Specific serine proteases selectively damage KCNH2 (hERG1) potassium channels and IKr

Sridharan Rajamani, Corey L. Anderson, Carmen R. Valdivia, Lee L. Eckhardt, Jason D. Foell, Gail A. Robertson, Timothy J. Kamp, Jonathan C. Makielski, Blake D. Anson, Craig T. January

Abstract

KCNH2 (hERG1) encodes the α-subunit proteins for the rapidly activating delayed rectifier K+ current (IKr), a major K+ current for cardiac myocyte repolarization. In isolated myocytes IKr frequently is small in amplitude or absent, yet KCNH2 channels and IKr are targets for drug block or mutations to cause long QT syndrome. We hypothesized that KCNH2 channels and IKr are uniquely sensitive to enzymatic damage. To test this hypothesis, we studied heterologously expressed K+, Na+, and L-type Ca2+ channels, and in ventricular myoctyes IKr, slowly activating delayed rectifier K+ current (IKs), and inward rectifier K+ current (IK1), by using electrophysiological and biochemical methods. 1) Specific exogenous serine proteases (protease XIV, XXIV, or proteinase K) selectively degraded KCNH2 current (IKCNH2) and its mature channel protein without damaging cell integrity and with minimal effects on the other channel currents; 2) immature KCNH2 channel protein remained intact; 3) smaller molecular mass KCNH2 degradation products appeared; 4) protease XXIV selectively abolished IKr; and 5) reculturing HEK-293 cells after protease exposure resulted in the gradual recovery of IKCNH2 and its mature channel protein over several hours. Thus the channel protein for IKCNH2 and IKr is uniquely sensitive to proteolysis. Analysis of the degradation products suggests selective proteolysis within the S5-pore extracellular linker, which is structurally unique among Kv channels. These data provide 1) a new mechanism to account for low IKr density in some isolated myocytes, 2) evidence that most complexly glycosylated KCNH2 channel protein is in the plasma membrane, and 3) new insight into the rate of biogenesis of KCNH2 channel protein within cells.

  • rapidly activating delayed rectifier potassium current
  • enzymes
  • myocyte isolation

selective cleavage of specific ion channel proteins by distinct proteases has been previously recognized, and the resulting alterations in channel function have provided new insights into structure-function relations for these channels. For example, the intracellular application of pronase in axons removes normal Na+ channel inactivation consistent with selective proteolysis (3), and similar gating effects have been shown for inactivating K+ channels (25). In cultured rat neocortical neurons, a hyperpolarization-activated Na+-K+ current is sensitive to extracellular proteolysis by superfusing cells with trypsin (6). In single grassfrog saccular hair cells, papain treatment modifies Ca2+ and K+ currents to alter cell electrophysiological properties (2).

The isolation of cardiac myocytes also can result in abnormal electrophysiological properties. In atrial myocytes, Nattel and colleagues (55) showed that different cell isolation methods altered membrane currents, particularly delayed rectifier K+ channel current, IK. An important component of IK is the rapidly activating delayed rectifier K+ current, IKr, the congenital or pharmacological impairment of which can cause long QT syndrome; yet in an apparent paradox some isolated cardiac myocyte preparations lack IKr or it is very small (for example, see Refs. 27, 46).

In this study, we used electrophysiological and biochemical methods to study the sensitivity and specificity of multiple heterologously expressed K+, Na+, and L-type Ca2+ channels, as well as native K+ currents from ventricular myocytes, to degradation by extracellular proteases. We tested the hypothesis that KCNH2 K+ channels and IKr are uniquely sensitive to degradation by specific protease enzymes. A preliminary report of this work has appeared (40).

MATERIALS AND METHODS

Heterologous Expression: DNA Constructs and Transfection

KCNH2 (hERG1) K+ channels.

KCNH2 subunits coassemble to generate the pore-forming protein for IKr. (26, 44, 50). The generation of human embryonic kidney (HEK-293) cells stably expressing KCNH2 wild-type (WT) or N598Q channels has been previously described (17, 58).

KCNQ1 (hKvLQT1) + KCNE1 (hMinK) K+ channels.

KCNQ1 +KCNE1 subunits coassemble to form the slowly activating delayed rectifier K+ current, IKs (4, 43). A stable cell line expressing KCNQ1 + KCNE1 channels was generated with methods similar to those used to create the KCNH2 cell line (56). Current from coassembled KCNQ1 + KCNE1 channels was easily distinguished, confirming previous reports (4, 43, 56).

KCNJ2 (hKir2.1) K+ channels.

hKCNJ2 cDNA, which encodes the pore-forming protein for the inward rectifier K+ channel, IK1 (54), was cloned from a human heart library (13). COS-1 cells were transiently transfected with 5 μg/ml KCNJ2 cDNA using SuperFect, along with 300 ng/ml of green fluorescent protein (GFP) cDNA in a pRK5 vector (GFP-pRK5, Clonetech, Palo Alto, CA) as a marker for successful transfection (16). Cells were cultured in DMEM and studied 24 h later.

Mouse eag (meag) K+ channels.

HEK-293 cells were transiently transfected with 3 μg/ml of meag cDNA (24, 42) using Superfect (Qiagen) along with 300 ng/ml of GFP-pRK5. Cells were cultured as described for KCNH2-transfected cells and studied within 24 h of transfection.

SCN5a (hH1a) Na+ channels.

hH1a cDNA encodes the α-subunit for human cardiac Na+ channels (22). HEK-293 cells were transiently transfected as previously described (52). Cells were incubated in MEM and studied within 24 h of transfection.

rCav1.2 + hCavβ2cN2 + hCavα2δ1 (L-type) Ca2+ channels.

