Antagonistic roles of Wnt5 and the Drl receptor in patterning the Drosophila antennal lobe

Transcription

1 Antagonistic roles of Wnt5 and the Drl receptor in patterning the Drosophila antennal lobe Ying Yao 1,5, Yuping Wu 2,5, Chong Yin 2,5, Rie Ozawa 1, Toshiro Aigaki 3, Rene R Wouda 4, Jasprina N Noordermeer 4, Lee G Fradkin 4 & Huey Hing 1,2 Numerous studies have shown that ingrowing olfactory axons exert powerful inductive influences on olfactory map development. From an overexpression screen, we have identified as a potent organizer of the olfactory map in Drosophila melanogaster. Loss of resulted in severe derangement of the glomerular pattern, whereas overexpression of resulted in the formation of ectopic midline glomeruli. Cell type specific cdna rescue and mosaic experiments showed that functions in olfactory neurons. Mutation of the derailed () gene, encoding a receptor for Wnt5, resulted in derangement of the glomerular map, ectopic midline glomeruli and the accumulation of Wnt5 at the midline. We show here that functions in glial cells, where it acts upstream of to modulate its function in glomerular patterning. Our findings establish as an anterograde signal that is expressed by olfactory axons and demonstrate a previously unappreciated, yet powerful, role for glia in patterning the Drosophila olfactory map. The perception of odors requires that stimulus information be systematically organized in a map in the olfactory bulb. Indeed, axons of olfactory receptor neurons (ORNs) expressing a given odorant receptor sort out from other axons and terminate specifically in one of a field of discrete synaptic structures, termed glomeruli, in the olfactory bulb 1. Previous studies have shown that the pattern of glomeruli, or olfactory map, is a direct result of precise axon pathfinding and synaptogenesis 2,3. Recent work in several species has identified a number of molecules that are necessary for the proper development of the olfactory map. These molecules include transmembrane proteins such as Ncam-18, Neuropilin-1, Plexin A, Robo, the odorant receptors, Dscam and Cadherin, as well as cytoplasmic signaling molecules such as Src, Fyn, Dock and Pak An attempt to integrate these diverse mechanisms has led to a hierarchical model of ORN axon targeting, wherein ORN axons are sequentially guided toward their postsynaptic targets with an increasing degree of precision 5,8. A wealth of evidence shows that ingrowing ORN axons are important for directing the development of the glomeruli in Drosophila and other species. First, genetic or surgical disruption of the ORN axons blocks the formation of the glomeruli, indicating that ORN axons are necessary for glomerular development 9,12,13. Second, when the LIM kinase 1 (Limk1) gene is overexpressed in the ORNs, or when the ORNs are transplanted to ectopic sites, the axons direct the formation of glomeruli in ectopic positions, indicating that the presence of ORN axons is also sufficient for glomerular development Ithas been suggested that ORN axons have an intrinsic ability to specify glomerular development in their target tissues 17, but the underlying mechanisms for this are unknown. The Drosophila olfactory system is highly suited for use in unraveling the mechanisms of olfactory map development. The anatomy and development of the fly antennal lobe closely resembles that of the olfactory bulb of mammals, but is vastly simpler, containing only 43 glomeruli compared with B2, in the mouse. We conducted an overexpression screen for genes that disrupted the stereotyped anatomy of the Drosophila antennal lobe. We identified as a candidate regulator of antennal lobe development. The gene encodes a member of the large, evolutionarily conserved Wnt family of secreted proteins, which have well-established roles in embryonic patterning, cell proliferation and cell differentiation 18,19. Recent studies have shown that Wnt5 functions as a repulsive cue that routes axons through the correct commissures of the embryonic ventral midline and regulate their fasciculation 21, and is required for stabilizing a subset of axons in the adult brain 22.Weshowherethat functions in ORNs to regulate the patterning of the olfactory glomeruli. In the absence of, ORN axons do not target properly, and the glomerular map is disorganized. We also provide evidence that Drl, a Ryk-family receptor tyrosine kinase previously demonstrated to act as a repulsive neuronal Wnt5 receptor in the embryonic CNS, functions in glia to inhibit Wnt5 signaling via its extracellular Wnt inhibitory factor (WIF) domain. Our results demonstrate that Wnt5 is an anterograde signal by which ORN axons specify the organization of the olfactory map and that Drl, presented by glial cells, is a powerful modulator of this signal. 1 Neuroscience Program and 2 Cell and Developmental Biology, University of Illinois at Urbana-Champaign, 61 South Goodwin Avenue, Urbana, Illinois 6181, USA. 3 Biological Sciences, Tokyo Metropolitan University, 1-1 Minami-osawa, Hachioji-shi, Tokyo , Japan. 4 Laboratory of Developmental Neurobiology, Department of Molecular Cell Biology, Leiden University Medical Center, Einthovenweg, 23 RC Leiden, The Netherlands. 5 These authors contributed equally to this work. Correspondence should be addressed to H.H. (hing@life.uiuc.edu). Received 1 July; accepted 17 September; published online 14 October 7; doi:1.138/nn1993 NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION 1

2 Distance (µm) ARTICLES RESULTS ORNs overexpressing disrupt antennal lobe structure From a screen of 3,996 P{GS} Drosophila lines 23, we found that the overexpression of (P{GS1}1192) under the control of the SG18.1- Gal4 driver, which is preferentially expressed in the ORNs 9,24, profoundly altered the anatomy of the antennal lobe (Fig. 1). In the wild-type adult expressing GFP under the control of SG18.1-Gal4, the antennal lobes were spherical (B7 mm diameters) and were connected by a commissure B mm thick(fig. 1a). In the -overexpressing animals, the antennal lobes were malformed and glomerulus-like structures appeared in the commissure in 79% (42/53) of the antennal lobes, substantially enlarging the structure to B6 mm in thickness (Fig. 1b). We observed that a large percentage of the Or22a, Or and Or47b glomeruli in the -overexpressing animals were split into smaller subunits, many of which were displaced into the midline (Supplementary Fig. 1 online). The termination of ORN axons in ectopic positions indicates that ORN axons were misguided in the -overexpressing animals. Examination of the projection neurons showed that their dendritic arbors were disrupted, and a subset of them invaded the midline, where they innervated the ectopic glomeruli o a SG18.1 > P{GS1} 1192 d g k Or Or From glomerulus to the top of 71a 67b 59c Between two glomeruli 42a 47a-47b 67d b γ-glutamyltranserase gene P{GS1}124, 139 and 1192 insertion e f Or Or47a & Or47b D l p M h i j Percentage (%) Or71a Or71a Splitting in Crossing defect in 71a 67b 59c 42a 7K 47b 67d c m q Or in Or in 7K 7K Or47a in Or47a in n r gene deletion line Or47b in Or47b in Or47a & Or47b Or47a & Or47b Figure 1 The olfactory map is disrupted in the mutant. (a