Dynamics of propagating waves in the olfactory network of a terrestrial mollusk: An electrical and optical study

1. The procerebral (PC) lobe of the terrestrial mollusk Limax maximus contains a highly interconnected network of local olfactory interneurons that receives ipsilateral axonal projections from superior and inferior noses. This network exhibits an ~0.7-Hz intrinsic oscillation in its local field potential (LFP). 2. Intracellular recordings show that the lobe contains at least two classes of neurons with activity phase locked to the oscillation. Neurons in one class produce periodic bursts of spikes, followed by a period of hyperpolarization and subsequently a depolarizing afterpotential. There is a small but significant chance for a second burst to occur during the depolarizing afterpotential; this leads to a double event in the LFP. Bursting neurons constitute ~10% of the neurons in the lobe. 3. Neurons in the other class fire infrequently and do not produce periodic bursts of action potentials. However, they receive strong, periodic inhibitory input during every event in the LFP. These nonbursting cells constitute the major fraction of neurons in the lobe. There is a clear correlation between the periodic burst of action potentials in the bursting neurons and the hyperpolarization seen in nonbursting neurons. 4. Optical techniques are used to image the spatially averaged transmembrane potentials in preparations stained with voltage-sensitive dyes. The results of simultaneous optical and electrical measurements show that the major part of the optical signal can be interpreted as a superposition of the intracellular signals arising from the bursting and nonbursting neurons. 5. Successive images of the entire PC lobe show waves of electrical activity that span the width of the lobe and travel its full length along a longitudinal axis. The direction of propagation in the unperturbed lobe is always from the distal to the proximal end. The wavelength varies between preparations but is on the order of the length of the preparation. 6. One-dimensional images along the longitudinal axis of the lobe are used to construct a space-time map of the optical activity, from which we calculate the absolute contribution of bursting and nonbursting neurons to the optical signal. The contribution of the intracellular signals from the two cell types appears to vary systematically across the lobe; bursting cells dominate at middle and proximal locations, and nonbursting cells dominate at distal locations. 7. The direction and form of the waves can be perturbed either by microsurgical manipulation of the preparation or by chemical modulation of its synaptic and neuronal properties. These results suggest that the direction of propagation arises from dynamic properties of the network, as opposed to an architecture with unidirectional connectivity. Further, fragments of the lobe oscillate after surgical isolation, showing that there is not a single pacemaker region in the lobe. 8. The periodic LFP occasionally exhibits rapid, double events during which the propagating wave state is seen to transiently shift to one with nearly spatially synchronized activity along the length of the lobe. This indicates that the lobe can support a state with essentially no temporal phase gradient, as well as one with propagating waves. 9. Although details of the circuitry within the lobe are largely unknown, analogies between the dynamics reported here and theoretical understanding of excitable media suggest that the temporal phase gradient originates from a spatial gradient of excitability along the lobe. Consistent with this hypothesis is the observation that surgically isolated distal halves of the lobe oscillate faster than isolated proximal halves.