A high affinity switch for cAMP in the HCN pacemaker channels

Abstract

Binding of cAMP to Hyperpolarization activated cyclic nucleotide gated (HCN) channels facilitates pore opening. It is unclear why the isolated cyclic nucleotide binding domain (CNBD) displays in vitro lower affinity for cAMP than the full-length channel in patch experiments. Here we show that HCN are endowed with an affinity switch for cAMP. Alpha helices D and E, downstream of the cyclic nucleotide binding domain (CNBD), bind to and stabilize the holo CNBD in a high affinity state. These helices increase by 30-fold cAMP efficacy and affinity measured in patch clamp and ITC, respectively. We further show that helices D and E regulate affinity by interacting with helix C of the CNBD, similarly to the regulatory protein TRIP8b. Our results uncover an intramolecular mechanism whereby changes in binding affinity, rather than changes in cAMP concentration, can modulate HCN channels, adding another layer to the complex regulation of their activity.

Fig. 1. Hydrophobic and hydrophilic interactions of the D and E helices with the C-helix.

a Ribbon representation of human HCN1 tetrameric structure bound to cAMP (PDB: 6UQF). Helices D and E of each monomer are shown in yellow and labeled. Boxed: Blow-up showing residues L601, L602 of D-helix and D629 of E-helix engaging hydrophobic and hydrophilic interactions with I588 and R593 on the C-helix of their own subunit. Numbering refers to hHCN1 sequence. b Multiple sequence alignment and secondary structure elements in the region of the D and E helices of human and mouse HCN1 (hHCN1 and mHCN1, Gene ID: 348980 and 15165, respectively), human and mouse HCN2 (hHCN2 and mHCN2, Gene ID: 610 and 15166, respectively) and rabbit HCN4 (rbHCN4, Gene ID: 100009452). Residues shown in the blow-up in panel a, are highlighted in gray and in yellow. Symbols below denote residue identity (*) and conservative (:) or semi-conservative (.) substitutions. Arrowheads indicate the last residue of the truncated constructs, color-coded as follow: ΔC-term gray, ΔE green, ΔDE’ blue and ΔDE red. c Representative whole cell currents of HCN4 full length (HCN4 FL) and ΔC-term recorded at −30, −90 and −150 mV in control solution and with 1 µM cAMP in the patch pipette. Scale bars: 200 pA and 500 ms. Right: mean activation curves in control solution (full symbols) and with cAMP (empty symbols) from HCN4 FL (black) and ΔC-term (gray). Data are presented as mean ± SEM. Data fit with a Boltzmann function are plotted as solid (control) and dashed line (+1 µM cAMP). Calculated half-activation voltages (V_1/2) and inverse slope factors (k) are reported in Supplementary Table 2 together with the details on statistical analysis.

Fig. 2. Analysis of cAMP response and binding in HCN4 and HCN2 constructs deprived of the D and E helices.

a Ribbon representation of D and E helices, in yellow, shown with helices B and C of the CNBD, in gray, in the FL and in hypothetical representation of the truncated constructs ΔE, ΔDE’ and ΔDE, with the portions removed (PDB:6UQF). b Mean activation curves measured in control solution (full symbols) and with 5 µM of cAMP (empty symbols) for HCN4 (H4) FL (black), ΔE (green), ΔDE’ (blue) and ΔDE (red). Control (solid line) and 5 µM cAMP (dotted) line. FL lines are replotted, for comparison, in the other panels. Data are shown as mean ± SEM. Calculated half-activation voltages (V_1/2) and inverse slope factors (k) are reported in Supplementary Table 2 together with the details on statistical analysis. c, ΔV_1/2 as a function of cAMP concentration for the constructs, color-coded as in panel b. Data fit to the Hill equation (solid lines) yielded half maximal effective concentration (K_1/2) values of 1.35, 8.6, 9.3, 38 µM and Hill coefficients (nH) values of 0.9, 0.9, 1, 1 for HCN4 FL, ΔE, ΔDE’ and ΔDE constructs, respectively. Each data point is an average of n ≥ 3 experiments (exact numbers are provided in the source data file). Data are shown as mean ± SEM. d Representative binding curves of cAMP to purified human HCN2 CNBD constructs measured by ITC. CNBD ΔC-term includes helices DE, as shown in Fig. 1b. Top panels show heat changes (μcal/sec) following injections of cAMP into the chamber containing the protein. Bottom panels show the binding curve. The peaks were integrated, normalized to cAMP concentration, and plotted against the molar ratio (cAMP/CNBD). Solid red line represents a nonlinear least-squares fit to a single-site binding model yielding the equilibrium dissociation constant (K_D) and stoichiometry (N) values as shown for each representative sample. Mean K_D and N values for the group are reported in Supplementary Table 3. e Mean K_D values ± SEM (n = 3 experiments) from measurements shown in panel d, plotted on a logarithmic scale (see also Supplementary Table 3 for details on statistical analysis).

