A, Schematic of the velo gene locus showing a long (5852 base pairs (bp) CG10107-RA) and short (2540 bp CG10107-RB) velo transcript with the piggyBac (pBac) insertions (red triangles) used in this study. A third velo transcript (5867 bp CG10107-RC; not shown here) differs from the long transcript by 15 bp in the 5’UTR. pBacLL05207 and pBacLL05209 insertions originate from a previously described pBac collection and are inserted 490 bp upstream from the translational start in the first intron of the long transcript and 30 bp upstream from the translational start of the short transcript, respectively (Schuldiner et al., 2008). pBac(RB)e01260 originates from the Exelixis collection (Thibault et al., 2004) and is inserted 33 bp upstream from the second exon in the first intron of the short transcript. Exons are shown as black, UTRs as grey bars, and introns as lines. B, WT anterodorsal projection neurons (adPNs) and D, lateral projection neurons (lPNs) target a stereotyped set of glomeruli. C,E, velo mutant adPNs and lPNs exhibit reduced cell numbers, their dendrites are misorganize, innervating either incorrect glomeruli within (dashed circle in C) or targets outside the antennal lobe (AL) (arrowhead in E). F, WT DL1 PN dendrites innervate the DL1 glomerulus specifically. Asterisks and dashed circles in F–K, M–N denote cell bodies of single cell clones and the DL1 glomerulus innervated by these PNs, respectively. G–K, representative images of velo^LL05209 mutant single neurons arranged into 5 classes according to their dendrite phenotypes. Class 1 velo^LL05209 mutant single neurons innervate the DL1 glomerulus often more sparsely and additionally innervate adjacent glomeruli partially, most often DL5 and D (G), whereas class 2 dendrites target dorsolateral glomeruli partially but avoid DL1 (H). Class 3 velo^LL05209 mutant dendrites exhibit ventromedial innervation patterns, most typically mistargeting to VM6 (I), class 4 dendrites innervate structures outside the AL, mostly the suboesophagial ganglion (SOG) (arrowhead in J), and class 5 dendrites target non-stereotypically to multiple glomeruli within the AL (arrowheads in K). L, Quantification of the 5 distinct dendrite mistargeting classes is shown in percentages (Y-axis) for three velo alleles (X-axis). Numbers in brackets represent phenotypic class. Numbers on top of the bars indicate the number of velo mutant DL1 single neurons analyzed. M–N, During development at 50h APF (after pupa formation) WT DL1 PN dendrites innervate the DL1 glomerulus (M) while dendrites of a velo^e01260 mutant DL1 PN mistarget to the ventromedial area (N). All images are confocal z-projections, mCD8-GFP labeled dendrites and cell bodies generated by MARCM using Gal4-GH146 are marked in green, the presynaptic marker nc82 is shown in red. Scale bars: 20µm. Genotype: yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/+; Gal80 FRT2A/ FRT2A, FRT82B, y+ (B, D, F, M), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/+; Gal80 FRT2A/ velo^LL05209[DsRed] FRT2A, FRT82B, y+ (C, E, G–K), and yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/+; Gal80 FRT2A/ velo^e01260[w+] FRT2A, FRT82B (N).

A–H, PN dendrites (A,C,E, and G) and axons (B,D,F, and H) of single neurons generated by MARCM are shown in green stained with antibodies to CD8, red represents staining using the presynaptic marker nc82. White asterisks in A,C,E, and G denote single cell bodies. A, WT PN dendrites innervate the DL1 glomerulus specifically. B, WT PN axons project stereotypically into higher brain centers, the mushroom body calyx (MBC) and lateral horn (LH). Characteristic axon branches in the MBC and LH are within the dotted lines, an arrowhead points to the dorsal branch (B). C–F, Representative images for two velo^LL05209 single PNs. One exhibits the mildest dendrite mistargeting phenotype illustrated by dendrites spilling into adjacent glomuerli (C), the other mistargets ventromedially to the vicinity of VM6 (E). The corresponding axons form fewer branches in the MBC and lack the dorsal branch in the LH (D,F). G,H, Typical velo^e01260 single PNs mistarget ventromedially, often to the VM6 glomerulus (G) and their corresponding axon bears fewer branches in the MBC and LH or lacks them entirely (H). A–H represent confocal z-projections. Scale bars: 20µm. Genotypes: yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/+; Gal80 FRT2A/ FRT2A, FRT82B, y+ (A,B),yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/+; Gal80 FRT2A/ velo^LL05209[DsRed] FRT2A, FRT82B, y+ (C–F), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ +; Gal80 FRT2A/ velo^e01260[w+] FRT2A, FRT82B (G,H).

