Serrano (Sano) Functions with the Planar Cell Polarity Genes to Control Tracheal Tube Length

Epithelial tubes are the functional units of many organs, and proper tube geometry is crucial for organ function. Here, we characterize serrano (sano), a novel cytoplasmic protein that is apically enriched in several tube-forming epithelia in Drosophila, including the tracheal system. Loss of sano results in elongated tracheae, whereas Sano overexpression causes shortened tracheae with reduced apical boundaries. Sano overexpression during larval and pupal stages causes planar cell polarity (PCP) defects in several adult tissues. In Sano-overexpressing pupal wing cells, core PCP proteins are mislocalized and prehairs are misoriented; sano loss or overexpression in the eye disrupts ommatidial polarity and rotation. Importantly, Sano binds the PCP regulator Dishevelled (Dsh), and loss or ectopic expression of many known PCP proteins in the trachea gives rise to similar defects observed with loss or gain of sano, revealing a previously unrecognized role for PCP pathway components in tube size control.


Introduction
Multicellular animals employ tubular structures in organs to transport vital fluids and gases that sustain life. Examples of organs with prominent tubular architecture include the circulatory system, the lung and kidney in mammals, the secretory and respiratory organs in flies, and the excretory organ in worms. Proper development of tubular networks is critical for the function of several organs, evidenced by disruption of these networks being an underlying cause of common human diseases including cardiovascular disease, polycystic kidney diseases, and asthma.
The Drosophila trachea is a branched network of tubular epithelia that transports oxygen and other gases throughout tissues. The comparative simplicity and genetic tractability of this system has made it one of the most powerful model systems to dissect tubular epithelial morphogenesis. Tracheal formation begins as tracheal placodes invaginate from the epidermis during early embryogenesis. Through stereotypic cell migrations, cell shape changes, and rearrangements of cell-cell junctions, tracheal cells generate a tubular network that extends branches to all embryonic tissues [1][2][3][4].
In epithelia, cells are polarized along the apical/basal axis. In epithelial tubes, the apical surface of each cell faces the lumen, whereas the basal surface faces surrounding tissues and/or a basement membrane. In addition to apical/basal polarity, epithelial cells in most tissues require information about their orientation within the plane, orthogonal to the axis of apical/basal polarity, in order to generate polarized structures such as cilia, or to move or orient themselves in a directed fashion. This type of polarization is referred to as planar cell polarity (PCP). In vertebrates, PCP is involved in diverse patterning events, including convergence extension during gastrulation, neural tube closure, inner ear sensory hair morphogenesis, and hair follicle orientation [23]. In Drosophila, PCP biases cell orientation in several adult epithelial tissues and has been implicated in ovarian border cell migration [24][25][26][27]. In many contexts, both in vertebrates and in Drosophila, a conserved PCP pathway -the Frizzled (Fz) pathwaymediates local cell-cell interactions that instruct neighboring cells to adopt appropriate polarity [24][25][26][27].
In Drosophila, loss or overexpression of PCP proteins causes disorganization of wing hairs and bristles on the thorax and/or alteration in the orientation of ommatidia in the compound eye. Analysis of such phenotypes revealed an evolutionarily conserved set of genes that control planar polarity -the ''core'' PCP factors. These factors include: Fz, a seven-pass transmembrane receptor [28]; Dishevelled (Dsh), an adaptor protein that acts downstream of Fz [29][30][31]; Flamingo/Starry night (Fmi/Stan), a cadherinfamily member with a seven-pass transmembrane domain [32,33]; Strabismus/Vang Gogh (Stbm/Vang), a four-pass transmembrane protein [34,35]; and Prickle (Pk) and Diego (Dgo), each cytoplasmic proteins that are associated with the apical membrane during PCP signaling [36,37].
PCP pathway activity itself leads to polarized enrichment and distribution of core components in all Drosophila tissues analyzed to date. In pupal wing cells, core PCP proteins localize apicolaterally, partially overlapping with cellular junctions [17,38], and each protein is enriched in a distal and/or proximal location in the cells during prehair formation [33,37,[39][40][41][42]. The function of each core PCP protein is essential for the asymmetric accumulation of the other proteins.
The PCP signal from Fz/Dsh directs asymmetric cytoskeletal reorganization and polarized cell morphology, in part by activating RhoA/Rho1 [43] and its downstream effector, Drosophila Rho-associated kinase, Drok [44]. In the wing, RhoA signals via Drok, which regulates myosin II activity via phosphorylation of Spaghetti squash (Sqh), a Drosophila homolog of nonmuscle myosin II regulatory light chain (MRLC) [44,45]. Additional PCP regulators include Fat (Ft) and Dachsous (Ds), two protocadherins that can interact in a heterophilic fashion across cell boundaries [46,47], and the Golgi kinase Four jointed (Fj) [48,49]. Fj and Ds are expressed in a gradient in the eye and wing, making these proteins attractive candidates for providing upstream global polarity cues [46,47]. Alternatively, the Ft/Ds group may function in parallel to the core PCP proteins [50].
Recently, a role for PCP genes in regulating tube length and diameter by orienting cell divisions was demonstrated in vertebrate renal and gut epithelia [51,52], but whether the PCP components affect tube geometry in Drosophila remains unknown. Here we identify serrano (sano), a novel protein that affects tracheal tube length in Drosophila. sano mutant embryos have elongated tracheal dorsal trunks (DTs), whereas overexpression of Sano results in shortened DTs. Sano directly binds the core PCP component Dsh, and tracheal morphology and geometry are similarly affected by alterations in Sano activity and PCP signaling. Our results implicate for the first time the PCP mutants in Drosophila tubular morphogenesis.

