Binding of Correolide to Kv1 Family Potassium Channels

Correolide, a novel nortriterpene natural product, potently inhibits the voltage-gated potassium channel, Kv1.3, and [3H]dihydrocorreolide (diTC) binds with high affinity (K d ∼ 10 nm) to membranes from Chinese hamster ovary cells that express Kv1.3 (Felix, J. P., Bugianesi, R. M., Schmalhofer, W. A., Borris, R., Goetz, M. A., Hensens, O. D., Bao, J.-M., Kayser, F., Parsons, W. H., Rupprecht, K., Garcia, M. L., Kaczorowski, G. J., and Slaughter, R. S. (1999) Biochemistry 38, 4922–4930). Mutagenesis studies were used to localize the diTC binding site and to design a high affinity receptor in the diTC-insensitive channel, Kv3.2. Transferring the pore from Kv1.3 to Kv3.2 produces a chimera that binds peptidyl inhibitors of Kv1.3 with high affinity, but not diTC. Transfer of the S5 region of Kv1.3 to Kv3.2 reconstitutes diTC binding at 4-fold lower affinity as compared with Kv1.3, whereas transfer of the entire S5-S6 domain results in a normal Kv1.3 phenotype. Substitutions in S5-S6 of Kv1.3 with nonconserved residues from Kv3.2 has identified two positions in S5 and one in S6 that cause significant alterations in diTC binding. High affinity diTC binding can be conferred to Kv3.2 after substitution of these three residues with the corresponding amino acids found in Kv1.3. These results suggest that lack of sensitivity of Kv3.2 to diTC is a consequence of the presence of Phe382 and Ile387 in S5, and Met458 in S6. Inspection of Kv1.1–1.6 channels indicates that they all possess identical S5 and S6domains. As expected, diTC binds with high affinity (K d values 7–21 nm) to each of these homotetrameric channels. However, the kinetics of binding are fastest with Kv1.3 and Kv1.4, suggesting that conformations associated with C-type inactivation will facilitate entry and exit of diTC at its binding site. Taken together, these findings identify Kv1 channel regions necessary for high affinity diTC binding, as well as, reveal a channel conformation that markedly influences the rate of binding of this ligand.

useful in defining various structural and functional properties of potassium channels (2). Two of these peptides, margatoxin (MgTX) 1 and charybdotoxin (ChTX), have been critical in identifying K v 1.3 as a target for development of novel immunosuppressants (3)(4)(5)(6). In the course of screening for small molecule K v 1.3 inhibitors to support these efforts, a novel nortriterpene, correolide, was isolated from the plant Spachea correa (7) and shown to be a potent K v 1.3 blocker (8). Correolide is a selective inhibitor of the K v 1 family of potassium channels, displaying very low affinity for other ion channels and membrane proteins. However, its selectivity for K v 1.3 appears to be limited and in functional assays it blocks other K v 1 channels with 4 -20-fold lower potency. A tritium-labeled derivative of correolide, [ 3 H]dihydrocorreolide (diTC; Table I), binds reversibly and saturably to K v 1.3; the stoichiometry of the binding reaction suggests that one molecule of correolide binds per channel tetramer; the binding sites for correolide and peptidyl blockers are distinct (8). Because of the large structural diversity of voltage-gated potassium channels (9 -11), knowledge of specific molecular determinants that determine high affinity binding of modulators among different channel proteins is crucial for development of potent and selective therapeutic agents.
In this study, we investigate the molecular basis for interaction of diTC with K v 1.3 and design a high affinity receptor in the diTC-insensitive channel, K v 3.2. Our results indicate that residues located in the S 5 and S 6 region are responsible for the high affinity interaction of diTC with K v 1.3 and that the lack of sensitivity of K v 3.2 to diTC is due to the presence of Phe 382 and Ile 387 in S 5 , and Met 458 in S 6 . Although all K v 1 channels display conserved residues in the S 5 and S 6 domains and diTC binds with high affinity to all of these channels, the kinetics of ligand association and dissociation are much faster with K v 1.3 and K v 1.4 than with other K v 1 channels. The energy barrier for diTC to enter or leave its binding site may be less for K v 1.3 and K v 1.4 as a consequence of a change in channel conformation induced by C-type inactivation, which is a common feature of these two channels. A preliminary report of these findings has been made in abstract form (12).