A–B, Postmitotic expression of UAS-velo-HA in single DL1 velo mutant PNs with Gal4-GH146 results in localization of the Velo-HA fusion protein to the nucleus (inset in A), and rescues velo^e01260 mutant dendrite phenotypes in 20/22 cases (A) and axon (B) phenotypes as quantified. C–D, Quantification of the PN axon morphogenesis phenotypes in the MBC (C) and the dorsal branch in the LH (D) are illustrated as percentages of phenotypes on the Y- and genotypes on the X-axis. C, Single DL1 PNs are divided into groups that contain 0–2 or more than 3 branches in the MBC. D, Additionally, single PNs are examined for the presence of a dorsal branch in the LH. Numbers on top of the bars indicate the number of single PNs analyzed. E, VeloWT-HA, VeloC>S-HA or GFP were expressed in S2 cells and detected by Western blot. Expression levels were similar for both WT and mutant proteins. α-Tubulin was used as a loading control. F–G, MARCM experiments expressing a catalytically dead Velo protease UAS-velo^C>S-HA in velo^e01260 mutant PNs fail to rescue PN dendrite in 8/8 (F) and axon (G) phenotypes in 5/8 cases, and even causes premature axon termination outside the LH in 3/8 (arrowhead in G). A–B, F–G, confocal z-projections stained with anti-CD8 (green), nc82 (red), and anti-HA (inset in A,F), except insets in A,F are single confocal sections. Asterisks in A,F mark single cell bodies. Scale bars: 20µm. Genotypes: yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ UAS-velo-HA^#108L; Gal80 FRT2A/ velo^e01260[w+] FRT2A, FRT82B (A,B), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ +; Gal80 FRT2A/ velo^e01260[w+], FRT2A, FRT82B UAS-velo^C>S-HA^#1.2 (F,G).

A, Phylogenetic tree displaying the relationship between budding yeast (Sc), human (Hs), and Drosophila (Dm) homologs of SUMO proteases. The number at each branch represents the confidence in that branch, as generated by bootstrapping analysis. For details about the making of this tree, see Materials and Methods. The ScUlp1-related family is illustrated in red, ScUlp2-related members in green letters. B, Schematic of the CG12717 gene locus indicating the insertions of the P-element P(XP)d05069 and pBac(RB)e01706a lines (Thibault et al., 2004) (red triangles) used to generate the deletion mutant CG12717^ex#6 by FLP-mediated excision. Brackets illustrate the extent of deletion. Exons are shown as black, UTRs as grey bars, and introns as lines. C–D, CG12717^ex#6 adPNs and single DL1 PNs target, like WT PNs, target a stereotyped set of glomeruli and to the DL1 glomerulus, respectively. E–F, Typical velo^e01260 single PNs mistarget ventromedially (VM), often to the VM6 glomerulus (E) which is rescued by adding UAS-DmUlp1-HA into the same mutant neuron in 12/14 PNs (F). Inset in F demonstrates subcellular distribution of the DmUlp1 fusionprotein. Asterisks in D,E, and F denote single cell bodies. H–J, velo^e01260 adPN cell numbers are reduced (H), a phenotype that is significantly rescued by introducing a UAS-DmUlp1-HA transgene (I). adPNs are encircled by dotted lines. J, DmUlp1-HA (in red and inset) localizes to the nucleus and cytoplasm. G, Quantification of adPN numbers in velo^e01260 neuroblast clones and after introducing various UAS rescue transgenes. Numbers on top of the bars indicate the number of analyzed anterodorsal neuroblast clones. Error bars indicate SEM. **p<0.00002 in two-tailed t test. C–F, H–I, confocal z-projections stained with anti-CD8 (green), nc82 (red), and anti-HA (insets in F,J), except J and insets in J and F are single confocal sections. Scale bars: 20µm. Genotypes: yw UAS-mCD8GFP, CG12717^ex#6, FRT19A/ yw Gal80, hsFlp¹²², FRT19A; Gal4-GH146 UAS-mCD8GFP/ + (C,D), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ +; Gal80 FRT2A/ velo^e01260[w+], FRT2A, FRT82B (E,H,J), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ UAS-DmUlp1-HA^#5.1; Gal80 FRT2A/ velo^e01260[w+], FRT2A, FRT82B (F,G,I,J), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ UAS-velo-HA^#108L; Gal80 FRT2A/ velo^e01260[w+], FRT2A, FRT82B and yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ UAS-CG12717-HA^#4; Gal80 FRT2A/ velo^e01260[w+], FRT2A, FRT82B (G).

