The T-box-encoding Dorsocross genes function in amnioserosa development and the patterning of the dorsolateral germ band downstream of Dpp

The BMP family member Dpp has a key role in dorsoventral axis formation and is responsible for the establishment of positional identities in dorsal and lateral areas of the early Drosophila embryo. Cell identities that are determined by Dpp include those of the extra-embryonic amnioserosa in the dorsalmost region of the embryo, dorsal epidermis and peripheral nervous system (PNS) in dorsolateral regions of the ectoderm, as well as dorsal vessel, dorsal somatic and visceral muscles in the dorsal mesoderm(Irish and Gelbart, 1987; Ray et al., 1991; Ferguson and Anderson, 1992a; Staehling-Hampton et al.,1994; Frasch,1995). In addition to promoting dorsal epidermal and PNS fates in the dorsolateral ectoderm, Dpp acts to suppress the formation of neurons of the central nervous system in the same area.

For a better understanding of these activities, we need to consider that Dpp exercises some of its functions sequentially at different stages of development, during which dpp changes its own pattern of expression(St Johnston and Gelbart,1987). In particular, during blastoderm and gastrulation stages,Dpp acts in a dose-dependent fashion to establish positional information in dorsal and lateral areas of the embryo and to specify amnioserosa tissue(Ferguson and Anderson, 1992a; Ashe et al., 2000). Although dpp is expressed uniformly around ∼40% of the dorsal circumference of the embryos during this stage, the activity of Dpp is modulated along the dorsoventral axis by diffusion of the secreted gene product as well as by positive and negative regulators of the signaling pathway. Negative regulators include Short gastrulation (Sog) and Brinker(Brk), both of which are expressed ventrolaterally(Ferguson and Anderson, 1992b; Francois et al., 1994; Jazwinska et al., 1999). Whereas Sog and its vertebrate homolog Chordin are secreted molecules that inhibit BMP signaling via binding to the ligand (reviewed by Garcia Abreu et al., 2002),Brk appears to be a nuclear factor that interferes with the signaling output via binding to regulatory sequences of Dpp target genes(Sivasankaran et al., 2000; Kirkpatrick et al., 2001; Rushlow et al., 2001; Zhang et al., 2001). By contrast, specification of amnioserosa fates in the dorsal 10% of embryonic cells requires maximal signaling activities that involve Sog as a positive regulator of Dpp in conjunction with Twisted gastrulation (Tsg) as well as a second, uniformly-distributed BMP ligand, Screw (Scw)(Arora et al., 1994; Mason et al., 1994; Ashe and Levine, 1999; Decotto and Ferguson, 2001). Dpp, Sog and Tsg are thought to be present in a diffusible trimolecular complex that serves to carry and release active Dpp prefentially into dorsalmost areas where tsg is expressed(Decotto and Ferguson,2001).

After gastrulation, dpp expression ceases in the developing amnioserosa and becomes restricted to a broad stripe of cells in the dorsolateral ectoderm along the elongated germ band(St Johnston and Gelbart,1987). During this period, the dorsally migrating cells of the mesoderm reach the dpp-expressing area of the ectoderm, thus allowing Dpp to induce dorsal mesodermal cell fates across germ layers(Staehling-Hampton et al.,1994; Frasch,1995). In addition, Dpp is thought to act in the continuing patterning processes within the dorsolateral ectoderm during this stage, which lead to the specification of tracheal as well as particular epidermal and sensory organ progenitors. Both in the dorsal mesoderm and dorsolateral ectoderm, Dpp must act in combination with additional patterning molecules that provoke differential responses of cells to the Dpp signal. For example,in the dorsal mesoderm, the presence or absence of Wingless (Wg) activity determines whether cells will respond to Dpp by forming heart and dorsal somatic 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 how Dpp signals pattern the embryo and how they are integrated with other patterning processes, it is crucial to study the regulation of Dpp target genes. To date,detailed molecular studies have been described for three targets that are induced during early embryogenesis, namely the homeobox genes zerknüllt (zen), tinman (tin) and even-skipped (eve). zen is required for the specification of the amnioserosa downstream of Dpp. Accordingly, the expression of zen in a dorsal on/ventral off pattern, although initially Dpp-independent, requires low levels of Dpp activity for its maintenance and high Dpp activities for its subsequent refinement to areas of the prospective amnioserosa (Doyle et al.,1986; Rushlow and Levine,1990). Likewise, tin is required for the specification of all dorsal mesodermal tissues and eve for the normal differentiation of specific pericardial cells and dorsal somatic muscles in a Dpp-dependent manner. (Bodmer, 1993; Azpiazu and Frasch, 1993; Su et al., 1999). All three genes have in common the presence of multiple binding sites for intracellular Dpp effectors, the Smad proteins Mad and Medea, in their regulatory regions,which are essential for mediating the inductive activity of Dpp. However, in addition to these Smad-binding sites, each of these genes has a characteristic set of additional regulatory sequences that, at least in part, explain its particular spatial and tissue-specific response to Dpp signals. For example, zen contains binding sites for Brk in addition to the Smad sites(Rushlow et al., 2001). It appears that the antagonistic activities of the Brk and Smad sites and the differential ratios of Brk versus active Smad proteins along the dorsoventral embryo axis determine the ventral border of Dpp-dependent zen domain during cellularization stages. The Smad sites but not the Brk sites are also required for zen induction in the prospective amnioserosa during the cellularized blastoderm stage (Rushlow et al., 2001). The mesodermal Dpp targets tin and eve require Smad-binding sites and, in addition, binding sites for Tin, which serve to target the Dpp response to the mesoderm(Xu et al., 1998; Halfon et al., 2000; Knirr and Frasch, 2001). Further, the Dpp-responsive enhancer of eve contains functionally important binding sites for regulators that restrict its activity to segmental subsets of dorsal mesodermal cells, including the Wg effector Pangolin (Pan)(Halfon et al., 2000; Knirr and Frasch, 2001).

