Molecular and functional characterization of common polymorphisms in HERG (KCNH2) potassium channels

Blake D. Anson, Michael J. Ackerman, David J. Tester, Melissa L. Will, Brian P. Delisle, Corey L. Anderson, Craig T. January

Abstract

Long QT syndrome (LQTS) is a cardiac repolarization disorder that can lead to arrhythmias and sudden death. Chromosome 7-linked inherited LQTS (LQT2) is caused by mutations in human ether-a-go-go-related gene (HERG; KCNH2), whereas drug-induced LQTS is caused primarily by HERG channel block. Many common polymorphisms are functionally silent and have been traditionally regarded as benign and without physiological consequence. However, the identification of common nonsynonymous single nucleotide polymorphisms (nSNPs; i.e., amino-acid coding variants) with functional phenotypes in the SCN5A Na+ channel and MiRP1 K+ channel β-subunit have challenged this viewpoint. In this report, we test the hypothesis that common missense HERG polymorphisms alter channel physiology. Comprehensive mutational analysis of HERG was performed on genomic DNA derived from a population-based cohort of sudden infant death syndrome and two reference allele cohorts derived from 100 African American and 100 Caucasian individuals. Amino acid-encoding variants were considered common polymorphisms if they were present in at least two of the three study cohorts with an allelic frequency >0.5%. Four nSNPs were identified: K897T, P967L, R1047L, and Q1068R. Wild-type (WT) and polymorphic channels were heterologously expressed in human embryonic kidney cells, and biochemical and voltage-clamp techniques were used to characterize their functional properties. All channel types were processed similarly, but several electrophysiological differences were identified: 1) K897T current density was lower than the other polymorphic channels; 2) K897T channels activated at more negative potentials than WT and R1047L; 3) K897T and Q1068R channels inactivated and recovered from inactivation faster than WT, P967L, and R1047L channels; and 4) K897T channels showed subtle differences compared with WT channels when stimulated with an action potential waveform. In contrast to K897T and Q1068R channels, P967L and R1047L channels were electrophysiologically indistinguishable from WT channels. All HERG channels had similar sensitivity to block by cisapride. Therefore, some HERG polymorphic channels are electrophysiologically different from WT channels.

  • ion channels
  • genetics
  • long QT syndrome
  • arrhythmia
  • sudden death

congenital long qt syndrome (LQTS) is a primary ion channelopathy characterized by abnormalities involving several cardiac ion channel α- or β-subunits or ankyrin-B, with K+ channel defects accounting for the majority of LQTS. LQT2 is caused by mutations in human ether-a-go-go-related gene (HERG; KCNH2), which encodes the α-subunit of the channel mediating the rapidly activating component of the delayed rectifier K+ current (IKr) (5, 21, 29). Most LQT2-causing gene defects are missense mutations, resulting in pathogenic single amino acid residue changes. The functional consequence of LQT2-linked mutations is thought to be a net reduction in IKr leading to a diminished “repolarization reserve” (19).

In addition to rare mutations, common polymorphisms also exist. Common polymorphisms have been defined as nucleotide substitutions found in both control and patient populations, usually at a frequency of ∼1% or greater (31). When viewed in the context of pathological mutations, the presence of common nonsynonymous single nucleotide polymorphisms (nSNPs) in apparently healthy populations suggests that they are well tolerated and likely to have wild-type (WT)-like physiology. However, the identification of common nSNPs in the cardiac SCN5A Na+ channel and the MiRP1 K+ channel β-subunit that alter channel physiology and drug sensitivity has challenged this viewpoint (25, 28). Indeed, these particular nSNPs in critical channels of the heart have a functional phenotype in vitro and may mediate genetic susceptibility to fatal ventricular arrhythmias in the setting of myocardial infarction or exposure to QT-prolonging medications.

Two nSNPs have been found within HERG (10, 11, 31). While more prevalent than disease-linked mutations, polymorphisms may produce subtle functional changes in HERG channel physiology that may alter the QT interval. In fact, the most common nSNP identified to date, HERG K897T (lysine, K, to threonine, T, at amino acid position 897) has been associated with altered channel biophysics and QT interval, although results differ between investigative groups (3, 10, 17, 18, 23).

