Ceramide modulates HERG potassium channel gating by translocation into lipid rafts

Abstract

Human ether-à-go-go-related gene (HERG) potassium channels play an important role in cardiac action potential repolarization, and HERG dysfunction can cause cardiac arrhythmias. However, recent evidence suggests a role for HERG in the proliferation and progression of multiple types of cancers, making it an attractive target for cancer therapy. Ceramide is an important second messenger of the sphingolipid family, which due to its proapoptotic properties has shown promising results in animal models as an anticancer agent. Yet the acute effects of ceramide on HERG potassium channels are not known. In the present study we examined the effects of cell-permeable C(6)-ceramide on HERG potassium channels stably expressed in HEK-293 cells. C(6)-ceramide (10 microM) reversibly inhibited HERG channel current (I(HERG)) by 36 +/- 5%. Kinetically, ceramide induced a significant hyperpolarizing shift in the current-voltage relationship (DeltaV(1/2) = -8 +/- 0.5 mV) and increased the deactivation rate (43 +/- 3% for tau(fast) and 51 +/- 3% for tau(slow)). Mechanistically, ceramide recruited HERG channels within caveolin-enriched lipid rafts. Cholesterol depletion and repletion experiments and mathematical modeling studies confirmed that inhibition and gating effects are mediated by separate mechanisms. The ceramide-induced hyperpolarizing gating shift (raft mediated) could offset the impact of inhibition (raft independent) during cardiac action potential repolarization, so together they may nullify any negative impact on cardiac rhythm. Our results provide new insights into the effects of C(6)-ceramide on HERG channels and suggest that C(6)-ceramide can be a promising therapeutic for cancers that overexpress HERG.

Fig. 1.

Ceramide (Cer) inhibits human ether-à-go-go-related gene (HERG) current. A: HERG channel current (I_HERG) was elicited once every 20 s by a 1-s depolarizing step to +20 mV. Peak tail current was measured 50 ms after repolarization to −40 mV and plotted against time. I_HERG was allowed to stabilize in the control (Cntl) solution before a switch of the solution to 10 μM Cer for 2.5 min was made. The smooth line is a linear regression fit to the control data that was used to estimate control current amplitude at the time points used to measure Cer-induced inhibition. B: currents from the same cell as in A are shown before, during, and after recovery (Recov) from Cer treatment. Note the faster deactivation during the Cer treatment. C: the inactive analog dihydro-C₆-ceramide (DihydroCer; 10 μM) fails to affect HERG current. This time course is from a different cell than that used in A. D: inhibition of peak tail current (as described in A) by Cer (n = 10) and DihydroCer (n = 7). **P < 0.01.

Fig. 2.

Cer hyperpolarizes HERG activation. A: activation-voltage relationships were recorded before, during, and after 10 μM Cer. Currents were measured upon repolarization to −40 mV following 5-s voltage steps ranging from −60 to 40 mV and are plotted versus the step voltage. The smooth lines represent single Boltzmann equation fits to generate half-activation voltage (V_1/2) = −12, −19, and −10 mV; slope factor = 8, 8, and 8; and maximum current = 1.1, 0.8, and 0.8 nA for Cntl, Cer, and Recov, respectively. B: activation-voltage relationships shown in A were normalized to tail current generated after the +40-mV step to highlight the left shift in V_1/2. The smooth lines are Boltzmann equation fits with V_1/2 and slope, the same as those in A. C: activation-voltage relationship recorded before, during, and after 10 μM DihydroCer treatment. Fit parameters were V_1/2 = −3, 0, and 2 mV; slope factor = 8, 8, and 8; and maximum current = 0.8, 0.8, and 0.8 nA for Cntl, DihydroCer, and Recov, respectively. D: normalized relationships (described in B) for the data shown in C. E: the negative shift in V_1/2 induced by Cer. The change in V_1/2 was calculated as the difference between test (either Cer, n = 21, or DihydroCer, n = 15) and Cntl, where Cntl was the average V_1/2 measured before application and after Recov. **P < 0.01.

