The T-box-encoding Dorsocross genes function in amnioserosa … · demonstrates that the three Doc genes have largely redundant functions during amnioserosa development, as well as

INTRODUCTION

The BMP family member Dpp has a key role in dorsoventralaxis formation and is responsible for the establishment ofpositional identities in dorsal and lateral areas of the earlyDrosophilaembryo. Cell identities that are determined by Dppinclude those of the extra-embryonic amnioserosa in thedorsalmost region of the embryo, dorsal epidermis andperipheral nervous system (PNS) in dorsolateral regions of theectoderm, as well as dorsal vessel, dorsal somatic and visceralmuscles in the dorsal mesoderm (Irish and Gelbart, 1987; Rayet al., 1991; Ferguson and Anderson, 1992a; Staehling-Hampton et al., 1994; Frasch, 1995). In addition to promotingdorsal epidermal and PNS fates in the dorsolateral ectoderm,Dpp acts to suppress the formation of neurons of the centralnervous system in the same area.

For a better understanding of these activities, we need toconsider that Dpp exercises some of its functions sequentiallyat different stages of development, during which dpp changesits own pattern of expression (St Johnston and Gelbart, 1987).In particular, during blastoderm and gastrulation stages, Dppacts in a dose-dependent fashion to establish positional

information in dorsal and lateral areas of the embryo and tospecify amnioserosa tissue (Ferguson and Anderson, 1992a;Ashe et al., 2000). Although dppis expressed uniformly around~40% of the dorsal circumference of the embryos during thisstage, the activity of Dpp is modulated along the dorsoventralaxis by diffusion of the secreted gene product as well as bypositive and negative regulators of the signaling pathway.Negative regulators include Short gastrulation (Sog) andBrinker (Brk), both of which are expressed ventrolaterally(Ferguson and Anderson, 1992b; Francois et al., 1994;Jazwinska et al., 1999). Whereas Sog and its vertebratehom*olog Chordin are secreted molecules that inhibit BMPsignaling via binding to the ligand (reviewed by Garcia Abreuet al., 2002), Brk appears to be a nuclear factor that interfereswith the signaling output via binding to regulatory sequencesof Dpp target genes (Sivasankaran et al., 2000; Kirkpatrick etal., 2001; Rushlow et al., 2001; Zhang et al., 2001). Bycontrast, specification of amnioserosa fates in the dorsal 10%of embryonic cells requires maximal signaling activities thatinvolve Sog as a positive regulator of Dpp in conjunction withTwisted gastrulation (Tsg) as well as a second, uniformly-distributed BMP ligand, Screw (Scw) (Arora et al., 1994;

Dpp signals are responsible for establishing a variety of cellidentities in dorsal and lateral areas of the early Drosophilaembryo, including the extra-embryonic amnioserosa aswell as different ectodermal and mesodermal cell types.Although we have a reasonably clear picture of how Dppsignaling activity is modulated spatially and temporallyduring these processes, a better understanding of how thesesignals are executed requires the identification andcharacterization of a collection of downstream genes thatuniquely respond to these signals. In the present study, wedescribe three novel genes, Dorsocross1, Dorsocross2andDorsocross3, which are expressed downstream of Dpp inthe presumptive and definitive amnioserosa, dorsalectoderm and dorsal mesoderm. We show that these genesare good candidates for being direct targets of the Dppsignaling cascade. Dorsocross expression in the dorsalectoderm and mesoderm is metameric and requires acombination of Dpp and Wingless signals. In addition, atransverse stripe of expression in dorsoanterior areas of

early embryos is independent of Dpp. The Dorsocross genesencode closely related proteins of the T-box domain familyof transcription factors. All three genes are arranged in agene cluster, are expressed in identical patterns in embryos,and appear to be genetically redundant. By generatingmutants with a loss of all three Dorsocross genes, wedemonstrate that Dorsocross gene activity is crucial for thecompletion of differentiation, cell proliferation arrest, andsurvival of amnioserosa cells. In addition, we show that theDorsocross genes are required for normal patterning of thedorsolateral ectoderm and, in particular, the repression ofwinglessand the ladybirdhomeobox genes within this areaof the germ band. These findings extend our knowledge ofthe regulatory pathways during amnioserosa developmentand the patterning of the dorsolateral embryonic germband in response to Dpp signals.

Key words: T-box, Amnioserosa, Dorsal ectoderm, Dpp, wingless,Drosophila

SUMMARY

The T-box-encoding Dorsocross genes function in amnioserosa development

and the patterning of the dorsolateral germ band downstream of Dpp

Ingolf Reim, Hsiu-Hsiang Lee and Manfred Frasch*

Brookdale Department of Molecular, Cell and Developmental Biology, Mount Sinai School of Medicine, New York NY 10029, USA*Author for correspondence (e-mail: [emailprotected])

Accepted 17 April 2003

3188

Mason et al., 1994; Ashe and Levine, 1999; Decotto andFerguson, 2001). Dpp, Sog and Tsg are thought to be presentin a diffusible trimolecular complex that serves to carry andrelease active Dpp prefentially into dorsalmost areas where tsgis expressed (Decotto and Ferguson, 2001).

After gastrulation, dppexpression ceases in the developingamnioserosa and becomes restricted to a broad stripe of cellsin the dorsolateral ectoderm along the elongated germ band (StJohnston and Gelbart, 1987). During this period, the dorsallymigrating cells of the mesoderm reach the dpp-expressingarea of the ectoderm, thus allowing Dpp to induce dorsalmesodermal cell fates across germ layers (Staehling-Hamptonet al., 1994; Frasch, 1995). In addition, Dpp is thought to actin the continuing patterning processes within the dorsolateralectoderm during this stage, which lead to the specification oftracheal as well as particular epidermal and sensory organprogenitors. Both in the dorsal mesoderm and dorsolateralectoderm, Dpp must act in combination with additionalpatterning molecules that provoke differential responses ofcells to the Dpp signal. For example, in the dorsal mesoderm,the presence or absence of Wingless (Wg) activity determineswhether cells will respond to Dpp by forming heart and dorsalsomatic muscle progenitors versus visceral muscle progenitors(Wu et al., 1995; Azpiazu et al., 1996; Carmena et al., 1998).

In order to obtain more insight into the mechanisms of howDpp signals pattern the embryo and how they are integratedwith other patterning processes, it is crucial to study theregulation of Dpp target genes. To date, detailed molecularstudies have been described for three targets that are inducedduring early embryogenesis, namely the homeobox geneszerknüllt (zen), tinman (tin) and even-skipped(eve). zen isrequired for the specification of the amnioserosa downstreamof Dpp. Accordingly, the expression of zen in a dorsalon/ventral off pattern, although initially Dpp-independent,requires low levels of Dpp activity for its maintenance and highDpp activities for its subsequent refinement to areas of theprospective amnioserosa (Doyle et al., 1986; Rushlow andLevine, 1990). Likewise, tin is required for the specification ofall dorsal mesodermal tissues and eve for the normaldifferentiation of specific pericardial cells and dorsal somaticmuscles in a Dpp-dependent manner. (Bodmer, 1993; Azpiazuand Frasch, 1993; Su et al., 1999). All three genes have incommon the presence of multiple binding sites for intracellularDpp effectors, the Smad proteins Mad and Medea, in theirregulatory regions, which are essential for mediating theinductive activity of Dpp. However, in addition to these Smad-binding sites, each of these genes has a characteristic set ofadditional regulatory sequences that, at least in part, explain itsparticular spatial and tissue-specific response to Dpp signals.For example, zencontains binding sites for Brk in addition tothe Smad sites (Rushlow et al., 2001). It appears that theantagonistic activities of the Brk and Smad sites and thedifferential ratios of Brk versus active Smad proteins along thedorsoventral embryo axis determine the ventral border of Dpp-dependent zendomain during cellularization stages. The Smadsites but not the Brk sites are also required for zeninductionin the prospective amnioserosa during the cellularizedblastoderm stage (Rushlow et al., 2001). The mesodermal Dpptargets tin and everequire Smad-binding sites and, in addition,binding sites for Tin, which serve to target the Dpp responseto the mesoderm (Xu et al., 1998; Halfon et al., 2000; Knirr

and Frasch, 2001). Further, the Dpp-responsive enhancer of evecontains functionally important binding sites for regulators thatrestrict its activity to segmental subsets of dorsal mesodermalcells, including the Wg effector Pangolin (Pan) (Halfon et al.,2000; Knirr and Frasch, 2001).

In the present study, we introduce three novel genes thatrespond to Dpp signals in the prospective amnioserosa, dorsalectoderm and dorsal mesoderm, and are good candidates forbeing direct targets of the Dpp signaling cascade. The threegenes, Dorsocross1 (Doc1), Dorsocross2 (Doc2) andDorsocross3(Doc3), which are present in a gene cluster, areclosely related members of the T-box family of genes andpresumably arose by relatively recent duplications from acommon ancestor. The Dorsocross (Doc) genes are expressedin essentially identical patterns within several areas that receivehigh levels of Dpp signals, including the prospectiveamnioserosa during the cellularized blastoderm stage, thedorsolateral ectoderm and dorsal mesoderm during germ bandelongated stages and areas that span the compartment borderin wing discs. We show that Doc expression in the prospectiveamnioserosa depends on dpp and zen, whereas the metamericexpression in the dorsolateral ectoderm and dorsal mesodermdepends on a combination of dppand wg. Our genetic analysisdemonstrates that the three Doc genes have largely redundantfunctions during amnioserosa development, as well as duringdorsolateral ectoderm and dorsal mesoderm patterning. Wefocus on the role of the Doc genes in the amnioserosa anddorsolateral ectoderm. We show that they are essential for fulldifferentiation and maintenance of the amnioserosa, includingthe arrest of cell proliferation in this tissue. Owing to therequirement of a functional amnioserosa for normal germ bandretraction, loss of Doc activity produces embryos with apermanently extended germ band. Hence, Doc genes are newmembers of the u-shaped family of genes. All genes of thisfamily, which also includes hindsight(hnt; peb – FlyBase),serpent(srp), tail-up (tup), u-shaped(ush), epidermal growthfactor receptor (Egfr) and insulin-like receptor(InR), arecomponents of a regulatory network that controls normaldevelopment and functioning of the amnioserosa (Frank andRushlow, 1996; Goldman-Levi et al., 1996; Yip et al., 1997;Lamka and Lipsh*tz, 1999). In addition to the amnioserosa, theDoc genes are required for the normal patterning of thedorsolateral ectoderm, which includes the repression of wgandladybird (lb) expression within this area. These findingsprovide valuable insight into the mechanisms of how Dppsignals are executed during the development of theamnioserosa and the patterning of dorsolateral areas of theembryonic germ band.

MATERIALS AND METHODS

cDNA cloning and northern blots660-850 bp DNA fragments from the non-conserved 3′-region of thepredicted genes CG5133(Doc1), CG5187 (Doc2) and CG5093(Doc3) were amplified by PCR from genomic DNA and clonedinto pCRII-TOPO (Invitrogen) to obtain gene-specific probes. Thefollowing primer pairs were used: GTTCGCTAAGGGTTTCCGCG-AGTC and GCAAATAGTTTTGCATTTTCTACGGATTC for Doc1;GCGCTGCAAACGCAAGATGTCTTCATC and GCCGATATGCT-GAAGCCCTTGCTCCTT for Doc2; and GTCGAGATGCAAAC-GGAAGATCAATGAC and GTTTCATCCCAGAAATAGCTCCAT-

I. Reim, H.-H. Lee and M. Frasch

3189Dorsocross genes in dorsal tissue development

CGAATTCA for Doc3. A plasmid library containing cDNAs from 4-to 8-hour-old Drosophila melanogasterembryos in pNB40 (Brownand Kafatos, 1988) was screened with the respective radioactivelylabeled fragments and the plasmid DNAs of isolated clones weresequenced (GeneWiz, New York). The cDNAs shown in Fig. 1Acorrespond, from top to bottom, to clones Doc1-a1.1, Doc2-c6.2,Doc2-c1.1, Doc2-c12.2 and Doc3-b4.1.

