cDNA Cloning and Functional Characterization of the Mouse Ca2+-gated K+ Channel, mIK1

We have cloned from murine erythroleukemia (MEL) cells, thymus, and stomach the cDNA encoding the Ca2+-gated K+ (KCa) channel, mIK1, the mouse homolog of hIK1 (Ishii, T. M., Silvia, C., Hirschberg, B., Bond, C. T., Adelman, J. P., and Maylie, J. (1997) Proc. Natl. Acad. Sci.(U. S. A. 94, 11651–11656). mIK1 mRNA was detected at varied levels in many tissue types. mIK1 KCa channel activity expressed inXenopus oocytes closely resembled the Kca of red cells (Gardos channel) and MEL cells in its single channel conductance, lack of voltage-sensitivity of activation, inward rectification, and Ca2+ concentration dependence. mIK1 also resembled the erythroid channel in its pharmacological properties, mediating whole cell and unitary currents sensitive to low nm concentrations of both clotrimazole (CLT) and its des-imidazolyl metabolite, 2-chlorophenyl-bisphenyl-methanol, and to low nm concentrations of iodocharybdotoxin. Whereas control oocytes subjected to hypotonic swelling remained swollen, mIK1 expression conferred on oocytes a novel, Ca2+-dependent, CLT-sensitive regulatory volume decrease response. Hypotonic swelling of voltage-clamped mIK1-expressing oocytes increased outward currents that were Ca2+-dependent, CLT-sensitive, and reversed near the K+ equilibrium potential. mIK1 mRNA levels in ES cells increased steadily during erythroid differentiation in culture, in contrast to other KCa mRNAs examined. Low nanomolar concentrations of CLT inhibited proliferation and erythroid differentiation of peripheral blood stem cells in liquid culture.

. mIK1 mRNA was detected at varied levels in many tissue types. mIK1 K Ca channel activity expressed in Xenopus oocytes closely resembled the K ca of red cells (Gardos channel) and MEL cells in its single channel conductance, lack of voltage-sensitivity of activation, inward rectification, and Ca 2؉ concentration dependence. mIK1 also resembled the erythroid channel in its pharmacological properties, mediating whole cell and unitary currents sensitive to low nM concentrations of both clotrimazole (CLT) and its des-imidazolyl metabolite, 2-chlorophenyl-bisphenyl-methanol, and to low nM concentrations of iodocharybdotoxin. Whereas control oocytes subjected to hypotonic swelling remained swollen, mIK1 expression conferred on oocytes a novel, Ca 2؉ -dependent, CLT-sensitive regulatory volume decrease response. Hypotonic swelling of voltage-clamped mIK1-expressing oocytes increased outward currents that were Ca 2؉ -dependent, CLT-sensitive, and reversed near the K ؉ equilibrium potential. mIK1 mRNA levels in ES cells increased steadily during erythroid differentiation in culture, in contrast to other K Ca mRNAs examined. Low nanomolar concentrations of CLT inhibited proliferation and erythroid differentiation of peripheral blood stem cells in liquid culture.
Terminal differentiation of erythroid precursor cells is marked by enucleation and reduction in cell volume. A major component of cell volume reduction is achieved by reduction of cell K ϩ content. Mature, circulating erythrocytes retain two major ion transport pathways mediating K ϩ efflux (1). These are: 1) electroneutral K-Cl cotransport and 2) a voltage-insensitive, Ca 2ϩ -activated potassium (K ϩ ) channel of intermediate conductance (2)(3)(4), also known as the Gardos channel (5). The Gardos channel is thought to play a major role in volume regulation in normal (6) and sickle (SS) 1 human erythrocytes (7,8).
Especially in the chronically hypoxic environment of adherent or trapped sickle cells, the Gardos channel appears to mediate the major component of K ϩ loss from the erythrocyte (9), leading to an increased concentration of intracellular hemoglobin S, and exponentially decreasing the lag time for accelerated hemoglobin S polymerization (10). The Gardos channel's biophysical and pharmacological properties have been characterized in excised inside-out human red cell membrane patches, in which Ca 2ϩ -activated K (K Ca ) currents show inwardly rectifying properties with a unitary slope conductance ranging from 15 to 40 picosiemens, depending on the ionic conditions used (11)(12)(13). The channel is sensitive to block by charybdotoxin (14 -16), but insensitive to the SK channel blocker, apamin, and to the K ATP channel blockers, the antihypoglycemic drugs (17). 2 Sickle cell disease is a lifelong illness in which severity varies for poorly understood reasons not only among but within families, and even during an individual patient's clinical course. It is very likely that the pathological consequences of the autosomal recessive hemoglobin S mutation underlying sickle cell disease (18) are influenced by the polypeptide products of many yet undefined modifier genes. The central role of the erythroid K Ca channel in SS red cell dehydration has suggested it as a strong candidate modifier gene in sickle cell disease. We have hypothesized that K Ca channel blockade could serve as a useful adjunct therapy of sickle cell disease (7,8,16,19).
A subset of antifungal imidazole drugs was found potently to inhibit K ϩ and Rb ϩ flux in normal and SS human red blood cells. Clotrimazole was the most potent of those tested (7,20), blocking calcium ionophore A23187-induced 86 Rb ϩ influx and displacing 125 I-ChTX binding to red cells with equivalent ID 50 values of ϳ30 nM. The combined results of 86 Rb flux and 125 I-ChTX binding studies led to the proposal that CLT inhibited Ca 2ϩ -activated K transport by direct binding to the external surface of the Gardos channel (7), in contrast to earlier (20) and continuing suggestions (21) that CLT blocks K ϩ conductance via its inhibitory effects on cytochrome P450 enzymes. We subsequently showed that an inwardly rectifying K Ca channel from murine erythroleukemia (MEL) cells (22) was inhibited directly not only by charybdotoxin and CLT, but also by the major des-imidazolyl metabolite of CLT, 2-chlorophenyl-bisphenyl-methanol, incapable of inhibiting cytochromes P-450 (16).
The potency of CLT blockade of the erythroid K Ca channel and the status of CLT as a drug already long in clinical use for other indications recommended consideration of CLT as an erythroid K Ca channel blocker for adjunct therapy of sickle cell disease (19). Indeed, oral administration of CLT was shown to inhibit the erythroid K Ca channel in vivo, and to diminish formation of dehydrated dense cells in a mouse model of sickle cell disease (23) and in a phase I clinical trial with sickle cell disease patients (8). More recently, a complementary approach to prevention of red cell dehydration in patients with sickle cell disease via inhibition of K-Cl cotransport (24) provided the first preliminary evidence for long term clinical benefit in a group of less stable patients (25), suggesting that therapeutic inhibition of the erythroid K Ca channel might yield similar (and possibly additive) clinical benefit.