sano Encodes a Cytoplasmic Protein That Is Apically Enriched
An enhancer trap screen for lines with expression in the developing salivary gland and trachea identified rp395, a P-element insertion that expresses b-gal throughout the salivary gland, in trachea, and in several other embryonic tissues, including the hindgut, midgut endoderm, CNS midline, posterior spiracles, and epidermis ( Figure 1A-1D). Cloning and characterization of the flanking region revealed that the rp395 P element had inserted after nucleotide 14 of the RC and RD transcripts of serrano (sano; CG12758), two of five alternatively spliced transcripts, designated RA-RE ( Figure 1M). Northern analysis revealed a single size transcript of 4.6 kb, first detected in 4-8-hour embryos and reaching peak levels in 8-12-hour embryos ( Figure S1). The transcript was detected at all subsequent developmental stages, but was not detected in RNA isolated from cultured Drosophila Schneider (S2) cells. With minor exceptions, the endogenous sano transcripts recapitulate the pattern of rp395 b-gal expression ( Figure 1E-1H).
Sano expression requires the transcription factors Sex combs reduced (Scr) in the salivary gland, Trachealess (Trh) in the trachea, and Single-minded (Sim) in the CNS midline ( Figure 1I-1L); sano expression was not affected by loss of transcription factors including fork head, huckebein, or CrebA that are expressed early in salivary gland formation (data not shown). Early transient tracheal expression of sano was observed in trh mutant cells also deficient for programmed cell death (Df(3L)H99), suggesting that initial tracheal expression is in part trh-independent and complete loss of sano expression in trh mutants is due to tracheal cell death ( Figure S2). Since other known regulators of tracheal development, including ventral veinless/drifter, trachea defective/apontic, breathless, branchless, and rhomboid, did not affect sano expression (data not shown), initial sano expression could be regulated by factors that initiate trh expression. Since Trh and Sim bind the same consensus DNA sequence [53,54], regulation of sano expression by these proteins could be direct.
All predicted sano splice forms encode the same 778-residue ORF ( Figure 1M). Sano is highly conserved in arthropods (Figure 2A), and is a member of a largely uncharacterized family of proteins with members from cnidarians to mammals that includes the recently identified Themis protein (also known as Gasp). Themis/Gasp is a cytosolic thymocyte-adaptor protein that binds Grb2 and is required for positive selection of thymocytes [55][56][57][58][59][60]. Because we were unable to generate antiserum that detected endogenous Sano, we cloned and expressed both untagged and C-terminally tagged (GFP or HA) Sano in flies under Gal4/UAS control [61]. In both tracheal and salivary gland cells, each version of overexpressed Sano localized diffusely in the cytoplasm, with enrichment at apical membranes, colocalizing with the apical membrane markers Crumbs (Crb) and Stranded at Second (SAS) (Figure 2B-2D; [62][63][64]). During late embryogenesis and in the 3 rd instar larval salivary gland, however, Sano-GFP also localized to nuclei ( Figure 3F, 3F9, and 3G and data not shown). Neither untagged (detected with Sano antiserum) nor HAtagged Sano could be detected in nuclei at any stage. Taken altogether, these experiments suggest that Sano is an apically enriched cytoplasmic protein that may also sometimes localize to nuclei, a localization similar to that reported for the mammalian Themis/Gasp protein [55][56][57][58][59][60].

sano Affects Tracheal Tube Length
Three independent loss-of-function knock-out sano alleles, sano KO1 , sano KO2 and sano KO3 , were generated by homologous recombination

