The need to detect and process sensory cues varies in different behavioral contexts. Plasticity in sensory coding can be achieved by the context-specific release of neuromodulators in restricted brain areas. The context of aversion triggers the release of dopamine in the insect brain, yet the effects of dopamine on sensory coding are unknown. In this study, we characterize the morphology of dopaminergic neurons that innervate each of the antennal lobes (ALs; the first synaptic neuropils of the olfactory system) of the moth Manduca sexta and demonstrate with electrophysiology that dopamine enhances odor-evoked responses of the majority of AL neurons while reducing the responses of a small minority. Because dopamine release in higher brain areas mediates aversive learning we developed a naturalistic, ecologically inspired aversive learning paradigm in which an innately appetitive host plant floral odor is paired with a mimic of the aversive nectar of herbivorized host plants. This pairing resulted in a decrease in feeding behavior that was blocked when dopamine receptor antagonists were injected directly into the ALs. These results suggest that a transient dopaminergic enhancement of sensory output from the AL contributes to the formation of aversive memories. We propose a model of olfactory modulation in which specific contexts trigger the release of different neuromodulators in the AL to increase olfactory output to downstream areas of processing.
- antennal lobe
- biogenic amines
while searching for food and mates, animals encounter a huge diversity of sensory stimuli that the nervous system must encode and decide whether, and in what manner, to respond. This task is made all the more difficult because most resources have patchy distributions and varying reward values. This variability establishes different behavioral contexts in which sensory information is encoded by the nervous system. The nervous system must therefore adjust its activity so that behavioral output is maximally beneficial for survival within the present context. This plasticity is often accomplished when a given context triggers the release of specific neuromodulators within restricted brain regions from a small population of neurons. Modulators can modify neural processing by altering the efficacy with which neurons respond to stimuli or communicate with each other without directly causing inhibition or excitation (Kupfermann 1979). For instance, in the mammalian brain neurons from the ventral tegmental area project to many other areas and release dopamine (DA) within the context of reward (Alcaro et al. 2007; Berridge 2007; Schultz 2007; Wise 2004).
The insect olfactory system has evolved several modulatory systems to maximize foraging efficiency for resources that are patchy in their distribution. For instance, the flowers of Datura wrightii (a host plant of the moth Manduca sexta) open in the early evening, making them a temporally patchy resource. Foraging moths must therefore be maximally sensitive to olfactory cues at a specific time. The levels of serotonin in the antennal lobes (ALs; the first synaptic neuropil of the olfactory system) increase at this time (Kloppenburg et al. 1999), and serotonin increases the sensitivity and responsiveness of AL neurons (Dacks et al. 2008; Kloppenburg et al. 1999; Kloppenburg and Hildebrand 1995), suggesting that serotonin acts as a circadian modulator of olfactory sensitivity (Kloppenburg and Mercer 2008).
Resources can also be patchy in their reward value. For instance, there is natural variability in the nicotine (which is repellant) content of floral nectar in the flowers of the host plants of Manduca (Kessler and Baldwin 2007; Kessler et al. 2008). An association may then be formed between the nicotine content of the nectar and the floral features so that those flowers can be avoided in the future. DA release in brain areas downstream of the ALs has been demonstrated to act as a molecular signal for the occurrence of an aversive stimulus during olfactory learning in insects (Aso et al. 2010; Beggs et al. 2007; Beggs and Mercer 2009; Claridge-Chang et al. 2009; Mizunami et al. 2009; Selcho et al. 2009; Unoki et al. 2005; Vergoz et al. 2007; Wright et al. 2010; Zhang et al. 2008). However, the effects of DA on odor-evoked responses in the ALs are unknown, and the consequences of the effects of DA in the AL on behavior have not been examined. We therefore sought to explore the role of DA as a modulator of olfactory coding in the AL of Manduca within the context of aversive learning.
MATERIALS AND METHODS
Moths were reared in the Department of Neuroscience at the University of Arizona under a 17:7-h light-dark cycle as described previously (Christensen and Hildebrand 1987).
For DA immunoreactivity (DA-ir), brains were labeled with a protocol identical to that published by Dacks and Nighorn (2011), which also included preadsorption controls for the DA antibody with Manduca brain tissue. For immunochemical labeling of Manduca tyrosine hydroxylase (TH-ir), brains were dissected in insect saline and fixed overnight in 4% paraformaldehyde at 4°C. Brains were then washed in PBS, embedded, and sectioned as described previously (Dacks and Nighorn 2011) except for AMIRA reconstruction, for which whole mount brains were used. Sections were washed in PBS with 0.5% Triton X-100 (PBST), blocked for 1 h in PBST with 2% IgG-free BSA, and incubated for 2 days in 1:10,000 rabbit anti-Manduca TH (a generous gift from Dr. Maureen Gorman at Kansas State University) in PBST with 1% Triton X-100 and 50 mM sodium azide (PBSAT). Sections were then washed in PBST, blocked, and incubated overnight in 1:1,000 goat anti-rabbit Cy3 (Jackson Immunoresearch) in PBSAT. Tissue was then washed in PBST, cleared, and mounted as described previously (Dacks and Nighorn 2011). The full characterization of the anti-Manduca TH antibody is described by Gorman et al. (2007). In Western blots a second smaller, weaker band was observed (Gorman et al. 2007). We observed some very weak labeling in addition to the strong labeling that matched the observed DA-ir. We therefore only traced TH-ir processes in AMIRA (see below) that also matched the DA-ir processes innervating the AL. Anterograde fills of olfactory receptor neurons (ORCs) projecting to the macroglomerular complex (MGC) were performed by cutting the tips of the sex pheromone-sensitive sensillae of males and placing droplets of dextran-Texas red dissolved in insect saline over the stumps. The entire antenna was then coated in petroleum jelly to avoid desiccation (Molecular Probes), and the moths were allowed to rest overnight so that the dye could be taken up and transported throughout the ORCs. Brain were then dissected and processed as described above.