r) Preparations of wild-type () flies and mutants were stained with nc82 (magenta) and antibody to GFP (green). Antennal lobes of wild-type (a), -overexpressing (P{GS1}1192) (b) and 4 (d) animals were visualized using GFP driven by SG18.1-Gal4. Commissures are indicated with arrows. (c) Schematic diagram of the locus showing the P{GS} lines recovered from our screen and the 4 deletion breakpoints. (e) Superimposed outlines of antennal lobes from wild-type flies and the 4 mutants (n ¼ 12). The 4 mutant antennal lobes were flattened dorsomedially (D, dorsal; M, medial), producing a characteristic heart-shape appearance. (f) Plots of the positions of Or, Or47a and Or47b glomeruli in the wild type and 4 mutant. Preparations of wild-type (g j) and 4 mutant (k n) antennal lobes in which specific glomeruli are highlighted with GFP. In the 4 mutant, the glomeruli were closer to the antennal lobe dorsal edge or were split into smaller structures (arrowheads), compared with wild type. ORN axons frequently took meandering paths (arrow) to their targets and failed to cross the midline (double arrowheads). (o) Histograms showing the distances between various glomeruli and the antennal lobe dorsal edge or between two glomeruli in the same antennal lobes (s). Values are mean ± s.e.m. (p) Quantification of the glomerular-splitting and midline-crossing defects in the 4 mutant. Single confocal sections showing the projection neuron dendritic arrangement in the wild type (q) and 4 mutant (r), labeled by GH164-Gal4 driving UAS-mGFP. For each group, n ¼ Scale bar, mm. 2 ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

3 a b e c 26 h 5 h 18 h 18 h d (Supplementary Fig. 1). Thus, overexpression in the ORNs disrupted the stereotyped structure of the antennal lobe and elicited the formation of ectopic glomeruli at the midline. is necessary for antennal lobe structure To determine whether normally functions in the antennal lobes, we examined the antennal lobes of the 4 null mutant (Fig. 1c), which is homozygous viable 21.Inthe 4 mutant expressing GFP under the control of SG18.1-Gal4, many antennal lobes (37/55 antennal lobes, B67%) were not connected by commissures (Fig. 1d,p) and appeared misshapen; being flattened dorsally and elongated ventrally, producing a characteristic heart shape (Fig. 1d,e,k n). We investigated the glomerular arrangement of eight ORN subclasses. In the wild type, these glomeruli are located at stereotyped positions in the antennal lobe 25,26 (Fig. 1g j). In the 4 mutant, the dorsal edge of the antennal lobe appeared to be closer to many of these glomeruli (Fig. 1f,k n,o). To test the possibility that the antennal lobes were distorted, we measured the distance between glomeruli in the same antennal lobe (Or47a to Or47b, and Or67d; Fig. 1j,n,o). In the wild type, the distance between the Or47a and Or47b glomeruli was ± 1.6 mm (mean ± s.e.m., n ¼ 32). In the 4 mutant, this distance decreased to 8.7 ± 2.73 mm (n ¼ 32, P ¼.1 compared with wild type). Thus, the 4 antennal lobes showed a characteristic collapse of the dorsomedial region of the antennal lobe (Fig. 1e). Examination of the ORN fibers also showed that ORN axons frequently took meandering paths toward their targets (arrows, Fig. 1k,l), looped back on the ipsilateral glomeruli (double arrowheads, Fig. 1n), stalled before the commissure, projected aberrantly to dorsal regions of the 26 h 5 h f g h i j Percentage (%) k 1 Crossing midline 1 Project dorsally 8 Absent hapf Adult 4 5 hapf Adult (n = 18) (n = 26) (n = 5) (n = 55) Figure 2 Wnt5 functions during antennal lobe development and localizes to projection neuron dendrites. (a d) Preparations of wild type (a,c) and 4 /Y hemizygote (b,d) in which GFP was driven by SG18.1-Gal were stained with nc82 (magenta) and GFP (green) antibodies. Figures show projected confocal sections. (a,c) Inthewild type, contralateral axons crossed the midline (arrow) and the antennal lobes had spherical shapes. (b,d) In the mutant, contralateral axons projected aberrantly toward dorsal regions of the brain (arrows). (e) Quantification of the commissural defects seen in the 4 mutant at 4 5 hapf or adulthood. (f h) Preparations of 18-hAPF pupal brains expressing GFP (green) under the control of the pan-olfactory driver pbl- Gal4 were stained with antibodies to Wnt5 (red) and Elav (blue). Figures show single confocal sections. Asterisks indicate three (of the four) spheroidal structures that accumulated Wnt5 and are surrounded by masses of ORN axons. (i k) Preparations of 18-hAPF pupal brains expressing GFP (green) under the control of GH146-Gal4 were stained with antibodies to Wnt5 (red) and Elav (blue). Projection neuron dendrites overlapped largely with the niduses of Wnt5 staining (red). Dorsal is up and lateral is to the right in f k. Scale bars, mm ina d and 1 mm inf k. brain (arrows, Fig. 2) or terminated on ectopic glomeruli (arrowheads, Fig. 1k,l,n,p). In summary, the loss of resulted in aberrant targeting of the ORN axons and characteristic shifts of dorsomedial glomeruli leading to the antennal lobes acquiring a distinct heartshaped appearance. We also examined the projection neuron dendrites by expressing GFP under the control of the GH146-Gal4 driver. In the wild type, GH146-Gal4 labeled a subset of projection neuron dendrites, revealing an invariant pattern in the antennal lobe neuropil (Fig. 1q). In the 4 mutant, this pattern was disrupted and many dendritic arbors were displaced ventrally (Fig. 1r). Thus, mutation of the gene disrupted the targeting of the ORN axons and the positioning of the projection neuron dendrites. functions in antennal lobe development To determine when the observed defects arise in the 4 mutant, we evaluated the antennal lobe structure of the 4 mutant during the pupal stages. In the control 4 /+ heterozygotes at 26 h after puparium formation (hapf), the antennal lobe had a smooth neuropil (Fig. 2a). ORN axons had arrived at the outer surface of the antennal lobes and were projecting across the midline in a commissure that was B5 mm in width. At 5 hapf, the antennal lobe neuropil was well partitioned into glomeruli and a thick commissure connected the left and right antennal lobes (Fig. 2c). In the hemizygous 4 /Y mutant, the antennal lobes were oval shaped at 26 hapf, as in the wild type, although in many brains, axons appeared to project dorsally instead of across the midline (Fig. 2b). At 5 hapf, the antennal lobes appeared misshapen, with many showing the distinctive heart shape seen in the adult (Fig. 2d). Contralateral axons failed to decussate and instead projected dorsally in 85% of the brains examined (22/26; Fig. 2e). By the adult stage, many of the brains no longer showed dorsal projections, indicating that the wayward axons had likely retracted or degenerated (Fig. 2e). These results indicate that NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION 3

