Analysis of the Cyclic Nucleotide Binding Domain of the HERG Potassium Channel and Interactions with KCNE2*

Mutations in the cyclic nucleotide binding domain (CNBD) of the human ether-a-go-go -related gene (HERG) K 1 channel are associated with LQT2, a form of heredi-tary Long QT syndrome (LQTS). Elevation of cAMP can modulate HERG K 1 channels both by direct binding and indirect regulation through protein kinase A. To assess the physiological significance of cAMP binding to HERG, we introduced mutations to disrupt the cyclic nucleotide binding domain. Eight mutants including two naturally occurring LQT2 mutants V822M and R823W were constructed. Relative cAMP binding capacity was reduced or absent in CNBD mutants. Mutant homotetramers carry little or no K 1 current despite normal protein abundance and surface expression. Co-ex-pression of mutant and wild-type HERG resulted in currents with altered voltage dependence but without dominant current suppression. The data from co-expres-sion of V822M and wild-type HERG best fit a model where one normal subunit within a tetramer allows nearly normal current expression. The presence of KCNE2, an accessory protein that associates with HERG, however, conferred a partially dominant current suppression by CNBD mutants. Thus KCNE2 plays a pivotal role in determining the phenotypic severity of some forms of LQT2, which suggests that the CNBD of HERG may be involved in its interaction with KCNE2. and reverse primer an Eco RI site within the -GCTCGTATGTTGTGTGGAAT-TGTGAGCGG). 300-base pair construct HERG-psp64 D mutation 815–825.


Mutations in the cyclic nucleotide binding domain (CNBD) of the human ether-a-go-go-related gene (HERG)
K ؉ channel are associated with LQT2, a form of hereditary Long QT syndrome (LQTS). Elevation of cAMP can modulate HERG K ؉ channels both by direct binding and indirect regulation through protein kinase A. To assess the physiological significance of cAMP binding to HERG, we introduced mutations to disrupt the cyclic nucleotide binding domain. Eight mutants including two naturally occurring LQT2 mutants V822M and R823W were constructed. Relative cAMP binding capacity was reduced or absent in CNBD mutants. Mutant homotetramers carry little or no K ؉ current despite normal protein abundance and surface expression. Co-expression of mutant and wild-type HERG resulted in currents with altered voltage dependence but without dominant current suppression. The data from co-expression of V822M and wild-type HERG best fit a model where one normal subunit within a tetramer allows nearly normal current expression. The presence of KCNE2, an accessory protein that associates with HERG, however, conferred a partially dominant current suppression by CNBD mutants. Thus KCNE2 plays a pivotal role in determining the phenotypic severity of some forms of LQT2, which suggests that the CNBD of HERG may be involved in its interaction with KCNE2.
The HERG protein contains a putative cyclic nucleotide binding domain (CNBD) in its cytoplasmic carboxyl terminus (2). A splice-acceptor mutation resulting in a C-terminal truncation lacking the entire CNBD was one of the HERG mutants that was originally identified as a cause of LQT2 (3). More recently, new LQT2 mutations have been discovered in the putative CNBD (1,6,7).
Although the cardiac I kr was originally thought to be insensitive to cAMP-dependent regulation (3,8), recent studies provided evidence that HERG channels can be modulated via cAMP-dependent protein kinase (PKA) phosphorylation pathway (9 -11). We reported that cAMP could regulate HERG by direct interaction or through PKA-mediated phosphorylation of the channel (12). In this study, we further investigate the direct effect of cAMP on HERG channel properties through mutagenesis designed to disrupt the ability of HERG to bind cAMP.