The cDNAs encoding the pore-forming L-type Ca2+ channel rCav1.2 (rabbit α1c) subunit plus the auxiliary hβ2cN2 and hα2δ1 subunits were transiently transfected in a 1:1:1 molar ratio into HEK-293 cells using a calcium phosphate method (Invitrogen) as previously described (16). Cells were incubated with the cDNAs for 4 h, washed four times with PBS, incubated overnight in DMEM, and studied 24 h after transfection.

Canine Ventricular Myocytes

Canine ventricular myocytes were isolated as previously described (23) using a protocol approved by the University of Wisconsin-Madison. Briefly, following harvest, the hearts were perfused via the coronary circulation with ice-cold nominally Ca2+-free cardioplegia solution. Ventricular tissue chunks were cut into small pieces and agitated in enzyme solution (collagenase, type II, 1 mg/ml, Worthington Chemicals, Lakewood, NJ, plus hyaluronidase, 0.5 mg/ml, Sigma Chemicals, St. Louis, MO) to obtain single ventricular myocytes. Isolated myocytes were stored in Tyrode solution before study. For these experiments, we did not distinguish between cells of subendocardial, midmyocardial, and subepicardial origins.

Patch-Clamp Recording Technique

For electrophysiological study, cultured cells were mechanically harvested by trituration from culture dishes without the use of enzymes, washed twice with standard MEM, and stored in the same medium for later use the same day. Whole cell membrane current was recorded as previously described (20). Pipette resistances were 2–4 MΩ except for recording of Na+ current (INa), where they were 1–1.5 MΩ. Series resistance compensation was used in all experiments, and recordings of (INa) were leak subtracted (52). Computer software (pCLAMP, Axon Instruments, Union City, CA) was used to generate voltage-clamp protocols and acquire data. Patch-clamp amplifier (Axopatch 200A or B, Axon Instruments) data sampling rates varied from 5 to 100 kHz, depending on the ion channel studied. Data anaylsis was performed using pCLAMP (Axon Instruments) and Origin (MicroCal, Northhampton, MA) software. Transfected cells and canine myocytes were studied in small volume chambers mounted on an inverted microscope, and experiments were performed at either 23 ± 1 or 37 ± 1°C. Bath solution flow rates allowed for complete perfusate exchange within 1 min. The effect of an enzyme on ion currents was tested by first recording current for control (no enzyme) conditions followed by the presence or absence of a specific enzyme. In each experiment, after we obtained a stable whole cell recording and initiating the voltage-clamp protocol, control data were recorded for 1.5 min to obtain baseline values. After this, at time 0 (t0), perfusion of the experimental chamber was continued with the control solution or switched to enzyme-containing solution. Current density measurements were normalized to a value recorded for control conditions 1 min before t0.

Solutions

The bath and pipette solutions used in each experimental series have been previously published (13, 16, 39, 52, 56) and are given in Tables 1 and 2.

View this table:
Table 1.

Bath solutions

View this table:
Table 2.

Pipette solutions

Enzymes

Six enzyme preparations were tested (Table 3). Protease XIV, protease XXIV, proteinase K, trypsin, and collagenase are serine peptidases, whereas hyaluronidase is a glycosidase. The enzyme concentrations are typical of those used for cardiac myocyte isolation. Each enzyme was freshly dissolved in the appropriate bath solution used at the time of the experiment.

View this table:
Table 3.

Enzymes

Western Blot Analysis

Western blot analysis of KCNH2 protein was performed using whole cell lysates of transfected HEK-293 cells as previously described (1, 57). Two rabbit polyclonal anti-KCNH2 antibodies were used, one raised against 181 amino acid residues of the COOH terminus (57) and one raised against amino acids 96–270 in the NH2 terminus (Santa Cruz Biotechnology, Santa Cruz, CA). Image density analysis was performed using Bio-Rad GS-700 Imaging Densitometer and Molecular Analyst software (Bio-Rad, Hercules, CA). Each Western blot analysis contained a KCNH2 WT control lane (see Fig. 2B, inset). In each blot, image intensity of the subsequent mature (155-kDa band) protein was normalized to the intensity of the KCNH2 WT control lane.

Statistics

Data are given as average ± SE and were analyzed using a Student's paired t-test with P < 0.05 considered significant. Recovery of KCNH2 current after enzyme exposure was fit with a single-exponential equation, y = y0 + Aex/t, where A is current amplitude, t is a time constant, and y0 is the offset.

RESULTS

Reduction of KCNH2 Current by Specific Proteases

Figure 1A shows control data and the effect of protease XXIV on heterologously expressed KCNH2 channel current (IKCNH2) studied at 23 ± 1°C. As shown in Fig. 1A, inset, cells were depolarized from −80 to 20 mV for 4 s, tail current was elicited by a repolarizing step to −50 mV for 5.7 s, and the protocol was repeated every 15 s. In each cell studied, after obtaining baseline current values for 1.5 min, at t0 (arrow) perfusion of the experimental chamber was continued with Tyrode solution (control) or it was switched to Tyrode solution containing protease XXIV. Figure 1A, inset, shows example IKCNH2 traces recorded from one cell for control conditions and after 7 and 17 min of exposure to protease XXIV. Enzyme exposure eliminated IKCNH2. The figure shows the averaged normalized peak tail IKCNH2 amplitude plotted against time. For control conditions at t0, the peak tail current amplitude was 1,330 ± 251 pA and the holding current was −12 ± 8 pA (n = 4 cells). After 19 min of control recordings, the peak tail current amplitude was 1,140 ± 142 pA (P > 0.05 compared with peak tail current amplitude at t0) and the holding current was −21 ± 15 pA (P > 0.05 compared with holding current at t0), indicating minimal rundown of IKCNH2 and stable recording conditions. In the presence of protease XXIV, IKCNH2 declined with a time to 50% reduction in peak tail current amplitude (t1/2) of 294 ± 49 s (n = 7 cells). In these cells, at t0 the peak tail current amplitude was 949 ± 107 pA and the holding current was −13 ± 6 pA. After 19 min of enzyme exposure, the peak tail current amplitude was nearly abolished to 32 ± 17 pA (P < 0.05 compared with peak tail current amplitude at t0), whereas the holding current was stable at −30 ± 12 pA (P > 0.05 compared with holding current at t0). The stable holding current indicates that protease XXIV did not affect cell membrane and pipette seal integrity under conditions that abolished IKCNH2. The effect of protease XXIV could not be washed out for time periods up to 20 min, the maximum recording duration. In other experiments, we studied untransfected HEK-293 cells and exposed them to protease XXIV using a similar protocol (n = 3 cells), and the enzyme had no effect on the electrical properties of these cells (data not shown).