In the present study, we have cloned the cDNA encoding mIK1, the mouse homolog of hIK1 (28 -30), and report the tissue distribution of mIK1 mRNA, and the lack of N-glycosylation of in vitro translated mIK1 polypeptide. We have functionally expressed mIK1 in Xenopus oocytes, and studied its biophysical and pharmacological properties at the levels of whole cell and unitary currents. We have shown that hypotonic swelling of oocytes leads to activation of mIK1 currents and to the novel volume regulatory response of regulatory volume decrease (RVD). Both the currents and the volume regulation are inhibited by CLT and require Ca 2ϩ . We have also investigated the developmental profiles of mIK1 and other K ϩ channel mRNAs in ES cells undergoing hematopoietic differentiation along the erythroid lineage. Finally, we have demonstrated that 10 nM CLT retards erythroid differentiation of human peripheral blood stem cells, consistent with a role of IK1 in this process.

MATERIALS AND METHODS
Inhibitors and Chemicals-CLT was purchased from Sigma. The CLT metabolite, 2-chlorophenyl-bisphenyl-methanol, was the kind gift of R. Lombardy (Pharm-Eco Laboratories, Lexington, MA). Z1-Iodo-ChTX was synthesized by the MIT Biopolymers Facility, using N-Butoxycarbonyl-O-bromobenzyl-3-mono-iodo-L-tyrosine (Peninsula Laboratories, Belmont, CA) in place of tyrosine in position 36. The final product used was purified by high performance liquid chromatography, confirmed by amino acid analysis, and by inhibition of A23187-induced 86 Rb influx into normal human red cells (15,17). All salts were of analytical grade.
Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR)-Total RNA from MEL cells and mouse tissues (freshly resected kidney, stomach, spleen, distal colon, proximal colon, and epididymis) was prepared using the RNeasy kit (Qiagen, Chatsworth, CA). Total RNA from ES cells and ES-derived colonies was prepared using the RNAsol reagent (Biotecx, Houston, TX). A mouse tissue total RNA panel (liver, brain, thymus, heart, testis, ovary, and embryo) was purchased from Ambion (Austin, TX).
Reverse transcription was performed with the First Strand DNA Synthesis Kit (Ambion) using 1 g of total RNA. 5% of the reaction volume was used for hot start PCR in a total reaction volume of 50 l, using either Taq DNA polymerase (Qiagen), Expand High Fidelity PCR system (Boehringer Mannheim), or Taq DNA polymerase (Promega, Madison, WI) in the suppliers' recommended buffers.
PCR mixes lacking only primers were preheated at 82°C for 1 min, after which appropriate primers (Table I) were injected into the mix through mineral oil. The complete reaction mixes were denatured for 3 min at 95°C, then cycled through these conditions: denaturation for 45 s at 94°C, annealing for 2 min at 60°C (or, as indicated, in the presence of Q-solution (Qiagen), 52°C), and elongation for 2-3 min at 72°C. Final extension of 10 min at 72°C was terminated by rapid cooling to 4°C after the indicated number of cycles. PCR products were separated in 1% agarose gels for analysis and purified as necessary with the QIAquick Gel Extraction Kit (Qiagen). Control amplification experiments were performed on RNA samples in which reverse transcriptase was omitted.
DNA sequence analysis and data base searches were carried out with the GCG suite of programs (University of Wisconsin Genetics Computing Group, Madison, WI).
cDNA Cloning, Reconstruction, Transcription, and Translation of mIK1-With hIK1 sequence information (GenBank AF022150), four degenerate oligonucleotides were designed (Table I), two corresponding to N-and C-terminal peptides (IK1.F, IK1.R) and two to internal peptide sequences of low degeneracy (IK1.IF, IK1.IR). With the combined use of the Expand High Fidelity PCR system and of Q solution (Qiagen), 32 amplification cycles produced sufficient mIK1 cDNA for sequencing and cloning. 3 Overlapping PCR products for mIK1 were sequenced on both strands using an ABI 373 DNA Sequencer, then cloned into the "T-vector" plasmid, pCR 2.1 (Invitrogen, Carlsbad, CA).
Search of the EST data base identified 12 anonymous clones from mouse embryo, lymph node, melanoma, and cultured myotubes that encoded parts of mIK1: W30402, W45910, W79984, W82293, W99968, AA033013, AA042002, AA140298, AA185916, AA185547, AA265296, and AA592555. These clones helped verify nucleotide sequences encoding the N and C termini of mIK1 polypeptide. They also assisted in design of the primers IK1.5UTRF and IK1.3UTRR (Table I), used for PCR amplification of cDNAs from MEL cells and mouse stomach that encompassed mIK1 initiator and terminator codon sequences. All PCR products and their subclones were sequenced on both strands.
In vitro translation of mIK1 from capped cRNA in the presence of 3 Not only these and other degenerate primers, but also perfectly matched hIK1 primers failed to amplify hIK1 cDNA from a positive control tissue, human placenta (28), under a range of standard conditions. The same was true for many mIK1 primer pairs used with mouse cDNA. However, the addition of Q solution (Qiagen) to the PCR reaction mix allowed successful amplification of mIK1 fragments with multiple sets of primer pairs from many mouse tissue sources. The unusually high GC content of mIK1 (62% overall, with clusters up to 82% of 79 nucleotides) is likely related to these observations. Tran 35 S-Label (ICN, Costa Mesa, CA) was performed with the rabbit reticulocyte lysate (nuclease-treated) system (Promega) in the presence or absence of canine pancreatic microsomal membranes (Promega), according to the manufacturer's protocol. N-Deglycosylation of in vitrotranslated polypeptide with peptidyl-N-glycosidase F (PNGase F, Promega), SDS-polyacrylamide gel electrophoresis, and autoradiography were as described previously (32,33).