Images were collected with a Zeiss 510 Meta laser scanning confocal microscope equipped with argon and green HeNe lasers and appropriate filters. The Zeiss LSM Image Browser was used to create image stacks and to adjust contrast and brightness. CorelDRAW X4 (Corel, Ottawa, ON, Canada) was used to organize all images and figures. AMIRA reconstructions were created in Amira 4.1.2 from confocal scans of a whole mount brain labeled with the Manduca TH antibody with a custom AMIRA plug-in for reconstruction of three-dimensional branching patterns (Evers et al. 2005) generously provided by Dr. Felix Evers (University of Cambridge). The processes within the AL were not reconstructed, to avoid obscuring the branching patterns of these neurons within the rest of the brain.
Cloning of Manduca dopamine receptors.
Degenerate PCR and RT-PCR were performed as described by Dacks et al. (2006). Antennal lobe cDNA was isolated by cutting out the ALs only (which is relatively easy to do because of the large size of the ALs in Manduca) for mRNA extraction. Degenerate PCR primers were designed based on sequences from Bombyx mori, Drosophila melanogaster, Papilio xuthus, and Apis mellifera with Primer Premiere 4.1 (Premiere Biosoft International, Palo Alto, CA). Degenerate PCR primer sequences used to clone each of the MsDA receptors were 5′-CGTGATCTCCCTGGACMGNTAYTGGGC-3′ and 5′-GAACACCATCACGAACAGAGGNARRTARAA-3′ for the MsINDR, 5′-ACCGCYWSNATCTTCAACYTSTG-3′ and 5′-CCAGCANAVNADGAAGACGCCCAT-3′ for the MsDop1, and 5′-TGYTGGBTNCCNTTYTT-3′ and 5′-GTRTADATNAYNGGRTT-3′ for the MsDop2 receptors. Brain and AL cDNA were generated with the Omniscript RT kit (Qiagen, Valencia, CA), and AccuPrime Pfx Supermix (Invitrogen) was used to generate initial fragments for all three DA receptors. Rapid amplification of cDNA ends (RACE) was used to generate the full-length sequence for the MsINDR and MsDop1 receptors with the use of the SMARTer PCR Synthesis Kit (Clontech, Mountain View, CA) to generate the cDNA with a universal tag sequence and Advantage 2 Polymerase Mix (Clontech) to generate 5′ and 3′ fragments with touchdown PCR. The RACE primer sequences used were as follows: 5′-ACGGTGGTGAGTGAGAACAG-3′ for the 5′ MSINDR fragment and 5′-TTCATAGTCTGCTGGCTGCCATTC-3′ for the 3′ MsINDR fragment and 5′-GCATAGCAGTAGAGCCTGCAATAT-3′ for the 5′ MsDop1 fragment and 5′-GACTTTTGCAGGAGTCAACGACTTGC-3′ for the 3′ MsDop1 fragment. Despite many attempts with RACE, only a 144-nucleotide fragment of the MsDop2 receptor was obtained.
Sequence alignments in Fig. 3 were constructed with the program ClustalW (Combet et al. 2000) (http://npsa-pbil.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_clustalw.html). The GenBank accession numbers for the sequences used for the sequence alignments in Fig. 3 were MsINDR (JN_117928), MsDop1 (JN_117929), MsDop2 (not applicable), DmDD2R (NP_001014758), AmDop2 (NP_001011567), AmDop1 (NP_001011595), AmDop3 (NP_001014983), BmDopR2 (NP_001108338), BmDopR1 (NP_001108459), TcDop1 (not published, XP_971542), and TcINDR (not published, XM_967686).
Multichannel extracellular recordings.
Multichannel extracellular recordings were performed as described previously (Dacks et al. 2008). Briefly, 2- to 5-day-old moths were secured in plastic tubes with dental wax, and the cuticle and muscles lying above the brain were removed. The perineural sheath was removed from the AL, and physiological saline with 8.55 g/l sucrose and 200 μM ascorbic acid (to reduce the oxidation of DA) was immediately superfused over the brain. Pilot experiments found no effect of 200 μM ascorbic acid on AL responses. Sixteen-channel extracellular electrode arrays (NeuroNexus Technologies, Ann Arbor, MI; catalog no. 434-3mm 50-177) were inserted into the AL in parallel with the antennal nerve. Extracellular activity was acquired with a RX5 Pentusa base station (Tucker-Davis Technologies, Alachua, FL) and a RP2.1 real-time processor (Tucker-Davis Technologies), and spike data were extracted from the recorded signals and digitized at 25 kHz with the Tucker-Davis Technologies data-acquisition software. Threshold and gain settings were adjusted independently for each channel, and spikes were captured in the tetrode recording configuration: any waveform that passed threshold on one channel triggered the capture of waveforms recorded on the other three channels on the same shank. Offline Sorter v.3 (Plexon Neurotechnology Research Systems, Dallas, TX) was used to sort extracellular waveforms, and spikes were assigned timestamps to create raster plots and calculate perievent histograms in Neuroexplorer v.3 (Plexon Neurotechnology Research Systems).