Fig. 1.

Effect of different enzymes on KCNH2 current (IKCNH2). A: averaged normalized peak tail IKCNH2 amplitude for control conditions (□) and after protease XXIV (▪) exposure studied at 23 ± 1°C. A, inset: voltage-clamp protocol and example IKCNH2 traces for control conditions and after protease XXIV exposure. B: averaged normalized peak tail IKCNH2 amplitude for control conditions (□) and after protease XXIV exposure in normal extracellular Ca (Cao; 1.8 mmol/l, ▪) or 0 mM Cao (▴), protease XXIV washout (normal Cao, ○), or 140 μg/ml protease XXIV (normal Cao, ▾) exposure studied at 37 ± 1°C. B, inset: example IKCNH2 traces for control conditions and after protease XXIV (normal Cao) exposure. C: averaged normalized peak tail IKCNH2 amplitude for protease XIV (•) and proteinase K (Prot K; ⧫) exposure. C, inset: example IKCNH2 traces for control conditions and after protease XIV exposure. D: effect of collagenase type II (○), hyaluronidase (◂), and trypsin (▸) on normalized averaged peak tail IKCNH2. D, inset: example IKCNH2 traces for control conditions and after hyaluronidase exposure.

Figure 1B shows control recordings and the effect of protease XXIV studied at 37 ± 1°C. The procedure was the same as in Fig. 1A. Figure 1B, inset, shows example current traces from one cell for control conditions and after 2 and 5 min of exposure to protease XXIV, which rapidly abolished IKCNH2. The plot of averaged normalized peak tail IKCNH2 for control (n = 5 cells) conditions confirmed minimal current rundown. The addition of protease XXIV at t0 (arrow) resulted in the rapid disappearance of IKCNH2 with a t1/2 of 63 ± 16 s (n = 7 cells). Reducing the protease XXIV concentration by half to 140 μg/ml slowed the disappearance of IKCNH2 (t1/2 of 183 ± 18 s, n = 3 cells). Ca2+ is a cofactor for many enzymes, and with myocyte isolation procedures it is often omitted. Tyrode solution containing protease XXIV (280 μg/ml) that was nominally Ca2+ free slowed only slightly the enzyme-induced decay in IKCNH2 (t1/2 of 94 ± 15 s, n = 7 cells). We tested the effect of brief superfusion of protease XXIV for ∼2 min (washout) followed by return to enzyme-free solution. After the initial decline in IKCNH2, protease XXIV washout resulted in the stabilization of the current at an intermediate amplitude, demonstrating that proteolytic degradation of KCNH2 channels can be graded. In other experiments, after control recordings for 1.5 min, at t0 cells were held continuously at −80 mV while protease XXIV was superfused and after 3 min of enzyme exposure the voltage-clamp protocol was resumed. In these cells, the control peak tail IKCNH2 was 1,678.2 ± 95.8 pA, whereas with the first pulse after protease XXIV exposure the peak tail IKCNH2 had decreased to 339.5 ± 71.4 pA (n = 4 cells, data not shown), which shows that protease XXIV did not require channel activation for the rapid disappearance of IKCNH2. In another set of experiments, changing the holding potential from −80 to −120 mV during the voltage-clamp protocol did not alter the protease XXIV-induced disappearance of IKCNH2 (n = 4 cells, data not shown), which suggests that a marked negative shift in the voltage dependence of IKCNH2 activation or inactivation did not account for its disappearance. Finally, we tested the effect of heat-inactivated protease XXIV (90°C for 10 min). Under these conditions, peak tail IKCNH2 values were similar to control values (data not shown).

Protease XIV and proteinase K also degrade IKCNH2, as shown in Fig. 1C. The procedure was the same as that used in Fig. 1A. Figure 1C, inset, shows example IKCNH2 traces from one cell for control conditions and after 5 min exposure to protease XIV, which rapidly abolished IKCNH2. The plotted data show the effect of protease XIV and proteinase K on averaged normalized peak tail IKCNH2 amplitude studied at 37 ± 1°C. Protease XIV or proteinase K was applied at t0 (arrow) and caused IKCNH2 to disappear with a t1/2 of 63 ± 13 (n = 4 cells) and 197 ± 30 s (n = 4 cells), respectively. At t0 the peak tail IKCNH2 amplitudes were 838 ± 73 pA and 1,268 ± 326 pA, respectively, and the holding currents were −30 ± 9 pA and −40 ± 8 pA, respectively. After 7 min of protease XIV exposure and 12 min of proteinase K exposure, the peak tail IKCNH2 amplitudes were 17 ± 13 and 183 ± 60 pA (P < 0.05 compared with respective peak tail current amplitudes at t0), and the holding currents were −42 ± 8 and −45 ± 9 pA (P > 0.05 compared with respective holding currents at t0). The stability of the holding currents indicates that neither enzyme affected cell membrane and pipette seal integrity under conditions that abolished IKCNH2.