Expression of mIK1 in Xenopus Oocytes-Female Xenopus laevis were purchased from NASCO (Madison, WI), maintained at room temperature in running distilled water, and fed with frog brittle (NASCO). Oocytes were manually defolliculated after collagenase digestion of ovarian segments (34), then were microinjected with water or with 10 ng mIK1 cRNA and incubated at 19°C for 2-7 days in ND-96 with 2.5 mM sodium pyruvate and 500 g/ml gentamicin. ND-96 contained (in mM): 96 NaCl, 2 KCl, 1.8 MgCl 2 , 1 CaCl 2 , and 5 HEPES hemisodium, pH 7.40. All experiments with oocytes were conducted at 21°C. Two-microelectrode Studies-Oocytes injected 2-7 days previously with cRNA or with water were placed in a 1-ml chamber (Model RC-11, Warner Instrument Corp, Hamden, CT) on the stage of a dissecting microscope and impaled with microelectrodes under direct view. Current-injecting and potential-sensing microelectrodes were pulled from borosilicate glass with a Narishige puller, filled with 3 M KCl, and had resistances of 2-3 megohms. They were used with a Geneclamp 500 amplifier (Axon Instruments, Foster City, CA) interfaced to a Phoenix 386 computer with a TL-1 AD/DA board (Axon Instruments). PCLAMP 6.0.3 software was used to control data acquisition and to analyze data (Axon Instruments, Burlingame, CA). The isotonic ND-96 solution was 212 mosM; hypotonic bath solution (ND-70) was made by reducing [NaCl] from 96 to 70 mM, yielding an osmolarity of 160 mosM. Membrane potentials and currents were expressed as means Ϯ S.E.
Single-channel currents were recorded using standard techniques (35). All voltages refer to the cell interior referenced to the patch 5Ј-GAAGCTGGCTGAGCCCCAAGAC-3Ј Designed from hIK1 amino acid sequence, found to amplify mouse cDNA. b From EST clones allowing extension in both 5Ј-and 3Ј-untranslated regions of mIK1. c Degenerate oligonucleotide designed to amplify mIRK1, mIRK2, and mIRK3. d Designed from rat sequences, found to amplify mouse cDNA. e EST clone AA288411 contains C terminus and part of 3Ј-untranslated region of mSK2.
FIG. 1. Alignment of deduced amino acid sequences of mIK1 and hIK1. Transmembrane spans (S1-S6) and the pore (P) region are indicated above the sequences. Underlined residues are those assigned to transmembrane regions by all reports (28 -30); boldface residues not underlined are those so assigned by at least one report. Boldface italics show putative leucine zipper sequences.
pipette. Currents were measured with a 10-gigohm headstage, low pass-filtered at 1 kHz (Axopatch 1-D, Axon Instruments, Burlingame, CA), digitized at 5 kHz, and stored on the hard drive of an Hewlett Packard 486-DX computer. PCLAMP 6.0.3 software was used to control data acquisition via a Digidata 1200 interface and to analyze data (Axon Instruments, Burlingame, CA). Data were acquired continuously using FETCHEX subroutines.
The pipette and bath solution compositions used in studies of outside-out patches were as described (27). Inside-out or outside-out recordings were made with symmetrical solutions containing 116 mM potassium gluconate, 4 mM KCl, 10 mM HEPES, adjusted to pH 7.2 with KOH. The intracellular solution was supplemented with CaCl 2 to give a free calcium concentration of 5 M (assuming a stability constant of 15.9 M Ϫ1 for calcium gluconate). To obtain intracellular calcium concentrations below 1 M, 1 mM EGTA was added to the bath solution and CaCl 2 was added as calculated using published stability constants. After a cell-attached gigaseal patch was established, whole cell configuration was attained by gentle suction. The outside-out patch configuration was attained subsequently by slow withdrawal of the micropipette from the cell body. Cell-attached patches were excised to attain the inside-out configuration. A voltage ramp protocol (ϩ100 mV to Ϫ100 mV, duration 2600 ms) was used to test sensitivity to activation by calcium ([Ca 2ϩ ] i ) in inside-out patch configuration and sensitivity to inhibition by extracellular antagonists in the outside-out patch configuration. Relative effects of these agents (I/I o ) were quantitated at a membrane potential (V m ) of Ϫ100 mV. Absolute and relative currents were expressed as means Ϯ S.E. Values for inhibitor ID 50 were determined from best fit of the Langmuir isotherm or Hill equation; ED 50 and Hill coefficient for Ca 2ϩ were determined from best fit of the Hill equation (28).
Control recordings were made for the generation of I-V plots and the determination of   (36). 530 nm fluorescence emission images resulting from timed excitations at 495 nm were recorded at the equatorial plane of the oocyte. Acquired images were loaded into Image-1 software and replayed with enhanced intensity and contrast in contour display mode for measurement of oocyte diameter. Acquired images were calibrated to the cross hatch lines of a standard hemocytometer as described previously (36). Oocyte volume was calculated from measured diameter, assuming spherical geometry, and was expressed as mean % Ϯ S.D. of original volume in isotonic medium. Oocyte volume so calculated differed by Ͻ0.5% from that computed by the Image-1 software from the equatorial spheroid area.
Global estimates of oocyte intracellular calcium concentration ([Ca 2ϩ ] i ) were obtained with two ion-sensitive dyes. Oocytes loaded for 45 min with 5-10 M Fura-2-AM (Molecular Probes) were excited alternately at 340 and 380 nm (37,38). Excitation ratio images were collected at 15-90 s intervals at an emission wavelength of 510 nm. In vitro calibration of the Fura-2 free acid fluorescence ratio was performed (38) using a value for the K d of Ca 2ϩ binding to Fura-2 of 224 nM, and with R min and R max determined at 10 nM and 40 M free Ca 2ϩ , respectively.
Alternatively, oocytes injected with Calcium Green-Dextran 70 kDa (Molecular Probes) to a final intracellular concentration of 3.5-7 M were excited at 490 nm and imaged at 530 nm. Oocyte [Ca 2ϩ ] i was determined from relative fluorescence intensity, and calculated from the equation below. With both methods, the transition from isotonic to hypotonic medium was marked by small fluctuations in fluorescence ratio (Fura-2) or intensity (Calcium Green), consistent with previously reported spatial and temporal oscillations of [Ca 2ϩ ] i (39,40). However, these oscillations could not be clearly resolved over the large sampled areas within the nonconfocal equatorial plane. The reported values of [Ca 2ϩ ] i therefore represent averages of global (putatively cytosolic) free Ca 2ϩ concentrations, and were expressed as means Ϯ S.D.
