Dendritic Ih normalizes temporal summation in hippocampal CA1 neurons

Most neurons of the mammalian CNS receive information through synaptic input located predominately within dendritic arbors. Here in the dendrites, thousands of excitatory and inhibitory synaptic inputs are blended together to form a coherent output response. Whereas synaptic inputs are widely distributed across complicated dendritic arbors, action potential output usually occurs in a relatively localized region of the soma or proximal axon1,2. Because of this arrangement, distances between widespread synaptic inputs and the final integration site can vary greatly. Under most conditions, large variation in synaptic distance causes the amplitude and temporal characteristics of signals from even similar inputs to differ at the final integration site3–6. This location-dependent synaptic variability dictates that the effect of any given synapse on output depends upon its location. Under passive conditions, the amplitude and kinetics of a given excitatory postsynaptic potential (EPSP) vary inversely with the distance of the synapse from the soma; thus, EPSPs from distal synapses will be both smaller and slower than their proximal counterparts3–6. Cable filtering by the dendrites also renders the temporal integration of synaptic activity dependent on synaptic location, with greater temporal summation occurring at the soma for the slower distal input3–6. This location dependence of synaptic input heavily influences the impact of a particular input on the action potential output of the neuron, perhaps even altering the basic mode of firing (coincidence detection or temporal integration)3. Location-dependent synaptic variability could create specific computational problems for CNS neurons. Distortion by dendritic arbors might interfere with a neuron’s ability to accurately determine the total amount of incoming synaptic activity or increase the variability of action potential discharge, or it might decrease the memory-recall performance of artificial neuronal networks7,8. Therefore, mechanisms for lowering the dependence of synaptic effectiveness on input location might be highly advantageous (but see ref. 9). Many mechanisms, usually involving dendritic ion channels, have been proposed for reducing the location dependence of synaptic input (reviewed in refs. 10, 11). However, the role of dendritic ion channels in the integration of spatially and temporally distributed synaptic input has not been well studied. The non-uniform distributions of voltagegated ion channels in CA1 pyramidal dendrites could reduce the distorting effects of the dendritic arbor on temporal summation. One channel type in particular, the hyperpolarization-activated (Ih) channel, seems to be uniquely well suited to influencing the integration of subthreshold synaptic input. The density of these channels increases nearly sevenfold from the soma to the distal apical dendrites of CA1 pyramidal neurons; a similar nonuniform density has been hypothesized to be present in neocortical pyramidal neurons12–15. Because these inwardly conducting channels deactivate during membrane depolarizations, an effectively outward current is generated during synaptic activity12–16. This current could actively reduce local temporal summation occurring at more distal input sites and remove the dependence of temporal integration in CA1 pyramidal neurons on location12. To characterize the location dependence of temporal synaptic integration, I used whole-cell, patch-clamp recordings from the soma and apical dendrites of hippocampal CA1 pyramidal neurons in adult rat slices. Hippocampal CA1 pyramidal neurons receive a fairly homogeneous synaptic input via the Schaffer collateral pathway. In this pathway, axons from CA3 pyramidal neurons make excitatory synaptic contacts over several hundred micrometers of the apical dendritic arbor17,18. The distributed nature of the Schaffercollateral pathway makes this an excellent system in which to investigate location dependence of synaptic input. Experiments with a specific Ih channel blocker indicated that these voltage-gated ion channels are capable of greatly reducing variability in temporal summation (but not in EPSP amplitude) that exists because of spatial differences in input location. The functional impact of this spatial normalization of synaptic integration was also investigated by monitoring the response properties of CA1 pyramidal cells as the degree of location-dependent synaptic variability was altered.