Generation of Dorsocross antibodiesC-terminal cDNA fragments containing the variable coding region ofeach Dorsocross gene were cloned into the expression vector pQE-9(Qiagen) cut with BamHI and HindIII. Fragments were obtained byPCR using the cDNA clones Doc1-a1.1, Doc2-c6.2 and Doc3-b4.1 astemplates and restriction site-introducing primers. In the recombinantproteins Doc1CTD, Doc2CTD and Doc3CTD, a N-terminal His-tag(MGRSH6GS) is fused to an arginine residue at position 249, 246 and249, respectively. Proteins were produced in E. coli M15{pRep4}.Expression was induced by 1 mM IPTG at mid-log phase and bacteriawere grown for another 3-4 hours at 37°C, except for Doc1CTD thatwas produced at 30°C. Ni-NTA-agarose (Qiagen) purified proteinswere used for immunization (Covance Research Products). Anti-Doc2(made in rabbit) is specific for Doc2 in Drosophilaembryos as shownby Doc1, Doc2 and Doc3 misexpression. Anti-Doc1 antibody (madein rat) reacts with Doc1 in Drosophilaembryos, but only weakly atwild-type Doc1 levels. Unexpectedly, the two guinea pig antibodiesderived from separate Doc2 and Doc3 immunizations both recognizeboth Doc2 and Doc3. Therefore, we name these antibodies anti-Doc2+3 and anti-Doc3+2, with anti-Doc2+3 working better for Doc3detection.

Generation of UAS-Dorsocross transgenesTransformation plasmids were constructed by subcloning the cDNAsDoc1-a1.1, Doc2-c6.2 and Doc3-b4.1 into pP{UAST} (Brand andPerrimon, 1993) using EcoRI or blunted EcoRI and NotI cloning sites.Doc1-a1.1 was cloned as EcoRI-NotI fragment, Doc2-c6.2 as SmaI-NotI fragment and Doc3-b4.1 as HindIII (blunted)-NotI fragment. TheHindIII and NotI sites are pNB40 vector-derived and the EcoRI andSmaI sites are located in the 5′UTR of Doc1and Doc2, respectively.Several independent lines were established for each construct usingstandard transformation methods (Rubin and Spradling, 1982). GAL4inducible expression of Dorsocross proteins was confirmed byimmunostaining with the antibodies described above.

Mutagenesis by male recombinationIn order to create deletions uncovering all three Dorsocross genes amutagenesis screen was performed using the male recombinationmethod (Preston et al., 1996). The closest available P-insertion,EP(3)3556, was used to trigger male recombination in the presenceof active transposase. EP(3)3556is a hom*ozygous viable insertion inthe 5′region of smg (Dahanukar et al., 1999), which is located about8 kb upstream of Doc3 (see Fig. 6A). F0 males subject torecombination/mutagenesis carried both a X-chromosomaltransposase source and the targeted P-insertion flanked by geneticmarkers on the third chromosome (y w H{P∆2-3} HoP8/Y; ruEP(3)3556 th st cu sr es ca/+). F0 males were crossed to ru h th st cusr es Pr ca/TM6B, Bri Tbfemales. The F1 generation was screenedfor non-balancer males carrying a recombinant third chromosome.The ru marker is expected to be retained along with the P-insertion ifrecombination causes a deletion extending towards the Dorsocrossgene cluster (Preston et al., 1996). Recombinants were crossedindividually to Df(3L)Scf-R11/TM3, Sb eve-lacZfemales forproducing TM3, eve-lacZbalanced stocks and for complementationanalysis with Df(3L)Scf-R11.

A total of 23 recombinants (13 ru, 6 th st cu sr es ca and fourunusual marker combinations) were obtained out of 12400 scorednon-balancer F1 males. These recombinants were derived from 18 of168 individual F0 crosses and from each of these crosses only one ru

male was taken to ensure the recovery of individual events. Five outof 10 candidate rulines were hom*ozygous lethal and two of them,Df(3L)DocA and Df(3L)DocB, removed the entire Dorsocross genecluster. An analogous mutagenesis screen with EP(3)584, which isinserted ~80 kb downstream of Doc1, yielded Df(3L)EP584MR2(Fig. 5A).

Molecular characterization of deficienciesThe breakpoints of Df(3L)DocA, Df(3L)DocBand Df(3L)EP584MR2were mapped by sequencing of inverse PCR products recovered fromthe 5′ end of the retained Pinsertion. Inverse PCR was performedas described by E. J. Rehm (http:/www.fruitfly.org/p_disrupt/inverse_pcr.html) and Huang et al. (Huang et al., 2000) using Pwht1and Plac1 primers. The removal of genes by various deficiencies (Fig.5A) was confirmed by PCR from hom*ozygous mutant embryos asdescribed in Duan et al. (Duan et al., 2001). PCR from equally treatedheterozygous embryos and PCR amplification of sequences notaffected by the deficiencies served as positive controls for primers andtemplate, respectively. PCR amplification and sequencing of DNA tothe left of EP(3)3556demonstrated no change in sequences flankingthe 3′P-end.

TUNEL staining and BrdU labeling of embryosApoptotic cells were labeled by terminal deoxynucleotidyltransferase (TdT)-mediated dUTP nick end-labeling (TUNEL) usingcomponents of the ApopTag® Peroxidase kit S7101 (IntergenCompany/Serologicals Corporation). Rehydrated embryos (about 30µl) were treated with 10 µg/ml Proteinase K for 1 minute, rinsedquickly three times and washed another five times for 3 minutes withPBT. Embryos were postfixed in 3.7% formaldehyde in PBT for 20minutes, washed five times with PBT and twice for 20 minutes with30 µl equlibration buffer. TdT reaction was performed over night at37°C using 50 µl buffer/TdT mixed in a ratio 7:3 and supplementedwith 0.3% Triton X-100. The reaction was stopped by a 20 minutewash in 1/34 diluted stop buffer. Detection of incorporatedDigoxigenin-nucleotides using sheep-anti-Dig and biotinylated anti-sheep antibodies, the VectaStain ABC elite kit (Vector Laboratories)and Tyramide Signal Amplification (TSA) reagents (NEN/PerkinElmer Life Sciences) was essentially as for in situ hybridization usingdigoxigenin-labeled probes (Knirr et al., 1999).

BrdU labeling and detection was as described by Shermoen(Shermoen, 2000). Embryos were labeled by 30 minutes incubationin 1 mg/ml BrdU in PBS. For detection mouse anti-BrdU antibody(Becton-Dickinson, 1:200) and Cy3-anti-mouse antibody (JacksonImmunoResearch Laboratories, 1:200) were used. For doublestaining, standard antibody staining using the VectaStain ABC elitekit and TSA fluorescence substrates were performed prior to theTUNEL reaction or BrdU detection.

Drosophila strains and crosses y w or Oregon R were used as wild-type controls and stocks wereobtained from the Bloomington stock collection unless notedotherwise. The following UAS/GAL4 driver lines were used:P{GAL4-nanos.NGT} 40 (Tracey et al., 2000), P{en2.4-GAL4} e22c,P{ZKr-GAL4}#8 (Frasch, 1995), P{dpp.blk1-GAL4} 40C.6,P{w+mW.hs=GawB}c381 (Manseau et al., 1997) (which drivesamnioserosa expression from stage 9; I.R. and M.F., unpublished) andUAS-dpp#5 (Frasch, 1995). The following previously describedmutant alleles were also used: dppH46, hntE8, pnr1, CyO slp∆34B,srpP{PZ}01549, tup1, ush2, wgCX4and zen7. For male recombination andmapping experiments, we used the linesy w H{P∆2-3} HoP8, ‘ru cuca’ and ‘ru Pr ca’, ru Df(3L)Scf-R11,Scf-R6 th st cu sr e ca(Koppand Duncan, 1997),smg1 (Dahanukar et al., 1999),Df(3L)29A6 kniri-1

pp, EP(3)3556, and EP(3)584(Exelixis). For ectopic expressionexperiments, embryos were collected at 28°C from crosses of GAL4-carrying females with UAS construct-carrying males. All othercrosses were performed at room temperature (22-25°C).

3190

RNA interference experimentsSense and antisense RNA was transcribed from non-conserved 3′fragments of Doc1, Doc2and Doc3in pCRII-TOPO (see above) andhybridized in injection buffer (5 mM KCl, 10 mM sodium phosphate,pH 7.8) to generate dsRNA (Kennerdell and Carthew, 1998). A mixof Doc1, Doc2and Doc3dsRNA (~100-300 pl of 4 mg/ml each) wasinjected ventrally into pre-blastoderm embryos that had beendechorionated and mounted onto double-sided sticky tape underHalocarbon 700 oil. An Eppendorf FemtoJet automatic injector andEppendorf Femtotip injection needles were used for injections.Embryos were allowed to develop at 18°C until the desired stage wasreached. For immunostaining, embryos were transferred to standardheptane/formaldehyde fixation solution in a small drop of oil. Theheptane phase and total fixation solution were exchanged twice toremove traces of oil. After 20 minutes incubation, the formaldehydesolution was replaced by PBS. For the manual removal of the vitellinemembranes, embryos were spread on agar plates, transferred todouble-sided sticky tape and covered with PBS. After devitellinization0.05% Tween was added and embryos were transferred into reactiontubes for standard staining procedures. Cuticle preparations weremade 2-3 days after injection. Embryos were passed through aceticacid/glycerol (4:1) overlaid with heptane, transferred to a mesh, rinsedwith heptane and PBT and mounted in standard Hoyer’s medium.

Staining of embryos and imaginal discsAntibody staining of embryos using DAB, double fluorescent stainingand in situ hybridization in combination with fluorescent antibodystaining were carried out as described previously (Knirr et al., 1999).Dorsocross in situ hybridization probes were made by in vitrotranscription from 3′fragments cloned into pCRII-TOPO (see above).The race in situ hybridization probe and zencDNA were a gift fromC. Rushlow (Frank and Rushlow, 1996). Antibody staining ofimaginal discs was essentially the same as for embryos, except thatimaginal discs attached to inverted larval heads were fixed in 3.7%formaldehyde in PBS, dehydrated in methanol, and rehydrated using70%, 50% and 30% methanol in PBT before blocking and staining.

The following antibodies were used: rabbit anti-Doc2 (1:2000),guinea pig anti-Doc2+3 (1:400 to 1:600), guinea pig anti-Doc3+2(1:600), rat anti-Doc1 (1:200), rabbit anti-Bap (1: 500) (Zaffran et al.,2001), rabbit anti-β-galactosidase (Promega; 1:1500), rabbit anti-Phospho-Smad1/PMad (1:2000; gift from C.-H. Heldin), rat anti-Cf1a(1:3500; gift from W. A. Johnson, University of Iowa), guinea piganti-Kr (1:400) (Kosman et al., 1998), guinea pig anti-Slp (1:20)(Kosman et al., 1998), affinity-purified rabbit anti-C15 (1:25 to 1:50,TSA-indirect, NEN; S. Grimm and M.F., unpublished), mousemonoclonal anti-Lbe (1:10) (Jagla et al., 1997), and rabbit anti-phospho-Histone H3 (1:600; Upstate Biotechnology, NY).Monoclonal mouse antibodies obtained from the DevelopmentalStudies Hybridoma Bank, University of Iowa: anti-α-tubulin 12G10(1:10), anti-Futsch 22C10 (1:250), anti-Wg 4D4 (1:40, in imaginaldiscs 1:400), anti-En/Inv 4D9 (1:4), anti-Hnt 27B8 1G9 (1:30) andanti-β-gal 40-1a (1:40).