Culture and Erythroid Differentiation of ES Cells-CCE murine embryonal stem cells maintained on primary mouse embryo fibroblasts were transferred onto gelatin-coated dishes in the presence of 100 U/ml leukemia inhibitory factor (Genetics Institute, Cambridge, MA) one day before a differentiation experiment. Two ES subclones, B2 and A20, were used (42). 4 In vitro differentiation of ES cells at a density of 3000 cells/ml was performed as described previously (42)(43)(44)(45) in 0.9% methyl cellulose, 400 units/ml interleukin-1, 100 units/ml interleukin-3, 2 units/ml erythropoietin, 50 ng/ml Kit ligand, 5 ϫ 10 Ϫ4 M monothioglycerol, and 20% fetal calf serum. The cell suspension was cultured in bacterial plates and incubated at 37°C in a humidified chamber with 5% CO 2 . After the indicated times of differentiation in culture, embryoid bodies were pooled from several dishes, rinsed several times in 150 mM NaCl, 20 mM sodium phosphate, pH 7.4, and harvested for RNA extraction.
Culture and Differentiation of Human Peripheral Blood Stem Cells-Two-stage erythroid cultures derived from normal human peripheral blood were set up as described by Fibach et al. (46,47). CLT was added at the beginning of the second phase of culture (upon addition of erythropoietin), on culture day 5. Cells were harvested and counted at the indicated times between culture days 9 and 19. Cell viability was determined by Trypan Blue exclusion. Cytocentrifuge preparations from the experiment of Fig. 10 were stained with May Grunwald's Giemsa and benzidine-HCl for erythroid developmental staging (46,47). In three additional experi- 4 The B2 and A20 ES cell clonal lines are Ϫ/Ϫ and Ϫ/ϩ, respectively, for the GDI-D4 gene. They are completely normal in adherence, spreading, proliferation, and erythroid differentiation in methylcellulose, as judged by colony-forming efficiency, histochemical indices of erythroid phenotype and hemoglobinization, embryonic and adult globin mRNA induction, and secondary erythroid colony formation. These cell lines are also normal in differentiation of megakaryocyte, mast cell, and myeloid lineages (42).  (Table I). Equal RNA loading was confirmed by RT-PCR of glyceraldehyde-3-phosphate dehydrogenase cDNA (25 cycles, not shown). B, 35 S-labeled mIK1 polypeptide translated from cRNA by rabbit reticulocyte lysate in the absence and presence of dog pancreatic microsomes. ments, staging assessed by morphological criteria in hematoxylin-eosinstained cytocentrifuge preparations gave similar results.

Molecular
Cloning of mIK1 cDNA-The mIK1 cDNA was cloned by homology-based RT-PCR supported by data base searches, as described under "Materials and Methods." The compiled mIK1 sequence has been assigned GenBank accession number AF042487. mIK1 nucleotide sequences from MEL cells, mouse thymus, and mouse stomach were identical.
The mIK1 amino acid sequence shares 88% identity with that of hIK1 (Fig. 1). The 425 amino acids of mIK1 conform to the topographical model of 6 transmembrane domains plus pore region proposed by Kohler et al. for the SK family of calciumactivated K channels (27), to which hIK1 (28 -30) and mIK1 are most closely related. The putative pore region of mIK1 is identical to that of hIK1, including the canonical signature sequence for K ϩ -selectivity, GYG (48). Indeed, only 1 amino acid distinguishes mIK1 and hIK1 between the putative exofacial ends of their S4 transmembrane helices until well beyond their endofacial ends of S6 (Fig. 1). The underlined N-glycosylation sequon between S5 and the P-region is also conserved.
Major regions of sequence divergence between mIK1 and hIK1 include S2, S3 and its adjacent second exofacial loop, and the putatively cytoplasmic C-terminal domain. Though the po-tential 5-leucine zipper is conserved in the divergent mIK1 C-terminal domain, the less extensive amino acid changes in the short, putatively cytoplasmic mIK1 N-terminal region do disrupt another potential leucine zipper (Fig. 1). mIK1 mRNA and Polypeptide-As has been previously noted for hIK1 mRNA (28), mIK1 mRNA was detected in thymus, colon, and stomach ( Fig. 2A). With the enhanced sensitivity of RT-PCR, mIK1 mRNA was also detected in kidney, liver, testis, ovary, heart, and brain. In addition, among tissues not tested in the human, mIK1 mRNA was present in whole embryo, epididymis, as well as in spleen, murine MEL cells ( Fig. 2A), and murine ES cells (Fig. 9A). mIK1 cRNA in vitro translated by reticulocyte lysate yielded a homogeneous polypeptide of ϳ40 kDa (Fig. 2B), somewhat lower than predicted from the calculated apoprotein molecular mass of 47,783 Da. The presence of dog pancreatic microsomes led to no increase in mass (Fig. 2B). Incubation of the membrane-associated mIK1 polypeptide with PNGase-F led to no reduction in M r for mIK1 polypeptide, under conditions in which the glycoprotein anion exchanger AE1 (33) was N-deglycosylated (data not shown). Thus, in vitro translation in the presence of microsomes provided no evidence for utilization of the N-glycosylation sequon in mIK1. This apparent absence of glycosylation may result from the sequon's distal proximity to the P-region, at most 6 amino acids in contrast to the 12 amino acids on either side required for optimal N-glycosylation (49).
Pharmacological Characterization of Recombinant mIK1 Expressed in Xenopus Oocytes-Xenopus oocytes were injected with water or with 10 ng of capped cRNA encoding mIK1, and examined 2-5 days later. mIK1 expression hyperpolarized resting membrane potential (V m ) measured ϳ3-5 min after impalement (Ϫ74 Ϯ 2 mV; n ϭ 49) compared with that of waterinjected oocytes (Ϫ37 Ϯ 1 mV; n ϭ 13, p Ͻ 0.0001, unpaired t test). This change was consistent with a shift toward E K of the oocyte V m normally dominated by Cl Ϫ conductance, and as predicted for increased expression of a K ϩ channel with some basal activity.