The same experimental protocol was used for all concentrations of DA applied and the saline-only controls in which saline with 200 μM ascorbic acid was applied from different superfusion containers. AL neurons were stimulated with ten 200-ms pulses of a five-component D. wrightii blend (Riffell et al. 2009a) separated by 10 s at a 1:1,000 and then a 1:10 dilution in mineral oil. After the two sets of stimuli, 2 min of spontaneous activity was recorded. The ALs were then superfused with dopamine at 5 × 10−5 M (n = 23), 5 × 10−6 M (n = 6), or 5 × 10− 7 M (n = 6) or with physiological saline (n = 6) (a total of 106 neurons from 41 moths) with sucrose and 200 μM ascorbic acid for 5 min. The 200 μM ascorbic acid was added to the saline to prevent the breakdown of DA and was included for all conditions. The stimulation protocol was then repeated and another 2 min of spontaneous activity recorded. The ALs were then superfused with physiological saline for 15 min, the stimulation protocol was again repeated, and another 2 min of spontaneous activity was recorded.
Analysis of multichannel data.
After spikes were assigned timestamps, the number of spikes elicited by each odor stimulus was determined by counting the number of spikes that occurred in a 1-s window after the presentation of the stimulus. A neuron was considered to respond if the firing rate after stimulation rose above a threshold of 1.96 times the standard deviation of background firing rate. If the firing rate remained above the threshold for >1 s (which occurred infrequently), then only those spikes elicited in that 1-s period were counted. To calculate the duration of inhibitory responses, the number of 1-ms bins in which the firing rate dropped to zero within 500 ms after stimulus presentation were counted. The duration of the postexcitatory inhibitory phase (or “I2”) was calculated as the amount of time in which there were no spikes following an odor-evoked excitatory burst. All measures of background activity [mean interspike interval, coefficient of variation (CV), CV2, peak interspike interval, and mean firing rate] were calculated in MATLAB, and CV2 was calculated as described previously (Holt et al. 1996).
Nectar collection and analysis.
To provide a naturalistic context for the neural basis of aversive learning, we used the nectar and floral odors from the Manduca host plants D. wrightii and D. discolor. Many plants increase their toxic alkaloid content in leaf tissue and nectar when experiencing herbivory (Adler et al. 2006; Kessler et al. 2008), which subsequently can lead to decreased visitation by the floral visitors (Gegear et al. 2007). The decreased visitation has been shown to be mediated by aversive learning by the pollinators (Gegear et al. 2007; Wright et al. 2010). To examine the effects of herbivory on nectar chemical composition, we placed screen cages on D. wrightii and D. discolor branches that did not have flower buds. Manduca larvae were placed inside the screen cages and allowed to feed ad libitum on the vegetative tissue for 3 days. After 3 days, nectar standing crops were collected from flowers on the other branches of the plant at dusk (2000 PST) from at least 10 newly opened flowers of each species with 1-ml syringes. Control experiments were conducted in parallel with plants that had the screen cages but did not have larvae feeding on the vegetative tissue.
Individual nectar samples were partitioned into water-soluble and non-water-soluble fractions by adding 500 μl of methylene chloride. The methylene chloride fraction was extracted and concentrated by gently blowing nitrogen gas over the sample until 20 μl remained. Each sample was stored in a 2-ml borosilicate glass vial with a Teflon-lined cap at −80°C until analysis. Samples (1 μl) were analyzed with a gas chromatography-mass spectrometric detection system (GC-MS) consisting of an HP 7890A GC and a 5975C Network Mass Selective Detector (Agilent Technologies, Palo Alto, CA). A DB1 GC column (J&W Scientific, Folsom, CA; 30 m, 0.25 mm, 0.25 μm) was used, and helium was used as carrier gas at constant flow of 1 ml/min. The initial oven temperature was 50°C for 5 min, followed by a heating gradient of 6°C/min to 250°C, which was held isothermally for 6 min. Chromatogram peaks were identified tentatively with the aid of the NIST mass spectral library (∼120,000 spectra) and verified by chromatography with authentic standards (when available). Peak areas for each compound were integrated with ChemStation software (Agilent Technologies) and are presented in terms of nanograms per microliter of nectar.
Pharmacology and focal microinjection-conditioning experiments.
To evaluate the influence of DA within the AL on olfactory learning, DA receptor antagonists were focal-microinjected into the AL of 4-day-old male moths. As described previously (Lei et al. 2009), moths were restrained in a plastic tube 30 min prior to scotophase and kept at room temperature in the light awaiting surgery and injection. The head capsule was descaled and then opened, and the ALs were exposed for microinjection. Injection was accomplished via quartz pipettes (OD 1.0 mm, ID 0.70 mm, Sutter Instruments, San Diego, CA) pulled with a model P-2000 laser puller (Sutter Instruments) and clipped to allow solution passage. Pipettes were filled with the solution to be injected and connected with an output line of a dual-channel Picospritzer (Picospritzer II, General Valve, East Hanover, NJ). Pipettes were inserted into the center of each AL, and two drops (mean diameter ± SD: 82 ± 12.1 μm) were administered in quick succession. The diameter was determined before the experiment by placing the tip of the pipette into a beaker of mineral oil and measuring the droplet size with an ocular micrometer. After injection the cuticle window was repositioned and sealed with myristic acid (Sigma), and the moths were allowed to recuperate for 30 min before testing. We previously found that drugs delivered in this manner do not diffuse into other regions of the brain, and that the saline- and vehicle-injected moths do not behave significantly differently from noninjected moths (Lei et al. 2009).
