Activity-dependent heteromerization of the hyperpolarization-activated, cyclic-nucleotide gated (HCN) channels: role of N-linked glycosylation

Abstract

Formation of heteromeric complexes of ion channels via co-assembly of different subunit isoforms provides an important mechanism for enhanced channel diversity. We have previously demonstrated co-association of the hyperpolarization activated cyclic-nucleotide gated (HCN1/HCN2) channel isoforms that was regulated by network (seizure) activity in developing hippocampus. However, the mechanisms that underlie this augmented expression of heteromeric complexes have remained unknown. Glycosylation of the HCN channels has been implicated in the stabilization and membrane expression of heteromeric HCN1/HCN2 constructs in heterologous systems. Therefore, we used in vivo and in vitro systems to test the hypothesis that activity modifies HCN1/HCN2 heteromerization in neurons by modulating the glycosylation state of the channel molecules. Seizure-like activity (SA) increased HCN1/HCN2 heteromerization in hippocampus in vivo as well as in hippocampal organotypic slice cultures. This activity increased the abundance of glycosylated HCN1 but not HCN2-channel molecules. In addition, glycosylated HCN1 channels were preferentially co-immunoprecipitated with the HCN2 isoforms. Provoking SA in vitro in the presence of the N-linked glycosylation blocker tunicamycin abrogated the activity-dependent increase of HCN1/HCN2 heteromerization. Thus, hippocampal HCN1 molecules have a significantly higher probability of being glycosylated after SA, and this might promote a stable heteromerization with HCN2.

Fig. 1

Neuronal activity bursts (seizures) increase the membrane expression of heteromeric HCN1/HCN2 channels. (a) Immunoprecipitation (IP) with an antiserum to the HCN2 channel isoform followed by western blot (WB) analysis for HCN1 shows enhanced co-association of HCN1/HCN2 in hippocampi from animals subjected to experimental seizures (SA), as quantified in (b). The immunoreactive HCN1 band has an apparent molecular weight of ~125 kDa. The specificity of the immunoprecipitation is supported by the absence of immunoreactive band when IgG is substituted for anti-HCN2. The correspondence of the co-precipitated band to native HCN1 is demonstrated by comparing it to a WB from the same tissue. (c and d) The increased co-association of HCN1 and HCN2 isoforms is also apparent when immunoprecipitation (IP) is performed using an antiserum to HCN1 and the precipitated channels probed by WB analysis for HCN2. (e) The specificity of the HCN1 and HCN2 antisera is evident from the absence of an immunoreactive band of the appropriate molecular weight in brains from mice lacking these channels (HCN1−/− and HCN2−/−). (f) The efficiency of precipitating HCN channels with the antisera used was evaluated using escalated amounts of each anti-serum. Input and HCN2 immunoreactive bands precipitated with 1, 5, and 10 µL anti-HCN2 are shown. Efficiency for this antibody was 20.4 ± 2.8%, 66.1 ± 8.1%, and 36% (n = 2), respectively. Efficiency of HCN1 antiserum in precipitating the same isoform was 9.7%, 30.5%, and 26.5%, respectively (not shown).

Fig. 2

Seizure activity increases the proportion of glycosylated HCN1 channel molecules, but not of the HCN2 isoform. (a) Western blot (WB) analysis of HCN1 reveals two distinct HCN1-immunoreactive bands of different molecular weights (MW): The lower (L) band, migrating at ~105 kDa is the predicted MW of the protein moiety, and the higher band (H) migrates at ~125 kDa. Quantification of the optical density of the HCN1 bands demonstrates an increased H/L ratio from 1.8 in control tissue to 2.2 (n = 14, p < 0.001). (b) Seizure activity does not influence the ratio of the high and low apparent MW bands. (c and d) Treatment with N-glycanase eliminates the high MW band immunoreactive for HCN1 (c), or HCN2 (d), demonstrating that the high MW band is a result of N-glycosylation of the channel molecules.