hIK1 (28 -30) and the murine Gardos (K Ca ) channel (16) were activated by Ca 2ϩ and inhibited by CLT. Therefore, we assessed the response to CaCl 2 injection into oocytes at a holding potential of Ϫ50 mV chosen to discriminate K ϩ from Cl Ϫ currents in the ND-96 bath (27,50). The peak currents elicited by CaCl 2 injection (arrows b) were outward in mIK1-expressing oocytes (ϩ600 Ϯ 139 nA, n ϭ 6; Fig. 3A) and significantly differed from the inward currents (Ϫ216 Ϯ 63 nA, n ϭ 4; Fig.  3B) recorded in control oocytes (p ϭ 0.002, unpaired t test). Similar results were observed in oocytes expressing hIK1 (data not shown). These results suggest that an outward K ϩ current developed in response to CaCl 2 injection in mIK1-expressing oocytes, likely greater in magnitude than the inward Cl Ϫ current stimulated in control oocytes. 5 The identity of the mIK1-associated K ϩ current elicited by Ca 2ϩ injection was further tested pharmacologically. The Gardos channel is potently inhibited by CLT (7,8,16). This property was subsequently demonstrated in outside-out patch for recombinant hIK1 (28) and confirmed in two-electrode voltage clamp recordings (data not shown). We have shown that inhibition of the Gardos channel by CLT-related drugs does not require inhibition of cytochrome P450 lipid oxidases (16,51). Therefore, 2-chlorophenyl-bisphenyl-methanol, the major in vivo des-imidazolyl metabolite of CLT (51), was tested for its 5 The whole cell currents shown in the protocol shown in Fig. 4 were recorded 2-3 days after cRNA injection. The larger currents shown in Fig. 8 were recorded from oocytes 2-7 days after cRNA injection. mIK1 current amplitude in the oocytes increased with time post-cRNA injection.
Inhibition of mIK1 by 2-chlorophenyl-bis-phenylmethanol was investigated further in outside-out patch recordings. As shown in Fig. 4A, inwardly rectifying currents were elicited by ramped voltages between ϩ100 and Ϫ100 mV. These currents were progressively inhibited by increasing concentrations of 2-chlorophenyl-bis-phenylmethanol to the bath, with an ID 50 value of 14 Ϯ 7 nM (Fig. 4B).
The Gardos channel is also inhibited by the scorpion venom component, charybdotoxin (ChTX, 16, 28). However, whereas the maxi-K channel of skeletal muscle (52) and whole cell K ϩ currents of lymphocytes (53) were much less potently inhibited by 125 I-ChTX than by ChTX, mIK1 was inhibited by synthetic iodo-ChTX with an ID 50 value of 4 Ϯ 3 nM (Fig. 4C; ID 50 was 9 Ϯ 5 nM from a Langmuir fit). This value was very similar to those with which both cold iodo-ChTX (data not shown) and 125 I-ChTX inhibited A23187-activated 86 Rb ϩ influx into red cells (15) and with which bound 125 I-ChTX was displaced from intact red cells by ChTX (15). Thus, mIK1-mediated channel activity in Xenopus oocytes was sensitive to inhibition not only by CLT, but also by its major des-imidazolyl metabolite, and not only to ChTX (data not shown) but also by iodo-ChTX. These results are fully consistent with the pharmacological properties of the erythroid Gardos channel.
Inside-out patches containing mIK1 were voltage-ramped from ϩ100 to Ϫ100 mV in the presence of 0.1, 0.2, 0.5, 1.0, and 10 M Ca 2ϩ . As shown in Fig. 5C, mIK1 was half-maximally activated by 158 Ϯ 8 nM free Ca 2ϩ , and the Hill coefficient for activation was 0.9 Ϯ 0.05. Thus, in voltage independence (see above) and in Ca 2ϩ sensitivity, mIK1 resembled the erythroid Gardos channel (3). These properties also resembled those of hIK1 (28) expressed in Xenopus oocytes (K 0.5 300 nM, Hill coefficient for activation by Ca 2ϩ of 1.7), but differed more substantially from the Hill coefficients of 2.7 and 3.2 determined for hIK1 in HEK293 cells (29,30).
mIK1 Expression Confers on Oocytes CLT-sensitive Regulatory Volume Decrease-Native oocytes lack the ability to respond to mild (36) or extreme hypotonic swelling (54) with RVD. As shown in Fig. 6A, three oocytes previously injected with water swelled upon exposure to mildly hypotonic medium, and failed to return to their isotonic volumes. Whereas the resting membrane potential of native oocytes is dominated by Cl Ϫ conductances, mIK1-expressing oocytes show a large shift in membrane potential toward E K . Therefore, we hypothesized that mIK1 expression might confer on oocytes a novel RVD response. Fig. 6B shows that mIK1-injected oocytes subjected to mild hypotonic swelling indeed recovered their original isotonic volume in the continued presence of hypotonic medium, but this RVD response was prevented when exposure to hypotonicity occurred in the presence of 10 M CLT (Fig. 6C). Moreover, mIK1-associated RVD was abrogated in BAPTA-AMloaded oocytes exposed to hypotonic medium containing EGTA (Fig. 6D). These results and similar results for oocytes expressing hIK1 are summarized in Table II, and demonstrate that IK1 expression conferred on oocytes a novel, Ca 2ϩ -dependent, CLT-sensitive RVD response. mIK1-mediated, CLT-inhibited, Ca 2ϩ -dependent K ϩ Conductance Is Activated by Hypotonic Swelling of Xenopus Oocytes-Hypotonic swelling activates an endogenous Cl Ϫ current in Xenopus oocytes (50). However, this activation (Fig. 7A) is not accompanied by RVD (Fig. 6A). We reasoned that RVD in IK1-expressing oocytes (Fig. 6B, Table II) likely was mediated by concomitant activation of heterologous IK1 in coordination with endogenous Cl Ϫ channels. The left panels of Fig. 7A show that hypotonic swelling of control oocytes previously injected with water led to enhanced current that displayed no inhibition by subsequently added CLT. As summarized in Fig. 7B, currents in control oocytes (n ϭ 9) measured at ϩ20 mV were ϩ88 Ϯ 14 nA in isotonic medium, increased to ϩ276 Ϯ 50 nA in hypotonic medium (p Ͻ 0.01), but were not reduced by addition of 10 M CLT (317 Ϯ 43 nA, p Ͼ 0.05, ANOVA).
In contrast, the right panels of Fig. 7A show that outward currents measured in isotonic medium in oocytes previously injected with mIK1 cRNA (n ϭ 30) were larger than in control oocytes (p Ͻ 0.03), were also activated by hypotonicity, and a Percent recovery to isotonic value from hypotonic V max , measured 45 min after transition from isotonic to hypotonic medium. *, p Ͻ 0.05; **, p Ͻ 0.01 compared to water-injected oocytes; ***, p Ͻ 0.01 compared to mIK1 cRNA-injected oocytes studied in the absence of clotrimazole, and p Ͼ 0.1 compared to water-injected oocytes.