RESULTS

Three novel T-box genes are clustered in thechromosomal region 66F1-2Sequence information from the Berkeley DrosophilaGenomeProject (Adams et al., 2000) revealed that three novel, closelyrelated T-box encoding genes are clustered within ~40 kb ofgenomic sequences at 66F1 to 66F2 on chromosome arm 3L.In reference to their peculiar patterns of expression inblastoderm embryos (see below), these genes have beennamed, from proximal to distal, Dorsocross1 (Doc1;

previously Tb66F2) (Lo and Frasch, 2001), Dorsocross2(Doc2) and Dorsocross3(Doc3) (Fig. 1A). The gene clusteralso includes an unrelated predicted gene, CG5194, whichmaps between Doc2and Doc3.

cDNAs for the three Doc genes, which were isolated froman early embryonic cDNA library (see Materials and Methods),encode proteins of 391 amino acids (Doc1), 469 aminoacids (Doc2) and 424 amino acids (Doc3), respectively.Comparisons between cDNA and genomic sequences indicatethat Doc2 encodes at least three different mRNA products,which appear to be generated from alternative transcriptionstart sites. Among these, Doc2 variants A and B encodeidentical polypeptides, whereas variant C does not encode anylong open reading frame (Fig. 1A). The data from northernanalysis indicate that the longest cDNAs obtained for eachgene are close to full length if the polyA tails are taken intoaccount (1.7 kb transcripts versus 1500 bp cDNA for Doc1, 2.0kb transcripts versus 1759 bp cDNA for Doc2A, and 1.8 kbtranscripts versus 1681 bp cDNA for Doc3) (data not shown).For Doc2, these data indicate that variant A (1.75 kb,presumably corresponding to the 2.0 kb transcripts) isexpressed much more strongly than the other two variants. Inaddition, we note that splicing occurs at identical positionswithin the open reading frames of Doc1, Doc2and Doc3,although most introns in Doc3are much smaller as comparedwith those in the other two genes (Fig. 1A).

Sequence comparisons show that the three Doc proteinsshare high degrees of similarity within their T-box domainsequences (>95% amino acid identities) as well as within shortsequence stretches extending N- and C-terminally from thesedomains (Fig. 1B). The N-terminal regions of the polypeptidesup to the T-box domains are moderately conserved (>40%amino acid identities), whereas the C-terminal regions containonly few short stretches of additional sequence similarity (data

I. Reim, H.-H. Lee and M. Frasch

Fig. 1.Genomic clustering, gene structure and phylogenetic analysisof the T-box genes Doc1, Doc2and Doc3. (A) Arrangement of thethree Dorsocross genes within a genomic region of about 40 kb.CG5194is a predicted gene with no similarity to any known gene.The exon structures of Doc cDNAs are depicted below with thecoding sequences hatched and the T-box domains in black. The exon-intron structure with the T-box spanning exons 2 to exons 5 isconserved among Doc1, Doc2variant A and Doc3. (B) ClustalX-generated alignment of T-box domains from T-box genes ofDrosophila melanogaster(Doc1, Doc2, Doc3, omb/optomotor-blind,H15, H15r/H15-related/CG6634, org-1/omb-related gene 1,byn/brachyenteron/trg) and human (marked Hs). AdditionalTbx6/16-related members of the T-box family, which appear to forma separate subgroup, are included from zebrafish (Dr) and Xenopuslaevis(Xl). The T-box core sequence was N- and C-terminallyextended in order to include amino acids partially conserved betweensubfamily members. (C) Phylogenetic N-J tree derived fromClustalX analysis, based on the alignment shown in B and using1000 bootstrap trials (bootstrap values at tree node representconfidence values; branches with values below 700 are generallyconsidered as less reliable and below 500 as unreliable. Barrepresents amino acid exchanges as a fraction of 1). Caenorhabditiselegans(Ce) Tbx9 was included as an outgroup member. GenBankAccession Numbers are, for Doc2A, AAM11544; for Doc2B,AAM11545; and for Doc3, AAM11543. A Doc1 protein sequenceidentical to ours has previously been submitted by R. Murakami andT. Hamaguchi (AB035412).

3191Dorsocross genes in dorsal tissue development

top

T-B

c3D

oc2

c1C

c2va

ria

c2va

ria

c2va

ria

1 k

8 k

CeTbx9

HsTbr1

HsTbr2

HsTbx21

1000

HsBra/T

HsTbx19

Byn

975

1000

HsTbx18

HsTbx22

HsTbx1

Org-1

HsTbx20

H15

H15r 1000

998

1000

650

HsTbx4

HsTbx5

HsTbx2

HsTbx3

Omb

1000

514

669

DrTbx16/spadetail

XlTbx16/VegT

DrTbx6

990

HsTbx6

Doc2

Doc3

Doc1 936

1000

488

603

684

281

726

945

998

0.05

x6su

bfa

mily

x2su

bfa

mily

x1su

bfa

mily

Tsu

bfa

mily

rsu

bfa

mily

Veg

Tsu

bfa

mily

Doc

oc3

Doc

sTbx

rTbx

lTbx

16/V

egT

DrT

bx16

/spt

b H

sTbx

2 H

sTbx

3 H

sTbx

5 H

HsT

bx20

HsT

bx1

HsT

bx22

HsT

bx18

HsT

bx19

HsB

ra/T

HsT

bx21

HsT

br1

HsT

br2

x6su

bfa

mily

x2su

bfa

mily

x1su

bfa

mily

Tsu

bfa

mily

rsu

bfa

mily

Veg

Tsu

bfa

mily

H15

Org

-1

1020

3040

5060

7080

9010

W-

-S

W-

-V

G-

-L

D-

-C

-V

G-

-L

D-

-C

G-

-L

N-

-H

G-

-L

G-

-L

-Q

-V

-Q

-V

-Q

-I

-V

P-

-R

-I

R-

-I

-R

-A

-I

R-

-I

R-

-V

I-

-I

N-

-L

-I

-E

-T

-I

-Y

-P

-I

E-

T-

-R

-I

R-

-R

G-

-S

E-

-R

Q-

-I

-N

-Q

-E

Doc

oc3

Doc

sTbx

rTbx

lTbx

16/V

egT

DrT

bx16

/spt

b H

sTbx

2 H

sTbx

3 H

sTbx

5 H

HsT

bx20

HsT

bx1

HsT

bx22

HsT

bx18

HsT

bx19

HsB

ra/T

HsT

bx21

HsT

br1

HsT

br2

x6su

bfa

mily

x2su

bfa

mily

x1su

bfa

mily

Tsu

bfa

mily

rsu

bfa

mily

Veg

Tsu

bfa

mily

H15

Org

-1

110

Q Q Q M M M M M M M M M R R L M M M M M M M M M M

3192

not shown). Additional sequence comparisons with T-boxdomains from vertebrates and phylogenetic analysis show thatthe Doc T-box domains are most closely related to those frommembers of the Tbx6 subfamily of T-box proteins (Fig. 1B,C)(Papaioannou, 2001) (see Discussion).

Dorsocross expression is prominent in dorsaltissues during embryogenesisNorthern analysis with gene-specific probes showed that allthree Doc genes display similar expression profiles duringdevelopment with maximal levels occurring between 2 and 12hours of embryonic development and lower levels during lateembryonic, larval and pupal stages. The only significantdifference among the three genes in this assay was theexpression of Doc1mRNA in adult males, which was notobserved for Doc2and Doc3(data not shown).

The spatial expression patterns of Doc products in embryoswere examined by whole-mount in situ hybridization withgene-specific probes and whole-mount immunocytochemistryusing antibodies raised against the unique C-terminal regionsof the Doc proteins (see Materials and Methods). As all threegenes were found to have essentially identical expressionpatterns (with some minor differences regarding the relativelevels of expression in different tissues; data not shown), wewill henceforth collectively refer to them as ‘Doc genes’.As expected, Doc proteins are exclusively nuclear duringinterphase.

The initial expression of Doc genes is observed at thecellular blastoderm stage in a transverse stripe encompassingthe dorsal ~40% of the embryonic circumference within theprospective head region. Shortly later, a narrow longitudinalstripe of expression appears, which ultimately extends all alongthe dorsal midline of the embryo, and the joint domains forma cross-shaped pattern of Doc expression in dorsal areas of theearly embryo (Fig. 2A). The domain of the transverse stripe islocated anteriorly to the cephalic furrow forming duringgastrulation (Fig. 2B; Fig. 3G) and largely corresponds toprocephalic neuroectoderm. The cells of this domain continueDoc expression until stage 11, when the segregation ofprocephalic neuroblasts is completed (Fig. 2B-D) (Campos-Ortega and Hartenstein, 1997). By contrast, the cells from thedorsal longitudinal domain within the trunk region give rise toamnioserosa, which maintains strong Doc expression untilstage 15 (Fig. 2C-F). In addition, the cells from the anteriorand posterior termini within this longitudinal stripe contributeto regions of the anterior and posterior digestive tract andmaintain expression until stage 11.

During stages 9 and 10, a new pattern of Doc expressionemerges within dorsolateral areas of the germ band fromparasegment (PS) 1 to PS13, which consists of 13 rectangularcell clusters (Fig. 2D). This metameric expression includesectodermal as well as underlying mesodermal cells. In theectoderm, Doc expression is excluded from the dorsalmostcells near the amnioserosa at this stage, whereas in theectoderm the metameric Doc expression extends to thedorsalmost areas of this germ layer (Fig. 2D; see below). Todetermine the segmental register of Doc expression in themesoderm we co-stained with antibodies against the POUdomain transcription factor Cf1a (Vvl – FlyBase), which markthe tracheal placodes (Anderson et al., 1995). As shown in Fig.2E, Doc and Cf1a are expressed in mutually exclusive

domains, which implies that Doc expression encompassesprospective tissues of the lateral epidermis and dorsolateralsensory organs. After stage 11, the segmental expression in theepidermis is modified to form segmental stripes that areinterrupted in dorsolateral regions. Within these stripes, Docexpression is largely found in posterior areas of the anteriorcompartments of each segment and there is a gradeddistribution of Doc expression with increasing levels towardsthe posterior of each stripe (Fig. 2F,G). The dorsal epidermalexpression domains now extend to the amnioserosa, and afterdorsal closure the bilateral domains merge at the dorsal midline(Fig. 2F-H).

Additional sites of Doc expression during lateembryogenesis include the dorsal pouch in the embryonic head(Fig. 2H), the anterior pair of Malpighian tubules (Fig. 2I) andthe pentascolopidial chordotonal sensory organs (Fig. 2J). Themesodermal expression of the Doc genes, which will bepresented elsewhere in more detail along with functional data,is observed in areas between the expression domains of thehomeobox gene bagpipe (bap) at stage 10 (Fig. 2K). Thislocation defines them as dorsal areas of the mesodermal A (orslp) domains (Azpiazu et al., 1996; Riechmann et al., 1997),which include the dorsal somatic and cardiogenic mesoderm.During early stage 11, additional Doc expression initiates inthe caudal visceral mesoderm, which contains the founder cellsof the longitudinal muscles of the midgut (Fig. 2L) (San Martinet al., 2001; Klapper et al., 2002). As reported previously forDoc1, two out of six bilateral cardioblasts in each segment ofthe dorsal vessel, which are tin negative and svppositive, alsoexpress the Doc genes (Fig. 2H) (Lo and Frasch, 2001).

Doc expression along the dorsal midline dependson dpp and zenAs peak levels of Dpp activity are known to be required forcell fate determination at the dorsal midline, we tested whetherthere is a correlation between Dpp activity and dorsallongitudinal Doc expression during blastoderm stages. Asshown in Fig. 3A, double-staining for Doc mRNA andphosphorylated Mad (PMad) indicates a close correlationbetween cells containing high levels of PMad and Doc productswithin the dorsal-longitudinal stripe. In addition, faint Docsignals that are modulated in a pair-rule pattern extend intoareas that receive lower Dpp inputs and lack detectable PMad(Fig. 3A). As predicted, both PMad and Doc expression in thedorsal-longitudinal stripe, but not the dorsal-transverse headstripe of Doc expression, are absent in dpp-null mutantembryos (Fig. 3B). Conversely, in blastoderm embryos withubiquitous Dpp expression (UAS-dppactivated by maternallyprovided nanos-GAL4), we observe a significant widening ofthe dorsal-longitudinal stripes of PMad and Doc expression,during which the correlation between high PMad and DocmRNA levels is still maintained (Fig. 3C).

