication of this latter study was that _i for channels containing four inactivation domains should approach ~25-28 msec. This is identical with the limiting _i observed here. In contrast, in hippocampus (Hicks and Marrion, 1998) and skeletal muscle (Pallotta, 1985) the inactivation of BK channels is markedly slower (_i approximately hundreds of milliseconds). In frog hair cells (Armstrong and Roberts, 1999) the inactivation of a Ca²⁺-dependent K⁺ current is faster (_i ~3 msec). Thus, the properties of the apparent _i observed for the Slo3 currents are identical to those of BK_i channels in chromaffin cells but differ from those observed in the other tissues.

The inability of QX-314 to hinder the Slo3 inactivation process is similar to the lack of effect of QX-314 on chromaffin cell (Solaro et al., 1997) and RIN cell (Li et al., 1999) BK_i inactivation. In comparison, tetraethylammonium and the Shaker ball peptide cause inactivated BK channels in hippocampal neurons to recover from inactivation (Hicks and Marrion, 1998), suggesting a competition between the inactivation domain and the blockers. This again suggests that the inactivation mechanism in the hippocampal cells differs from that resulting from the action of the 3 subunit. Thus, although the 3 subunit may account for BK inactivation in RIN and chromaffin cells, the molecular basis for inactivation in other tissues remains unknown.

One property of the Slo3 currents does vary somewhat from BK_i currents in chromaffin cells. Specifically, the voltage of half-activation at 10 µM Ca²⁺ for Slo3 channels is approximately 20 to 30 mV. This is negative to that reported for the activation of BK_i currents in native chromaffin cells by 10 µM Ca²⁺, which was ~3 mV (Prakriya et al., 1996). Although a difference in methodology might account for these differences, two other factors may contribute to this difference. First, the stoichiometry of the assembly of - and -subunits may influence the gating range of native BK channels. Second, there is evidence that the ability of -subunits to influence the gating range depends on the identity of the -subunit splice variant (Ramanathan et al., 1999). Either or both of these factors may contribute to differences in the gating range between native BK channels in chromaffin cells and channels arising from the coexpression of 3 and -subunits. The possibility that <4 -subunits may be associated with each BK channel in chromaffin cells is consistent with the observation that, on average, BK_i channels contain only two to three inactivation domains per channel (Ding et al., 1998). Furthermore, chromaffin cells express multiple -subunit splice variants (Saito et al., 1997), one of which may affect the ability of an associated -subunit to shift gating (Ramanathan et al., 1999). A complete answer to this issue will require a determination of the exact molecular composition(s) of BK channels in chromaffin cells.

The role of -subunits in defining CTX sensitivity

Despite the general acceptance of CTX as a blocker of BK channels, there is little detailed information about the effective IC₅₀ values for block in native tissues. This certainly stems in part from the challenges associated with the slow dissociation of CTX. Some studies also have described CTX-resistant BK channels (Reinhart et al., 1989; Bielefeldt et al., 1992). Our previous studies on chromaffin and RIN cells (Ding et al., 1998; Li et al., 1999) demonstrated a reduced CTX sensitivity (IC₅₀ ~50-100 nM) for BK_i channels relative to noninactivating BK channels (IC₅₀ ~2-4 nM). Although our results with CTX only provide a qualitative comparison of the effect of -subunits in defining the CTX sensitivity of different Slo channels, the similar resistance of the BK_i and Slo3 channels to CTX is consistent with the proposed role of 3 subunits in the formation of BK_i channels. Our results also suggest that there are differences between the 1 and 3 subunits in determining CTX sensitivity. Residues on the 1 subunit responsible for influencing the affinity of CTX for the Slo complex have been determined (Hanner et al., 1998). Surprisingly, the residues proposed to account for CTX interaction with the 1 subunit are identical in the 3 subunit. Thus, other differences between the 1 and 3 subunits must account for the differences in CTX sensitivity.

The mechanism of BK_i inactivation

The present results indicate unambiguously that the N-terminal domain of the 3 subunit is responsible for the inactivation of the Slo3 channels. Although inactivation of BK_i channels shares some features in common with N-terminal-mediated inactivation (Solaro et al., 1997), there are also clear differences. Unlike inactivation mediated by the N terminal of the ShakerB -subunit (Choi et al., 1991; Demo and Yellen, 1991), inactivation mediated by the tethered 3 N-terminal domain involves a binding site not influenced by occupancy of the ion permeation pathway by QX-314. Thus, the 3 N-terminal domain appears to block the BK channel at a position probably not homologous with the Shaker pore site acted on by the ShakerB N-terminal structures.

During the revision of this manuscript, a paper describing the identical human -subunit appeared (termed 2 in Wallner et al., 1999). Our results share a number of observations with this other study but differ on some interesting points. First, in the other work, Northern blots did not reveal the 3 subunit in adrenal tissue. In contrast, our results support the view that the 3 subunit can account specifically for BK channel inactivation in RIN cells and chromaffin cells. Specifically, we showed by PCR that the 3 subunit is found in pancreatic and chromaffin cell RNA and that the 3 subunit is present in a RIN cell library. It is possible that low message levels may result in the difficulty of detecting the 3 subunits by Northern blots. Second, our results indicate that QX-314 does not compete with the native inactivation process mediated by the 3 subunit, similar to the properties of BK_i channels in native cells (Solaro et al., 1997). However, Wallner et al. (1999) report that TEA does compete with a free 3 N-terminal ball peptide for blockade of the Slo channel. To address this potential discrepancy, we therefore have examined the ability of QX-314 to compete with a 26-amino-acid 3 N-terminal