Granular Cell Integration in the Mormyromast System
The juxtalobar input provides regular all-or-none EPSPs to the granular cell layer that are phase locked to the motor command that drives the electric organ discharge. Although this single EPSP is not sufficient to generate an action potential in the postsynaptic granular cell, it creates a time-window within which appropriately timed afferent inputs are enhanced so that they are more likely to elicit a spike in the post-synaptic granular cell (Bell and Grant, 1992; figures III-2a, III-2b).
Extra- and intracellular recordings from mormyromast afferents indicate that stimulus intensity is coded for by latency-to-fire in afferents arising from single receptors (Szabo and Hagiwara, 1967; Bell, 1989). Szabo and Hagiwara provided early theoretical justification for the likelihood of a latency-to-fire code for stimulus intensity in mormyromast afferents: intensity maps precisely and continuously onto spike latency. Spike number and frequency codes are inherently less precise because they are not continuously scaled, and a multifiber code would suffer from decreased spatial resolution (Szabo and Hagiwara, 1967).
The refractory period (time window following an action potential during which the neuron cannot generate another spike) in the fine central terminals of these mormyromast afferents, whose narrow axonal branches limit speed of recovery to resting potential, is large enough to allow only the first spike of an afferent burst to be relayed. This observation is an additional argument for a latency code, since latency/timing is the only available signal coded for by a single spike (Bell and Grant, 1992).
Due to the presence of electrical synapses (coupling between neurons that allows for the direct transfer of electrical current) from afferents onto granular cells, electrical current is bidirectionally conducted along the afferents (Slesinger and Bell, 1985). This bidirectional conduction allows leakage of primary afferent and juxtalobar EOCD EPSPs, as well as LMI mediated IPSPs, back into the afferent fiber from the granular cell. It is unclear whether such leakage, or backwash, is merely an artifact or whether it might play some feedback role to modify incoming signals (figure 3). Regardless, this backwash has a practical effect that facilitates the study of this system: whereas granular cells are difficult to record from intracellularly due to their small size, postsynaptic potentials in the granular cell can be detected via intracellular recordings from the large diameter primary afferents that make gap juction (electrical) synapses onto the cell bodies of granular cells (Bell and Grant, 1992).
The minimum time required for an EOD reafferent to reach the granular cell layer is coincident with the arrival of the EOCD via juxtalobar input (Bell and Grant, 1992). Objects influencing environmental conductances and impedances will maximally increase the probability of granular cell spiking for the granular cell receiving afferent input from the receptor closest to the object (figure 4). The latency-to-fire of these afferents is dependent on stimulus intensity: receptors more distant from an environmental object will receive a lower intensity signal and therefore their afferents will fire late and either fail to summate with the EOCD or produce a granular cell burst of shorter duration.
The function of LMI inhibition is thought to be similar to lateral inhibition in the mammalian retina; an object with high conductance near one area of the skin will suppress firing in granular cells representing more distal adjacent areas of the skin. Both LMI dendrites and axons are myelinated and probably conduct action potentials (Han et al., 2000). Synapses from LMI projections onto granular cell bodies are GABAergic, and the IPSPs resulting from these inhibitory connections have been shown in the granular layer both with extracellular field potential recordings and intracellular recordings (Han et al., 2000). Although it has been suggested that these IPSPs act to enhance the small differences in latency between the arrival of EOCD and primary afferent EPSPs at granular cells, such a hypothesis has not yet been fully tested.