Branching Out: A New Idea for Dendritic Function. Focus on “Coincidence Detection in Pyramidal Neurons Is Tuned by Their Dendritic Branching Pattern”

Nelson Spruston

For decades, neuroscientists have marveled at the complexity and intricacy of dendritic trees and pondered their function. At first, it was proposed that dendrites served as a way to increase the surface area of a neuron, allowing it to receive more synaptic inputs. Later, with the emergence of cable theory, dendrites were viewed by some as an inconvenience, due to the loss of current along the dendrites between the synapse and the action potential initiation zone in the axon. When it became clear that dendrites are not passive structures, but contain numerous voltage-gated channels, dendrites gained credibility as structures likely to increase the computational complexity of single neurons.

Both dendritic structure and the ion channels expressed in dendrites are incredibly diverse. The dendrites of cerebellar Purkinje cells are virtually devoid of sodium channels (Stuart and Häusser 1994). On its own, this would limit action potential propagation in Purkinje dendrites, but modeling studies suggest that the extensive branching of the Purkinje cell dendritic tree further limits action potential conduction in these fan-like structures (Vettter et al. 2001). By contrast, both the expression of sodium channels and the limited branching of the mitral cell primary dendrites contribute to axon-like action potential conduction in these olfactory neurons (Bischofberger and Jonas 1997; Chen et al. 1997).

The message is clear: differences in structure and ion channel expression both contribute to functional diversity across different classes of neurons. But what about within a single class of neurons; to what extent do dendritic structure and ion channel expression contribute to variability within a population of neurons? A new study, by Schaefer et al. 2003, reported in this issue of the Journal of Neurophysiology (p. 3143), provides strong evidence that subtle differences in dendritic structure may contribute substantially to neuronal function.

The question of how dendritic structure contributes to function is difficult to address, primarily because manipulating dendritic structure is difficult to do, without stepping outside the boundary of normal biological variability. Schaefer and colleagues (2003) solve the problem by studying layer V pyramidal neurons and relating structural and functional variability within this population of neurons. They use a powerful modeling approach to demonstrate that subtle structural differences within the population are sufficient to explain the functional variability they measured in the same neurons they modeled.

The functional property they measured experimentally is called “BAC firing”. In layer V neurons, appropriately timed excitatory postsynaptic potentials (EPSPs) and backpropagating action potentials can lead to the initiation of dendritic calcium spikes, or Backpropagation Activated Calcium spikes (Larkum et al. 1999). The degree of coupling is variable. In some neurons the backpropagating action potential produces a large reduction in the threshold for a dendritic spike (strongly coupled). In other cells the backpropagating action potential produces only a modest threshold reduction (weakly coupled).

To determine whether differences in dendritic structure contributed to the variability in BAC firing observed in their experiments, Schaefer and colleagues reconstructed the dendritic morphology of 29 neurons. They then created computer models of the neurons containing a single set of voltage-gated conductances. Having chosen this set of channels for its ability to produce BAC firing, they found that the population of modeled neurons fully captured the range of variability observed in their experimental measurements of BAC firing. Because each of the neurons contained an identical set of channel conductances, they concluded that morphological variability alone was sufficient to reproduce normal functional variability, so they began a search for the relevant morphological features.

Using a procedure they call “dendritic fingerprinting”, Schaefer and colleagues (2003) identified oblique dendritic branches as the culprit. Branches emerging from the proximal apical dendrite (<140 μm from the soma) enhanced BAC coupling, whereas branches emerging from more distal regions of the apical dendrite diminished coupling. The branches act as either current sources or current sinks. Proximal branches provide a current source to the main dendrite; though current is initially drawn away from the main apical dendrite, branches close to the axon are depolarized enough to activate more sodium and calcium channels and provide further excitatory current back to the main dendrite. Distal dendrites act as current sinks; depolarized less by the backpropagating action potential, fewer sodium and calcium channels are activated, so distal branches tend to draw current away from the main apical branch. The result of these current sources and sinks are dendritic branches that enhance or diminish BAC coupling, respectively.

These findings unveil a new capacity for dendritic function. Not only are subtle variations in dendritic morphology rendered functionally relevant, but also a new mechanism for plasticity is implied: neurons may fundamentally alter their integrative function simply through growth or retraction of just a few small dendritic branches. With recent advances in long-term structural and functional imaging of dendrites in vivo (Grutzendler et al. 2002; Trachtenberg et al. 2003), it may soon be possible to see (literally) whether such plasticity is part of the nervous system's remarkable capacity for change.