For DA receptor antagonist experiments (see Fig. 8, B and D), moths were injected with either a mixture of 10−8 M SCH39166 (a D1 receptor antagonist) and 10−8 M L-741,626 (a D2 receptor antagonist) (both from Tocris Bioscience) or vehicle (moth saline) or were not injected. Receptor antagonists were selected for their extreme selectivity for DA receptors (Bowery et al. 1996; McQuade et al. 1991). In addition, we performed experiments in which the ALs of moths were injected with SCH23390 (see Fig. 8C) at 10−6 M and tested the ability of moths to form an aversive association between the Datura odor and the simulated toxic nectar (see below). SCH23390 (Tocris) was selected for its effectiveness in blocking the Dop1- and INDR-type receptors in the silk moth Bombyx mori (Ohta et al. 2009). Because of the high degree of sequence similarity for the DA receptors between Manduca and Bombyx, SCH23390 is likely an effective Dop1- and INDR-type receptor antagonist in Manduca. However, it should be noted that SCH23390 has not been directly tested on serotonin or octopamine receptors in moths, although SCH23390 is ∼25 times more effective at blocking the effects of DA compared with octopamine in crude membrane preparations from cockroach brain (Orr et al. 1987). All solutions included 0.1 μM DMSO. For the dopamine injection experiments (see Fig. 8E), the ALs were injected with dopamine-HCl at 5 × 10−5 M diluted in moth saline with 200 μM ascorbic acid (to prevent the breakdown of DA), moth saline with 200 μM ascorbic acid (vehicle), or saline alone.
We used a forward-paired conditioning paradigm to examine the effects of the pharmacological manipulations while the animal learned to associate an odor with an aversive stimulus. In the forward-paired condition, the artificial mimic of the D. wrightii floral scent was delivered in a 3-s pulse. One second after odor onset, the unconditioned stimulus, which was either 1 μl of 0.1 M quinine or the “toxic nectar” (30 μM scopolamine in 20% sucrose), was applied to the proboscis for ∼2 s. A 10-min intertrial interval separated each training trial, and moths were trained over seven or eight trials. After the conditioning trials were completed, a test trial was performed during which only the D. wrightii mixture was presented to assess the behavioral responses as a result of the conditioning treatment. The ability of moths to form aversive association in backward- or random-pairing training paradigms was not tested because of the inability of moths to form appetitive associations with these training procedures (Daly and Smith 2000). The toxic nectar used in the forward-paired paradigms is based on the scopolamine levels shown in Table 1 and Fig. 8A and sucrose levels normally found in nectar, as herbivory did not affect sugar concentrations. Fourteen to forty-nine moths were used for each drug and control treatment group (n = 247 total moths).
All statistical analyses were performed with GraphPad Prism 5.01 (GraphPad Software, La Jolla, CA) or SPSS 18.0 (IBM, Armonk, NY). For electrophysiological measures a D'Agostino and Pearson omnibus normality test was applied, and if data were normally distributed a single-factor repeated-measures ANOVA with a Tukey honestly significant difference (HSD) post hoc test was calculated. If data were not normally distributed, a Kruskal-Wallis test with a Dunn's multiple comparison post hoc test was performed (P < 0.05 was assigned as a significance threshold for both tests). For all normalizations, response measures of a given neuron (such as peak firing rate or inhibition duration) were normalized to the maximal response for that neuron to all of the odor stimuli under all of the treatment conditions. In all behavioral experiments, repeated-measures binary logistic regression modeling was used to analyze proboscis extension response (PER) between treatments. Fisher's exact test was used to make specific pairwise comparisons among test odors and treatments within trials.
Antennal lobes of Manduca are innervated by two pairs of dopaminergic neurons.
To identify dopaminergic input to the ALs of Manduca we labeled the brains of adult moths for both DA and Manduca TH, the rate-limiting enzyme essential to the production of DA. The AL is comprised of spherical neuropil called “glomeruli” that receive input from ORCs on the antennae. Olfactory information is transmitted to downstream brain areas via the projection neurons, and glomeruli are interconnected by a diverse population of local interneurons. DA-ir and TH-ir were observed in every glomerulus of the AL (Fig. 1, A and D, respectively), and the DA-ir and TH-ir fibers innervated the entire glomerular volume (Fig. 1, B and D, respectively) with the exception of the distal area of the pheromone-sensitive MGC of male moths (Fig. 1, C, E, and F). The glomeruli that comprise the MGC are distinct from the other glomeruli in the AL in that they receive input from ORCs that are selectively tuned to single components of the female sex pheromone. Anterograde dye fills of the pheromone-sensitive ORCs on the antennae demonstrate a lack of overlap between the TH-ir processes and the receptor neurons innervating the MGC (Fig. 1F). The widespread nature of the projections of the DA-ir/TH-ir neurons within the AL suggests that the activation of these neurons likely affects odor-evoked responses in all glomeruli, rather than a few specialized glomeruli. Both DA-ir and TH-ir revealed four neurites innervating the ALs (Fig. 1, C and D). These four processes were traced from the ALs in whole mount TH-ir preparations to two pairs of cell bodies (Fig. 2, A–C) in the dorsal protocerebrum, a central brain region in insects. Each pair of cells projects from the dorso-posterior protocerebrum to both ALs as well as other higher areas of processing, including the lateral horn, forming an archlike structure that bridges both ALs (Fig. 2, D–F). Because of the arching morphology of these neurons, we refer to them as the dopaminergic arching (DAAr) neurons.