Previous studies of the CNBD in cyclic nucleotide-dependent protein kinase and bacterial cAMP-regulated catabolic gene activator protein (CAP) identified six invariant key amino acid residues for cyclic nucleotide binding (13). Three of these are glycine residues that are essential for maintenance of the ␤-barrel structure that is required to form a pocket for cyclic nucleotide binding. A glutamic acid residue forms a hydrogen bond with the ribose 2Ј-OH of cAMP and an arginine residue interacts with the phosphate of cAMP to form a salt bridge (14). Mutagenesis of any of these six invariant amino acid residues in CAP or in type I cAMP-dependent kinase impairs or eliminates cAMP binding (15)(16)(17)(18)(19)(20). Homologous amino acid residues are also present in the CNBD of HERG. Here we describe the functional consequences of mutations at these homologous residues (G806D, E807(K/Q), and R823(W/Q)) designed to alter the ability of the channel to bind cAMP. Because HERG assembles in a tetramer to form functional channels, we also investigated the effects of co-expression of CNBD mutants with wild-type HERG. We found that CNBD mutants lose specific binding affinity for cAMP. They do not express currents as homotetrameric channels. Moreover, the loss of function is not because of defective protein biosynthesis or trafficking. Mutant HERG proteins do not have a dominant-negative effect on wild-type current but do alter voltage-dependent gating. Coexpression with the accessory subunit KCNE2, converts the CNBD mutants into partially dominant-negative suppressors.

MATERIALS AND METHODS
Mutagenesis of HERG cDNA-Epitope tagging of HERG with c-Myc has been described (4). Site-directed missense mutagenesis was performed by an overlap extension PCR strategy (21). The mutated XhoI-FseI fragments were subcloned into a HERG-myc-pCI-neo plasmid. An inframe deletion mutation ⌬NBD was constructed using a one-step PCR. The forward primer begins at 2662 of HERG cDNA (5Ј-TACTGT-GACCTACACAAGATCCATC) and the reverse primer spanned an EcoRI site within the psp64 vector (5Ј-GCTCGTATGTTGTGTGGAAT-TGTGAGCGG). The resulting PCR product was digested with FseI and the ϳ300-base pair fragment was purified and ligated into the parent construct HERG-psp64 digested with StuI and FseI sites. The resulting ⌬NBD mutation lacks the coding sequence for amino acids 815-825. N-terminal GFP-tagged HERG plasmids were constructed by inserting WT-HERG or CNBD mutants into pEGFPC3 at BamHI-HindIII sites. Mutant constructs were verified by DNA sequence analyses.
Transfection of cDNA into Cultured Cells-CHO cells were maintained in Ham's F-12 media supplemented with 10% fetal calf serum and penicillin/streptomycin at 37°C under 5% CO 2 . Gene transfer was performed using 1-5 g of Qiagen Midiprep purified plasmid DNA. Cells were electroporated at 225 V, 72 ohms, and 1800 microfarads in cytomix media (22). Cells were studied after 36 -72 h after transfection. A plasmid containing the cDNA for GFP was combined with HERG plasmids (in a ratio of 1:5 GFP/HERG) to allow identification of transfected cells as previously described (4). In co-transfection experiments current amplitudes were normalized each day to a group of cells transfected with HERG alone to control for day-to-day variations in transfection efficiency.
Electrophysiology-Whole-cell patch clamp (23) membrane currents were recorded as previously described (12). Extracellular solution was NaCl 150 mM, 1.8 mM CaCl 2 , 4 mM KCl, 1 mM MgCl 2 , 5 mM glucose, and 10 mM HEPES, pH 7.4, osmolarity ϳ300 -330 mM. Pipette solution was 120 mM KCl, 4 mM Mg-ATP, 2 mM MgSO 4 , 5 mM EGTA, 0.5 mM CaCl 2 , 10 mM HEPES, pH 7.2. Osmolarity was adjusted to maintain the pipette solution at 20 -40 mM less than the extracellular solution. Currents were evoked by 2-4 s depolarizing steps to various levels from a holding potential of Ϫ70 mV followed by a repolarizing step to Ϫ40 mV and then to Ϫ120 mV briefly to measure outward and inward tail currents. Current densities were calculated by dividing maximal tail current peaks by cell capacitance measurements. Voltage activation data were plotted as peak tail current amplitudes against the test potential values and were fitted to a Boltzmann function, I ϭ 1/(1 ϩ exp[(V1 ⁄2 ϪV)/k]), where I is the measured tail current, V is the applied membrane voltage, V1 ⁄2 is the voltage at half-maximal activation, and k is the slope factor.
