(a-h) The developing antennal lobes at 48 hrs APF stained by antibodies against a synaptic marker nc82 (a, d and g), Caps (b, e and h) and caps-Gal4-driven mCD8-GFP (c and f). Caps is differentially expressed in the wild-type developing antennal lobe as shown in a single anterior (a-c) and posterior section (d-f) from the same triple-labeled brain. (g-h) Caps staining is absent in caps^c28fs/Df(3L)6118 trans-heterozygous mutant antennal lobes.
(i-k) Expression of a Flp-out GFP reporter UAS>stop>mCD8GFP at the intersection of caps-Gal4 and PN-specific GH146-Flp in the adult antennal lobe. Three single sections in anterior, central, and posterior regions of the antennal lobe are shown. (Magenta: nc82; Green: mCD8GFP)
(l) Schematic representation of the glomerular innervation pattern of caps-Gal4-expressing PNs across three different sections of the antennal lobe. Three groups of glomeruli are indicated in the legend to the left. See Fig. S2 for details. Antennal lobes in this and all subsequent images are shown such that dorsal is up and medial is to the left. Scale bars represent 10 μm.

(a and b) Dendrite targeting of neuroblast MARCM clones of wild-type and caps^−/− PNs. (a) Wild-type lateral neuroblast clones stereotypically target dendrites to 12 glomeruli. (b) caps^−/− lateral neuroblast clones exhibit selective dendrite mistargeting: arrowhead indicates the ectopic innervation of VM7 where wild-type PNs do not innervate; arrow indicates the loss of innervation of wild-type glomerular target VA4.
(c) Quantification of dendrite targeting defects in lateral neuroblast MARCM clones of control and caps mutant. The Y-axis represents the percentage of the antennal lobes that exhibit mistargeting phenotypes. Light-grey color, dark-grey color, and black color represent the percentages of the antennal lobes: that innervate ectopic glomeruli in addition to the normal glomerular targets, that only exhibit loss of innervation of the normal glomerular targets, and that exhibit both defects, respectively. (Control: n=9; caps^−/−: n=12.)
(d) Glomerular innervation specificity of lateral neuroblast MARCM clones of control and caps mutants analyzed in (c). The left 12 columns are the normal glomerular targets of lateral PNs; green bars: the percentage of antennal lobes in which an individual glomerulus is innervated; white bars: the percentage of antennal lobes in which a normally innervated glomerulus is not innervated. The right 34 columns are glomeruli that are normally not innervated by lateral PNs; black bars: the percentage of antennal lobes in which an ectopic glomerulus is innervated. All glomeruli are color-coded at the bottom based on the expression of Caps in the corresponding PNs as indicated. ND, not determined for Caps expression (GH146-negative).

(a-c, h-j) Normal dendrite targeting of caps^−/− MARCM clones in DL1 single cells (h), DA1 neuroblasts (i) and VA1d/DC3 neuroblasts (j), compared with wild-type (a-c). These PNs are all Caps-negative. n=20 for both wild-type and caps^−/−.
(d-g, k-n) Defective dendrite targeting of VC1, VC2, VA4 and DM1 PNs in single-cell caps^−/− clones (k-n) compared with wild-type (d-g). These PNs are all Caps-positive. Their identities are determined as described in the Methods.
(o-r) Rescue of dendrite targeting of four Caps-positive PNs by expressing a UAS-caps transgene only in the single-cell clones.
(s-v) Quantification (s, u) and glomerular innervation specificity (t, v) of dendrite targeting defects in single-cell clones of control and caps mutant analyzed in (d-g, k-r). VC2, VA4 and DM1 were analyzed together (see Methods). (s, u) Y-axes represent the percentage of PNs in particular classes that exhibit dendrite mistargeting phenotypes. (t, v) Each black bar indicates the percentage of antennal lobes in which an ectopic glomerulus is innervated. All glomeruli are color-coded at the bottom as in Fig. 2d. VC1 and VC2/VA4/DM1 classes are omitted in t and v, respectively, as they are the ones from which clones were made. (Wild-type: VC1 n=15, VC2 n=15, VA4 n=15, DM1 n=10; caps^−/−: VC1 n=7; VC2/VA4/DM1 n=19; Rescue: VC1 n=6, VC2 n=5, VA4 n=6, DM1 n=4).
(w) Schematic representation of the glomerular innervation pattern of individual caps^−/− VC1 PNs across three different sections of the antennal lobe. Red glomeruli are the ectopic targets of caps^−/− VC1 dendrites.

(a-d) MARCM overexpression of Caps in anterodorsal (b) and lateral (d) neuroblast clones shows severe mistargeting phenotypes, compared to wild type neuroblast clones (a and c). (Control: n=9; Caps misexpression: n=10).
(e-f) Caps misexpression in DA1, DC3 and VA1d PNs using Mz19-Gal4 shows strong mistargeting of dendrites to VA1lm and VA4 (arrowheads), both of which are normally innervated by Caps-positive PNs. (Control: n=40; Caps misexpression: n=48).
(g-i) Misexpression of Caps in single DL1 PNs using GH146-Gal4 causes mistargeting of dendrites to ectopic glomeruli (arrowheads in h-i).
(j) Quantification of glomerular innervation pattern of individual DL1 PNs misexpressing Caps. Each black bar indicates the percentage of antennal lobes in which an ectopic glomerulus is innervated. All glomeruli are color-coded at the bottom as in Fig. 2d. (Control: n=20; Caps misexpression: n=19).
(k) Schematic representation of the glomerular innervation pattern of individual Caps-misexpressing DL1 PNs across three different sections of the antennal lobe. Red glomeruli: ectopic targets of DL1 dendrites misexpressing Caps.