The expansion of PMad upon uniform ectopic expression ofdpp includes the prospective mesoderm, although notventrolateral areas of the blastoderm embryo (Fig. 3D).However, high PMad in the prospective mesoderm does nottrigger ectopic Doc expression, suggesting either the presenceof a ventral repressor or the requirement for a co-activator indorsal areas. A candidate for a co-activator is the homeoboxgene zerknüllt (zen). Double in situ hybridization shows thatthe appearance of dorsal Doc mRNAs coincides with the time

I. Reim, H.-H. Lee and M. Frasch

3193Dorsocross genes in dorsal tissue development

when zenmRNA levels increase in the areas of the presumptiveamnioserosa as a result of high Dpp inputs (Fig. 3E). When therefinement of zenexpression is completed, there is an exactcorrespondence in the widths of the Doc and zenexpressiondomains, although Doc expression extends more posteriorly(Fig. 3F). As shown in Fig. 3H (compare with Fig. 3G), theactivity of zenis necessary for normal levels of Doc expressionin the dorsal-longitudinal stripe, because in zen mutantembryos there are only low residual levels of Doc productspresent in this domain. These observations suggest that Docexpression along the dorsal midline of blastoderm embryosrequires the combined activities of dppand zen.

Metameric Doc expression in dorsal ectoderm andmesoderm requires Dpp + WgThe known distribution of dppmRNA during its second phaseof expression in the dorsolateral ectoderm of stage 9-11embryos (St Johnston and Gelbart, 1987) suggests that Docexpression in the dorsolateral ectoderm and mesoderm duringthese stages is also dependent on Dpp activity. As expected

from the known fate map shifts in dppmutants, these domainsof Doc expression are missing in dpp-null mutant embryos(data not shown). Notably, the exact coincidence between theventral borders of the domains of dorsolateral Doc expressionand high nuclear PMad (Fig. 4A,B) suggests that Docexpression is directly controlled by Dpp-activated Smadproteins in the ectoderm and mesoderm during this stage.Additional evidence for this hypothesis comes fromexperiments with ectopic expression of dpp in the ventralectoderm of the Krüppel domain (by virtue of a modified Kr-GAL4driver) (Frasch, 1995), which results in the concomitantexpansion of PMad and the Doc expression stripes towards theventral midline (Fig. 4C).

In addition to the inputs from dpp, metameric Docexpression in dorsolateral areas of the germ band must dependon the activity of segmental regulators. A direct comparisonwith the expression of engrailed(en) shows that the clusters ofDoc expression straddle the compartmental borders. AlthoughDoc expression overlaps with en in the P compartments, abouttwo-thirds of the Doc expressing cells of each cluster are

Fig. 2.Embryonicexpression pattern ofDorsocross. NuclearDorsocross proteins weredetected by immunostainingwith anti-Doc3+2 antibody(see Materials and Methods)and visualized either withDAB (brown in A-D,F-I,L)or fluorescent secondaryantibodies (J). DorsocrossmRNA was detected by insitu hybridization withspecific probes for Doc3(E)or Doc1(K). Images fromfluorescent staining arecombined ectodermal (E),subepidermal (J) ormesodermal (K) confocalsections. Views are lateral(B,D,E,G,I-K) ordorsolateral (C,F) withdorsal upwards and anteriortowards the left, or dorsalwith anterior towards the left(A,H,L). Embryos areoriented the same way in allfigures and all non-flourescent images are takenusing Nomarski optics.(A) Blastoderm stage embryo, (B) stage 7 embryo and (C) stage 8 embryo showing Dorsocross protein in the anlagen of the amnioserosa (as)and the procephalic neuroectoderm (pne). Amnioserosa nuclei are distinguishable by their larger size. (D) At stage 10, a metameric expressionpattern (PS4, parasegment 4) has emerged in addition to continued expression in the amnioserosa. (E) Patches of Dorsocross expression (green)alternate with the tracheal placodes labeled by anti-Cf1a antibody staining (red) in the dorsolateral ectoderm of the embryonic trunk. (F,G) Atstage 14, epidermal stripes are visible dorsally and ventrolaterally. (H) Enlarged view of stage15-16 embryo that has completed dorsal closure,showing expression in dorsally fused epidermal stripes (de) and in the dorsal pouch (dp), and two pairs of cardioblasts per segment (cb).(I) Stage 16 embryo focussed on Doc expression in Malpighian tubules (arrowhead). (J) Dorsocross expression in the pentascolopidialchordotonal organs (arrowheads, green), which are marked by their staining with mab 22C10 (red). Dorsocross-positive cells are identified asligament cells based on their ventral juxtaposition to 22C10-labeled LCh5 neurons (arrowheads). Some epidermal Doc staining is also presentin this image. (K) The metameric expression as seen in D includes the dorsal mesoderm, where Dorsocross-expressing cells (green) alternatewith visceral mesoderm precursors that express bap(red). The combined sections include ectodermal expression seen in more radial areas [seebroken line between mesoderm (ms) and ectoderm (ec)]. (L) Stage 11 embryo focussed on the bilateral caudal visceral mesoderm anlagen(cvm) close to the posterior tip of the germ band.

3194

located in posterior areas of the A compartments (Fig. 4D). Inagreement with this allocation, we find that the metameric Docdomains are exactly centered on the stripes of Wingless (Wg)expression (Fig. 4E). The observed correlation of thesegmental registers of Wg and Doc makes wga good candidatefor an upstream regulator of Doc. As shown in Fig. 4F,dorsolateral Doc expression in the ectoderm and mesoderm iscompletely absent if wgis inactive. By contrast, deletion ofsloppy paired(slp), a known target of wgin the mesoderm anda wg feedback regulator in the ectoderm (Lee and Frasch,2000), results in a reduction, but not a complete loss ofmetameric DOC expression (data not shown). Hence, slpprobably affects Doc indirectly through its effect onectodermal wg expression. Altogether, our data suggest thatmetameric Doc expression in the ectoderm and mesoderm istriggered by the intersecting activities of Wg and Dpp.

The Doc genes are required for full differentiationand maintenance of amnioserosa cellsThe similarities in sequence and expression of the three Docgenes suggested functional redundancy among these genes.Because our molecular analysis of available deficiencies at66E-F showed that none of them uncovered all three genes(Fig. 5A) we used the flanking P-insertions EP(3)3556 and

EP(3)584 in attempts to delete the entire Doc gene cluster viamale recombination-induced mutagenesis (see Materials andMethods). Molecular mapping of the obtained deletionsdemonstrated that two of them, Df(3L)DocAand Df(3L)DocB,which were generated with the distally located insertionEP(3)3556and cause embryonic lethality, deleted all three Docgenes (Fig. 5A). As Df(3L)DocAdeletes the smallest numberof additional genes (CG5087, CG5194, CG5144, ArgkandCG4911), we describe the phenotypic analysis in the presentstudy using this deficiency, although the salient phenotypes arevery similar between Df(3L)DocA andDf(3L)DocB.

Additional genetic analysis showed that it is possible toobtain a small number of viable adult escapers with thegenotype Df(3L)Scf-R11/Df(3L)DocA, which indicates thatCG5087 is not absolutely required for viability, and thatDoc1 and Doc2 can functionally substitute for the loss ofDoc3. Similarly, the full viability of flies with the genotypeDf(3L)DocA/Df(3L)EP584MR2(Fig. 5A) shows that CG4911and the 5′exons of Argk(preceding the large intron) arealso not essential. Furthermore, we determined thatembryos with the genotypes Df(3L)Scf-R11/Df(3L)DocAandDf(3L)DocA/Df(3L)29A6 (which causes pupal lethality) do notdisplay any of the phenotypes described below for Df(3L)DocAhom*ozygous embryos. In summary, our genetic analysis shows

I. Reim, H.-H. Lee and M. Frasch

Fig. 3.Dorsocross expression along the dorsalmidline of blastoderm stage embryos requiresdppand zen. (A-D) Double-fluorescent stainingfor Doc3RNA and nuclear Phospho-Mad(PMad) protein. An antibody specific for theactivated (phosphorylated) form of the Dppeffector Mad allows monitoring Dpp activity.All views are dorsal, except the lateral view inD. The longitudinal stripe is absent fromhom*ozygous dppH46 embryos (B) whencompared with the wild type (A). (C,D) Uponinduction of ectopic Dpp via P{GAL4-nanos.NGT}40and UAS-dpp, nuclear PMadand expression of Doc3(as well as of Doc1andDoc2, not shown) appear in a significantlybroader longitudinal stripe when comparedwith A. PMad, but not Doc3, is also detected inventral cells (D). (E) Double in situhybridization for Doc3mRNA (green) and zenmRNA (red) shows that the dorsal stripe of Docexpression appears during the time whenrefined zenexpression is detected on top of theweaker broad zenpattern. (F) During mid stage5, when zentranscripts in the broad dorsaldomain have disappeared, and dorsal Doc3andzenare expressed at peak levels, the widths ofthe Doc3and zendomains are identical.(G) Dorsolateral view of heterozygouszen7/CyOembryo and (H) hom*ozygous zen-

mutant (zen7, also known as zenW36) stainedwith anti-Doc3+2 antibody. Very littleDorsocross protein is detectable along thedorsal AP axis in zenmutants. By contrast,expression is maintained in the head stripe and at the termini (foregut and hindgut primordia, more prominently at later stages). Embryos inA,B,G,H were analyzed at the beginning of gastrulation, when wild-type embryos have a fully formed dorsal stripe.

3195Dorsocross genes in dorsal tissue development

that the loss of either Doc3or Doc1can be compensated forby the remaining two Doc genes in embryos and that thephenotypes described herein are a consequence of the loss ofall three of the Doc genes. However, we can not rule out acontribution of CG5194, which encodes a 128 amino acidpredicted ORF with no known hom*ology, to the observedphenotypes.

Because of the prominent Doc expression in the primordiaand developing tissue of the amnioserosa we used theamnioserosa marker Krüppel(Kr) to examine whether the Docgenes are required for the development of this extra-embryonictissue. These experiments demonstrated that hom*ozygousDf(3L)DocA mutant embryos (henceforth called DocAmutants) fail to express Krin the amnioserosa at any stage,whereas CNS expression of Kris not affected (Fig. 5B). Toconfirm that this observed phenotype is due to the loss of Docgene function we diminished Doc gene functions by usingRNA interference (RNAi) as an independent assay andperformed rescue experiments with DocAmutants (see below).As shown in the example of Fig. 5C, injection of a mixture ofequimolar amounts of dsRNAs for all three Doc genes (seeMaterials and Methods) frequently results in a completeabsence of Krexpression in the amnioserosa. The remainingembryos display strongly reduced numbers of Kr-containingnuclei in this tissue (data not shown). These phenotypescorrelate with the observed absence or severe reduction of Docprotein levels in Doc RNAi embryos (data not shown). Bycontrast, mock-injected embryos display normal expressionof Kr in the amnioserosa (Fig. 5D). Hence, the strongestphenotype obtained by RNAi mimics the observed DocAmutant phenotype, confirming that the lack of Kr expression inDocA mutant embryos is specifically due to the loss of theactivity of all three Doc genes.

Besides the effects on Krexpression, DocAmutant andRNAi-treated embryos share several morphological defects.The extending germ band is unable to displace the amnioserosafully towards the anterior and the posterior germ band is

therefore forced to bend underneath the amnioserosa. Of note,germ band retraction is strongly disrupted, which can beclearly seen in stage 14 embryos (Fig. 5E) and in cuticlepreparations of unhatched first instar larvae (Fig. 5H,I;compare with J). This phenotype is shared with previouslydescribed genes of the u-shaped (ush) group, which affect themaintenance of the amnioserosa (Frank and Rushlow, 1996).Kr expression and the germ band retraction defects in DocAmutant embryos can be partially rescued by expressing any ofthe three Doc genes with an early amnioserosa-specific driver(Fig. 5F,G). Rescue with Doc2 (Fig. 5F) is consistently moreefficient when compared with Doc1(data not shown) and Doc3(Fig. 5G), although it is not known whether this difference isdue to a higher intrinsic activity or a more efficient expressionof Doc2 protein in this assay.

An additional phenotype consists of reductions in the sizeof the embryonic head in DocAmutants and RNAi-treatedembryos, which is apparent from stage 12 onwards and resultsin reduced head structures and a frequent failure of headinvolution at later stages (Fig. 5B,C,H,I, and data not shown).This phenotype is probably due to excessive cell death as aconsequence of the absence of Doc activity in the procephalicneuroectoderm and other dorsal areas of the embryonichead (Fig. 2B,C and data not shown). The observed headphenotypes, as well as the aberrant shape of the filzkörper(Fig. 5I), are also reminiscent of similar phenotypes ofembryos mutant for genes of the ush group (Frank andRushlow, 1996).

