The neural substrate of spectral preference in Drosophila

Abstract

Drosophila vision is mediated by inputs from three types of photoreceptor neurons; R1-R6 mediate achromatic motion detection, while R7 and R8 constitute two chromatic channels. Neural circuits for processing chromatic information are not known. Here, we identified the first-order interneurons downstream of the chromatic channels. Serial EM revealed that small-field projection neurons Tm5 and Tm9 receive direct synaptic input from R7 and R8, respectively, and indirect input from R1-R6, qualifying them to function as color-opponent neurons. Wide-field Dm8 amacrine neurons receive input from 13-16 UV-sensing R7s and provide output to projection neurons. Using a combinatorial expression system to manipulate activity in different neuron subtypes, we determined that Dm8 neurons are necessary and sufficient for flies to exhibit phototaxis toward ultraviolet instead of green light. We propose that Dm8 sacrifices spatial resolution for sensitivity by relaying signals from multiple R7s to projection neurons, which then provide output to higher visual centers.

Figure 1. The histamine chloride channel Ort is expressed in subsets of lamina and medulla neurons

(A) A schematic illustration of the Drosophila visual system, including the eye (Eye) and four optic neuropils: lamina (La), medulla (Me), lobula (Lo), and lobula plate (Lp). The outer photoreceptors, R1–R6 (pink), terminate in the lamina and synapse with lamina neurons (LN: green). The central photoreceptors, R7 (red) and R8 (purple), project axons to the medulla strata M6 and M3, respectively. Three selected types of medulla neurons are shown: transmedulla (Tm) neurons arborize in various medulla strata and project axons to distinct lobula strata; distal medulla (Dm) amacrine neurons extend processes in distal medulla strata; T and C (T/C) neurons extend axons into the medulla and lobula (T2 neurons) or lamina (C2 neurons, not shown). Medulla and lobula strata marked by anti-FasIII antibody are colored cyan. (B) The ort promoter driver labels subsets of medulla neurons. The ort^C1-3-LexA::VP16 driver was used to drive the expression of rCD2::GFP, a membrane-tethered GFP marker (green) in the lamina and medulla neurons that are postsynaptic to photoreceptors. Photoreceptor axons were visualized using MAb24B10 antibody (red). Anti-FasIII antibody (blue), which labels distinct medulla and lobula strata, was used as a stratumspecific landmark. (C,D) High magnification views of (B) showing the medulla (C) and lobula neuropil (D). (D) The GFP-labelled transmedulla neurons project axons to strata Lo1, Lo2, and Lo5 of the lobula (Lo1, 2 and 5), forming a topographic map. (E) Promoter analysis of the ort gene. The ort genomic structure shown as a linear cartoon with boxes representing exons and lines representing introns and intergenic sequences. Comparative genomic analysis identifies four blocks of sequences, C1–C4 (red, shown above the genomic structure), that are highly conserved among twelve species of Drosophila (see Figure S2A). The ort promotor (C1–C3) with the ort or hs70 3’ UTR region (purple and grey, respectively) was fused to either the yeast transcription factor Gal4 (dark grey box) or the chimeric transcription factor LexA::VP16 (blue box) to generate various ort promoter drivers (as indicated). Orange box: coding region; cyan box: 5’-UTR; purple box: 3’-UTR. Scale bar: 20 µm in (B); 10 µm in (C, D).

Figure 2. Ort-expressing neurons mediate phototaxis and a normal preference for UV