(a-c) Expression of an intersectional reporter UAS>stop>mCD8GFP for caps-Gal4 and ey-Flp. (ey-Flp is expressed in ORN precursors but not PNs or their precursors). Magenta: nc82; Green: mCD8GFP.
(d) Schematic representation of the glomerular innervation pattern of caps-Gal4 expressing ORNs across three different sections of the antennal lobe. Gray glomeruli: targets of caps-Gal4 positive ORNs; white glomeruli: targets of caps-Gal4 negative ORNs.
(e) Schematic comparison of the innervation pattern of Caps-positive PNs and ORNs. See also Fig. S2.
(f-i) Axon targeting of ORNs in the antennal lobe is not affected in caps^−/− ORNs. Or67b ORNs innervating VA3 and Or22a ORNs innervating DM2 are shown. In these experiments, only caps^−/− axons of a particular class of ORNs are visualized by the Gal4 lines, whereas all the cells in the central brain, including all PNs, are heterozygous because ey-Flp restricts recombination in the olfactory system to the peripheral organs.

(a-d) Mistargeting of DL1 PN dendrites caused by Caps misexpression is observed at early developmental time points. Dotted lines bisect the antennal lobe into dorsolateral and ventromedial parts. Arrowheads in (b) and (d) point to DL1 dendrites that cross the dotted line and invade the ventromedial part of the antennal lobe.
(e) Quantification of mistargeting across the dorsomedial-ventrolateral midline of the antennal lobe (dotted lines in a-d) at different developmental stages. Adult: n=12; 16 hrs APF: n=10; 48 hrs APF: n=5.
(f-j) Caps misexpression by Mz19-Gal4 causes a segregation of dendritic field as a consequence of dendrite mistargeting to VA1lm and VA4 (arrowheads in g), compared to continual dendritic field from control Mz19-Gal4 labeled PNs (f). Caps misexpression by Mz19-Gal4 in ORN-ablated animals causes dendrite segregation (i) compared with control Mz19 PNs in ORN-ablated animals which forms a continual field (h). (j) Quantification of the dendrite segregation phenotypes caused by Mz19-Gal4 misexpression of Caps, with and without ORN ablation. Caps misexpression without ORN ablation: n=48; Caps misexpression with ORN ablation: n=18.

(a) Normal dendrite targeting of a single DL1 PN in a caps whole animal mutant (caps^c28fs/Df(3L)6118).
(b-c) Dendrite mistargeting of a single DL1 PN misexpressing Caps in an otherwise wild-type background (b) or an otherwise caps^c28fs/Df(3L)6118 mutant background (c).
(d) Quantification of DL1 targeting defects for genotypes in a-c. (DL1 labeling in wild-type background: n=20; DL1 labeling in trans-heterozygous mutant background: n=20; Caps misexpression in a otherwise wild-type background: n=8; Caps misexpression in a otherwise trans-heterozygous mutant background: n=9). Grey bars represent the percentage of partial mistargeting events in which PN dendrites still partially innervate DL1 and black bars represent the percentage of full mistargeting events in which no dendrites innervate DL1.

(a) Quantification of dendrite targeting defects in lateral neuroblast MARCM clones of control, single, and double mutants of caps and trn as indicated. The Y-axis and the colored bars are represented as described in Fig. 2c. (Wild-type: n=31; caps^c28fs FRT2A: n=30; trn^28.4 FRT2A: n=18; trn^Δ17 caps^c28fs FRT2A: n=17; trn^28.4 caps^Δ1 FRT80B: n=11)
(b) Quantification of dendrite targeting defects in single-cell MARCM clones of control, single, and double mutants of caps and trn as indicated. The Y-axis and the bars are represented as described in Fig. 7d. (For VC1, wild-type: n=20; caps^c28fs FRT2A: n=7; trn^28.4 FRT2A: n=5; trn^Δ17 caps^c28fs FRT2A: n=5; trn^28.4 caps^Δ1 FRT80B: n=5. For VC2/VA4/DM1, wild-type: n=30, caps^c28fs FRT2A: n=19, trn^28.4 FRT2A: n=13, trn^Δ17 caps^c28fs FRT2A: n=16, trn^28.4 caps^Δ1 FRT80B: n=13).
(c) Glomerular innervation specificity of lateral neuroblast MARCM clones of trn single mutant and caps, trn double mutants analyzed in (a). The Y-axis and the colored bars are represented as described in Fig. 2d.
(d-h) Dendrite targeting of single-cell VC1 MARCM clones of control, single, and double mutants of caps and trn as indicated. Arrowheads indicate the ectopic innervation of DA2 where wild-type VC1 PNs do not innervate.