To obtain more information about the particular role of theDoc genes in the specification and/or differentiation of theamnioserosa we analyzed the distribution of additionalamnioserosa markers in DocAmutant embryos. For the ushgroup gene hnt(Yip et al., 1997) we find a strong reduction ofexpression, with significant levels of Hnt protein only beingdetected in nuclei along the posterior margin of theamnioserosa (Fig. 6A, Fig. 7A, compare with Fig. 6B andFig. 7B, respectively). By contrast, the expression of the

Fig. 4. Combinatorial inputs from Dpp and Wgsignaling regulate metameric Dorsocross expressionin the dorsolateral ectoderm and mesoderm.(A,B) Doc3 in situ hybridization (green) and anti-PMad antibody staining (red) in wild-type stage 10embryos seen from lateral (A) or ventral (B) views(only trunk regions are shown in this figure;v, ventral midline). Doc3expression is activatedwithin the nuclear PMad domain. (C) Ventralextension of the ectodermal PMad domain byectopic Dpp leads to ventral extension of Doc3patches. Ectopic Dpp was driven from a UAS-dpptransgene by ZKr-GAL4 in the ectodermal Krüppeldomain (Frasch, 1995). (D,E) Wild-type stage 10-11embryos double-stained with anti-Doc3+2 antibody(green) plus anti-En/Invected (red) (D) and anti-Doc3+2 antibody (green) plus anti-Wg (red) (E).Ectodermal Doc expression is centered around theWg-stripes and overlaps partially with the En stripes(yellow ectodermal signals). (F) hom*ozygous wgCX4

mutant embryo stained as in E. Metameric Docexpression is absent from the ectoderm andmesoderm, although amnioserosa expression is not affected. All images are combined confocal sections within the specified germ layer. Doc-expressing cells flanking the amnioserosa (as) in D, and to lesser extent in E, are mesodermal.

3196

amnioserosa marker race(Ance – FlyBase) (Tatei et al., 1995)is initiated normally in the primordium of the amnioserosa ofDocA mutant embryos, suggesting that the expression of therace upstream activator zerknüllt(zen) is also not disrupted

(data not shown). However, after embryonic stage 9, raceexpression is gradually lost in the amnioserosa of DocAmutantembryos and its residual mRNA distribution closely followsthat of Hnt (Fig. 7A, compare with 7B).

I. Reim, H.-H. Lee and M. Frasch

Fig. 5.Mutagenesis of the Dorsocross locus and phenotypic rescue experiments. (A) Genomic map of the Dorsocross region showing knownand predicted genes, deficiencies and P-insertions. Numbers indicate base pair coordinates of genomic sequences on chromosome 3L (BGPDrelease 3). Slashes indicate the omission of sequences owing to space limitations. Genes above the genomic line are transcribed from left toright and those below from right to left (introns are not displayed). EP(3)3556was used for generating Df(3L)DocAand Df(3L)DocB, andEP(3)584 (male-sterile insertion in bol, located within the omitted sequence) for Df(3L)EP584MR2. Relevant breakpoints of previously knowndeficiencies were mapped by PCR from hom*ozygous embryos. Dashes indicate uncertainty ranges of breakpoints. In situ hybridizationconfirmed the presence of mRNAs of all three Doc genes in Df(3L)Scf-R6, absence of Doc3mRNA in Df(3L)Scf-R11, absence of Doc1mRNAin Df(3L)29A6, and absence of all three Doc mRNAs in Df(3L)DocAand Df(3L)DocBhom*ozygotes. Df(3L)DocAand Df(3L)DocBcomplement female sterility associated with smg1, whereas Df(3L)Scf-R6and Df(3L)Scf-R11do not. Therefore upstreamsmgsequences(hatched), including alternative smgstart sites, are dispensable. The sequences at the breakpoints of Df(3L)DocA, Df(3L)DocBandDf(3L)EP584MR2 will be made accessible in FlyBase. (B-D) Mid stage 12 embryos stained with anti-Kr antibody. (B) Embryo hom*ozygousfor Df(3L)DocA (DocAmutant; composite of two focal planes), which lacks Kr in the amnioserosa (arrowheads). Kr in the nervous systemserves as an internal staining control (white asterisks). (C) Doc1+2+3RNAi embryo, which is a wild-type embryo injected with a mix of Doc1,Doc2and Doc3dsRNA (3′-fragments downstream of T-box) and shows a phenotype as with DocAmutants. (D) Control wild-type embryoinjected with buffer only. (E) Stage 14 DocAmutant embryo showing absence of Kr staining in the amnioserosa region (arrowhead) andincomplete germ band retraction. Somatic muscle staining of Kr in E-G is denoted by black asterisks. (F) Stage 14 DocAmutant embryo withforced expression of Doc2in the early amnioserosa (via c381-GAL4). Kr expression in the amnioserosa is rescued to a significant degree(arrow), as well as extended temporally. Retraction defects are fully rescued in this and the majority of other embryos. (G) Stage 14 DocAmutant embryo with forced expression of Doc3as in F. There is some rescue of Kr expression in the amnioserosa (arrow) but little rescue of thegerm band retraction defects. (H-J) Cuticle preparations of unhatched first instar larvae visualized by dark-field optics. (H) Df(3L)DocAmutantshave a u-shaped phenotype owing to the failure of germ band retraction. A similar cuticle phenotype is observed in Doc1+2+3RNAi embryos(I), but not in the wild-type control (J). as, amnioserosa; fk, filzkörper.

3197Dorsocross genes in dorsal tissue development

We also examined the expression of a novel amnioserosamarker, which is encoded by the homeobox gene C15. In thenormal situation, C15is expressed in the amnioserosa fromstage 7 until stage 17, when the amnioserosa undergoesapoptosis (Fig. 6D,F,H, and data not shown) (Campos-Ortegaand Hartenstein, 1997). In addition, from early stage 10onwards there is a narrow domain of expression at the leadingedge of the dorsal germ band, which later becomes segmental(Fig. 6F,H). In DocAmutant embryos, the level of C15expression in early amnioserosa cells is unaltered, whichallows us to use C15 protein as a marker for the developmentof this tissue in the absence of Doc activity.

Until stage 9, the large majority of amnioserosa nuclei inDocA mutant embryos appear large and flattened as in wild-type embryos (Fig. 6C, compare with 6D). Together with datafrom α-tubulin staining (not shown), this observationindicates that the amnioserosa cells begin to acquire thenormal features of a squamous epithelium (data not shown).However, the amnioserosa does not display a properly foldedmorphology during stages 8-10, and the posterior germ bandis forced to bend towards the inside in DocAmutant embryos(Fig. 6C, compare with 6D, and data not shown). In addition,some small nuclei become detectable within the amnioserosaduring this stage (Fig. 6C, arrow). Altogether, theseobservations indicate that the amnioserosa initiates itsdifferentiation process in the absence of Doc gene activity butfails to complete it, thus leading to morphological andfunctional abnormalities of this tissue towards the end of

germ band elongation. Much stronger alterations can beobserved during subsequent stages, when there are anincreasing number of C15-stained amnioserosa nuclei withmuch smaller diameters than regular amnioserosa nuclei. Atlate stage 12, almost all amnioserosa cells feature smallnuclei that are difficult to distinguish from dorsal epidermalcells (Fig. 6E, Fig. 7C, compare with Fig. 6F, Fig. 7D,respectively). Co-staining for raceindicates that it ispredominantly the cells with the small nuclei that lose raceexpression, while most normally-sized nuclei are stillsurrounded by racesignals (Fig. 7C, compare with D). Fromthis stage onwards, non-stained ‘holes’ appear in theamnioserosa and the number of C15-stained amnioserosanuclei decreases prematurely. Hence, unlike wild-typeembryos, stage 14 DocA mutant embryos are not covereddorsally by C15-stained amnioserosa cells (Fig. 6G, comparewith 6H). In addition to the observed alterations in theamnioserosa, the C15 expression domain at the leadingedge of the epidermis appears significantly broadened (Fig.6E).

We tested whether the increasing number of smaller nucleiin the amnioserosa of DocAmutant embryos is connectedwith abnormal cell divisions. As shown in Fig. 7E, the M-phase marker phospho-Histone H3 can be detected innumerous amnioserosa nuclei of DocAmutant embryos afterstage 10, which is not seen in wild-type embryos (Fig. 7F).In addition, there is significant incorporation of BrdU inamnioserosa nuclei of DocAmutant embryos (particularly

Fig. 6.Loss of Dorsocross causes abnormaldevelopment and premature breakdown of theamnioserosa. hom*ozygous Df(3L)DocAmutantembryos (A,C,E,G) and heterozygous control embryos(B,D,F,H) were stained for amnioserosa markers asindicated. (A,B) Expression of hnt, as determined withanti-Hnt antibodies, initiates in the absence ofDorsocross (A), but fails to the reach wild-type levelsespecially in anterior areas of the amnioserosa by stage9 (arrowheads; compare with B). Note comparablelevels of Hnt in midgut primordia (amg, pmg) in A andB. (C-H) Anti-C15 antibody staining of stage 9 (C,D),late stage 12 (E,F) and stage 14 (G,H) embryos. C15expression is not reduced in early DocAmutants (C)and large flattened nuclei are present in theamnioserosa similar to the wild-type situation (D).However, in DocAmutants there are some abnormallysmall nuclei in the amnioserosa (arrow) and germ bandelongation is slightly aberrant. (E) C15 staining of astage 12 DocAmutant embryo reveals a decrease in thesize of most amnioserosa nuclei (arrow) and abroadening of the ectodermal C15 expression domain(arrowhead) when compared with F. Around stage 14,no large amnioserosa cells expressing C15can befound in DocAmutants (G), there is a dorsal holelacking C15 (arrowhead) and the germ band is notretracted. In the control embryo (H), C15expression inthe amnioserosa is still strong during this stage and inthe dorsal ectoderm it shows a well-defined segmentedpattern. DocAmutants were initially identified viaabsence of balancer-derived anti-β-galactosidase staining. Various morphological features (reduction of dorsal head structures, incomplete germband extension associated with an inwardly kinked posterior germ band, absence of germ band retraction, yolk displacement) were found to beconsistently present in mutants, which allows reliable discrimination of mutant and control embryos without anti-β-galactosidase staining afterstage 8.

3198

in the small nuclei; Fig. 7G), whereas no incorporationis observed in wild-type embryos (Fig. 7H). Mitotic

spindles are also present in the amnioserosa of DocAmutants (Fig. 7I, compare with J). These observationsindicate that the normal G2 arrest of amnioserosa cellshas been released and the cells re-enter the cell cycle.We also examined whether the subsequentdisappearance of small C15-stained amnioserosa nucleiin DocA mutant embryos is a result of prematureapoptosis of cells in this tissue. This possibility wasconfirmed by the results of TUNEL labelingexperiments, which produced signals in manyamnioserosa nuclei from 12 onwards. Most of theTUNEL-labeled nuclei have reduced or are lackingC15 expression (Fig. 7K, compare with 7L, whichshows that wild-type amnioserosa nuclei at latestage 12 are not apoptotic). Altogether, theseobservations suggest that loss of Doc activity preventsthe normal differentiation of the amnioserosa to a fullyfunctional tissue, suspends the cell cycle block of

amnioserosa cells, and causes premature apoptotic cell deathin this tissue.