Although the AL is uniformly innervated by DA-ir/TH-ir fibers, the effects of DA on AL neurons could be heterogeneous depending on the DA receptors expressed in the AL. We cloned homologs of the DA 1-type receptor (Dop1), DA 2-type receptor (Dop2), and invertebrate DA receptors (INDRs) from Manduca, all of which share high levels of sequence identity with other insect DA receptors (Fig. 3; MsINDR; 92% sequence identity with B. mori, MsDop1; 93% sequence identity with B. mori, MsDop2 fragment; 100% sequence identity with D. melanogaster). RT-PCR of AL cDNA revealed that all three Manduca DA receptor homologs are expressed in the AL (Fig. 4) and thus at least these three receptors are potential targets for any modulatory effects of DA in the AL. Thus the ALs of Manduca express several DA receptors and receive dopaminergic input from four centrifugal neurons.
Responses of the majority of antennal lobe neurons are enhanced by dopamine.
In Drosophila (Yu et al. 2004) and Manduca (Daly et al. 2004) the responses of AL neurons are transiently enhanced after the formation of aversive or appetitive associations, respectively. This is striking as it suggests that despite the opposing natures of appetitive and aversive contexts, there is an increase in the responsiveness of the AL associated with the formation of either type of olfactory association. Similarly, both octopamine (Barrozo et al. 2010) and serotonin (Dacks et al. 2008; Kloppenburg et al. 1999) enhance odor-evoked responses of AL neurons in moths, despite being associated with different behavioral contexts. We therefore sought to determine the effects of DA on odor-evoked responses of AL neurons. We performed extracellular multichannel recordings of the responses of AL neurons to two concentrations of the innately attractive (Raguso and Willis 2002, 2005; Riffell et al. 2008, 2009a) odor of D. wrighti flowers, from which adult Manduca feed (Raguso et al. 2003; Riffell et al. 2008). DA increased the number of odor-evoked spikes (Fig. 5A) in 58% of responsive AL neurons (n = 30 of 52 neurons that responded to the Datura odor), and this effect was DA dose dependent (Fig. 5B). Of those neurons enhanced by DA, there was an average 58.5% increase in elicited spikes (Fig. 5C; single-factor repeated-measures ANOVA: P < 0.0001). Thus DA increased the magnitude of odor-evoked responses of most AL neurons. DA did not affect the responses of 31% of AL neurons and significantly reduced the responses of 11% of AL neurons (Fig. 5D). DA also decreased the threshold for activation of a few AL neurons (Fig. 5E), suggesting that DA may increase the sensitivity of AL neurons, resulting in more cells participating in the encoding of olfactory stimuli.
DA also affected the slow temporal dynamics of AL responses (i.e., firing patterns that evolve over hundreds of milliseconds), which can provide information about the identity of an odor (Laurent et al. 1998) and the physical structure of an odor plume (Vickers et al. 2001). The excitatory responses of Manduca PNs are often preceded by a rapid GABAA-dependent inhibition (referred to as “I1”) and followed by a period of spike suppression (“I2”) lasting anywhere between 10 ms and 1.5 s (Christensen et al. 1996). The I2 phase shortens the duration of PN responses after stimulus offset, allowing tracking of odor intermittency (Lei et al. 2009; Tripathy et al. 2010) that is both inherent in the odor plume structure (Vickers et al. 2001) and produced by the beating of the moth's wings (Sane and Jacobson 2006). DA caused a 24.5% decrease in the duration of I2 (Fig. 6, A and B; Kruskal-Wallis test: P < 0.0001) in 73.5% (25 of 34 units) of units that displayed the I2 phase. This effect was not due to the enhancement of the magnitude of the excitatory phase overwhelming the I2 phase, as higher odor concentrations (which increase response magnitudes) did not affect I2 duration (Fig. 6C). DA did not, however, affect the duration of purely inhibitory responses (Fig. 6D), suggesting that odor-evoked inhibition and the I2 phase are mediated by different cellular or network mechanisms. Furthermore, DA had no apparent effects on I1 phase. In addition to odor-evoked activity, 2 min of background activity was recorded before, during, and after DA application to determine whether DA affected the background activity of AL neurons. However, there was no effect of DA on several measures of background activity of AL neurons (mean interspike interval, CV, CV2, peak interspike interval, and mean firing rate; Fig. 6, E–I, respectively; Kruskal-Wallis test).
The global release of a neuromodulator should result in widespread effects on the representation of a stimulus by an ensemble of neurons within a network. However, DA did not homogeneously affect the responses of individual neurons and may therefore alter the population response to an odor in one of several ways. If DA sharpened the overall AL response, we would expect an enhancement of the strongest responses and a decrement of the weakest responses (Sachse and Galizia 2002), increasing the contrast or signal-to-noise ratio. Conversely, if DA broadened the AL responses, then DA would either decrease or have little effect on the strongest responses and increase the weaker responses (Olsen et al. 2007; Silbering et al. 2008), thus recruiting more cells to participate in the odor representation. In contrast to these two options, we observed that the relative strength of a neuron's response does not predict the modulatory effects of DA (Fig. 7A). Thus the ensemble response is not homogeneously sharpened or broadened but reorganized by the enhancement of most neurons across the spectrum of response strength, the decrement of some responses, and the recruitment of additional neurons. Comparing pre-DA responses to post-DA responses across the population (Fig. 7B), the best-fit line is positively shifted from a 1:1 pre- to post-DA response ratio, indicating that responses are increased across the population. However, the slope of this line is close to 1 (slope = 0.9756), indicating that the overall population response was increased and that neither significant sharpening (slope > 1) nor broadening (slope < 1) of the population response occurred.
