There is now strong evidence that learning and memory depend on long-term changes in the strength of synaptic transmission between neurons. However, a typical neuron receives 10,000 synaptic inputs that are distributed along complex dendritic processes emanating from the neuron's cell body. This synaptic information must be integrated by the neuron to determine whether it will fire an action potential, the output signal. It is now clear that neuronal dendrites are not passive recipients of synaptic information but active processors of this information to regulate neural function. Although the role of long-term synaptic plasticity in learning and memory has been widely studied, we know much less about the contribution of active dendritic processing to information storage.
Our laboratory has focused on the role of the hyperpolarization-activated channels—present in many neuronal dendrites—in regulating synaptic integration and long-term plasticity during learning and memory. These channels generate an excitatory inward cation current referred to as Ih. The Ih channels are encoded by a family of four genes, HCN1–4 (so named because the channels are hyperpolarization-activated, cation nonselective, and regulated by cyclic nucleotide binding). The HCN1 channel is expressed heavily in specific brain regions important for learning and memory, including the hippocampus, and is present in neuronal dendrites. In collaboration with Eric Kandel's laboratory (HHMI, Columbia University College of Physicians and Surgeons), we have been studying the importance of these channels for learning and memory by analyzing mice in which the HCN1 gene has been deleted through homologous recombination.
Surprisingly, mice in which the HCN1 gene has been selectively deleted from the forebrain, which includes the hippocampus, show an enhancement in their ability to learn to navigate to a hidden platform in a water maze. Why does removal of HCN1 improve spatial learning? To answer this question we are studying synaptic transmission and long-term synaptic plasticity in pyramidal neurons from the CA1 region of the hippocampus, the neurons that provide the major output of the hippocampus. These neurons receive two major sources of excitatory synaptic input. One set of inputs, the Schaffer collaterals, comes from hippocampal CA3 pyramidal neurons and terminates on regions of CA1 neuron dendrites that are relatively close to the cell body, an area with only moderate HCN1 channel expression. The second set of inputs, the perforant path, represents a direct connection from the neocortex and terminates on the distal dendritic regions of the CA1 neurons, an area where HCN1 expression is very high.
Synaptic transmission and long-term plasticity at the more proximal Schaffer collateral pathway are relatively unaffected by the deletion of HCN1. In contrast, we observed a significant enhancement in the integration of synaptic potentials and a striking increase in long-term potentiation (LTP) of synaptic transmission at the distal perforant path synapses. The differential effect of HCN1 deletion on the two synaptic inputs is consistent with the normal pattern of HCN1 expression in the dendrites of wild-type mice.
Why does HCN1 deletion enhance the induction of LTP at the perforant path synapses? Previous studies from other labs suggest that induction of LTP requires the firing of local calcium action potentials in the distal dendrites. We therefore imaged calcium in distal dendrites from wild-type and HCN1-knockout mice during stimulation of the perforant path inputs. Brief bursts of synaptic activity elicit large, long-lasting calcium signals that persist for 200–300 milliseconds, far longer than the 100-millisecond period of synaptic stimulation. In mice with a deletion of HCN1, the same synaptic stimulation induces a calcium spike with a twofold increase in duration, compared to the calcium signal in wild-type mice. We see a similar enhancement in the calcium signal using a pharmacological antagonist of HCN channel function.
The inhibitory effects of HCN1 channels are somewhat puzzling since the channels generate a depolarizing Ih current that should help excite a neuron. We found that this paradoxical inhibitory effect is caused, at least in part, by an action of HCN channels to make the resting potential of a neuron more positive. The depolarization, in turn, increases the extent of resting inactivation of voltage-gated calcium channels, which decreases their ability to participate in the firing of dendritic calcium action potentials.
Is this the only mechanism by which HCN1 and Ih exert an inhibitory effect on neural activity? Results from several groups suggest that HCN1 may inhibit subthreshold excitatory postsynaptic potentials (EPSPs) by acting as a shunt conductance, decreasing the resting membrane resistance and thereby decreasing the depolarization produced by a given synaptic current.
We examined the action of Ih on subthreshold EPSPs by comparing the peak voltage reached during an EPSP in the absence and presence of a pharmacological antagonist of Ih. Surprisingly, these results demonstrate that Ih has two distinct effects. For weak EPSPs it has an excitatory effect, making the peak EPSP voltage more positive. For strong EPSPs, however, Ih has an inhibitory effect, making the peak EPSP voltage more negative. How does Ih exert these dual excitatory-inhibitory actions?
Using a simple computer model, we find that, on its own, Ih always has a depolarizing effect on EPSP peak amplitude, as long as the EPSP is negative to the Ih reversal potential of –30 mV. Since threshold for firing a spike is around –50 mV, Ih should always be excitatory, regardless of EPSP amplitude. The computer model results thus imply that the inhibitory effect of Ih must result from its interaction with other voltage-gated channels.
So how does Ih inhibit the subthreshold EPSP? We next used a slightly more complex model that contained a voltage-gated delayed-rectifier potassium conductance, in addition to Ih. Under these conditions we were able to reproduce the dual excitatory-inhibitory effects of Ih as a function of synaptic strength. We found thatthe depolarization of the resting membrane by Ih led to an increase in the resting activation of the potassium conductance. The net effect of Ih on peak EPSP voltage thus depends on the relative magnitude of the direct excitatory contributions of this current compared with the inhibitory effects of the additional potassium conductance activated by Ih to hyperpolarize the membrane. For small EPSPs, the direct excitatory action of Ih wins out, because the inward driving force on excitatory current through these channels is much greater than the outward driving force on inhibitory potassium flow. In contrast, for large EPSPs, the outward driving force on potassium becomes larger, leading to a predominant inhibitory action.
We confirmed the predictions of this model in experiments where we blocked the delayed rectifier M-type potassium channel. With these channels blocked, Ih had only an excitatory effect on EPSP peak voltage. Interestingly, both Ih and M current have been shown to be regulated during epilepsy. Thus it may be of interest to target the interaction of these two channels as a novel therapeutic approach to neurological disorders.