1493RESEARCH ARTICLE

INTRODUCTIONEstablishing and maintaining epithelial integrity is essential forembryonic development, organogenesis and tissue remodeling. Thekey characteristic of epithelial cells is asymmetrical specification ofmembrane domains marked by domain-specific proteins (Knust andBossinger, 2002; Nelson, 2003; Tepass, 2002). The epithelialmorphogenic mechanism, although with some variations in differentepithelial tissues, is highly conserved from worm to mammal. Thecrucial initial step in establishing epithelial polarity is thespecification of the apical domain, which is defined by the functionof a complex containing atypical PKC (aPKC), Bazooka (Baz;mammalian and worm PAR-3) and PAR-6 (for a review, see Suzukiand Ohno, 2006). The PAR complex is initially recruited byactivated Cdc42 to the apical domain. The three proteins wereoriginally thought to function as a complex; however, recentevidence indicates that Baz might be required first to recruit theaPKC–PAR-6 complex to the subapical domain juxtaposed to thefuture adherens junction (AJ) (Harris and Peifer, 2007). The PARcomplex is required for the localization of another apical complexcontaining Crumbs (Crb), Stardust (Std; mammalian Pals1) andDiscs lost (Dlt; mammalian Patj). The PAR- and Crb-containingcomplexes occupy the apical-most region of the lateral membrane,just apical to the AJs.

The apical complexes in turn restrict the localization of a thirdcomplex comprising Scribble (Scrib; mammalian Scribble/Vartul),Discs large (Dlg) and Lethal(2) giant larvae (Lgl) (Betschinger et al.,

2005; Bilder et al., 2003) to the basolateral domain, while Lgl alsoantagonizes the apical components and prevents their spreading tothe basolateral side (Yamanaka et al., 2003). The antagonistic actionof apical and basolateral complexes helps define the apicolateral locieventually occupied by AJs. It is not yet completely clear how theinitial localization of the apical complex is achieved.

The VHL tumor-suppressor gene mutations are the genetic causeof the familial VHL disease. Germline mutations in VHL predisposethe patients to several benign and malignant tumors, including renalcell carcinoma (RCC, kidney cancer), hemangioblastoma(overgrowth of blood vessels in the retina and central nervoussystem) and pheochromocytoma (tumors in the adrenal glands).VHL protein has been shown to function as an E3 ubiquitin ligase.Among its best-documented targets is the alpha subunit of thehypoxia-inducible factor (HIF-). Therefore, the canonical tumor-suppressor function of VHL is modulation of the normal oxygen-sensing mechanism that regulates angiogenic response andmetabolic switch to glycolysis (Kaelin, 2008). However, how thisfunction correlates with the origin of epithelial tumors such as RCCis unclear, although it is thought that HIF-independent mechanismsmight be involved (Frew and Krek, 2007).

VHL is evolutionarily conserved. In Drosophila, the VHL genehas been implicated in tracheal tubule development and HIF-regulation in the embryos based on biochemical and RNAinterference-mediated phenotypic studies (Adryan et al., 2000;Arquier et al., 2006; Aso et al., 2000; Mortimer and Moberg, 2009).In this report, we generated the first genomic Drosophila VHLmutant and examined the function of VHL in epithelialmorphogenesis using a model epithelium – the follicle cells in theegg chamber. We show that VHL regulates the proper localizationand stability of aPKC in the follicle cells and that this function is, atleast in part, mediated by the action of VHL on microtubule (MT)stability. Without VHL function, MTs and aPKC are destabilized,resulting in epithelial defects. These results establish adevelopmental function of the VHL gene that is relevant to its tumor-suppressor activity.

Development 137, 1493-1503 (2010) doi:10.1242/dev.042804© 2010. Published by The Company of Biologists Ltd

1Dipartimento di Biologia Evoluzionistica Sperimentale, Università di Bologna, ViaSelmi 3, 40126 Bologna, Italy. 2Department of Pathology and Laboratory Medicine,and Hollings Cancer Center, Medical University of South Carolina, Charleston, SC29425, USA.

*Present address: Boston University School of Medicine, Hem/Onc Section,650 Albany Street, EBRC440, Boston, MA 02118, USA†Authors for correspondence (giuseppe.gargiulo@unibo.it; tienh@bu.edu)

Accepted 23 February 2010

SUMMARYMutations in the human von Hippel-Lindau (VHL) genes are the cause of VHL disease, which displays multiple benign and malignanttumors. The VHL gene has been shown to regulate angiogenic potential and glycolic metabolism via its E3 ubiquitin ligase functionagainst the alpha subunit of hypoxia-inducible factor (HIF). However, many other HIF-independent functions of VHL have beenidentified and recent evidence indicates that the canonical function cannot fully explain the VHL mutant cell phenotypes. Many ofthese functions have not been verified in genetically tractable systems. Using an established follicular epithelial model inDrosophila, we show that the Drosophila VHL gene is involved in epithelial morphogenesis via stabilizing microtubule bundles andaPKC. Microtubule defects in VHL mutants lead to mislocalization of aPKC and subsequent loss of epithelial integrity. Destabilizingmicrotubules in ex vivo culture of wild-type egg chambers can also result in aPKC mislocalization and epithelial defects. Importantly,paclitaxel-induced stabilization of microtubules can rescue the aPKC localization phenotype in Drosophila VHL mutant follicle cells.The results establish a developmental function of the VHL gene that is relevant to its tumor-suppressor activity.

KEY WORDS: Von Hippel-Lindau tumor-suppressor gene, Drosophila, Epithelial morphogenesis

Drosophila VHL tumor-suppressor gene regulates epithelialmorphogenesis by promoting microtubule and aPKC stabilitySerena Duchi1, Luca Fagnocchi1, Valeria Cavaliere1, Anita Hsouna2, Giuseppe Gargiulo1,† and Tien Hsu2,*,†

DEVELO

PMENT

1494

MATERIALS AND METHODSDrosophila strains and geneticsy1, w67c23 was used as the wild-type stock in this study. The Drosophila VHL1

mutant was generated by replacing the wild-type copy, via homologousrecombination (Rong and Golic, 2000), with a deletion that removes 81codons encompassing the first two in-frame AUGs. The mutant stock ismaintained with the balancer chromosome CyO, Dfd-YFP that carries a YFPreporter gene directed from the Dfd promoter (expressed in mandibles). TheDf(2R)en-A is a chromosomal deletion that uncovers the region 47D6-48F8on the second chromosome, including the genomic VHL locus at 47E5-47E6. Follicle cell clones mutant for VHL were generated with site-directedmitotic recombination using the Flp/FRT system (Xu and Rubin, 1993). Themutant VHL alleles were crossed onto chromosome 2R carrying the yeastrecombination site FRTG13 at the base of the chromosome arm (at position42B). The mitotic mutant clone was induced by crossing the VHL, FRTG13

fly with a transgenic strain carrying the wild-type FRTG13 chromosome (witha ubiquitin promoter-driven GFP reporter gene, ubi-GFP) and the UAS-Flprecombinase transgene under control of the T155-Gal4 driver. The genotypeof the VHL mutant mosaic female is (using VHL1 as an example): w67c23;P{FRTwhs}G13, VHL1/P{FRTwhs}G13, P{ubi-GFP}; P{T155-Gal4},P{UAS-Flp}/+. T155-Gal4 directs expression in all follicle cells throughoutoogenesis. FRT42D, aPKCk06403/CyO is a strong loss-of-function allele andwas a gift of W. M. Deng (Florida State University, USA) (Tian and Deng,2008). The VHL full-length cDNA (Adryan et al., 2000) and the VHLYH

variant (single amino acid substitution mutation of tyrosine to histidine atposition 51, equivalent to human Y98) were cloned in the UAS-basedpGateway vector tagged with three HA epitopes at the 3� end. Theseexpression vectors were used to transform the y w flies. Cy2-GAL4 waskindly provided by T. Schüpbach (Princeton University, USA), and used forUAS-driven expression in follicle cells. btl-GAL4 was used to drive theexpression of VHL in the tracheal and glial cells. Unless specified, all flystocks are from the Bloomington Stock Center.