In the present study, we introduce three novel genes that respond to Dpp signals in the prospective amnioserosa, dorsal ectoderm and dorsal mesoderm,and are good candidates for being direct targets of the Dpp signaling cascade. The three genes, Dorsocross1 (Doc1), Dorsocross2(Doc2) and Dorsocross3 (Doc3), which are present in a gene cluster, are closely related members of the T-box family of genes and presumably arose by relatively recent duplications from a common ancestor. The Dorsocross (Doc) genes are expressed in essentially identical patterns within several areas that receive high levels of Dpp signals, including the prospective amnioserosa during the cellularized blastoderm stage, the dorsolateral ectoderm and dorsal mesoderm during germ band elongated stages and areas that span the compartment border in wing discs. We show that Doc expression in the prospective amnioserosa depends on dpp and zen, whereas the metameric expression in the dorsolateral ectoderm and dorsal mesoderm depends on a combination of dpp and wg. Our genetic analysis demonstrates that the three Doc genes have largely redundant functions during amnioserosa development, as well as during dorsolateral ectoderm and dorsal mesoderm patterning. We focus on the role of the Doc genes in the amnioserosa and dorsolateral ectoderm. We show that they are essential for full differentiation and maintenance of the amnioserosa,including the arrest of cell proliferation in this tissue. Owing to the requirement of a functional amnioserosa for normal germ band retraction, loss of Doc activity produces embryos with a permanently extended germ band. Hence,Doc genes are new members of the u-shaped family of genes. All genes of this family, which also includes hindsight (hnt; peb– FlyBase), serpent (srp), tail-up(tup), u-shaped (ush), epidermal growth factor receptor (Egfr) and insulin-like receptor(InR), are components of a regulatory network that controls normal development and functioning of the amnioserosa(Frank and Rushlow, 1996; Goldman-Levi et al., 1996; Yip et al., 1997; Lamka and Lipshitz, 1999). In addition to the amnioserosa, the Doc genes are required for the normal patterning of the dorsolateral ectoderm, which includes the repression of wg and ladybird (lb) expression within this area. These findings provide valuable insight into the mechanisms of how Dpp signals are executed during the development of the amnioserosa and the patterning of dorsolateral areas of the embryonic germ band.

660-850 bp DNA fragments from the non-conserved 3′-region of the predicted genes CG5133 (Doc1), CG5187(Doc2) and CG5093 (Doc3) were amplified by PCR from genomic DNA and cloned into pCRII-TOPO (Invitrogen) to obtain gene-specific probes. The following primer pairs were used: GTTCGCTAAGGGTTTCCGCGAGTC and GCAAATAGTTTTGCATTTTCTACGGATTC for Doc1; GCGCTGCAAACGCAAGATGTCTTCATC and GCCGATATGCTGAAGCCCTTGCTCCTT for Doc2; and GTCGAGATGCAAACGGAAGATCAATGAC and GTTTCATCCCAGAAATAGCTCCATCGAATTCA for Doc3. A plasmid library containing cDNAs from 4- to 8-hour-old Drosophila melanogaster embryos in pNB40(Brown and Kafatos, 1988) was screened with the respective radioactively labeled fragments and the plasmid DNAs of isolated clones were sequenced (GeneWiz, New York). The cDNAs shown in Fig. 1A correspond, from top to bottom, to clones Doc1-a1.1, Doc2-c6.2, Doc2-c1.1, Doc2-c12.2 and Doc3-b4.1.

C-terminal cDNA fragments containing the variable coding region of each Dorsocross gene were cloned into the expression vector pQE-9 (Qiagen) cut with BamHI and HindIII. Fragments were obtained by PCR using the cDNA clones Doc1-a1.1, Doc2-c6.2 and Doc3-b4.1 as templates and restriction site-introducing primers. In the recombinant proteins Doc1^CTD,Doc2^CTD and Doc3^CTD, a N-terminal His-tag(MGRSH₆GS) is fused to an arginine residue at position 249, 246 and 249, respectively. Proteins were produced in E. coli M15{pRep4}. Expression was induced by 1 mM IPTG at mid-log phase and bacteria were grown for another 3-4 hours at 37°C, except for Doc1^CTD that was produced at 30°C. Ni-NTA-agarose (Qiagen) purified proteins were used for immunization (Covance Research Products). Anti-Doc2 (made in rabbit) is specific for Doc2 in Drosophila embryos as shown by Doc1, Doc2 and Doc3 misexpression. Anti-Doc1 antibody (made in rat) reacts with Doc1 in Drosophila embryos, but only weakly at wild-type Doc1 levels. Unexpectedly, the two guinea pig antibodies derived from separate Doc2 and Doc3 immunizations both recognize both Doc2 and Doc3. Therefore, we name these antibodies anti-Doc2+3 and anti-Doc3+2, with anti-Doc2+3 working better for Doc3 detection.

Transformation plasmids were constructed by subcloning the cDNAs Doc1-a1.1,Doc2-c6.2 and Doc3-b4.1 into pP{UAST}(Brand and Perrimon, 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. The HindIII and NotI sites are pNB40 vector-derived and the EcoRI and SmaI sites are located in the 5′ UTR of Doc1 and Doc2, respectively. Several independent lines were established for each construct using standard transformation methods(Rubin and Spradling, 1982). GAL4 inducible expression of Dorsocross proteins was confirmed by immunostaining with the antibodies described above.

In order to create deletions uncovering all three Dorsocross genes a mutagenesis screen was performed using the male recombination method(Preston et al., 1996). The closest available P-insertion, EP(3)3556, was used to trigger male recombination in the presence of active transposase. EP(3)3556 is a homozygous viable insertion in the 5′ region of smg (Dahanukar et al.,1999), which is located about 8 kb upstream of Doc3 (see Fig. 6A). F0 males subject to recombination/mutagenesis carried both a X-chromosomal transposase source and the targeted P-insertion flanked by genetic markers on the third chromosome (y w H{PΔ2-3} HoP8/Y; ru EP(3)3556 th st cu sr e^s ca/+). F0 males were crossed to ru h th st cu sr e^s Pr ca/TM6B, Bri Tb females. The F1 generation was screened for non-balancer males carrying a recombinant third chromosome. The ru marker is expected to be retained along with the P-insertion if recombination causes a deletion extending towards the Dorsocross gene cluster (Preston et al.,1996). Recombinants were crossed individually to Df(3L)Scf-R11/TM3, Sb eve-lacZ females for producing TM3, eve-lacZ balanced stocks and for complementation analysis with Df(3L)Scf-R11.

A total of 23 recombinants (13 ru, 6 th st cu sr e^sca and four unusual marker combinations) were obtained out of 12400 scored non-balancer F1 males. These recombinants were derived from 18 of 168 individual F0 crosses and from each of these crosses only one ru male was taken to ensure the recovery of individual events. Five out of 10 candidate ru lines were homozygous lethal and two of them, Df(3L)DocA and Df(3L)DocB, removed the entire Dorsocross gene cluster. An analogous mutagenesis screen with EP(3)584, which is inserted ∼80 kb downstream of Doc1, yielded Df(3L)EP584MR2 (Fig. 5A).