Fig. 3.

Cer speeds HERG deactivation. A: HERG current traces were normalized to peak tail current to highlight the faster deactivation induced by 10 μM Cer (gray curved line) compared with Cntl and Recov (black curved line). B: deactivation was not altered by DihydroCer (10 μM; gray curved line). C and D: %change in deactivation τ at −60 mV induced by Cer and DihydroCer for the fast (C) and slow (D) components. The %change was calculated using control deactivation τ that was the average of values measured before and after Recov from lipid application (either Cer, n = 27, or DihydroCer, n = 20). **P < 0.01.

Fig. 4.

I_HERG is affected by physiological ceramide concentrations. A: dose-response relationships for the effect of Cer on slow and fast τ, V_1/2, and inhibition. The %effect values were normalized to effect measured at 20 μM Cer. The smooth lines are fits to the data using the Hill equation to yield EC₅₀ values of 2.6, 2.8, 10.2, and 9.4 μM, and Hill coefficient values were 2.4, 2.7, 1.1, and 0.9 for fast τ, slow τ, V_1/2, and inhibition, respectively. The applied Cer concentrations were 1, 3, 5, 10, and 20 μM (n = 3, 3, 8, 8, and 6, respectively). B: Cer inhibits HERG window current. The holding potential was set to the voltage (usually −20 mV) generating maximum window current, which was measured before, during, and upon recovery from 10 μM Cer (n = 5). To confirm that HERG channels produced the window current, inhibition by a known HERG blocker, terfenadine (0.1 μM), is also presented (n = 5).

Fig. 5.

Acute Cer treatment does not affect HERG protein levels. A: HEK-293 cells were treated with either 10 μM Cer or DihydroCer for the durations indicated. Densitometric values of the mature 155-kDa band were normalized with GAPDH levels and are shown below a representative gel (n = 3). B: HEK-293 cells stably expressing HERG were treated with 10 μM Cer for the durations indicated. Cells were then treated with biotin to bind proteins expressed on the cell surface, followed by treatment with streptavidine beads. Protein bound to streptavidine beads was separated, and equal protein was loaded on the gel and probed for HERG. A representative blot is shown above, with normalized and averaged densitometer values shown in the bar graph below (n = 3). All data were normalized to the Cntl value for that run, and only the value at 24-h Cer treatment was statistically different from Cntl. *P < 0.05.

Fig. 6.

Cer localizes HERG channels to lipid rafts. HERG expressing HEK-293 cells was treated with either Cntl solution (A) or 10 μM Cer (B) for 10 min, and lipid microdomains were isolated by detergent-free sucrose gradient method (highlighted by rectangles). The 12 fractions from top to bottom were probed for HERG and the lipid raft marker caveolin 1 (Cav1) with anti-HERG and anti-Cav1 antibodies. C: lipid raft fractions (fractions 4 and 5) were quantified and expressed as %total protein in the 12 fractions (n = 3). *P < 0.05.

Fig. 7.

Cholesterol depletion abrogates the effect of Cer on HERG kinetics. HERG-expressing cells were pretreated for 30 min with either 5 mM methyl-β-cyclodextrin (MβCD; n = 5) or 5 mM MβCD conjugated with cholesterol (Chol; n = 6) prior to Cer application. Cer-induced changes in V_1/2 (A), fast deactivation τ (B), and slow deactivation τ (C) are depicted. *P < 0.05; **P < 0.01.

Fig. 8.

The model reproduces the kinetic effects of ceramide on I_HERG. A: the left shift in the V_1/2 and the I_HERG inhibition (11%) induced by Cer model (●) are compared with values from the Cntl model (□). Currents were measured at −40 mV following 5-s steps to the indicated voltage. The smooth lines represent single Boltzmann equation fits to generate V_1/2 = −4.5 and −10.2 mV and slope = 9 and 9 for Cntl and Cer, respectively. B: model reproduces the Cer-induced acceleration of HERG deactivation.