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes and the pCI-neo vector were bought from Promega. The pEGFP-N1 vector was from CLONTECH, and Pfu DNA polymerase from Stratagene. TsA-201 cell line, a subclone of the human embryonic kidney cell line, HEK293, which expresses the SV40 T antigen, was a gift of Dr. Robert DuBridge. All tissue culture media were from Life Technologies, Inc., serum was from Hyclone, and the FuGENE transfection reagent was from Roche Molecular Biochemi-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. cals. HEK293 cells stably transfected with homotetrameric K v 1.1, K v 1.3, K v 1.4, K v 1.5, and K v 1.6 channels were obtained from Professor Olaf Pongs (Zentrum fü r Molekulare Neurobiologie, Hamburg, Germany). HEK293 cells stably transfected with K v 1.2 were prepared as described (8). K v 3.2 DNA was a generous gift of Dr. Bernardo Rudy, New York University Medical Center. Correolide, diTC (29 Ci/mmol), and dihydrocorreolide (diHC) were prepared as described previously (8). Hongotoxin1-A19Y/Y37F (HgTX 1 A19Y/Y37F) was prepared and radioiodinated as described (13). Charybdotoxin and stichodactyla toxin were purchased from Peninsula Laboratories. GF/C glass fiber filters were obtained from Whatman, and polyethylenimine was from Sigma. All other reagents were obtained from commercial sources and were of the highest purity commercially available.
Mutant Channel Constructs-A 9E10 c-Myc tag was introduced at the C terminus of K v 1.3 and K v 3.2 using an oligonucleotide cassette containing a HindIII and NotI restriction site. In addition, the 5Јuntranslated region of K v 3.2 was replaced with the corresponding sequence of K v 1.3 to improve channel expression. Chimeric cDNAs of K v 1.3 and K v 3.2 were generated using the gene splicing with overlap extension technique (14). Chimeras were constructed as follows: in K v 3.2 (P..K v 1.3), the linker between transmembrane domain S 5 and S 6 of K v 3.2 (referred to as P) was replaced with the corresponding region from K v 1.3; in K v 3.2 (S 5 , P..K v 1.3), both the transmembrane domain S 5 and P of K v 3.2 were replaced with those from K v 1.3; in K v 3.2 (P, S 6 ..K v 1.3), both P and transmembrane domain S 6 of K v 3.2 were replaced with those from K v 1.3; in K v 3.2 (S 5 -S 6 ..K v 1.3) both P and transmembrane domains S 5 and S 6 of K v 3.2 were replaced with the corresponding segments from K v 1.3. Site-directed mutagenesis was performed using the "overlap extension" technique (15). Polymerase chain reaction was carried out using proofreading Pfu DNA polymerase and the integrity of all constructs was verified by nucleotide sequencing (automated sequencer, ABI 377).
Transfection of TsA-201 Cells and Membrane Preparation-The procedures for handling TsA-201 cells, their transfection with FuGENE transfection reagent, and preparation of membrane vesicles have been previously described (16). The final membrane pellet was resuspended in 100 mM NaCl, 20 mM Hepes-NaOH, pH 7.4. Aliquots were frozen in liquid N 2 and stored at Ϫ70°C. 125 I-HgTX 1 A19Y/Y37F Binding-The interaction of 125 I-HgTX 1 A19Y/Y37F with TsA-201 membranes was measured in a medium consisting of 50 mM NaCl, 5 mM KCl, 20 mM Tris-HCl, pH 7.4, 0.1% bovine serum albumin. For saturation experiments, membranes were incubated with increasing concentrations of 125 I-HgTX 1 -A19Y/ Y37F in a total volume of 4 ml for 20 h at room temperature. Separation of bound from free ligand was achieved using filtration protocols as described (13).