4 a b c d /Y;SG > /Y;7K > e f g i Distance between Or47b and the top of (µm) /Y; SG > /Y /Y; 7K > functions during development to regulate ORN axon projection and antennal lobe development. Wnt5 protein is localized in the developing antennal lobes To ascertain the distribution of Wnt5 protein, we stained pupal brains with an antibody to Wnt5 (ref. 21) that did not stain 4 mutant brains (Supplementary Fig. 2 online), which demonstrates its specificity. We stained pupae expressing GFP under the control of pbl-gal4, a pan olfactory neuron driver 11. We observed ORN axons penetrating into the antennal lobe as early as 18 hapf (arrows, Fig. 2f,g). At this time, Wnt5 immunolabeling was seen in four spheroidal structures (B7 mm in diameter, asterisks, Fig. 2h), each surrounded by profuse axonal terminals (Fig. 2f). This pattern of Wnt5 expression remained relatively unchanged up to 42 hapf (Supplementary Fig. 2). To identify the spheroidal structures, we stained 18-hAPF animals expressing GFP under the control of GH146-Gal4, which labels the projection neuron dendrites, with the Wnt5 antibody. Wnt5 immunoreactivity largely coincided with the projection neuron dendrites, indicating that Wnt5 was localized to projection neuron dendrites during glomerular development (Fig. 2i k). Besides the dendritic localization, we show below that Wnt5 also accumulated in the region of the antennal commissure in the 2 mutant. Wnt5 staining was greatly reduced by 7 hapf, when glomerular development was largely complete (data not shown). Our immunolocalization results indicate that Wnt5 is localized to incipient glomeruli during the period of ORN axon targeting and glomerular development. D j M /Y /Y; SG > /Y; 72OK > k and /Y Or47b in Or47b in /Y mutant Or47b in rescue of /Y mutant l /Y; SG > Distance between 7K and the top of (µm) /Y; 72OK > /Y;GH146 > is required in the ORNs To delineate the cell type in which functions, we first restored function to specific cell types in the 4 mutant background. We employed a UAS- transgene 21, which expresses at a low level and therefore does not disrupt antennal lobe development (Supplementary Fig. 1). Three criteria were specifically assessed: the shape of the antennal lobe, the position of the Or47b glomerulus and the presence or absence of the commissure. In the 4 mutant, the antennal lobe was heart shaped (compare Fig. 3a,b), the Or47b glomerulus was located ±.8 mm (n ¼ 14) from the dorsal edge of the antennal lobe compared with the wild-type glomerulus, which was located ± 1.3 mm (n ¼ ) from the dorsal edge (P o.1; Fig. 3), and the commissure was present in only 3% of the brains. When we expressed UAS- using the GH146-Gal4 driver, which is expressed in projection neurons, the mutant phenotype was not rescued (Fig. 3h). No rescue was observed when UAS- was expressed under the control of MZ317-Gal4, which is expressed in glia (data not shown). When UAS- was expressed under the control of SG18.1-Gal4, the 4 mutant phenotype was rescued to a significant degree. The rescued antennal lobes were spherical, instead of heart-shaped (Fig. 3c,f), the Or47b glomerulus was located.12 ± 1.4 mm from the dorsal edge of the antennal lobe (n ¼ 24, P ¼.12 compared with mutant; Fig. 3e,g), and the antennal lobes were connected by a commissure in 55% (11/) of the brains. Expression of UAS- under the control of the 72OK-Gal4 driver, which is expressed in a subset of ORNs 7, also significantly rescued the mutant h Percentage (%) Splitting in Splitting in Crossing defect in Crossing defect in Figure 3 functions in the ORNs. (a h) Antennal lobes from wild-type flies (a), 4 mutants (b) and 4 mutants expressing UAS- under the control of various drivers (c,d,h) were stained with antibody to GFP (green), to reveal the Or47b axons, and nc82 (magenta). Expression of UAS- with SG18.1-Gal4 (c) and72ok-gal4 (d) rescued the 4 mutant phenotype, but expression with GH146-Gal4 did not (h). (e) Histograms show the distances between the Or47b glomerulus and the top edge of the antennal lobe in the various genotypes. (f) Superimposed antennal lobe outlines from various genotypes (n ¼ 12). (g) Plots showing the Or47b glomerulus positions in the various genotypes. (i,j) Representative antennal lobes with MARCM clones of wild-type and 4 mutant 72OK neurons. In antennal lobes with 4 mutant 72OK axons, the 72OK glomerulus was split into smaller structures (arrowheads) and contralateral axons (arrows) looped back on the ipsilateral targets instead of crossing the midline. (k) A plot of the positions of the 72OK glomerulus in the wild-type and 4 mosaic animals. (l) Quantification of the distance between the 72OK glomerulus and the top edge of the antennal lobe, the splitting of glomerulus and the absence of the contralateral tract in wild-type and 4 mosaic animals. Values are mean ± s.e.m. All figures show projected confocal sections. SG ¼ SG18.1-Gal4; 72OK¼ 72OK-Gal4. Scale bar, mm. 4 ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

5 s a e f g h p q r Percentage (%) 1 -mutant phenotype b c d phenotype. The rescued antennal lobes had a wild-type shape (Fig. 3d,f), the Or47b glomerulus was located ± 1.2 mm from the dorsal edge of the antennal lobe (n ¼ 16, P ¼.1 compared with mutant; Fig. 3e,g) and the antennal lobes were connected by a commissure in 77% (1/13) of the brains. Next, we investigated whether is required in the ORNs for proper antennal lobe development. We employed the mosaic analysis with a repressible cell marker (MARCM) system 27 to induce and examine clones of wild-type or 4 homozygous 72OK cells. Large clones were examined, as small clones were likely to be rescued by Wnt5 secreted from wild-type cells adjacent to the clones. In control animals with wild-type clones, the antennal lobes had a spherical shape (Fig. 3i). The wild-type 72OK ORNs projected to their expected positions (49.14 ± 3.5 mm from dorsal surface of antennal lobe, n ¼ 14; Fig. 3k,l) and formed glomeruli with stereotyped shapes. Their contralateral axons formed a distinct fascicle that projected normally across the midline. In animals bearing 4 clones, the antennal lobes were misshapen, with a heart-shaped appearance (Fig. 3j). The mutant 72OK ORNs projected to a slightly more dorsal position than that of the control (44.16 ± 4.5 mm, n ¼ 18, P ¼.2; Fig. 3k,l) and formed distorted and split glomeruli (Fig. 3j,l). Their contralateral axons frequently appeared defasciculated and (72.2%, 13/18 brains) failed to project across the midline (Fig. 3j,l). Our cell type specific cdna rescue and mosaic experiments therefore indicated that functioned in the ORNs for appropriate antennal lobe patterning. is necessary for patterning the glomerular map The capacity of Wnt5 to regulate glomerular patterning focused our attention on its potential downstream signaling pathway. Drl has been shown to act as a receptor for the repulsive Wnt5 cue in axons that cross i j k l m n o Or Or67b Or42a Or71a 22a 59c Medial Targeting Unbalance 42a 92a 47a 33c 67b 71a 43a 47b Lateral t L D V M L D V 71a M Figure 4 The olfactory map is disrupted in the mutant. (a r) Antennal lobes of wild type (a d, m o) and 2 mutants (e l,p r) were stained with antibody to GFP (green) and nc82 (magenta). (a d) Wild-type antennal lobes expressing GFP in the Or, Or67b, Or42a and Or71a glomeruli. (e l) 2 mutant antennal lobes showing the corresponding glomeruli. Arrowheads indicate the split glomeruli, arrows indicate the wandering ORN axons and asterisks denote the ectopic midline