Injection of dopamine receptor antagonists into the AL prevents formation of an aversive association.
Given the clear effects of DA on the AL responses of Manduca, and the fact that the release of DA within the mushroom bodies of Drosophila is necessary (Schwaerzel et al. 2003) and sufficient (Claridge-Chang et al. 2009) for aversive learning, we investigated the role of DA within the AL in aversive learning. Aversive learning assays in insects have predominantly relied on electric shock as an aversive stimulus. Here we take advantage of the natural variation in the amounts of aversive compounds in the floral nectar of the host plants of Manduca (Kessler et al. 2008). Plant herbivory has cascading effects on plant physiology that, in turn, influence diverse processes. Herbivory alters the volatile chemicals emitted by host plants of Manduca (Kessler et al. 2008) as well as the time of day at which these plants flower (Kessler et al. 2010). We therefore tested whether herbivory could also alter the chemical content of host plant nectar. Manduca caterpillars were allowed to feed on the vegetative (but not floral) tissue for 3 days, and then the nectar content of the flowers was collected and analyzed with a GC-MS system. Herbivory increased the amount of the tropamine alkaloids (e.g., scopolamine) in the floral nectar of both D. wrightii and D. discolor (Fig. 8A and Table 1). This, in combination with natural variability in the nectar content of nicotine, which is repellant to Manduca (Kessler et al. 2008) and honeybees (Kohler et al. 2012), suggests that there are instances in the field in which pollinators encounter flowers with aversive nectar. We therefore tested whether moths could form an association between an appetitive olfactory stimulus (the Datura floral blend) (Riffell et al. 2009b) and an aversive tastant (a synthetic blend of the nectar of herbivorized Datura plants), and whether the formation of such an association could be blocked by DA receptor antagonists.
Naive Manduca extend their proboscis in response to the Datura floral blend in both a tethered feeding assay (see materials and methods) and flight assays (Raguso and Willis 2002, 2005; Riffell et al. 2009a, 2009b), indicating that the Datura odor is an innately attractive stimulus. This feeding response was eliminated when a synthetic nectar mimicking the chemical composition of the nectar from larva-damaged plants was repeatedly paired with the Datura odor (Fig. 8B; Fisher's exact test; P < 0.005) in a forward-paired learning paradigm. Trained moths did not respond to the Datura blend 2 h after training (Fig. 8B; Fisher's exact test; P < 0.005), thus representing a learned aversion for the Datura blend. To test the necessity of DA in the AL for aversive memory formation, DA receptor antagonists were injected directly into the ALs of adult moths. Moths injected with either a mixture of D1 and D2 receptor antagonists (Fig. 8B; SCH39166 and L-741,626, respectively) or a D1 receptor antagonist (Fig. 8C; SCH23390) responded in the same proportion at the end of the training protocol as at the start, and significantly more than control animals (Fig. 8B; logistic regression: χ12 > 7.15, P < 0.01). In particular, animals injected with the mixture of D1 and D2 receptor antagonists produced significantly more PER responses than control animals at the end of the training protocol (Fisher's exact test; P < 0.001) and had elevated responses compared with animals injected with the D1 receptor antagonist SCH23390 (Fig. 8C; logistic regression: χ12 = 6.84, P = 0.009). In control experiments in which moths injected with saline only were repeatedly exposed to either the Datura blend or a mineral oil blank without exposure to an aversive stimulus, the proportion of moths responding to either olfactory stimulus remained unchanged (Fig. 8D). This indicated that the decrease in the proportion of moths responding to the Datura odor when it is paired with the aversive stimulus is due to associative processes. Moths injected with the DA receptor antagonists continued to produce feeding responses at the same proportion 2 h after training (Fig. 8, B, C, and E). The DA receptor antagonists could not have prevented the detection of the Datura odor, as injected moths still produced feeding responses when the odor was presented. Furthermore, the injection of the DA receptor antagonists did not cause a nonassociative increase in the PER that masked the decrease in responses, as there was no difference between treatments in the proportion of naive moths responding at the start of the experiment (trial 0; Fig. 8B, C, and E). The injection of the mixture of DA receptor antagonists also prevented the formation of an aversive association between the Datura floral blend and a high-concentration bitter tastant, 0.1 M quinine (Fig. 8E), both immediately after training and 2 h later (Fisher's exact test; P < 0.001). When DA was injected directly into the ALs, very few moths exhibited feeding responses to the Datura odor shortly after the injections (Fisher's exact test; P < 0.01), whereas the majority of moths injected with either saline or vehicle extended their proboscis in response to the Datura odor (Fig. 8F). However, 2 h after injection moths in all treatment groups responded at the same proportion (Fig. 8F).
The nervous system must adjust its activity to best suit the behavioral state of the individual animal, which, in turn, is often established by the external and internal environments. In this study, we explored the effect of an aversive food stimulus on the encoding of an associated odor in the primary olfactory system. The nectar resources that Manduca encounters while foraging in the field vary in their reward value, with some expressing aversive compounds (like nicotine and scopolamine). The nervous system of Manduca must be able to accentuate sensory cues associated with aversive stimuli. We demonstrate that, in addition to the established role of mediating the formation of aversive olfactory associations in higher brain areas, DA plays a more transient role of enhancing the response to olfactory stimuli in the primary sensory neuropil.