Author Summary
Tubular organ formation is a ubiquitous process required to sustain life in multicellular organisms. In this study, we focused on the tracheal system of the fruit fly, Drosophila melanogaster, and identified Serrano (Sano) as a novel protein expressed in several embryonic tubular organs, including trachea. sano loss results in over-elongated trachea, whereas Sano overexpression causes shortened trachea, suggesting that sano is required for proper tracheal tube length. Interestingly, Sano overexpression results in typical planar cell polarity (PCP) defects in many adult tissues and pupal wing cells. The PCP pathway is highly conserved from flies to mammals and it has been known to control cell polarity within the plane of epithelial tissues. Importantly, we found that Sano binds Dishevelled (Dsh), a key PCP regulator, and loss or ectopic expression of many known PCP proteins in the trachea give rise to similar defects observed with loss or gain of sano, suggesting a new role for the PCP genes in tube length control. Interestingly, the changes in tube length and PCP defects in the wing were linked to changes in apical domain size, suggesting that Sano and the PCP components affect either membrane recycling and/or the linkage of the membrane to the cytoskeleton. [65]. PCR analysis confirmed that exons common to all five splice forms were replaced with the mini-white + gene ( Figure S3A and S3B). sano mRNA was not detected in sano homozygotes or in embryos transheterozygous for each sano allele over a deficiency that removes sano and nearby genes, indicating that the sano alleles are null ( Figure 1N and 1O; Figure S3C, S3D, S3E, S3F, S3G, and S3H; data not shown). Each sano allele is homozygous lethal, and lethal over the sano deficiency, with death occurring during the 2 nd instar larval stage. The sano lethality was partially rescued by expression of the Sano ORF under the control of a heat-shock promoter (HS-Sano) induced during larval stages (11/45 viable adults when heat shocked at 58-70 hr AEL).
Most features of salivary gland and tracheal development appear normal in sano mutant embryos (data not shown). Interestingly, however, staining with 2A12, a marker of tracheal lumen after stage 13, revealed that the dorsal trunk (DT) in sano mutants is more elongated and convoluted than in wild type (WT; Figure 3A-3D). Measurements of DT lengths from confocal projections of 2A12 staining from lateral views of stage 16 embryos revealed that sano mutant DTs are significantly (,12%) longer than wild type ( Figure 3E). Tracheal cell numbers in the dorsal trunk of sano mutants (14.760.6, N = 5, metamere 4) were comparable to those of WT (15.260.4, N = 5, metamere 4; p.0.5, t-test), indicating that the elongated DT phenotype is not due to increased cell numbers.
Conversely, Sano overexpression using btl-Gal4 caused shortened DTs with discontinuous staining with either 2A12 or SAS ( Figure 3F and 3F9). The UAS-Sano-GFP distributions in tracheal cells revealed that cells connecting adjacent segments of the DT (fusion cells) contact each other basally, but that the tracheal lumens and apical membranes are discontinuous. Fusion cell markers including Dysfusion (Dys), a bHLH-PAS transcription factor [66], and Arf-like-3 (Arl3), a small GTPase [67], were normally expressed in the discontinuous region of the DTs, indicating that fusion cells are not transformed to another fate ( Figure 3G; data not shown). No increase in apoptosis was detected in the Sano-overexpressing trachea ( Figure S5), and tracheal cell numbers in the btl-Gal4:UAS-Sano trachea (1660.7, N = 7, metamere 4) were comparable to WT (15.260.4, N = 5, metamere 4; p.0.1, t-test), indicating that the shortened DT phenotype is not due to reduced numbers of tracheal cells. At 25uC, 100% of btl-Gal4:UAS-Sano embryos showed apical disconnection of DTs in more than one metamere, whereas neither btl-Gal4 nor UAS-Sano alone had the shortened apical DT phenotype (Table 1). Sano overexpression also caused mismigration and/or failure of other tracheal branches to connect (data not shown). btl-Gal4-driven  Sano expression in the trachea of sano null mutants rescued the elongated DT phenotypes observed in sano mutants ( Figure 3E) and alleviated the gain-of-function phenotype of shortened DTs ( Table 1), suggesting that an optimal dose of Sano is critical for proper tube length and that tube length is inversely related to Sano levels. Sano overexpression also reduced salivary gland lumenal length (88.762.0mm (WT) vs. 58.965.4mm (Sano-overexpressing glands), N = 5 for each genotype; p,0.01, t-test), suggesting that Sano has generalized effects on tube length ( Figure 3H and 3I).
Tracheal tube size is controlled neither by the number nor the overall size of the individual cells [11]. Nonetheless, mutations in several genes have been discovered that, like loss of sano, lead to tracheal tube overelongation. Most of these known genes either regulate chitin synthesis or encode components of the septate junction, an invertebrate structure that has trans-epithelial barrier functions analogous to the vertebrate tight junction [16,[18][19][20][21][22]68]. To test whether sano function is linked to either category of known genes affecting tube length, we analyzed luminal chitin using a fluorescent chitin binding protein (CBP) and a fluorescent chitin binding lectin (Wheat Germ Agglutinin; WGA) [21]. Both reagents revealed that the chitin cable, an extracellular scaffold upon which the tracheal branches elongate, is normal in sano mutants ( Figure S4A and S4B; data not shown). vermiform (verm) encodes an apically-secreted chitin-binding protein with predicted polysaccharide deacetylase activity [12,18,22]. Verm staining in sano mutant trachea was indistinguishable from WT ( Figure S4C and S4D). We conclude that tracheal length defects in sano mutants are not a consequence of detectable alterations in chitin biogenesis.
Septate junction proteins, including Coracle (Cor), Neurexin IV (NrxIV), and Fasciclin 3 (Fas3), localized normally to the basolateral domain of sano mutant tracheal cells, suggesting that septate junctions are intact ( Figure S4E and S4F; data not shown). A 10 kDa dextran dye exclusion assay indicated that barrier function of septate junctions is intact in sano mutants ( Figure S4G, S4H, and S4I). Thus, neither septate junction function nor chitin cable assembly is disrupted in sano mutants, suggesting another mechanism for the elongated tracheal phenotype.