glomeruli. Antennal lobes from wild-type flies (m) and 2 mutants (p) expressing GFP (green) under the control of SG18.1-Gal4 were stained with nc82 (magenta). Ectopic protrusions were visible at the midline of the 2 mutant (asterisk). In the wild-type adult, projection neuron dendrites, visualized with GH146-Gal4 driving GFP (green), were arranged in a stereotyped pattern (n) and never terminated at the midline (o). In the 2 mutant, the dendritic pattern was severely deranged (q) and dendrites innervated ectopic glomeruli in the midline (asterisk, r). (s) Quantification of the targeting (dark bars) or the unbalanced (light bars) defects in the various ORN subtypes. (t) Plots of the positions of the ectopic Or and Or71a glomeruli (magenta dots) in the 2 mutant. All panels, except for n, o, q and r, show projected confocal sections. Scale bars, mm in a d and 1 mm ine h. the embryonic midline. To determine whether also functions in the antennal lobe, we examined the antennal lobes of the 2 null mutant 28 (Fig. 4). In 39% (22/56) of the mutant brains, ectopic glomeruli developed at the midline (asterisks, Fig. 4a f,m,p,r), whereas abnormal protrusions extended from the dorsal-medial corner of the antennal lobes of the remainder (61%). In each of the 12 subclasses of ORNs that we analyzed, ectopic midline glomeruli were seen at low frequencies, suggesting that all subclasses have an equal tendency to terminate at the midline. Plotting the positions of the Or and Or71a glomeruli in different antennal lobes revealed no consistent pattern, indicating that the glomeruli were randomly situated in the mutant (Fig. 4t). In addition to the glomerular defects, ORN axons frequently took circuitous routes to their targets (arrows, Fig. 4g,i k). Strikingly, a number of axons terminated unilaterally (confirmed by unilateral antennal ablations, Supplementary Fig. 3 online), resulting in the loss of presynaptic structures from a single antennal lobe (Fig. 4i k). To better assess the data, we grouped the glomerular positioning and splitting defects as targeting defects (Fig. 4e h,l) and the unilateral loss (or reduction) of glomeruli as unbalanced defects (Fig. 4i k). Both types of defects were present in each of the 11 ORN subclasses (Fig. 4s). We also inspected the organization of the projection neurons in the 2 mutant using the GH146-Gal4 marker. Unlike in the wild-type adult, where GH146-Gal4-expressing cells arborized in specific glomeruli, the organization of the dendritic arbors appeared chaotic in the 2 mutant adult (Fig. 4n,q), and a subset of arbors even invaded the midline, where they targeted the ectopic midline glomeruli (Fig. 4o,r). Loss of therefore leads to disruption of the glomerular arrangement and midline invasion by projection neuron dendrites. This midline invasion of projection neuron dendrites is highly reminiscent of the overexpression phenotype (Supplementary Fig. 1). When we examined the antennal lobes of the 2 mutant at 4 hapf (Fig. 5), we NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION 5

6 a 4 h observed fine dendritic processes growing into the midline, which did not occur in the wild type, indicating that the defects arose during development (Fig. 5a,b). Drl is localized to glial cells and projection neurons To ascertain the localization of the Drl protein in the olfactory system, we stained developing brains with an antibody to Drl. We first stained animals expressing GFP under the control of Repo-Gal4, which labels b c d e 3 h f g h i j k Or Repo > wif GH146 > 3 h Or GH146 > 4 h Repo > Repo > Repo > Repo > Repo > cyto Repo > cyto Repo > cyto Repo > cyto Repo > wif Or22a a b c d e f g h i j k l m n o q t r u v Percentage (%) s Or Repo > K371A 1 -mutant phenotype p Figure 5 Drl functions during antennal lobe development and localizes to TIFR glia. (a) In the wild type at 4 hapf, projection neuron dendrites, visualized with GH146-Gal4 driving GFP (green), were confined to the antennal lobes. (b) An equivalent section of 2 mutant showed projection neuron dendrites extending toward the midline (arrows). (c g) Frontal optical sections of a wild-type 3-hAPF pupal brain expressing GFP (green) under the control of Repo-Gal4 stained with an antibody to Drl (magenta). Antennal lobes are indicated by the dotted outlines. (c e) The Drl antibody intensely stained the TIFR (brackets). (f,g) A more posterior section, at the level of the antennal commissure, shows a bundle of glial processes that connects the two antennal lobes and also expressed the Drl protein (arrows). (h) No Drl staining was found in the 2 mutant brain. (i k) Antennal lobe from a 3-hAPF wild-type animal expressing GFP (green) under the control of GH146-Gal4 stained with antibodies to Drl (red) and Elav (blue). Drl distribution colocalized largely with the projection neuron dendrites. Dorsal is up and lateral is to the right in i k. All the panels show single confocal sections. Scale bars, 1 mm. glial cells. At 3 hapf, GFP was highly expressed by the transient interhemispheric fibrous ring (TIFR) 29, a toroidal glial structure lying in the mid-sagittal plane of the brain (brackets, Fig. 5c,e). The ventral edge of the TIFR is closely associated with and bridges the antennal lobes. In more posterior sections, at the level of the antennal commissure, a thick bundle of glial processes was seen to pass from one antennal lobe to the other (Fig. 5f). The TIFR was strongly stained by the Drl antibody (Fig. 5d), including the bundle of glial processes linking the antennal lobes (Fig. 5g). We next stained brains from animals expressing GFP under the control of GH146-Gal4.At 3 hapf, GFP highlighted the projection neuron cell bodies and their dendritic arbors in the antennal lobe neuropil (Fig. 5j). Drl immunolabeling largely overlapped with the dendritic GFP staining, indicating that Drl was localized to a large subset of the projection neuron dendrites (Fig. 5i k). The Drl antibody did not stain the 2 null mutant brains (Fig. 5h). Or47b 22a 47b 22a 47b 22a 47b 22a 47b FL GH146 FL Targeting Unbalance K371A INTRA Repo WIF Figure 6 Drl functions in glial cells and its WIF domain is essential. (a u) Preparations of wildtype flies (a d) and 2 mutants (e u) expressing Or-mGFP, Or-mGFP, Or22a-mGFP and Or47b-mGFP were stained with antibody to GFP (green) and nc82 (magenta). (a d) Glomeruli had predictable shapes and were located in invariant topographic positions in the wild type. (e h) In the 2 mutant, glomeruli were ectopically positioned, split or missing. (i l) Expression of UAS- under the control of the Repo-Gal4 driver strongly rescued the mutant phenotype. Glomeruli were located in their stereotyped positions. (m p) Expression of a truncated Drl protein lacking the cytoplasmic domain using the Repo-Gal4 driver rescued the mutant phenotype. (q,r) Expression of a truncated Drl protein lacking the extracellular WIF domain failed to rescue the mutant phenotype. Glomeruli were split or missing. (s) Expression of a kinase-dead Drl protein (K371A) also rescued the mutant phenotype. (t,u) ExpressionofUAS- under the control of GH146-Gal4 partially rescued the mutant phenotype. Although the glomerular defects remained, protrusions from the antennal lobes were absent. (v) Quantification of the mutant defects found in the various genotypes. Dark bars represent targeting defects, whereas light bars represent unbalanced defects. FL, full-length Drl protein. All images are projections of confocal z-series. Scale bars, mm. n Z ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