(A–D) Wild-type (A), ort (B) and various mutant flies were tested for fast phototaxis towards UV or green light. The intensity-response curves were measured by recording the percentage of flies choosing UV or green light of various intensities over dark. Light intensity was shown as a logarithmic scale and error bars (standard deviations) represent the variations among trials. (A) Wild-type flies exhibited phototactic responses to UV in a simple intensity-dependent fashion, resulting in a sigmoidal intensity-response curve. In contrast, phototactic response towards green light was not monotonous because the response was reduced at high intensities of green light. Compared with dark adaptation (dotted lines), light adaptation (solid lines) decreased sensitivity to both UV and green light by approximately two orders of magnitude. (B) Compared with wild-type, ort mutants exhibited a significantly reduced phototactic response under dark- or light-adapted conditions. In ort mutants, phototaxis towards UV appeared to be affected more severely than that towards green light. (C–D) Histograms of light sensitivity of wild-type and various mutant flies under dark-(C) or light-adapted (D) conditions. Light sensitivity, defined as the negative logarithm of the minimal light intensity required to attract 75% of the test flies, was calculated from the intensity-response curves. Note that fast phototaxis towards UV or green light was driven primarily by the broad-spectrum R1–R6 photoreceptors since this behavior was significantly affected by the inactivation of R1–R6 (NinaE mutants), but not R7 cells (Rh3,4->shi^ts). (E–H) Wild-type, ort, HisCl1, and various mutant flies tested for phototactic preference to UV over green light. The intensity-response curves were measured by varying UV intensity while keeping the green light intensity constant (see Experimental Procedures for details). (E–G) The P.I. for each genotype was calculated from the numbers of flies choosing UV (N_UV) or green (N_G) light by the following formula: P.I. = [N_UV − N_G] / [N_UV + N_G]. The UV/green intensity ratio (E–G) is shown as a logarithmic scale. Error bars (standard deviations) represent the variations among trials. (E) ort mutants had a reduced preference for UV. Wild-type (wt) flies exhibited phototactic preference to UV in an intensity-dependent fashion, resulting in a sigmoidal intensity-response curve. For ort mutants (ort^1/1 and ort^1/US2515), the intensity-response curve was shifted to the right. Note that normal UV preference requires R7s but not R1–R6 as sevenless (sevE2) mutants, but not NinaE mutants, exhibited low UV preference. HisCl1 ort double-null mutants, like norpA³⁶, a phototransduction mutant, chose UV and green light indiscriminately over a broad range of UV/green intensity ratios. (F) The expression of Ort driven by ort^C1-4-Gal4 restored normal UV preference in ort mutants. In contrast, UAS-ort alone or reinstating Ort function in the achromatic channels L1 and L2 failed to restore UV preference in ort mutants. Positive control (wt) and negative control (sev) were from those described in (E). (G) Ort-expressing neurons are required for normal UV preference. Shi^ts1 expressed in Ort-expressing neurons or R7s blocks their synaptic transmission. At a restrictive temperature (33°C), ort^C1-4->shi^ts1 flies exhibited lower UV preference compared with wild-type controls. Inactivating R7s using Rh3,4->shi^ts1 resulted in an even greater reduction in UV preference. In contrast, inactivating L1 and L2 using L1L2-> shi^ts1 did not affect UV preference. Wild-type control (wt) at 22°C was from that described in (E). (H) Histogram of the relative attractiveness of UV over green light (Attr_UV/G) for each genotype. Attr_UV/G was calculated from the UV/green intensity ratio at which flies exhibited phototaxis to UV and green lights with equal frequency (isoluminance point, P.I.=0), based on the following formula: Attr_UV/G = −log (UV/green ratio at the isoluminance point). The difference between the Attr_UV/G of the wild-type and ort mutants (or ort^C1-4->shi^ts1 flies) was statistically significant (*p<0.00001), whereas the difference between the wild-type and rescued ort mutants (or HisCl1 mutants) was not (p>0.1).

Figure 3. Ort is expressed in subsets of transmedulla neurons

Axonal and dendritic projections (green) of single Ort-expressing transmedulla neurons were examined in flies carrying ort^1–3-Gal4, hs-Flp and UAS>CD2,y+>CD8-GFP transgenes. R7 and R8 photoreceptor axons, visualized with MAb24B10 antibody (red), served as landmarks for medulla columns. Medulla and lobula strata were identified using R7 and R8 terminals and anti-FasIII immunolabeling (cyan, see Figure S5). Four Tm types, including Tm5, Tm2, Tm20, Tm9, were identified based on their dendritic morphologies (A’–F’) and stratum-specific axon terminations (A”– F”). Tm5 was further categorized into three subtypes: Tm5a, b, and c (A–C; see text for details). (A-A’’) The Tm5a neuron extends a single main dendritic branch (arrowhead), which runs along the photoreceptor terminals and extends multiple fine processes in strata M3 and M6 (A’). Its axon terminal in the Lo5 stratum is hook-shaped (A”). (B–B”) The Tm5b neuron extends 2 or 3 main dendritic branches (arrowheads) with fine processes spanning ~5 columns in strata M3, M6, and also M8, (B’); its axon terminates in stratum Lo5 (B”). (C–C”) The Tm5c neuron extends a single main dendritic branch with multiple fine processes, which span multiple columns in strata M3 and M6. The most distinguishable features of Tm5c are the dendritic arbors in the superficial part of the M1 stratum (arrowhead, C’) and the presence of axon terminals (arrowhead, C”) in both strata Lo4 and Lo5. (D–F”) Tm2 (D–D”), Tm20 (E–E”), and Tm9 (F–F”) form type-specific dendritic arbors largely confined to a single medulla column, and project their axons to specific lobula strata. (A’–F’, A”–F”) High magnification views of (A–F) in the medulla (A’–F’) and lobula (A”–F”), respectively. Scale bars: in (A), 20 µm (for A–F); in (A’), 5 µm (for A’–F’); in (A”), 5 µm (for A”-F”). (G) Schematic diagram illustrating the dendritic and axonal morphologies of Tm neurons. All are shown in dorso-ventral view (as in A–F”) except Tm2, which is in approximately medio-lateral view.

