A central neural circuit for experience-independent olfactory and courtship behavior in Drosophila melanogaster

P[GAL4] line GH146 was described in ref. 16. UAS-reporter strains were UAS-lacZ (4), UAS-tau (17), UAS-tra (18), and the UAS-TeTxLC line TNT-E (called TNT for simplicity) (5). These lines and the wild-type strain Canton-S (CS) were used as controls in behavioral tests.

β-Galactosidase and antibody staining of whole mount preparations and sections was done according to ref. 8. Anti-TAU antibody was from Sigma, and anti-TeTxLC was kindly provided by H. Niemann (Universität Hannover, Hannover, Germany). The Vectastain ABC system was used for detection of primary antibodies (Vector Laboratories).

Olfactory choice assays were carried out in the dark by using a T-maze apparatus (19). About 50 males or 50 females, 4–5 d old, were inserted into the apparatus, which was connected to two constant air streams, one of which carried the test substance. After 90 sec, the flies in both tubes were counted. A response index was calculated as No − Nc/No + Nc, No and Nc being the number of animals inside the odor and control tubes, respectively. Odors tested were ethyl acetate (Fluka), trans-2-hexenal (Aldrich), 2-hexanol (Aldrich), and 1-octen-3-ol (Fluka). All substances were diluted in paraffin oil (Fluka). To determine the threshold of response for GH146/TNT heterozygotes, we first tested a range of dilutions. The concentration leading to clearly abnormal behavior was then used for additional tests.

Tests were done as described in ref. 20. Single 3-d-old females and males were starved for 24 h, anesthetized with CO₂, then fixed on slides with tape and modeling clay and left for 2 h in a humid box to recover. Proboscis extension was elicited by touching the tarsi of the forelegs with a drop of sucrose solution. Each fly was tested five times by using the same sucrose concentration. To avoid habituation, we waited at least 30 sec between tests.

Rotation of the visual field around a fly leads to syndirectional turning of the head (21). The angle of the head roll response was measured by gluing 2-d-old females to a manipulator and placing them in an arena. The stimulus was a grating of moving stripes (60° pattern wavelength, 1.2 Hz contrast frequency). The flies were videotaped, and the angle between head positions for clockwise and counterclockwise stimulation was measured.

Pattern-induced orientation of walking flies was analyzed in an illuminated arena (29-cm diameter, 64-cm height) (22). Flies with wings cut off were placed in the center of a circular arena and allowed to walk freely. The arena was divided in 12 30° segments. The segment in which flies crossed a measuring circle (19-cm diameter) was recorded. A fixation event was counted when the segment in which they crossed the measuring circle contained a vertical stripe, the position of which was changed halfway during experiments for each animal. The fixation efficiency was then calculated. The stimuli consisted of two opposing stripes of 10° or 60°, respectively.

Courtship was tested in small circular Teflon chambers (diameter 11 mm, height 2 mm) (23). Four- to five-day-old males and intact females were introduced without anesthesia by aspiration into the chambers. Courtship behavior of single fly pairs was observed under a binocular microscope for 10 min or until copulation occurred. A courtship index (CI) was calculated as the percentage of time the male spent performing courtship behavior (24). GH146/TNT were tested in white or red light (Kodak Safe-light filter no. 1; light transmission is zero between 220 and 610 nm and >10% between 610 and 900 nm), the latter condition to rule out visual contributions to courtship (23). The tests of GH146/UAS-tra were performed under white light only because of the lack of significant effects. To exclude reciprocal courtship, object flies were anesthetized on ice and decapitated (18).

Experiments, which resulted in normally distributed data, were evaluated by using ANOVA. For not normally distributed data, statistical analysis was done by using the nonparametric method of Kruskal-Wallis (25).

The expression pattern of line GH146 as shown by TNT, GFP, TAU, and lacZ reporters (5, 17), was qualitatively similar but varied somewhat in intensity (Fig. 1 A–D). To exclude the possibility that TeTxLC expression causes cell death during postembryonic development, we coexpressed TAU and TeTxLC in a GH146-dependent manner. TAU staining in these flies was qualitatively the same as TAU staining in flies lacking TeTxLC expression (Fig. 1 B and C), suggesting that neurons develop normally even if they express TeTxLC. Unlike larval chemosensory afferents labeled by another P[GAL4] line (8), none of the neurons expressing TeTxLC in GH146 had swollen terminal arbors.