In the present work, we studied whether nSNPs alter HERG channel function. To test this, we functionally characterized four nSNPs found in genomic DNA derived from a population-based cohort of sudden infant death syndrome (SIDS) and subsequently in two anonymous reference allele cohorts of distinct ethnicity (100 African-American and 100 Caucasian individuals) (1). Two of these amino acid variants were the previously identified arginine to leucine at amino acid position 1047 (R1047L) and K897T polymorphisms. The other two variants, proline to leucine at amino acid position 967 (P967L) and glutamine to arginine at amino acid position 1068 (Q1068R), are novel nSNPs. Western blot and whole cell patch-clamp techniques were used to investigate the effects of the amino acid changes on HERG channels heterologously expressed in a human cell line. The data show that some polymorphisms alter HERG channel physiology.

MATERIALS AND METHODS

Allelic variants.

Genomic DNA was derived from three separate cohorts: 1) a population-based study of 93 SIDS cases (1); 2) 100 healthy, anonymous African-American individuals; and 3) 100 healthy, anonymous Caucasian individuals. Genomic DNA from healthy individuals was obtained from the Human Genetic Cell Repository sponsored by the National Institute of General Medical Sciences and the Coriell Institute for Medical Research (Camden, NJ) (1). Allelic variants in HERG were identified using PCR amplification, denaturing high-performance liquid chromatography, and DNA sequencing (1). In this study, a nSNP was considered common if present in at least two of the three study cohorts with an allelic frequency >0.5%.

Site-directed mutagenesis and heterologous expression.

The appropriate nucleotide changes (A2690C, C2900T, G3140T, and A3203G) resulting in K897T, P967L, R1047L, and Q1068R polymorphisms, respectively, were engineered into WT HERG cDNA in the pCDNA3 vector (Invitrogen; Carlsbad, CA), and the integrity of the construct was verified by DNA sequencing. WT and polymorphic HERG channels were stably and transiently expressed in human embryonic kidney (HEK)-293 cells as described previously (32). HEK-293 cells were cultured in modified MEM (GIBCO/Invitrogen; Carlsbad, CA) with G418 (Invitrogen) antibiotic and kept at 37°C in 5% CO2.

Western blotting, electrophysiology, and cisapride.

HEK-293 cells were grown to similar confluency before transient transfection with equal amounts of cDNA. Forty-eight hours after transfection, cells expressing each channel variant were prepared in parallel for Western blot analysis as previously described (32). Equal volumes of cells or total protein were electrophoresed on a 7.5% SDS-polyacrylamide gel, transferred to nitrocellulose membranes, and probed with an antibody specific for the distal COOH terminal of HERG (32).

A tight-seal whole cell recording technique was used for biophysical analysis of WT and polymorphic HERG channels (33). Pipettes had resistances between 1.5 and 2.5 MΩ when filled with recording solution. Series resistance compensation was ≥70% in all experiments. Cell capacitance was estimated from the area within the capacitance transient. During action potential voltage-clamp recordings, HERG currents were separated from endogenous HEK-293 currents through off-line subtraction of the current response after application of 5 μM E-4031 (a specific HERG channel blocker) from the response obtained before drug application (33). Data were acquired with an Axopatch-2C amplifier controlled by Clampex 8.0 (Axon Instruments; Union City, CA). Current traces fit with exponential functions were acquired at 20 kHz with all other data acquired at 2 kHz.

Cisapride (Research Diagnostics; Flanders, NJ) stock concentration was 10 μM in ethanol. Working concentrations were obtained through serial dilutions with recording saline and used within 48 h. E-4031 (Alomone Labs; Jerusalem, Israel) was diluted in recording saline to a working concentration of 5 μM.

Statistics.

Data are given as means ± SE. ANOVA analyses determined whether significant biophysical differences existed between the channel variants at P < 0.05. When the ANOVA yielded a significant F-value, differences between specific channel variants were ascertained through pair-wise comparisons with Fisher's protected least-significant difference (LSD) test, which takes into account the effect of multiple comparisons. For the test, group i was determined to be different from group j if |ȳiȳj| ≥ LSD, where ȳi and ȳj are the means of groups i and j and LSD is calculated as (tα/2, dferror) × MSerror(1/ni+1/nj), where tα/2 is the t-statistic at P < 0.05/2, dferror is the degree of freedom (total number of measurements in the pairwise comparison − 2), MSerror is the intragroup error calculated from ANOVA, and ni and nj are the numbers of measurements in groups i and j, respectively. Differences in allelic frequency were analyzed using Fisher's exact test.

RESULTS

Identification of four HERG polymorphisms.

