Probing S4 and S5 segment proximity in mammalian hyperpolarization-activated HCN channels by disulfide bridging and Cd2+ coordination

Abstract

We explored the structural basis of voltage sensing in the HCN1 hyperpolarization-activated cyclic nucleotide-gated cation channel by examining the relative orientation of the voltage sensor and pore domains. The opening of channels engineered to contain single cysteine residues at the extracellular ends of the voltage-sensing S4 (V246C) and pore-forming S5 (C303) domains is inhibited by formation of disulfide or cysteine:Cd(2+) bonds. As Cd(2+) coordination is promoted by depolarization, the S4-S5 interaction occurs preferentially in the closed state. The failure of oxidation to catalyze dimer formation, as assayed by Western blotting, indicates the V246C:C303 interaction occurs within a subunit. Intriguingly, a similar interaction has been observed in depolarization-activated Shaker voltage-dependent potassium (Kv) channels at depolarized potentials but such an intrasubunit interaction is inconsistent with the X-ray crystal structure of Kv1.2, wherein S4 approaches S5 of an adjacent subunit. These findings suggest channels of opposite voltage-sensing polarity adopt a conserved S4-S5 orientation in the depolarized state that is distinct from that trapped upon crystallization.

Figure 1. DTT treatment reveals spontaneous disulphide bridging in V246C but not in HCN1-R (C303 alone)

Representative TEVC current recordings (Upper panels) and mean steady-state tail current activation curves (Lower panels) obtained from cells expressing HCN1-R (a) or V246C (b) channels. In both cases, cells were maintained in DTT-free Barths solution. Currents were measured before (control) and after (DTT) acute exposure to DTT (10 mM applied by bath perfusion for 3 min) in response to the I-V protocol described in Methods. In this and subsequent figures the most negative voltage activation potential is indicated next to the records. In the lower panels, the smooth lines show best fits of the Boltzmann equation to the mean ± s.e.m. data wherein the V_1/2, slope (both in mV) and number of observations before (control) and after (post-DTT) were HCN1-R control: -77.6 ±1.5, 7.9 ± 0.5 (n = 4, Figure 1a, filled circles); HCN1-R post-DTT: -79.3 ± 1.4, 8.3 ± 0.4 (n = 4, Figure 1A, open circles); V246C control: -90.6 ± 1.5, 13.4 ± 0.5 (n = 10, Figure 1b, filled circles); V246C post DTT: −83.7 ± 1.3, 10.1 ± 0.5 (n = 10, Figure 1b, open circles). In HCN1-R neither the V_1/2 nor slope were significantly affected by DTT (p = 0.074 and 0.11, respectively by paired Student's t-test) while in V246C DTT significantly affected both parameters with p = 0.001 for V_1/2 and 1.4×10^-4 for slope (paired Student's t-test).

Figure 2. Cu-Phenanthroline (CuPhen) treatment induces disulphide bridge formation in V246C but not in HCN1-R (C303 alone) channels

Representative TEVC current recordings (Upper panels) and mean steady-state tail current activation curves (Lower panels) obtained from cells expressing HCN1-R (a) or V246C (b) channels. In both cases, cells were maintained in Barths solution supplemented with DTT to prevent spontaneous formation of disulfide bonds. Records were obtained before (control) and after acute exposure to CuPhen (30 µM for 2 min) in DTT-free recording solution. Following collection of the CuPhen treated I-V relationship, V246C channels were re-exposed to DTT (10 mM for 5-8 min) after which another I-V relationship was collected (DTT in Figure 2B). Smooth lines in the lower panels represent fits of the Boltzmann equation to the mean ± s.e.m. data wherein V_1/2, slope (both in mV) and number of observations were: HCN1-R control: -73.2 ± 0.6, 8.8 ± 0.7 (n = 4, Figure 2a, filled circles); HCN1-R post-CuPhen: -73.3 ± 0.7, 8.5 ± 0.6 mV (n = 4, Figure 2a, open squares); V246C control: −75.1 ± 1.6, 10.7 ± 1.1 (n = 5, Figure 2b, filled circles); V246C post CuPhen: −95.9 ± 2.9, 14.7 ± 0.7 (n = 5, Figure 2b, open squares). V246C post DTT: −82 ± 3.6, 12.7 ± 0.6 (n = 5, Figure 2b, open circles). In HCN1-R, neither V_1/2 (p = 0.66) nor slope (p = 0.07) were significantly altered by exposure to CuPhen (paired Student's t-test) while in V246C both parameters were found to be significantly altered by CuPhen treatment (p = 2 × 10^-5 and 1.4 × 10^-4 for V_1/2 and slope respectively; paired Student's t-test).

