Compensatory Anion Currents in Kv1.3 Channel-deficient Thymocytes*

Kv1.3 is a voltage-gated potassium channel with roles in human T cell activation/proliferation, cell-mediated cytotoxicity, and volume regulation and is thus a target for therapeutic control of T cell responses. Kv1.3 is also present in some mouse thymocyte subsets and splenocytes, but its role in the mouse is less well understood. We report the generation and characterization of Kv1.3-deficient (Kv1.3–/–) mice. In contrast to wild-type cells, the majority of Kv1.3–/– thymocytes had no detectable voltage-dependent potassium current, although RNA and protein for several potassium channel subunits were found in the thymocyte population. Surprisingly, the level of chloride current in the Kv1.3–/– thymocytes was increased approximately 50-fold over that in wild-type cells. There were no abnormalities in lymphocyte types or absolute numbers in thymus, spleen, and lymph nodes and no obvious defect in thymocyte apoptosis or T cell proliferation in the Kv1.3–/– animals. The compensatory effects of the enhanced chloride current may account for the apparent lack of immune system defects in Kv1.3–/–mice.

Studies with selective drug inhibitors have suggested Kv1.3 to be involved in naïve T cell activation and lymphoblast proliferation (5,6,(12)(13)(14), cell-mediated cytotoxicity (15,16), T cell volume regulation (6,17), and thymocyte development (14,18). A proposed common role for Kv1.3 in these functions is to maintain the conditions necessary for a sustained rise in intracellular Ca 2ϩ . Following stimulation of the T cell receptor, production of inositol 1,4,5-triphosphate causes a transient release of Ca 2ϩ from intracellular stores followed by a sustained Ca 2ϩ influx that is required for full activation (19,20). Kv1.3 channels help to maintain a negative membrane potential (Nernst potential for K ϩ is about Ϫ85 mV) and a large driving force for Ca 2ϩ entry through Ca 2ϩ -release-activated Ca 2ϩ channels (21).
We have now generated Kv1.3-deficient (Kv1.3 Ϫ/Ϫ ) mice to determine whether they have compromised thymocyte development or peripheral T cell activation. Despite an increase in levels of transcripts and protein for certain other potassium channel subunits (Kv1. 1 and Kv1.4) in the total Kv1.3 Ϫ/Ϫ thymocyte population, deletion of the Kv1.3 gene results in the complete loss of voltage-dependent potassium current in the majority of thymocytes. Surprisingly, this is accompanied by an approximately 50-fold increase in chloride conductance. Because chloride channels would be expected to maintain negative membrane potentials in lymphocytes, this compensatory change in membrane permeability may account for the apparent absence of abnormalities in the immune responses of Kv1.3 Ϫ/Ϫ animals.

MATERIALS AND METHODS
Generation of Kv1.3-deficient Mice-The Kv1.3 gene was isolated from a lambda Fix II 129Sv/J library (Stratagene) using a 5Ј region of the rat homologue (GenBank TM number m30441) as a probe. The identity of the clone was confirmed by restriction mapping and sequencing. The rat Kv1.3-coding sequence (GenBank TM number m30441) possesses particular PvuII and SacI sites that are unique among members of the Kv1 family, as well as ScaI (used in the targeting strategy), HincII, PstI, and SmaI sites.
The targeted region spans an 8.2-kbp region between an upstream BamHI site and the 3Ј-end of the lambda Fix II genomic clone (see Fig.  1). The BamHI site was subsequently eliminated by Klenow polymerase end-filling and re-ligation. The construct was then linearized with XhoI by partial digestion, and a herpes simplex virus thymidine kinase gene * This work was supported in part by grants from the Natural Sciences and Engineering Research Council (NSERC) (to L. C. S.), the Canadian Institutes for Health Research (Grant MT-13657), the National Institutes of Health (Grant DC-01919) (to L. K. K.), and the Howard Hughes Medical Institute (to R. A. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (22) was inserted into the vector XhoI site at the 3Ј-end of the targeted region after Klenow end-filling.