The breakpoints of Df(3L)DocA, Df(3L)DocB and Df(3L)EP584MR2 were mapped by sequencing of inverse PCR products recovered from the 5′ end of the retained P insertion. Inverse PCR was performed as described by E. J. Rehm(http:/www.fruitfly.org/p_disrupt/inverse_pcr.html)and Huang et al. (Huang et al.,2000) using Pwht1 and Plac1 primers. The removal of genes by various deficiencies (Fig. 5A)was confirmed by PCR from homozygous mutant embryos as described in Duan et al. (Duan et al., 2001). PCR from equally treated heterozygous embryos and PCR amplification of sequences not affected by the deficiencies served as positive controls for primers and template, respectively. PCR amplification and sequencing of DNA to the left of EP(3)3556 demonstrated no change in sequences flanking the 3′ P-end.

Apoptotic cells were labeled by terminal deoxynucleotidyl transferase(TdT)-mediated dUTP nick end-labeling (TUNEL) using components of the ApopTag® Peroxidase kit S7101 (Intergen Company/Serologicals Corporation). Rehydrated embryos (about 30 μl) were treated with 10 μg/ml Proteinase K for 1 minute, rinsed quickly three times and washed another five times for 3 minutes with PBT. Embryos were postfixed in 3.7% formaldehyde in PBT for 20 minutes, washed five times with PBT and twice for 20 minutes with 30 μl equlibration buffer. TdT reaction was performed over night at 37°C using 50 μl buffer/TdT mixed in a ratio 7:3 and supplemented with 0.3% Triton X-100. The reaction was stopped by a 20 minute wash in 1/34 diluted stop buffer. Detection of incorporated Digoxigenin-nucleotides using sheep-anti-Dig and biotinylated anti-sheep antibodies, the VectaStain ABC elite kit (Vector Laboratories) and Tyramide Signal Amplification (TSA) reagents (NEN/Perkin Elmer Life Sciences) was essentially as for in situ hybridization using digoxigenin-labeled probes (Knirr et al.,1999).

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

y w or Oregon R were used as wild-type controls and stocks were obtained from the Bloomington stock collection unless noted otherwise. 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 drives amnioserosa expression from stage 9; I.R. and M.F., unpublished) and UAS-dpp#5 (Frasch,1995). The following previously described mutant alleles were also used: dpp^H46, hnt^E8, pnr¹, CyO slp^Δ34B, srp^P{PZ}01549, tup¹, ush², wg^CX4 and zen⁷. For male recombination and mapping experiments, we used the lines y w H{PΔ2-3} HoP8, `ru cu ca'and `ru Pr ca', ru Df(3L)Scf-R11, Scf-R6 th st cu sr e ca (Kopp and Duncan,1997), smg¹(Dahanukar et al., 1999), Df(3L)29A6 kni^ri-1 p^p, EP(3)3556, and EP(3)584 (Exelixis). For ectopic expression experiments, embryos were collected at 28°C from crosses of GAL4-carrying females with UAS construct-carrying males. All other crosses were performed at room temperature(22-25°C).

Sense and antisense RNA was transcribed from non-conserved 3′fragments of Doc1, Doc2 and Doc3 in pCRII-TOPO (see above)and hybridized in injection buffer (5 mM KCl, 10 mM sodium phosphate, pH 7.8)to generate dsRNA (Kennerdell and Carthew,1998). A mix of Doc1, Doc2 and Doc3 dsRNA(∼100-300 pl of 4 mg/ml each) was injected ventrally into pre-blastoderm embryos that had been dechorionated and mounted onto double-sided sticky tape under Halocarbon 700 oil. An Eppendorf FemtoJet automatic injector and Eppendorf Femtotip injection needles were used for injections. Embryos were allowed to develop at 18°C until the desired stage was reached. For immunostaining, embryos were transferred to standard heptane/formaldehyde fixation solution in a small drop of oil. The heptane phase and total fixation solution were exchanged twice to remove traces of oil. After 20 minutes incubation, the formaldehyde solution was replaced by PBS. For the manual removal of the vitelline membranes, embryos were spread on agar plates,transferred to double-sided sticky tape and covered with PBS. After devitellinization 0.05% Tween was added and embryos were transferred into reaction tubes for standard staining procedures. Cuticle preparations were made 2-3 days after injection. Embryos were passed through acetic acid/glycerol (4:1) overlaid with heptane, transferred to a mesh, rinsed with heptane and PBT and mounted in standard Hoyer's medium.

Antibody staining of embryos using DAB, double fluorescent staining and in situ hybridization in combination with fluorescent antibody staining were carried out as described previously (Knirr et al., 1999). Dorsocross in situ hybridization probes were made by in vitro transcription from 3′ fragments cloned into pCRII-TOPO (see above). The race in situ hybridization probe and zen cDNA were a gift from C. Rushlow (Frank and Rushlow, 1996). Antibody staining of imaginal discs was essentially the same as for embryos, except that imaginal discs attached to inverted larval heads were fixed in 3.7% formaldehyde in PBS, dehydrated in methanol, and rehydrated using 70%, 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 pig anti-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), mouse monoclonal 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 Developmental Studies Hybridoma Bank, University of Iowa: anti-α-tubulin 12G10 (1:10), anti-Futsch 22C10(1:250), anti-Wg 4D4 (1:40, in imaginal discs 1:400), anti-En/Inv 4D9 (1:4),anti-Hnt 27B8 1G9 (1:30) and anti-β-gal 40-1a (1:40).

Sequence information from the Berkeley Drosophila Genome Project(Adams et al., 2000) revealed that three novel, closely related T-box encoding genes are clustered within∼40 kb of genomic sequences at 66F1 to 66F2 on chromosome arm 3L. In reference to their peculiar patterns of expression in blastoderm embryos (see below), these genes have been named, from proximal to distal, Dorsocross1 (Doc1; previously Tb66F2)(Lo and Frasch, 2001), Dorsocross2 (Doc2) and Dorsocross3 (Doc3)(Fig. 1A). The gene cluster also includes an unrelated predicted gene, CG5194, which maps between Doc2 and Doc3.

cDNAs for the three Doc genes, which were isolated from an early embryonic cDNA library (see Materials and Methods), encode proteins of 391 amino acids(Doc1), 469 amino acids (Doc2) and 424 amino acids (Doc3), respectively. Comparisons between cDNA and genomic sequences indicate that Doc2encodes at least three different mRNA products, which appear to be generated from alternative transcription start sites. Among these, Doc2variants A and B encode identical polypeptides, whereas variant C does not encode any long open reading frame (Fig. 1A). The data from northern analysis indicate that the longest cDNAs obtained for each gene are close to full length if the polyA tails are taken into account (1.7 kb transcripts versus 1500 bp cDNA for Doc1,2.0 kb transcripts versus 1759 bp cDNA for Doc2A, and 1.8 kb transcripts 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) is expressed much more strongly than the other two variants. In addition, we note that splicing occurs at identical positions within the open reading frames of Doc1, Doc2 and Doc3, although most introns in Doc3 are much smaller as compared with those in the other two genes(Fig. 1A).