Additional enzymes typically used for myocyte isolation or to harvest cells from culture dishes, collagenase (type II), hyaluronidase, or trypsin, were also studied. Figure 1D, inset, shows IKCNH2 records from one cell recorded for control conditions and after 10 min exposure to hyaluronidase, which had no effect on IKCNH2. Figure 1D shows averaged normalized peak tail IKCNH2 amplitude plotted against time. Trypsin (n = 4 cells) caused a small reduction in peak tail IKCNH2 amplitude, whereas IKCNH2 was insensitive to collagenase type II (n = 4 cells) and hyaluronidase (n = 5 cells) exposure.

Digestion and Reappearance of Mature KCNH2 Protein

The previous findings show that IKCNH2 is rapidly abolished by protease XIV, protease XXIV, and proteinase K exposure without damaging the electrophysiological integrity of the cells. We then tested the hypothesis that the mature KCNH2 channel protein was the target for proteolysis and that it and IKCNH2 would reappear after protease treatment. HEK-293 cells expressing KCNH2 WT channels were exposed to protease XXIV for 30 min, after which enzyme activity was stopped by the addition of 6 mM phenylmethylsulfonyl fluoride and 25 mM EDTA in ice-cold Tyrode solution, followed by three washes in culture media. Cells were then allowed to recover under the original cell culture conditions. At different times during the recovery period, cells were subjected to electrophysiological or Western blot analysis. Figure 2A shows families of IKCNH2 traces recorded at 37 ± 1°C for control conditions and after 4 and 32 h of recovery. From a holding potential of −80 mV, 4-s-long depolarizing steps were applied in 10 mV increments to between −70 and 70 mV, followed by a 5.7-s-long repolarizing step to −50 mV to obtain tail current. Averaged peak tail IKCNH2 density measurements, recorded after steps from 70 to −50 mV, were obtained for control conditions and at different times of reculture and are plotted in Fig. 2B. The data confirm that protease XXIV exposure abolished IKCNH2 and that IKCNH2 gradually reappeared as cells recovered to reach a current density similar to that of control cells. When fit as a single-exponential process, peak tail IKCNH2 density recovered with a time constant of 13.2 h.

Fig. 2.

Degradation and recovery of KCNH2 channel protein and IKCNH2. A: voltage-clamp protocol along with IKCNH2 traces recorded for control conditions and after 4 and 32 h of reculture. B: plot of averaged IKCNH2 peak tail current density recorded for control conditions and at different times during recovery. B, inset: Western blot analysis of KCNH2 wild-type (WT) channel for control conditions and after 30 min protease XXIV exposure probed with the COOH terminus anti-KCNH2 antibody. Filled arrows, protein bands at 135 and 155 kDa. See text for further details.

During biogenesis, KCNH2 WT channels undergo N-linked core- and complex glycosylation at residue N598 to give on Western blot analysis in HEK-293 cells protein bands of ∼135 kDa [immature protein located predominately in the endoplasmic reticulum (ER)] and ∼155 kDa (mature protein), respectively (57). Figure 2B, inset, shows Western blot analysis of the effect of protease XXIV on KCNH2 WT channels detected using a COOH-terminal anti-KCNH2 antibody. WT channels for control conditions (lane 1) show two protein bands of 135 and 155 kDa (filled arrows) as well as a faint protein band at ∼60 kDa. After 30 min of protease XXIV exposure (lane 2), the 155-kDa protein band was eliminated whereas the 135-kDa protein band was unchanged, and a stronger band appeared at ∼60–64 kDa consistent with the COOH terminus-containing degradation products from the 155-kDa protein. After cell reculture, Western blot analysis shows that the degradation products fade and the 155-kDa mature protein band gradually reappears (lanes 3–6). Similar findings were obtained in four additional Western blot analyses of KCNH2 protein from cells exposed to protease XXIV and two Western blot analyses of KCNH2 protein from cells exposed to protease XIV. These findings show that loss of IKCNH2 is associated with the disappearance of the mature KCNH2 protein and the appearance of COOH terminus-containing degradation products, and that this reverses with cell reculture. These findings suggest that exposed extracellular linker peptides (between S1–S2, S3–S4, S5–pore, and pore–S6) of mature KCNH2 channels in the plasma membrane may be the target for proteolytic degradation.

Image density analysis was performed on the 155-kDa bands of the five Western blots of KCNH2 WT channel protein exposed to protease XXIV. Each was normalized to the density of the respective 155-kDa WT control lane (see Fig. 2B, inset). The averaged image density increased progressively with cell reculture time, and when fit as a single-exponential process the 155-kDa band image density recovered with a time constant of 6.4 h.