7 Figure 7 functions upstream of to inhibit function. (a f) Preparations of various genotypes expressing GFP under the control of SG18.1-Gal4 were stained for GFP (green) and nc82 (magenta). (a) Antennal lobes of the 2 mutant. (b) Antennal lobes of the 2 mutant expressing UAS- under the control of SG18.1-Gal4. (c) Antennal lobes of the wild type expressing UAS- under the control of SG18.1-Gal4. (d f) Corresponding deep sections of a c, revealing the large ectopic glomeruli (asterisks) of b and e compared with a and d. (g) Wild-type antennal lobes with invariant shape and position of Or47b glomeruli. (h) The 4 antennal lobes had a characteristic heart shape and the Or47b axons failed to cross the midline (arrow). (i) Inthe 2 antennal lobes, the Or47b glomerulus was split and ectopic glomeruli appeared at the midline (asterisk). (j) In the ; double mutant, the antennal lobes had a heart-shaped appearance and the Or47b contralateral fibers failed to cross the midline (arrow). (k) Removal of a copy of suppressed the mutant phenotype. The antennal lobes had wild-type shapes and ectopic midline structures were absent. (l) Quantification of the percentage of antennal lobes showing the failure of Or47b axons to cross the midline. (m q) Superimpositions of the antennal lobe outlines from the various genotypes (n ¼ 6 for each genotype). Midline protrusions are indicated with an asterisk. a c and g k are projected confocal sections and d f are single confocal sections. Scale bar, mm. functions in glial cells To delineate the cell types in which functions, we sought to restore function to specific cell types in the 2 mutant background. Because Drl localized, in part, to projection neuron dendrites, we employed the GH146-Gal4 driver, which is active in two thirds of all projection neurons 3,31, to drive the expression of UAS-. We examined several criteria to assess genetic rescue (Fig. 6): the presence or absence of ectopic midline structures and the position and integrity of the Or22a, Or, Or and Or47b glomeruli (Fig. 6a d). In the 2 mutant, 1% of the brains showed aberrant midline structures (either midline glomeruli or protrusions) and 38% 94% of the brains showed defects in the positioning or integrity of the glomeruli (Fig. 6e h,v). Expression of UAS- under the control of GH146-Gal4 partially rescued the mutant phenotype (58% targeting defects; Fig. 6t,u,v). To determine whether is required in the projection neurons, we generated and examined -mutant projection neuron MARCM single-cell clones. Loss of from single projection neurons, however, did not disrupt the projection neuron dendrite morphology or the stereotyped arrangement of the glomeruli in the antennal lobes (Supplementary Fig. 4 online). To further evaluate whether functions in neurons during antennal lobe development, we expressed UAS- under the control of elav-gal4, a pan-neuronal driver, but this did not rescue the 2 mutant phenotype (data not shown). Next, we tested whether expression of in glia would rescue the 2 phenotype. Expression of UAS- under the control of the pan-glial driver Repo- Gal4 completely eliminated the midline defects (%) and strongly reduced the occurrence of targeting defects to 11 24% for different glomeruli (Fig. 6i l,v). We conclude that functions in the glia for the proper development of the antennal lobes. Drl s extracellular domain regulates glomerular patterning Our ability to rescue the mutant phenotype allowed us to probe the functions of the different domains of the Drl protein in glomerular patterning. To determine whether the Drl kinase activity is required for antennal lobe development, we mutated the conserved K371 residue, which is essential for kinase activity 32, to an alanine. Expression of the UAS- K371A transgene under the control of Repo-Gal4 strongly rescued the mutant phenotype (26% midline defects and 11% Or axon targeting defects compared with 1% and 35%, respectively, in the 2 mutant; Fig. 6s), indicating that kinase activity is not a b c / / ; SG18.1 > d e f g h essential for Drl function in antennal lobe development. To ascertain whether the cytoplasmic domain is necessary for antennal lobe development, we used the UAS- Dcyto transgene 33, which encodes a truncated Drl protein bearing only the extracellular and transmembrane domains. Notably, UAS- Dcyto retained substantial ability to rescue the mutant phenotype (33% midline defects, 23% targeting defects, Fig. 6m p,v). To determine whether the Drl WIF domain is needed for antennal lobe development, we generated a UAS- DWIF transgene, which encodes a Drl protein that only lacks the WIF domain. The loss of the WIF domain abolished the ability of to rescue the mutant phenotype (89% midline defects and 56% targeting defects; Fig. 6q,r,v). Taken together, these results indicate that Drl regulates antennal lobe development largely through its Wnt5-binding WIF domain. acts downstream of in antennal lobe patterning We next investigated the genetic relationship between and.the similarity of the midline defects seen in the loss-of-function and gain-of-function mutants suggested that the two genes might act antagonistically in antennal lobe development. To test this idea, we examined the effect of expressing a low level of in the 2 mutant background (Fig. 7). In the 2 mutant, midline glomeruli were found in 39% of the animals, with most showing only protrusions from the dorsomedial corner of the antennal lobes (Fig. 7a,d). Expression of a single copy of UAS- under the control of SG18.1-Gal4 in the wildtype background did not alter the antennal lobe structure (Fig. 7c,f and Supplementary Fig. 1). In contrast, expression of a single UAS- in the 2 mutant background resulted in the induction of large midline glomeruli in 83% of the animals (Fig. 7b,e). Thus, function is SG18.1 > /Y +/Y; / j k /Y; / /+; / m p n /Y +/Y; / q /Y; / /+; / l i Midline crossing (%) o % 44% /Y +/Y; / /Y; / /+; / NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION 7