diTC Binding-Binding of diTC to membranes was carried out in a medium consisting of 135 mM NaCl, 4.6 mM KCl, 20 mM Tris-HCl, pH 7.4, 0.02% bovine serum albumin. For saturation experiments, membranes were incubated in a total volume of 0.2 ml with increasing concentrations of diTC for 20 h at room temperature. To determine kinetics of ligand association, membranes were incubated with diTC for different periods of time at room temperature. Dissociation kinetics were initiated by addition of 10 M diHC to membrane-bound diTC followed by incubation at room temperature for different periods of time. Competition experiments were carried out in a total volume of 1 ml with 1 nM diTC in the absence or presence of increasing concentrations of diHC. Under these conditions, IC 50 values of inhibition are very close to K i values, because the concentration of diTC in the incubation assay is ϳ5-fold below K d . Separation of bound from free ligand was achieved using a filtration protocol as described (8). Triplicate samples were run for each experimental point. Standard deviation of the mean was typically less than 5%.
Data Analysis-Data from saturation, ligand association and dissociation experiments were analyzed as described (8,16). IC 50 values for peptide and diHC inhibition were determined using the equation: B eq ϭ (B max Ϫ B min )/(1ϩ(I/IC 50 ) nH ) ϩ B min , where B eq is the degree of binding at the ligand concentration tested with no inhibitors present, B min is the minimum amount of ligand bound at higher concentrations of inhibitor where the binding curve has leveled off, I is the inhibitor concentration, and IC 50 , the inhibition constant. For the data presented, B max was usually around 100% and B min was 0%.

RESULTS
Construction of K v 3.2/1.3 Channel Chimeras-Although diTC binds to K v 1.3 with high affinity, this ligand does not interact with members from other families of voltage-gated potassium channels such as K v 3.2 (8). Therefore, chimeras were constructed using regions of K v 1.3 and K v 3.2 to identify those domains that confer high affinity binding of diTC. This is an approach similar to that used to identify determinants responsible for block of potassium channels by peptide inhibitors (17,18).
K v 1.3 channels are inhibited with high affinity by peptidyl blockers such as ChTX (6,19,20), hongotoxin-1 (HgTX 1 ) (13), MgTX (21), kaliotoxin (22), and stichodactyla toxin (23). Furthermore, high affinity binding sites for radiolabeled ChTX (24), MgTX (25), and HgTX 1 (13) have been identified in membranes prepared from cells containing K v 1.3 channels. K v 3.2 channels, however, are not sensitive to any of the identified peptidyl blockers of potassium channels (2). To introduce a marker into K v 3.2 that could be used to quantitate levels of channel expression after transient transfection of TsA-201 cells, a chimera was constructed in which the pore region of K v 1.3 was transferred to K v 3.2 ( Fig. 1). Expression of this and other chimeras was also monitored by immunoblotting membrane preparations using an antibody against the c-Myc sequence that was added to the C terminus of these proteins. Such experiments indicate that wild type channels and all chimeric channels analyzed in these studies give similar levels of protein expression (data not shown).
To determine whether the chimeric proteins associate to form tetrameric structures known to be required for binding of peptides, binding of 125 I-HgTX 1 A19Y/Y37F was measured to membranes derived from TsA-201 cells transiently transfected with either K v 1.3 or K v 3.2(P..K v 1.3). Results of these experiments are shown in Fig. 2A. 125 I-HgTX 1 A19Y/Y37F binds to a single class of sites that display K d values of 0.29 and 3.8 pM for K v 1.3 and K v 3.2(P..K v 1.3), respectively. Thus, transferring the pore region of K v 1.3 yields a high affinity receptor for HgTX 1 in K v 3.2 that can be used to monitor levels of channel expression. The differences in HgTX 1 affinity noted between K v 1.3 and this chimera are similar to those found in studies where the pore of K v 1.3 was transferred to K v 2.1 to confer agitoxin 2 sensitivity to the later channel (18). In this study, we have not investigated whether these differences are due to effects on peptide association or dissociation kinetics.
