Subthalamic firing without an end, but now with a beginning

Firing action potentials is a metabolically expensive process, but some cell types generate them incessantly and independent of synaptic input. Neurons of the subthalamic nuclei (STN) continuously fire either regularly or in rhythmic bursts (Bevan & Wilson, 1999), and play a central role in the spatiotemporal organization of activity in the basal ganglia network, and movement control. Aberrant STN pacemaking has been associated with the occurrence of inappropriate voluntary movements. For example, high-frequency burst firing in the STN is associated with rigidity and akinesia, whereas reductions in firing frequency cause ballistic movements (Mink & Thach, 1993). The central role of the STN in orchestrating movements is further substantiated by the recognition of it being the primary target for deep brain stimulation to treat Parkinson's disease (Benabid, 2003). Given the important role of action potential firing of subthalamic neurons, the question arises where do the action potentials originate? In this issue of The Journal of Physiology, Atherton et al. (2008) examined individual STN neurons and unveil a unique subcellular origin for their endless firing.

The usual suspect for action potential origin is the axon initial segment (Eccles, 1957). As central axon diameters can be as small as ∼0.3 μm, they are in most cases not accessible for conventional patch-clamp recording techniques. Recently, the axon staged a remarkable comeback by the observation that some axons are cut during the slicing procedure forming swollen unmyelinated regions amenable to direct recording (Shu et al. 2006). Atherton and co-workers adopted this strategy, using 2-photon laser scanning microscopy of STN cells in vitro to visually target the cut ends of axons, enabling loose-seal cell-attached recordings of extracellular axonal spikes as far as ∼850 μm from the cell body. Plotting the spike latency between the axonal and somatic spike versus recording distance from the cell body showed that the minimum latency for firing an autonomous spike occurred in a region proximal to the cell body. By using subsequently immuno-histochemistry, electron microscopy and Golgi staining of individual STN axons they demonstrated that this region overlaps with the unmyelinated part of the axon between the cell body and myelin onset. This region also expressed ankyrin-G, a marker for axon initial segments and nodes of Ranvier, suggesting that STN pacemaking begins in the axon initial segment. This is consistent with the observed site of action potential initiation in the spontaneously firing Purkinje cell (Khaliq & Raman, 2006), although another study found action potential initiation in Purkinje cells to originate at the first node of Ranvier (Clark et al. 2005).

Is the axon initial segment sufficient for STN firing? The authors compared somatic action potential properties of cells with axons of variable length. Both the threshold and rate-of-rise of action potentials were not affected even when axons were as short as ∼25 μm. These data suggest that the first node of Ranvier, located at ∼105 μm from the cell body, does not play a direct role in action potential initiation. More evidence for this notion comes from experiments where the authors tested the contribution of Na⁺ channels to autonomous action potential generation. Using local applications of low extracellular Na⁺, as well as the Na⁺ channel blocker TTX, they systematically investigate α the role of Na⁺ channels located in the dendrites and axons up to ∼100 μm from the cell body in action potential generation. The results demonstrate that the maximum effect on action potential threshold was obtained when axonal Na⁺ channels ∼25 μm from the cell body were compromised, with some contribution of somatic Na⁺ channels as well. Taken together with the time-latency data these experiments support the idea that the initial segment is essential for on-going spontaneous action potential generation.

Are the spontaneous action potentials reliably forward propagated through the axon network? The authors show that at in vitro firing frequencies, 5–15 Hz, action potentials are faithfully transmitted through axons up to ∼600 μm from the soma. However, conduction failure is known to be highly frequency dependent (Debanne, 2004), and it remains to be tested whether STN axons also transmit action potentials reliably to target cells in the basal ganglia at higher frequencies, or whether STN axons instead act as a low-pass conduction filter. This may be a critical issue. The structural elements excited during deep brain stimulation are predominantly myelinated axons and evidence suggests that high-frequency protocols > 100 Hz replace spontaneous intrinsic firing with the exogenously induced patterns (Hammond et al. 2008). The cellular basis of this interaction is unknown but the mechanisms could converge at the level of the STN axon initial segment. With the convincing demonstration that spontaneous firing in STNs begins at the axon initial segment, Atherton et al.'s paper provides an important starting point to test hypotheses on the cellular basis of deep brain stimulation, which may ultimately improve and rationalize the choice of stimulation parameters used during deep brain stimulation to treat Parkinson's disease.