Fig. 3. Quantification of the cAMP content in the purified CNBDs of HCN1 and HCN2 subtypes.

Representative absorbance profiles, measured at λ = 254 nm, of cAMP molecules released by (a) HCN1 CNBD ΔC-term and CNBD ΔDE proteins (black and red lines, respectively) and (b) HCN2 CNBD ΔC-term and CNBD ΔDE proteins (black and red lines, respectively) after boiling. Mean molar ratio [protein]:[cAMP] ± SEM of 3 independent experiments are reported in the main text. Calibration curves are shown in the inset (Arb. units: arbitrary units).

Fig. 4. Analysis of the salt bridge interaction between the E-helix and the C-helix.

a Ribbon representation of the D and E helices, in yellow, together with the C-helix of the CNBD, in gray (PDB:6UQF). The residues forming the salt bridge interaction (dotted line) are shown as sticks and labeled. Numbering refers to hHCN2 (H2) and rbHCN4 (H4) sequences. b Mean activation curves, measured by patch clamp, of HCN4 D750 (purple) in control solution (full symbol, solid line) and with 5 µM cAMP in the pipette solution (empty symbol, broken line). Activation curves of wild type HCN4 are plotted in black, without symbols, for comparison (replotted from FL in Fig. 2b). Half-activation voltage values (V_1/2) and inverse slope factors (k) are reported in Supplementary Table 2 together with the details on statistical analysis. Data are shown as mean ± SEM. c Shift of V_1/2 as a function of cAMP concentration for HCN4 D750A construct (purple symbols). Each data point is an average of n ≥ 3 experiments (exact numbers are reported in the source data file). Data fit to the Hill equation (solid line) yielded half maximal effective concentration (K_1/2) of 10.4 µM and Hill coefficient (nH) of 0.8. Data are presented as mean ± SEM. Data for HCN4 FL (black line) and ΔE (green line) are replotted form Fig. 2c. d Representative ITC measurements of cAMP binding to purified hHCN2 CNBD D698A and R662A. Top panels show heat changes (μcal/sec) after cAMP injection into the chamber containing CNBD. Bottom panels show binding curves obtained from data displayed in the upper panel. The peaks were integrated, normalized to cAMP concentration, and plotted against the molar ratio (cAMP/CNBD). Solid red line represents a nonlinear least-squares fit to a single-site binding model yielding, in the present examples, equilibrium dissociation constant (K_D) and stoichiometry (N) values as shown. Mean K_D and N values are reported in Supplementary Table 3 together with the statistical analysis. e Calculated K_D values ± SEM for CNBD D698A and R662A (empty purple bars). Values for CNBD ΔC-term and ΔE are replotted as black and green bars without data points, from Fig. 2e.

Fig. 5. Helices D and E control cAMP affinity in HCN2.

a Representative current traces (recorded at −30, −90 and −130 mV) and mean activation curves of HCN2 FL (black), ΔDE (red) in control solution (full symbols) and with 1 µM cAMP in the pipette solution (empty symbols). Scale bar is 200 pA x 500 ms. Half-activation voltage (V_1/2) and inverse slope factor (k) values are reported in Supplementary Table 2 together with the details on statistical analysis. Data are shown as mean ± SEM. b ΔV_1/2 as a function of cAMP concentration for the constructs, color-coded as in panel a. Data fit to the Hill equation (solid lines) yielded half maximal effective concentration (K_1/2) values of 1.69 and 45.8 µM and Hill coefficients (nH) values of 1 and 1 for HCN2 FL and ΔDE constructs, respectively. Each data point is an average of n ≥ 3 experiments (the exact number is reported in the source data file). Data are shown as mean ± SEM. c Representative current traces (recorded at −30, −90 and −130 mV) and mean activation curves of HCN2 ΔE (green) and D671A (purple) in control solution (full symbols) and with 1 µM cAMP in the pipette solution (empty symbols). Scale bar is 200 pA x 500 ms. Half-activation voltage (V_1/2) and inverse slope factor (k) values are reported in Supplementary Table 2 together with the details on statistical analysis. Data are shown as mean ± SEM. d cAMP-induced shift in V_1/2 in constructs from panels a and c. Data are shown as mean ± SEM.