Several other K v 3.2 chimeras were constructed ( Fig. 1) and characterized initially for functional channel tetramerization. In these cases, either transmembrane domain S 5 , S 6 , or the complete S 5 -S 6 region of K v 1.3 was transferred to the K v 3.2(P..K v 1.3) chimera. Binding of 125 I-HgTX 1 A19Y/Y37F was monitored to these chimeras, and the results are shown in Fig. 2A. In all instances, high affinity binding of 125 I-HgTX 1 A19Y/Y37F occurred to a single class of sites as indicated by the fact that the Hill coefficients of all slopes shown in Fig. 2A are close to unity. The highest affinity for peptide was observed with the chimera in which the S 5 -S 6 region of K v 1.3 had been transferred. In fact, this chimera possesses the same affinity for HgTX 1 as does native K v 1.3. Binding of 125 I-HgTX 1 A19Y/Y37F to K v 3.2/1.3 chimeras was further characterized. Peptidyl blockers of K v 1.3 such as ChTX, MgTX, and stichodactyla toxin cause concentration-dependent inhibition of 125 I-HgTX 1 A19Y/Y37F binding (data not shown), and inhibitory potencies of these peptides for the various chimeras relative to the wild type channel scale with the ability of 125 I-HgTX 1 A19Y/Y37F to bind to these proteins. These data suggest that all K v 3.2 chimeras that were constructed can assemble to form tetrameric structures and that the molecular pharmacology of these chimeras with respect to the pore resembles closely that of native K v 1.3 in terms of their sensitivity to a number of different peptidyl inhibitors.
Binding unsuccessful. The resulting construct did not yield any detectable 125 I-HgTX 1 A19Y/Y37F binding, despite the fact that expression of the protein could be easily detected in Western blots. It is interesting that the chimera that contains the S 5 -S 6 region of K v 1.3 displays about 4-fold higher affinity for diTC than when S 5 is transferred to K v 3.2 by itself. This finding is consistent with the observation from competition binding experiments that a 4-fold loss in diHC affinity is observed when S 6 from K v 3.2 is transferred to K v 1.3 (data not shown), and suggests that differences in S 6 between the two channels are not the major contributor to correolide sensitivity.
Comparison of the sequences of K v 1.3 and K v 3.2 in S 5 and S 6 indicate that 16 nonidentical residues are present; 11 in S 5 and 5 in S 6 (Fig. 3). To determine whether any of these residues contribute to the differences observed in diTC binding, individual mutations were introduced in K v 1.3 to incorporate those amino acids present in K v 3.2. In these studies, binding parameters were determined with each mutant for both 125 I-HgTX 1 A19Y/Y37F and diTC, and the results are presented in Fig. 4. In these plots, a ratio of 1 indicates that the behavior of the mutant is identical to wild type, whereas a ratio of Ͼ1 reflects decreased binding as a result of the amino acid substitution. All mutants give similar levels of channel expression, as indicated by the maximum density of 125 I-HgTX 1 A19Y/Y37F binding sites, and all but one, V364I, displayed an identical affinity for 125 I-HgTX 1 A19Y/Y37F as wild type (Fig. 4A). These data indicate that, with the V364I exception, no major structural changes were conferred in the pore of the channel as a consequence of these mutations. In marked contrast, three mutations, L346 F and F351I, both in S 5 , and L422M in S 6 had significant effects on diTC binding, causing 25-, 3-, or 5-fold decreases in ligand affinity, respectively (Fig. 4B). There is also a modest effect of a 2-fold increase in K d with the mutation S362T. These data are consistent with previous observations regarding the K v 3.2/1.3 chimeras, and suggest that lack of high affinity interaction of diTC with K v 3.2 could be due to the combined effect of 3 nonconserved residues in K v 1.3, 2 located in S 5 and 1 in S 6 .