Sequence comparisons show that the three Doc proteins share high degrees of similarity within their T-box domain sequences (>95% amino acid identities)as well as within short sequence stretches extending N- and C-terminally from these domains (Fig. 1B). The N-terminal regions of the polypeptides up to the T-box domains are moderately conserved (>40% amino acid identities), whereas the C-terminal regions contain only few short stretches of additional sequence similarity (data not shown). Additional sequence comparisons with T-box domains from vertebrates and phylogenetic analysis show that the Doc T-box domains are most closely related to those from members of the Tbx6 subfamily of T-box proteins(Fig. 1B,C)(Papaioannou, 2001) (see Discussion).

Northern analysis with gene-specific probes showed that all three Doc genes display similar expression profiles during development with maximal levels occurring between 2 and 12 hours of embryonic development and lower levels during late embryonic, larval and pupal stages. The only significant difference among the three genes in this assay was the expression of Doc1 mRNA in adult males, which was not observed for Doc2and Doc3 (data not shown).

The spatial expression patterns of Doc products in embryos were examined by whole-mount in situ hybridization with gene-specific probes and whole-mount immunocytochemistry using antibodies raised against the unique C-terminal regions of the Doc proteins (see Materials and Methods). As all three genes were found to have essentially identical expression patterns (with some minor differences regarding the relative levels of expression in different tissues;data not shown), we will henceforth collectively refer to them as `Doc genes'. As expected, Doc proteins are exclusively nuclear during interphase.

The initial expression of Doc genes is observed at the cellular blastoderm stage in a transverse stripe encompassing the dorsal ∼40% of the embryonic circumference within the prospective head region. Shortly later, a narrow longitudinal stripe of expression appears, which ultimately extends all along the dorsal midline of the embryo, and the joint domains form a cross-shaped pattern of Doc expression in dorsal areas of the early embryo(Fig. 2A). The domain of the transverse stripe is located anteriorly to the cephalic furrow forming during gastrulation (Fig. 2B; Fig. 3G) and largely corresponds to procephalic neuroectoderm. The cells of this domain continue Doc expression until stage 11, when the segregation of procephalic neuroblasts is completed (Fig. 2B-D)(Campos-Ortega and Hartenstein,1997). By contrast, the cells from the dorsal longitudinal domain within the trunk region give rise to amnioserosa, which maintains strong Doc expression until stage 15 (Fig. 2C-F). In addition, the cells from the anterior and posterior termini within this longitudinal stripe contribute to regions of the anterior and posterior digestive tract and maintain expression until stage 11.

During stages 9 and 10, a new pattern of Doc expression emerges within dorsolateral areas of the germ band from parasegment (PS) 1 to PS13, which consists of 13 rectangular cell clusters(Fig. 2D). This metameric expression includes ectodermal as well as underlying mesodermal cells. In the ectoderm, Doc expression is excluded from the dorsalmost cells near the amnioserosa at this stage, whereas in the ectoderm the metameric Doc expression extends to the dorsalmost areas of this germ layer(Fig. 2D; see below). To determine the segmental register of Doc expression in the mesoderm we co-stained with antibodies against the POU domain transcription factor Cf1a(Vvl – FlyBase), which mark the 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 encompasses prospective tissues of the lateral epidermis and dorsolateral sensory organs. After stage 11, the segmental expression in the epidermis is modified to form segmental stripes that are interrupted in dorsolateral regions. Within these stripes, Doc expression is largely found in posterior areas of the anterior compartments of each segment and there is a graded distribution of Doc expression with increasing levels towards the posterior of each stripe (Fig. 2F,G). The dorsal epidermal expression domains now extend to the amnioserosa, and after dorsal closure the bilateral domains merge at the dorsal midline(Fig. 2F-H).

Additional sites of Doc expression during late embryogenesis include the dorsal pouch in the embryonic head (Fig. 2H), the anterior pair of Malpighian tubules(Fig. 2I) and the pentascolopidial chordotonal sensory organs(Fig. 2J). The mesodermal expression of the Doc genes, which will be presented elsewhere in more detail along with functional data, is observed in areas between the expression domains of the homeobox gene bagpipe (bap) at stage 10(Fig. 2K). This location defines them as dorsal areas of the mesodermal A (or slp) 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 in the caudal visceral mesoderm, which contains the founder cells of the longitudinal muscles of the midgut(Fig. 2L)(San Martin et al., 2001; Klapper et al., 2002). As reported previously for Doc1, two out of six bilateral cardioblasts in each segment of the dorsal vessel, which are tin negative and svp positive, also express the Doc genes(Fig. 2H)(Lo and Frasch, 2001).

As peak levels of Dpp activity are known to be required for cell fate determination at the dorsal midline, we tested whether there is a correlation between Dpp activity and dorsal longitudinal Doc expression during blastoderm stages. As shown in Fig. 3A,double-staining for Doc mRNA and phosphorylated Mad (PMad) indicates a close correlation between cells containing high levels of PMad and Doc products within the dorsal-longitudinal stripe. In addition, faint Doc signals that are modulated in a pair-rule pattern extend into areas that receive lower Dpp inputs and lack detectable PMad (Fig. 3A). As predicted, both PMad and Doc expression in the dorsal-longitudinal stripe, but not the dorsal-transverse head stripe of Doc expression, are absent in dpp-null mutant embryos(Fig. 3B). Conversely, in blastoderm embryos with ubiquitous Dpp expression (UAS-dpp activated by maternally provided nanos-GAL4), we observe a significant widening of the dorsal-longitudinal stripes of PMad and Doc expression, during which the correlation between high PMad and Doc mRNA levels is still maintained(Fig. 3C).

The expansion of PMad upon uniform ectopic expression of dppincludes the prospective mesoderm, although not ventrolateral areas of the blastoderm embryo (Fig. 3D). However, high PMad in the prospective mesoderm does not trigger ectopic Doc expression, suggesting either the presence of a ventral repressor or the requirement for a co-activator in dorsal areas. A candidate for a co-activator is the homeobox gene zerknüllt (zen). Double in situ hybridization shows that the appearance of dorsal Doc mRNAs coincides with the time when zen mRNA levels increase in the areas of the presumptive amnioserosa as a result of high Dpp inputs(Fig. 3E). When the refinement of zen expression is completed, there is an exact correspondence in the widths of the Doc and zen expression domains, although Doc expression extends more posteriorly (Fig. 3F). As shown in Fig. 3H (compare with Fig. 3G), the activity of zen is necessary for normal levels of Doc expression in the dorsal-longitudinal stripe, because in zenmutant embryos there are only low residual levels of Doc products present in this domain. These observations suggest that Doc expression along the dorsal midline of blastoderm embryos requires the combined activities of dppand zen.