Enzyme Effects on Other Human K+ Channels

We tested whether other human K+ channels have similar sensitivity to protease enzymes. We studied KCNQ1 + KCNE1 and KCNJ2 channels, which encode the native currents IKs and IK1, respectively. In these and subsequent experiments, we used protease XXIV because it abolished IKCNH2 with the shortest t1/2 (see Fig. 1) and it has fewer amino acid residue peptide bond targets for proteolytic digestion than protease XIV or proteinase K (12, 28). Figure 3A shows the effect of protease XXIV on KCNQ1 + KCNE1 channels. Figure 3A, left inset, shows control KCNQ1 + KCNE1 current (IKCNQ1+KCNE1) traces along with a current-voltage (I-V) relation of the averaged current amplitude measured at the end of the depolarizing steps. The current was elicited from a holding potential of −80 mV with 4-s-long depolarizing steps to between −70 and 70 mV in 10 mV increments, followed by a 1-s-long repolarizing step to −50 mV. Figure 3A, right inset, shows two example IKCNQ1 + KCNE1 traces (step to 60 mV) recorded during control conditions and after 10 min of protease XXIV exposure, and these are nearly identical. The averaged normalized peak IKCNQ1 + KCNE1 measured at the end of a step to 60 mV are plotted with the protocol applied every 15 s. Protease XXIV was added to the bath at t0 (arrow) and had no effect on IKCNQ1 + KCNE1 during 10 min of enzyme exposure (n = 5 cells).

Fig. 3.

Effect of protease XXIV on other ion channel currents. A: KCNQ1 +KCNE1 current (IKCNQ1+KCNE1) insensitivity to protease XXIV (▪). A, insets: example IKCNQ1 + KCNE1 traces. B: current-voltage (I-V) relations for KCNJ2 current (IKCNJ2) measured at the end of depolarizing steps to between −140 and 40 mV for the same cells for control conditions (□) and after 10 min of protease XXIV exposure (▪). C: current traces along with I-V relations of meag (•) and KCNH2 (▪) channel expressing cells. D: insensitivity of meag current (Imeag) to protease XXIV (▪) exposure. E: averaged normalized Na current (INa) for control conditions (□) and with exposure to collagenase (•) or protease XXIV (▪). E, inset: example INa traces for control conditions and after 10 min of protease XXIV exposure. F: insensitivity of L-type Ca2+ channels to protease XXIV (▪). F, inset: example Ba2+ current (IBa) traces for control conditions and after 10 min of protease XXIV exposure.

Figure 3B shows the effect of protease XXIV on KCNJ2 current (IKCNJ2). Figure 3B, inset, shows example IKCNJ2 traces. From a holding potential of −80 mV, cells were polarized for 100 ms to between −140 and 40 mV in 20-mV increments to record both inward and outward IKCNJ2. The I-V relation shows the averaged current measured at the end of the polarizing steps and demonstrates the typical marked inward rectification at more positive voltages. The I-V relations obtained for control conditions and after 10 min of protease XXIV exposure were identical (n = 3 cells).

Enzyme Effects on meag K+ Channels

Mouse eag channels belong to the eag gene superfamily of K+ channels that also include KCNH2 channels. These channels share some structural homology (53), although functionally they gate differently. Compared with KCNH2 channels, meag channels activate more rapidly, inactivate minimally, and the tail current is of small amplitude and decays rapidly (24, 42). We tested whether meag channels were sensitive to digestion by protease XXIV. Figure 3C, insets, show families of original current traces from cells expressing meag or KCNH2 channels. From −80 mV, 4-s-long depolarizing steps were applied from between −70 and 70 mV in 10-mV increments every 15 s. The I-V relation in Fig. 3C shows averaged meag current (Imeag) and IKCNH2 measured at the end of the depolarizing steps and illustrate the different functional properties of the two channel currents. The sensitivity of Imeag to protease XXIV is shown in Fig. 3D. For these experiments, cells were depolarized from −80 to 20 mV for 4 s with the protocol applied at 15-s intervals. Figure 3D, inset, shows example Imeag traces from one cell for control conditions and after 6 min of protease XXIV exposure. The plotted data show the averaged normalized current measured at the end of each step to 20 mV. Protease XXIV had no effect on Imeag (n = 5 cells).

Enzyme Effects on Na+ and Ca2+ Channels

We also studied heterologously expressed Na+ and Ca2+ channels. Figure 3E shows the effect of protease XXIV and collagenase type II on SCN5a Na+ channels. Cells were depolarized from a holding potential of −140 to −20 mV for 100 ms, and the protocol was applied every 6 s. Figure 3E, inset, shows two example INa traces recorded for control conditions and after 10 min of exposure to protease XXIV. Peak INa decreased by ∼7%. The plotted data show averaged normalized peak INa. At t0 (arrow), perfusion with control Tyrode solution was continued (control, n = 4 cells) or switched to Tyrode solution containing protease XXIV or collagenase type II. At the end of 10 min, control INa declined minimally (4%). Protease XXIV or collagenase had slightly greater effects on peak INa, reducing it by 17% (n = 5 cells) and 9% (n = 5 cells), respectively. Figure 3F shows the effect of protease XXIV on the L-type Ca2+ channels. Cells were depolarized from −80 to 20 mV for 50 ms, and the protocol was repeated every 15 s. Example Ba2+ current (IBa) records for control conditions and after 10 min of protease XXIV exposure are shown in Fig. 3F, inset. The plotted data show the normalized (to 1 min before t0) averaged peak IBa, and at t0 (arrow) protease XXIV was added to the Tyrode solution. Protease XXIV had little effect on IBa (n = 4 cells).