I. Reim, H.-H. Lee and M. Frasch

Fig. 7.Abnormal marker expression, cell cycle entry andpremature apopotosis of amnioserosa cells in Doc mutantembryos. Shown are confocal fluorescent microscopic imageswith merged amnioserosa scans. (A-D) racein situhybridization (green) and antibody staining (red) with anti-Hnt (A,B) and anti-C15 (C,D) at stage 12. raceexpressiondecreases in the amnioserosa of DocAmutants. Residualexpression of racegenerally correlates with residual Hntexpression (arrow in A) and large C15-stained nuclei (arrowin C; arrowheads, small nuclei). In addition, there areincreased levels of racemRNA in the posterior/dorsal head(asterisk) of DocAembryos. (E,F) Stage 10 embryos stainedfor racemRNA and phospho-Histone H3. Unlike in the wild-type (F), race-stained amnioserosa cells show nuclearphospho-Histone H3 staining in DocAmutants (E, arrowheads). (G,H) Stage 11 embryos after 30 minutes BrdU pulselabeling, double-stained with anti-BrdU (green) and anti-C15antibodies (red) to visualize amnioserosa nuclei. Overlappingsignals appear yellow. While normal amnioserosa cells arearrested in the G2 phase of cell cycle 14 and do notincorporate BrdU (H), BrdU incorporation is detected in afraction of amnioserosa cells of DocAmutants (arrowheads inG). BrdU incorporation in dorsal ectodermal cells flanking theamnioserosa and other domains is seen in both mutant andwild-type embryos at this stage. (I,J) Stage 11 embryosstained for C15 (red), DNA (Hoechst, blue) and α-tubulin(green) after fixation in the presence of taxol. Mitotic spindlesin the amnioserosa of DocAmutants (I, arrowheads) indicatedividing amnioserosa cells, which are not seen in wild-typeembryos (J). (K,L) Detection of apoptotic cell death in latestage 12 embryos by TUNEL is shown in green and stainingwith anti-C15 is shown in red. At this stage, there issignificant apoptosis within the amnioserosa in DocAmutants(K, arrowheads), but none in wild type (L), althoughapoptosis can be detected in the head and other regions ofwild-type embryos. Most apoptotic amnioserosa cells havealready lost C15 expression, but are clearly localized withinthe amniosera layer. Mutant embryos were identified bytriple-staining with anti-β-galactosidase (A-D; not shown) orby the absence of anti-Doc2 staining performed in parallel toanti-C15 staining (G,I,K). G,H are at 1.75× greatermagnification than A-F,K,L; I,J are at 2.5×greatermagnification than A-F,K,L.

3199Dorsocross genes in dorsal tissue development

Doc patterns the lateral ectoderm via repressing wgand ladybirdThe segmental stripes of wgexpression in the embryonic trunksegments initially span the entire dorsoventral extent of theectoderm, but at stage 11 they become interrupted indorsolateral areas (Baker, 1988). A comparison of Wg and Docexpression at this stage shows that the positions of themetameric ectodermal domains of Doc expression correspondto the areas in which the Wg stripes become interrupted (Fig.8A). Temporally, there is a brief overlap of ectodermal Wg andDoc expression during stage 10 until Wg expression isdownregulated within the Doc domains (see Fig. 4E). Incontrast to the wild-type situation (Fig. 8A,C), the Wg stripesremain continuous in DocAmutant embryos (Fig. 8B). Similarobservations were made with the homeobox gene productLadybird (Lb=Lbe + Lbl) (Jagla et al., 1997) as a marker. Inwild-type embryos after stage 11, Lb is also expressed in stripeddomains that are interrupted at the positions of the ectodermalDoc domains (Fig. 8D), whereas in DocAmutant embryos thereis ectopic expression in a pattern of continuous stripes (Fig. 8E,compare with F). These data show that Doc activity is requiredfor patterning events in the dorsolateral ectoderm, whichinclude the repression of wgand lbexpression in these areas.

Ectopic expression experiments with Doc genes provideadditional evidence for a repressive activity of Doc on wgexpression. Upon ectopic expression of Doc2 in all cells of theectoderm of wild type embryos, the ventral portions of the Wgstripes are lost (Fig. 9B, compare with 9A). However, thedorsal regions of the Wg stripes appear to be under differentregulation, because ectopic Doc results in a uniform domain ofdorsal Wg along the anteroposterior axis, albeit at lower levelsthan in wild-type embryos.

Ectopic expression experiments with Doc genes in imaginaldiscs further confirm their ability to repress wg. In third instarlarval wing discs, Doc genes are expressed in four distinct areasthat do not overlap with the wg expression domains.Specifically, two large Doc expression domains are located inthe centers of the dorsal and ventral regions of the prospectivewing blades and two smaller domains in prospective dorsalhinge and posterior notal regions, respectively (Fig. 9C). In legdiscs, low levels of Doc expression can be detected in regionsof the prospective body wall and proximal leg segments, whichalso do not express wg (Fig. 9E). Importantly, ectopicexpression of Doc2within the Dpp domains of imaginal discscauses wg expression to disappear in the corresponding areas(Fig. 9D,F). In agreement with the known role of wg in limbdevelopment (Lecuit and Cohen, 1997), its repression byectopic Doc results in the loss of distal structures of wings, legsand antenna of adult animals (Fig. 9G-I). Analogous ectopicexpression experiments with Doc1and Doc3in embryos anddiscs produced qualitatively similar, although weaker, effectsto Doc2.

DISCUSSION

The closely related T-box sequences, genomic clustering andvirtually identical expression patterns of the three Dorsocrossgenes suggest that they are derived from relatively recentduplications of a common progenitor gene. Accordingly, ourobservation that loss of Doc1 or Doc3does not cause any ofthe embryonic phenotypes seen upon loss of all three genesindicates that there is a large degree of functional redundancyamong these three genes. Phylogenetic analysis with the

Fig. 8. Dorsocross regulates the patterning of the dorsolateral ectoderm. (A) Fluorescent double-staining with anti-Doc3+2 antibodies (green)and anti-Wg (red). After an initial overlap of expression (see Fig. 4E), at stage 11 there is complementary expression of Doc and wgas the wgstripes become interrupted in the dorsolateral ectoderm. (B,C) Anti-Wg staining (brown; blue anti-β-galactosidase staining indicates TM3, eve-lacZ-balanced control embryos) of stage 11 embryos. Note persistence of continuous stripes in DocAmutants (B, arrowhead), which contrastswith interrupted stripes in control embryos (C, arrowheads). (D) Expression of Dorsocross (anti-Doc3+2, green) and ladybird (anti-Lbe, red) inthe embryonic trunk at stage 11/12. Doc patches are observed between Lb patches similar to the spatial arrangement seen in A. (E,F) Anti-Lbestaining (plus anti-β-galactosidase staining, blue) of stage 14 embryos. (E) Continuous lb stripes are visible in the ectoderm of DocAmutants.(F) Normal lbexpression pattern with separated dorsal patches and ventral stripes in the ectoderm in control embryos. Small groups of cellsexpressing lbin the lateral region correspond to muscle precursors.

3200

extended T-box domain sequences shows that the Doc genesare most closely related to the vertebrate Tbx6genes, whoseexpression in the paraxial mesoderm is reminiscent of theexpression of the Doc genes in the dorsal somatic mesoderm.However, the limited reliability of the branches separating theTbx6, VegT and Tbx2 subfamilies in the phylogenetic treeanalysis, the absence of Drosophilaorthologs of VegTandTbx4/5genes, as well as shared features of expression in thesomatic and/or precardiac and cardiac mesoderm seem tosupport the alternative possibility that the Doc, Tbx6, VegTandTbx4/5genes arose from a common ancestral gene by geneamplifactions after the divergence of the insect and vertebratelineages.

A prominent feature of the Doc genes is their expression inareas that receive inputs from Dpp, including the dorsalmostcells in blastoderm embryos, the dorsolateral ectoderm andmesoderm in the elongated germband, and distinct domainsspanning the compartment border of the wing disc. Indeed, ourgenetic data, together with the co-localization of Doctranscripts with active Mad in dorsal embryonic tissues, favorthe possibility that the Doc genes are direct targets of the Dppsignaling cascade. However, the Dpp signals are required to act

in combination with additional regulators during each of theseinductive events.

Regulation and function of the Doc genes duringamnioserosa developmentOur observations suggest that robust and stable induction ofDoc expression in a dorsal stripe requires the activity of thehomeodomain protein Zen as a co-activator of Dpp signals.The zengene features an early, broad expression domain alongthe dorsal embryonic circumference, which is initially Dppindependent but subsequently requires Dpp for it to bemaintained (Rushlow et al., 1987; Ray et al., 1991). Thereafter,its expression refines into a narrow dorsal domain in a processthat requires peak levels of Dpp (Rushlow and Levine, 1990;Ray et al., 1991; Rushlow et al., 2001). The activation of Docexpression occurs at the same time as the refinement of zenexpression and within the same narrow domain, which alsocoincides with high phospho-Mad levels (Rushlow et al.,2001). Although the maintenance and refinement of zen byDpp is zenindependent (Ray et al., 1991), we propose that Zensynergizes with peak signals of Dpp to trigger Doc geneexpression in a dorsal stripe. The requirement for this proposed

I. Reim, H.-H. Lee and M. Frasch

Fig. 9. Ectopic expression ofDorsocross represses wingless.(A) wgexpression in stage 11 controlembryo (UAS-Doc2/CyO).(B) Embryo of similar stageexpressing Doc2ectopically in thewhole ectoderm (e22c-GAL4/UAS-Doc2). Ventral wgexpression ismissing (see arrowhead) and dorsalwgexpression is almost continuous,although with reduced levels.(C,D) Dorsocross (red) and wg(green) expression in imaginal wingdiscs from 3rd instar larvae detectedby flourescent double-staining usinganti-Doc2+3 and anti-Wg antibodies.Dorsal is upwards and anteriortowards the left. (C) Wild-type wingdisc. (D) Wing discs ectopicallyexpressing Doc2in the dppexpression domain (UAS-Doc2/+;dpp.blk1-GAL4/+). Note theinterruption of wgexpression at theintersection of ectopic Doc2 and Wgstripes (arrowheads). (E,F) Leg discof 3rd instar stained as in C and D.(E) Wild-type leg disc with wgexpression in anterior/ventralterritories. Endogenous Doc proteinin dorsal proximal areas of the disc isdetected at low levels (redarrowheads). (F) Leg disc from UAS-Doc2/+;dpp.blk1-GAL4/+larvae. wgexpression is repressed by ectopicDoc2 in the central area thatnormally produces distal parts of theleg (arrowheads). (G) Wild-typewing. (H) Wing derived from a discwith a genotype as in D, which lacks distal structures (arrowheads). Wing veins appear broadened as well. (I) Adult male fly showing aphenotype of intermediate strength caused by ectopic dpp.blk1-driven Doc2. Arrowheads indicate aristaless antennae and shortened legs.

3201Dorsocross genes in dorsal tissue development

interaction between zenand dppwould explain the failure ofzento activate Doc genes in an early, broad domain as well asthe observed low levels of residual Doc expression in zenmutant embryos, which may be due to inputs from Dpp alone.Formally, this proposed mechanism would be analogous topreviously described inductive events in the early dorsalmesoderm, where the synergistic activities of thehomeodomain protein Tinman and activated Smads induce theexpression of downstream targets such as even-skipped (Halfonet al., 2000; Knirr and Frasch, 2001). The identification offunctional binding sites for Zen and Smads in Doc enhancerelement(s) will be necessary for demonstrating that ananalogous mechanism is active during induction of Doc geneexpression in a dorsal stripe. In the absence of such data, wecan not completely rule out that dorsal Doc expression iscontrolled indirectly by Dpp, possibly via the combinatorialactivities of zenand another high-level target gene of Dpp. Asmutations in several other genes that are expressed in the earlyamnioserosa, including pannier (pnr), hnt, srp, tupand ush, donot affect Doc expression until at least stage 12 (I.R. and M.F.,unpublished), these genes can be excluded as candidates forearly upstream regulators of Doc.

Unlike zen, which is expressed only transiently, Docexpression is maintained throughout amnioserosa development.Hence, the Doc genes provide a functional link between theearly patterning and specification events in dorsal areas of theblastoderm embryo and the subsequent events of amnioserosadifferentiation. The activity of zenis required for all aspects ofamnioserosa development that have been examined to date,including normal activation of C15(M.F., unpublished). Bycontrast, our data demonstrate that the Doc genes execute onlya subset of the functions of zen, which includes the activationof Kr and hnt, but not C15and early race, in amnioserosa cells.This interpretation is consistent with our failure to obtain asignificant increase of amnioserosa cells upon ectopicexpression of any of the Doc genes in the ectoderm orthroughout the early embryo (using e22c and nanos-GAL4drivers, respectively; I.R. and M.F., unpublished). The residualexpression of hntin some amnioserosa cells of Doc mutantembryos could be due to direct inputs from zenitself or from ayet undefined zendownstream gene acting in parallel with Doc.Nonetheless, the strong reduction of hnt expression in Docmutant embryos could largely account for their amnioserosa-related phenotypes, including the absence of Kr expression, thedecline of raceexpression, premature apoptosis and failure ofgerm band retraction. All of these phenotypes have also beenobserved in hntmutant embryos (Wieschaus et al., 1984; Frankand Rushlow, 1996; Lamka and Lipsh*tz, 1999; Yip et al.,1997). However, it is likely that Doc gene activity is requiredfor the activation not only of hntbut also of additional genes ofthe u-shaped group and that Doc genes exert some of theirfunctions in parallel with hnt. Some evidence for this notion isderived from the observation that loss of Doc activity has astronger effect on Krexpression than loss of hnt activity.