To test this idea further, two K v 3.2 mutants, K v 3.2F382L/ M458L, and K v 3.2F382L/I387F/M458L were constructed so that the corresponding residues in K v 1.3 were introduced into K v 3.2, and their diTC binding properties were compared with those of K v 1.3. Results of these experiments are presented in Fig. 5. diTC binds to a single class of sites that display K d values of 5.7, 20.9, and 10.2 nM for K v 1.3, K v 3.2F382L/M458L, and K v 3.2F382L/I387F/M458L, respectively (Fig. 5A). Thus, introduction of Leu, Phe, and Leu into K v 3.2 at positions 382, 387, and 458, respectively, reconstitutes a high affinity receptor for diTC; in the absence of these mutations, the K d of K v 3.2 for diTC is Ͼ1 M. As suggested from mutagenesis experiments with K v 1.3, the double mutant K v 3.2F382L/M458L has a 4-fold lower affinity for diTC than does K v 1.3. These results were also confirmed in competition experiments (Fig. 5B). diHC inhibits binding of diTC to these channels with K i values of 6, 19.2, and 9.4 nM for K v 1.3, K v 3.2F382L/M458L, and K v 3.2F382L/I387F/ M458L, respectively. All these data, taken together, strongly suggest that lack of sensitivity of K v 3.2 to diTC is due to the presence of only three amino acids; Phe 382 and Ile 387 in S 5 , and Met 458 in S 6 .
Interaction of DiTC with K v 1 Channels-Because the S 5 and S 6 domains are conserved among all K v 1 channels, it might be expected that diTC would bind to these proteins with similar K d values. Previously reported data indicate that diTC binds with high affinity to a single class of sites in human and rat brain membranes, a receptor most likely comprised of heteromultimers of K v 1.1/1.2/1.4 channels (8, 13), and that K d values determined from either equilibrium binding or ratio of rate constants are equivalent for both brain and K v 1.3 membranes. However, there are marked differences in the kinetics of ligand binding. Thus, rates of ligand association and dissociation are about 20-fold faster for K v 1.3 than with brain synaptic membranes. Because brain membranes do not contain a homogeneous population of any given K v 1 channel, the properties of diTC binding to HEK293 membranes containing either K v 1.1, K v 1.2, K v 1.3, K v 1.4, K v 1.5, or K v 1.6 homotetrameric channels were investigated next. Under equilibrium conditions, diTC binds in a saturable fashion and with similar K d values (7-21 nM) to a single class of sites present in each of these membrane preparations (Table I). These findings are consistent with the postulate that high affinity diTC binding would be nearly equivalent for all members of the K v 1 channel family.
The kinetics of diTC association to homomultimeric K v 1 channels were investigated next. At a ligand concentration close to K d , diTC associates with channels in each of these membrane preparations in a time-dependent fashion; binding kinetics are pseudo-first order (Fig. 6). However, the association rate constant (k ϩ1 ), varies greatly depending on the K v 1 channel under consideration. Thus, association rates of diTC with either K v 1.3 or K v 1.4 channels are about 10 -50-fold faster than those determined for the other K v 1 channels (Table I). The kinetics of ligand dissociation were also determined from each of these channels. In all cases, dissociation of diTC is a timedependent event that follows monoexponential kinetics, indicative of a first-order reaction (Fig. 7). Dissociation rate constant values (k Ϫ1 ) for the different K v 1 channels are listed in Table I. K d values determined from the ratio of k Ϫ1 versus k 1 are similar to those determined under equilibrium binding conditions, suggesting that diTC binds to homotetrameric K v 1 channels through a simple bimolecular reaction. As observed for k 1 , K v 1.3 and K v 1.4 are the channels that display the fastest dissociation kinetics. Differences in k Ϫ1 values between these and other K v 1 channels are about 5-25-fold. Thus, the affinity of diTC for all K v 1 channels, as determined under equilibrium binding conditions, is similar because relative differences in association and dissociation rate constants cancel each other. These data suggest that with some K v 1 channels, ligand must overcome a much larger energy barrier than is present in either K v 1.3 or K v 1.4 during binding and unbinding reactions. Inter-estingly, K v 1.3 and K v 1.4 are the only channels in the K v 1 family that undergo significant C-type inactivation, a process involving a conformational change of some residues in the outer vestibule of the channel that constricts the pore and causes blockade of the ion conduction pathway (26). It is possible that accessibility to the binding site for diTC becomes more facile after C-type inactivation occurs.