DA acts as a molecular mediator of aversive learning in higher brain areas in a variety of insect species (Aso et al. 2010; Beggs et al. 2007; Beggs and Mercer 2009; Claridge-Chang et al. 2009; Keene and Waddell 2005; Mizunami et al. 2009; Schwaerzel et al. 2003; Selcho et al. 2009; Unoki et al. 2005, 2006; Vergoz et al. 2007; Wright et al. 2010; Zhang et al. 2008) and has also been implicated in appetitive learning in insects (Kim et al. 2007; Krashes et al. 2009; Selcho et al. 2009). Despite the extensive study DA has received in olfactory learning, the effects of DA release in the primary olfactory neuropil were unknown. We found that two pairs of DA-ir/TH-ir neurons (the DAAr neurons) innervated widely throughout each AL, innervating all of the glomeruli, although only the proximal portion (with relation to the center of the AL) of the sex pheromone-sensitive MGC glomerulus, which is occupied by the ORCs (Fig. 1, E and F). Anterograde fills of the ORCs revealed a lack of overlap between the TH-ir processes and the ORCs, reminiscent of the innervation pattern of GABA (Reisenman et al. 2011) and serotonin (Sun et al. 1993) in the MGC of Manduca. This suggests that the MGC may be less subject to presynaptic modulation relative to the other glomeruli. In fruit flies and bees the ALs are innervated by dopaminergic neurons that are local to the ALs (Chou et al. 2010 and Kirchhof et al. 1999, respectively) and are thus morphologically quite distinct from the DAAr neurons. Neurons similar in morphology to the DAArs have been partially filled from the AL of Manduca (Homberg et al. 1988), and the location of the cell bodies of the DAAr neurons is similar to the PPL1 neurons in Drosophila, although the morphologies of the DAAr and the PPL1 neurons are not similar (Mao and Davis 2009).
Despite the widespread projections of the DAAr neurons in the AL, DA did not affect all AL neurons in the same manner, suggesting that the MsDA receptors may not be homogeneously expressed in AL neurons. In insects, activation of different DA receptors has opposing effect on cAMP levels (reviewed in Mustard et al. 2005), lending support to a model in which homogeneous DA release results in heterogeneous effects on neural responses. For instance, in Drosophila, selective dopamine 1-like receptor activation has been attributed to the modulation of cholinergic postsynaptic excitatory potentials (Yuan and Lee 2007). Thus the DA-induced decrease in response magnitude observed in some cells (Fig. 5D) may be attributable to a direct effect of the activation of specific DA receptors, although it could also be due to the enhancement of local interneurons that provide inhibitory input. Furthermore, different MsDA receptors could mediate distinct effects of DA on AL neuron responses such as the observed enhanced excitation (Fig. 5) or shortened postexcitation inhibition (Fig. 6). For instance, the Dop1-type receptor of Drosophila expression in the mushroom bodies is required for both appetitive and aversive conditioning (Kim et al. 2007), and rescuing the expression of this receptor in the gamma lobes of the mushroom bodies can fully restore the ability of flies to make aversive associations (Qin et al. 2012). In addition, the Drosophila INDR (DAMB) is highly expressed in the mushroom bodies in a manner highly overlapping with rutabaga adenylate cyclase (Han et al. 1996), which is thought to act as a molecular coincidence detector during classical conditioning. The framework of this model is similar to vertebrate systems in that a small number of neuromodulatory neurons act on diverse and heterogeneously expressed receptors in a sensory area. For instance, there is a diverse array of serotonin receptors expressed by different cell types within the olfactory bulb (McLean et al. 1995; Morilak et al. 1993; Pazos and Palacios 1985), which is innervated by centrifugal neurons from the raphe nuclei (McLean and Shipley 1987). Not only does serotonin have diverse effects on the response properties of the different cell types in the olfactory bulb (Hardy et al. 2005; Liu et al. 2012; Petzold et al. 2009), but it also facilitates, but is not necessary for, olfactory learning (Langdon et al. 1997). The heterogeneity of the effects of biogenic amines within any neural circuit underlines the importance of future studies in which the expression patterns and the effects of selective activation of the receptors are examined.
Different behavioral contexts can establish similar levels of arousal. Although DA, serotonin, and octopamine differ in the behavioral context in which they are released, they each produce a general increase in the magnitude of the majority of odor-evoked responses in the ALs of moths, although there are subtle differences in the effects of these modulators on odor-evoked responses such as the effects of DA on I2 duration (Fig. 6). It should also be reiterated that not all AL neuron responses were enhanced by either DA or serotonin. Serotonin increased the odor-evoked responses of 50% of AL neurons in Manduca, while causing a reduction in 6% of AL neurons (Dacks et al. 2008). Furthermore, in some instances the decreases in odor-evoked responses caused by serotonin were odor dependent, suggesting that serotonin modulated the inhibitory input to the AL neurons. There are multiple serotonin (Dacks et al. 2006) and DA receptors expressed in the AL, further complicating the consequences of modulator release, which could therefore be heterogeneous because of the different effects of the activation of each receptor type or the changes in the lateral input received by different neurons as has been shown for serotonin in insects and vertebrates (Dacks et al. 2008, 2009; Hardy et al. 2005; Liu et al. 2012; Petzold et al. 2009). In this study, the enhancement by DA selectively increases the activity of particular AL neurons. However, the responses of this subpopulation are uniformly elevated, increasing the proportion of neurons participating in the response without altering (e.g., broadening or sharpening) the signal-to-noise ratio of the population activity in the AL. Theoretically, this could result in an increase in the overall sensitivity of the olfactory system in a manner similar to lateral excitation evoked at low odor concentrations (Yaksi and Wilson 2010). In addition to enhancing odor-evoked responses, DA also affected the slow temporal patterning of olfactory-driven responses by decreasing the postexcitatory inhibition phase (I2). Although the effects of DA on the conductances of Manduca AL neurons are unknown, DA decreases a Ca2+-dependent K+ conductance in in vitro honeybee AL neurons (Perk and Mercer 2006), which could in theory result in a decreased postexcitatory inhibition duration. Whatever the biophysical properties modulated by DA, the decreased I2 phase could allow for a faster recovery from excitation, while still preserving the mechanism by which AL neurons can track the temporal dynamics of an odor stimulus (Lei et al. 2009; Tripathy et al. 2010). It should also be noted that it is not known whether the AL neurons affected by one amine are also affected by other amines. However, the large proportion of neurons in the AL affected by biogenic amines makes it likely that there must be some overlap.