One of the hallmarks of amnioserosa development is that thecells of this tissue never resume mitotic divisions after theblastoderm divisions (Campos-Ortega and Hartenstein, 1997).To a large extent, this cell cycle arrest is due to the absence ofexpression of cdc25/string in the prospective amnioserosa,which prevents the cells from entering M-phase and leads to G2arrest (Edgar and O’Farrell, 1989). In addition, the expression

of the Cdk inhibitor p21/Dacapo in the early amnioserosa isthought to contribute to the cell cycle arrest (de Nooij et al.,1996; Lane et al., 1996). Although a detailed description of theregulation of stringand dacapoexpression in dorsal embryonicareas is lacking, it has been reported that zenis required forrepressing dorsal stringexpression, which is expected to preventfurther cell divisions (Edgar and O’Farrell, 1990; Edgar et al.,1994). Notably, our observation that amnioserosa cells re-enterthe cell cycle in Doc mutant embryos demonstrate that Docgenes are required for the cell cycle block in addition to zen.Whereas zenmutant embryos feature ectopic cell divisions indorsal areas already from stage 8 onwards (Arora and Nüsslein-Volhard, 1992), in Doc mutants the amnioserosa cells resumemitosis only during and after stage 10, which is shortly after Zenprotein disappears. Thus, we hypothesize that the Doc genes takeover the function of zenin repressing stringand prevent celldivisions at later stages of amnioserosa development when Zenis no longer present. Overall, the phenotype of Doc mutantembryos suggests that amnioserosa differentiation, includingcell cycle arrest and the development of squamous epithelialfeatures, initiates in the absence of Doc activity but is notmaintained beyond stage 11. Thereafter, cell division resumesand there is a reversal of the partially differentiated state.Apoptotic events are not observed prior to stage 11 in Docmutants. However at later stages, many amnioserosa cells dieprematurely and the remaining cells are difficult to distinguishmorphologically from dorsal ectodermal cells.

Altogether, our studies have identified the Doc genes as newmembers of the u-shaped group of genes, which controlamnioserosa development, and provide new insights into theregulatory pathways in amnioserosa development downstreamof Dpp (summarized in Fig. 10) (see also Rusch and Levine,1997). In future studies, it will be necessary to define thespecific roles of the remaining genes of the u-shaped family,particularly ush, srp, tupand C15, in this regulatory frameworkin more detail.

Regulation and function of the Doc genes duringpatterning of lateral epidermis and dorsal mesodermUnlike in the presumptive amnioserosa, not all cells in the

Fig. 10.Summary of currently known regulatory networks duringamnioserosa development. Dorsocross genes act downstream of Dppand zen, and upstream of hntand Kr. Other pathways, which includeC15, act in parallel with Dorsocross (see Discussion).

dpp

zen Doc

racestringcell

prolif.

C15

hnt string?cell

prolif.

race

stage 5

stages 6-8

stages 6-14

stages 7-14

dors

albr

oad

amn

ios

ero

dors

alst

ripe

3202

dorsolateral ectoderm and dorsal mesoderm that receive highlevels of Dpp induce Doc expression. Rather, Wg signals arerequired in combination with Dpp in these tissues, such thatthe Doc genes are induced at the intersections of transverse Wgstripes and the dorsally restricted domain containing highphospho-Mad levels. The Doc stripes extend beyond the peaklevels of Wg on both sides of the Wg stripes, which indicatesthat the Doc genes are able to respond to relatively low levelsof diffusible Wg. In addition, the absence of Doc expressionin the dorsalmost cells of the ectoderm that receive Wg andDpp signals indicates the presence of a negative regulator thatprevents Doc induction in the ectoderm adjacent to theamnioserosa until stage 12. Together, these inputs restrict Docexpression to metameric quadrants that encompass the areas ofthe dorsolateral ectoderm between the tracheal placodes as wellas the underlying mesodermal cells.

Previous studies have shown that some of the effects of wgare mediated by its target gene sloppy paired(slp), includingthe feedback activation of wgin the ectoderm and therepression of bagpipe(bap) in the mesoderm (Cadigan et al.,1994; Lee and Frasch, 2000). However, the residual (althoughstrongly reduced) expression of the Doc genes in the germ bandof slpmutant embryos argues against a role of slp in mediatingthe function of wgto induce the Doc genes. Hence, the Docgenes may be direct targets of the Wg signaling cascade in theectoderm and mesoderm.

Our observations show that one of the important functionsof the Doc genes in the dorsolateral ectoderm is the repressionof wg expression. Although the expression of Doc initiallydepends on wg, the Doc genes subsequently exert a negativefeedback on wgexpression, which leads to the previouslyunexplained interruption of the wg stripes during stage 11.Because the ventral extent of the ectodermal Doc domainscorrelates with the ventral borders of high levels of P-Mad, weconclude that the dorsal limit of the ventral wgstripes at stage11 is determined indirectly by Dpp via Doc.

The maintenance of wgafter stage 10 has been shown todepend on two different positive feedback loops, one beingactive in the dorsal and the other in the ventral ectoderm. Thedorsal feedback loop is mediated by the ladybird homeoboxgenes (lb=lbeand lbl), whereas the ventral loop is mediated bythe Pax gene gooseberry(gsb) (Li and Noll, 1993; Jagla et al.,1997). The Doc genes must interrupt one or both of thesefeedback loops, although it is not clear whether the primaryblock is at the level of the wg gene or at the level of thetranscription factor-encoding genes lband/or gsb. Anothertarget for repression by the Doc genes in this pathway couldbe slp, which is required both dorsally and ventrally in wgfeedback regulation (Cadigan et al., 1994; Lee and Frasch,2000). We think that lbis unlikely to be the primary target ofDoc repression as the failure of wg repression temporallyprecedes the expansion of the lb stripes in Doc mutantembryos. Furthermore, our observation that the Doc genes canalso repress wgin other tissue contexts such as the imaginaldiscs, where gsb, lband slp are not components of a wgfeedback loop, seems to favor the mechanism of a directrepression of the wggene by Doc.

Taken together, our observations show that dynamicinteractions among positive and negative feedback loops, whichshare wg as a common component, are involved in thedorsoventral and anteroposterior patterning of the embryonic

ectoderm. The activity of the Doc genes in negatively regulatingwg and lb, as well as their potential positive effects on yetunknown targets in the dorsolateral ectoderm, are expected tobe important for the proper dorsoventral organization of thecuticle and sensory organs. In the mesoderm, the metamericexpression domains of the Doc genes during stages 9-11 includethe dorsal somatic and cardiac mesoderm. Notably, preliminaryanalysis has revealed defects in dorsal somatic muscle anddorsal vessel development in Doc mutant embryos, which weare currently examining in more detail. Finally, we note that theexpression pattern of the Doc genes in the embryonic epidermisis very reminiscent of the pattern of expression and activity ofanother T-box gene, optomotor-blind(omb), in the pupalepidermis (Kopp and Duncan, 1997). Doc and ombexpressionoverlap in the wing discs although, unlike omb, Doc expressionis interrupted near the Wg domains (Grimm and Pflugfelder,1996). Furthermore, it has been reported that dominantmutations in the gene Scruffy(Scf) and their revertantsgenetically interact with omb during abdominal cuticle andwing patterning (Kopp and Duncan, 1997). Because we havemapped the breakpoints of two Scfrevertants, Df(3L)Scf-R6andScf-R11, directly upstream and downstream, respectively, of theDoc3gene, it is tempting to speculate that the Scfphenotype iscaused by rearranged Doc3. Future studies will clarify therelationship between Scfand Doc genes and establish whetherthe T-box genes Doc and omb functionally interact duringpatterning of the adult cuticle and wings.

We thank the Bloomington stock center for fly stocks; K. Jagla, D.Kosman, J. Reinitz, C. Rushlow and the Developmental StudiesHybridoma Bank (University of Iowa/NICHD) for antibodies andprobes; and D. Kosman for the double-in situ hybridization protocol.This research was supported by a grant to M.F. from the NationalInstitutes of Health (HD30832). The Mount Sinai ConfocalMicroscopy Shared Resource Facility was supported, in part, withfunding from NIH-NCI shared resources grant (1 R24 CA095823-01).

REFERENCES

Adams, M. D., Celniker, S. E., Holt, R. A., Evans, C. A., Gocayne, J. D.,Amanatides, P. G., Scherer, S. E., Li, P. W., Hoskins, R. A., Galle, R. F.et al. (2000). The genome sequence of Drosophila melanogaster. Science287, 2185-2195.

Anderson, M., Perkins, G., Chittick, P., Shrigley, R. and Johnson, W.(1995). drifter, a DrosophilaPOU-domain transcription factor, is requiredfor correct differentiation and migration of tracheal cells and midline glia.Genes Dev.9, 123-137.

Arora, K., Levine, M. and O’Connor, M. (1994). The screwgene encodesa ubiquitously expressed member of the TGF-β family required forspecification of dorsal cell fates in the Drosophilaembryo. Genes Dev.8,2588-2601.

Arora, K. and Nüsslein-Volhard, C. (1992). Altered mitotic domains revealfate map changes in Drosophila embryos mutant for zygotic dorsoventralpatterning genes. Development114, 1003-1024.

Ashe, H. and Levine, M. (1999). Local inhibition and long-rangeenhancement of Dpp signal transduction by Sog. Nature398, 427-431.

Ashe, H., Mannervik, M. and Levine, M. (2000). Dpp signaling thresholdsin the dorsal ectoderm of the Drosophilaembryo. Development127, 3305-3312.

Azpiazu, N. and Frasch, M.(1993). tinmanand bagpipe: two homeo boxgenes that determine cell fates in the dorsal mesoderm of Drosophila. GenesDev.7, 1325-1340.

Azpiazu, N., Lawrence, P., Vincent, J.-P. and Frasch, M.(1996).Segmentation and specification of the Drosophila mesoderm. Genes Dev.10, 3183-3194.

Baker, N. E. (1988). Transcription of the segment-polarity gene wingless in

I. Reim, H.-H. Lee and M. Frasch

3203Dorsocross genes in dorsal tissue development

the imaginal discs of Drosophila, and the phenotype of a pupal-lethalwingless mutation. Development102, 489-497.

Bodmer, R. (1993). The gene tinmanis required for specification of the heartand visceral muscles in Drosophila. Development118, 719-729.

Brand, A. H. and Perrimon, N. (1993). Targeted gene expression as a meansof altering cell fates and generating dominant phenotypes. Development118,401-415.

Brown, N. H. and Kafatos, F. C.(1988). Functional cDNA libraries fromDrosophilaembryos. J. Mol. Biol.203, 425-437.

Cadigan, K., Grossniklaus, U. and Gehring, W. (1994). Localizedexpression of sloppy pairedprotein maintains the polarity of Drosophilaparasegments. Development8, 899-913.

Campos-Ortega, J. A. and Hartenstein, V. (1997). The EmbryonicDevelopment of Drosophila melanogaster. Berlin: Springer Verlag.

Carmena, A., Gisselbrecht, S., Harrison, J., Jimenez, F. and Michelson, A.(1998). Combinatorial signaling codes for the progressive determination ofcell fates in the Drosophilaembryonic mesoderm. Genes Dev.15, 3910-3922.

Dahanukar, A., Walker, J. A. and Wharton, R. P.(1999). Smaug, a novelRNA-binding protein that operates a translational switch in Drosophila.Mol. Cell 4, 209-218.

de Nooij, J., Letendre, M. and Hariharan, I. (1996). A cyclin-dependentkinase inhibitor, Dacapo, is necessary for timely exit from the cell cycleduring Drosophilaembryogenesis. Cell 87, 1237-1247.