Four polymorphisms were detected in multiple survey cohorts (Table 1). The K897T and R1047L polymorphisms were found with allelic frequencies for the minor allele of 13.8% and 1.5%, respectively, for the 586 alleles examined, similar to previous reports (10, 11, 31). However, the minor allele for both nSNPs was significantly more common among Caucasians than African Americans in both the reference cohort and population-based SIDS cohort. For example, the allelic frequency for T897 was 24% in the Caucasian cohort compared with 4% in the African-American cohort (P < 0.05). In addition, 6% of the Caucasian cohort were homozygous for T897. No African Americans in this study were homozygous for T897. Two novel polymorphisms, P967L and Q1068R, were each found in one of 34 African-American SIDS infants and 1 of 100 African-American controls, yielding an allelic frequency among the African American samples of 0.7%. Neither of these two variants were seen in the Caucasian samples.

View this table:
Table 1.

HERG polymorphisms in the study cohorts

Channel processing and trafficking.

Western blot analysis of WT, K897T, P967L, R1047L, and Q1068R protein is shown in Fig. 1A. Western blot data are shown for similar cell numbers (top) or equal amounts of total protein (bottom) loaded into each lane. All samples show the 135- and 155-kDa bands associated with normal HERG channel processing (32) and both blots show similar patterns. Thus polymorphic channels undergo biochemical processing similar to WT channels.

Fig. 1.

Channel processing and expression levels. A: Western blot analysis of wild-type (WT), K897T, P967L, R1047L, and Q1068R human ether-a-go-go-related gene (HERG) protein (lanes 1–5, respectively). Immunoblots show approximately equal cell numbers (top) or equal amounts of total protein (bottom) loaded into each lane. The bars in A and B represent the 135-kDa marker and the arrows mark the location of the 135- and 155-kDa bands. B: mean peak Itail density for each channel variant. The inset shows the voltage-clamp protocol (top trace) and representative WT current data (bottom trace). Peak tail current (Itail) (arrow, inset) was normalized to cell capacitance. Scale is 400 pA by 0.5 s.

Current levels for WT and polymorphic channels were assessed 24 h after transient transfection with equal amounts of appropriate cDNA (Fig. 1B). Tail currents (Itail) were measured by depolarizing cells from a holding potential of −80 to 20 mV for 2 s, followed by repolarization to −50 mV for 2.85 s to elicit Itail. Current density was calculated as peak Itail normalized to cell capacitance. Mean peak Itail densities for WT, K897T, P967L, R1047L, and Q1068R channels were 94.4 ± 18.3, 57.3 ± 7.2, 116.9 ± 18.3, 129.5 ± 14.5, and 126.1 ± 13.9 pA/pF (n = 19, 22, 20, 20, and 20 cells), respectively. K897T current density was not significantly lower than WT levels (P = 0.09) but was lower than that of P967L, R1047L, and Q1068R (P < 0.05). Thus differences exist between some Itail densities.

Channel biophysical properties.

During voltage-clamp protocols, as well as cardiac action potentials, HERG channels undergo distinct voltage- and time-dependent gating transitions. When heterologously expressed in HEK-293 cells and subjected to voltage-clamp analysis, WT and polymorphic channels showed qualitatively similar current waveforms (Fig. 2A). Cells were depolarized from a holding potential of −80 mV to voltages between −70 and 50 mV in 10-mV increments for 3 s, followed by a step to −50 mV for 2.85 s to elicit Itail. All channel variants exhibited outward currents in response to steps positive to −50 mV that peaked around 0 mV, declined upon further depolarization, and showed a prominent Itail upon repolarization to −50 mV.

Fig. 2.

Current-voltage relations. A: families of HERG current traces from channel variants recorded with the voltage-clamp protocol shown in the inset (see text). Scale is 200 pA by 1 s. B: normalized mean activation relations for channel variants. C and D: half-maximal voltage (V1/2) and slope factor derived from Boltzmann fits for each channel variant.