Immunoblot Analysis-CHO or HEK293 cells were transiently transfected using LipofectAMINE 2000 according to the manufacturer's instructions (Life Technologies, Inc.). Cells were harvested 24 h after transfection for biochemical analyses. Membrane preparations were enriched for plasma membrane proteins and Western blots were carried out as previously described (4,24). To control for transfection efficiency between groups of cells, cDNA for Myc-tagged connexin43 was included. Proteins were separated by SDS-PAGE on a 7% gel. Anti-Myc 9E10 monoclonal antibody ascites fluid was used for Western blots to detect Myc-tagged HERG proteins.
Antibodies-Anti-HERG polyclonal antisera was generated against a GST fusion protein constructed from a PCR fragment encoding amino acids 979 -1159 ligated into pGEX-KG. The GST⅐HERG fusion protein was produced in Escherichia coli strain BL21 and was partially purified on GSH-agarose. The protein was then identified by Coomassie Blue staining after SDS-PAGE, and the appropriate gel band was excised, electroeluted, and injected into New Zealand White rabbits. The resulting antisera was assayed for ability to recognize heterologously expressed HERG⅐Myc protein on immunoblot and immunoprecipitation and was verified by comparison to anti-Myc staining. Anti-Myc monoclonal antibody was used as ascites fluid diluted into immunoblot blocking solution (Tris-buffered saline, 10% nonfat dry milk, and 0.1% Tween 20).
Microscopy-CHO cells were transiently transfected with Lipo-fectAMINE (as above) with GFP⅐HERG plasmids. After 24 -48 h, the transfected cells were resuspended and replated on glass coverslip chambers (Nunc) and allowed to attach for at least 24 h prior to imaging. GFP fluorescent images were collected using an Olympus Iϫ70 microscope with 12 bit cooled Photometrics Sensys CCD camera. Image analysis was performed using NIH Image, ImagePro, and Photoshop 7.0 software.

Targeted Mutations of HERG CNBD Result in Functionally
Defective Channels-To investigate the importance of the HERG nucleotide binding domain, we constructed eight CNBD mutations targeted to disrupt cAMP binding (Fig. 1a). The selected mutant amino acid residues are homologous to invariant residues in other cyclic nucleotide-binding proteins that have been identified as essential for normal conformation and cAMP binding ability. Among these mutations, V822M and R823W are naturally occurring LQT2 mutations (1,6). ⌬NBD is an inframe deletion mutant HERG lacking 11 amino acids in the cyclic nucleotide binding site (⌬815-825).
We measured the whole cell K ϩ current from CHO cells transiently transfected with each HERG mutant. Typical whole cell current tracings are seen in Fig. 1b. Except for mutant R823Q, none of the mutations could produce voltagegated K ϩ current when expressed alone. We also did not detect any K ϩ current when mutant expressing cells were exposed to 1 mM CPT-cAMP. Our results confirmed previous reports that LQT2 homotetramers involving the cyclic nucleotide binding domain (V822M and S818L) do not form functional channels (25,26).
We have previously shown that cAMP decreased whole cell HERG current amplitude with minimal alteration in voltagedependent activation (12). This current inhibition was primar- ily achieved via PKA-mediated phosphorylation of the channel protein.