An XhoI/SalI neomycin resistance cassette from pMC1neopA (Stratagene) was then inserted into the XhoI site 5Ј of Kv1.3 in the opposite orientation to Kv1.3. This regenerates the XhoI site downstream of the neomycin resistance cassette (see Fig. 1). The 1.8-kbp XhoI/ScaI region of Kv1.3 was then excised, and the construct was re-ligated after Klenow end-filling. The left and right arms of the targeting construct are 4.5 and 1.8 kbp, respectively. The targeting vector was linearized at a NotI site (see Fig. 1), and 25 g was used to electroporate 10 7 W9.5 embryonic stem cells. Embryonic stem cells were then plated onto mitomycin C-treated embryonic fibroblasts, and drug selection was begun 24 h later with 2 M gancyclovir (Syntex) and 0.3 mg/ml G418 (Invitrogen).
Embryonic stem cell clones and mice were screened by BamHI-digest Southern blot analysis with probes X and Y (see Fig. 1). Probe X is a 1.5-kbp EcoRI region, and probe Y is a 0.5-kbp HincII/SalI fragment at the 3Ј-end of the genomic clone. Homologous recombinant embryonic stem cells were injected into C57BL/6 blastocysts and chimeric males were bred to C57BL/6 females. All mice were housed in specific pathogen-free conditions in accordance with institutional animal care and use guidelines.
Fluorocytometry-Cells were recovered into 10 ml of Bruff 's medium/5% fetal calf serum (FCS) 1 from thymus, spleen, and/or mesenteric lymph nodes by using the plunger of a syringe to tease the tissues between two pieces of 0.1-mm nylon mesh (Small Parts, Inc.). Splenocytes were centrifuged at 1000 rpm at 4°C for 5 min, re-suspended in 2 ml of erythroid cell lysis buffer (Biofluids), and then made up to 10 ml with PBS. All cell suspensions were then filtered through the 0.1-mm nylon mesh and centrifuged at 1000 rpm in a bench top centrifuge at 4°C for 5 min. Finally, cells were re-suspended in 10 ml of Bruff's/5% FCS for counting in a hemacytometer.
Thymocytes were then centrifuged at 1000 rpm for 5 min and resuspended in 0.1 ml of annexin V-staining buffer (BD Pharmingen) containing 1 g/ml 7-amino actinomycin D (7-AAD, Calbiochem), 20 g/ml annexin V-PE (BD Pharmingen), and 20 M 3,3Ј-dihexyloxacarbocyanine iodide (DiOC 6 , Molecular Probes). Samples were then left at room temperature in the dark for 20 min before being made up to 0.5 ml with annexin V-staining buffer for fluorocytometry, as above. Percentages of thymocytes that were "live" (7-AAD-low), annexin-V (Ann) positive or negative and DiOC 6 positive or negative were then determined from the fluorocytometry data and represented as mean percentage of thymocytes Ϯ S.D. from three mice per group. 7-AAD is a fluorescent DNA stain that allows discrimination of live cells from necrotic and lateapoptotic cells that have lost cell membrane integrity (23). Annexin V binds phosphatidylserine and as such is useful in fluorescent-conju- gated forms for visualizing aberrant phosphatidylserine exposure on the outer cell membrane surface that occurs early in apoptosis (24,25). DiOC 6 is a cationic, lipophilic fluorescent dye that is used to monitor changes in mitochondrial membrane integrity, which also constitutes an early event in apoptosis (26).
T Cell Stimulation Assays-Splenocytes were prepared in Bruff's/5% FCS as described above, and 0.2 ml of 5 ϫ 10 5 cell aliquots was placed in 96-well plates. Cells were cultured in a 37°C, 5% CO 2 incubator with or without 1-30 g/ml concanavalin A (Roche Applied Science) or 0.1 g/ml anti-CD3 (2C11). After 48 h, 1 Ci of [ 3 H]thymidine (PerkinElmer Life Sciences) was added to each well for a further 16 h. Thymidine incorporation was then determined with a Wallac plate cell harvester and scintillation counter.
