The cGMP-dependent protein kinase II Is an inhibitory modulator of the hyperpolarization-activated HCN2 channel

Abstract

Opening of hyperpolarization-activated cyclic nucleotide-gated (HCN) channels is facilitated by direct binding of cyclic nucleotides to a cyclic nucleotide-binding domain (CNBD) in the C-terminus. Here, we show for the first time that in the HCN2 channel cGMP can also exert an inhibitory effect on gating via cGMP-dependent protein kinase II (cGKII)-mediated phosphorylation. Using coimmunoprecipitation and immunohistochemistry we demonstrate that cGKII and HCN2 interact and colocalize with each other upon heterologous expression as well as in native mouse brain. We identify the proximal C-terminus of HCN2 as binding region of cGKII and show that cGKII phosphorylates HCN2 at a specific serine residue (S641) in the C-terminal end of the CNBD. The cGKII shifts the voltage-dependence of HCN2 activation to 2-5 mV more negative voltages and, hence, counteracts the stimulatory effect of cGMP on gating. The inhibitory cGMP effect can be either abolished by mutation of the phosphorylation site in HCN2 or by impairing the catalytic domain of cGKII. By contrast, the inhibitory effect is preserved in a HCN2 mutant carrying a CNBD deficient for cGMP binding. Our data suggest that bidirectional regulation of HCN2 gating by cGMP contributes to cellular fine-tuning of HCN channel activity.

Figure 1. Interaction between HCN2 and cGKII.

(A) Coimmunoprecipitation of HCN2 and cGKII in HEK293 cells. Lysates of HEK293 cells transfected with HCN2 and cGKII or cGKII alone were immunoprecipitated (IP) using a cGKII antibody and stained for HCN2 and cGKII as loading control. 500 µg protein was applied per lane. (B) Protein extracts of hypothalamic brain tissue from WT and HCN2-KO mice were immunoprecipitated using a cGKII antibody and analyzed in immunoblots (IB) for HCN2. Anti-cGKII served as loading control. (C) Schematic representation of full length HCN2 (862 amino acids) and myc-tagged HCN2-domains used for interaction studies. The calculated molecular size of the proteins is indicated. NT, N-terminus; TMR, transmembrane region; CT, complete HCN2 C-terminus; L, C-linker; CNBD, cyclic nucleotide-binding domain; dC, distal C-terminus. (D) GFP-Trap. Lysates of HEK293 cells coexpressing cGKII-GFP and myc-tagged portions of the HCN2 C-terminus were bound to GFP-tagged beads. Co-immunoprecipitated proteins were detected by immunoblotting with an anti-myc antibody. Anti-cGKII was used as loading control.

Figure 2. Colocalization of HCN2 and cGKII in neurons.

(A–D) Colocalization in primary neurons. Hippocampal neurons of neonatal mice (E16.5) were cotransduced with lentivirus expressing HCN2 and cGKII-myc, respectively. Neurons were stained with antibodies against myc (A) and HCN2 (B). Counterstaining was performed with Hoechst dye. (C) Merge of (A) and (B). (D) Negative control (nc). Merge of stainings in the absence of primary antibodies. Scale bar corresponds to 100 µm. (E–H) Immunohistochemical staining of coronal brain slices. Consecutive slices from wild-type mice were stained with anti-cGKII (E) or anti-HCN2 (F). The signal was amplified by tyramide signal amplification. Counter stain was performed with Hoechst 33342 nuclear dye. As negative control, coronal slices of cGKII-KO (G) and HCN2-KO mice (H) were used. Scale bar corresponds to 500 µm. (I, J) Higher magnification of cGKII (I) and HCN2 (J) staining in the hypothalamic region corresponding to the dotted white box as indicated in (E). Scale bar corresponds to 50 µm.

Figure 3. Phosphorylation of HCN2 by cGKII.

(A) In vitro phosphorylation of HCN2 by cGKII. Lysates of COS-7 cells expressing HCN2 and cGKII were incubated with [γ-32P]-ATP for the times indicated. After incubation, proteins were separated on SDS page and analyzed by autoradiography. The first lane represents a control reaction with a cell lysate lacking cGKII. (B) HCN channel constructs used for phosphorylation studies. The positions of the three putative cGKII phosphorylation sites (S641, S786 and S840) are indicated. The calculated molecular mass is given for each construct. (C) Phosphorylation assay of a HCN2 mutant lacking S786 and S840 (first lane) and the HCN2-S641A mutant. (D) Pulldown of phosphoproteins by TiO₂ beads. Lysates of cells expressing HCN2-CT or HCN2-CT-S641A in the presence or absence of cGKII, respectively, were incubated with TiO₂ beads. Proteins specifically bound to the beads were analyzed with an anti-myc antibody.

Figure 4. Regulation of voltage-dependence of HCN2 activation by cGKII.

(A) Voltage step protocol and family of current traces of a HEK293 cell transiently transfected with HCN2. (B–D) Normalized current-voltage (IV) dependence of HCN2 activation in the presence and absence of cGKII. The voltage-dependence was determined in the presence of 10 µM intracellular cGMP (B), 100 µM intracellular cGMP (C) and 1 µM intracellular cGMP (D). (E) IV curves of HCN2 in the presence or absence of cGKII at 2 µM intracellular cAMP. (F) IV curves determined at 10 µM intracellular cGMP from cells coexpressing cGKII and HCN2 or HCN2-S641A. (G) IV curves of HCN2 compared to the IV curve of an HCN2 mutant with functionally impaired cyclic nucleotide binding domain (HCN2-RT>EA) that was coexpressed with cGKII. Currents were measured in the presence of 10 µM cGMP. (H) Comparison of midpoint potentials (V_0.5) of wild type (WT) and HCN2 mutants (HCN2-S641A, HCN2-RT>EA). Channels were expressed alone or together with either wild type or catalytically inactive GKII (cGKII-D576A). V_0.5 was determined from the normalized IV curves in the presence (+) or absence (−) of 10 µM cGMP as indicated. In one set of experiments the cGKII was inhibited by the pharmacological blocker KT5823. *** = p<0.001.

Figure 5. Model of the bidirectional regulation of HCN2 gating by cGMP.

cGMP shifts the voltage-dependence of HCN2 activation to more positive voltage (+ΔV) via direct interaction with the CNBD of HCN2 and induces a hyperpolarizing shift (−ΔV) by activating cGKII that is bound to the channel.