To assess the effect of cAMP on mutant R823Q, a membrane-permeable cAMP analog CPT-cAMP was applied to the extracellular solution at a concentration of 500 M. HERG K ϩ current was recorded 3 min after CPT-cAMP addition. Fig.  2b shows that cAMP caused a current reduction of ϳ15% and a small negative shift in voltage-dependent activation (V1 ⁄2 (baseline) ϭ Ϫ20.44 Ϯ 0.39 mV; V1 ⁄2(cAMP) ϭ Ϫ26.86 Ϯ 0.16 mV, n ϭ 18).
Analysis of CNBD Mutant HERG Protein Expression-Several LQT2 mutations are known to cause reduced I kr by defective transport of mutant proteins to the plasma membrane or through enhanced protein degradation (24,25,27,28). When misfolded mutant HERG subunits assemble with wild-type protein, a dominant effect can be exerted where both proteins are subjected to early and rapid degradation (24). To assess whether the dysfunction of CNBD mutations was also caused by abnormal protein biosynthesis or trafficking, we examined the abundance and the membrane expression of c-Myc-tagged HERG and CNBD mutants. Anti-Myc immunoblot analysis showed that the CNBD mutants were expressed in abundance less than but comparable with wild-type HERG (Fig. 3a). There was no evidence of retention of CNBD mutant protein in endoplasmic reticulum or Golgi compartments when fractions enriched for microsomal membrane proteins were isolated from plasma membrane and subjected to immunoblotting (Fig. 3b).
Recent reports have shown that loss of function associated with misfolded and mistrafficked mutant HERG may be rescued by incubation at low temperatures (24,27,28). We examined the effect of decreasing culture temperature on the expression of CNBD mutations. Previous work demonstrated that HERG existed as two bands on SDS-PAGE analysis when expressed in HEK293 cells. A band migrating near 155 kDa represents the mature fully glycosylated channel protein, and a band near 135-kDa represents a core-glycosylated immature form of protein (29). As shown in Fig. 3c, both wild-type HERG and the CNBD mutants were expressed as two bands on Western blot. In all groups, both bands migrated slightly slower when grown at 30°C. No enhancement of mutant protein abundance occurred in cells that were grown at the lower temperature. Moreover, we were not able to detect any HERG current in CHO cells transfected with the V822M or ⌬NBD mutants after incubation at 30°C for 36 -72 h. The R823W mutant, however, occasionally produced a very small amount of current when the incubation temperature was decreased to 27-30°C for 24 -48 h. Among 12 examined cells, 6 exhibited small HERG currents (Fig. 3d). The average tail current density for wild-type and R823W at 30°C was 117.6 Ϯ 14.5 pA/pF and 8.25 Ϯ 2.7 pA/pF, respectively (p Ͻ 0.01). Despite comparable protein abundance, low temperatures resulted in a current that was expressed in only half of the cells with an amplitude that was ϳ7% of the wild-type current. This argues against misfolding and mistrafficking as a primary mechanism in the loss of function for R823W.
To further examine whether the CNBD mutant proteins are expressed on the plasma membrane, we visualized the location of HERG protein in living cells using N-terminal GFP fusion proteins for both wild-type and mutant HERG. GFP-tagged wild-type HERG produces K ϩ current similar to untagged wildtype HERG (data not shown). GFP-tagged ␤-adrenergic receptor (␤-AR) was used as a control for plasma membrane location (30). CNBD mutations exhibited a similar fluorescence pattern as wild-type HERG and ␤-AR (Fig. 4). As frequently seen with transient forced expression, a signal was seen in endoplasmic reticulum-Golgi locations in all cells in addition to cell surface locations. These results confirm that defective biosynthetic processing and protein transport are not the major causes of the channel dysfunction in CNBD mutations.