RT-PCR-RNA was prepared from thymocytes and brain using TRIzol (Invitrogen) and subjected to DNase I digestion (0.1 unit/ml, 15 min, 37°C, Amersham Biosciences) to eliminate genomic contamination. First strand cDNA was synthesized as described (Amersham Biosciences) using an oligo(dT)-based primer. The cDNA was then used for PCR with gene-specific primers for several K ϩ and Cl Ϫ genes and the housekeeping gene, ␤-actin (see Table I for primer sequences and predicted product sizes; see also Ref. 6). Amplification was performed in a Minicycler (MJ Research) using 35 (for K ϩ or Cl Ϫ channel genes) or 25 cycles (for ␤-actin), with parameters of 94°C for 30 s, 52°C for 30 s, and 72°C for 45 s. Band densities were quantified using the Quantity One software supplied with the Gel-Doc system (Bio-Rad). The identities of PCR-amplified fragments of the predicted sizes were confirmed by sequencing.
Western Blot Analysis-Thymocytes or brain tissues were lysed in 0.5 ml of cold ionic detergent buffer consisting of 20 mM Tris at pH 8, 100 mM NaCl, 0.5% Igepal, 0.1% sodium dodecyl sulfate, and 1% deoxycholate. Following a 20-min incubation on ice, lysates were cleared by centrifuging at 15,000 ϫ g for 20 min at 4°C. Protein concentrations were measured using the DC Protein Assay (Bio-Rad). Proteins (25 g/lane) were resolved by SDS-PAGE, transferred to nitrocellulose, blocked in 5% nonfat milk and incubated overnight at 4°C with the following antibodies: Kv1 4-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) and 4,4Ј-diisothiocyanatostilbene-2,2Ј-disulfonic acid (DIDS), and the K ϩ -channel blocker tetraethylammonium (TEA). Agitoxin-2 (AgTx-2) was from Alomone Laboratories. Stock solutions of NPPB (500 mM) and DIDS (500 mM) were dissolved in Me 2 SO and stored at Ϫ20°C. When Me 2 SO was used as the solvent, the maximal final concentration in the bathing solution was about 0.2%. Thus, control recordings included 0.2% Me 2 SO in the bath solution. Stock solutions of AgTx-2 (1 M) and charybdotoxin (50 M) were made in standard bath medium (see below) supplemented with 0.1% bovine serum albumin (BSA, Sigma) to prevent adhesion to surfaces, and stored at Ϫ20°C. Control recordings were done by adding 0.1% BSA to the bath solution.
Patch-clamp Electrophysiology-Thymocytes were prepared as above and then washed five times in RPMI 1640/L-glutamine supplemented with 10% fetal calf serum (both from Invitrogen). Whole-cell currents were measured using an Axopatch 200 amplifier and pCLAMP version 6.04 software (both from Axon Instruments). Patch electrodes of resistance 8 -12 M⍀ were pulled from thick-walled borosilicate glass (World Precision Instruments). During data acquisition, capacitive currents were canceled by analog subtraction, and all currents were filtered at 2 kHz via the amplifier. Data analysis was performed using pCLAMP and Origin (version 5, Microcal Software), and curve fitting used the iterative Levenberg-Marquardt algorithm of non-linear regression. All recordings were made at room temperature (18 -21°C). All data were corrected for junction potentials between bath and pipette solutions.
The standard bathing solution contained (mM): 145 NaCl, 5 KCl, 1 MgCl 2 , 1 CaCl 2 , 5 HEPES, adjusted with NaOH to pH 7.4. The standard pipette solution contained (mM): 145 K aspartate, 1 K 4 BAPTA, 5 HEPES, 1 MgCl 2 , 0.09 CaCl 2 , 2 K 2 ATP, adjusted to pH 7.2 with KOH. The resulting low Ca 2ϩ (10 nM) pipette solution allowed Cl Ϫ currents and voltage-gated K ϩ currents to be monitored without contaminating Ca 2ϩ -activated currents. Fresh K 2 ATP (Sigma) was added to pipette solutions just before use to help maintain channel activity during whole-cell recording. Occasionally, to isolate K ϩ currents, Cl Ϫ currents were reduced by substituting aspartate for most of the Cl Ϫ in the bath.

