Hyperpolarization activated cyclic-nucleotide gated (HCN) channels in developing neuronal networks

Abstract

Developing neuronal networks evolve continuously, requiring that neurons modulate both their intrinsic properties and their responses to incoming synaptic signals. Emerging evidence supports roles for the hyperpolarization-activated cyclic nucleotide-gated (HCN) channels in this neuronal plasticity. HCN channels seem particularly suited for fine-tuning neuronal properties and responses because of their remarkably large and variable repertoire of functions, enabling integration of a wide range of cellular signals. Here, we discuss the involvement of HCN channels in cortical and hippocampal network maturation, and consider potential roles of developmental HCN channel dysregulation in brain disorders such as epilepsy.

Figure 1. Molecular and functional characteristics of HCN channels and I_h

A) HCN channels are composed of four isoforms that can assemble as homomeric or heteromeric tetramers. Each of these isoforms, coded by the HCN1, 2, 3 and 4 genes, contains six transmembrane segments, with a positively charged S4 voltage sensor, similar to the voltage sensors of depolarization-activated potassium channels. HCN channels are non-selective cation channels that conduct primarily Na⁺ ions at the negative membrane potentials at which they activate (Robinson and Siegelbaum, 2003). A characteristic feature of the HCN channels is the presence of a 120-amino-acid cyclic nucleotide binding domain in the cytoplasmic carboxy terminus (CNBD) which mediates their responses to cyclic AMP. B) Note: I_h activation following an action potential produces a slow depolarization that may activate other cation channels (Ca²⁺, Na⁺) and thus trigger a new action potential. I_h then deactivates (modified from Pape, 1996). C) Current clamp recordings illustrate the stabilizing actions of I_h on the resting membrane potential (dashed line): A hyperpolarizing input elicits a slow, depolarizing “sag” in membrane potential, reflecting I_h activation (red trace). Similarly, a depolarizing input yields a hyperpolarizing “sag” in membrane potential, reflecting I_h deactivation (blue trace). Note the rebound de- and hyperpolarization at the end of the hyperpolarizing resp. depolarizing input (arrows; modified from Poolos et al., 2002).

Figure 2. The subcellular localization of HCN channels in cortical and hippocampal neurons is neuron-type- and isoform-specific

HCN channels can localize to somatic, dendritic and axonal compartments of neurons, and these locations differ among neuronal populations. In CA1 pyramidal cells (A), HCN channels preferentially occupy dendritic locations (Lörincz et al., 2002; Notomi & Shigemoto, 2004; Brewster et al., 2007a), whereas in GABAergic interneurons, somatic HCN expression is more pronounced (B; Notomi & Shigemoto, 2004; Brewster et al., 2007a). Axonal localization is found in certain interneurons (e.g. basket cells, B; Notomi & Shigemoto, 2004; Aponte et al., 2006; Brewster et al., 2007a) and in layer II stellate cells of entorhinal cortex (C; Bender et al., 2007). The subcellular distribution is isoform-specific. Thus, HCN1- and HCN2-, but not HCN4-containing channels were detected in axon terminals of interneurons that express all three isoforms (B; Aponte et al., 2006; Brewster et al., 2007a).

Figure 3. Presynaptic localization of functional HCN1 channels in perforant path axon terminals is age-specific

A) Robust expression in the termination zone of the medial perforant path in dentate gyrus molecular layer of immature (A, C), but not adult rats (B, D), suggests an age-specific presynaptic location of HCN1 channels in this pathway (inset in C: electron microscopy/immunogold-detection of HCN1 in an axon terminal; Bender et al., 2007). Blockade of these channels with the selective I_h-inhibitor ZD7288 resulted in increased short-term-depression (STD) of neurotransmission in medial perforant path of P10-14 rats after stimulation with 20 Hz (E, top panel), suggesting that the channels are active and influence transmitter release at the immature age. In slices from adult rats with little HCN1 in perforant path (B, D), no effect of ZD7288 on STD (expressed as ratio of fEPSP₈–₁₀/fEPSP₁) was detected (E, bottom panel; with permission of Journal of Neuroscience).

Figure 4. Activity-dependent changes in HCN channel expression and molecular rearrangements

Expression levels of HCN isoforms in developing hippocampus are differentially influenced by neuronal activity. Activity burst or seizures (red arrow in A) provoked reduction of the HCN1 isoform (fuchsia-colored spheres in A) and increased mRNA but not protein expression of the HCN2 isoform (yellow spheres in A). Decreased HCN1 protein levels further increased the stochastic probability of formation of HCN1/HCN2 heteromeric channels (A, green arrows), which may generate an h-current with distinctive biophysical properties (Chen S. et al., 2001). B) Seizure-induced expression changes of the HCN1 isoform (fuchsia-colored line) endured for months, suggesting that the normal developmental expression program of these channels (blue line) has been corrupted by the early-life seizures (red arrow).