8 Figure 8 Wnt5 is localized in the antennal lobe neuropil and midline commissure in the mutant animals. (a f) Preparations of 36-hAPF pupae of wild-type flies (a,d) and 2 mutants (b,c,e,f) expressing UAS-GFP under the control of Repo-Gal4 were stained for Wnt5 (red), GFP (green) and Elav (blue). (c,f) 2 mutant coexpressing UAS- under the control of Repo-Gal4. (a c) Anterior optical sections showing Wnt5 staining in the antennal lobe neuropil. (d f) Posterior sections showing Wnt5 staining in the commissure (dashed lines). The TIFR is indicated by a bracket in a. (g) Quantification of the levels of Wnt5 in the antennal lobe neuropils and commissures of the various genotypes. (h k) Preparations of wild-type (h,i) and 2 mutant (j,k) pupae expressing UAS-GFP under the control of GH146-Gal4 were stained for Wnt5 (red), GFP (green) and Elav (blue). Micrographs were taken at equivalent section planes (at the level of the commissure) and the antennal lobes are outlined with dashed lines. (j) At 3 hapf in the 2 mutant, Wnt5 was located medial to the antennal lobes (arrowheads), whereas Wnt5 remained in the confines of the antennal lobes in the wild type (h). (k) At 36 hapf in the 2 mutant, Wnt5 accumulation increased the antennal commissure (arrowhead). Projection neuron dendrites (arrows) began to transgress the antennal lobe boundaries and projected toward the region of Wnt5 accumulation, which was not seen in the wild type (i). All figures are of single confocal optical sections. Scale bar, 1 mm. strongly increased in the 2 mutant background, which is consistent with the notion that antagonizes during glomerular development and antennal lobe patterning. To determine the hierarchical order of and in the genetic pathway regulating antennal lobe development, we constructed animals bearing null mutations in both genes and expressing GFP in the Or47b axons. Heterozygosity for 2 had no effect on the homozygous 4 (/Y; /+) mutant phenotype (data not shown). In contrast, heterozygosity for 4 markedly suppressed both the dorsomedial antennal lobe protrusions and the glomerular defects that were typical of the 2 mutant (/+; /; Fig. 7i,k,o,q). Homozygosity for both the 4 and 2 null mutations resulted in antennal lobes having the characteristic mutant phenotype (/Y; /; Fig. 7g j). The dorsal-medial corners of the antennal lobes were collapsed, producing a heart-shaped appearance (Fig. 7m p), and the Or47b contralateral axons failed to cross the midline in 56% of the animals (compared with % in 2 mutants and 62% in 4 mutants; Fig. 7j,l). 4 is thus epistatic to 2, suggesting that functions downstream of and is repressed by in the signaling pathway that regulates antennal lobe development. Wnt5 protein accumulates at the midline in the mutant The antagonistic relationship between and led us to ask how they might function together to regulate antennal lobe development. The strong midline defects of the and mutants and the localization of Drl to the TIFR suggest a role for these proteins at the midline (Fig. 8). As mentioned above, we observed Wnt5 immunoreactivity in the wild-type antennal lobe neuropil at 36 hapf (1 ± 5.5%, n ¼ 8; Fig. 8a,g), but no detectable staining in the antennal commissure (47. ± 1.7%, n ¼ 8; Fig. 8d,g). We then stained for the Wnt5 protein in the 2 loss-of-function mutant. Of note, Wnt5 was found in the antennal lobe neuropil (96. ± 8.9%, n ¼ 16, P ¼.5165 compared with wild-type neuropil; Fig. 8b,g) and the antennal Superficial Commissure a b c d e f h j 36 h 36 h 36 h 3 h 36 h i k ;Repo > commissure, a structure enwrapped by processes of the TIFR glia (75.9 ± 8.5%, n ¼ 16, P o.1 compared with wild-type commissure; dashed lines, Fig. 8e,g). This result indicates that downregulates Wnt5 protein in the antennal commissure. To determine whether the downregulation of Wnt5 at the midline is a result of acting in glia, we restored specifically to glia in the 2 mutant. In 2 mutants expressing UAS- under the control of the Repo-Gal4 driver, Wnt5 staining in the antennal commissure was no longer observed (43.7 ± 4.7%, n ¼ 16, P ¼.25 compared with wild-type commissure; Fig. 8f,g). The midline accumulation of Wnt5 in the 2 mutant prompted us to ask how it might affect glomerular development in the mutant. At 3 hapf, as Wnt5 began to accumulate in the dorsomedial corners of the antennal lobes in the 2 mutant (arrowheads in Fig. 8j, compare with Fig. 8h), projection neuron dendrites were observed lateral to the domains of Wnt5 staining, confined to the antennal lobes (Fig. 8j). At 36 hapf, as high levels of Wnt5 accumulated in the commissure (arrowhead in Fig. 8k, compare with Fig. 8i), the projection neuron dendrites began to transgress the antennal lobe boundaries and projected medially toward the region of ectopic Wnt5 enrichment (arrows in Fig. 8k). Our results show that in the absence of, Wnt5 accumulates in the commissure, leading to the formation of ectopic midline glomeruli. DISCUSSION The mechanisms by which ingrowing axons sort into precise maps, such as those found in the olfactory glomeruli or the somatosensory barrels, are poorly understood. Deafferentation and transplantation experiments revealed that ingrowing axons are important for specifying the maps in the initially homogenous structures 15,16,34. However, little is known about how the ingrowing axons carry out these feats. In this report, we show that ingrowing ORN axons express Wnt5, which contributes to organizing the glomerular pattern of the Drosophila g Percentage intensity of Wnt5 mab staining (%) 1 1 ; Repo > Superficial Commissure 8 ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE

9 olfactory system. We also show that the Drl receptor tyrosine kinase acts in glial cells to modulate Wnt5 signaling. This previously unknown interaction between ORN axons and glia reveals an important function of ORN axon-glia interactions in regulating the precise neural circuitry of the Drosophila antennal lobes. The mutant had characteristic disruptions of the olfactory map. Many dorsomedial glomeruli were displaced ventrally (resulting in heart-shaped antennal lobes) and the antennal commissure failed to form. In contrast to the loss-of-function defects, overexpression of led to the displacement of glomeruli into the midline. Examination of the ORN axons in the mutant showed that they take circuitous paths to their targets and frequently misprojected to dorsal regions of the brain. Consistent with a role for in antennal lobe development, the antennal lobe defects appeared during the pupal stage, when ORN axon targeting and glomerular development occur. Our genetic mosaic and cell type specific rescue experiments indicated that is required in the ORNs. Antibody stainings indicated that the Wnt5 protein was enriched on the dendrites of the projection neurons, where it presumably accumulated subsequent to its secretion by ORNs. In addition to the projection neuron dendrites, Wnt5 also accumulated in the antennal commissure in the 2 mutant. We propose that Wnt5 is a signal by which ingrowing ORN axons direct the development of their target field. Mutation of the gene also produced disruptions of the olfactory map. However, unlike the stereotyped shifts of glomeruli seen in the mutant, the glomeruli were randomly positioned in or missing from one antennal lobe in the mutant. Furthermore, there was a strong tendency for glomeruli to form at the midline. As in the mutant, ORN axons took indirect routes to their targets. That functions in development is supported by the observation that antennal lobe defects were visible at 4 hapf, the time when ORN axon targeting and glomerular development take place. Antibody staining showed that the Drl protein was highly expressed by the projection neurons and TIFR glia, cells that are intimately associated with the ingrowing ORN axons. In the projection neurons, Drl was enriched in the dendrites of nascent glomeruli, four of which also appeared to accumulate Wnt5. The TIFR is a donut-shaped midsagittal structure located between the antennal lobes. Our histological studies