The known distribution of dpp mRNA during its second phase of expression in the dorsolateral ectoderm of stage 9-11 embryos(St Johnston and Gelbart,1987) suggests that Doc expression in the dorsolateral ectoderm and mesoderm during these stages is also dependent on Dpp activity. As expected from the known fate map shifts in dpp mutants, these domains of Doc expression are missing in dpp-null mutant embryos (data not shown). Notably, the exact coincidence between the ventral borders of the domains of dorsolateral Doc expression and high nuclear PMad(Fig. 4A,B) suggests that Doc expression is directly controlled by Dpp-activated Smad proteins in the ectoderm and mesoderm during this stage. Additional evidence for this hypothesis comes from experiments with ectopic expression of dpp in the ventral ectoderm of the Krüppel domain (by virtue of a modified Kr-GAL4 driver) (Frasch,1995), which results in the concomitant expansion of PMad and the Doc expression stripes towards the ventral midline(Fig. 4C).

In addition to the inputs from dpp, metameric Doc expression in dorsolateral areas of the germ band must depend on the activity of segmental regulators. A direct comparison with the expression of engrailed(en) shows that the clusters of Doc expression straddle the compartmental borders. Although Doc expression overlaps with en in the P compartments, about two-thirds of the Doc expressing cells of each cluster are located in posterior areas of the A compartments(Fig. 4D). In agreement with this allocation, we find that the metameric Doc domains are exactly centered on the stripes of Wingless (Wg) expression(Fig. 4E). The observed correlation of the segmental registers of Wg and Doc makes wg a good candidate for an upstream regulator of Doc. As shown in Fig. 4F, dorsolateral Doc expression in the ectoderm and mesoderm is completely absent if wg is inactive. By contrast, deletion of sloppy paired (slp), a known target of wg in the mesoderm and a wg feedback regulator in the ectoderm (Lee and Frasch,2000), results in a reduction, but not a complete loss of metameric DOC expression (data not shown). Hence, slp probably affects Doc indirectly through its effect on ectodermal wgexpression. Altogether, our data suggest that metameric Doc expression in the ectoderm and mesoderm is triggered by the intersecting activities of Wg and Dpp.

The similarities in sequence and expression of the three Doc genes suggested functional redundancy among these genes. Because our molecular analysis of available deficiencies at 66E-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 via male recombination-induced mutagenesis (see Materials and Methods). Molecular mapping of the obtained deletions demonstrated that two of them, Df(3L)DocA and Df(3L)DocB, which were generated with the distally located insertion EP(3)3556 and cause embryonic lethality, deleted all three Doc genes(Fig. 5A). As Df(3L)DocA deletes the smallest number of additional genes(CG5087, CG5194, CG5144, Argk and CG4911), we describe the phenotypic analysis in the present study using this deficiency, although the salient phenotypes are very similar between Df(3L)DocA and Df(3L)DocB.

Additional genetic analysis showed that it is possible to obtain a small number of viable adult escapers with the genotype Df(3L)Scf-R11/Df(3L)DocA, which indicates that CG5087 is not absolutely required for viability, and that Doc1 and Doc2 can functionally substitute for the loss of Doc3. Similarly, the full viability of flies with the genotype Df(3L)DocA/Df(3L)EP584MR2(Fig. 5A) shows that CG4911 and the 5′ exons of Argk (preceding the large intron) are also not essential. Furthermore, we determined that embryos with the genotypes Df(3L)Scf-R11/Df(3L)DocA and Df(3L)DocA/Df(3L)29A6 (which causes pupal lethality) do not display any of the phenotypes described below for Df(3L)DocAhomozygous embryos. In summary, our genetic analysis shows that the loss of either Doc3 or Doc1 can be compensated for by the remaining two Doc genes in embryos and that the phenotypes described herein are a consequence of the loss of all three of the Doc genes. However, we can not rule out a contribution of CG5194, which encodes a 128 amino acid predicted ORF with no known homology, to the observed phenotypes.

Because of the prominent Doc expression in the primordia and developing tissue of the amnioserosa we used the amnioserosa marker Krüppel(Kr) to examine whether the Doc genes are required for the development of this extra-embryonic tissue. These experiments demonstrated that homozygous Df(3L)DocA mutant embryos (henceforth called DocA mutants) fail to express Kr in the amnioserosa at any stage, whereas CNS expression of Kr is not affected(Fig. 5B). To confirm that this observed phenotype is due to the loss of Doc gene function we diminished Doc gene functions by using RNA interference (RNAi) as an independent assay and performed rescue experiments with DocA mutants (see below). As shown in the example of Fig. 5C,injection of a mixture of equimolar amounts of dsRNAs for all three Doc genes(see Materials and Methods) frequently results in a complete absence of Kr expression in the amnioserosa. The remaining embryos display strongly reduced numbers of Kr-containing nuclei in this tissue (data not shown). These phenotypes correlate with the observed absence or severe reduction of Doc protein levels in Doc RNAi embryos (data not shown). By contrast, mock-injected embryos display normal expression of Kr in the amnioserosa (Fig. 5D). Hence, the strongest phenotype obtained by RNAi mimics the observed DocA mutant phenotype, confirming that the lack of Krexpression in DocA mutant embryos is specifically due to the loss of the activity of all three Doc genes.

Besides the effects on Kr expression, DocA mutant and RNAi-treated embryos share several morphological defects. The extending germ band is unable to displace the amnioserosa fully 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 be clearly seen in stage 14 embryos (Fig. 5E) and in cuticle preparations of unhatched first instar larvae(Fig. 5H,I; compare with J). This phenotype is shared with previously described genes of the u-shaped (ush)group, which affect the maintenance of the amnioserosa(Frank and Rushlow, 1996). Kr expression and the germ band retraction defects in DocA mutant embryos can be partially rescued by expressing any of the three Doc genes with an early amnioserosa-specific driver (Fig. 5F,G). Rescue with Doc2(Fig. 5F) is consistently more efficient when compared with Doc1 (data not shown) and Doc3(Fig. 5G), although it is not known whether this difference is due to a higher intrinsic activity or a more efficient expression of Doc2 protein in this assay.