Enzyme Effects on IKr, IKs, and IK1 in Native Myocytes

We next tested the effect of protease XXIV on canine ventricular myocytes. Myocytes were held at −50 mV to voltage-inactivate Na+ channels, the Tyrode solution contained nitrendipine (10 μmol/l) to block L-type Ca2+ channels, and the studies were performed at 37°C. Figure 4A shows the voltage-clamp protocol. Depolarizing steps were applied to between −30 and 30 mV for 2 s in 5-mV increments, and tail current was measured after repolarization to −40 mV for 2.85 s, with the pulse sequence applied at 15-s intervals. At the holding potential of −50 mV, the whole cell current was outward, consistent with the outward component of IK1 (arrow; see also IKCNJ2 I-V relation in Fig. 3B). Depolarization resulted in an inward shift in baseline current due to inward rectification of IK1, which was followed by the development of additional outward current consistent with activation of IK, composed of IKr and IKs (arrow). Repolarization was followed by a tail current that gradually decayed, consistent with deactivation of IK, composed mainly of IKr (arrow). For control conditions the averaged peak tail current density measured for the step from 30 to −40 mV was 0.21 ± 0.04 pA/pF (n = 6 myocytes), with a half-maximal activation (V1/2) of −18.8 ± 1.1 mV. After protease XXIV exposure for ∼5 min, the data set was repeated with the principal changes being a decrease in the amplitude of IK activated during depolarizing steps and the complete loss of tail current (n = 6 myocytes). To confirm that protease XXIV had selectively removed IKr, the same experimental procedure was repeated in five additional myocytes, except that E-4031, a selective IKr blocker (5 μmol/l, Fig. 4B) was applied rather than enzyme. For control conditions, the averaged peak tail current density measured after the step from 30 to −40 mV was 0.24 ± 0.02 pA/pF with a V1/2 of −13.1 ± 2.4 mV. As shown in Fig. 4B, the principal effect of E-4031 was a decrease in the amplitude of IK activated during depolarizing steps, and complete loss of the tail current (n = 5 myocytes).

Fig. 4.

Effect of protease XXIV and E-4031 on rapidly activating delayed rectifier K+ current (IKr), slowly activating delayed rectifier K+ current (IKs), and inward rectifier K+ current (IK1). A and B: families of current traces for control conditions and after exposure to protease XXIV (A) or E-4031 (B). For clarity, currents are shown for 10-mV increments. C: effect of protease XXIV (▪) or E-4031 (•) exposure on averaged peak tail IKr density. D: effect of protease XXIV (▪) or E-4031 (•) on averaged holding current density. E and F: I-V relations for outward current density for control conditions (□), after protease XXIV (▪, E) or E-4031 (•, F) exposure, along with the subtraction I-V relation (▴, difference plot).

Between obtaining the control data and the enzyme or drug exposure data shown in Fig. 4, A and B, the voltage-clamp protocol was changed to a single depolarizing step to 20 mV for 2 s followed by repolarization to −40 mV to elicit IKr tail current. Figure 4C shows averaged peak tail IKr density measurements obtained using this protocol. Control data were recorded for 2.5 min, and at t0 (arrow) perfusion was switched to Tyrode solution containing protease XXIV or E-4031. Protease XXIV rapidly abolished peak tail IKr (t1/2 of 100 ± 20 s), and this was similar to the effect of E-4031. The time course of the effect protease XXIV on peak tail IKr is nearly the same as that found with IKCNH2 under similar conditions (see Fig. 1B).

The effects of protease XXIV and E-4031 on the holding current, which is composed mainly of IK1, are shown in Fig. 4D. In each myocyte, the holding current density (measured just before application of each depolarizing step used in Fig. 4C) was calculated, and the averaged data are plotted vs. time. For control conditions, the averaged holding current densities at t0 were 1.29 ± 0.23 pA/pF, and after ∼3 min of exposure to protease XXIV (n = 6 myocytes), it was 1.35 ± 0.29 pA/pF (P > 0.05). Similarly, the control holding current was 1.42 ± 0.26 pA/pF, and after ∼3 min of myocyte exposure to E-4031 (5 μmol/l, n = 5 myocytes), the averaged holding current was 1.28 ± 0.31 pA/pF (P > 0.05). Thus the holding current was not altered, and this is similar to the results with protease XXIV on IKCNJ2 (Fig. 3B).

Finally, we performed a subtraction protocol to separate IKr (measured as the E-4031-sensitive or protease XXIV-sensitive current) from IK. Figure 4, E and F, shows the I-V relations for outward current measured at the end of 2-s-long depolarizing steps to between −30 and 30 mV from a holding potential of −50 mV (example data shown in Fig. 4, A and B). The I-V relations are shown for control conditions and after protease XXIV exposure (Fig. 4E,) or E-4031 exposure (Fig. 4F). In Fig. 4, E and F, the difference plot was calculated by subtracting the respective data points for the two I-V plots. The two difference plots represent IKr and are similar (see also IKCNH2 I-V plot in Fig. 3C). The data support the conclusion that the persisting current, which is predominantly IKs, is not disrupted by protease XXIV, similar to the heterologous expression data shown in Fig. 3A, and that it is not blocked by E-4031 (29, 31, 45).