Activation current-voltage (I-V) relations for WT and polymorphic channels were determined by normalizing peak Itail obtained with the previously described series of activating steps to the maximal peak Itail. Normalized Itail values were averaged for each channel variant and plotted as a function of the activating voltage step (Fig. 2B). The voltage at which peak current was half-maximal (V1/2) and the slope factor of the relation for each channel variant were determined by fitting single cell activation I-V plots with the Boltzmann distribution (Fig. 2, C and D). Mean V1/2 values for WT, K897T, P967L, R1047L, and Q1068R channels were −10.4 ± 2.3, −16.2 ± 1.3, −14.7 ± 1.0, −9.7 ± 0.9, and −12.3 ± 1.2 mV with slope factors of 6.6 ± 0.3, 6.4 ± 0.1, 7.0 ± 0.1, 6.8 ± 0.2, and 5.9 ± 0.2 (n = 14, 10, 12, 13, and 9 cells), respectively. The V1/2 for K897T channels was different from that of WT and R1047L channels (P < 0.05) and the slope factor of Q1068R channels was different from that of P967L and R1047L channels (P < 0.05).

To measure channel inactivation, cells were depolarized from a holding potential of −80 to 50 mV for 1.5 s, hyperpolarized to −100 mV for 2.5 ms, and then stepped to test voltages between 60 and −40 mV in 10-mV decrements for 1.5 s. The test voltages generate large amplitude outward currents, the decline of which (arrow, Fig. 3A, inset) was fit as a single-exponential process to derive a time constant for channel inactivation (33). Inactivation time constants were plotted as a function of test voltage (Fig. 3A), and statistical analyses showed differences between channel variants (n = 13, 12, 12, 13, and 9 cells for WT, K897T, P967L, R1047L, and Q1068R channels, respectively). The most striking difference was that both K897T and Q1068R channels inactivated more rapidly than WT, P967L, or R1047L channels (P < 0.05) while being indistinguishable from each other. Specifically, K897T channel inactivation was faster than WT channels between −30 and 60 mV, faster than P967L channels between −40 and 60 mV, and faster than R1047L channels between −30 and 60 mV (P < 0.05). Q1068R channel inactivation was faster than WT channels at −20 mV and between 0 and 60 mV, faster than P967L channels between −30 and 60 mV, and faster than R1047L channels between −20 and 60 mV (P < 0.05). No significant differences were present between P967L, R1047L, or WT channel inactivation.

Fig. 3.

Channel inactivation and deactivation. A: channel inactivation. Mean time constants (τ) for channel inactivation are plotted as a function of test voltage. Inset shows a portion of the voltage-clamp protocol (top traces) and representative currents (bottom traces) highlighting inactivation (arrow). Scale is 1 nA by 10 ms. B: channel recovery from inactivation. Mean time constants for channel recovery from inactivation are plotted as a function of test voltage. Inset shows a portion of the voltage-clamp protocol (top traces) and representative currents (bottom traces) highlighting channel recovery (arrow). Symbols are as in A. Scale is 500 pA by 10 ms. C: channel deactivation. Mean time constants for channel deactivation are plotted as a function of test potential. Inset shows a portion of the voltage-clamp protocol (top traces) and representative currents (bottom traces) highlighting channel deactivation (arrow). Symbols are as in A with open and filled symbols representing the slow and fast time constants, respectively. Scale is 500 pA by 100 ms.

Recovery from inactivation was measured by depolarizing cells to 50 mV for 1.5 s, followed by hyperpolarizing test voltages between −120 and −30 mV in 10-mV increments for 3 s. The initial phase (“hook”) of Itail (arrow, Fig. 3B, inset) reflects recovery from inactivation and was fit as a single-exponential process (33). When the time constants of recovery from inactivation were plotted as a function of test voltage (Fig. 3B), statistical differences between channel variants were present (n = 8, 9, 10, 11, and 9 for WT, K897T, P967L, R1047L, and Q1068R channels, respectively). The most striking finding was that similar to their inactivation properties, K897T and Q1068R channels recover from inactivation faster than WT, P967L, and R1047L channels (P < 0.05) without being significantly different from each other. Specifically, K897T channel recovery was faster than WT channels between −120 and −50 mV, faster than P967L channels at −110 mV and between −80 and −50mV, and faster than R1047L channels between −120 and −110 mV and −80 and −50 mV (P < 0.05). Q1068R channel recovery was faster than WT channels between −120 and −30 mV, faster than P967L channels between −110 and −30 mV, and faster than R1047L channels between −120 and −30 mV (P < 0.05). No significant differences were found between WT, P967L, and R1047L channels.

Deactivation was measured using the same voltage-clamp protocol as recovery from inactivation except that the decay of Itail (arrow, Fig. 3C, inset) was fit as a double-exponential process (33). The fast and slow time constants derived from the fits represent the fast and slow components of channel deactivation. The mean fast and slow deactivation time constants were plotted as a function of test voltage and showed no consistent difference between the five channel variants (Fig. 3C).