showed that TIFR glial processes were closely associated with ORN axons that were projecting across the midline. Several observations indicated that functions in the TIFR to regulate function. First, removal of from single projection neuron clones did not disrupt the development and morphology of the projection neurons. Second, neuronal expression of in the 2 mutant background did not rescue the mutant phenotype. Third, expression of UAS- under the control of Repo-Gal4 strongly rescued the mutant phenotype, suggesting that functions in glial cells. Although we cannot rule out roles for Drl in the projection neurons, collectively, our observations suggest that functions predominantly in glial cells to regulate antennal lobe development. The phenotypic similarities between the loss-of-function and the -overexpressing mutants raise the intriguing possibility that the two genes act antagonistically in antennal lobe development. Indeed, expression of a weak transgene in the ORNs, which has no effect in the wild type, triggers the formation of ectopic glomeruli in the 2 mutant. Thus, and function in opposition to each other in antennal lobe development. To ascertain the relative positions of and in this signaling pathway, we generated animals carrying null mutations in both genes. We found that the 4 ; 2 double mutants had the characteristic phenotype. The gene is therefore epistatic to the gene, indicating that functions downstream of in antennal lobe development. This conclusion is also supported by the observation that, although the removal of a copy of the gene strongly suppressed the homozygous mutant phenotype, the removal of a copy of the gene had no effect on the homozygous mutant phenotype. The genetic data that downregulates function is further supported by our observation that the Wnt5 protein significantly accumulates in the commissure in the absence of Drl. Taken together, our genetic and histological data indicate that acts to inhibit the activity of during antennal lobe development. To probe the molecular mechanisms by which Drl regulates antennal lobe development, we mutated the various domains of Drl. We observed that neither disruption of the kinase activity nor deletion of the intracellular domain significantly impaired rescue by the transgene. In contrast, deletion of the extracellular WIF domain completely abolished Drl s ability to rescue the mutant phenotype. These results suggest that Drl regulates antennal lobe patterning predominantly through its extracellular WIF domain. How might Drl inhibit the function of Wnt5? One possibility is that Drl inhibits Wnt5 function simply by promoting Wnt5 s sequestration or endocytosis, thus limiting its interaction with another as yet unidentified receptor (Supplementary Fig. 5 online). This receptor might be one of the other Drosophila receptor tyrosine kinases or a member of the Frizzled family, one of which, frizzled 2 (fz2), interacts genetically with to stabilize axons of the Drosophila visual system 22. Alternatively, Drl may directly interact with another receptor and Wnt5, as has been observed previously for its mammalian ortholog Ryk and members of the Wnt and Frizzled families 35. This interaction could inhibit or alter the signal transduced from the membrane. However, we did not observe a requirement for Drl s cytoplasmic domain, suggesting that transduction of the Wnt5 signal by Drl alone is unlikely to have a major role in patterning the antennal lobes. How do glial cells interact with the ORN axons to specify the olfactory map? Our data suggest that the ingrowing ORN axons contribute to antennal lobe patterning through secretion of Wnt5 and that glial cells locally regulate Wnt5 actions through Drl. We propose the following working model for how Wnt5-Drl signaling might regulate glomerular patterning. Ingrowing ORN axons express Wnt5, which is important for the precise organization of the glomeruli and pathfinding of the ORN axons, such as those crossing the midline or projecting to the dorsomedial region of the antennal lobes. Normal antennal lobe development requires that the Wnt5 signal be locally attenuated by the TIFR glial cell expressed Drl protein. In the mutant, the lack of Wnt5 signaling results in the failure of ORN axons to cross the midline and the establishment of glomeruli in more ventral positions. In the mutant, Wnt5 accumulates at the midline and presumably inappropriately signals through another receptor, resulting in aberrant ORN axon targeting to the midline and the formation of ectopic glomeruli at the dorsomedial corner of the antennal lobe and at the midline. Further studies will hopefully help to unravel the precise mechanisms by which Wnt5 and Drl act together to specify the patterning of the Drosophila olfactory map. METHODS Experimental animals. The P{GS} Drosophila lines were generated by the Drosophila Gene Search Project (Metropolitan University, Tokyo) 36. The Or-Gal4 and Or-MCD8-GFP (Or-mGFP) lines were from L. Vosshall (The Rockefeller University) and B. Dickson (IMBA, Austria), the UAS- Dcyto line was from J. Dura (Institut de Genetique Humaine) 33 and the 72OK-Gal4 line was kindly provided by K. Ito (The University of Tokyo). The construction of NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION 9

10 the GH-mGFP line has been described 14. Other stocks, referenced throughout the text, were obtained from Bloomington Drosophila Stock Center. Transgenes. Full-length cdna (from Open Biosystems) was subcloned into the puast vector to generate UAS-. TocreateUAS- K371A, we generated a fragment of the coding region (base pairs ) bearing the K371A mutation by PCR, fused it in frame with the remainder of the coding region and then cloned it into puast. To generate UAS- DWIF, we generated a DNA fragment (nucleotide ) that lacked the WIF-encoding sequences by PCR, fused it with the first 6 nucleotides of the coding region, and then subcloned it into puast. Transgenic animals were generated by standard procedures. Immunohistochemistry. Adult (1 to 2 d old) or pupal brains were quickly dissected in cold phosphate-buffered saline (PBS, 13 mm NaCl, and 1 mm Na 2 HPO 4, ph 7.2) and fixed in PLP fixative (2% paraformaldehyde,.25% sodium periodate, 75 mm lysine-hcl and 37 mm sodium phosphate, ph 7.4) for 1 h. The fixed brains were washed with PBST (PBS with.5% Triton X-1) and stained with primary antibodies overnight. For Wnt5 staining, dissected brains were directly stained with antibody to Wnt5 in PBS (2.5 h at 4 1C), washed with PBS and goat serum, and fixed in PLP (1 h, 25 1C). Antibodies and dilutions are described here. mab nc82 (1: dilution) was a gift from A. Hofbauer 37, rabbit antibody to GFP (1:1 dilution) was obtained from Molecular Probes, and rat antibody to mcd8 (1:1 dilution) was from Caltag. Affinity-purified rabbit antibody to Wnt5 (ref. 21) and rabbit antibody to Drl were both used at 1:1 dilutions. The anti-drl antiserum was raised against a GST fusion protein that included Drl amino acids and was affinity purified against the same protein coupled to a column. The secondary antibodies, FITC-conjugated goat antibodies to rabbit, Cy3-conjugated goat antibodies to mouse and FITC-conjugated