Generation of Kv1.3-deficient Mice
were generated by homologous recombination in embryonic stem cells using conventional techniques (Fig. 1). Our strategy deletes a 1.8-kbp region of Kv1.3 that includes the promoter/transcriptional start site and a 5Ј-region of the coding sequence (see "Materials and Methods"). Kv1.3 Ϫ/Ϫ mice were initially obtained by interbreeding Kv1.3 ϩ/Ϫ mice. Kv1.3 Ϫ/Ϫ mice were born at a normal Mendelian ratio, and subsequent interbreeding of Kv1.3 Ϫ/Ϫ mice did not reveal any obvious lack of breeding or reduced litter sizes.
Whole-cell Currents in Kv1.3 ϩ/ϩ and Kv1.3 Ϫ/Ϫ Thymocytes-Thymocyte currents were recorded in the whole-cell patch clamp configuration. Most recordings were expected to be from CD4 ϩ CD8 ϩ thymocytes, because they represent ϳ85% of thymocytes in both Kv1.3 ϩ/ϩ mice and Kv1.3 Ϫ/Ϫ mice (see later and Table III below). Consistent with a previous report that CD4 ϩ CD8 ϩ thymocytes in Kv1.3 ϩ/ϩ mice have a large n-type current (11), the predominant current in Kv1.3 ϩ/ϩ thymocytes strongly resembled Kv1.3 ( Fig. 2A). Biophysical features in common with Kv1.3 (1, 30 -32) include time-dependent activation with depolarization at about Ϫ30 mV, inactivation during prolonged depolarizing steps, and use-dependent inactivation. Although we did not examine details of kinetics and voltage dependence of the Kv current, about 30% inactivation was seen (e.g. Fig. 2A) with voltage pulses of 200-ms duration (which are relatively short). Faster inactivation has been observed for the Kv1.3 current (formerly called "n"-type); e.g. a time constant of ϳ100 ms (at ϩ40 mV) in mouse thymocytes and ϳ180 ms in human T cells (8,9). However, Kv1.3 inactivation is affected by many factors, including temperature, internal and external divalent ions, and the type of internal anion. In particular, intracellular fluoride was previously used (8,9), and this accelerates inactivation (33) compared with aspartate, which we used in the present study. Also consistent with Kv1.3, there was complete block by AgTx-2, a scorpion toxin that blocks Kv1.3 at very low concentrations (K d ϳ 4 pM (34)). We found that 2 nM (not shown) and 5 nM AgTx-2 ( Fig. 2A, middle panel) blocked the time-dependent current completely. The current remaining after AgTx-2 treatment of Kv1.3 ϩ/ϩ cells was only about 20% of control levels and was insensitive to the anion channel blocker, NPPB (0.1 mM) ( Fig. 2A, lower panel). For Kv1.3 ϩ/ϩ thymocytes, only two channels among those whose transcripts were found (see below) are apparently inhibited by AgTx-2 (34): Kv1.3 (K d ϳ 4 pM) and Kv1.6 (K d ϳ 40 pM). Although Kv1.1 is also blocked (K d ϳ 40 pM), it was not detected in Kv1.3 ϩ/ϩ thymocytes. Because we used 2 or 5 nM AgTx-2, in principle, Kv1.6 expression could also contribute to the current in Kv1.3 ϩ/ϩ thymocytes. One possibility is that Kv1.6 may not form functional channels on the cell surface of thymocytes, because Kv1.6 transcripts were also present in Kv1.3 Ϫ/Ϫ cells, but there was no Kv current (see below). Although we cannot rule out a difference in Kv1.6 activity in the two mice strains, the simplest interpretation is that Kv1.3 accounts for the AgTx-2-sensitive current in thymocytes from wild-type mice.
In contrast to Kv1.3 ϩ/ϩ cells, no AgTx-2-sensitive current could be detected in cells from Kv1.3 Ϫ/Ϫ mice. Instead, a substantial time-independent current appeared (Fig. 2B), the properties of which closely matched those of the small conductance chloride current in human T cells (35). The current was blocked by 0.1 mM NPPB (Fig. 2B, middle panel), a well known blocker of whole-cell Cl Ϫ current in normal human T lymphocytes (35) and microglia cells (36), but only weakly by the Cl Ϫ channel blocker, DIDS (0.1 mM, not shown), also as previously seen in normal human T lymphocytes (35). NPPB also blocks some other Cl Ϫ currents, including cloned ClC-2 channels (37). It is not known to block voltage-gated K ϩ channels, and because the only current observed in the Kv1.3-deficient thymo- cytes was a time-and voltage-independent Cl Ϫ current, NPPB appears to be selective in this context. After block by NPPB, the remaining current reversed at about 0 mV (as expected for a non-selective leak current) and was comparable in amplitude to the leak current (AgTx-2-insensitive component) in Kv1.3 ϩ/ϩ thymocytes. The whole-cell current and conductance, which is the slope of the current-versus-voltage (I-V) relation, of Kv1.3 Ϫ/Ϫ thymocytes were outwardly rectifying (Fig. 2C). With the standard NaCl bath and K ϩ aspartate pipette solutions the I-V relation reversed at Ϫ50 mV, from which the calculated aspartate permeability is about 15% that of Cl Ϫ ; the same value calculated for the Cl Ϫ current in human T cells (35). To further corroborate the anion selectivity, NaCl in the bath was replaced with aspartate. The resulting reversal potential shifted to 0 mV (Fig. 2C), as predicted for an anion channel.

-deficient mice
Relative mRNA levels were determined by RT-PCR with gene-specific primers (n ϭ 3 animals; at least four independent PCRs per sample). Analyses were done using total thymocyte populations. CD4 ϩ CD8 ϩ cells in both Kv1.3 ϩ/ϩ mice and Kv1.3 Ϫ/Ϫ mice represented about 85% of the total thymocyte population, with a relatively normal distribution of other subsets (see Table III). Band densities were quantified using the Quantity One software program (Gel-Doc system, Bio-Rad).
AgTx-2-sensitive Kv1.3 current was 505 Ϯ 208 pA/pF (n ϭ 7); similar in amplitude to that previously reported (11). Although ϳ85% of the cells are CD4 ϩ CD8 ϩ , other cells may have been included in the small sample sizes. This would contribute to the variability, because different thymocyte subsets express different amounts of Kv1.3 current. None of the Kv1.3 ϩ/ϩ thymocytes had detectable NPPB-sensitive anion currents (n ϭ 7). In contrast, none of the Kv1.3 Ϫ/Ϫ thymocytes (n ϭ 9) had detectable AgTx-2-sensitive Kv1.3 current, but all had a substantial anion current (289 Ϯ 72 pA/pF at ϩ50 mV, n ϭ 9) that was almost as large as the Kv1.3 current in Kv1.3 ϩ/ϩ thymocytes. All cells tested had comparably small leak currents (the AgTx-2-insensitive and NPPB-insensitive portion): 62 Ϯ 15 pA/pF (n ϭ 7) in Kv1.3 ϩ/ϩ thymocytes and 54 Ϯ 15 pA/pF (n ϭ 9) in Kv1.3 Ϫ/Ϫ thymocytes. Further corroboration that this was simply the leak current was that the remaining current was not inhibited by 100 M gadolinium, a blocker of non-selective cation currents (see Ref. 38) or by 5 nM apamin, a blocker of the SK2 Ca 2ϩ /calmodulin-activated K ϩ channel (39), and the only SK channel found by using RT-PCR in these cells.  Table III). Potassium channel subunits were selected for study if they had previously been found in immune cells or tissues (32,40). In addition, several previously cloned Cl Ϫ channels were assessed as possibly underlying the lymphocyte Cl Ϫ current (1,35). mRNAs, determined by RT-PCR with gene-specific primers (Table I), were compared with a positive control; i.e. ␤-actin mRNA levels in the same samples (n ϭ 3 animals; Ͼ4 independent tests per sample) and summarized in Table II as present or undetectable.
Immune Cell Profile of Kv1.3-deficient Mice-Counting and fluorocytometry of thymocytes and splenocytes from 6-to 10week-old Kv1.3 ϩ/ϩ mice and Kv1.3 Ϫ/Ϫ mice with several antibodies failed to reveal any significant differences in relative or absolute numbers of thymocytes or splenic B cells and T cells (Table III). This included the presence of a relatively normal frequency of splenic CD62L-low/CD44-high memory CD4 T cells. The small difference in absolute numbers of thymocytes was not significant, nor was there a significant difference in absolute numbers of any thymocyte subset (data not shown, but see Table III).
Splenocytes were then employed to test the significance of Kv1.3 deficiency on T cell proliferation capacity in response to the polyclonal T cell mitogen, concanavalin A (ConA). Fig. 5 shows that splenocytes from 6-to 8-week-old Kv1.3 Ϫ/Ϫ mice responded normally to ConA. Equivalent results were obtained with the polyclonal T cell mitogen anti-CD3 (data not shown).
Although the role of ion channels in apoptosis is largely unexplored, Kv1.3 current is reduced during CD95/Fas-induced apoptosis in the human T cell line, Jurkat (44,45). Moreover, Fas stimulation caused tyrosine phosphorylation of Kv1.3 (44,45), a modification that inhibits Kv1.3 activity (3,44,46). Blockade of Kv1.3 activity might conceivably be a prerequisite to K ϩ depletion from cells, which occurs early in apoptosis and stimulates a molecule in the apoptosis pathway, interleukin-converting enzyme (47).
Thus, we next challenged thymocytes in vitro with various apoptosis-inducing agents to examine whether or not Kv1.3 is involved in one or other pathway of thymocyte apoptosis. Dexamethasone, a synthetic glucocorticoid hormone, causes apoptosis in thymocytes (48) in a manner attributable to early ceramide generation caused by the activation of an acidic sphingomyelinase (49). Inhibition of ceramide generation inhibited caspase activation and thymocyte death (49). Like dexamethasone, anti-CD3 antibody treatment causes thymocyte apoptosis via caspase 9 activation, whereas Fas-induced apo-  ptosis employs a caspase 8 pathway (50). Staurosporine is a protein kinase inhibitor that induces apoptosis with cytoplasmic features that appear to be independent of caspase activation (such as externalization of phosphatidylserine and loss of mitochondrial membrane integrity), although nuclear features of staurosporine-induced apoptosis are caspase activation-dependent (51). Finally, UV irradiation causes, among other things, DNA damage leading to the expression of Fas ligand and subsequent apoptosis (52).
Thymocyte apoptosis was assessed using annexin V and DiOC 6 (see "Materials and Methods"). These assays did not reveal any significant difference between Kv1.3 ϩ/ϩ and Kv1.3 Ϫ/Ϫ thymocytes in anti-CD3-, anti-Fas-, UV-, or druginduced apoptosis (Fig. 6). DISCUSSION We have generated Kv1.3 Ϫ/Ϫ mice by deleting a large promoter region and the N-terminal third of the Kv1.3 coding sequence. This resulted in complete loss of Kv1.3 as demonstrated by RT-PCR, Western blot analysis, and whole-cell current recordings. Despite the loss of Kv1.3, Kv1.3 Ϫ/Ϫ mice had a normal distribution of lymphocytes in the thymus and spleen (Table III) as well as mesenteric lymph nodes (data not shown). There were also no apparent abnormalities in thymocyte development or apoptosis, or in the ability of peripheral T cells to proliferate. These findings might at first seem to contradict studies suggesting that Kv1.3 has a critical role in the immune system. For example, margatoxin, which blocks both Kv1.3 and Kv1.1 (53,54), caused thymic atrophy in mature mini-swine (14). In addition, there are aspects of Kv1.3 function that we have not yet addressed. For example, selective blockade of Kv1.3 ameliorated CD4 ϩ T cell-mediated experimental allergic encephalomyelitis in a rat model (55).