An action potential voltage-clamp protocol was used to further compare the biophysical properties of K897T and WT channels (Fig. 4). As previously described (33), depolarization by the action potential protocol first modestly increases HERG current through channel activation and subsequent inactivation. During repolarization, HERG current increases further to its maximal level through rapid recovery of channels from inactivation. After repolarization, HERG current gradually decreases through channel deactivation. For each cell, current was normalized to total charge (pC) calculated by integrating the area under the current response (from 0 to 700 ms; upward arrows, Fig. 4A). For comparisons between WT and K897T channels, normalized current responses were averaged for all cells expressing each channel type (n = 8 cells each). The mean current response for WT and K897T HERG channels is shown in Fig. 4A, with the voltage-clamp protocol superimposed in light gray. Recordings from single cells expressing WT (top) or K897T (bottom) channels are shown in the left insets. Whereas the normalized peak current values are virtually identical (4.54 ± 0.24 and 4.49 ± 0.21 pA/pC for WT and K897T, respectively), the two waveforms are not. Subtracting the normalized mean WT response from that of K897T (Diff trace, Fig. 4A, right inset) suggests that K897T currents are smaller during the upstroke of the current response with depolarization and larger during the decay of the current response after repolarization relative to WT currents. The statistical significance of this difference was analyzed by partitioning the current responses into sequential 100-ms windows (i.e., 0–100 ms, 100–200 ms, etc), determining the time point of maximal difference between the two current waveforms (downward arrows, Fig. 4A), and then assaying whether K897T was smaller during the upstroke and larger during the downstroke of the current response relative to the WT current level. During the upstroke of the current response, K897T current levels of 2.20 ± 0.15 and 3.43 ± 0.17 pA/pC at 256.3 and 329.2 ms, respectively, were significantly smaller than the corresponding WT current levels of 2.53 ± 0.06 and 3.80 ± 0.11 pA/pC (P < 0.05). During the decay of the current response after repolarization, the K897T current level of 1.66 ± 0.28 pA/pC at 418.3 ms was significantly larger than the WT current level of 1.12 ± 0.09 pA/pC (P < 0.05). All other assayed time points showed statistically similar current levels (Fig. 4B).

Fig. 4.

Action potential voltage clamp. A, middle: mean normalized current response of WT and K897T channels when subjected to a rabbit ventricular action potential voltage-clamp protocol (light gray). HERG currents are shown as E-4031-subtracted records. Upward arrows delineate the time period over which current was integrated for charge calculation and downward arrows denote the position of maximal difference between the two responses for sequential 100-ms windows (see text). *Positions of significant differences. Scale is 0.5 pA/pC by 50 ms. Left insets show responses from single cells expressing WT (top) or K897T (bottom) channels. Right inset shows the difference between the mean normalized WT and K897T response currents superimposed over the action potential waveform. B: mean normalized action potential current response values for WT and K897T channels at time points of greatest difference between the two current responses over sequential 100-ms windows.

Cisapride sensitivity.

HERG channel block is a common mechanism for drug-induced LQTS. The drug binding domain is located in the pore-S6 region of HERG (12, 14), whereas heightened sensitivity to block has been linked to a polymorphism in the MiRP1 subunit protein (25). We tested whether high-affinity drug block of HERG channels was altered by the HERG polymorphisms. Cisapride is a prokinetic gastrointestinal drug that was voluntarily withdrawn from the marketplace because of acquired LQTS due to high-affinity HERG channel block (7). We studied its ability to block WT, K897T, P967L, R1047L, and Q1068R channels by using a protocol similar to that of Mohammad and colleagues (15). From a holding potential of −80 mV, channels were activated by a depolarizing step to 10 mV for 20 s, followed by a repolarizing step to −50 mV for 2.5 s to elicit Itail. After control data were obtained, each cell was exposed to progressive increases of 2, 5, 10, and 50 nM cisapride, and residual (steady-state block) Itail was averaged for all cells expressing each channel variant (Fig. 5A). Normalized concentration-response data from each cell were fit with the Hill equation to obtain the concentration at which peak Itail was reduced by 50% (IC50). The averaged IC50 values for each channel variant were 4.3 ± 0.2, 4.5 ± 0.4, 4.5 ± 0.6, 3.9 ± 0.3, and 6.0 ± 1.0 nM with mean Hill coefficients of 0.97 ± 0.06, 1.00 ± 0.03, 0.97 ± 0.05, 0.99 ± 0.08, and 0.99 ± 0.04 for WT, K897T, P967L, R1047L, and Q1068R channels, respectively (Fig. 5, B and C; n = 5, 4, 5, 5, and 5 cells, respectively). There were no significant differences in either the IC50 values or Hill coefficients.