Site of KCNH2 Channel Proteolysis

To understand better the potential site(s) of proteolytic cleavage, we used Western blot analysis to determine the approximate size of degradation products detected using antibodies directed against the COOH terminus or NH2 terminus of KCNH2 protein. In addition to KCNH2 WT channels, we also examined proteolysis in the KCNH2 engineered mutation N598Q, which substitutes glutamine for asparagine at amino acid 598 in the S5-pore extracellular linker. KCNH2 N598Q channel protein does not undergo N-linked glycosylation, has a molecular mass of ∼132 kDa (close to that of the native protein), and traffics into the plasma membrane to generate IKCNH2 (17). Figure 5 shows Western blot analyses of the WT and N598Q channel proteins. For WT channels (lanes 1 and 5), two protein bands are present at 135 and 155 kDa (filled arrows). After 30 min of protease XXIV exposure, the 155-kDa protein band was eliminated, whereas the 135-kDa protein band was unchanged (lanes 2 and 6). These data also show the appearance of degradation products at ∼60–64 kDa detected using COOH-terminal anti-KCNH2 antibody (lane 2) and at ∼63–67 kDa detected using the NH2-terminal anti-KCNH2 antibody (lane 6). With the NH2-terminal, but not COOH-terminal, anti-KCNH2 antibody, faint smaller molecular mass bands were also occasionally detected. For KCNH2 N598Q channel protein (lanes 3 and 7), a single protein band is present at ∼132 kDa (open arrows), which represents protein located in intracellular organelles and the plasma membrane (38). After 30 min of protease XXIV exposure, the 132-kDa protein band remained, but degradation products were present at ∼60–64 kDa (lane 4) detected with the COOH-terminal anti-KCNH2 antibody and at ∼63–67 kDa (lane 8) detected using the NH2-terminal anti-KCNH2 antibody. Thus, for KCNH2 WT and N598Q channel proteins, the principal degradation products detected with either the COOH-terminal or NH2-terminal anti-KCNH2 antibodies were of comparable molecular mass, with the NH2 terminus containing fragments being slightly larger. These findings were confirmed with seven additional Western blot analyses using COOH-terminal anti-KCNH2 antibody and with 9 additional Western blot analyses using NH2-terminal anti-KCNH2 antibody.

Fig. 5.

Western blot analyses of KCNH2 WT and N598Q channel proteins for control condition and after 30 min protease XXIV exposure probed with COOH terminus and NH2 terminus anti-KCNH2 antibodies. For WT channels (lanes 1 and 5), 2 protein bands are present at 135 and 155 kDa (filled arrows). For KCNH2 N598Q channel protein (lanes 3 and 7), a single protein band is present at ∼132 kDa (open arrows). Degradation bands are highlighted with dotted boxes. See text for further details.

DISCUSSION

Cardiac single cell isolation was first reported in 1960 by Harary and Farley (21), who used trypsin to isolate rat single beating myocytes. Although various mechanical (e.g., trituration) and ionic (e.g., Ca2+ free) conditions have been employed, enzymatic myocyte isolation methods are dominant, mainly using collagenase, protease XIV, protease XXIV, trypsin, and/or hyaluronidase (9, 15, 34, 36, 47, 48, 51). There are several reports of minimal or no IK or IKr in enzymatically dissociated cardiac myocytes, including from human heart. For example, Beuckelmann and coworkers (5) reported a small IKr recorded from ventricular myocytes isolated from failing human hearts and its absence in nonfailing human ventricular myocytes, and Schaffer and colleagues (46) reported that IKr was present in only 2 of 34 ventricular myocytes isolated from human tissue. Konarzewska and colleagues (27) obtained similar findings in isolated nonfailing human ventricular myocytes and speculated that the use of protease XXIV during myocyte isolation might have altered IK. Yue and colleagues (55) studied two isolation techniques for canine atrial myocytes: chunk and perfusion methods. IK, as well as transient outward currents (Ito1 and Ito2), were measured. Both Ito1 and Ito2 were observed in 90% of cells isolated by either technique. IK was present in 142 of 144 myocytes isolated by a perfusion technique, whereas only 3 of 68 myocytes showed IK when isolated by a chunk technique. These reports of small-amplitude IKr in some isolated cardiac myocyte preparations represent a potential paradox as congenital or pharmacological reduction in IKr can cause long QT syndrome, and KCNH2 proteins can be detected in isolated heart cells and myocardium from a number of mammalian species (26, 37). Our findings of selective proteolysis of KCNH2 protein to reduce IKr in isolated native heart cells and IKCNH2 in heterologously expressed channels provide a mechanism that may account for this paradox. Although unlikely, we cannot exclude that proteases also can occlude the KCNH2 channel external vestibule as has been shown for some ether-a-go-go-related gene (ERG) toxins (35).

Western blot analysis employing COOH and NH2 terminus-directed anti-KCNH2 antibodies of intact and protease-treated WT and N598Q channels was used to investigate the site(s) of proteolytic cleaving by protease XXIV. Both the COOH- and NH2-terminal antibodies detected KCNH2 protein fragments of similar size (∼60–64 and ∼63–67 kDa, respectively, Fig. 5). For the KCNH2 N598Q engineered mutation, which does not undergo N-linked glycosylation, the molecular mass of the two degradation fragments, when added together, is close to that predicted for the native KCNH2 protein (∼127 kDa). By using the KCNH2 amino acid sequence beginning with the COOH or NH2 terminus, the molecular masses calculated for N598Q protein fragments are consistent with selective cleaving of amino acid peptide bonds within the S5-pore linker. Figure 6A shows this 39-amino acid-long linker and the amino acid residues whose peptide bonds are putative targets for proteolysis by protease XXIV [aspartate, histidine or serine, see Sigma Chemicals literature (28)]. Cleaving at these sites is predicted to generate COOH terminus-containing KCNH2 fragments of ∼60.1 to ∼63.4 kDa and NH2 terminus-containing KCNH2 fragments of ∼63.2 to ∼66.5 kDa (SwissProt tools), molecular masses that are close to those we found. In comparison, if proteolytic cleaving occurred at the serine in the pore-S6 linker (Fig. 6A), predicted COOH and NH2 terminus-containing KCNH2 fragments would have different molecular masses (∼57.2 and ∼67.7 kDa, respectively), and if proteolytic cleaving occurred in the S1-S2 (residue S428) or the S3-S4 (residues S515 or S517) linkers, then even larger predicted molecular mass differences would be expected (>14 kDa). In KCNH2 WT channel protein, the finding that the COOH and NH2 terminus-directed antibodies detected protein fragments similar to those found with N598Q suggests that the oligosaccharides added during N-linked core- and complex-glycosylation are lost during proteolysis. One possibility is that proteolytic cleaving occurs in the S5-pore region at sites both proximal and distal to the N598 residue. Hence the glycoprotein would not be detected by the antibodies, and this is illustrated in Fig. 6B. Another possibility is that proteolytic cleaving occurs within the N-linked glycosylation consensus motif (N-X-S/T), which in KCNH2 channels is N598-S599-S600, as both serine peptide bonds are putative targets for proteolysis by protease XXIV. Disruption of the KCNH2 N-linked glycosylation consensus motif could then destabilize the glycopeptide bond on N598 with subsequent loss of the oligosaccharides, and this is illustrated in Fig. 6C. The resolution of the Western blot method does not distinguish between these two possibilities. It is not likely that proteases act to simply deglycosylate KCNH2 channels, since direct modification of N-linked glycosylation produces different biochemical patterns and does not disrupt IKCNH2 (10, 17, 58).