Sano as a PCP Regulator
sano is dynamically expressed in larval imaginal discs, structures that give rise to much of the adult during metamorphosis ( Figure S6). Overexpression of Sano using several imaginal disc-specific Gal4 drivers caused planar polarity defects. For example, in the wild-type adult thorax, bristles point posteriorly, whereas in Sano-overexpressing adult thoraces, the bristles displayed altered orientations ( Figure 4A and 4B). In the WT wing, each cell produces a single distally-oriented, actin-rich protrusion (a trichome, a.k.a. a ''hair''). All Sano-overexpressing wing cells exhibited swirling hair patterns ( Figure 4C and 4D). Sano overexpression in the eye caused ommatidial polarity defects, including misoriented and symmetrical photoreceptor phenotypes, as well as abnormal photoreceptor numbers (Figure 4E and 4F; data not shown), with about 14.5% of the ommatidia showing defects (137/916, N = 5). Polarity defects observed with Sano overexpression are similar to those observed when PCP genes are mutated or overexpressed [33,36,37,43,[69][70][71][72][73], suggesting that Sano perturbs PCP. Next we examined Sanooverexpressing pupal wing cells. Phalloidin staining of actin-rich prehairs at 32 hours after puparium formation (APF) revealed that hair formation is delayed in Sano-overexpressing cells ( Figure 5A), as observed in dsh mutant clones or in dgo pk double mutant clones [74]. Phalloidin staining of the slightly older pupal wings (at 33-34 hours APF) revealed Sano-overexpressing cells with prehairs the same size as surrounding wild-type hairs but with altered polarity ( Figure 5B). Sano overexpression sometimes produced multiple wing hairs, another PCP phenotype ( Figure S7B). Wild-type hairs near some sano-overexpressing clones exhibited non-cell-autonomous polarity defects ( Figure 5B, arrows; Figure S7C and S7D), distinct from those near fz or stbm/Vang mutant clones; wild-type cells proximal to fz clones or distal to stbm/Vang clones have reversed hair polarity [34,35,75], which was not observed with Sano overexpression. All Sano overexpression clones that produced nonautonomous phenotypes mapped either between veins 3 and 4, distal to the anterior crossvein, or between veins 4 and 5, distal to the posterior crossvein ( Figure S7A; N.100 clones examined), both regions of which are sensitive to PCP alteration [49].
Core PCP proteins are asymmetrically localized in pupal wing cells during prehair formation and show typical ''zigzag'' localization patterns on the apical surfaces of the pupal wings [33,36,37,39,40,42]. When a PCP gene is mutated or overexpressed, other PCP proteins are typically mislocalized. Sano overexpression in pupal wings through either ptc-Gal4-driven expression or in Sano-overexpressing clones resulted in the mislocalization of all PCP proteins examined. Fmi, normally localized to both the proximal and distal sides of wing cells during prehair formation, was observed around the entire perimeter ( Figure 5C and 5D). A similar mislocalization was observed with Stbm and Pk ( Figure S8A; data not shown). Fz and Dsh, which normally localize to the distal side of the apical surface, exhibited reduced apical membrane distribution with Sano overexpression ( Figure 5E, Figure S8B).
sano loss-of-function mutant clones in cells giving rise to adult tissues such as thorax and wing did not result in PCP phenotypes (data not shown). Similarly, actin prehairs of sano mutant clones in pupal wing cells always pointed distally as in WT ( Figure S9A and S9B). Since mutations of some PCP genes, such as ft, show polarity defects only in very large clones [46], we induced sano mutant clones at earlier time points to generate a range of sizes of clones missing sano function. Even very large clones did not exhibit PCP defects (data not shown). However, although it was rare, when we induced clones relatively early (36-48 hours after egg laying (AEL), we obtained only twin spots (,5%, N = ,70), suggesting that the sano mutant cells either died or were eliminated from the wing epithelium ( Figure S9C and S9D). On the other hand, sano null eye clones had defects characteristic of loss of known PCP genes, including misoriented ommatidia and loss of asymmetry ( Figure 4G; Table 2). sano null eye clones also often had abnormal numbers of photoreceptors ( Figure 4G; Table 2). In 3 rd instar eye discs, the expression of BarH1, a marker for the R1 and R6 photoreceptors [76], showed ommatidial misrotation, consistent with the adult phenotype ( Figure 4H and 4I), and the expression of md0.5-lacZ, a marker for the R4 photoreceptor, was absent or significantly reduced in sano null clones, consistent with a cell fate change of R4 to R3, which has been observed with some PCP mutants, including fz and dsh ( Figure 4J and 4K; [77]). Our data suggest that although sano overexpression disrupts PCP signaling in multiple tissues, loss of sano results in a range of defects that are limited to fewer tissues.