Co-expression of CNBD Mutants with Wild-type HERG Produces Functional
Heterotetrameric Channels-To assess the interactions between wild-type and CNBD mutant proteins, we co-expressed equal amounts of wild-type and mutant HERG in CHO cells and performed whole cell patch clamp recordings. Compared with wild-type HERG homotetramer, co-expression of wild-type with mutant HERG did not significantly alter current amplitudes (Table I and Fig. 5a). Alterations in V1 ⁄2 observed in currents from cells co-expressing wild-type HERG and CNBD mutants are evidence that support heterotetrameric channel assembly (Table I). When cAMP was applied during whole cell patch clamp measurements, the degree of current amplitude suppression seen in cells transfected with wild-type HERG alone was comparable with cells expressing both wild-type and CNBD mutants (Fig. 5a). This is consistent with our finding that cAMP-dependent current suppression is because of PKA-mediated phosphorylation of HERG (12).
Association of KCNE2 Uncovers an LQT2 Phenotype for CNBD HERG Mutants-The current belief is that HERG is complexed to small subunits of the KCNE family to form native I kr channels in vivo (4,5). Furthermore, we reported that both KCNE1 and KCNE2 altered the effects of cAMP regulation of the HERG channel (12). To examine the effects of CNBD mutations on I kr , we co-expressed wild-type HERG and CNBD mutants with KCNE2. KCNE2 failed to induce currents from the functionally defective CNBD mutants in the absence of co-expressed wild-type HERG. The presence of KCNE2, however, did confer a suppressive effect on CNBD mutants when co-expressed with wild-type HERG. The exception to this was the R823Q mutant that can sustain current as a homotetramer (Table I and Fig. 5b). Compared with expression of wild type alone, current density for co-transfection with V822M, R823W, and ⌬NBD were decreased 35.3%, 49.6, and 44.6%, respectively. Whereas this is a significant suppression, it does not represent a completely dominant-negative effect as seen with several other LQT2 mutants (24,31). To exclude the possibility that co-expression KCNE2 with wild-type and mutant HERG would lower the expression of channel protein, we examined the abundance of total HERG protein by immunoblot. As seen in Fig. 5c, the total HERG protein abundance was comparable in the wild-type alone and mutant co-expression groups when KCNE2 was present. Assuming random assembly of an equal amount of wild-type and mutant subunits a binomial distribution would predict that a completely dominant mutation should reduce the current amplitude by Ͼ90% whereas the CNBD mutants range from ϳ35-50% current reduction. The cAMP effects on wild-type and mutant heteromultimers are altered when KCNE2 is co-expressed (Table I and Fig. 5b). In cells expressing WT-HERG-KCNE2, cAMP induced a leftward shift in voltage-dependent activation and did not significantly alter current amplitudes, which is consistent with our previous report (12). Addition of cAMP, however, causes an augmentation of current density for WT-V822M-KCNE2 channel and a suppression of the others. The V1 ⁄2 for WT-CNBD mutant-KCNE2 were all shifted to more negative voltages by cAMP. Fig. 6 shows a more detailed analysis of the effects of the LQT2 mutant V822M on WT HERG current. CHO cells were transfected with constant amount (5 g) of wild-type HERG cDNA with varying ratios of V822M. Expression of V822M alone produced no current (Fig. 1b), whereas co-transfection of equal amounts of wild-type and V822M cDNA (50% wild type) yielded a current similar in magnitude to that of wild-type HERG (Fig. 5a and Table I). Significant current suppression was only noted when the fraction of transfected mutant cDNA to wild-type cDNA was greater than 0.5. When current ampli-tudes were plotted against the fraction of wild-type HERG cDNA (WT/[WT ϩ V822M]), the data best fit a model where mutant subunits exert no dominant effect over co-assembled wild-type subunits. This is described by the binomial expression for tetramer association, I ϭ I WT (1 Ϫ (f mut ) 4 ), where I is the macroscopic current amplitude, I WT is the current amplitude when all subunits are wild type and f mut is the fraction of subunits that are mutant (as shown by the solid curve on Fig.  6c). This equation is reduced from the general expression, where f WT is the fraction of subunits that are wild type, f mut is the fraction of mutant subunits, and Z x is the fraction of wildtype current passed by channels with x mutant subunits. For a completely dominant-negative suppression, Z x will be 0 whenever x Ն 1, and the binomial expression would be reduced to I ϭ I WT ϫ (f WT ) 4 (Fig. 6c, dotted curve). When co-expressed with KCNE2, however, mutant HERG subunit showed partial dominant-negative suppression of current (Fig. 5b). Fig. 6c also showed the current density at varying ratios of V822M HERG associated with KCNE2 with a current expression intermediate between a completely dominant and non-dominant phenotype. These data points were not well fit to simple binomial expressions where the Z x was varied to values between 0 and 1. If we assume that through the association of KCNE2, mutant HERG subunits become dominant-negative, then the predicted current density will be best described by a function where a fraction of channels that are assembled with KCNE2 shows a dominant-negative effect, and the remaining channels behave in a completely non-dominant manner. Such a model may be expressed as where P is the fraction of mutant HERG associated with KCNE2. Our experimental data are best fit to this equation with P ϭ 0.36 (Fig. 6c, dashed curve) corresponding to 36% of HERG channels associated with KCNE2.