Figure 4. Ort-expressing Tm neurons receive multi-channel inputs in the medulla and are presynaptic at both the medulla and lobula

(A–F) The distribution of presynaptic terminals of single Ort-expressing Tm neurons was examined in flies carrying ort^1–3-Gal4, hs-Flp, TubP->Gal80>, UAS-mCD8GFP (green) and UAS-synaptotagmin-HA (red). Localization of the presynaptic reporter, Synaptotagmin-HA, was visualized using anti-HA antibody. R7 and R8 photoreceptors were visualized using MAb24B10 antibody (blue). Tm cell types are as indicated. IsoSurface representations of medulla arborization (A’–F’) and lobula terminals (A-F”) were generated using Imaris software. Synaptotagmin-HA was localized to the tips of the axon terminals and dendritic arbors, the latter especially in the proximal medulla strata (M7 for Tm5c; M8 for Tm5a, Tm5b, and Tm20; M9 for Tm2). (G) Profiles of R7, R8, L3, Tm5 and Tm9 reconstructed in three dimensions from a single medulla column. The white square box indicates the contact site between L3 and both Tm5 and Tm9 shown in (J). Although the partially reconstructed profile resembles Tm5a, the subtype reconstructed is still not certain. (H–J) Synaptic contacts between R7 and Tm5 (H), R8 and Tm9 (I), and L3 and both Tm9 and Tm5 (J) Arrowheads point to T-bar ribbons in presynaptic elements, in the presumed direction of transmission. Scale bar: in (A), 5 µm (for A–F); in (H), 500nm (for H–J)

Figure 5. Glutamatergic and cholinergic Ort-expressing neurons confer UV and green preference, respectively

(A,B) The combinatorial drivers, ort^C1-3∩vGlut (A) and cha∩ort^C1-3 (B) are expressed in distinct neuron subsets in the adult optic lobe. The ort^C1-3∩vGlut and cha∩ort^C1-3 drivers express the EGFP marker (green) in those Ort-expressing neurons with either a glutamatergic or cholinergic phenotype. The ort^C1-3∩vGlut driver labeled L1, Tm5c, and Dm8 neurons while the cha∩ort^C1-3 driver was expressed in L2, C2, Tm2, Tm9 and Tm20. Lobula plate neurons (arrowhead, A), which do not normally express Ort, were also labeled by the combinatorial drivers. Photoreceptor axons visualized with MAb24B10 antibody (red); specific neuropil strata marked with FasIII antibody (cyan). Scale bar: 20 µm in (A) for (A–B). (C–D) Sufficiency of glutamatergic or cholinergic Ort-expressing neurons for UV and green light preference. Restoring Ort function in phenotypically glutamatergic Ort-expressing neurons (ort^C1-3∩vGlut->Ort) rescued the UV phototatic defects in ort mutant flies, while restoring phenotypically cholinergic Ort-expressing neurons (cha∩ort^C1-3->Ort) rendered a stronger green preference. (C) Intensity-response curves for UV/green spectral preference were measured as described in Figure 2. ort^1/US2515, wild-type and negative control sev are from those described in Figure 2E. (D) Histogram of the relative attractiveness of UV over green light (Attr_UV/G) calculated from (C). The differences between ort mutants and those rescued with ort^C1-3∩vGlut->Ort (and cha∩ort^C1-3->Ort) are highly significant (*p<0.00001). Error bars indicate standard deviations. ort^1/US2515, wild-type and negative control sev data are from those reported in Figure 2E.

