Searching for a particular pattern of neural activity responsible for a defined behavior is challenging because of the difficulty of establishing a causal link. In this study we confronted this problem by successfully disrupting MGC-PNs' ability to generate discrete bursts of action potentials and to follow repeated odor pulses that mimic the intermittency of natural odor plumes. Such a bursting response pattern was also observed in a previous study in which the moth was exposed to a pheromone plume and the electroantennogram (EAG) and firing activity of MGC-PNs were simultaneously recorded [21
]. The discontinuous nature of wind-borne plumes was clearly demonstrated in that study by the individual EAG peaks that were found to be tightly correlated with the bursting responses of the PNs [21
]. These findings suggest that MGC-PNs resolve the temporal discontinuity of a pheromone plume, which is known to be crucial for the flight behavior of a male moth seeking an unseen source of sex pheromone [6
]. The bursts of spikes were locked to the haphazard, high-frequency contacts with pheromone filaments in the plume. A missing link, established in this study, was the causal relationship between the PNs' bursting response pattern and the odor-modulated flight behavior of the moth.
Bicuculline methiodide effectively and reversibly disrupted the ability of PNs to encode intermittent odor pulses (Figure ), consistent with previous work, which also suggested that such disruption may result from antagonizing GABAA
receptors in PNs [29
]. This disruptive effect has now been more carefully quantified in the current study. The autocorrelation-based PFI was significantly lower for bicuculline-treated than untreated neurons for odor-delivery rates of up to 5 pulses s-1
(Figure ), implying that the bicuculline treatment would affect the orientation behavior if a moth encountered odor filaments at frequencies of 5 pulses s-1
or fewer in a natural plume. Through dynamic scaling of the turbulent conditions in our wind tunnel, we were able to control the filament frequency of the odor plume in the range of 1.98–2.5 pulses s-1
as determined by EAG recordings, tracer plume experiments and anemometry (supplemental Figure 4 in Additional data file 1), and the estimated filament-encounter frequency was about 4 pulses s-1
(Additional data file 1: experimental procedures and supplemental Table 1). Because of the boundary-layer effect around the moth antennae, which prolongs the pheromone concentration decay time [31
], the ORC activation frequency may be further decreased from the encounter frequency, although biological and physical phenomena, including three-dimensional turbulence, kinematics of the moth flight (change in velocity, acceleration), and interaction between air movement generated by the moth wing-beat and the wind velocity [32
], make accurate determination of the ORC activation frequencies difficult, if not impossible.
In our experiments, the flight-track analysis showed that although the unoperated and saline-injected animals spent most of the time heading directly toward the odor source, the bicuculline-injected moths were unsuccessful at steering a zero-degree track angle relative to the odor source despite being capable of making upwind progress (Figure ). As a consequence, a significantly lower percentage of bicuculline-injected moths exhibited close hovering, source contact, and abdomen curling (Figure ). These behavioral modifications are best explained by the alteration of PN response pattern caused by the action of bicuculline. Although clarifying the exact cellular mechanisms of bicuculline effects is beyond the scope of this study, our data suggest that these effects did not originate from the ORCs (supplemental Figure 5a–c in Additional data file 1) and were calcium dependent (supplemental Figure 5d–h in Additional data file 1).
According to a model proposed by Baker [11
] based on studies of lepidopteran species, phasically modulated neural responses are responsible for generating upwind surges on contact with a pheromone plume, and separate tonic responses (resulting from non-olfactory input) are responsible for activating an internal counterturning program, the behavioral output of which is the cross-wind casting. Moreover, the tonic response can be inhibited by the odor-induced phasic response. Observations of Drosophila melanogaster
differ noticeably from findings with moths in showing upwind surge even with a homogeneous odor cloud [27
]. Our results, however, support the Baker model.
The bursting response generated by PNs upon contact with each odor filament is a critical component of the olfactory code responsible for upwind surges. In a natural odor plume, the arrival of odor packets at appropriate frequencies produces a series of fused upwind surges, which often appear as approximately straight flight tracks toward an odor source (Figure ). Transforming the discrete bursting response to prolonged excitation using bicuculline caused the moth to lose orientation toward the odor source and to perform the counterturning behavior more frequently (Figure ). The correlation between the prolonged excitation of PN response and the increased casting behavior suggests that this response pattern may function to shut down the upwind surge and unmask the internal tendency for casting. The internal counterturning program may be autonomously activated by non-olfactory stimuli at a center downstream from the AL, which may use a gating mechanism to filter the AL outputs carried by PNs. When there is no phasic (or bursting) input to this center, it may produce alternating antiphasic signals [34
] that drive the casting behavior. The bursting responses of PNs, caused by intermittent stimulation, then inhibit the internal counterturning program, thus producing upwind surges. On the other hand, when the circuitry of this center is overloaded with PN inputs (prolonged excitation), it may become adapted and leave its alternating antiphasic output unmodulated. Behavioral experiments of moths in a homogeneous plume with unidirectional wind support this hypothesis [7
]. In such an environment the animal receives long-lasting stimulation, which may cause heterogeneous response patterns among PNs. Some PNs may produce a continuous spiking response matching the stimulus duration [35
], and others may produce random bursts within the stimulation period [29
]. In either case the PN population as a whole may effectively cause their target neurons to adapt, resulting in casting behavior. Conversely, in nature the PN population may be entrained by stimulus dynamics, and thus only phasically activate their target neurons, resulting in upwind surge. Although bicuculline treatment altered the spontaneous spiking pattern of MGC-PNs (Figure ; supplemental Figure 1 in Additional data file 1), these changes did not seem to affect the moth's crosswind casting behavior. Our data therefore suggest that the spontaneous firing pattern of MGC-PNs, whether or not modulated by drug treatment, contributes little if at all to the activation and sustaining of the counterturning program.
To determine the relationship between MGC-PNs' pulse-following ability and the pheromone-modulated orientation behavior of male moths, it is important to ask if the treatment with bicuculline also caused other changes, such as an altered firing rate, that might contribute to the moth's inability to track the odor plume in the wind tunnel. Experimental results showed, however, that the bicuculline treatment did not significantly change the response magnitude over a large range of pheromone concentrations (supplemental Figure 3 in Additional data file 1). Moreover, bicuculline treatment had no detectable effect on ORC activities whereas it did affect simultaneously recorded PNs (supplemental Figure 5a–c in Additional data file 1). This was probably due to some differences in ion conductances between ORCs and PNs. Thus, we conclude that the ability of PNs to respond to olfactory stimuli and encode odor concentrations (that is, to increase their firing rate proportionally to increasing odor concentration) are not affected by bicuculline treatment. Instead, the temporal response pattern is the feature that is significantly modified by the treatment.
Although bicuculline may not affect only the neurons associated with MGC, non-MGC neurons are unlikely to contribute to pheromone-mediated behaviors as pheromonal stimuli do not cross-excite non-MGC glomeruli [36
]. Moreover, the experiments in which bicuculline was injected into the MGC of male moths that were subsequently tested for flight responses to behaviorally effective mixtures of floral odorants demonstrated that the drug-injected moths behaved as well as the unoperated animals in the wind tunnel (Figure ). These findings suggest that little, if any, of the drug diffused beyond the MGC within the test time window, perhaps owing to the glial investment that ensheathes each glomerulus in the AL [38