PCP Mutants Have Elongated DTs
To determine if sano affects tube length by altering PCP signaling, we asked if other PCP mutants have tracheal length  defects, including null mutants of the core PCP genes fz, dsh, fmi, dgo, stbm, and pk, the ft/ds group of PCP regulator genes fj, ft, and ds, and the PCP downstream effectors rhoA, Drok, zip and sqh. For dsh, a key hub in canonical Wingless (Wg)/Wnt signaling and in Fz-dependent PCP signaling, we used the dsh 1 allele, which is defective for only its PCP function [69,70]. Interestingly, many PCP mutants had tracheal length defects, exhibiting similar elongated DT phenotypes as loss of sano ( Figure 6A-6F and 6I). Among the core PCP genes, fz, dsh and fmi had elongated DTs, whereas dgo, pk and stbm mutant embryos had normal DTs. Among the ft/ds PCP regulator group, fj and ds had elongated DTs. Among the PCP downstream effectors, rhoA and zip mutant embryos showed elongated DTs, revealing a potential role for the cytoskeleton in tracheal elongation. Drok mutant embryos also   have convoluted trachea, but overall tracheal length was comparable to WT. ft and sqh mutant embryos had shorter DTs than WT, but the DTs were contiguous ( Figure 6I). Overexpression of Dsh or a constitutively-active form of RhoA in the trachea caused shortened DT defects with discontinuities similar to sano overexpression ( Figure 6G and 6G9; [78]), further implicating this pathway in apical cell surface elongation. Sano and Dsh are both cytoplasmic proteins, and Sano binds Dsh in yeast two-hybrid assays and co-immunoprecipitation (co-IP) ( Figure 6J and 6K), providing a physical link between Sano and PCP proteins that is consistent with genetic interactions between dsh and sano; double mutants of sano and dsh 1 have elongation defects similar to those of sano or dsh 1 alone, suggesting that Sano and Dsh act in a common pathway ( Figure 6I). Moreover, reduction of PCP function of Dsh (dsh 1 /+) suppressed the Sano overexpression phenotype in the thorax, a finding also consistent with Sano acting through Dsh ( Figure S10). The apical enrichment of Dsh in the late embryonic trachea and Fmi localization to the adherens junctions ( Figure  S11) is consistent with PCP proteins acting at the apical membrane. These data suggest that Sano affects tube length by impinging on Dsh activity, likely through its role in PCP signaling. Also consistent with this model is our finding that sano;ft double mutant trachea have DT lengths that are intermediate between those of ft and sano mutants alone ( Figure 6I).

Sano Overexpression Results in Smaller Apical Domains
Since wing epithelial cells become hexagonally packed prior to PCP proteins regulating hair formation [79], we examined cell shape in Sano-overexpressing wing cells. As observed with other PCP mutants, Sano-overexpressing cells often assume a pentagonal shape instead of the typical hexagonal shape ( Figure 7A; [79]). Sano overexpressing cells also have smaller apical domains than surrounding wild-type cells ( Figure 7A; here, we define the apical domain as the area circumscribed by the zonula adherens, where E-Cad localizes). Using E-Cad staining, we measured the apical domain perimeters from several examples of single Sanooverexpressing cells and found a 29-41% decrease in the perimeters of the Sano-overexpressing cells compared to their wild-type neighbors ( Figure 7A and 7B).
A decrease in apical domain size with Sano overexpression was also seen in adult tissues. For example, ptc-Gal4-driven Sano expression at the anterior-posterior wing margin resulted in a decreased distance between wing veins L3 and L4 compared to wildtype wings (Figure 7C and 7D; 178.2962.10 pixels vs. 210.9164.58 pixels, N = 3 for each genotype; p,0.005, t-test). Wing cell numbers in the area demarcated by veins L3, L4, the anterior crossvein, and an imaginary line starting from the tip of posterior crossvein and perpendicular to L3, did not reveal a significant difference in cell number between WT and ptc-Gal4:UAS-Sano wings (354.3616.9 vs. 360617.3, n = 3 for each genotype; p.0.5, t-test), indicating that the decrease of the adult wing size is due to a decrease in apical domain size. Likewise, global wing expression of Sano, using the MS1096-Gal4 [80], resulted in a decrease in overall wing size ( Figure 7E). To ask if sano affects tracheal tube length through changes in apical domain size, we examined E-Cadherin staining of WT, sano mutant, and Sano-overexpressing trachea. We also examined E-Cadherin staining in PCP mutants with altered tracheal tube length. Although it was difficult to ascertain differences in apical domain size of individual cells between WT and sano mutant trachea, which are expected to be at most ,12% different, we observed a marked decrease in apical domain size in the Sano-overexpressing tracheal cells ( Figure 7G). A similar decrease in apical domain size was observed in Dsh-overexpressing tracheal cells (data not shown). Moreover, the tracheal cells of rhoA, one of the PCP mutants, had larger apical domains than WT ( Figure 7H), indicating that the changes in tube length observed with Sano and other PCP genes are due to altered cell geometry and not altered cell arrangement. In escargot (esg) mutant trachea, where infrequent DT breaks occur, apical domain size was comparable to WT, suggesting that the smaller apical domain size observed with sano and dsh overexpression is not a due to a failure of adjacent DT segments to fuse ( Figure 7I and 7J).