Ex vivo culture of egg chambersEx vivo culture of egg chambers has been previously described (Prasad etal., 2007). Briefly, ovaries were dissected at room temperature in PBS andcultured at 25°C in oxygenated Schneider’s Complete Medium Cocktail[Schneider’s Medium, heat-inactivated 15% fetal bovine serum (FBS), 0.6�penicillin/streptomycin mix (Gibco/Invitrogen), 0.2 mg/ml fresh insulin, pH6.95-7.0]. Paclitaxel and nocodazole (Sigma) were dissolved in DMSO andused at indicated concentrations. After a 5-hour incubation at 25°C, theovaries were gently collected, washed and processed for immunostaining.

ImmunohistochemistryOvaries were dissected at room temperature in PBS, pH 7.5, and fixed in 4%paraformaldehyde or as specified (methanol was used for Crb staining).Subsequent washes and incubations were performed in PBT (1� PBScontaining 0.1% Triton X-100). Primary antibody incubation was performedat 4°C overnight and antibodies were diluted in PBT containing 3% BSA.Mouse monoclonal primary antibodies used were: -tubulin [1:25; E7,Developmental Studies Hybridoma Bank (DSHB)], Arm (1:10; N2A71,DHSB), Crb (1:5; Cq4, DSHB) and GFP (1:1000; sc-9996, Santa Cruz).Primary rabbit polyclonal antibodies used were: cleaved caspase 3 (1:50;Cell Signaling), phospho-histone H3 (1:200; Upstate), aPKC (1:200; C20,Santa Cruz), HA (sc-805, Santa Cruz), Lgl (1:500) (Grifoni et al., 2004) andBazooka (1:2000) (Wodarz et al., 1999). Rabbit polyclonal anti-VHLantibody was generated against full-length bacterially expressed His-taggedprotein. This antibody is not suitable for wholemount immunostaining of theegg chambers. AlexaFluor 546-phalloidin and ToPro-3 are fromInvitrogen/Molecular Probes. AlexaFluor 488, 546 or 633-conjugatedsecondary antibodies (Invitrogen/Molecular Probes) were used at 1:200dilution and incubation performed for 2 hours at room temperature.Propidium Iodide and DAPI are from Sigma. Egg chambers were mountedin Fluoromount G (Electron Microscopy Sciences) and analyzed with theTCS SL Leica confocal system or with an Olympus IX70 microscopeequipped with the FluoView 300 confocal capability. Digital images wereprocessed using the Photoshop software without biased manipulations.

ImmunoprecipitationOvaries from 50-150 females were collected in 1� PBS and rinsed twicewith 1� PBS. Then 180 l protein extraction buffer (50 mM HEPES pH 7.6,100 mM KCl, 1 mM EGTA, 1 mM MgCl2, 1% Triton-X 100, 10% glycerol)and 20 l of 10� protein inhibitor cocktail (Roche) were added. The ovarieswere homogenized manually on ice using an Eppendorf tissue grinder, thensonicated on ice four times with 2-second pulses and 2-second intervals andcentrifuged at 6000 g for 20 minutes at 4°C. The supernatant was stored at–20°C in 50 l aliquots and a small portion was used to perform a proteinquantification assay. 500 g of protein extracts were used for eachimmunoprecipitation in a total volume of 500 l of protein extraction buffer.The extracts were pre-adsorbed with equilibrated 50 l of 50% protein A/Gagarose bead slurry (Pierce). The extracts were then incubated with ~3 gof anti--tubulin or anti-GFP (either mouse or rabbit) before 50 l ofequilibrated protein A/G beads were added. Purified IgG was used ascontrol. The precipitated protein complexes were then subjected to westernblotting. Antibodies used in western blots were: mouse anti--tubulin(1:250; see above), mouse anti--actin (1:2000; Sigma), rabbit anti-Bazooka(1:5000; see above) and rabbit anti-VHL (1:10,000; see above).

RESULTSDrosophila VHL mutantThe VHL mutation was generated using the homologousrecombination strategy (Rong et al., 2002). Homozygous VHL1

mutants are sluggish after hatching and die at the end of first instarlarval stage. Wild-type, heterozygous and homozygous first instarlarvae were hand-picked and subjected to genomic DNA PCR. Thehomozygous mutant animals show complete loss of the wild-typegene copy (Fig. 1A). The VHL1 allele, when paired with a deficiencychromosome encompassing the VHL locus (at 47E), shows the samelate first instar lethality, suggesting that VHL1 is a null mutant. Thelethal phenotype can be rescued by expressing a wild-type VHLcDNA under the control of the hsp70 promoter (data not shown).Therefore, VHL gene truncation is the only major genetic defect inthe VHL1 allele.

Epithelial defects in VHL mutant follicle cellsThe Drosophila follicle cells exhibit the typical epithelial polarityexemplified by markers such as the apical PAR complex, AJcomponents and the basolateral Lgl complex (Bilder et al., 2003;Tanentzapf et al., 2000). Establishment of the epithelial polaritybegins soon after follicle cells diverge from the somatic stem cellsand encircle the germ cells (stage 1). The epithelium reachesmaturity at mid-oogenic stages (after stage 6 at ~30 hours of eggchamber development; total developmental time ~70 hours), whenfollicle cells cease to proliferate. To examine the phenotype in thefollicular epithelium, an adult tissue, we generated mosaic mitoticmutant clones using the Flp/FRT system. Mutant clones areidentified by a lack of GFP expression. In an initial survey ofpotential egg chamber phenotypes, prominent epithelial defectswere observed at stage 10 (about 50-55 hours into egg chamberdevelopment; Fig. 1B-D). The most notable morphologicalabnormalities were the piling-up of follicle cells (Fig. 1B,C, arrows)and stretched follicular epithelium (brackets in Fig. 1C,D). The samephenotypes were observed with the VHL deficiency Df(2R)en-A(Fig. 1E,F). To quantify the phenotypes, 100 clones of various sizesat stage 9-10 were analyzed. Clone sizes were categorized based onthe number of cells in an optical cross-section. As the penetrance insingle-cell clones is variable, probably because of phenotypic rescueby the neighboring epithelial cells, we considered only clones largerthan two cells. Thirty-eight percent of clones showed stacking offollicle cells (Fig. 1B, arrows) and the more severe, multilayeringphenotypes (more than three layers of cells; Fig. 1C, arrow); 42%

RESEARCH ARTICLE Development 137 (9)

DEVELO

PMENT

showed flattening/stretching of epithelium (Fig. 1C,D, brackets);and the remaining 20% showed swelling (see below). We alsodetermined that VHL loss-of-function mutation did not affect cell

proliferation or viability. As mitotic recombination in aheterozygous progenitor cell (VHL1/ubi-GFP) generates one VHLhomozygote (no GFP) and one wild type (two copies of GFP), thenumbers of high GFP-expressing cells and GFP-negative cellsshould be equal if the mutant cells do not exhibit an altered rate ofproliferation or viability. We calculated the ratio of the cell numberwithin the VHL1 mutant clones versus the cell number within thesister wild-type clones (number of clones examined25) andobtained a mean value of 0.96 (± 0.16 s.d.). Importantly, the ratiobetween the cell number in mutant and wild-type sister clones didnot change as a function of the clone size (Fig. 1G). Furthermore,staining for mitosis and apoptosis markers phospho-histone H3 andcleaved caspase 3, respectively, also showed that VHL mutation doesnot affect cell proliferation or caspase-mediated apoptosis (see Fig.S1 in the supplementary material).

Loss of VHL in follicle cell clones affects cellpolarityMultilayering is a typical terminal phenotype for polarity defects inepithelial cells. In homozygous VHL1 clones, the aPKC protein wasmislocalized or greatly diminished from the apical membranedomain, which abuts the germ cell complex. This defect occurredearly in oogenesis in 82% (n28) of the stage 3-4 clones (~10-24hours of egg chamber development; stage 4 as an example in Fig.2A). At later stages (>stage 8; >40 hours of egg chamberdevelopment) in homozygous VHL1 clones, mislocalization ofaPKC was more pronounced, showing either cytoplasmiclocalization or absence in 96% (n27) of stage 10 clones or 83%(n29) of stage 8-9 clones (Fig. 2B,C). Interestingly, in manymutant cells [61% (n28) of stage 8-9 clones], aPKC was detectedin the perinuclear region (Fig. 2C�, sharp arrows), suggesting adefect in translocation of the protein. We next analyzed thedistribution of the aPKC partner Bazooka (Baz). Baz localizationwas largely normal in VHL mutant cells at the same stage 3-4, withonly a slight decrease in the apical domain and some spreading tothe lateral domain (50%, n28; Fig. 2D). The overall levelsdiminished significantly later in 83% (n29) of stage 9-10 clones(Fig. 2E). Perinuclear accumulation was also detected in 61%(n28) of stage 10 clones (Fig. 2E). The time difference inlocalization defects of these two proteins indicates that Bazmislocalization at later stage is probably an indirect effectdownstream of mislocalized aPKC.

The PAR complex has been shown to be a master regulator ofepithelial polarity (Suzuki and Ohno, 2006). We thereforeexamined representative markers of different apicolateralmembrane domains. The subapical region (SAR) can be visualizedby Crumbs (Crb), a transmembrane protein that interacts withaPKC and is required for the establishment of this domain (Tepasset al., 1990; Wodarz et al., 1995). It has been shown thatphosphorylation of Crb by aPKC in Drosophila embryonicblastodermal cells is necessary for Crb stabilization in the apicaldomain (Sotillos et al., 2004). In homozygous VHL1 clones at stage4, apical localization of Crb was only slightly reduced (but notabsent; Fig. 2F,F�). Co-staining of Crb and aPKC at stage 4 showedthat in the VHL1 clone, aPKC is undetectable while Crb is fadingbut still present in the apical domain (see Fig. S2A in thesupplementary material). Later at stage 8, Crb was mislocalized tothe basal domain and formed large aggregates in 89% (n28) ofstage 8-10 clones (Fig. 2G,G�, arrows).

AJs are the mature epithelial cell landmark and can be visualizedby the Drosophila -catenin encoded by the armadillo (arm) locus(Peifer, 1993; Tanentzapf et al., 2000). Arm distribution was normal

1495RESEARCH ARTICLEVHL regulates epithelial morphogenesis

Fig. 1. Drosophila VHL mutant follicle cells exhibit epithelialdefects. (A)VHL1 mutant allele contains a deletion removing the first twoin-frame AUG codons. PCR primers (black line fragments) fordistinguishing wild-type and VHL1 genomic segments are shown withexpected PCR fragment sizes. Genomic DNAs from first-instar larvae ofthe indicated genotypes were subject to PCR amplification using the VHL-specific primers indicated on the left or rp49-specific primers as control.(B-D)VHL1 mutant clones were generated as described in Materials andMethods. Relevant cell types are marked in B for reference herein. FC,follicle cells; NC, nurse cells; O, oocyte. The apical surface of the folliclecells is in contact with the germ cell complex (nurse cells and the oocyte).Mutant cells are identified by a lack of GFP expression (dotted lines orbrackets). Egg chambers were stained for DNA (ToPro-3 in B and DAPI inC,D). The mutant epithelia show stacking of cells (arrows in B), piling-up(more than three layers of cells; arrow in C) and stretching (brackets inC,D). (E,F)Mutant clones of Df(2R)en-A encompassing the VHL locuswere generated as above. Egg chambers were stained for DNA(Propidium Iodide). Stacking of epithelial cells is seen (arrows).(G)Quantification of cell numbers in mutant clones compared with thesister wild-type cell clones. Twenty-five clones are presented withincreasing sizes. There are no significant size differences between mutantand wild-type clones. Scale bars: 20m.

DEVELO

PMENT

1496

at stage 4 (Fig. 2H,H�). Co-staining of aPKC and Arm confirmedthat, in VHL mutant cells at stage 4, apical aPKC is greatlydiminished but Arm appears normal (see Fig. S2B in thesupplementary material). Arm distribution became abnormal in VHLmutant cells after stage 6 of oogenesis. As shown in Fig. 2I,I�, lossof VHL caused a pronounced alteration of the epithelial monolayer(swelling as an example) and Arm overaccumulated to the apicaland basolateral membrane domains in 89% (n27) of stage 8-10clones (Fig. 2I�, arrows). The pattern of Arm spreading is consistentwith the notion that apical and SAR polarity proteins are crucial forthe formation and restriction of AJ at the apicolateral domain of theepithelial cells.

In Drosophila follicle cells, as in other epithelial tissues,antagonistic interaction between the PAR complex and thebasolateral Lgl complex is necessary for the establishment andmaintenance of epithelial polarity (Bilder et al., 2003; Hutterer etal., 2004; Tanentzapf and Tepass, 2003). Lgl can compete withBaz/PAR-3 for binding to aPKC-PAR-6 and destabilize the PARcomplex (Yamanaka et al., 2003). Conversely, aPKCphosphorylates Lgl, resulting in its dissociation from the

membrane domain (Betschinger et al., 2005). In wild-typeepithelial cells, Lgl is located in the lateral membrane domain withenrichment toward the basal side. At stage 4 in the homozygousVHL1 clones, Lgl was slightly reduced but present on thebasolateral domain (Fig. 2J,J�), but greatly reduced from themembrane domain in 90% (n29) of stage 8-10 clones (Fig.2K,K�). Interestingly, it has been shown that a functional butmislocalized aPKC-PAR-6 complex could still phosphorylate Lglin the cytoplasm (Hutterer et al., 2004). It is therefore probable thatthe mislocalized aPKC in VHL mutant cells can phosphorylate Lgland prevent its membrane localization.

We have previously shown that RNA interference-mediatedknock-down of VHL resulted in defects in tracheal developmentduring embryogenesis (Adryan et al., 2000). Homozygous VHL‘escaper’ mutant flies (~10% of adult progenies from heterozygouscross) can be recovered by re-expression of the VHL cDNAsequence from the breathless (btl) gene promoter, which is specificfor the trachea and glial cells. As btl is not expressed in the folliclecells, we reasoned that the rescued adult VHL1 homozygous femalesshould exhibit severe epithelial defects in the follicle cells and serveas a confirmation of the VHL clonal phenotypes. We first confirmedthat overexpression of VHL from the btl promoter in wild-typefemales did not cause epithelial defects (Fig. 3A). Conversely, theVHL homozygous females rescued by btl>VHL were sterile, inwhich 70% of the ovarioles contained only degenerated eggchambers. In the rest, we observed very severe epithelial defects intheir egg chambers, including prominent expansion of theepithelium into the germ cell proper (Fig. 3B,C, arrows). In othercases, the entire epithelium became multilayered as shown in theoptical cross-section through the center of the egg chamber in Fig.3D. The invading follicle cells showed a loss of apical specificationas Arm was expressed evenly throughout the cell peripheries (Fig.3B,C, arrows). In the severely multilayered epithelium (Fig. 3D),

RESEARCH ARTICLE Development 137 (9)

Fig. 2. Altered epithelial marker expression in VHL mutant cells.Egg chambers containing VHL1 clones (no GFP) were dissected andstained for indicated epithelial markers (red). (A,A�) A stage-4 eggchamber showing significantly reduced aPKC in mutant cells (singlecells are marked by empty arrows and multi-cell clones are marked bythe dotted line) compared with the neighboring normal cells. (B,B�) Astage-10A egg chamber showing a mutant clone (dotted line) with noaPKC expression (empty arrow). (C-C�) A stage-9 egg chamber showingcytoplasmic accumulation (C�) and perinuclear accumulation (sharparrows in C�) of aPKC. Dotted lines mark the mutant clones. (D,D�) Astage-4 egg chamber. The VHL1 mutant clone (dotted line) shows somelateral spreading (arrow) but retains the proper apical localization ofBaz. (E,E�) A stage-10A egg chamber. VHL1 mutant clones (dotted lines)either lose Baz expression (empty arrows) or show perinuclearlocalization of Baz (sharp arrows in �). (F,F�) A stage-4 egg chamber.The VHL1 mutant clone (dotted line) shows slightly reduced apical Crb.(G,G�) A stage-8 egg chamber. The VHL1 mutant clone (dotted line)shows basally localized Crb and disrupted epithelium (arrows). (H,H�) Astage-4 egg chamber. The VHL1 mutant clones (dotted lines) shownormal distribution of Arm. (I,I�) A stage-9 egg chamber. The VHL1

mutant clone (dotted line) shows apical and lateral overaccumulation ofArm and swelling of the epithelium (arrows). (J,J�) A stage-4 eggchamber. The VHL1 mutant clones (dotted lines) show slightly reducedLgl (empty arrows). (K,K�) A stage-10B egg chamber. The VHL1 mutantclone (dashed line) shows greatly diminished expression of Lgl (emptyarrow). Scale bars: 20m.

DEVELO

PMENT

Arm overaccumulated in the apical domain in the inner layer facingthe germline complex and spread throughout the lateral membranedomains in the stacking layers. We also observed a collapsed nursecell complex (Fig. 3D, bracket), which, in the wild-type eggchamber, occupies the anterior half of the egg chamber at stage 10.This might be the result of germline cytoskeletal defects or the resultof structural defects in the squamous follicle cells that cover thenurse cell complex. In all egg chambers examined, aPKC wasmislocalized (cytoplasmic localization or spreading throughout thelateral membrane) and greatly reduced.

MT defectsIt has been shown that human VHL can promote MT stabilityagainst MT destabilizing agents colcemid and nocodazole(Hergovich et al., 2003; Lolkema et al., 2004), although thephysiological role of this activity has not been elucidated. Wetherefore examined whether MT organization is affected inDrosophila VHL mutant follicle cells. In many organ-associatedepithelial cells such as the follicle cells, MTs are not centrosome-anchored, but are instead organized in cortical bundles parallel to theapicobasal axis, with the plus-end localized basally (Bartolini andGundersen, 2006). In follicle cells, an apical meshwork of MTs alsoexists. As shown in Fig. 4, VHL1 mutant cells already displayedeither disorganization or loss of MTs in pre-stage 6 clones (Fig. 4A-B�), in step with loss of apical aPKC (Fig. 4A�,B�). Loss of MTsalong the lateral cortex became more obvious in later egg chambers(Fig. 4C). The MT defects were observed in all VHL1 clones. It hasbeen shown that unbundled MTs are lost during fixation withstandard immunostaining protocol (e.g. 4% paraformaldehyde) butcan be preserved and detected using more extensive fixation/cross-linking conditions (e.g. 8% paraformaldehyde) (Theurkauf, 1994).Surprisingly, even with 8% paraformaldehyde fixation, VHL mutantcells still exhibited loss of MTs (Fig. 4D,D�), suggesting that eitherthe stability of MTs in VHL1 mutant cells was severely compromisedor the plus-end polymerization became defective. In all VHL1

homozygous egg chambers from btl>VHL-rescued mutant females,MT was barely detectable (Fig. 3E,F).

VHL functions via stabilizing MTsDrosophila aPKC can control epithelial polarity duringepithelialization of embryonic blastoderm by regulating MTorganization (Harris and Peifer, 2007). In human cells, VHL hasbeen shown to interact with the PAR-3–PAR-6–aPKC complex(Schermer et al., 2006), which in turn regulates MT–cortexinteractions and coordinates growth of cortical MTs. Conversely,because MTs are a major transport route for epithelial markers(Musch, 2004; Rodriguez-Boulan et al., 2005), the disruption of MTmight indirectly affect aPKC localization. To distinguish betweenthese two possibilities, we first analyzed MT distribution in aPKCmutant clones. Homozygous aPKC mutant clones, geneticallymarked by the absence of GFP and verified by lack of aPKCantibody staining, showed disorganized MT structure but notdiminished MTs (Fig. 4E,F). Even at later stages (>stage 8) whenDrosophila VHL mutant clones showed consistent loss of MTs,aPKC mutant cells exhibited a disorganized MT network but no lossof MTs (Fig. 4G). Closer examination showed that instead oforganized cortical bundling (Fig. 4G�), the aPKC mutant cellsexhibited a dotted pattern of MT structure (Fig. 4E�,G�, emptyarrowheads), indicating a truncated MT structure. Other aspects ofthe VHL mutant phenotypes, such as piling up (Fig. 4E, arrows) andstretching or gaps in epithelium (Fig. 4F,H, brackets), wereobserved. Therefore, the aPKC mutant exhibits some, but not all, ofthe phenotypes in the VHL1 mutant.

We next examined whether disruption of MTs can directlyinfluence aPKC localization. Wild-type egg chambers weredissected and cultured ex vivo in the presence of DMSO (control),paclitaxel (MT stabilizing agent) or nocodazole (MT destabilizingagent). After a 5-hour incubation, stage 8-9 egg chambers wereexamined for expression of aPKC and -tubulin. Egg chamberstreated with DMSO (Fig. 5A) showed normal organization of theMT network (Fig. 5A�) and normal epithelial structure (Fig. 5A�),with apically enriched aPKC (Fig. 5A). Paclitaxel increased thelevels of MT bundles throughout the egg chamber (Fig. 5B�),

1497RESEARCH ARTICLEVHL regulates epithelial morphogenesis

Fig. 3. Homozygous VHL1 egg chambers show severe epithelialdefects. Lethality of homozygous VHL1 is rescued by btl-driven VHLtransgene. (A)A control egg chamber from wild-type femalesexpressing VHL from the btl promoter. (B-F)Ovaries from surviving VHL1

homozygous females expressing btl>VHL. Egg chambers wereprocessed for staining for Arm and aPKC (A-D) or -tubulin and aPKC(E,F) and ToPro-3. (B,C)A stage-7 and a stage-8 egg chamber,respectively, showing piling-up of follicle cells and expansion of theepithelium into the germ cell proper (arrows). (D)A stage-10B eggchamber. The entire follicular epithelium is multilayered. The nurse cellcomplex has collapsed (bracket). (E,F)Two examples of rescued VHLmutant egg chambers showing greatly diminished MTs and aPKC. Allimages were confocal optical cross-sections through the center of theegg chambers. Scale bars: 20m.DEVELO

PMENT

1498

especially in the nurse cells (yellow arrow). Follicle cells weresomewhat thickened but maintained the monolayer characteristics(Fig. 5B�). aPKC localization was also normal (Fig. 5B). Bycontrast, nocodazole treatment greatly diminished MTs throughoutthe epithelium (Fig. 5C�,D�,E�) and caused stacking (Fig. 5C�,E�,sharp arrows). Concomitantly, aPKC was either diminished ormislocalized, not only in cells that showed epithelial defects (Fig.5C, sharp arrows), but also in morphologically normal epithelium(Fig. 5C, arrows; Fig. 5D). Actin filaments showed somedisorganization but were largely intact (Fig. 5E), indicating thatnocodazole treatment for 5 hours does not cause generaldisintegration of the follicular epithelium. In nocodazole-treatedsamples, follicle cells without disintegration of epithelial structurewere shown to demonstrate that aPKC and MT disruption is thedirect result of nocodazole treatment, not downstream indirecteffects from epithelial destruction. Our crucial observation is thatthere is 90% (n20) correlation between MT disruption and aPKCreduction. We therefore conclude that MT destabilization candirectly cause aPKC mislocalization similar to that observed in theVHL mutant in vivo. It is interesting to note that these defectsoccurred in stage 9-10 egg chambers within 5 hours of nocodazoletreatment without cell division (follicle cell division ceased afterstage 6). This suggests that maintenance of epithelial integrity is adynamic and continuous process, for which MTs are cruciallyimportant.

To further demonstrate that VHL function in follicle cells ismediated by MT organization, we examined whether stabilizingMTs can rescue VHL phenotypes. As shown in Fig. 5F,G, ex vivocultured egg chambers containing VHL1 mutant clones (Fig. 5F�,G�,

dashed lines) were treated with either solvent (DMSO) or paclitaxel.Homozygous VHL1 clones from stage 9 egg chambers without drugtreatment showed that MTs are compromised in the mutant cells,with significant loss along the lateral cortex (Fig. 5F�), consistentwith the in vivo VHL phenotype. In 100% (n20) of VHL1 clones,aPKC was almost completely lost from the apical domain (Fig. 5F).By contrast, in 80% of the homozygous VHL1 clones (n20) fromegg chambers treated with paclitaxel, MTs recovered and theircorrect distribution was restored (Fig. 5G�). Importantly, withinthese rescued clones, the correct apical localization of aPKC wasinvariably (100%) re-established (Fig. 5G,G�, arrows).

Because aPKC can also regulate MT organization, we testedwhether paclitaxel can also rescue the MT phenotype in aPKCmutant cells. Interestingly, only 45% of the aPKC mutant clones(n20) could be rescued under the same treatment condition as forthe VHL1 clones described above (Fig. H,I). This indicates thataPKC regulates overlapping as well as different aspects of MTorganization than VHL. Taken together, the above resultsdemonstrate that the epithelial function of VHL is mainly mediatedvia stabilizing MTs, although we cannot exclude the possibility thatVHL can also influence the stability of aPKC as 45% of aPKCmutant clones can be similarly rescued and 20% of the VHL mutantscannot be rescued by paclitaxel treatment.

VHL regulates MTs and aPKC stabilityThe functional relationship among VHL, aPKC and MTs was furthertested in cultured Drosophila S2 cells and in ovaries expressing theVHL full-length sequence or VHL RNA interference construct(VHLi) under heat-shock promoter control. Protein extracts were

RESEARCH ARTICLE Development 137 (9)

Fig. 4. MT defects in VHL1 mutant cells.(A-D�) Ovaries containing VHL1 clones (no GFP) weredissected and egg chambers stained for -tubulin andaPKC (A-B�) or -tubulin and nuclei (ToPro-3; C-D�). Theovary in D was fixed with 8% formaldehyde. All otherswere with 4% formaldehyde. (A-A�) A stage-4 eggchamber showing VHL1 clones (dotted lines) withmislocalized and diminished MTs (A�). aPKC is reducedin these mutant cells (A�). (B-B�) A stage-5 egg chambershowing a stretched epithelium (bracket). A VHL1 clone(dotted lines) shows diminished MTs and aPKC (emptyarrows) compared with normal cells (arrows). Instretched epithelial cells (bracket), residual MTs occupyvery thin cytoplasmic space, thus appearing condensedand stained strongly. (C,C�) A stage-10A egg chamber.A VHL1 clone (dotted line) shows loss of MTs along thelateral cortex. (D,D�) A stage-7 egg chamber. Even with8% formaldehyde fixation, MTs are still lost in the VHL1

clone (dotted line). (E-H�) Ovaries containing aPKCclones were dissected and egg chambers stained for -tubulin, GFP and DNA (ToPro-3 in E) or aPKC (F-H).(E,E�) A stage-4 egg chamber. MTs are not diminished inthe mutant cells (dotted line) but show a dotted pattern(empty arrowheads). Arrows point to mutant regionsthat show the piling-up phenotype. (F,F�) A stage-6 eggchamber containing two aPKC clones (dotted lines),which show abnormal cell shape or stretchedepithelium (brackets). MTs are not diminished in eithercase. (G-G�) A stage-9 egg chamber containing 2 aPKCclones (dotted lines). The aPKC mutant clones showdisorganized and truncated (dotted lines) MT pattern(empty arrowhead) compared with the normal bundles(G�). (H,H�) A stage-9 egg chamber showing a gap inthe epithelium (brackets). Scale bars: 20m.

DEVELO

PMENT

subjected to western blotting. Overexpression of full-length VHLincreased the VHL protein level modestly and expression of VHLisignificantly reduced VHL protein levels in both S2 cells (Fig. 6A)and in ovaries (Fig. 6B). In S2 cells, overexpression of VHLincreased the levels of aPKC and PAR-6, whereas knockdown ofVHL greatly reduced their levels (Fig. 6A). However, the effects ofaltered VHL levels on the levels of aPKC and PAR-6 were presentbut less pronounced in vivo (Fig. 6B). Also interestingly, althoughVHL knockdown did not alter the total cellular tubulin level in S2cells, in ovaries, VHL knockdown greatly reduced tubulin levels,consistent with the immunostaining results (Fig. 4). Perhaps in cell

culture, the MTs are less prone to dynamic instability and thereforemore stable. This is particularly true when compared with follicularepithelium, in which the epithelial sheet movement at stage 9,accompanied by rapid growth into a columnar shape, requiresextensive remodeling and turnover of cytoskeletons.

We next examined whether VHL can interact with tubulin andaPKC in vivo. For this purpose, we used transgenic flies expressingaPKC-GFP fusion protein, which has been shown to be fullyfunctional (Tian and Deng, 2008). Ovarian extracts expressingaPKC-GFP fusion proteins were processed for immunoprecipitation.As shown in Fig. 6C, aPKC-GFP pulled down its known partner

1499RESEARCH ARTICLEVHL regulates epithelial morphogenesis

Fig. 5. MT instability leads to aPKC localization andepithelial defects. (A-E�) y w (representing wild type),(F-G�) VHL1 clone-containing or (H-I�) aPKC mutant clone-containing egg chambers were cultured ex vivo andtreated with DMSO (A,F), paclitaxel (3.2M for B and6.4M for G-I) or nocodazole (20M, C-E). Five hoursafter drug treatment, the egg chambers were processedfor staining for -tubulin, nuclei (ToPro-3) and either aPKCor actin (Phalloidin) as indicated. (A-A�) A stage-9 eggchamber treated with DMSO showing normal epithelium.(B-B�) A paclitaxel-treated stage-10A egg chambershowing increased level of MT bundles, especially in thegerm cells (yellow arrow). aPKC expression is normal (B).There is some thickening of the follicle cells but theepithelium maintains the monolayer morphology (B�).(C-C�) A nocodazole-treated stage-10A egg chambershowing diminished aPKC and MTs either in a regionshowing stacking of follicle cells (sharp arrows) or in amorphologically normal region of the epithelium (arrows).(D-D�) A nocodazole-treated stage-9 egg chambershowing diminished aPKC (D) and MTs (D�) in the entireepithelium (also see merged view in D�). (E-E�) Anocodazole-treated stage-8 egg chamber. MTs arediminished (E�) and the epithelium shows multilayeredphenotypes (sharp arrows). Actin filament staining showsthat although the epithelial structure is disrupted, the drugdosage used does not cause general disintegration of thefollicle cells. (F-F�) Enlarged view of a DMSO-treated stage-9 egg chamber containing a VHL1 clone (dashed lines;identified as lack of GFP in F�). MTs are diminished alongthe lateral cortex of the mutant cells (inset) and aPKC isalso diminished (F,F�). (G-G�) Enlarged view of a paclitaxel-treated stage-9 egg chamber containing a VHL1 clone(dashed lines; lack of GFP in G�). The MT bundles arerestored in the mutant cells and aPKC regains the apicallocalization (arrows). The percentage in G� indicates that80% of the VHL1 mutant clones (n20) are rescued by thetreatment. (H-I�) Egg chambers containing aPKC mutantclones (dotted lines; lack of aPKC in H and I and lack ofGFP in H� and I�) were treated with paclitaxel as above. Thetreatment restores MT bundles (arrow in H�) in 45% of theclones (H�). The rest still exhibit disorganized and dottedMT pattern (inset in I�). Scale bars: 20m.

DEVELO

PMENT

1500

Baz, but not tubulin (left panels), while VHL was co-immunoprecipitated by either -tubulin or aPKC-GFP (right panels).Interestingly, VHL interactions with MTs and aPKC are probablynot within the same complex as the majority of aPKC does notassociate with microtubules (aPKC-GFP does not bring downtubulin). These results are consistent with a direct regulatoryfunction of VHL on MTs and suggest an additional VHL function indirectly regulating aPKC stability.

To test more specifically the significance of reduced aPKCprotein level in VHL mutant cells, we examined whetheroverexpression of aPKC can rescue the VHL knockdown phenotype.UAS-aPKC-GFP; hs-VHLi flies were crossed with flies from theCy2-Gal4 line, which expresses the Gal4 transgene in all the folliclecells covering the oocyte from stage 8 (Queenan et al., 1997). Heat-shock-induced VHL knockdown was activated by heat-treatment offlies at 37°C for 1 hour per day for 4 days. After 12 hours at 29°Cfrom the fourth heat-shock treatment (29° promotes the expressionaPKC-GFP transgene), ovaries from females carrying hs-VHLi andUAS-aPKC-GFP or hs-VHLi alone (both with Cy2-Gal4) weredissected and analyzed. The piling up phenotype was tabulated fromstage 4 to 9. When quantifying the extent of rescue by exogenouslyexpressed aPKC-GFP, only cells expressing GFP were examined.That is, an egg chamber containing pile-up follicle cells in the non-GFP-expressing region was not counted; these pile-up follicle cellscannot be regarded as ‘not rescued’ as they do not express aPKC-GFP. In 50% of stage 4 and 5 hs-VHLi egg chambers analyzed(n30), follicular epithelium lost the coherent single-layered shapeafter heat-shock treatments and endogenous aPKC was eitherreduced or absent in the affected epithelium (see Fig. S3 in thesupplementary material). The same percentage of mutants wasobtained for the hs-VHLi flies with UAS-aPKC-GFP expression(n30) in stage 4-5 egg chambers. This is expected as the Cy2-Gal4driver is only expressed in the follicular epithelium after stage 8.

This set of early eggs serves as a convenient internal control.Conversely, 60% of stage 9 egg chambers analyzed (n36) from hs-VHLi females displayed a multilayered phenotype, whereas 43% ofstage 9 egg chambers from UAS-aPKC-GFP-expressing flies (n36)displayed the same phenotype. The extent of rescue, althoughnotable, is not statistically significant (based on Pearson’s Chi-squared test, P0.2357; see Fig. S3 in the supplementary material).The partial rescue might be because of an insufficient expressionlevel of aPKC-GFP. This is not probable, however, as the sameCy2-Gal4-driven aPKC-GFP can rescue an aPKC mutant (data notshown). More plausible is that the direct stabilizing (versus properlocalization) action of VHL on aPKC plays an appreciable but minorrole in epithelial morphogenesis.

A disease-related VHL mutant defective in MTstabilization cannot rescue the aPKC localizationphenotypeA few disease-causing mutations in human VHL have been shown toameliorate the ability of VHL to promote MT stability in culturedcells (Hergovich et al., 2003). One of these amino acid substitutionsaffects a highly conserved tyrosine residue (Y98 to histidine). Wegenerated the equivalent mutation in Drosophila VHL (Y51 to H inthe Drosophila counterpart) as a hemagglutinin fusion protein. Thewild-type VHL-HA transgene (VHLwt-HA) is functional as it canrescue VHL1 phenotypes in 71% of the VHL mutant clones (n24),whereas the Y-to-H mutant (VHLYH-HA) cannot in 100% of theclones examined (n26; see Fig. S4 in the supplementary material).Anti-HA staining in VHLwt-HA-expressing egg chambers showedpunctate cytoplasmic localization of the wild-type VHLwt-HA fusionprotein (see Fig. S5A in the supplementary material). There wassubstantial co-localization of VHLwt-HA and the cortical MTbundles along the lateral membrane, in agreement with theimmunoprecipitation result, although not all VHL punctae wereassociated with MTs (see Fig. S5A in the supplementary material).Consistent with the property of the human Y98H mutant protein,Drosophila VHLYH-HA retained, at least partly, its ability to co-localize with the MT (see Fig. S5B in the supplementary material).Importantly, the expression levels of the wild-type and YH mutantVHL proteins were the same in the follicle cells, measured by ImageJsoftware analysis of confocal images of two-cell clusters from fivedifferent same-stage egg chambers of each strain. VHL also showedlocalization along the MT bundles in cultured S2 cells (see Fig. S5Bin the supplementary material). The effects of the wild-type and YHmutant VHL on MT stability were further tested ex vivo. As shownbefore, nocodazole treatment on a normal egg chamber disrupted theMT organization and reduced the overall levels of MT bundles(compare Fig. 7A,B). Overexpression of VHLwt-HA counteracted thedestructive effects of nocodazole (compare Fig. 7B-D). By contrast,the VHLYH-HA mutant had only limited effect (compare Fig. 7E,F).The MTs in these follicle cells contained only a mesh in the apicalregion. These observations are quantified in Fig. 7G. Interestingly,without intact MT bundles, the VHLYH protein itself becamemislocalized. It did not colocalize with the disorganized MT andbecame more nuclear (Fig. 7F, insets). This indicates that cytoplasmicdistribution of the VHL protein is itself MT-dependent.

DISCUSSIONThe role of VHL and MTs in epithelialmorphogenesisIn this report, we show that Drosophila VHL is important forestablishing and maintaining epithelial integrity via its regulation ofMT and aPKC stability. We observed MT disruption and epithelial

RESEARCH ARTICLE Development 137 (9)

Fig. 6. VHL interacts with MTs and aPKC. Head-to-head VHL cDNAduplex (dVHLi) or full-length VHL (dVHL) were cloned into the pCaSpe-hsvector and transfected into Schneider (S2) cells or used to generatetransgenic flies. (A)Lysates from S2 cells transfected with control plasmidor vectors expressing full-length VHL or VHL duplex were western-blotted for indicated proteins. (B)Ovaries from females carrying hs-VHL(two lines) or VHLi were dissected and protein lysates processed forwestern blotting with the indicated antibodies. (C)Protein lysates fromovaries carrying Cy2-Gal4/UAS-aPKC-GFP were processed forimmunoprecipitation (IP) and probed with the antibodies indicated onthe left. Input denotes western blots of total protein lysates without IP.

DEVELO

PMENT

phenotypes early in oogenesis. This indicates that MT bundles indeveloping epithelial cells are crucial for epithelial development andare under pressure from dynamic instability. Without stabilizingactivity provided by VHL, MTs are disorganized and ultimatelydisintegrate, resulting in loss of epithelial integrity. We show thatdisrupted MTs interfere with proper localization of aPKC, which inturn leads to mislocalization of downstream epithelial markers andepithelial defects. Our ex vivo experiment also demonstrates thatepithelial defects can occur within a short time (relative to the entireoogenesis time frame) after destabilizing MTs in non-proliferatingepithelial cells. This indicates that the maintenance of epithelialintegrity is a dynamic and continuous process even in a stableepithelium, for which MTs are crucially important. Previous studiesusing RNA interference-mediated knockdown demonstrated amorphogenic role of VHL in trachea development (Adryan et al.,2000; Mortimer and Moberg, 2009). The tracheal phenotypes appearto be the result of elevated cell motility and ectopic chemotacticsignaling. Therefore, the tracheal function of VHL might bemediated via different VHL targets in a tissue-specific context.Alternatively, regulation of MT stabilization might also be theunderlying mechanism. We favor a separate, tissue-specific functionfor VHL as the tracheal defects in VHL knockdown can be relievedby decreased expression of breathless, which encodes thechemotactic signaling receptor in the trachea. The two VHLfunctions, however, are not necessarily mutually exclusive. Thesedifferent organ systems might in the future serve as a model fortesting whether the various functions assigned to VHL are tissue-specific and context-dependent.

Human VHL has been shown to translocate aPKC to MTs,thereby influencing MT reorganization (Schermer et al., 2006). Inthis study, we show that the aPKC mutant can affect MT

organization but not stability, whereas VHL can influence both.Conversely, we show that disruption of MTs alone can result inaPKC mislocalization resembling that observed in VHL mutantcells. Importantly, paclitaxel-induced MT stabilization can rescueaPKC localization in VHL mutant follicle cells. We thereforeconclude that a major function of VHL in the follicular epithelium isregulation of MT stability. Loss of MTs leads to aPKCmislocalization and degradation. Conversely, our results alsoindicate that part of the VHL epithelial functions might be mediatedby its direct effect on aPKC stability, as exogenously expressedaPKC-GFP fusion protein can partially rescue (not statisticallysignificant) the VHL mutant phenotype. Indeed, we also demonstratethat VHL can co-immunoprecipitate with tubulin or aPKC, and that,at least in S2 cells, aPKC levels can be affected by VHL levelswithout affecting tubulin. Taken together, it appears that theepithelial function of VHL is mediated through stabilization of MT,with an auxiliary role in directly stabilizing aPKC.

It has been suggested that VHL interacts with MTs via thekinesin 2 family of motors (Lolkema et al., 2007). Future studiesusing our epithelial system should also address this issue in vivo.Also interestingly, we show that the YH mutant can associate withMT but has little MT-stabilizing activity. This suggests that the YHmutant might be defective in recruiting other proteins, possiblyincluding aPKC, that are important for regulating MT functions. Inlight of the role of Drosophila VHL in regulating MT stability, afunction presumably important for all cells, it is curious that thetissue-specific btl-driven VHL expression can rescue thehomozygous lethality of VHL1. We have shown that trachealdefects are the major embryonic phenotype observed in VHLmutant (Adryan et al., 2000) (data not shown). In the course ofattempting to rescue the tracheal phenotype with btl-driven VHL

1501RESEARCH ARTICLEVHL regulates epithelial morphogenesis

Fig. 7. A VHL disease-related mutantdefective in MT stabilization cannotrestore MT stability againstnocodazole. Control flies (Cy2-Gal4) orflies expressing VHLwt-HA or VHLYH-HAwere conditioned at 25°C with live yeastfor 1 day and incubated at 29°C for 5days to induce transgene expressionbefore the ovaries were dissected andegg chambers cultured ex vivo. Culturedegg chambers were treated with DMSO(A,C,E) or nocodazole (B,D,F) for 5hours and processed for staining for -tubulin, HA and nuclei (ToPro-3). Apicalsides of the follicle cells are marked inthe insets (ap). Nocodazole treatmentsignificantly reduced MTs in control eggchambers (compare B with A). VHLwt-HAcan restore the stability of MTs in thepresence of nocodazole (compare Dwith C and B), whereas VHLYH-HAcannot (compare F with E).(G)Quantification of MT phenotype.Statistical analysis is based on Pearson’sChi-squared test, in which P<0.01 isconsidered significant. Scale bars:20m.

DEVELO

PMENT

1502

(A.H. and T.H., unpublished), we noted the appearance of rescuedhomozygous adults. This indicates that the MT stabilizing functionof VHL is not required in all tissues. It is possible that althoughVHL can enhance MT stability, by itself it is not an essential factorfor MT polymerization. As such, some tissues might be lessdependent on VHL levels. In the follicular environment, MTrearrangement, including depolymerization and repolymerization,is crucial when the entire epithelial sheet moves over the germ cellcomplex while the cells grow increasingly columnar. MTstabilization facilitated by VHL might be of particular importanceduring this process.

Tumor-suppressor function of VHLThe best-documented function of VHL is its E3 ubiquitin ligaseactivity that targets the alpha subunit of the HIF transcription factor.This activity provides an elegant mechanistic explanation for thehypervascularity of many of the VHL tumors and for a potentialcontributor to the metabolic switch to glycolysis, as HIF canupregulate pro-angiogenic factors such as vascular-endothelialgrowth factor and components in the glucose metabolic pathway.However, recent evidence has suggested that VHL is amultifunctional protein. It can function as a regulator of matrixdeposition (Ohh et al., 1998), integrin assembly (Esteban-Barraganet al., 2002), endocytosis (Champion et al., 2008; Hsu et al., 2006),kinase activity (Yang et al., 2007), senescence (Young et al., 2008),protein stabilities (Chitalia et al., 2008; Roe and Youn, 2006) andtight junction formation (Calzada et al., 2006; Harten et al., 2009),among many others (Frew and Krek, 2007). Whether tight junctiondisassembly in VHL mutant cells is HIF-dependent is stillunresolved; however, other – HIF-independent – functions appearto facilitate protein stability or activity instead of destabilizing themas a ubiquitin ligase. Such chaperon/adaptor function has also beenimplicated in promoting stability of MTs (Hergovich et al., 2003;Lolkema et al., 2004). The MT-stabilizing function, althoughpotentially highly significant, has so far only been linked to ciliumbiogenesis and mitotic spindle orientation in cultured RCC and renaltubule cells (Esteban et al., 2006; Lolkema et al., 2008; Lutz andBurk, 2006; Schermer et al., 2006; Thoma et al., 2009). Thephysiological and developmental significance of this function hasnot been elucidated in vivo. Indeed, it is unclear how loss of manyof these HIF-independent functions contributes to VHL tumorformation because of a lack of tractable genetic models.

One crucial element in tumorigenesis is the breakdown ofepithelial integrity that ultimately leads to epithelial-to-mesenchymal transition. This report provides the first demonstrationof a potential tumor-suppressor function for VHL in regulatingepithelial morphogenesis via its role in promoting MT stability.Future studies should exploit further this genetic system forelucidating how a myriad of disease-related VHL point mutationsmight differentially influence such function.

AcknowledgementsWe thank W. M. Deng (Florida State University), D. Grifoni (University ofBologna), T. Schüpbach (Princeton University) and A. Wodarz (Georg-August-University Göttingen) for generous gifts of antibodies and fly stocks. K.Lavenburg contributed to the initial cloning of the Drosophila VHL genomicsequence for subsequent mutagenesis. The work was supported by grantsfrom the National Institutes of Health (PO1CA78582 and RO1CA109860) toT.H., a grant from the University of Bologna (RFO 2007) to G.G. and V.C. and aMarco Polo Fellowship from the University of Bologna to S.D. Deposited inPMC for release after 12 months.

Competing interests statementThe authors declare no competing financial interests.

Supplementary materialSupplementary material for this article is available athttp://dev.biologists.org/lookup/suppl/doi:10.1242/dev.042804/-/DC1

ReferencesAdryan, B., Decker, H. J., Papas, T. S. and Hsu, T. (2000). Tracheal development

and the von Hippel-Lindau tumor suppressor homolog in Drosophila. Oncogene19, 2803-2811.

Arquier, N., Vigne, P., Duplan, E., Hsu, T., Therond, P. P., Frelin, C. andD’Angelo, G. (2006). Analysis of the hypoxia-sensing pathway in Drosophilamelanogaster. Biochem. J. 393, 471-480.

Aso, T., Yamazaki, K., Aigaki, T. and Kitajima, S. (2000). Drosophila von Hippel-Lindau tumor suppressor complex possesses E3 ubiquitin ligase activity. Biochem.Biophys. Res. Commun. 276, 355-361.

Bartolini, F. and Gundersen, G. G. (2006). Generation of noncentrosomalmicrotubule arrays. J. Cell Sci. 119, 4155-4163.

Betschinger, J., Eisenhaber, F. and Knoblich, J. A. (2005). Phosphorylation-induced autoinhibition regulates the cytoskeletal protein Lethal (2) giant larvae.Curr. Biol. 15, 276-282.

Bilder, D., Schober, M. and Perrimon, N. (2003). Integrated activity of PDZprotein complexes regulates epithelial polarity. Nat. Cell Biol. 5, 53-58.

Calzada, M. J., Esteban, M. A., Feijoo-Cuaresma, M., Castellanos, M. C.,Naranjo-Suarez, S., Temes, E., Mendez, F., Yanez-Mo, M., Ohh, M. andLandazuri, M. O. (2006). von Hippel-Lindau tumor suppressor protein regulatesthe assembly of intercellular junctions in renal cancer cells through hypoxia-inducible factor-independent mechanisms. Cancer Res. 66, 1553-1560.

Champion, K. J., Guinea, M., Dammai, V. and Hsu, T. (2008). Endothelialfunction of von Hippel-Lindau tumor suppressor gene: control of fibroblastgrowth factor receptor signaling. Cancer Res. 68, 4649-4657.

Chitalia, V. C., Foy, R. L., Bachschmid, M. M., Zeng, L., Panchenko, M. V.,Zhou, M. I., Bharti, A., Seldin, D. C., Lecker, S. H., Dominguez, I. et al.(2008). Jade-1 inhibits Wnt signalling by ubiquitylating beta-catenin andmediates Wnt pathway inhibition by pVHL. Nat. Cell Biol. 10, 1208-1216.

Esteban, M. A., Harten, S. K., Tran, M. G. and Maxwell, P. H. (2006).Formation of primary cilia in the renal epithelium is regulated by the von Hippel-Lindau tumor suppressor protein. J. Am. Soc. Nephrol. 17, 1801-1806.

Esteban-Barragan, M. A., Avila, P., Alvarez-Tejado, M., Gutierrez, M. D.,Garcia-Pardo, A., Sanchez-Madrid, F. and Landazuri, M. O. (2002). Role ofthe von Hippel-Lindau tumor suppressor gene in the formation of beta1-integrinfibrillar adhesions. Cancer Res. 62, 2929-2936.

Frew, I. J. and Krek, W. (2007). Multitasking by pVHL in tumour suppression.Curr. Opin. Cell Biol. 19, 685-690.

Grifoni, D., Garoia, F., Schimanski, C. C., Schmitz, G., Laurenti, E., Galle, P. R.,Pession, A., Cavicchi, S. and Strand, D. (2004). The human protein Hugl-1substitutes for Drosophila lethal giant larvae tumour suppressor function in vivo.Oncogene 23, 8688-8694.

Harris, T. J. and Peifer, M. (2007). aPKC controls microtubule organization tobalance adherens junction symmetry and planar polarity during development.Dev. Cell 12, 727-738.

Harten, S. K., Shukla, D., Barod, R., Hergovich, A., Balda, M. S., Matter, K.,Esteban, M. A. and Maxwell, P. H. (2009). Regulation of renal epithelial tightjunctions by the von Hippel-Lindau tumor suppressor gene involves occludin andclaudin 1 and is independent of E-cadherin. Mol. Biol. Cell 20, 1089-1101.

Hergovich, A., Lisztwan, J., Barry, R., Ballschmieter, P. and Krek, W. (2003).Regulation of microtubule stability by the von Hippel-Lindau tumour suppressorprotein pVHL. Nat. Cell Biol. 5, 64-70.

Hsu, T., Adereth, Y., Kose, N. and Dammai, V. (2006). Endocytic function of vonHippel-Lindau tumor suppressor protein regulates surface localization offibroblast growth factor receptor 1 and cell motility. J. Biol. Chem. 281, 12069-12080.

Hutterer, A., Betschinger, J., Petronczki, M. and Knoblich, J. A. (2004).Sequential roles of Cdc42, Par-6, aPKC, and Lgl in the establishment of epithelialpolarity during Drosophila embryogenesis. Dev. Cell 6, 845-854.

Kaelin, W. G., Jr (2008). The von Hippel-Lindau tumour suppressor protein: O2sensing and cancer. Nat. Rev. Cancer 8, 865-873.

Knust, E. and Bossinger, O. (2002). Composition and formation of intercellularjunctions in epithelial cells. Science 298, 1955-1959.

Lolkema, M. P., Mehra, N., Jorna, A. S., van Beest, M., Giles, R. H. and Voest,E. E. (2004). The von Hippel-Lindau tumor suppressor protein influencesmicrotubule dynamics at the cell periphery. Exp. Cell Res. 301, 139-146.

Lolkema, M. P., Mans, D. A., Snijckers, C. M., van Noort, M., van Beest, M.,Voest, E. E. and Giles, R. H. (2007). The von Hippel-Lindau tumour suppressorinteracts with microtubules through kinesin-2. FEBS Lett. 581, 4571-4576.

Lolkema, M. P., Mans, D. A., Ulfman, L. H., Volpi, S., Voest, E. E. and Giles, R.H. (2008). Allele-specific regulation of primary cilia function by the von Hippel-Lindau tumor suppressor. Eur. J. Hum. Genet. 16, 73-78.

Lutz, M. S. and Burk, R. D. (2006). Primary cilium formation requires von hippel-lindau gene function in renal-derived cells. Cancer Res. 66, 6903-6907.

RESEARCH ARTICLE Development 137 (9)

DEVELO

PMENT

Mortimer, N. T. and Moberg, K. H. (2009). Regulation of Drosophila embryonictracheogenesis by dVHL and hypoxia. Dev. Biol. 329, 294-305.

Musch, A. (2004). Microtubule organization and function in epithelial cells. Traffic5, 1-9.

Nelson, W. J. (2003). Adaptation of core mechanisms to generate cell polarity.Nature 422, 766-774.

Ohh, M., Yauch, R. L., Lonergan, K. M., Whaley, J. M., Stemmer-Rachamimov, A. O., Louis, D. N., Gavin, B. J., Kley, N., Kaelin, W. G., Jr andIliopoulos, O. (1998). The von Hippel-Lindau tumor suppressor protein isrequired for proper assembly of an extracellular fibronectin matrix. Mol. Cell 1,959-968.

Peifer, M. (1993). The product of the Drosophila segment polarity gene armadillois part of a multi-protein complex resembling the vertebrate adherens junction.J. Cell Sci. 105, 993-1000.

Prasad, M., Jang, A. C., Starz-Gaiano, M., Melani, M. and Montell, D. J.(2007). A protocol for culturing Drosophila melanogaster stage 9 egg chambersfor live imaging. Nat. Protoc. 2, 2467-2473.

Queenan, A. M., Ghabrial, A. and Schupbach, T. (1997). Ectopic activation oftorpedo/Egfr, a Drosophila receptor tyrosine kinase, dorsalizes both the eggshelland the embryo. Development 124, 3871-3880.

Rodriguez-Boulan, E., Kreitzer, G. and Musch, A. (2005). Organization ofvesicular trafficking in epithelia. Nat. Rev. Mol. Cell Biol. 6, 233-247.

Roe, J. S. and Youn, H. D. (2006). The positive regulation of p53 by the tumorsuppressor VHL. Cell Cycle 5, 2054-2056.

Rong, Y. S. and Golic, K. G. (2000). Gene targeting by homologousrecombination in Drosophila. Science 288, 2013-2018.

Rong, Y. S., Titen, S. W., Xie, H. B., Golic, M. M., Bastiani, M.,Bandyopadhyay, P., Olivera, B. M., Brodsky, M., Rubin, G. M. and Golic, K.G. (2002). Targeted mutagenesis by homologous recombination in D.melanogaster. Genes Dev. 16, 1568-1581.

Schermer, B., Ghenoiu, C., Bartram, M., Muller, R. U., Kotsis, F., Hohne, M.,Kuhn, W., Rapka, M., Nitschke, R., Zentgraf, H. et al. (2006). The vonHippel-Lindau tumor suppressor protein controls ciliogenesis by orientingmicrotubule growth. J. Cell Biol. 175, 547-554.

Sotillos, S., Diaz-Meco, M. T., Caminero, E., Moscat, J. and Campuzano, S.(2004). DaPKC-dependent phosphorylation of Crumbs is required for epithelialcell polarity in Drosophila. J. Cell Biol. 166, 549-557.

Suzuki, A. and Ohno, S. (2006). The PAR-aPKC system: lessons in polarity. J. CellSci. 119, 979-987.

Tanentzapf, G. and Tepass, U. (2003). Interactions between the crumbs, lethalgiant larvae and bazooka pathways in epithelial polarization. Nat. Cell. Biol. 5,46-52.

Tanentzapf, G., Smith, C., McGlade, J. and Tepass, U. (2000). Apical, lateral,and basal polarization cues contribute to the development of the follicularepithelium during Drosophila oogenesis. J. Cell Biol. 151, 891-904.

Tepass, U. (2002). Adherens junctions: new insight into assembly, modulation andfunction. BioEssays 24, 690-695.

Tepass, U., Theres, C. and Knust, E. (1990). crumbs encodes an EGF-like proteinexpressed on apical membranes of Drosophila epithelial cells and required fororganization of epithelia. Cell 61, 787-799.

Theurkauf, W. E. (1994). Microtubules and cytoplasm organization duringDrosophila oogenesis. Dev. Biol. 165, 352-360.

Thoma, C. R., Toso, A., Gutbrodt, K. L., Reggi, S. P., Frew, I. J., Schraml, P.,Hergovich, A., Moch, H., Meraldi, P. and Krek, W. (2009). VHL loss causesspindle misorientation and chromosome instability. Nat. Cell Biol. 11, 994-1001.

Tian, A. G. and Deng, W. M. (2008). Lgl and its phosphorylation by aPKCregulate oocyte polarity formation in Drosophila. Development 135, 463-471.

Wodarz, A., Hinz, U., Engelbert, M. and Knust, E. (1995). Expression of crumbsconfers apical character on plasma membrane domains of ectodermal epitheliaof Drosophila. Cell 82, 67-76.

Wodarz, A., Ramrath, A., Kuchinke, U. and Knust, E. (1999). Bazooka providesan apical cue for Inscuteable localization in Drosophila neuroblasts. Nature 402,544-547.

Xu, T. and Rubin, G. M. (1993). Analysis of genetic mosaics in developing andadult Drosophila tissues. Development 117, 1223-1237.

Yamanaka, T., Horikoshi, Y., Sugiyama, Y., Ishiyama, C., Suzuki, A., Hirose,T., Iwamatsu, A., Shinohara, A. and Ohno, S. (2003). Mammalian Lgl forms aprotein complex with PAR-6 and aPKC independently of PAR-3 to regulateepithelial cell polarity. Curr. Biol. 13, 734-743.

Yang, H., Minamishima, Y. A., Yan, Q., Schlisio, S., Ebert, B. L., Zhang, X.,Zhang, L., Kim, W. Y., Olumi, A. F. and Kaelin, W. G., Jr (2007). pVHL acts asan adaptor to promote the inhibitory phosphorylation of the NF-kappaB agonistCard9 by CK2. Mol. Cell 28, 15-27.

Young, A. P., Schlisio, S., Minamishima, Y. A., Zhang, Q., Li, L., Grisanzio, C.,Signoretti, S. and Kaelin, W. G., Jr (2008). VHL loss actuates a HIF-independent senescence programme mediated by Rb and p400. Nat. Cell Biol.10, 361-369.

1503RESEARCH ARTICLEVHL regulates epithelial morphogenesis

DEVELO

PMENT