Fig. 6.

A: extracellular linker structure for KCNH2, KCNQ1, and meag channel proteins. B and C: alternative models for sites of proteolytic cleaving of the KCNH2 S5-pore linker protein. Only the S5-pore–S6 portion of the KCNH2 protein is shown. Putative amino acid targets for protease XXIV proteolysis are shown as enlarged dots, and OS indicates N-linked oligosaccharide complex.

We used the approach of selective proteolysis of mature KCNH2 channels to provide new insight into the location and rates of protein processing in cells. Biosynthesis of multimeric ion channels is incompletely understood and involves their translation in the ER, proper protein folding and assembly of subunits, posttranslational processing including glycosylation, maturation of the glycoprotein complex through the secretory pathway to reach the plasma membrane, and retrieval of these proteins from the plasma membrane for degradation (14, 19, 30). The finding that the mature protein and IKCNH2 were both quickly abolished by exogenous proteases supports the idea that most complexly glycosylated channel protein is in the plasma membrane. This suggests rapid transport of the 155-kDa WT protein into the plasma membrane after complex glycosylation in the Golgi, as well as rapid processing of the mature channel on retrieval from the plasma membrane, i.e., it does not accumulate in an internal compartment(s) protected from exogenous proteases but rather it must be efficiently degraded in proteosomes (18) or possibly recycled back to the plasma membrane similar to the cystic fibrosis transmembrane conductance regulator protein (49). In contrast, the reappearance of the 155-kDa band and IKCNH2 after proteolysis is relatively slow (mean time constants of 6.4 and 13.2 h, respectively). One explanation is that an important rate-limiting step in KCNH2 channel biogenesis is ER-to-Golgi protein transport, and this agrees with previous qualitative findings obtained using pulse-chase analysis (17, 57). A potential limitation is that these findings were obtained in a HEK-293 cell KCNH2 overexpression model, which may limit comparison with native cardiac myocytes.

The S5-pore linker of KCNH2 channels has been suggested to be unique among Kv channels. It is unusually long compared with most Kv channels, lacks specific residues implicated in the formation of hydrogen bonds that are conserved in other Kv channels, is postulated to assume an α-helical structure spanning amino acid residues 583–594 (the “turret”), and together these are thought to promote C-type inactivation (32, 35). Figure 6A shows the amino acid residues of the S5-pore extracellular linker of the KCNH2, KCNQ1, and meag K+ channel proteins we studied, along with the putative target sites for proteolysis by protease XXIV. The meag S5-pore linker is longer (48 amino acids) than the KCNH2 or KCNQ1 S5-pore linkers (39 and 17 amino acids, respectively). The S5-pore linkers in KCNH2 and meag channels have low (∼30%) homology and vary in the putative sites for proteolytic sensitivity. These differences in primary structure, plus the putative secondary α-helical domain in the S5-pore linker of KCNH2, may provide unique structural sites for its selective proteolytic degradation. In addition, ion channel proteins are increasingly being recognized to reside in specific microdomains in cell membranes, including in lipid rafts such as caveoli, and it is possible that specialized microdomains also might regulate access of extracellular enzymes to channel linker peptides (8).

On the basis of their catalytic mechanism, proteases are subdivided into five classes: aspartic, cysteine, metalloproteinase, serine, and unclassified or threonine (41), and the enzymes may be secreted or cell membrane bound. Within the human genome there are more than 500 genes that encode proteases or protease-like molecules, with the metalloproteinases and serine proteases comprising the largest classes (33). Within the human cardiovascular system, a variety of endogenous proteases are present that mediate roles in blood clotting and thrombolysis, angiogenesis, maturation of cytokines and prohormones, breakdown of intracellular proteins and apoptosis. Proteases have been implicated in cardiovascular disease. For example, human matrix metalloproteinases are thought to play an important role in remodeling of the extracellular matrix (7), specific human serine proteases are important in thrombolysis (33), and they have also been implicated in the regulation of amiloride-sensitive epithelial Na+ channels (11). It is not known whether specific endogenous proteases exist that can modify IKr; however, if endogenous proteases exist that can act similarly on IKr, this potentially could represent a novel regulatory pathway for modifying channel density that could have therapeutic potential, as well as pathological consequences, for arrhythmia generation.

GRANTS

This work was supported, in part, by National Heart, Lung, and Blood Institute Grants R01-HL-60723 (C. T. January), R01-HL-68868 (G. A. Robertson), R01-HL-61537 (T. J. Kamp), and R01-HL-71092 (J. C. Makielski). S. Rajamani received support by a postdoctoral fellowship training award from the American Heart Association, Northland Affiliate.

Acknowledgments

We thank Brian P. Delisle for helpful discussion.

Footnotes

  • 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