Discussion
Here, we report the discovery of Sano, a novel cytoplasmic protein enriched in the apical domains of developing tubular organs and other epithelia. Loss of sano results in overelongated tracheal tubes, whereas increased Sano shortens tracheal tubes, frequently leading to failures in dorsal trunk fusion. In adult tissues, Sano overexpression leads to characteristic PCP defects including misorientation of hairs on the adult wing, mislocalization of core PCP proteins in pupal wing cells, misorientation of thoracic bristles and loss of asymmetry in the rhabdomeres of the ommatidia. Based on clonal analysis, loss of sano does not cause PCP defects in the wing but is required for normal ommatidial organization. Further support for Sano functioning as a PCP regulator is provided by our finding that mutations in both global and core PCP genes, as well as in the PCP downstream effectors, result in tracheal length defects similar to those seen with loss of sano. Based on comparisons of Sano-overexpressing and wild-type cells, Sano limits apical membrane domain size, suggesting that Sano and known PCP regulators control the linkage of the subapical cytoskeleton to the apical membrane and/or modulate apical membrane trafficking by regulating levels of endocytosis and exocytosis.

Identification of a Novel Protein Affecting PCP
Since the discovery that the Fz pathway controls PCP, many additional PCP components have been identified, including core factors, several PCP regulators, and general and tissue-specific downstream effectors [24,25,27]. Sano overexpression causes PCP defects in adult epithelial tissues as well as mislocalization of core PCP proteins. In wing cells, sano null cells appeared normal although we very occasionally obtained twin spot-only clones, suggesting a role for Sano in cell survival or in epithelial maintenance. It is unclear whether this function is related to PCP. On the other hand, sano loss in the eye gave rise to a range of defects, some of which are typical of PCP mutants, including loss of R4 cell specification, ommatidial misorientation, and loss of equatorial asymmetry.
The direct physical interaction between Dsh and Sano ( Figure 6J and 6K) provides potential mechanistic insight into Sano function. The interaction between Dsh and Sano appears quite different from that between Dsh and Naked cuticle (Nkd), a Wingless (Wg) antagonist that also gives rise to PCP defects when overexpressed. Dsh participates both in canonical Wg/Wnt signaling and in Fz-dependent PCP signaling [29][30][31]. Like Sano, Nkd directly binds Dsh, and overproduced Nkd causes polarity defects and limits Wg signaling activity presumably by sequestering, degrading and/or modifying Dsh and thus blocking its participation in PCP [81]. Unlike Nkd, however, Sano overexpression does not cause defects typical of those seen when canonical Wg signaling is blocked. Moreover, whereas Nkd overexpression blocks Dsh activity, our studies of tube length control suggest that Dsh and Sano act in the same direction: gain or loss of Dsh mimics the gain or loss of Sano in the trachea. Similarly, dsh sano double mutants have the same tracheal length defects as each single mutant ( Figure 6I). Likewise, overexpression of either Dsh or Sano in the eye using sev-Gal4 causes similar changes in ommatidial polarity and rotation [82]). Finally, we showed that reduced dsh function suppresses the Sano overexpression PCP phenotypes in the thorax ( Figure S10). Overall, the interaction and genetic data suggest that Sano and Dsh act together in a common pathway.