The injection of DA receptor antagonists prevented the formation of an aversive olfactory association, suggesting that DA increases the likelihood that signals from the ALs are associated with input encoding the aversive stimulus in higher-order associative centers. However, DA injection into the AL alone was not sufficient to induce an aversion to a specific odor. There are two possible explanations for this result. This may have been due to a lack of temporal pairing between the DA injection and the presentation of the Datura odor. We attempted to perform repeated forward pairings of DA injection into the AL with presentation of the Datura odor; however, this had adverse consequences for the health of the moths, which began to extend their proboscis in response to the injections alone after a few trials. Although it is possible that a tight temporal correlation between the DA application and the odor stimulus is required for the formation of an aversion to the Datura odor, it is also possible that the effects of DA on AL neurons are not sufficient to induce the formation of an aversive memory. DA release within the AL may serve as a sensitizing modulator within the context of aversive learning, increasing AL output and occurring in combination with the release of DA in the mushroom bodies observed in Drosophila (Aso et al. 2010; Claridge-Chang et al. 2009; Schwaerzel et al. 2003). The MB-MP1 dopaminergic neurons in the mushroom bodies of Drosophila are sufficient (Aso et al. 2010) but not necessary for the formation of aversive olfactory associations (Krashes et al. 2009), so it is conceivable that the release of DA in the ALs of Manduca could be necessary but not sufficient for aversive learning. Furthermore, rescue of the rutabaga adenylate cyclase exclusively in the mushroom bodies of Drosophila is sufficient to restore aversive learning (Mao et al. 2004; Zars et al. 2000), further suggesting that the ALs themselves are not the site of memory formation in the brain, despite the observations that olfactory representations in the AL change during learning. However, DA injection did cause an immediate decrease in feeding responses to the normally appetitive Datura floral odor, similar to an aversive response. Thus there may be some information encoded in DA-modulated AL output provided in the short term about the aversive context in which an odor is encountered, perhaps encoded in the identity of neurons that are selectively enhanced by DA.
We therefore propose that the release of modulatory amines in the AL from a small number of neurons that reside outside the AL serves to enhance olfactory responses in a particular behavioral context. This enhancement occurs in tandem with the processing of relevant contextual cues in other brain regions, leading to appropriate behavioral responses. Because resources are patchy in their temporal and spatial distribution as well as their reward value, there are several modulatory systems that alter olfactory processing within these different contexts. In addition to the proposed role of serotonin as a circadian modulator of olfactory sensitivity in moths (Kloppenburg and Mercer 2008), serotonin has been proposed in honeybees to signal the malaise induced after consumption of toxic foods (Wright et al. 2010) and octopamine is thought to signal rewarding stimuli during the formation of appetitive olfactory associations (Hammer and Menzel 1998; Schwaerzel et al. 2003). We found that DA enhances the odor-evoked responses of AL neurons and that DA receptor antagonists injected directly into the AL prevent the formation of aversive olfactory associations. Our data suggest that, similar to octopamine and serotonin, DA enhances olfactory responses in the ALs, but within a different context, that of aversion.
This work was supported by National Institutes of Health Grant DC-42092 to A. J. Nighorn and NIH Training Grant 1 K12 Gm00708 to A. M. Dacks.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: A.M.D., J.A.R., and A.J.N. conception and design of research; A.M.D., J.A.R., and A.J.N. performed experiments; A.M.D., J.A.R., J.P.M., S.L.G., and A.J.N. analyzed data; A.M.D., J.A.R., J.P.M., and A.J.N. interpreted results of experiments; A.M.D., J.P.M., S.L.G., and A.J.N. prepared figures; A.M.D. drafted manuscript; A.M.D., J.A.R., J.P.M., S.L.G., and A.J.N. edited and revised manuscript; A.M.D., J.A.R., J.P.M., S.L.G., and A.J.N. approved final version of manuscript.
We thank K. Teter and O. Zaninovich for cloning the MsDARs, L. Oland, N. Gibson, and P. Jansma for technical assistance with immunocytochemistry and confocal microscopy, M. Gorman for supplying the rabbit anti-MsTH antibody, F. Evers, C. Duch, and M. Landis for assistance with AMIRA, E. Constantopolis, B. Medina, and E. Lutz for help with behavioral experiments, and A. Paulk, and P. Dacks for constructive comments on the manuscript.
- Copyright © 2012 the American Physiological Society