Fig. 6. helix E controls cAMP affinity in HCN1.

a Representative current traces (recorded at −30, −70 and −120 mV) of HCN1 FL (black), ΔE (green), D629A (purple) and R549E (gray). b Activation curves of HCN1 FL, ΔE, D629A and R549E; colors as in panel a. V_1/2 and k values are reported in Supplementary Table 2 together with the details on statistical analysis. Data are shown as mean ± SEM. c cAMP-induced shift in V_1/2 in constructs from a. For HCN1 constructs, the reference value is that of the mutant R549E, which is virtually cAMP-insensitive. Data are presented as mean ± SEM.

Fig. 7. ITC validation of the hydrophobic interactions between D-helix and C-helix.

a Structure of D and E helices (in yellow) along with the C-helix, shown in gray. Side chains of residues involved in the hydrophobic interactions are shown as sticks. Van der Waals surfaces are displayed as spheres (PDB:6UQF). b Examples of ITC thermogram obtained by titrating purified human HCN2 CNBD L670G/L671G with and without the mutation D698A in E-helix with cAMP (see Fig. 4). Top panels show heat changes (μcal/sec) during successive injections of cAMP. Bottom panels show binding curves obtained from data displayed in the upper panel. The peaks were integrated, normalized to cAMP concentration, and plotted against the molar ratio (cAMP/CNBD). Solid red line represents a nonlinear least-squares fit to a single-site binding model yielding, in the present examples, equilibrium dissociation constant (K_D) and stoichiometry (N) values as shown. Mean K_D, N values and statistical analysis are reported in Supplementary Table 3. c Mean K_D values ± SEM from measurements shown in panel B. Values for CNBD ΔC-term and ΔDE are replotted as black and red bars, without data points, from Fig. 2e.

Fig. 8. Trip8b_nano binding to CNBD constructs with and without D and E helices.

a Left, superposition of the unbound (PDB: 5U6O, blue) and cAMP-bound (PDB: 6UQF, gray) structures of HCN2 CNBD; these constructs do not include helices DE. Center, structure of HCN1 (PDB:6UQF) highlighting the salt bridge, conserved in HCN subtypes, between Asp629 (red sticks) on helix E and Arg593 (blue sticks) on helix C. Right, structural model of HCN2 CNBD (light blue) in complex with TRIP8b_nano peptide (orange). Asp50 and 57 (red sticks) on the TRIP8b_nano peptide form salt bridges with Arg 662 (blue sticks) of C-helix. b Examples of ITC thermogram obtained by titrating purified human HCN2 CNBD ΔC-term, ΔE and with TRIP8b_nano peptide. Upper panel, heat changes (μcal/sec) during successive injections of TRIP8b peptide. Lower panel, binding curves obtained from data displayed in the upper panel. The peaks were integrated, normalized to TRIP8b peptide concentration and plotted against the molar ratio (TRIP8b_nano/CNBD). Solid red line represents a nonlinear least-squares fit to a single-site binding model yielding, in the present examples, equilibrium dissociation constant (K_D) and stoichiometry (N) values as shown. Mean K_D,N values and statistical analysis are reported in Supplementary Table 3. c Dissociation constant (K_D) values ± SEM of CNBD ΔC-term, ΔE, ΔDE’ and ΔDE. d Mean activation curves of HCN4 FL (black) and ΔE (green) in control solution (full circles), with 15 µM and 60 µM cAMP for FL and ΔE respectively (empty circles), and with cAMP + 1 µM of purified TRIP8b_nano in the patch pipette (diamonds). Lines show Boltzmann fitting to the data. Half-activation voltage (V_1/2) and inverse slope factor (k) values are reported in Supplementary Table 4 together with the details on statistical analysis. Data are presented as mean ± SEM (where not visible, the error bars are within the symbol). e Mean shift due to TRIP8b_nano ± SEM calculated from data in (d).

Fig. 9. Activation-induced affinity increase in HCN2 and ∆DE measured by confocal patch-clamp fluorometry (cPCF).

a Representative confocal images of pipette tips carrying excised macropatches from X. laevis oocytes. Either full-length mHCN2 (left) or ∆DE (right) channels were expressed. The green fluorescence signal of the patch is caused by binding of 0.25 µM f1cAMP to the binding sites of the functional channels. The red signal in the background is caused by the reference dye Dy647 (5 µM), used to subtract the background intensity of unbound f1cAMP. The scale bar is 10 μm. The microscope settings were similar throughout all wildtype and ∆DE recordings. For the sake of a better visibility of the ΔDE patch in exported images, the brightness of the green channel was increased during image processing, letting the original data unchanged. b Simultaneously measured current (black) and fluorescence traces (green) for full-length mHCN2 (left) and ∆DE (right). The voltage protocol is shown above. Fluorescence intensities at −30 mV were normalized to 1.