An additional phenotype consists of reductions in the size of the embryonic head in DocA mutants and RNAi-treated embryos, which is apparent from stage 12 onwards and results in reduced head structures and a frequent failure of head involution at later stages (Fig. 5B,C,H,I, and data not shown). This phenotype is probably due to excessive cell death as a consequence of the absence of Doc activity in the procephalic neuroectoderm and other dorsal areas of the embryonic head(Fig. 2B,C and data not shown). The observed head phenotypes, as well as the aberrant shape of the filzkörper (Fig. 5I), are also reminiscent of similar phenotypes of embryos mutant for genes of the ush group (Frank and Rushlow,1996).

To obtain more information about the particular role of the Doc genes in the specification and/or differentiation of the amnioserosa we analyzed the distribution of additional amnioserosa markers in DocA mutant embryos. For the ush group gene hnt(Yip et al., 1997) we find a strong reduction of expression, with significant levels of Hnt protein only being detected in nuclei along the posterior margin of the amnioserosa(Fig. 6A, Fig. 7A, compare with Fig. 6B and Fig. 7B, respectively). By contrast, the expression of the amnioserosa marker race(Ance – FlyBase) (Tatei et al., 1995) is initiated normally in the primordium of the amnioserosa of DocA mutant embryos, suggesting that the expression of the race upstream activator zerknüllt (zen) is also not disrupted (data not shown). However, after embryonic stage 9, race expression is gradually lost in the amnioserosa of DocAmutant embryos and its residual mRNA distribution closely follows that of Hnt(Fig. 7A, compare with 7B).

We also examined the expression of a novel amnioserosa marker, which is encoded by the homeobox gene C15. In the normal situation, C15 is expressed in the amnioserosa from stage 7 until stage 17, when the amnioserosa undergoes apoptosis (Fig. 6D,F,H, and data not shown)(Campos-Ortega and Hartenstein,1997). In addition, from early stage 10 onwards there is a narrow domain of expression at the leading edge of the dorsal germ band, which later becomes segmental (Fig. 6F,H). In DocA mutant embryos, the level of C15 expression in early amnioserosa cells is unaltered, which allows us to use C15 protein as a marker for the development of this tissue in the absence of Doc activity.

Until stage 9, the large majority of amnioserosa nuclei in DocAmutant embryos appear large and flattened as in wild-type embryos(Fig. 6C, compare with 6D). Together with data from α-tubulin staining (not shown), this observation indicates that the amnioserosa cells begin to acquire the normal features of a squamous epithelium (data not shown). However, the amnioserosa does not display a properly folded morphology during stages 8-10, and the posterior germ band is forced to bend towards the inside in DocA mutant embryos(Fig. 6C, compare with 6D, and data not shown). In addition, some small nuclei become detectable within the amnioserosa during this stage (Fig. 6C, arrow). Altogether, these observations indicate that the amnioserosa initiates its differentiation process in the absence of Doc gene activity but fails to complete it, thus leading to morphological and functional abnormalities of this tissue towards the end of germ band elongation. Much stronger alterations can be observed during subsequent stages, when there are an increasing number of C15-stained amnioserosa nuclei with much smaller diameters than regular amnioserosa nuclei. At late stage 12,almost all amnioserosa cells feature small nuclei that are difficult to distinguish from dorsal epidermal cells(Fig. 6E, Fig. 7C, compare with Fig. 6F, Fig. 7D, respectively). Co-staining for race indicates that it is predominantly the cells with the small nuclei that lose race expression, while most normally-sized nuclei are still surrounded by race signals(Fig. 7C, compare with D). From this stage onwards, non-stained `holes' appear in the amnioserosa and the number of C15-stained amnioserosa nuclei decreases prematurely. Hence, unlike wild-type embryos, stage 14 DocA mutant embryos are not covered dorsally by C15-stained amnioserosa cells(Fig. 6G, compare with 6H). In addition to the observed alterations in the amnioserosa, the C15 expression domain at the leading edge of the epidermis appears significantly broadened(Fig. 6E).

We tested whether the increasing number of smaller nuclei in the amnioserosa of DocA mutant embryos is connected with abnormal cell divisions. As shown in Fig. 7E,the M-phase marker phospho-Histone H3 can be detected in numerous amnioserosa nuclei of DocA mutant embryos after stage 10, which is not seen in wild-type embryos (Fig. 7F). In addition, there is significant incorporation of BrdU in amnioserosa nuclei of DocA mutant embryos (particularly in the small nuclei; Fig. 7G), whereas no incorporation is observed in wild-type embryos(Fig. 7H). Mitotic spindles are also present in the amnioserosa of DocA mutants(Fig. 7I, compare with J). These observations indicate that the normal G2 arrest of amnioserosa cells has been released and the cells re-enter the cell cycle. We also examined whether the subsequent disappearance of small C15-stained amnioserosa nuclei in DocA mutant embryos is a result of premature apoptosis of cells in this tissue. This possibility was confirmed by the results of TUNEL labeling experiments, which produced signals in many amnioserosa nuclei from 12 onwards. Most of the TUNEL-labeled nuclei have reduced or are lacking C15 expression (Fig. 7K, compare with 7L, which shows that wild-type amnioserosa nuclei at late stage 12 are not apoptotic). Altogether, these observations suggest that loss of Doc activity prevents the normal differentiation of the amnioserosa to a fully functional tissue, suspends the cell cycle block of amnioserosa cells, and causes premature apoptotic cell death in this tissue.

The segmental stripes of wg expression in the embryonic trunk segments initially span the entire dorsoventral extent of the ectoderm, but at stage 11 they become interrupted in dorsolateral areas(Baker, 1988). A comparison of Wg and Doc expression at this stage shows that the positions of the metameric ectodermal domains of Doc expression correspond to the areas in which the Wg stripes become interrupted (Fig. 8A). Temporally, there is a brief overlap of ectodermal Wg and Doc expression during stage 10 until Wg expression is downregulated within the Doc domains (see Fig. 4E). In contrast to the wild-type situation (Fig. 8A,C), the Wg stripes remain continuous in DocA mutant embryos (Fig. 8B). Similar observations were made with the homeobox gene product Ladybird (Lb=Lbe + Lbl)(Jagla et al., 1997) as a marker. In wild-type embryos after stage 11, Lb is also expressed in striped domains that are interrupted at the positions of the ectodermal Doc domains(Fig. 8D), whereas in DocA mutant embryos there is ectopic expression in a pattern of continuous stripes (Fig. 8E, compare with F). These data show that Doc activity is required for patterning events in the dorsolateral ectoderm, which include the repression of wg and lb expression in these areas.