Figure 6. Wide-field Dm8 amacrine neurons are required for UV preference

(A) Various ort promoter constructs containing different multispecies-conserved regions (C1–C3). The ort locus structure is as described in Figure 1E. (B–D) Expression patterns of ort^C2-GAL4 (B), ort^C3-GAL4 (C) and ort^C2∩vGlut (D) drivers in adult optic lobes. These drivers were used to express the mCD8-GFP marker in different subsets of Ort-expressing neurons (see text for details). A few neurons in the lobula and lobula plate (arrowheads), which do not normally express Ort, were labeled by ort^C2-GAL4 and ort^C3-GAL4, respectively (B, C). (D) The combinatorial driver ort^C2∩vGlut labeled Dm8 as well as sparse L1 cells (arrowheads in lamina cortex). Photoreceptor axons visualized with MAb24B10 antibody (red); specific medulla and lobula strata immunolabeled with anti-FasIII (cyan). Scale bar: 20 µm in (B), for (B–D). (E–H) Restoring or blocking different subsets of Ort-expressing neurons affects UV/green preference. Intensity-response curves were measured as described in Figure 2. (E) Sufficiency of ort^C2 and ort^C3 neurons for UV/green preference. Restoring the function to the ort^C2 neuron subset rescued UV preference in ort mutants while restoring it to the ort^C3 subset rendered a stronger green preference. ort^1/US2515, wild-type and negative control sev data are from Figure 2E. (F) Requirement for ort^C2 and ort^C3 neurons for UV/green preference. Blocking the ort^C2, but not ort^C3, neuron subset in wild-type background reduced UV preference. ort^1/US2515, wild-type and negative control sev data are from Figure 2E. (G) Requirement for, and sufficiency of, the Dm8 neurons for UV/green preference. Restoring Ort expression in the Dm8 neurons using the ort^C2∩vGlut driver rescued UV preference defects in ort or HisCl1 ort double-null mutants. Conversely, inactivating the Dm8 neurons caused a significant reduction in UV preference. ort^1/US2515, wild-type and negative control sev data are from Figure 2E. (H) Histogram of the relative attractiveness of UV over green light (Attr_UV/G) calculated from (E–G). The differences between ort mutants and after ort function is rescued in ort^C2->Ort (or ort^C3->Ort or ort^C2∩vGlut->Ort) are statistically significant (*p<0.00001), as are those between the wild-type and ort^C2->shi^ts1 (or ort^C2∩vGlut->shi^ts1). Error bars indicate standard deviations.

Figure 7. Amacrine Dm8 neurons receive direct synaptic input from multiple R7 neurons

(A–A’) Single Dm8 neuron clones were generated using ort^C2-Gal4, hs-Flp, and UAS->CD2>mCD8GFP and visualized with anti-GFP antibody (green). Photoreceptor axons were visualized with MAb24B10 antibody (red). Dm8 neurons extend large processes in medulla stratum M6 (arrow, inset) where they are postsynaptic to 13–16 R7s (A’) and presynaptic to Tm5s. In addition, each Dm8 extends small centrifugal processes to stratum M4 where they are presynaptic to Tm9 (double arrows, A, B). Demonstration of synaptic relations is shown from EM in later panels. (Inset) A low magnification view of (A). The arrowhead and arrow indicate the Dm8 neuron shown in (A). (A’) Isosurface representation of processes of a single Dm8 neuron in a proximo-distal view. (B) Distribution of presynaptic sites of a single Dm8 neuron. Presynaptic reporter synaptotagmin-HA (red) was localized to the Dm8 processes in strata M6 and M4 (double arrows). (C) Dm8 is postsynaptic to R7s. A single EM section from stratum M6 shows Dm8 processes marked by an EM marker HRP-CD2 and stained with DAB. R7 terminal was identified based on its vesicle-laden ultrastructure, the presence of capitate projections (not shown) and its location in stratum M6. Presynaptic T-bar ribbon (arrowhead) in R7 profile is juxtaposed to postsynaptic elements of Dm8 with electron-dense membranes. (D) Serial-EM reconstruction of processes of three Dm8 neurons (pink, yellow and blue) and corresponding R7 terminals (orange). The processes of Dm8 neurons tile stratum M6 with partial overlapping so that each R7 is presynaptic to one or two Dm8 cells. (E–F) Profiles of R7 (orange), Dm8 (pink), Tm5 (green) and Tm9 (beige) reconstructed from a single medulla column. (G,H) Single EM sections show that Dm8 is presynaptic to Tm5 (G) and Tm9 (H). Presynaptic T-bar ribbons in Dm8, indicated by arrowheads, point in the presumed direction of transmission. Scale bar: 5 µm in (A, B); 500 nm in (C); 200 nm in (G) for (G,H)

Figure 8. Medulla circuits in chromatic information processing

Summary diagram of synaptic connections between photoreceptor neurons and their first-order interneurons. Small-field projection neurons, Tm5 and Tm9, receive inputs from the chromatic channels, R7 and R8, respectively, as well as from the achromatic channel L3. Wide-field amacrine neuron Dm8 receives input from multiple R7s and is presynaptic to Tm5 and Tm9. In addition, R7 receives direct input from R8.