FIG. 6. A, volume of water-injected oocytes during transition from isotonic to hypotonic medium, with no evident volume regulation. B, volume of five mIK1expressing oocytes during transition from isotonic to hypotonic medium, exhibiting RVD. C, volume of five mIK1-expressing oocytes during transition from isotonic to hypotonic medium in the presence of clotrimazole (10 M). RVD has been inhibited. D, volume of five mIK1-expressing oocytes loaded with BAPTA-AM and incubated in EGTA-containing medium (Ͻ10 nM [Ca 2ϩ ] o ) prior to and during the transition from isotonic to hypotonic medium. Hypotonic swelling is enhanced, and RVD is absent. displayed sensitivity to inhibition by clotrimazole. As summarized in the right panel of Fig. 7B, currents in mIK1 measured at ϩ20 mV increased from ϩ496 Ϯ 100 nA in isotonic medium to ϩ1157 Ϯ 269 nA in hypotonic medium (p Ͻ 0.001), then fell to ϩ657 Ϯ 144 nA upon exposure to CLT in hypotonic medium (p Ͻ 0.01, ANOVA). Thus, CLT inhibited (at ϩ20 mV) 75% of the mIK1-associated oocyte current elicited by hypotonic swelling. It is also evident in the right panels of Fig. 7 (A and B) that whereas hypotonicity reduced total inward current in mIK1expressing oocyte, inward current was restored upon addition of CLT, likely representing unmasking of the chloride current activated by hypotonic swelling. Fig. 7C plots the potential dependence of isotonic minus hypotonic difference currents and of hypotonic minus clotrimazole difference currents. Note that, whereas E rev for the hypotonicity-induced difference current was ϳϪ26 mV in control oocytes, E rev was ϳϪ80 mV in mIK1-expressing oocytes. Similarly, whereas the CLT-induced difference current was negligible in control oocytes, the substantial CLT-induced difference current in mIK1-expressing oocytes (Fig. 7C) displayed an E rev of Ϫ103 mV (n ϭ 30, r ϭ 0.99) and in hIK1-expressing oocytes (data not shown) an E rev of Ϫ99 mV (n ϭ 6, r ϭ 0.87). Both approximate the predicted E K of Ϫ96 mV, supporting the hypothesis of K ϩ selectivity for mIK1-mediated currents elicited by hypotonic swelling of oocytes. Fig. 8 demonstrates that the outward current activated in mIK1-expressing oocytes by hypotonicity requires Ca 2ϩ . mIK1expressing oocytes bathed in isotonic zero-Ca 2ϩ medium exhibited an E rev of Ϫ29 mV (n ϭ 5). This value changed minimally to Ϫ33 mV upon exposure to hypotonic in the continued absence of extracellular Ca 2ϩ , consistent with the activation of I clswell (50) and/or other Cl Ϫ conductances. However, subsequent addition of 2 mM Ca 2ϩ to the hypotonic extracellular medium elicited a substantial outward current characterized by an E rev of Ϫ88 mV (close to the predicted E K ) and by sensitivity to inhibition by 10 M CLT (data not shown). Thus, the CLT-sensitive outward current activated by hypotonic swelling of mIK1-expressing oocytes was characterized by a Ca 2ϩ requirement.
Hypotonic Swelling of Oocytes Increases Intracellular [Ca 2ϩ ] i -The ability of hypotonic swelling to activate the Ca 2ϩsensitive, voltage-independent mIK1 channel suggested that hypotonicity might also elevate [Ca 2ϩ ] i in the oocyte. The resting global [Ca 2ϩ ] i in isotonic medium as estimated by Fura-2 fluorescence ratio did not differ significantly (p Ͼ 0.1) between oocytes previously injected with water (106 Ϯ 12 nM) or with mIK1 cRNA (94 Ϯ 11 nM, n ϭ 5). Both control and mIK1expressing oocytes responded to 25 min hypotonic swelling with gradual increases in global [Ca 2ϩ ] i (7.4 -9.0% as reported by Calcium Green Dextran and 3.9 -4.5% as reported by FIG. 7. Hypotonic bath solutions activate CLT-sensitive mIK1 currents. A, two-electrode voltage clamp traces of hypotonicity-activated currents in control (left) and mIK1-expressing ocytes (right). Each oocyte was subjected to sequential I-V protocols in ND-96 (isotonic), hypotonic medium, and in hypotonic medium containing 10 M CLT for each oocyte. Hypotonic currents were sampled after 8-min exposure to hypotonic medium. CLT currents were sampled 2 min after exposure to CLT. Whereas control oocyte currents were mildly outwardly rectified and insensitive to CLT, mIK1-associated currents were strongly outwardly rectified and CLT-sensitive. B, current-voltage relationships for control (left, n ϭ 9) and mIK1-expressing oocytes (right, n ϭ 21) in isotonic ND-96 (open circles), hypotonic medium (closed circles), and hypotonic medium plus 10 M CLT (open triangles). Expression of mIK1 is associated with a negative shift of E rev of ϳ50 mV that is partially reversed by CLT. C, current-voltage relationships of difference currents in control (left) and mIK1-expressing oocytes (right). Expression of mIK1 is associated with a negative shift of E rev for hypotonic minus isotonic difference currents (open diamonds) of ϳ60 mV. Hypotonic minus CLT difference currents (closed diamonds) in mIK1-expressing oocytes are significant at resting potential, but negligible in control oocytes. Fura-2; p Ͻ 0.01 compared with isotonic values) that were indistinguishable in rate and magnitude.
Regulation of mRNAs Encoding mIK1 and Other K Channels during Erythroid Differentiation-mIK1 mRNA was almost as abundant in MEL cells as in spleen ( Fig. 2A), did not differ consistently among uninduced and dimethyl sulfoxide-induced MEL cells of different clonal origins (data not shown), and so agreed with our earlier report that CLT-sensitive K ϩ currents of intermediate conductance were present in both induced and uninduced MEL cells (16). However, even uninduced MEL cells are developmentally committed to the erythroid lineage (16,22). Therefore, to phenotype expression of mIK1 and other K ϩ channel mRNA expression in erythroid development, we turned to ES cells (42)(43)(44), comparing independent erythroid inductions in two clonal cell lines, B-2 and A-20 (Fig. 9).