The most intensely stained brain elements of line GH146 are about 100 PNs of the AL (16). Roughly 90 of them establish dendritic arborizations in one of the 43 AL glomeruli (Fig. 1 A–E) (26) and send their axon via the antennocerebral tract to the MB calyx and the LPR (Fig. 1 A–D, F). The remaining 5–10 labeled PNs are mostly of the polyglomerular type, whose fibers bypass the calyx and extend directly to the LPR (27). Three lines of evidence suggest that the labeled PNs represent the majority of this cell type in Drosophila: (i) the presence of stained arbors in most of the glomeruli, (ii) the tight labeling of the antennocerebral tract, and (iii) an extrapolation from the known numbers of PNs per glomerulus in other insects (16). Only few additional elements are labeled by GH146 in the adult: (i) two to three interneurons with cell bodies near the LPR (Fig. 1 A and D) and fibers toward the antennocerebral tract, (ii) a small set of MB interneurons (Fig. 1 A, E, and H), (iii) a subset of lamina monopolar cells projecting to three medulla layers—most likely L1 and L2 (Fig. 1 A, B, F, and G)—and of neurons arborizing in one layer of the proximal medulla and lobula, as well as in specific lobula plate layers, (iv) a few bilateral neurons in the posterior brain (Fig. 1 B and F), and (v) a single descending interneuron with arborizations in the leg neuromeres (not shown). Expression occurs also in glial cells along all major peripheral nerves (Fig. 1 A and I), as shown by the binding pattern of the glial-specific Repo antibody (28) (data not shown). We note the complete absence of label in olfactory and gustatory receptor neurons.

To assess an olfactory function of PNs, we used the T-maze assay, which allowed us to test the behavioral responses of GH146 flies expressing TeTxLC to different odors. Flies were given the choice between different test substances (ethyl acetate, 1-octen-3-ol, 2-hexanol, trans-hexenal) and paraffin oil as a control (Fig. 2). All four chemicals were previously shown to elicit behavioral and/or electrophysiological responses (29–31). For all tested odorants, flies expressing TeTxLC (GH146/TNT) were impaired in odorant detection when tested with lower (10⁻³ or 10⁻⁴) concentrations, as their response indices were significantly different from controls. The same effect was observed at higher concentration (10⁻²) for 2-hexanol but not the other odorants. We conclude that GH146/TNT flies are unable to correctly process odor information at lower concentrations and thus show impaired experience-independent olfactory behavior.

Studies have postulated the presence of projections from the gustatory centers of the suboesophageal ganglion via the antennocerebral tract to the MB calyx (14, 32). Although line GH146 shows no reporter gene expression in such a pathway (see above), we investigated the ability of GH146/TNT flies to respond to gustatory signals by using the proboscis extension assay. The responses to different sucrose concentrations are shown in Fig. 3. TeTxLC expression clearly had no negative effect on sucrose perception. GH146/TNT flies achieved the highest mean response (3.95) when compared with wild-type CS (3.10) and the parental lines GH146 (2.85) and TNT (3.60). We conclude that the neuronal circuit labeled in line GH146 is not necessary for this type of gustatory response.

To study the function of TeTxLC-expressing neurons in the visual system (see above), we investigated visually induced head roll and landmark fixation in walking. GH146/TNT flies showed no head roll response to a grating of moving stripes (60° pattern wavelength, 1.2 Hz contrast frequency) (Fig. 4A). In addition optomotor responses in walking were absent for two different stimuli (24° pattern wavelength, 3 Hz contrast frequency, and 360° pattern wavelength, 0.2 Hz contrast frequency), and GH146/TNT flies show no visually induced landing response (data not shown). Regarding landmark fixation, GH146/TNT flies were tested for their ability to detect stationary objects of different width. When 10° stripes were presented, GH146/TNT flies walked randomly, whereas CS, GH146, and TNT controls showed normal fixation behavior (Fig. 4B). However, when the stripe width was extended to 60°, GH146/TNT flies walked toward the stripes as often as CS, TNT, and GH146 controls (Fig. 4C). We conclude that TeTxLC expression in GH146 leads to a complete abolishment of visual behaviors that require motion detection, whereas landmark fixation is only partially impaired.

Male courtship is triggered by various sex-specific sensory cues, including visual and chemical stimuli (24, 33, 34). Some of the neural elements responsible for this complex behavior have been described (18, 35, 36). We addressed the question whether GH146/TNT males exhibit altered courtship performance. Courtship tests under white light showed a strongly reduced CI for GH146/TNT males (data not shown). As this effect could be because of visual and/or olfactory impairments, we tested courtship behavior also under red light, to avoid visual input (Fig. 5A) (23). To exclude reciprocal courtship, we additionally used decapitated females (18). Under these conditions, males should largely rely on chemosensory cues. Decapitated and intact females elicited very reduced CIs from GH146/TNT males compared with controls. The low courtship levels were because of a general “disinterest” in the majority of these males, rather than to courtship breakoff after normal onset. However, locomotor activity of GH146/TNT males, measured as the crossing frequency of a center line in the courtship chamber, was not significantly different from wild-type CS (ANOVA, P = 0.28, data not shown). We conclude that GH146/TNT males show impaired experience-independent courtship.

To understand whether the circuit labeled in line GH146 includes sexually dimorphic neural elements, we feminized the cells in male flies by using the UAS-tra reporter (18). These chromosomally male flies courted decapitated female targets normally and were not attracted by decapitated male flies (Fig. 5B). Hence, we find no evidence of sexual dimorphism.