A, Schematic representation of reactions catalyzed by SUMO proteases in the protein sumoylation pathway. The processing reaction requires a SUMO protease (blue) and involves cleavage of a peptide bond within the SUMO (yellow) pro-peptide to expose the C-terminal di-glycine motif. E1, E2 and E3 are the activating enzyme, conjugating enzyme and SUMO ligase, respectively, and are required for SUMO conjugation (black arrow). The lesswright (lwr) gene encodes the only E2 enzyme in Drosophila. SUMO de-conjugation requires cleavage of the amide bond between mature SUMO and the target lysine within the substrate (green). Chain editing is chemically identical to de-conjugation, although it is distinguished by cleavage of one or more SUMOs from a poly-SUMO chain. SUMO catalyzed reactions are indicated by blue arrows. B, WT DL1 PN dendrites innervate the DL1 glomerulus specifically. C–D, PNs mutant for the original P-element l(2)SH0182 (G) exhibit normal DL1 innervation (C), while SUMO^ex77 DL1 PN dendrites fail to innervate the DL1 glomerulus (D). E–F, lwr¹³ DL1 PNs mistarget their dendrites ventromedially from DL1, avoiding DL1 (E); this phenotype is fully rescued by introducing a UAS-lwr transgene in 12/12 single neurons (F). B–F, confocal z-projections stained with nc82 as a presynaptic marker (red) and anti-CD8 (green) for PN single cell bodies (asterisks) and their dendrites. Dashed circles outline the DL1 glomerulus. Scale bar: 20µm. G, Schematic of the SUMO gene locus showing the P-element insertion l(2)SH0182 (red triangle) 10 base pairs (bp) from the transcriptional start. l(2)SH0182 was used for the imprecise excision to generate the SUMO^ex77 deletion allele (shown by brackets). The positions of primers to screen for imprecise excision mutations are denoted with red arrows. Exons are shown as black, UTRs as grey bars, and introns as lines. H, 1% Agarose gel showing a ~2,000 bp DNA PCR product amplified with the primers shown in G from WT fly genomic DNA (lane 1). From SUMO^ex77 heterozygous fly genomic DNA two fragments are amplified denoted by black arrowheads (lane 2+3): 2,000 bp for the wt and 800 bp for the deletion chromosome, indicating a ~1,200 bp deletion in SUMO^ex77 mutants. Standard DNA ladder is loaded in right lane. Genotypes: yw hsFlp¹²² UAS-mCD8GFP; FRT40A Gal4-GH146 UAS-mCD8GFP/ FRT40A (B),yw hsFlp¹²² UAS-mCD8GFP; FRT40A Gal4-GH146 UAS-mCD8GFP/ l(2)SH0182 FRT40A (C), yw hsFlp¹²² UAS-mCD8GFP; FRT40A Gal4-GH146 UAS-mCD8GFP/ SUMO^ex77 FRT40A (D), yw hsFlp¹²² UAS-mCD8GFP; FRT40A Gal4-GH146 UAS-mCD8GFP/ lwr¹³ FRT40A, FRTG13 (E), yw hsFlp¹²² UAS-mCD8GFP/ UAS-lwr; FRT40A Gal4-GH146 UAS-mCD8GFP/ lwr¹³ FRT40A, FRTG13 (F), w¹¹¹⁸ and yw; SUMO^ex77 FRT40A/CyO (H).

A, Two possible models for mechanisms of Velo action. Model 1 depicts a role for Velo in SUMO maturation, cleaving immature SUMO into its active form. Model 2 shows a role for Velo in SUMO de-conjugation by cleaving SUMO either from mono-sumoylated substrates or from SUMO chains. B,E, A typical example (representing 38%) for a MARCM generated velo^LL05209 single PN mistargeting its dendrites ventromedially. C,E, This mistargeting phenotype is significantly reverted by reducing SUMO gene dosage by half resulting in 4% of ventromedial mistargeting cases while milder DL1 targeting defects increase from 24% to 80%. D,E, This suppression effect is also seen after reduction of lwr gene dosage in velo^LL05209 single PNs causing an increase in milder DL1 and a decrease in more severe ventromedial targeting errors. E, Suppression effect is quantified by plotting distinct phenotypic classes in percent on the Y-axis for various genetic combinations shown on the X-axis. Numbers in brackets represent the phenotypic class as shown in Fig. 1. Numbers on top of the bars indicate the number of DL1 single neurons analyzed. Asterisks in (B–D) denote single cell bodies, images are confocal z-projections stained with a presynaptic marker nc82 (red) and anti-CD8 (green). Scale bar: 20µm. Genotypes: yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ +; Gal80 FRT2A/ velo^LL05209[DsRed], FRT2A, FRT82B (B,E), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ SUMO^ex77; Gal80 FRT2A/ velo^LL05209[DsRed], FRT2A, FRT82B (C,E), yw hsFlp¹²² UAS-mCD8GFP; Gal4-GH146 UAS-mCD8GFP/ lwr¹³; Gal80 FRT2A/ velo^LL05209[DsRed], FRT2A, FRT82B (D,E).