Fig. 3

The high MW species of HCN1 isoforms is preferentially co-immunoprecipitated with the HCN2 isoform. Two samples of hippocampal homogenates (S1 and S2) have been subjected to immunoprecipitation either with anti-HCN1 (left panels) or with anti-HCN2 sera. The precipitate was then probed for HCN1 immunoreactivity using western blots. Two species of the HCN1 channels, of low and high apparent MWs, were precipitated with anti-HCN1, as expected. In contrast, the HCN1 species that co-immunoprecipitated with anti-HCN2 was primarily of a higher MW, i.e. glycosylated. This indicates that stable complexes (heteromeric channels) with HCN2 are formed mainly by glycosylated HCN1 channel molecules.

Fig. 4

The distribution of HCN1 channels and the augmentation of their heteromerization with HCN2 isoforms are preserved in the hippocampal organotypic slice culture. (a): Immunocytochemistry demonstrates the similarity of HCN1 expression pattern in vivo and in the organotypic hippocampal slice cultures (in vitro). Note the strong immunoreactive signal in the distal dendritic fields of area CA1 in both preparations (white arrows). (b and c) Seizure-like activity generated by exposing the cultures to a low convulsant dose of the glutamate receptor agonist, kainic acid (see main text), increases the co-immunoprecipitation of HCN1 with the HCN1 isoform, as quantified in (c) (n = 4 per group). This recapitulates the findings in vivo, suggesting that the in vitro system can be used to study the underlying mechanisms. (d) HCN1 channels exist in species of different apparent MW (left gel panel), and the higher MW species are a result of glycosylation. Treatment with N-glycanase eliminated the ‘higher MW’ HCN1-immunoreactive band. In addition, treatment with the enzyme endo H, that preferentially cleaves mannose-type glycosyl moieties, failed to eliminate the ‘heavier’ band, suggesting that HCN1 glycosylation is not of the high-mannose type. DG denotes dentate gyrus.

Fig. 5

Tunicamycin can be used in organotypic hippocampal cultures to prevent HCN channels glycosylation, and the treatment permit evaluation of seizure-like activity (SA). (a) Exposure of cultures to 10 µg/mL of the glycosylation blocker tunicamycin does not compromise their viability. Using fluorojade as a sensitive method for neuronal injury and death, no increase in cell death was found in treated cultures compared with controls. (b) Tunicamycin applied for 72 h prevents glycosylation of HCN1 channels in a dose-dependent manner. The large majority of HCN1 channels are de-glycosylated at the dose used here 10 µg/mL. (c) Activation of the principal hippocampal neurons by SA takes place also in the presence of tunicamycin. The immediate early gene c-fos, an established marker of neuronal activation (Labiner et al. 1993) is minimally expressed in control hippocampal cultures (ctrl, left panel). A 3 h exposure to KA leads to robust expression of c-fos, and this has been found to correlate with electrophysiological seizures (Labiner et al. 1993; Richichi et al. 2007). c-fos is expressed to a similar degree in hippocampal cultures maintained in the absence (SA; middle panel) or presence (SA + Tu; right panel) of tunicamycin (10 µg/mL).

Fig. 6

Inhibition of N-linked glycosylation abolishes the increased heteromerization of HCN1 channels by seizure-like network activity. (a) Quantitative analysis of the extent of heteromerization of HCN1 channels in organotypic hippocampal slice cultures was carried out using co-immunoprecipitation (IP), calibrated to the total amount of the HCN1 channels in the tissue. Actin was used to compare protein content among groups. As shown in the representative gels, in the absence of tunicamycin seizure-like activity (SA) increased the optical density of the HCN1 immunoreactive band that precipitated with antiserum to HCN2. However, after cultures were maintained in Tunicamycin (see text and Fig. 5), SA no longer enhanced the co-association of HCN1/HCN2 channels. (b) Quantitative analysis of the percentage of heteromerization (n = 5 per group).