Flight motor networks modulate primary olfactory processing in the moth Manduca sexta

Abstract

Nervous systems must distinguish sensory signals derived from an animal's own movements (reafference) from environmentally derived sources (exafference). To accomplish this, motor networks producing reafference transmit motor information, via a corollary discharge circuit (CDC), to affected sensory networks, modulating sensory function during behavior. While CDCs have been described in most sensory modalities, none have been observed projecting to an olfactory pathway. In moths, two mesothoracic to deutocerebral histaminergic neurons (MDHns) project from flight sensorimotor centers in the mesothoracic neuromere to the antennal lobe (AL), where they provide the sole source of histamine (HA), but whether they represent a CDC is unknown. We demonstrate that MDHn spiking activity is positively correlated with wing-motor output and increased before bouts of motor activity, suggesting that MDHns communicate global locomotor state, rather than providing a precisely timed motor copy. Within the AL, HA application sharpened entrainment of projection neuron responses to odor stimuli embedded within simulated wing-beat-induced flows, whereas MDHn axotomy or AL HA receptor (HA-r) blockade reduced entrainment. This finding is consistent with higher-order CDCs, as the MDHns enhanced rather than filtered entrainment of AL projection neurons. Finally, HA-r blockade increased odor detection and discrimination thresholds in behavior assays. These results establish MDHns as a CDC that modulates AL temporal resolution, enhancing odor-guided behavior. MDHns thus appear to represent a higher-order CDC to an insect olfactory pathway; this CDC's unique nature highlights the importance of motor-to-sensory signaling as a context-specific mechanism that fine-tunes sensory function.

Keywords: active sampling; antennal lobe; ascending neuron; corollary discharge; olfaction.

Fig. 1.

MDH activity is correlated with flight-motor patterns. (A) Schematic of key components of the moth CNS, including the AL, subesophageal zone (SEZ), and AMMC and the pterothoracic ganglion (PTG), which includes the fused mesothoracic and metathoracic neuromeres. Also highlighted is our experimental approach, which included the simultaneous intracellular recording of an MDHn (blue) and suction electrode recording of the IIN1b nerve fiber (red), while wing-motor output is driven from the flight central pattern generator (red circle) via bath-applied chlordimeform (50 µM). (B) HA immunolabeling (green) of the MDHns with the intracellularly recorded MDHn filled with Alexa 568 (magenta; magnification = 20×). Inset below is a zoom-in of two distinct cell bodies labeled. (Left) Both laser channels. (Center) Alexa 568 channel showing a single filled cell body and primary neurite. (Right) HA channel showing the two cell bodies and primary neurites of the MDHn pair. Complete spatial overlap confirms the recording was of a MDHn. (C_i) Superimposition of the smoothed instantaneous spike rate of the recorded MDHn (blue) and the raw extracellular recording of the IIN1b fiber (red). Dashed rectangle highlights the time sample shown in C_ii, which shows the raw spike trains for both traces. Note that the MDH spike rate always increases just before and during bouts of wing motor output. (D) Plot of z-score normalized spike rate for MDH (blue) and IIN1b (red) across 10 min of continuous recording, demonstrating that as IIN1b activity increases over time, so too does MDH spike rate (r = 0.71). (E) Scatterplot of z-score normalized spike rate of MDH and IIN1b. Linear regression (n = 4 recordings/738 points; R² = 0.09, r = 0.30). (F) Mean spike rate from epochs where the IIN1b was quiet versus producing wing-motor output from the recording highlighted in C. Error bars represent the SE. Statistical comparisons between states indicates corresponding significant increase in both IIN1b and MDH (Welch’s t test; n = 8 recording segments; *P < 0.05). (G) Cross-correlation between MDH and IIN1b firing rates using Gaussian smoothing windows ranging in width from 2 ms to 1,000 ms. Note that for smoothing widows within typical spike integration times (2–5 ms), there is no correlation between measures.

Fig. 2.

HA enhances entrainment of AL PNs to rapidly pulsed odor. (A) To evaluate the effect of MDHn HA release on the ability of AL PNs to entrain to pulsed stimuli, we performed three experiments, each in separate groups of animals. For all experiments, multichannel electrodes were placed into the AL and multiunit recordings were made while the ipsilateral antenna was stimulated with a block of five 500-ms-long stimulation at 20-Hz pulse trains every 2 min for a total of 15 presentations. After the first block of pulse trains, animals were challenged with an experimental treatment. (Left) In the first experiment, to disrupt HA-r function we bath-applied 50 µM cimetidine (CIM) in saline vehicle continuously over the course of the experiment. (Center) In the second experiment of animals, to remove intrinsic HA input from the MDHns the neck connective was cut, thereby axotomizing the MDHns. (Right) In the third experiment of animals, direct bath application of HA (50 µM) in saline vehicle was used to simulate increased MDHn output during flight. Exemplar peristimulus rasters and histograms for the baseline responses (before) and during/after cimetidine (B), neck connective cut (D), and HA (F) treatments. Mean integrated power from 18 to 22 Hz by time across all recorded neurons that entrained to the pulsed odor at some point during cimetidine (C), neck connective cut (E), and HA (G) treatments. Error bars represent the SE. Results plotted as a function of time since treatment. Power was normalized by dividing mean power from each block by the mean baseline (pretreatment block) power. Arrows indicated the first block where there was a significant difference in power between experimental and control treatments (Welch’s t test for two samples with unequal variance; P < 0.05). Regressions are second order polynomials. Red rectangle (B) highlights the loss of responses to the first two pulses as a consequence of cimetidine relative to pretreatment.

Fig. 3.

HA-r blockade disrupts behavioral measures of olfactory acuity. (A) Acquisition of the conditioned feeding response to a single odor (2-hexanone) as a function of conditioning trial for groups of moths in the detection threshold assay. Twenty-four hours later, one group of moths was bilaterally injected with either 50 µM cimetidine (CIM) in saline vehicle or the saline vehicle without drug (saline) in a blind manner, then tested. (B) Conditioned feeding response as a function of odor concentration for the CIM and saline groups. Open and filled arrowheads indicate detection threshold concentrations, for the saline and CIM groups, respectively, as defined by the lowest concentration odor yielding a significant increase in response relative to the blank (one-tailed paired t test; n = 60; P < 0.001). (C) Acquisition of the differential conditioned feeding response to the CS⁺ and CS⁻ stimuli for CIM- and saline-injected groups. Moths were first differentially conditioned to one of the two odorants (2-hexanone or 2-octanone). Both odors were used as the CS⁺ and CS⁻ in separate but equally sized groups to counterbalance odor-dependent effects; for display, data were pooled by CS⁺ and CS⁻. (D) Discrimination index [(CS⁻) − (CS⁺)] displayed by concentration for the CIM- and saline-injected groups. Open and filled arrowheads indicate discrimination threshold, the concentration at which there was a significant differential response to the CS⁺ and CS⁻ odors using a one-tailed paired t tests (saline controls: P = 0.03; n = 46; CIM injected: P = 0.05; n = 43). All Regression lines are third-order polynomials and all error bars represent the SE.