Fig. 5.

Channel sensitivity to cisapride. A: normalized residual peak Itail (symbols) for WT and polymorphic channels recorded in the presence of 0, 2, 5, 10, and 50 nM cisapride fit to the Hill equation (solid lines). The inset shows the voltage-clamp protocol (top) and representative currents (bottom) highlighting peak Itail (arrow) at each drug concentration. Scale is 200 pA by 0.5 s. B and C: mean IC50 and Hill coefficient values, respectively, for WT and polymorphic channels.

DISCUSSION

We report here the first functional characterization of four amino acid coding variants or nSNPs within the HERG K+ channel gene and compared their biochemical and biophysical properties to those of WT HERG channels. The polymorphisms were identified by surveying three cohorts: a population-based cohort of 93 SIDS cases and genomic DNA derived from two anonymous African-American and Caucasian cohorts containing 100 apparently healthy individuals each. Two of the polymorphisms, HERG K897T and R1047L, have been previously identified and were found here with cumulative allelic frequencies of 13.8% and 1.5%, respectively. The identification of K897T and R1047L by multiple laboratories clearly supports their classification as common polymorphisms (8, 10, 11, 31). The identification of HERG P967L and Q1068R have not been reported previously and may be more common among African Americans than Caucasians.

Functional analysis is a key step to understanding the impact of amino acid coding variants on protein function, and a strength of our study is that we compared functional characteristics of five channels (four polymorphisms and WT) under identical experimental conditions. Western blot analysis showed the 135- and 155-kDa bands associated with normal HERG channel processing. Thus unlike many LQT2-associated disease-causing mutations that fail to traffic normally (32), all of the polymorphic channels underwent biochemical processing similar to WT channels.

Current density measurements from polymorphic and WT channels showed significant differences. The most important finding was that HERG K897T channels generated less current than the other polymorphic channels, whereas they did not achieve significantly different levels with respect to WT channels.

Biophysical differences between WT and polymorphic channels were also present. The V1/2 of channel activation for K897T is ∼5–6 mV negative to that of WT and R1047L channels, and thus K897T channels open at more negative voltages. Furthermore, K897T and Q1068R channels inactivate and recover from inactivation faster than the other channel variants, showing ∼10- to 20-mV shifts in the voltage dependence of these time constants. Responses to the action potential voltage-clamp protocol suggest that the biophysical changes present in K897T channels may result in slight alterations (10–30%) of HERG current during the action potential. Compared with WT channels, the small decrease in K897T channel current during the action potential could cause a subtle increase in action potential duration, whereas after repolarization the small increase in current could alter terminal repolarization and the subsequent return of excitability. Channel deactivation was not consistently different between WT and polymorphic channels, nor were the IC50 values or Hill coefficients for block by cisapride. Thus in addition to varying from each other, some polymorphic channels (K897T and Q1068R), but not all (P967L and R1047L), have detectable differences when compared with WT channels. Our data also suggest that the allelic frequency of each polymorphism was not a reliable predictor of altered channel function.

The four HERG polymorphisms reside in the distal COOH terminal of the channel protein. Eleven LQT2-linked missense mutations have also been identified within the COOH terminal (2, 6, 13, 22, 27, 31). Three of these, R752W, S818L, and V822M, have been functionally characterized and shown to be trafficking defective (6, 16, 32). A portion of the COOH terminal region containing amino acids 1018–1122 has been shown to be necessary for HERG channel expression (9), whereas an overlapping region consisting of amino acids 1063–1159, as well as the point mutation V822M, disrupt binding to GM-130, a resident Golgi protein thought to be involved in protein trafficking (20). Thus noncontinuous regions of the COOH terminal are involved in regulating channel trafficking and the small differences in current densities observed in this study could arise from perturbations in positive and/or negative regulation of channel trafficking.