Figure 3. High-affinity Cd²⁺ binding sites form in V246C but not HCN1-R (C303 alone) channels

Representative TEVC current recordings (Upper panels) and mean steady-state tail current activation curves (Lower panels) obtained from cells expressing HCN1-R (a) or V246C (b) channels. In both cases, cells were maintained in Barths solution supplemented with DTT to prevent spontaneous formation of disulfide bonds. Records were obtained before (control) and after acute exposure to Cd²⁺ for greater than 2 min at the indicated concentrations. Smooth lines in the lower panels represent fits of the Boltzmann equation to the mean ± s.e.m. data wherein V_1/2, slope (both in mV) and number of observations were: HCN1-R control: -75.9 ± 1.8, 8.4 ± 1.1 (n = 5, Figure 3A, filled circles); HCN1-R with Cd²⁺ (30µM, applied for >2 minutes): -74.1 ± 1.6, 8.4 ± 0.9 (n = 5, Figure 3a, open circles); V246C control: -78 ± 2.9, 11.8 ± 1 mV (n = 4, Figure 3b, filled circles); V246C with Cd²⁺ (2µM, applied for >2 minutes): −97.1 ± 2.4, 12.2 ± 0.5 (n = 4, Figure 3B, open circles). In HCN1-R Cd²⁺ exposure bid not significantly alter either gating parameter (p = 0.21 and 0.89 for V_1/2 and slope, respectively; paired Student's t-test) while in V246C the V_1/2 (p = 5 × 10^-5) but not the slope (p = 0.65) were significantly altered in the presence of Cd²⁺ (paired Student's t-tests).

Figure 4. Cd²⁺coordination in V246C channels is favoured in the channel closed state

Peak amplitude of currents elicited in response to 3 s steps to −125 mV (each normalized to the amplitude of the first step) are plotted as a function of recording time. Cells were maintained in Barths solution supplemented with DTT to prevent spontaneous formation of disulfide bonds but the recording solution was DTT-free throughout. Hashed boxes indicate the period of Cd²⁺ application. In each case, a single exponential function was fit to the onset and decay of Cd²⁺ action to determine the time constants of block and unblock (see Methods). Mean values of time constants when channels were predominantly open were: inhibition time constant = 115 ± 13.4 s (n = 6, 10µM Cd²⁺), unblock time constant = 24.6 ± 4.5 s (n = 5). Time constants when channels were predominantly closed were: inhibition time constant = 68 ± 10.8 (n = 6, 2µM Cd²⁺), unblock time constant = 60.6 ± 17.8 s (n = 5). The on rates determined in open and closed channels were significantly different (p = 0.02) but the off rates were not (p = 0.09; using Student's t-tests). Although in this example there is only partial recovery of HCN1 current after washout of 10 µM Cd2+, on average there is nearly complete recovery to 92.2 ± 4.1 % of initial current level (n = 5).

Figure 5. Reoxidation of V246C with Copper phenanthroline does not enhance formation of intersubunit disulphide bonds

a. Western blot of oocyte lysates obtained from cells injected with HCN1-ΔC_term (ΔC), V246C (V) and cells not injected with cRNA (U). The upper image (30 s exposure) shows that V246C is present predominantly as the monomer doublet (Blue arrow and M) irrespective of the pretreatment (indicated by the matrix below the lower blot). At this exposure none of the V246C bands exceeded the bin depth of the camera. The lower image (300 second exposure) shows that dimers (Red arrow and D) were present in all V246C lysates. At this exposure, none of the dimer bands exceeded the bin depth of the camera. HCN1-ΔC is included to show the position of monomers and dimers but exposure times are not optimized to ensure that the signals from this lysate are in the linear range of the imager. b. Densitometric analysis of blot shown in a. Background corrected density for the monomer doublet and dimer were calculated from the 30 s exposures and the dimer density as a function of the total (monomer doublet plus dimer) is shown as mean ± s.e.m. Horizontal brackets and text indicate that DTT pretreatment significantly lowered the dimer presence but that this was not reversed by reoxidation with CuPhen as determined by one-way ANOVA analysis (ns, not significant).

Figure 6. Model for HCN1 structure based on X-ray crystal structure of Kv1.2

Intra- and inter-subunit distances between V246 and C303 in the HCN1 channel were determined using a homology model based on the crystal structure of Kv1.2-Kv2.1 (33). To simplify the presentation, only one subunit containing S1-S6 (red) and a neighboring subunit containing S5-S6 (blue) are shown. Four K⁺ ions in the selectivity filter are shown as green balls. V246 in the S4 segment together with the C303 residue in S5 of that same subunit and a C303 residue of a neighboring subunit are shown as ball-and-stick figures. The two dashed lines represent the atomic distances between the C-α atoms of V246 and C303 of the same subunit (43.1 Å) and C303 of neighboring subunit (14.0 Å), respectively.