Decotto, E. and Ferguson, E.(2001). A positive role for Short gastrulationin modulating BMP signaling during dorsoventral patterning in theDrosophilaembryo. Development128, 3831-3841.

Doyle, H. J., Harding, K., Hoey, T. and Levine, M.(1986). Transcriptsencoded by a hom*oeo box gene are restricted to dorsal tissues of Drosophilaembryos. Nature323, 76-79.

Duan, H., Skeath, J. B. and Nguyen, H. T.(2001). DrosophilaLame duck,a novel member of the Gli superfamily, acts as a key regulator of myogenesisby controlling fusion-competent myoblast development. Development128,4489-4500.

Edgar, B., Lehman, D. and O’Farrell, P.(1994). Transcriptional regulationof string (cdc25): a link between developmental programming and the cellcycle. Development120, 3131-3143.

Edgar, B. and O’Farrell, P. (1989). Genetic control of cell division patternsin the Drosophilaembryo. Cell57, 177-187.

Edgar, B. A. and O’Farrell, P. H. (1990). The three postblastoderm cellcycles of Drosophilaembryogenesis are regulated in G2 by string. Cell62,469-480.

Ferguson, E. and Anderson, K. (1992a). decapentaplegic acts as amorphogen to organize dorsal-ventral pattern in the Drosophila embryo.Cell 71, 451-461.

Ferguson, E. L. and Anderson, K. V.(1992b). Localiced enhancement andrepression of the activity of the TGF-ß family member, decapentaplegic, isnecessary for dorsal-ventral pattern formation in the Drosophilaembryo.Development114, 583-597.

Francois, V., Solloway, M., O’Neill, J., Emery, J. and Bier, E.(1994).Dorsal-ventral patterning of the Drosophilaembryo depends on a putativenegative growth factor encoded by the short gastrulationgene. Genes Dev.8, 2602-2616.

Frank, L. and Rushlow, C. (1996). A group of genes required formaintenance of amnioserosa tissue in Drosophila. Development122, 1343-1352.

Frasch, M. (1995). Induction of visceral and cardiac mesoderm by ectodermalDpp in the early Drosophilaembryo. Nature374, 464-467.

Garcia Abreu, J., Coffinier, C., Larrain, J., Oelgeschlager, M. and deRobertis, E. (2002). Chordin-like CR domains and the regulation ofevolutionarily conserved extracellular signaling systems. Gene287, 39-47.

Goldman-Levi, R., Miller, C., Greenberg, G., Gabai, E. and Zak, N.(1996).Cellular pathways acting along the germband and in the amnioserosa mayparticipate in germband retraction of the Drosophila melanogasterembryo.Int. J. Dev. Biol.40, 1043-1051.

Grimm, S. and Pflugfelder, G.(1996). Control of the gene optomotor-blindin Drosophilawing development by decapentaplegicand wingless. Science271, 1601-1604.

Halfon, M., Carmena, A., Gisselbrecht, S., Sackerson, C., Jimenez, F.,Baylies, M. and Michelson, A. (2000). Ras pathway specificity isdetermined by the integration of multiple signal-activated and tissue-restricted transcription factors. Cell 103, 63-74.

Huang, A. M., Rehm, E. J. and Rubin, G. M. (2000). Recovery of DNAsequences flanking P-element insertions: inverse PCR and plasmid rescue.

In DrosophilaProtocols(ed. W. Sullivan, M. Ashburner and R. S. Hawley),pp. 429-438. Cold Spring Harbor, New York: Cold Spring HarborLaboratory Press.

Irish, V. F. and Gelbart, W. M. (1987). The decapentaplegicgene is requiredfor dorsal-ventral patterning of the Drosophilaembryo. Genes Dev.1, 868-879.

Jagla, K., Jagla, T., Heitzler, P., Dretzen, G., Bellard, F. and Bellard, M.(1997). ladybird, a tandem of homeobox genes that maintain late winglessexpression in terminal and dorsal epidermis of the Drosophila embryo.Development124, 91-100.

Jazwinska, A., Rushlow, C. and Roth, S.(1999). The role of brinker inmediating the graded response to Dpp in early Drosophila embryos.Development126, 3323-3334.

Kennerdell, J. R. and Carthew, R. W. (1998). Use of dsRNA-mediatedgenetic interference to demonstrate that frizzledand frizzled 2act in theWingless pathway. Cell95, 1017-1026.

Kirkpatrick, H., Johnson, K. and Laughon, A. (2001). Repression of dpptargets by binding of Brinker to Mad sites. J. Biol. Chem.276, 18216-18222.

Klapper, R., Stute, C., Schomaker, O., Strasser, T., Janning, W.,Renkawitz-Pohl, R. and Holz, A.(2002). The formation of syncytia withinthe visceral musculature of the Drosophilamidgut is dependent on duf, snsand mbc. Mech. Dev.110, 85-96.

Knirr, S., Azpiazu, N. and Frasch, M.(1999). The role of the NK-homeoboxgene slouch(S59) in somatic muscle patterning. Development126, 4525-4535.

Knirr, S. and Frasch, M. (2001). Molecular integration of inductive andmesoderm-intrinsic inputs governs even-skipped enhancer activity in asubset of pericardial and dorsal muscle progenitors. Dev. Biol.238, 13-26.

Kopp, A. and Duncan, I.(1997). Control of cell fate and polarity in the adultabdominal segments of Drosophilaby optomotor-blind. Development124,3715-3726.

Kosman, D., Small, S. and Reinitz, J.(1998). Rapid preparation of a panelof polyclonal antibodies to Drosophilasegmentation proteins. Dev. GenesEvol. 208, 290-294.

Lamka, M. and Lipsh*tz, H. (1999). Role of the amnioserosa in germ bandretraction of the Drosophila melanogasterembryo. Dev. Biol.214, 102-112.

Lane, M., Sauer, K., Wallace, K., Jan, Y., Lehner, C. and Vaessin, H.(1996). Dacapo, a cyclin-dependent kinase inhibitor, stops cell proliferationduring Drosophiladevelopment. Cell 87, 1225-1235.

Lecuit, T. and Cohen, S.(1997). Proximal-distal axis formation in theDrosophila leg. Nature388, 139-145.

Lee, H. and Frasch, M.(2000). Wingless effects mesoderm patterning andectoderm segmentation events via induction of its downstream target sloppypaired. Development127, 5497-5508.

Li, X. and Noll, M. (1993). Role of the gooseberrygene in Drosophilaembryos: maintenance of winglessexpression by a wingless-gooseberryautoregulatory loop. EMBO J.12, 4499-4509.

Lo, P. C. and Frasch, M.(2001). A role for the COUP-TF-related gene seven-up in the diversification of cardioblast identities in the dorsal vessel ofDrosophila. Mech. Dev.104, 49-60.

Manseau, L., Baradaran, A., Brower, D., Budhu, A., Elefant, F., Phan, H.,Philip, A. V., Yang, M., Glover, D., Kaiser, K., Palter, K. and Selleck, S.(1997). GAL4 enhancer traps expressed in the embryo, larval brain,imaginal discs, and ovary of Drosophila. Dev. Dyn.209, 310-322.

Mason, E., Konrad, K., Webb, C. and Marsh, J.(1994). Dorsal midline fatein Drosophila embryos requires twisted gastrulation, a gene encoding asecreted protein related to human connective tissue growth factor. GenesDev.8, 1489-1501.

Papaioannou, V.(2001). T-box genes in development: From hydra to humans.Int. Rev. Cytol.207, 1-69.

Preston, C., Sved, J. and Engels, W.(1996). Flanking duplications anddeletions associated with P-induced male recombination in Drosophila.Genetics144, 1623-1638.

Ray, R., Arora, K., Nüsslein-Volhard, C. and Gelbart, W. M.(1991). Thecontrol of cell fate along the dorsal-ventral axis of the Drosophilaembryo.Development113, 35-54.

Riechmann, V., Irion, U., Wilson, R., Grosskortenhaus, A. and Leptin, M.(1997). Control of cell fates and segmentation in the Drosophilamesoderm.Development124, 2915-2922.

Rubin, G. M. and Spradling, A. C. (1982). Genetic transformation ofDrosophilawith transposable element vectors. Science 218, 348-353.

Rusch, J. and Levine, M.(1997). Regulation of a dpp target gene in theDrosophilaembryo. Development127, 303-311.

Rushlow, C., Frasch, M., Doyle, H. and Levine, M.(1987). Maternal

3204

regulation of zerknüllt: a hom*oeobox gene controlling differentiation ofdorsal tissues in Drosophila. Nature330, 583-586.

Rushlow, C. and Levine, M.(1990). Role of the zerknüllt gene in dorsal-ventral pattern formation in Drosophila. Adv. Genet.27, 277-307.

Rushlow, C., Colosimo, P., Lin, M.-C., Xu, M. and Kirov, N. (2001).Transcriptional regulation of the Drosophilagene zenby competing Smadand Brinker inputs. Genes Dev.15, 340-351.

San Martin, B., Ruiz-Gomez, M., Landgraf, M. and Bate, M.(2001). Adistinct set of founders and fusion-competent myoblasts make visceralmuscles in the Drosophilaembryo. Development128, 3331-3338.

Sherman, A. W. (2000). BrdU labeling of chromsomes. In DrosophilaProtocols (ed. W. Sullivan, M. Ashburner and R. S. Hawley), pp. 57-65.Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press.

Sivasankaran, R., Vigano, M., Muller, B., Affolter, M. and Basler, K.(2000). Direct transcriptional control of the Dpp target omb by the DNAbinding protein Brinker. EMBO J.19, 6162-6172.

St Johnston, R. D. and Gelbart, W. M.(1987). Decapentaplegictranscriptsare localized along the dorsal-ventral axis of the Drosophilaembryo. EMBOJ. 6, 2785-2791.

Staehling-Hampton, K., Hoffmann, F. M., Baylies, M. K., Rushton, E. andBate, M. (1994). dppinduces mesodermal gene expression in Drosophila.Nature372, 783-786.

Su, M., Fujioka, M., Goto, T. and Bodmer, R.(1999). The Drosophilahomeobox genes zfh-1 and even-skippedare required for cardiac-specificdifferentiation of a numb-dependent lineage decision. Development126,3241-3251.

Tatei, K., Cai, H., Ip, Y. and Levine, M.(1995). Race: a Drosophila hom*ologof the angiotensin converting enzyme. Mech. Dev.51, 157-168.

Tracey, W. D., Jr, Ning, X., Klingler, M., Kramer, M. and Gergen, J. P.(2000). Quantitative analysis of gene function in the Drosophila embryo.Genetics154, 273-284.

Wieschaus, E., Nüsslein-Volhard, C. and Jürgens, G.(1984). Mutationsaffecting the pattern of the larval cuticle in Drosophila melanogasterIII.Zygotic loci on the X-chromosome and fourth chromosome. Roux’s Arch.Dev. Biol.193, 296-307.

Wu, X., Golden, K. and Bodmer, R. (1995). Heart development inDrosophilarequires the segment polarity gene wingless. Dev. Biol.169, 619-628.

Xu, X., Yin, Z., Hudson, J., Ferguson, E. and Frasch, M.(1998). Smadproteins act in combination with synergistic and antagonistic regulators totarget Dpp responses to the Drosophilamesoderm. Genes Dev.12, 2354-2370.

Yip, M., Lamka, M. and Lipsh*tz, H. (1997). Control of germ-band retractionin Drosophila by the zinc-finger protein HINDSIGHT. Development124,2129-2141.

Zaffran, S., Küchler, A., Lee, H. H. and Frasch, M.(2001). biniou(FoxF),a central component in a regulatory network controlling visceral mesodermdevelopment and midgut morphogenesis in Drosophila. Genes Dev.15,2900-2915.

Zhang, H., Levine, M. and Ashe, H.(2001). Brinker is a sequence-specifictranscriptional repressor in the Drosophilaembryo. Genes Dev.15, 261-266.

I. Reim, H.-H. Lee and M. Frasch

The T-box-encoding Dorsocross genes function in amnioserosa … · demonstrates that the three Doc genes have largely redundant functions during amnioserosa development, as well as - [PDF Document] (2024)

References