The ES cells formed embryoid bodies of which over 80% matured into colonies containing hematopoietic cells detectable by phase contrast microscopy. We and others have previously shown that early markers of erythroid differentiation, such as GATA-1, are detected in embryoid bodies as early as day 3-4 after initiation of in vitro differentiation (42)(43)(44)(45). Expression patterns of the erythroid-specific later markers ␤-globin and AE1 (band 3 anion exchanger) were nearly identical in three inductions, evident at day 6 (accompanied by the onset of microscopically evident hemoglobinization) and increasing thereafter. The K ϩ channel mRNAs displayed uninduced levels and patterns of induction that varied among independent experiments to a greater degree than did those of the erythroid specific markers. Nonetheless, they displayed distinct patterns.
IK1 mRNA was present at low (B2-1) or very low level in uninduced cells (B2-2 and A20). After induction, however, IK1 mRNA levels consistently increased between days 6 and 9 after induction and remained elevated, as confirmed with two sets of primer pairs. This pattern contrasted with the relatively unchanging mRNA levels of ␤-actin (Fig. 9A) and glyceraldehyde-3-phosphate dehydrogenase (Fig. 9B) in uninduced and induced cells.
The pattern of increase in mIK1 mRNA during erythroid differentiation of ES cells was not paralleled by most other tested K ϩ channel mRNAs, many of which were detectable at low or very low level in ES cells (Fig. 9B). Among K Ca mRNAs, ␣Slo was present in uninduced cells, decreased as erythroidspecific gene expression was activated on day 6, and increased again by day 12. This biphasic pattern was shown with two sets of primer pairs. In contrast, SK1 and SK2 mRNAs were unchanged during the differentiation process (Fig. 9B), whereas SK3 (data not shown) was undetectable after 45 cycles amplification. Each K ca PCR amplimer presented in Fig. 9 encompassed exon-intron boundaries. In addition, no PCR products were amplified from reactions from which reverse transcriptase was omitted (data not shown). 6 Effect of the IK1 Inhibitor, CLT, on Stem Cell Proliferation and Erythroid Differentiation-Two-stage liquid cultures of human CD34 Ϫ /CD38 ϩ stem cells from peripheral blood were examined to test the effect of CLT on the second stage of culture in the presence of erythropoietin. As shown in Fig. 10A, increasing concentrations of CLT proportionately inhibited cell proliferation at all times tested. Cellular progression along the erythroid differentiation pathway was retarded in the presence of the lowest tested concentration of CLT, 10 nM (Fig. 10B). Cell proliferation was only slightly inhibited at this concentration of CLT at the times when retardation of developmental progression was most pronounced on culture days 13 (CLT day 8) and 16 (CLT day 11). DISCUSSION We have cloned the mIK1 intermediate conductance K Ca channel from MEL cells and from murine thymus and stomach (Fig. 1). mIK1 mRNA was widely but not ubiquitously expressed in mouse tissues (Fig. 2). mIK1 expressed in Xenopus oocytes mediated whole oocyte K ϩ -selective currents (Fig. 3), as well as inwardly rectifying, Ca 2ϩ -activated, voltage-insensitive, unitary K ϩ currents in inside-out patches (Fig. 5) similar to those previously reported for the erythroid K Ca (Gardos) channel (2)(3)(4)16) and for hIK1 (28). Inhibition of mIK1-mediated currents by CLT did not require the presence of the imidazole group responsible for cytochrome P450 inhibition by CLT. Moreover, inhibition of mIK1 was not attenuated by the substitution of 3-iodotyrosine for tyrosine at residue 36 of ChTX (Fig. 4), a critical residue for inhibition of skeletal muscle 6 Among K ir channel mRNAs (data not shown), ROMK was undetectable after 38 cycles amplification, and IRK1 (36 cycles) steadily increased from very low levels at day 0 to low levels at days 10 -12. In contrast, mRNA levels of IRK2 (33 cycles) and IRK3 (36 cycles) steadily decreased during the observed periods. Though reproduced in three ES lines through multiple amplifications, and with appropriate control RT-minus amplifications, these results are to be interpreted with caution. The amplified regions of these K ir cDNAs are encoded by genomic segments uninterrupted by introns. maxi-K channels as defined by systematic ChTX mutagenesis (55). Hypotonic swelling of oocytes activated mIK1-mediated Ca 2ϩ -dependent outward currents sensitive to CLT (Figs. 7 and 8) in parallel with small but significant increases in global [Ca 2ϩ ] i , and conferred on oocytes the novel property of Ca 2ϩdependent, CLT-sensitive RVD (Fig. 6, Table II). mIK1 was present at low levels in uninduced ES cells, and exhibited a sustained increase in abundance during erythroid differentiation of ES cells (Fig. 9). The ability of 10 nM CLT to retard erythroid differentiation (Fig. 10) is consistent with a physiological role for IK1 K Ca channel activity in this process.
Molecular Cloning and Tissue Distribution-Several features of the mIK1 sequence merit discussion. Three mIK1 EST clones from two cDNA libraries encode reliable 5Ј-untranslated sequence to nucleotide Ϫ98. Although this region lacks upstream in-frame terminator codons, the residues surrounding the initiator codon of our fully functional mIK1 cDNA constitute a favorable Kozak consensus initiation site. It is also pertinent that an in-frame terminator resides at nucleotide Ϫ252 in the hIK1 cDNA (30).
By analogy to other K ϩ channels, it is likely that IK1 forms a functional tetramer, and in the current study in Xenopus oocytes a homotetramer. The N-terminal cytoplasmic tail (56), among other channel regions (57), has been proposed to contribute to subunit oligomerization of the Kv channels. However, the lack of conservation of the candidate leucine zipper sequence in this region of mIK1 (L25 in hIK1 versus V25 in mIK1) suggests some other mechanism for oligomerization possibly mediated through this portion of the polypeptide. The conserved leucine zipper near IK1 C termini may suggest a possible role for this region in protomer oligomerization or in other protein-protein interactions. The S6 helix is remarkable for conservation of four Cys residues spaced so as to allow speculation of possible disulfide bonding within or between channel protomers. The extreme C termini of mIK1 and hIK1 are not conserved, and neither displays a consensus PZD binding domain (58). Although mIK1 displayed only 88% identity to hIK1, this lower than usual degree of identity between orthologous polypeptides from different mammalian species arises almost entirely from several delimited regions. The presence of IK1 mRNA in mouse colon (Fig. 2) and in T84 cells (data not shown), suggests that it may mediate or contribute to the ChTX-and CLT-sensitive K Ca current present in T84 cells (59,60) required for Ca 2ϩ -mediated secretagoguestimulated transepithelial chloride secretion (59 -61). The wide distribution of IK1 mRNA suggests other physiological functions for this channel, such as that suggested in lymphocyte blast transformation (30).