cAMP Binding Capacity of Wild-type HERG Protein and CNBD Mutations-Our previous work demonstrated that cAMP binds to HERG protein with a dissociation constant of ϳ 40 M (12). We measured the specific [ 3 H]cAMP binding of wild-type HERG and the CNBD mutants in the presence and absence of excess unlabeled cAMP. As shown in Fig. 7, in the presence of 400 M unlabeled cAMP, only the wild-type HERG protein demonstrated a significant decrease in cAMP binding (p Ͻ 0.05), indicating a specific affinity for cAMP. The LQT2 mutant V822M and CNBD deletion mutant ⌬NBD did not demonstrate specific binding to cAMP. The binding capacity for R823Q and R823W were significantly reduced compared with wild-type HERG. The loss of specific [ 3 H]cAMP binding capacity of HERG CNBD mutants suggests that the ability to bind cAMP is essential for normal channel function. In this study we have examined the function of CNBD in HERG K ϩ channels in a heterologous mammalian expression system. We selectively introduced mutations at the key sites in the cyclic nucleotide binding domain, which we predicted would disrupt cAMP binding. With the exclusion of R823Q, we found that the CNBD mutants do not form functional channels as homotetramers despite normal protein expression. Immunoblot analyses and microscopy of GFP fusion proteins demonstrated that CNBD mutant HERG surface expression was similar to wild-type HERG at physiological temperature. Although reduced incubation temperatures have been shown to rescue several other LQT2 mutants (24,28), this failed to restore function to our CNBD mutants. Thus, defective protein transport to the plasma membrane does not appear to be the major reason for the functional defect in CNBD mutants. Furthermore, we show that CNBD mutant proteins have a decreased or absent capability to bind cAMP when compared with wild-type HERG. This suggests that the capacity to bind cAMP may be essential for normal HERG channel function.
Although CNBD mutant homotetramers did not express voltage-gated K ϩ currents, when co-expressed with wild-type HERG, functional heterotetrameric channels were detected. Unlike several other LQT2 mutations, the CNBD mutants that we examined demonstrated no dominant-negative suppression of wild-type current. This finding is consistent with previous reports of other C-terminal mutations (26,28). The resulting current density values in our experiments co-expressing V822M with wild-type HERG at varying molar ratios is best fit to a model where mutant subunits are entirely non-dominant. We speculate that only one subunit capable of binding cAMP is needed for full channel function. Whether cAMP must actually bind to the channel for it to be active, however, is still not known.