A Role for PCP Signaling in Tube Size Control
It is intriguing that loss-of-function mutations in many, albeit not all, PCP genes result in similar tube elongation defects observed with loss of sano ( Figure 6). PCP signaling can provide directional cues at the single cell level, such as directions on where to place the single hair within a Drosophila wing cell, or at the level of cell groups, such as controlling the organization of mechanosensory bristles in the Drosophila thorax and arrangement of photoreceptors in the Drosophila eye. PCP signaling also controls the behavior of cell populations undergoing extensive rearrangements, such as the dynamic morphogenetic changes that occur during body axis elongation in Drosophila and vertebrates and in ovarian border cell migration [23,27,83]. A recent study has implicated mammalian Fat4, a vertebrate homologue of the Drosophila global PCP protein Fat, in promoting renal tubule elongation through its effects on oriented cell divisions [51]. In those studies, loss of Fat4 led to shorter renal tubules, a defect exacerbated by simultaneous loss of one copy of Vangl2, a vertebrate homologue of the core PCP protein Stbm/Vang. Consistent with this finding, our studies reveal that, in the trachea, mutations in the proteins that negatively regulate Fat (Ds and Fj) and the Stbm/Vang complex (Fz, Dsh) have the opposite defect: longer tubes. In the case of the trachea, a tissue whose final cell divisions occur much earlier in development than when Sano affects tube length, the effects of the PCP pathway are on cell shape rather than on the orientation of cell division. Whether the subcellular mechanisms by which PCP genes regulate oriented cell divisions in vertebrates and apical membrane elongation in flies are similar or distinct is not clear, but the parallels in the two systems provide evidence for evolutionarily conserved functions for PCP genes in tubular architecture.
The finding of a role for PCP genes in tube length control raises two crucial questions. (1) Are PCP proteins asymmetrically localized in tubular epithelia in the same way they are in wing, eye and border cells [23,26,27]? (2) How do PCP genes regulate tube length? We examined the subcellular localization of Dsh and Fmi in the tracheal cells, where Dsh localizes mainly in the cytoplasm and is enriched at the apical domain at later stages, and Fmi localizes to the adherens junctions ( Figure S11). Unfortunately due to the irregular shape of tracheal cells and the three-dimensional structure of the tracheal tube, we could not determine with adequate resolution whether the PCP proteins are asymmetrically distributed. However, our data provides new insight into how PCP affects tracheal tube size. In the trachea, loss of sano or PCP function resulted in tubes that were 7-15% longer than WT based on apical domain measurements (2A12 staining). Since the sano trachea have the same number of cells as wild-type, each tracheal cell, on average, must have an apical domain that is approximately 12% longer than wild-type. Although an accurate measurement of apical dimensions in the trachea could not be obtained due to the shape and curvature of the tube, in rhoA mutant trachea, where the elongated DT defects were most obvious, the apical domains of the DT cells were consistently larger than WT ( Figure 7H). Similarly, E-Cad staining in dsh and ds mutant tracheal cells revealed slightly larger apical domains (data not shown). Importantly, overexpression of Sano and Dsh resulted in tubes that were so much shorter than WT that the individual segments were often too short to anastomose. Examination of E-Cad staining in Sano-or Dshoverexpressing trachea revealed markedly smaller apical domains in these cells, consistent with the decreased apical domain size observed in wing cells overexpressing Sano ( Figure 7G; data not shown). Thus, PCP components appear to control overall tube length by limiting the size of the apical domain. This activity could be mediated by increased linkage of the plasma membrane to the underlying cytoskeleton and/or by direct effects on plasma membrane growth by modulating relative levels of exoctyosis and endocytosis. In support of a link between PCP signaling and regulated vesicle trafficking, Rab11/Sec5-dependent recycling of E-cadherin has been implicated in junctional remodeling during hexagonal packing of wing cells, wherein the polarized recruitment of Sec5 is through the PCP protein Fmi [79]. Importantly, tracheal tube elongation has also been linked to regulated vesicle trafficking through Rab11 [64,84] and through core components of the secretory machinery [85,86].
There are two potential inconsistencies with a model that Sano functions to control tube length through its effects on PCP signaling: (1) Loss of sano does not give rise to overt PCP defects in all adult tissues and (2) not all of the components of the PCP signaling pathway disrupt tube length when their function is missing. Indeed, loss of ft, which is expressed early in the tracheal primordia, appears to have effects opposite those of loss of sano on tube length. Although the tissue-specificity of sano's role in PCP could reflect functional redundancy, it is also possible that sano and the large subset of the PCP genes that do have tracheal defects may function through novel, non-canonical, pathways to control tube length. In either case, it will be exciting to unravel the details of Sano's interactions with the cellular machinery to control apical domain size.

Immunohistochemistry
Embryo fixation and staining were performed as described [87] except for the a-E-Cad staining, for which embryos were fixed in 4% paraformaldehyde in PBS and devitellinized with ethanol. 3 rd instar rp395 larval discs were dissected and fixed in 2% paraformaldehyde in PBS for 20 minutes, incubated with primary antibody overnight (4uC) and then with the appropriate secondary antibody for two hours (RT).

Whole-Mount in situ Hybridization on Embryos and Imaginal Discs
In situ hybridizations were performed as described by [88]. The pPB3 cDNA, isolated by screening a cDNA library provided by L. Kauvar, was used to generate an anti-sense digoxigenin-labeled sano RNA probe.