One possible explanation for the apparent lack of abnormality in Kv1.3 Ϫ/Ϫ mice is that the loss of all detectable voltagedependent potassium current in most Kv1.3 Ϫ/Ϫ thymocytes was accompanied by the emergence of a major component of chloride current whose amplitude matched that of the Kv1.3 current in wild-type cells. In addition, Kv1.3 Ϫ/Ϫ thymocytes also appear to have up-regulated mRNA for Kv1.1, Kv3.1, and SK2, suggesting that there was a concerted compensatory increase in expression of these genes as a result of the loss of Kv1.3 currents. One role of Kv1.3 channels in T cells is to maintain a negative membrane potential necessary for Ca 2ϩ influx into thymocytes (21). Because the Nernst potential for chloride is also negative (ϳϪ30 to Ϫ40 mV (56)), in principle, Cl Ϫ channels could play a similar role. Although it would be desirable to compare the membrane potential of individual thymocytes from Kv1.3 ϩ/ϩ and Kv1.3 Ϫ/Ϫ mice, this is problematic. Thymocytes have very small maximal currents and even smaller currents at the resting potential; i.e. they have a much higher membrane resistance than the electrode seal resistance. ⌻hus, the leak around the patch pipette will depolarize the cells from their true membrane potential value.
We also observed greatly increased Kv1.1 protein in Kv1.3 Ϫ/Ϫ total thymocytes, without detecting a corresponding current. The presence of mRNA and even protein for an ion channel does not guarantee a corresponding membrane current for several reasons. (i) There are many examples of Kv channels not being present on the plasma membrane; e.g. because of trapping in the endoplasmic reticulum or other internal compartments or mis-processing. Thus, even if protein is detected on a Western blot, it may not be on the cell surface. For instance, we found that Kv1.5 protein in rat microglia was restricted to an intracellular location (2). (ii) Even if the channel is on the cell surface, channel modifications may inhibit the current; e.g. the inhibition of Kv1.3 and 1.5 by tyrosine phos-phorylation (46,57). (iii) CD4 ϩ CD8 ϩ thymocytes, which represent about 85% of all thymocytes, may lack these currents, and Kv1.1 in total Kv1.3 Ϫ/Ϫ thymocyte preparations might have been restricted to CD4 Ϫ CD8 Ϫ thymocytes, where Kv1.1 is normally expressed (18).
The prevalent chloride current in Kv1.3 Ϫ/Ϫ thymocytes had some properties that are the same as the small conductance chloride current in normal human T cells (35). Both lacked time-or voltage-dependent gating, were outwardly rectifying, had a calculated aspartate permeability of about 15% that of Cl Ϫ , and were fully blocked by 0.1 mM NPPB but only weakly blocked by 0.1 mM DIDS. The molecular identity of the chloride channel has not been determined, but it bears several similarities to the cloned ClC-3 channel, which produces a voltageinsensitive outwardly rectifying current with an iodide Ͼ chloride Ͼ Ͼ aspartate selectivity sequence. However, other properties diverge from ClC-3, which is strongly inhibited by 0.1 mM DIDS (100 M) and inactivates at potentials above ϩ80 mV (58). The lymphocyte current is less similar to ClC-2, which has a chloride Ͼ bromide Ͼ iodide selectivity sequence and is closed unless activated by a strong hyperpolarization (59) or a hypo-osmotic shock (60). One intriguing possibility is that these two proteins, like other ClC family members, form heteromultimers with novel properties (61).
In conclusion, deletion of the Kv1.3 gene results in the complete loss of voltage-dependent potassium current in the majority of thymocytes. This is accompanied by an increase of approximately 50-fold in chloride conductance. Because chloride channels would be expected to maintain negative membrane potentials in lymphocytes, this compensatory change in membrane permeability may account for the apparent absence of abnormalities in the immune responses of Kv1.3 Ϫ/Ϫ animals. Our findings also highlight the need for caution when interpreting studies on channel-deficient mice because of the possibility of compensatory channel up-regulation.