goat antibodies to rat, were obtained from Jackson Laboratories and were used at 1:1 dilution. Statistic analysis. Statistical comparisons were performed using Microsoft Excel and GraphPad Prism4. Tests on independent groups were two-tailed Student s t-test. Values and error bars indicate mean ± s.e.m., and denote P o.5,.1 and.1, respectively. Note: Supplementary information is available on the Nature Neuroscience website. ACKNOWLEDGMENTS We thank the Bloomington Drosophila Stock Center for the fly lines, A. Hofbauer for the generous gift of the nc82 antibody and W. Zhou for construction of the UAS- transgenic fly line. This work was supported by grants from the US National Institutes of Health and the US National Institute on Deafness and other Communication Disorders (DC548-1), the Roy J. Carver Charitable Trust (#3-27) (H.H.), ASPASIA (Netherlands Organization for Scientific Research) and Pionier grants (J.N.), and a Genomics grant (L.F. and J.N.) from the Nederlandse Organisatie voor Wetenschappelijk Onderzoek. AUTHOR CONTRIBUTIONS Y.Y., Y.W., C.Y. and R.O. conducted experiments in the laboratory of H.H. R.R.W. conducted experiments in the laboratory of J.N.N. L.G.F., Y.Y., Y.W. and H.H. analyzed the data. T.A. provided the P{GS} lines screened by the H.H. lab. H.H. and L.G.F. wrote the manuscript with contributions from the other authors. Published online at Reprints and permissions information is available online at reprintsandpermissions 1. Mombaerts, P. Axonal wiring in the mouse olfactory system. Annu. Rev. Cell Dev. Biol. 22, (6). 2. Dynes, J.L. & Ngai, J. Pathfinding of olfactory neuron axons to stereotyped glomerular targets revealed by dynamic imaging in living zebrafish embryos. Neuron, (1998). 3. Potter, S.M. et al. Structure and emergence of specific olfactory glomeruli in the mouse. J. Neurosci. 21, (1). 4. Miyasaka, N. et al. Robo2 is required for establishment of a precise glomerular map in the zebrafish olfactory system. Development 132, (5). 5. Lin, D.M. & Ngai, J. Development of the vertebrate main olfactory system. Curr. Opin. Neurobiol. 9, (1999). 6. Hummel, T. et al. Axonal targeting of olfactory receptor neurons in Drosophila is controlled by Dscam. Neuron 37, (3). 7. Hummel, T. & Zipursky, S.L. Afferent induction of olfactory glomeruli requires N-cadherin. Neuron 42, (4). 8. Key, B. & St John, J. Axon navigation in the mammalian primary olfactory pathway: where to next? Chem. Senses 27, (2). 9. Ang, L.H., Kim, J., Stepensky, V. & Hing, H. Dock and Pak regulate olfactory axon pathfinding in Drosophila. Development 13, (3). 1. Lattemann, M. et al. Semaphorin-1a controls receptor neuron-specific axonal convergence in the primary olfactory center of Drosophila. Neuron. 53, (7). 11. Sweeney, L.B. et al. Temporal target restriction of olfactory receptor neurons by Semaphorin-1a/PlexinA-mediated axon-axon interactions. Neuron. 53, 185 (7). 12. Oland, L.A., Orr, G. & Tolbert, L.P. Construction of a protoglomerular template by olfactory axons initiates the formation of olfactory glomeruli in the insect brain. J. Neurosci. 1, (199). 13. Stout, R.P. & Graziadei, P.P. Influence of the olfactory placode on the development of the brain in Xenopus laevis (Daudin). I. Axonal growth and connections of the transplanted olfactory placode. Neuroscience 5, (198). 14. Ang, L.H. et al. Lim kinase regulates the development of olfactory and neuromuscular synapses. Dev. Biol. 293, (6). 15. Graziadei, P.P. & Kaplan, M.S. Regrowth of olfactory sensory axons into transplanted neural tissue. 1. Development of connections with the occipital cortex. Brain Res. 1, (198). 16. Schneiderman, A.M., Matsumoto, S.G. & Hildebrand, J.G. Trans-sexually grafted antennae influence development of sexually dimorphic neurones in moth brain. Nature 298, (1982). 17. Oland, L.A. & Tolbert, L.P. Multiple factors shape development of olfactory glomeruli: insights from an insect model system. J. Neurobiol. 3, (1996). 18. Moon, R.T., Bowerman, B., Boutros, M. & Perrimon, N. The promise and perils of Wnt signaling through beta-catenin. Science 296, (2). 19. Cadigan, K.M. & Nusse, R. Wnt signaling: a common theme in animal development. Genes Dev. 11, (1997).. Yoshikawa, S., McKinnon, R.D., Kokel, M. & Thomas, J.B. Wnt-mediated axon guidance via the Drosophila Derailed receptor. Nature 422, (3). 21. Fradkin, L.G. et al. The Drosophila Wnt5 protein mediates selective axon fasciculation in the embryonic central nervous system. Dev. Biol. 272, (4). 22. Srahna, M. et al. A signaling network for patterning of neuronal connectivity in the Drosophila brain. PLoS Biol. 4, e438(6). 23. Zhang, D. et al. Misexpression screen for genes altering the olfactory map in Drosophila. Genesis 44, (6). 24. Jhaveri, D., Sen, A. & Rodrigues, V. Mechanisms underlying olfactory neuronal connectivity in Drosophila-the atonal lineage organizes the periphery while sensory neurons and glia pattern the olfactory lobe. Dev. Biol. 226, (). 25. Fishilevich, E. & Vosshall, L.B. Genetic and functional subdivision of the Drosophila antennal lobe. Curr. Biol. 15, (5). 26. Couto, A., Alenius, M. & Dickson, B.J. Molecular, anatomical, and functional organization of the Drosophila olfactory system. Curr. Biol. 15, (5). 27. Lee, T. & Luo, L. Mosaic analysis with a repressible neurotechnique cell marker for studies of gene function in neuronal morphogenesis. Neuron 22, (1999). 28. Dura, J.M., Taillebourg, E. & Preat, T. The Drosophila learning and memory gene linotte encodes a putative receptor tyrosine kinase homologous to the human RYK gene product. FEBS Lett. 37, (1995). 29. Simon, A.F., Boquet, I., Synguelakis, M. & Preat, T. The Drosophila putative kinase linotte (derailed) prevents central brain axons from converging on a newly described interhemispheric ring. Mech. Dev. 76, (1998). 3. Jefferis, G.S., Marin, E.C., Stocker, R.F. & Luo, L. Target neuron prespecification in the olfactory map of Drosophila. Nature 414, 4 8 (1). 31. Stocker, R.F., Heimbeck, G., Gendre, N. & de Belle, J.S. Neuroblast ablation in Drosophila P[Gal4] lines reveals origins of olfactory interneurons. J. Neurobiol. 32, (1997). 32. Hanks, S.K., Quinn, A.M. & Hunter, T. The protein kinase family: conserved features and deduced phylogeny of the catalytic domains. Science 241, (1988). 33. Taillebourg, E., Moreau-Fauvarque, C., Delaval, K. & Dura, J.M. In vivo evidence for a regulatory role of the kinase activity of the linotte/derailed receptor tyrosine kinase, a Drosophila Ryk ortholog. Dev. Genes Evol. 215, (5). 34. Schlaggar, B.L. & O Leary, D.D. Potential of visual cortex to develop an array of functional units unique to somatosensory cortex. Science 252, (1991). 35. Lu, W., Yamamoto, V., Ortega, B. & Baltimore, D. Mammalian Ryk is a Wnt coreceptor required for stimulation of neurite outgrowth. Cell 119, 97 18(4). 36. Toba, G. et al. The gene search system. A method for efficient detection and rapid molecular identification of genes in Drosophila melanogaster. Genetics 151, (1999). 37. Hofbauer, A. Eine Bibliothek monoklonaler Antikorper gegen das Gehirn von Drosophila melanogaster. (University of Wurzburg, Wurzburg, 1991). 1 ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE