Developmental Biology - UNC DEPARTMENT OF BIOLOGY

N. Xu et al. / Developmental Biology 321 (2008) 88–100transcription factor, Fork head (Fkh) (Myat and Andrew, 2000a; Myatand Andrew, 2000b). Hairy and Huckebein-dependent transcriptionalregulation of the apical determinant protein, Crumbs (Crb) isnecessary for apical membrane generation during gland invagination.After invagination is complete, the distal tip of the gland contacts theoverlying circular visceral mesoderm (CVM) and migrates with thedistal tip cells elongating and extending protrusions in the direction ofmigration (Bradley et al., 2003). The entire gland then turns to alignitself along the anterior–posterior axis before migrating furtherposteriorly. In this study, we show that Rho1 GTPase regulatessalivary gland invagination by maintaining apical localization of Crb,Drosophila atypical PKC (DaPKC) and Stardust (Sdt) and that thisoccurs partially through regulation of crb RNA level and apicallocalization of the transcript and by inducing apical constriction andcell shape change through Rho-kinase (Rok). The Rho–Rok signalingpathway is required again during gland migration for contraction anddorsal migration of the proximal half of the gland that allows theentire gland to turn and migrate posteriorly.Materials and methodsDrosophila strains and GeneticsCanton-S flies were used as wild-type controls. The following flylines were obtained from the Bloomington Stock Center and aredescribed in FlyBase (http://flybase.bio.indiana.edu/): Rho1K02107b(Rho1K), Rho11B, Rho172F, wingless (wg)-GAL4, engrailed (en)-GAL4,armadillo (arm)-GAL4, UAS-rok-CAT, UAS-rok-CAT-KG, rok2and UASmouse CD8GFP (mCD8GFP). UAS-rokRNAi was obtained from theVienna Drosophila Research Center (VDRC). crb11A22 and UAS-crbWTwere gifts of E. Knust, UAS-Rho1N19 and UAS-Rho1V12were gifts of N.Perrimon, UAS-Rho1WT was a gift of N. Harden, UAS-actinGFP, bagpipe(bap)-GAL4 and twist (twi)-GAL4 were gifts of M. Baylies and crb11A2Df(3L)H99 was a gift of D. Bilder. fork head (fkh)-GAL4 was used todrive salivary gland specific expression (Henderson and Andrew,2000).Antibody staining of embryosEmbryos were fixed and processed for antibody staining aspreviously described (Reuter et al., 1990). The following antiserawere used at the indicated dilutions: rat dCREB-A antiserum at1:10,000 for DAB staining and 1:1250 for fluorescence; rabbit Fkhantiserum (a gift from M. Stern and S. Beckendorf) at 1: 1000; rabbitDaPKC antiserum (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at1:500; mouse Crumbs antiserum (Developmental Studies HybridomaBank, DSHB; Iowa City, IA) at 1:100 for DAB and 1:10 forfluorescence; rabbit Stardust antiserum at 1:500 (a gift from E.Knust); mouse Neurotactin antiserum at 1:10 (DSHB); mouse αspectrin antiserum at 1:10 (DSHB); mouse Fasciclin III (FasIII)antiserum at 1:20 (DSHB); mouse β-galactosidase (β-gal) antiserum(Promega; Madison, WI) at 1:10,000 for DAB staining and 1:500 forfluorescence; rabbit phospho-myosin light chain (p-MLC) antiserumat 1:20 (Cell Signaling Technology, Danvers, MA), rabbit Bazookaantiserum at 1:1000 (a gift from A. Brand), mouse GFP antiserum at1:20,000 (Roche Diagnostics, Indianapolis, IN) and Alexa488conjugated anti-GFP at 1:50 (Invitrogen Molecular Probes, OR).Appropriate biotinylated-(Jackson Immunoresearch Laboratories,Westgrove, PA), AlexaFluor488- or Rhodamine-(Molecular Probes,Eugene, OR) conjugated secondary antibodies were used at a dilutionof 1:500. F-actin was labeled with AlexaFluor488-phalloidin (Molecular Probes). Whole-mount stained embryos were mounted inmethyl salicylate (Sigma, St. Louis, MO), 85% glycerol with 2.5% Npropylgalate or Aqua Polymount (Polysciences, Inc., Warrington, PA).Embryos were visualized on a Zeiss Axioplan 2 microscope withAxiovision Rel 4.2 software (Carl Zeiss, Thornwood, NY) and thick89(1 μm) fluorescent images were acquired on a Zeiss Axioplanmicroscope (Carl Zeiss) equipped for laser scanning confocalmicroscopy at the Rockefeller University Bio-imaging ResourcesCenter (New York, NY).RNA in situ hybridizationIn situ hybridization with antisense digoxigenin-labeled RNAprobes for crumbs was performed as previously described (Lehmannand Tautz, 1994). crumbs and β-galactosidase cDNAs were used astemplates for generating antisense digoxygenin-labeled RNA probesas previously described (Myat and Andrew, 2002). Embryos weremounted in 70% glycerol before visualization as described above forantibody staining.Reverse transcription (RT) and real-time PCR analysesUAS-Rho1N19 UAS-actinGFP/CFL flies were crossed to armadilloGAL4 flies and heterozygous and homozygous embryos weremanually selected with a Zeiss Stereo Discovery V12 ZoomMicroscope (Carl Zeiss). Total mRNA was extracted according tomanufacturer's instructions using the QIAshredder and RNeasy minikit from Qiagen (Valencia, CA). Reverse transcription (RT) wasperformed according to manufacturer's instructions using the OneStep RT-PCR kit (Qiagen). Primer sequences used for quantification ofcrb transcript were Dcrb5-3 (5′ CGCAGTCCTCTCGCCTTCTTCTAC 3′)and Dcrb3-3 (5′TGGTGCGCGAATACAGTTCCGCC 3′). Reference controlwas rp49 amplified with specific primers, rp49-5-2 (5′ ATGACCATCCGCCCAGCATACAGG 3′) and rp49-3-2 (5′ CTCGTTCTCTTGAGAACGCAGGCG 3′). All primers used were generated by Invitrogen(Carlsbad, CA). Intensity of the crb and control rp49 PCR productswas measured with NIH's ImageJ software and the ratio wascalculated. Real-time PCR was performed at the Weill CornellMicroarray Core Facility.Live imagingLive imaging analysis was performed on the LSM 5 LIVE confocalsystem (Carl Zeiss) equipped with a Diode 488-100 laser. Images wereacquired using either a 20X or 40X lens objective every 5 min at a scanspeed between 1 and 4 for the duration of the recording period.Embryos were adhered to double-sided tape, covered in Halocarbonoil (SIGMA) and maintained at 25 C during the recording.Scoring of salivary gland invagination and migration phenotypesTo score gland invagination phenotypes, stage 14 embryos stainedfor dCREB-A were scored for glands that did not invaginate at all(None), partially invaginated with some cells that formed a tube andother cells that remained at the ventral surface (Partial) or completelyinvaginated (Complete). To score gland migration phenotypes, stage14 embryos stained for dCREB-A were scored for glands that hadcompletely turned, incompletely turned, only the distal tip turned ordistal tip did not turn at all.Quantification of fluorescence intensity and extent of apical-basalcontractionFor quantification of fluorescence intensity in salivary glandplacodes (Figs. 7A and D), stage 11 en-GAL4 UAS-mCD8GFP and wgGAL4 UAS-mCD8GFP embryos were double stained for GFP and DaPKC.Three sets of Z series each consisting of three to five 0.5 μm thickoptical sections were acquired by LSM confocal microscopy and theprojected image of each of the Z series was analyzed by ImageJsoftware. Identical areas measuring 6.07 μm in width and 4.95 μm inlength were selected and the ratio of the mean total signal intensity

90N. Xu et al. / Developmental Biology 321 (2008) 88–100(pixel) for GFP and DaPKC was obtained. For quantification of p-MLCfluorescence intensity in migrating salivary glands (Figs. 7I and J),stage 12 embryos were double stained for p-MLC and Fkh. Pixelintensity measurements of an area 18 μm in width and 17 μm in lengthwere performed as described above.For quantification of extent of apical–basal contraction, live imagesof Rho11B heterozygous and homozygous embryos at stage 11 werefirst acquired as described above. The distance between the apical andbasal membranes of proximal gland cells at the beginning and end ofthe recording were measured with LSM 510 software and the averagecalculated. P values were obtained by STATA software two-wayANOVA analysis (StataCorp, TX).ResultsRho1 GTPase is required for salivary gland invagination and migrationTo understand how Rho1 GTPase regulates salivary gland invagination, we analyzed embryos mutant for three different alleles ofRho1, Rho1K02107b (Rho1K), Rho11B and Rho172F and found that glandinvagination was defective in all three alleles. In Rho1K homozygousembryos, majority of glands failed to invaginate and gland cellsremained at the ventral surface of the embryo (Figs. 1D–F and J) incontrast to heterozygous embryos (Figs. 1A–C and J). Invaginationdefects in Rho1K mutant glands were first observed in late stage 11. InFig. 1. Rho1 GTPase is required for salivary gland invagination and migration. Rho1K heterozygous salivary glands invaginate (A, arrow) and migrate posteriorly (B, arrow) to formelongated glands (C, arrow). Rho1 K homozygous glands begin to invaginate (D, arrow) but do not continue invaginating and cells remain at the ventral surface of the embryo (E and F,arrows). In embryos heterozygous for Rho11B, salivary glands invaginate and form elongated glands (G, arrow) whereas in Rho11B homozygous embryos, proximal gland cells do notinvaginate (H, arrow) and the distal cells do not migrate (H, arrowhead). In embryos expressing dominant negative Rho1N19 specifically in the gland (I), some cells invaginate to form atube (I, arrow) whereas others fail to invaginate (I, arrowhead). (J) Wild-type, mutant and recombinant embryos stained with dCREB-A were scored at stage 14 for glands thatcompletely invaginated (Complete), partially invaginated (Partial) or did not invaginate (None). All embryos were stained for dCREB-A to mark nuclei of salivary gland cells and βgalactosidase to distinguish heterozygous from homozygous embryos.

N. Xu et al. / Developmental Biology 321 (2008) 88–100those Rho1K mutant glands that did invaginate, invagination alwaysbegan in the correct dorsal–posterior position (data not shown). InRho11B homozygous embryos, majority of glands partially invaginated(Figs. 1H and J); however, the internalized portion of the gland failedto turn and migrate posteriorly unlike heterozygous glands that turnedand migrated completely (Fig. 1G). Rho172F homozygous embryosshowed an identical phenotype to Rho11B homozygous embryos wheremajority of glands invaginated but failed to turn and migrate posteriorly (data not shown). Gland invagination and migration defectswere also observed in embryos homozygous for Df(2R)Jp1, a deficiencythat deletes the entire Rho1 gene and trans-heterozygotes of Rho1Kand Df(2R)Jp1 (data not shown). Furthermore, expression of thedominant-negative Rho1N19 mutation specifically in the gland withfkh-GAL4 phenocopied the Rho1K loss of function phenotype with themajority of gland cells failing to invaginate (Figs. 1I and J).To confirm that the gland invagination defects observed in Rho1Kmutant embryos were due to loss of Rho1 function in the gland, weexpressed wild-type Rho1 (Rho1WT) specifically in glands of Rho1Khomozygous embryos with fkh-GAL4 and obtained a substantialrescue; the percentage of non-invaginated glands decreased from80% to 27% and the percentage of completely invaginated glandsincreased from 2% to 58% (n 126 glands; Fig. 1J). Expression of Rho1WTspecifically in salivary glands of wild-type embryos with fkh-GAL4 hadno effect on gland invagination (data not shown). Together, these dataindicate that the invagination defects observed in Rho1K homozygousembryos were due to lack of Rho1 function in salivary gland cells. TheRho1K allele is due to a P-element insertion in the first intron withinthe coding region (Magie et al., 1999) whereas the Rho11B allele is animprecise excision removing the coding region C-terminal to aminoacid 52. Although no Rho protein was detected in Rho11B mutantembryos (Magie and Parkhurst, 2005), the phosphate binding loop andthe entire effector domain that mediates binding of Rho1 to effectorproteins (Self et al., 1993) are retained within the initial 52 amino acidsof Rho11B, suggesting the possibility that this mutant protein mayretain some activity. Both the Rho1K and Rho11B alleles are described asstrong alleles and were used in the studies described here.Since loss of zygotic Rho1 function resulted in failure of gland cellsto invaginate, we tested whether overexpression of constitutivelyactive Rho1, Rho1V12, in gland cells would accelerate gland invagination. In control embryos (fkh-GAL4 /CFL), a small group of gland cellsinvaginated at any given time to form elongated glands with a singlelumen (Fig. S1A and B). In contrast, in Rho1V12 mutant embryos (fkhGAL4 UAS-Rho1V12) the entire placode invaginated simultaneouslyand cells were cuboidal shaped unlike the pyramidal shaped controlcells (Fig. S1C). Continuous expression of Rho1V12throughout glanddevelopment resulted in glands with multiple cyst-like lumena (Fig.S1D) instead of a single lumen characteristic of heterozygous glands(Fig. S1B). These data demonstrate that Rho1 activity is necessary andsufficient for gland invagination.Rho1 activity is required for epithelial shape and apical polaritySalivary gland cells of wild-type (Myat and Andrew, 2000a) andRho1K heterozygous embryos (Fig. 2A) are elongated and columnarwith prominent cortical F-actin. In contrast, the salivary glandepithelium of Rho1K homozygous embryos consisted of mesenchymalshaped cells with disorganized cortical F-actin (Fig. 2B). The loss ofepithelial morphology in Rho1K mutant embryos suggested thatepithelial apical/basolateral (A/B) polarity might also be lost. Ininvaginating wild-type salivary glands, A/B polarity was maintainedthroughout the entire invagination process as indicated by the apicallocalization of Drosophila atypical Protein Kinase C (DaPKC) andCrumbs (Crb) and basolateral localization of Neurotactin (Nrt) (Fig.S2). In early salivary gland placodes of Rho1K heterozygous andhomozygous embryos, the apical protein, Crb, was localized in theapical membrane (Figs. 2C and D). However, upon initiation of91invagination, apical localization of Crb was lost in the non-invaginatedgland cells of Rho1K homozygous embryos (Figs. 2F and H), whereasit was maintained in all gland cells of heterozygous embryos(Figs. 2E and G).To test whether apical polarity in general was not maintained inRho1K mutant embryos or the defect was specific to Crb, we analyzedthe localization of other apical proteins, DaPKC, Stardust (Sdt) andBazooka (Baz) in Rho1K heterozygous and homozygous embryos. DaPKCand Sdt colocalized with Crb at the apical membrane of Rho1Kheterozygous and homozygous glands prior to invagination (data notshown) but were lost from the apical membrane simultaneously withCrb at the onset of invagination in Rho1K homozygous glands (Fig. S3). Incontrast, Baz maintained its apical localization in the non-invaginatedgland cells of Rho1K homozygous embryos (Figs. 3B and B″

Developmental Biology 321 (2008) 88–100 ⁎ Corresponding author. Fax: 1 212 746 8175. E-mail address: mmm2005@med.cornell.edu (M.M. Myat). 1 BCMB Program of Weill Graduate School of Medical Sciences at Cornell University, New York, NY, USA. 2 Present address: New York Medical College, Valhalla, NY 10595, USA.