Our model of a single cAMP binding subunit resulting in a functional channel has similarities to the activation of PKA and cyclic nucleotide-gated channels. The regulatory subunit of PKA has two cAMP binding sites, and mutational analysis demonstrates that activation of type I PKA requires only one of the cAMP binding sites to be capable of binding cAMP for preservation of function (14). Cyclic nucleotide-gated ion channels also contain a CNBD at the C terminus, analogous to HERG. Binding of a cyclic nucleotide to a single subunit is sufficient to produce significant activation of the channel. All four channel subunits, however, must bind to cyclic nucleotide for full current activation (32).
How the CNBD affects channel activity is still unknown. One possibility is that CNBD binding of cAMP is required for activation of the channel through a conformational change in the channel protein. Kupershmidt et al. (33) reported that upon expressing a recombinant HERG that entirely lacked the CNBD, a I kr -like current was produced. Deletion of the entire CNBD however, is more likely to cause ancillary structural changes and channel perturbations in addition to loss of cAMP binding compared with point mutations that leave the CNBD intact but unable to bind cAMP. Liu et al. (32) have proposed a model where the CNBDs within a tetramer of cyclic nucleotide-gated channels act as ligand-bound dimers. Such a situation seems unlikely for HERG given the absence of dominant current suppression when CNBD mutants were co-expressed with wild-type HERG. For HERG CNBD to act as ligand-bound dimers we would have expected that increasing ratios of V822M/wild-type HERG would have suppressed the current in a fashion consistent with a 2-fold stoichiometry.
Another role that CNBD may play is to modulate HERG FIG. 5. Functional co-expression of CNBD mutants and wildtype HERG. a, summary tail current density data from cells expressing wild-type HERG (WT) or wild-type HERG with CNBD mutants in a 1:1 molar ratio. Open bars represent the current amplitude at baseline and filled bars show the amplitude after cAMP treatment. b, similar to a, but for cells also co-expressing KCNE2 plasmid. *, p Յ 0.05 by t test versus WT before treatment with cAMP. c, immunoblot analysis of HERG and KCNE2 protein expression. Cells were co-transfected with 1 g of untagged wild-type HERG cDNA, 1 g of Myc-tagged wild-type or CNBD mutant HERG cDNA and 1 g of HA-tagged KCNE2 cDNA. The membrane was probed with HERG-specific polyclonal antiserum to demonstrate the total HEG protein abundance (top gel). Re-probing with anti-Myc antibody shows the expression of the Myc-tagged subunits (middle gel) and re-probing with anti-HA antibody shows the expression of KCNE2 (bottom gel). current by interacting with KCNE2. The interaction between HERG and a small integral membrane subunit such as KCNE2 can alter HERG channel current to more resemble native I Kr (5). Moreover, we have shown that the co-expression of either minK or KCNE2 can significantly alter the cAMP-dependent response of HERG activity to favor direct effects as opposed to PKA-mediated effects (12). Although co-expression of CNBD mutant with WT HERG does not suppress current in a dominant-negative fashion, the addition of KCNE2 results in a mutant-dependent decrease in peak current density of about 40%. These results support a role for the CNBD of HERG in KCNE2 control of I Kr . Clearly, further experiments are necessary to define the functional and structural interaction between the CNBD and KCNE2 in modulation of channel activity. Furthermore, it now appears that KCNE2 may play a pivotal role in mechanisms that create the LQTS phenotype for some LQT2 mutations. That the CNBD mutants were only partially dominant over wildtype function may be consistent with reports of another CNBD LQT2 mutant (26), and the observation that the clinical presentations of C-terminal mutations are less severe than those involving transmembrane and pore regions of HERG (7). Our present data however, suggest that association of KCNE2 with the HERG channel uncovers the dominant phenotype of CNBD LQT2 mutants. The degree of suppression we observed with differing ratios of mutant and wild-type HERG are best explained by multiple channel populations, some complexed with KCNE2 and some not. The stoichiometry of the KCNEs complexed to their respective ␣-subunits is not absolutely known; therefore, our results may alternatively represent variable degrees of mutant suppression resulting from different combinations of subunits.