Relation between IK1 and the Erythroid K Ca (Gardos) Channel-The strong functional similarities between recombinant IK1 and the Gardos channel suggest that IK1 likely is or contributes to erythroid K Ca channel activity. This is especially true with respect to the pharmacological profile of the recombinant and native channels. However, the differences between recombinant and native function, including the more extreme inward rectification of mIK1, demand consideration of two possibilities. It is possible that mIK1 function is influenced by native oocyte K ϩ channels such as GIRK or I SK (minK), or that overexpression of mIK1 elicits atypical expression of endogenous oocyte channels (62). This possibility, as well as differences in kinase or phosphatase activities, may relate to the different apparent cooperativities of activation by Ca 2ϩ in 293 cells with Hill coefficients of 2.7-3.2 (29,30), compared with erythrocytes (3) and oocytes (28 and this work) with Hill coefficients of 0.9 -1.7. It is also possible that IK1 must interact with another erythroid channel subunit (of ␣ or ␤ type) to reconstitute with complete fidelity the native Gardos (39 -41) in the vicinity of plasmalemmal IK1 channels. The importance of juxtaplasmalemmal [Ca 2ϩ ] i in activation of IK1 is further suggested by its activation upon restoration of extracellular Ca 2ϩ (Fig. 9). In parallel with hypotonic activation of IK1, several endogenous oocyte Cl Ϫ channels may also be activated, including I Clswell (50) and two types of I ClCa (64). How might the combined activities of these channels lead to RVD?
Heterologous expression in the RVD-deficient T lymphoid cell line CTLL-2 of voltage-gated K ϩ channel Kv1.3 conferred the capacity for RVD (65), likely by complementing a limiting endogenous K ϩ conductance. RVD was elicited in Xenopus oocytes expressing mIK1 likely by similar complementation of the limiting endogenous K ϩ conductance. The net K ϩ efflux (3.77 nmol) required to achieve the measured 74% recovery from a 6% peak increase in oocyte volume (Table II) would require a time-averaged mean outward K ϩ current of 284 -331 nA, assuming [K ϩ ] i ϭ 91-104 mM in the maximally swollen oocyte and in the effluent, an operational K ϩ space ϭ 900 nl, and 20 min recovery time (Fig. 6). As shown in Fig. 7, mean mIK1associated outward CLT-sensitive current at Ϫ60 mV was 243 nA. Fig. 8 shows mean mIK1-associated outward Ca 2ϩ -dependent currents at Ϫ60 mV of Ϫ309 nA. Taken together with the concurrent ϳ100 nA of inward current attributable to I Clswellmediated Cl Ϫ efflux (50), and the increased electrical driving force for Cl Ϫ efflux in hyperpolarized mIK1-expressing oocytes, we conclude that net conductive flux of K ϩ plus Cl Ϫ in mIK1expressing oocytes could reasonably account for most or all of the observed RVD.
Red cell volume decrease following Gardos channel activation by prostaglandin E 2 also required parallel endogenous K ϩ and Cl Ϫ conductances, as suggested by inhibition not only by ChTX and CLT, but also by DIDS (66). However, though activation of the Gardos channel by A23187 or by PGE2 leads to volume decrease, erythrocytes effect RVD in response to hypotonic swelling via electroneutral K-Cl cotransport (67) rather than via the Gardos channel. Thus, the cellular machinery required to transduce to the IK1 channel the major hypotonic swelling signal (whether elevation in [Ca 2ϩ ] i or another signal) may be lacking in the human erythrocyte.
A recent report has suggested that a large increase in oocyte Cl Ϫ conductance secondary to expression and hyperpolarization-induced activation of the Cl Ϫ channel ClC-2 can also contribute to volume regulation in the voltage-clamped Xenopus oocyte, attenuating or abolishing oocyte swelling induced by extreme hypotonicity (68). However, the ability of ClC-2 to confer RVD was not tested in this study.
Possible Role of IK1 in Erythroid Differentiation-mIK1 mRNA increased in abundance during erythroid differentiation of ES cells (Fig. 9A). Other K Ca channel mRNAs tested either decreased or remained unchanged during ES cell erythroid differentiation (Fig. 9B). A possible requirement for or influence of IK1 in erythroid differentiation was tested by assessing the effect of increasing concentrations of CLT on the proliferation and erythroid differentiation of human peripheral blood CD34 Ϫ /CD38 ϩ stem cells. CLT inhibited stem cell proliferation with an ID 50 ϳ30 nM (Fig. 10A). The observed ID 50 of ϳ30 nM was Ͼ10-fold more potent than previously observed for inhibition by CLT of mitogen-stimulated [ 3 H]thymidine incorportation and intracellular Ca 2ϩ signaling in fibroblasts and tumor cells (69), but close to that observed for inhibition of IK1 (16, 28, and this work). Stem cell differentiation along the erythroid pathway was retarded by 10 nM CLT (Fig. 10B), a concentration at which inhibition of proliferation was modest. In addition, preliminary experiments indicate that 1 M CLT partially inhibits erythroid colony formation by murine ES cells. 7 IRK-like currents and unspecified IRK family mRNAs have also been implicated in the in vitro cytokine-driven expansion of human CD33 Ϫ /CD34 ϩ hemopoietic progenitor cells along both erythroid and myeloid pathways (70). IRK1 mRNA, was indeed present at very low levels in resting ES cells, and showed a modest, steady increase during erythroid differentiation (data not shown).
Conclusion-We have cloned the mIK1 K Ca channel, examined the tissue distribution of mIK1 mRNA, defined biophysi-cal and pharmacological characteristics of mIK1 expressed in Xenopus oocytes, determined profiles of mIK1 and other K ϩ channel transcripts in ES cells during erythroid differentiation, and shown that pharmacological inhibition of IK1 inhibits both proliferation and erythroid differentiation of stem cells. In its erythroid expression, Ca 2ϩ dependence, voltage independence, unitary conductance and rectification, pharmacological profile, and ability to mediate net volume decrease, mIK1 resembles the Gardos channel more closely than any other cloned K ϩ channel. Future experiments will test whether mIK1 is both necessary and sufficient to reconstitute Gardos channel function, and test more extensively the role of IK1 in erythroid differentiation. We have also shown that mIK1 can confer upon oocytes the novel response of RVD. Thus, in addition to its previously established utility in studying RVI (36), the oocyte will be a useful system for the more general study of cell volume regulation.