Ectopic expression experiments with Doc genes provide additional evidence for a repressive activity of Doc on wg expression. Upon ectopic expression of Doc2 in all cells of the ectoderm of wild type embryos,the ventral portions of the Wg stripes are lost(Fig. 9B, compare with 9A). However, the dorsal regions of the Wg stripes appear to be under different regulation, because ectopic Doc results in a uniform domain of dorsal Wg along the anteroposterior axis, albeit at lower levels than in wild-type embryos.

Ectopic expression experiments with Doc genes in imaginal discs further confirm their ability to repress wg. In third instar larval wing discs, Doc genes are expressed in four distinct areas that do not overlap with the wg expression domains. Specifically, two large Doc expression domains are located in the centers of the dorsal and ventral regions of the prospective wing blades and two smaller domains in prospective dorsal hinge and posterior notal regions, respectively(Fig. 9C). In leg discs, low levels of Doc expression can be detected in regions of the prospective body wall and proximal leg segments, which also do not express wg(Fig. 9E). Importantly, ectopic expression of Doc2 within the Dpp domains of imaginal discs causes wg expression to disappear in the corresponding areas(Fig. 9D,F). In agreement with the known role of wg in limb development(Lecuit and Cohen, 1997), its repression by ectopic Doc results in the loss of distal structures of wings,legs and antenna of adult animals (Fig. 9G-I). Analogous ectopic expression experiments with Doc1and Doc3 in embryos and discs produced qualitatively similar,although weaker, effects to Doc2.

The closely related T-box sequences, genomic clustering and virtually identical expression patterns of the three Dorsocross genes suggest that they are derived from relatively recent duplications of a common progenitor gene. Accordingly, our observation that loss of Doc1 or Doc3 does not cause any of the embryonic phenotypes seen upon loss of all three genes indicates that there is a large degree of functional redundancy among these three genes. Phylogenetic analysis with the extended T-box domain sequences shows that the Doc genes are most closely related to the vertebrate Tbx6 genes, whose expression in the paraxial mesoderm is reminiscent of the expression of the Doc genes in the dorsal somatic mesoderm. However,the limited reliability of the branches separating the Tbx6, VegT and Tbx2 subfamilies in the phylogenetic tree analysis, the absence of Drosophila orthologs of VegT and Tbx4/5 genes, as well as shared features of expression in the somatic and/or precardiac and cardiac mesoderm seem to support the alternative possibility that the Doc, Tbx6, VegT and Tbx4/5 genes arose from a common ancestral gene by gene amplifactions after the divergence of the insect and vertebrate lineages.

A prominent feature of the Doc genes is their expression in areas that receive inputs from Dpp, including the dorsalmost cells in blastoderm embryos,the dorsolateral ectoderm and mesoderm in the elongated germband, and distinct domains spanning the compartment border of the wing disc. Indeed, our genetic data, together with the co-localization of Doc transcripts with active Mad in dorsal embryonic tissues, favor the possibility that the Doc genes are direct targets of the Dpp signaling cascade. However, the Dpp signals are required to act in combination with additional regulators during each of these inductive events.

Our observations suggest that robust and stable induction of Doc expression in a dorsal stripe requires the activity of the homeodomain protein Zen as a co-activator of Dpp signals. The zen gene features an early, broad expression domain along the dorsal embryonic circumference, which is initially Dpp independent but subsequently requires Dpp for it to be maintained(Rushlow et al., 1987; Ray et al., 1991). Thereafter,its expression refines into a narrow dorsal domain in a process that requires peak levels of Dpp (Rushlow and Levine,1990; Ray et al.,1991; Rushlow et al.,2001). The activation of Doc expression occurs at the same time as the refinement of zen expression and within the same narrow domain,which also coincides with high phospho-Mad levels(Rushlow et al., 2001). Although the maintenance and refinement of zen by Dpp is zenindependent (Ray et al.,1991), we propose that Zen synergizes with peak signals of Dpp to trigger Doc gene expression in a dorsal stripe. The requirement for this proposed interaction between zen and dpp would explain the failure of zen to activate Doc genes in an early, broad domain as well as the 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 to previously described inductive events in the early dorsal mesoderm, where the synergistic activities of the homeodomain protein Tinman and activated Smads induce the expression of downstream targets such as even-skipped(Halfon et al., 2000; Knirr and Frasch, 2001). The identification of functional binding sites for Zen and Smads in Doc enhancer element(s) will be necessary for demonstrating that an analogous mechanism is active during induction of Doc gene expression in a dorsal stripe. In the absence of such data, we can not completely rule out that dorsal Doc expression is controlled indirectly by Dpp, possibly via the combinatorial activities of zen and another high-level target gene of Dpp. As mutations in several other genes that are expressed in the early amnioserosa, including pannier (pnr), hnt, srp, tupand ush, do not affect Doc expression until at least stage 12 (I.R. and M.F., unpublished), these genes can be excluded as candidates for early upstream regulators of Doc.

Unlike zen, which is expressed only transiently, Doc expression is maintained throughout amnioserosa development. Hence, the Doc genes provide a functional link between the early patterning and specification events in dorsal areas of the blastoderm embryo and the subsequent events of amnioserosa differentiation. The activity of zen is required for all aspects of amnioserosa development that have been examined to date, including normal activation of C15 (M.F., unpublished). By contrast, our data demonstrate that the Doc genes execute only a subset of the functions of zen, which includes the activation of Kr and hnt,but not C15 and early race, in amnioserosa cells. This interpretation is consistent with our failure to obtain a significant increase of amnioserosa cells upon ectopic expression of any of the Doc genes in the ectoderm or throughout the early embryo (using e22c and nanos-GAL4 drivers, respectively; I.R. and M.F., unpublished). The residual expression of hnt in some amnioserosa cells of Doc mutant embryos could be due to direct inputs from zen itself or from a yet undefined zen downstream gene acting in parallel with Doc. Nonetheless, the strong reduction of hnt expression in Doc mutant embryos could largely account for their amnioserosa-related phenotypes,including the absence of Kr expression, the decline of raceexpression, premature apoptosis and failure of germ band retraction. All of these phenotypes have also been observed in hnt mutant embryos(Wieschaus et al., 1984; Frank and Rushlow, 1996; Lamka and Lipshitz, 1999; Yip et al., 1997). However, it is likely that Doc gene activity is required for the activation not only of hnt but also of additional genes of the u-shaped group and that Doc genes exert some of their functions in parallel with hnt. Some evidence for this notion is derived from the observation that loss of Doc activity has a stronger effect on Kr expression than loss of hnt activity.