Kinetics of diTC Binding to K v 1.3 Channel Mutants with Altered C-type Inactivation-The idea that C-type inactivation induces a conformational change in K v 1.3 that modifies the accessibility of diTC to its binding site was tested by investigating two channel mutants, K v 1.3/G375Q and K v 1.3/H399N, that display altered inactivation kinetics. In K v 1.3, residues Gly 375 and His 399 are located in the external vestibule of the channel and appear to be directly involved in the conformational change that leads to channel inhibition (27). The mutant K v 1.3/G375Q is 8-fold slower in achieving C-type inactivation than wild type K v 1.3, whereas K v 1.3/H399N is 50-fold faster in this respect. Importantly, under equilibrium conditions, both K v 1.3 mutants and the wild type channel bind diTC with identical affinities when the reaction is monitored in membranes prepared from TsA-201 cells transiently expressing these proteins (data not shown). However, kinetics of diTC dissociation are markedly different among these various channels (Fig. 8).
Ligand dissociation follows first-order kinetics with t 1/2 values of 59, 18, and 305 min for K v 1.3, K v 1.3/H399N, and K v 1.3/ G375Q, respectively. Thus, diTC dissociation kinetics are 5-fold slower from a protein where the rate of C-type inactivation is diminished, whereas it is 3-fold faster from a K v 1.3 mutant displaying enhanced inactivation. Because equilibrium binding studies demonstrate that the affinity of diTC for all these K v 1.3 channels is the same, it is predicted that the kinetics of ligand association will also be modified accordingly. These data are consistent with the idea that diTC accessibility to and from its binding site is influenced by the conformational change eliciting C-type inactivation.

DISCUSSION
Correolide is the first potent, small molecule, natural product inhibitor of a voltage-gated potassium channel to be described. It displays remarkable specificity among ion channels in that it only interacts with K v 1 family channels. The results presented in this manuscript address two important aspects concerning the interaction of diTC with potassium channels. The first deals with the molecular basis for high affinity interaction of diTC with K v 1 channels, and in this context it has been possible to confer high affinity diTC binding to the diTCinsensitive channel, K v 3.2. Second, with respect to K v 1 channels, which possess a conserved diTC binding domain and to which diTC binding is of equivalent high affinity, the kinetics of ligand binding track with the ease by which these channels enter the C-type inactivated state.
Studies with chimeric channels constructed from domains of K v 1.3 and the diTC-insensitive channel, K v 3.2, indicate that it is possible to confer correolide sensitivity to K v 3.2 by transferring only the S 5 region of K v 1.3. These data suggest that differences between K v 1.3 and K v 3.2 in S 5 are a major determinant in the lack of sensitivity of the latter channel to this inhibitor. Site-directed mutagenesis studies have identified three nonconserved residues that may account for the lack of sensitivity of K v 3.2 to diTC. High affinity interaction of diTC with K v 3.2 has been achieved by mutating these residues to the corresponding amino acids found in K v 1.3. Two of these residues are located in S 5 , whereas the third is in S 6 . Interestingly, differences in the S 6 regions of these two channels appear to play a minor role in diTC binding, despite the fact that residues of S 6 have been shown to contribute to formation of the pore (28,29). Two important questions arise from these studies. Is the binding domain for diTC restricted primarily to the S 5 region? Alternatively, are residues of S 5 critical for conferring a special conformation to a binding domain that is located in another region of the channel? Comparison of the S 5 and S 6 regions of all K v 1 channels indicates that there are no differences among any of these proteins. However, ligand binding kinetics are markedly different between channels possessing C-type inactivation and those channels that do not display this property. It appears that conformational changes induced by C-type inactivation make drug access to the diTC binding domain easier. Thus, it is possible that the specific residues that contribute to the binding site for diTC are common to many channels, but that the three-dimensional structure of the re-FIG. 6. Binding of diTC to K v 1 channels: association kinetics. Membranes prepared from HEK293 cells stably transfected with either K v 1.1, K v 1.2, K v 1.3, K v 1.4, K v 1.5, or K v 1.6 channels were incubated with diTC for the indicated periods of time at room temperature. Nonspecific binding, determined in the presence of 10 M diHC, is time-invariant and has been subtracted from the experimental points. Specific binding data have been fit to the pseudo-first order association reaction as described under "Experimental Procedures."

