Effects of Mechanosensory Input on the Tracking of Pulsatile Odor Stimuli by Moth Antennal Lobe Neurons

Abstract

Air turbulence ensures that in a natural environment insects tend to encounter odor stimuli in a pulsatile fashion. The frequency and duration of odor pulses varies with distance from the source, and hence successful mid-flight odor tracking requires resolution of spatiotemporal pulse dynamics. This requires both olfactory and mechanosensory input (from wind speed), a form of sensory integration observed within the antennal lobe (AL). In this work, we employ a model of the moth AL to study the effect of mechanosensory input on AL responses to pulsatile stimuli; in particular, we examine the ability of model neurons to: (1) encode the temporal length of a stimulus pulse; (2) resolve the temporal dynamics of a high frequency train of brief stimulus pulses. We find that AL glomeruli receiving olfactory input are adept at encoding the temporal length of a stimulus pulse but less effective at tracking the temporal dynamics of a pulse train, while glomeruli receiving mechanosensory input but little olfactory input can efficiently track the temporal dynamics of high frequency pulse delivery but poorly encode the duration of an individual pulse. Furthermore, we show that stronger intrinsic small-conductance calcium-dependent potassium (SK) currents tend to skew cells toward being better trackers of pulse frequency, while weaker SK currents tend to entail better encoding of the temporal length of individual pulses. We speculate a possible functional division of labor within the AL, wherein, for a particular odor, glomeruli receiving strong olfactory input exhibit prolonged spiking responses that facilitate detailed discrimination of odor features, while glomeruli receiving mechanosensory input (but little olfactory input) serve to resolve the temporal dynamics of brief, pulsatile odor encounters. Finally, we discuss how this hypothesis extends to explaining the functional significance of intraglomerular variability in observed phase II response patterns of AL neurons.

Keywords: SK channel; antennal lobe model; moth olfactory dynamics; odor plume tracking; odor pulse; olfaction; sensory integration.

Figure 1

Schematic of model AL network. The model contains 6 glomeruli (columns); squares represent PNs (10 per glomerulus), and circles represent LNs (6 per glomerulus). Arrow heads indicate excitation, while bar heads indicate inhibition. Within a glomerulus, all cell types form synapses with each other (with cell type-specific connection probabilities), while glomerular cross-talk is mediated only via LN→PN synapses (cross-glomerular connection probability is identical for all glomerular pairs). An odor is simulated via delivery of excitatory stimulus current to all cells within a subset of three glomeruli (solid incoming arrows), while mechanosensory input is simulated via delivery of excitatory stimulus current to all cells within all glomeruli (dashed incoming arrows).

Figure 2

Model responses to 4 Hz stimulus pulses in the odor only (A), mechanosensory only (B), and additive sensory integration (C) stimulus scenarios. Odor is simulated by sending stimulus current to only glomeruli 1, 2, and 3, while mechansensory input is simulated by sending stimulus current to all glomeruli. Left panels show spike rasters for the entire network, while right panels show the membrane potential, fast inhibitory current, slow inhibitory current, and SK current for a representative PN within the network (in the presence of odor stimulus, the PN is taken from a glomerulus receiving odor input). Stimulus pulses are 50 ms in length. Black bars indicate stimulus.

Figure 3

(A) Length of spiking response vs. stimulus pulse length for an odor receiving PN (left) and a nonodor receiving PN (right) in the three stimulus scenarios (odor only, mechanosensory only, and additive sensory integration). Responses are averaged over 50 trials. Note that in the presence of odor (odor only and additive scenarios), the odor-receiving PN has a response length that reflects the length of the incoming stimulus pulse, whereas a PN that does not receive odor has a relatively fixed response length regardless of stimulus pulse length. To exemplify this further, we illustrate the spike raster for a single odor-receiving (B) and non odor-receiving (C) PN at various pulse lengths during the odor only stimulus scenario. The odor-receiving PN has a response that reflects the length of the incoming pulse, whereas the non odor-receiving PN is unstimulated and does not respond. See section 4 for details.

Figure 4

Response slope, defined as the slope of the best fit line for a response length vs. pulse length plot (i.e., plots as in Figure 3A), as a function of the strength of the SK current (left), fast inhibitory synapses (center), and slow inhibitory synapses (right) within the model. A larger (more positive) response slope indicates a greater sensitivity to pulse length, or, in other words, a greater ability of the temporal length of the spiking response to be indicative of the temporal length of the stimulus pulse. In the odor only and additive stimulus scenarios, the PN is taken from an odor-receiving glomerulus; data are shown for a single sample PN and averaged over 50 trials. Values on the x-axes are normalized by the standard strength employed in the model (i.e., 1 represents the standard strength employed for each current in the model).

Figure 5

Spike rasters of the model network for 50 ms stimulus pulses delivered at a frequency of 3 Hz (left) or 7 Hz (right), in the odor only (A), mechanosensory only (B), and additive sensory integration (C) stimulus scenarios. Odor is simulated by sending stimulus current to only cells within glomeruli 1, 2, and 3, while mechansensory input is simulated by sending stimulus current to cells within all glomeruli. Within each glomerulus, the bottom 10 cells are PNs while the top six cells are LNs. Black bar represents stimulus.

Figure 6

Pulse Following Index as a function of pulse frequency for a single odor-receiving and a single non odor-receiving PN from the network, in the odor only (A), mechanosensory only (B), and additive sensory integration (C) stimulus scenarios. The pulse following index is a measure of the ability of the PN response to track the temporal dynamics of pulsatile stimulus delivery, with values above zero signifying sensitivity to pulsatile dynamics and a value of zero implying a lack of discernment of pulse structure (see section 4 for details). Data are averaged over 50 trials.

Figure 7

Pulse Following Rate as a function of the strength of the SK current (left), fast inhibitory synapses (center), and slow inhibitory synapses (right) for a single odor-receiving (A) and a single non odor-receiving (B) PN within the network. Data are shown for the odor only, mechanosensory only, and additive sensory integration stimulus scenarios. The pulse following rate is a measure akin to a cutoff frequency—i.e., the highest pulse frequency at which the PN response can track the temporal structure of stimuli delivered in the form of a train of pulses (see section 4 for details). Values on the x-axis are normalized by the standard values of the SK current, fast inhibitory synapses, or slow inhibitory synapses within the network, with a value of 1 indicating standard strengths. In the standard network, SK current strengths vary from PN to PN, and are drawn from a Gaussian distribution; in the simulations in this figure, SK current strengths are fixed across PNs, and the standard value of the SK current strength is taken as the mean of Gaussian employed in the standard network. Data are averaged over 50 trials.

Figure 8

Pulse frequency tracking in the presence of varied odor tuning across glomeruli. In these simulations, odor is simulated by delivering odor-induced stimulus current to glomeruli 1, 2, 3, 4, 5, and 6 at 0, 0.2, 0.4, 0.6, 0.8, and 1, respectively, times the strength of the odor-induced stimulus current in the standard network (for those glomeruli in the standard network that are stimulated by odor). The total level of odor-induced excitation to the entire network is similar to that in the standard network, except with excitation delivered in a graded manner across glomeruli rather than three glomeruli receiving fixed positive excitation and the other three glomeruli receiving no excitation. (A) Pulse following index of net PN activity within a glomerulus plotted as a function of pulse frequency for several glomeruli, in the absence (left; Odor) or presence (right; Additive) of mechanosensory input. (B) Pulse following rate of net PN activity within a glomerulus for each glomerulus in the network, in the absence (Odor) or presence (Additive) of mechanosensory input. Data are averaged over 50 trials.