One of the hallmarks of amnioserosa development is that the cells of this tissue never resume mitotic divisions after the blastoderm divisions(Campos-Ortega and Hartenstein,1997). To a large extent, this cell cycle arrest is due to the absence of expression of cdc25/string in the prospective amnioserosa, which prevents the cells from entering M-phase and leads to G2 arrest (Edgar and O'Farrell,1989). In addition, the expression of the Cdk inhibitor p21/Dacapo in the early amnioserosa is thought to contribute to the cell cycle arrest(de Nooij et al., 1996; Lane et al., 1996). Although a detailed description of the regulation of string and dacapoexpression in dorsal embryonic areas is lacking, it has been reported that zen is required for repressing dorsal string expression,which is expected to prevent further cell divisions(Edgar and O'Farrell, 1990; Edgar et al., 1994). Notably,our observation that amnioserosa cells re-enter the cell cycle in Doc mutant embryos demonstrate that Doc genes are required for the cell cycle block in addition to zen. Whereas zen mutant embryos feature ectopic cell divisions in dorsal areas already from stage 8 onwards(Arora and Nüsslein-Volhard,1992), in Doc mutants the amnioserosa cells resume mitosis only during and after stage 10, which is shortly after Zen protein disappears. Thus, we hypothesize that the Doc genes take over the function of zenin repressing string and prevent cell divisions at later stages of amnioserosa development when Zen is no longer present. Overall, the phenotype of Doc mutant embryos suggests that amnioserosa differentiation, including cell cycle arrest and the development of squamous epithelial features,initiates in the absence of Doc activity but is not maintained beyond stage 11. Thereafter, cell division resumes and there is a reversal of the partially differentiated state. Apoptotic events are not observed prior to stage 11 in Doc mutants. However at later stages, many amnioserosa cells die prematurely and the remaining cells are difficult to distinguish morphologically from dorsal ectodermal cells.

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

Unlike in the presumptive amnioserosa, not all cells in the dorsolateral ectoderm and dorsal mesoderm that receive high levels of Dpp induce Doc expression. Rather, Wg signals are required in combination with Dpp in these tissues, such that the Doc genes are induced at the intersections of transverse Wg stripes and the dorsally restricted domain containing high phospho-Mad levels. The Doc stripes extend beyond the peak levels of Wg on both sides of the Wg stripes, which indicates that the Doc genes are able to respond to relatively low levels of diffusible Wg. In addition, the absence of Doc expression in the dorsalmost cells of the ectoderm that receive Wg and Dpp signals indicates the presence of a negative regulator that prevents Doc induction in the ectoderm adjacent to the amnioserosa until stage 12. Together, these inputs restrict Doc expression to metameric quadrants that encompass the areas of the dorsolateral ectoderm between the tracheal placodes as well as the underlying mesodermal cells.

Previous studies have shown that some of the effects of wg are mediated by its target gene sloppy paired (slp), including the feedback activation of wg in the ectoderm and the repression of bagpipe (bap) in the mesoderm(Cadigan et al., 1994; Lee and Frasch, 2000). However, the residual (although strongly reduced) expression of the Doc genes in the germ band of slp mutant embryos argues against a role of slp in mediating the function of wg to induce the Doc genes. Hence, the Doc genes may be direct targets of the Wg signaling cascade in the ectoderm and mesoderm.

Our observations show that one of the important functions of the Doc genes in the dorsolateral ectoderm is the repression of wg expression. Although the expression of Doc initially depends on wg, the Doc genes subsequently exert a negative feedback on wg expression, which leads to the previously unexplained interruption of the wg stripes during stage 11. Because the ventral extent of the ectodermal Doc domains correlates with the ventral borders of high levels of P-Mad, we conclude that the dorsal limit of the ventral wg stripes at stage 11 is determined indirectly by Dpp via Doc.

The maintenance of wg after stage 10 has been shown to depend on two different positive feedback loops, one being active in the dorsal and the other in the ventral ectoderm. The dorsal feedback loop is mediated by the ladybird homeobox genes (lb=lbe and lbl),whereas the ventral loop is mediated by the Pax gene gooseberry(gsb) (Li and Noll,1993; Jagla et al.,1997). The Doc genes must interrupt one or both of these feedback loops, although it is not clear whether the primary block is at the level of the wg gene or at the level of the transcription factor-encoding genes lb and/or gsb. Another target for repression by the Doc genes in this pathway could be slp, which is required both dorsally and ventrally in wg feedback regulation(Cadigan et al., 1994; Lee and Frasch, 2000). We think that lb is unlikely to be the primary target of Doc repression as the failure of wg repression temporally precedes the expansion of the lb stripes in Doc mutant embryos. Furthermore, our observation that the Doc genes can also repress wg in other tissue contexts such as the imaginal discs, where gsb, lb and slp are not components of a wg feedback loop, seems to favor the mechanism of a direct repression of the wg gene by Doc.

Taken together, our observations show that dynamic interactions among positive and negative feedback loops, which share wg as a common component, are involved in the dorsoventral and anteroposterior patterning of the embryonic ectoderm. The activity of the Doc genes in negatively regulating wg and lb, as well as their potential positive effects on yet unknown targets in the dorsolateral ectoderm, are expected to be important for the proper dorsoventral organization of the cuticle and sensory organs. In the mesoderm, the metameric expression domains of the Doc genes during stages 9-11 include the dorsal somatic and cardiac mesoderm. Notably, preliminary analysis has revealed defects in dorsal somatic muscle and dorsal vessel development in Doc mutant embryos, which we are currently examining in more detail. Finally, we note that the expression pattern of the Doc genes in the embryonic epidermis is very reminiscent of the pattern of expression and activity of another T-box gene, optomotor-blind (omb), in the pupal epidermis (Kopp and Duncan,1997). Doc and omb expression overlap in the wing discs although, unlike omb, Doc expression is interrupted near the Wg domains (Grimm and Pflugfelder,1996). Furthermore, it has been reported that dominant mutations in the gene Scruffy (Scf) and their revertants genetically interact with omb during abdominal cuticle and wing patterning(Kopp and Duncan, 1997). Because we have mapped the breakpoints of two Scf revertants, Df(3L)Scf-R6 and Scf-R11, directly upstream and downstream,respectively, of the Doc3 gene, it is tempting to speculate that the Scf phenotype is caused by rearranged Doc3. Future studies will clarify the relationship between Scf and Doc genes and establish whether the T-box genes Doc and omb functionally interact during patterning 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 Studies Hybridoma Bank(University of Iowa/NICHD) for antibodies and probes; and D. Kosman for the double-in situ hybridization protocol. This research was supported by a grant to M.F. from the National Institutes of Health (HD30832). The Mount Sinai Confocal Microscopy Shared Resource Facility was supported, in part, with funding from NIH-NCI shared resources grant (1 R24 CA095823-01).