TABLE I
The structure of diTC Values of k 1 or k Ϫ1 for diTC binding to homotetrameric K v 1 channels were calculated as described in the text. K d values for diTC were also determined under equilibrium binding conditions as described. These values are the average of 2-4 determinations carried out with different membrane preparations. ceptor differs between these channels. On the other hand, the pathway for drug entering its receptor site could be the ratedetermining step for binding. For some K v 1 channels, this energy barrier represents an additional 2 kcal/mol, but for other channels, such as K v 3.2, it could be much larger so as to prevent drug binding at all. Site-directed mutagenesis studies along S 5 and S 6 in K v 1.3 would be required to identify those specific residues that contribute to the diTC binding pocket. With use of the recently disclosed high resolution structure of the KcsA channel (28), it may be possible to model the structure of K v 1.3. Such a model could be useful in general for probing the molecular determinants that control diTC binding to potassium channels. Correolide and several structural analogs were isolated from the plant Spachea correa (7) based on their ability to block K v 1.3 potently in a functional efflux assay (8). These compounds also inhibit lymphocyte activation mediated by Ca 2ϩdependent stimulation of T cells in in vitro assays, as well as in a mini-pig in vivo delayed-type hypersensitivity test. 2 The mechanism by which correolide and its analogs suppress T cell activation is directly related to blockade of K v 1.3. Like peptide inhibitors, correolide partially depolarizes T cells (8). This depolarization leads to diminution of the Ca 2ϩ signaling that occurs upon activation of the T cell receptor complex. As expected, the properties of correolide and its analogs resemble those of peptidyl inhibitors of K v 1.3, such as MgTX, in immunological assays (4,5). Given these observations, correolidederived agents might be candidates for development as novel immunosuppressant drugs for treatment of graft rejection and autoimmune diseases.
The development of a safe immunosuppressant requires a minimum side effect profile. For agents targeting voltage-gated potassium channels, this is a monumental task given the large diversity of these proteins and their wide distribution throughout the body. Peptidyl inhibitors of K v 1 channels are not statedependent channel blockers and are generally neurotoxic. Although the safety profile of correolide and its analogs requires further investigation, preliminary results from evaluation of  H399N (q), or K v 1.3/G375Q (OE) were incubated with 10 nM diTC. Dissociation kinetics were initiated by addition of 10 M diHC followed by incubation at room temperature for different periods of time. Specific binding data have been fit to a single monoexponential decay corresponding to a first-order reaction. two correolide analogs that, after acute administration, block a delayed-type hypersensitivity reaction in mini-pigs, do not reveal a significant toxic side effect profile. 2 Perhaps correolide achieves functional selectivity in vivo as a blocker of T cell K v 1.3 channels. Such functional selectivity may be related to the mechanism by which correolide blocks K v 1.3. Preliminary studies indicate that this agent interacts with high affinity with either open or C-type inactivated, but not with the closed state of the channel (8,30). K v 1.3 exists primarily in the inactivated state in T cells with a resting potential of approximately Ϫ50 mV (4). If correolide would preferentially associate with the inactivated form of the channel, it may be possible to shift the channel equilibrium to the drug-bound form that is nonconductive. In firing neurons, however, the drug-sensitive inactivated form of the channel may be present for only a very brief period of time; drug binding might be negligible under these conditions, given the slow onset of correolide action observed with brain native channels (8). Although more detailed studies are necessary to discern the mechanism by which correolide interacts with potassium channels, it is tempting to speculate that the state-dependent interaction of this agent, together with its kinetics of binding, may be important for limiting toxicity in vivo.