Identification of Sano
sano was identified in a P-element expression screen in Corey Goodman's laboratory. We obtained the rp395 line because of its salivary gland and tracheal expression. Sano was independently identified in an EP screen for genes that when misexpressed alter the eye phenotype of Dsh+Nkd overexpression (S. Silva, G. Celik, C.-C. C. and K.A.W., unpublished).

sano Null Alleles
Null sano mutants were generated by homologous recombination [65]. Genomic fragments upstream and downstream of the sano ORF were amplified by PCR and cloned into pW25, which carries white + , the recognition site for I-SceI endonuclease, and FRT sites. The construct was injected into embryos by Rainbow Transgenic Flies, Inc. Transformants were crossed to flies carrying hs-I-SceI and hs-Flp and progeny were heat shocked (37uC) for 1 hour 48-72 hours AEL.
HS-Sano. UAS-Sano-GFP and UAS-Sano-HA were created using the Drosophila Gateway Vector system (Carnegie Institution).

Sano Antibody
A PCR fragment spanning the sano ORF was amplified and cloned into the pProEx expression vector (Life Technologies, Inc.). The construct was transformed into BL21-DE3 cells, from which Sano inclusion body preparations were made. Recombinant fulllength protein was further purified from an SDS-polyacrylamide gel slice as described [90]. Rat polyclonal antibodies were generated by Covance, Inc. and used at a dilution of 1:50.

Phalloidin/Antibody Staining of Pupal Wings
Pupae were fixed 32-34 hours APF in 4% paraformaldehyde in PBS overnight (4uC). Pupal wings were dissected and washed several times in 0.5% PBST (0.5% Triton X-100 in PBS) and incubated with phalloidin-568 (Molecular Probes, 1:1000) for one hour (ice). For antibody staining, pupae were fixed at 28 hours APF in 4% paraformaldehyde in PBS for one hour (4uC). Pupal wings were dissected and washed in 0.1% PBST. Wings were incubated in primary antibodies overnight (4uC) and then in secondary antibodies for two hours (ice).

Adult Eye Sections
sevenless-Gal4 (sev-Gal4) was used to overexpress Sano in the eye. sano mutant eye clones were generated using ey-FLP. sano mutant cells were w + /w + , which can easily be distinguished from w + /w 2 heterozygous cells and from w 2 /w 2 twin spots in whole eyes. Since it is difficult to distinguish w + /w + versus w + /w 2 in thin sections, however, we chose only eyes with large mutant clones and adjacent w 2 /w 2 twin spots for sectioning. Fixation and semithin sectioning of the adult eyes were slightly modified from [93]. Sections from at least five independent eyes were analyzed for each genotype.

Dorsal Trunk (DT) Length and Pupal Wing Cell Perimeter Measurements
Embryos were stained with 2A12 and projections from lateral views of confocal sections of the DT lumen of st. 16 embryos (at the four equal-compartment midgut stage) were traced from the starting point of metamere one to the point where the last transverse connective (TC) meets the DT in metamere nine using the Image J program (NIH). At least ten samples were measured and normalized to the length of the embryo for each genotype. An average length from three independent measurements of each sample was calculated.
Pupal wings were stained for E-Cadherin and the perimeter of pupal wing cells overexpressing Sano-GFP and of their wild-type neighbors were measured by Image J.
Co-Immunoprecipitation and Western Blotting da-Gal4/UAS-Sano-HA; dsh-GFP/+ embryos were used for co-IP, and da-Gal4/UAS-Sano-HA and dsh-GFP embryos were used as controls. The embryos were collected and homogenized in radioimmunoprecipitation (RIPA) buffer (Cell Signaling) including protease inhibitor cocktail (Roche). A small aliquot of the cleared supernatant was used for the Western to check the protein input with a-GFP and a-HA. Dynabeads Protein G (Invitrogen) was incubated with mouse a-HA (Roche) or rabbit a-GFP (Molecular Probes) for 10 minutes at RT. After several washes with PBTw (0.01% Tween-20 in 16 PBS), the remaining supernatant was incubated with antibody-bound Dynabeads Protein G for 20 minutes at RT. The beads were washed three times with RIPA buffer, and boiled in SDS sample buffer to elute the proteins. Bound antigen was detected by enhanced chemiluminescence (GE Healthcare). The antibodies for Western blotting were used at the following concentrations: rat a-HA (Roche, 1:2,000), mouse a-GFP (Roche, 1:2,000). Co-Ips were repeated three times with the same results.

Northern Hybridization
The developmental Northern blot was prepared as described [94] and hybridized with a Bgl II/Not I fragment from pPB3 cDNA labeled by random priming.

Dextran Injections
Fluorescence-labelled 10kDa dextran (Molecular Probes) injections were performed as described [22,95], using wild-type embryos as a negative control and the mega mutant as a positive control.