HERG channel inactivation is altered by missense mutations in the pore region as well as deletions of the NH2 terminal (24, 26, 30). The effects of K897T and Q1068R demonstrated here provide evidence that the COOH terminal may also affect HERG channel inactivation. Thus, similar to the cardiac SCN5A Na+ channel where the COOH terminal of the channel is host to multiple functional domains that regulate inactivation and current expression (4), the COOH terminal of HERG channels may also be involved in multiple aspects of channel physiology. This unexpected role of the HERG COOH terminal region stresses the importance of functional characterization of genetic variants.

Our data are the first to provide functional analysis regarding the biophysical properties of three of the HERG polymorphisms, P967L, R1047L, and Q1068R. Other studies have provided conflicting data regarding K897T channel biophysical properties. Scherer and colleagues (23) did not detect differences between K897T and WT current amplitude, biophysical properties, or sensitivity to drug block by the antihistamine terfenadine. Bezzina and colleagues (3) also found no differences in current amplitude or channel inactivation properties but did find K897T channels to have a hyperpolarized shift in channel activation as well as faster activation and deactivation kinetics at specific voltages. Paavonen and colleagues (17) did not detect changes in channel activation properties but did find decreased K897T protein levels on Western blot, slower channel deactivation and inactivation kinetics, and a hyperpolarized shift in steady-state inactivation. While our results provide new data, they do not resolve the reported differences, and in part the discrepancies may arise from differences in expression systems and/or experimental protocols.

The potential clinical impacts of P967L, R1047L, or Q1068R are unknown, whereas those associated with K897T status have been reported. Similar to discrepancies in K897T channel biophysics, the reports of K897T clinical phenotypes are not in complete agreement. Laitinen et al. (10) suggested a correlation between K897T and QT interval in female LQT1 patients. Pietilla et al. (18) found K897T to be associated with a longer maximal QTc interval and increased dispersion of ventricular repolarization in females but not males within a random middle-aged study group. Paavonen et al. (17) did not find gender-specific differences in resting QTc intervals for K897T carriers but did show an increased QTc interval during exercise for a subset of carriers with a prolonged baseline QTc. Bezzina et al. (3), on the other hand, found females homozygous for T897 to have shorter QTc intervals than either heterozygous or noncarriers. While our results provide potential mechanisms for altered IKr in K897T carriers (altered channel kinetics and slight changes in current levels), they do not resolve the apparent clinical or gender differences. The different biophysical and clinical phenotypes attributed to HERG K897T highlight the need for consistent experimental protocols when investigating the potentially subtle effects of polymorphisms.

Our data also provide insight in another important area, which is that the four HERG polymorphisms we studied do not directly convey heightened sensitivity to block by cisapride, a drug known to cause acquired LQTS by high-affinity HERG channel block (15). Thus unlike the T8A polymorphism in MiRP1 (a putative HERG channel β-subunit) that conveys heightened channel sensitivity to drug block (25), these HERG α-subunit polymorphisms do not do this. However, subtle modification of IKr by HERG polymorphisms such as K897T, whether through altered current density or channel kinetics, could indirectly contribute to proarrhythmic effects of HERG K+ channel blockers by reducing the repolarization reserve.

In summary, we studied WT and four nonsynonymous single nucleotide polymorphisms found in HERG K+ channels. Functional characterization of these polymorphisms demonstrated their WT-like biochemical processing with normal trafficking to the cell surface membrane. The expressed currents resemble WT current, although small differences in current density and biophysical properties were detectable within some of the polymorphisms. Sensitivity to block by cisapride was similar for WT and all polymorphic channels. These data suggest that some HERG polymorphisms are WT channel like in their physiological properties, whereas others have detectable differences. These findings may provide a basis for new insight into clinical phenotypes and arrhythmia risk prediction of polymorphism carriers; however, further correlative studies are clearly needed.

Limitations.

All polymorphisms were studied as homomeric and not hetermeric channels. Extrapolation of our results to heterozygous carriers must done cautiously and is beyond the scope of this work. DNA samples for these studies were obtained from either a SIDS-based tissue cohort or genomic samples obtained from apparently healthy individuals. Relevant clinical data do not exist for these cohorts to permit comparison of the HERG channel physiology with cardiac phenotypes.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Grants RO1-HL-42569 (to M. J. Ackerman) and HL-60723 (to C. T. January).

Acknowledgments

The authors gratefully acknowledge Dr. Ronald Gagnon, Department of Biostatistics, University of Wisconsin, for statistical consultation.

Footnotes

  • * B. D. Anson and M. J. Ackerman contributed equally to this work.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

REFERENCES

View Abstract