Common and Distinct Roles of DFos and DJun During Drosophila Development

See allHide authors and affiliations

Science  24 Oct 1997:
Vol. 278, Issue 5338, pp. 669-672
DOI: 10.1126/science.278.5338.669


The Drosophila homolog of c-Jun regulates epithelial cell shape changes during the process of dorsal closure in mid-embryogenesis. Here, mutations in the DFos gene are described. In dorsal closure, DFos cooperates with DJun by regulating the expression of dpp; Dpp acts as a relay signal that triggers cell shape changes and DFos expression in neighboring cells. In addition to the joint requirement of DFos and DJun during dorsal closure, DFos functions independently of DJun during early stages of embryogenesis. These findings demonstrate common and distinct roles of DFos and DJun during embryogenesis and suggest a conserved link between AP-1 (activating protein-1) and TGF-β (transforming growth factor–β) signaling during epithelial cell shape changes.

The AP-1 transcription factor complex is activated in response to various stimuli, including cytokines, growth factors, and cellular stresses like ultraviolet irradiation or heat shock. Active AP-1 complexes are dimers of basic-region leucine zipper (bZip) transcription factors, formed predominantly by members of the Jun and Fos gene families (1). In vertebrates, the number of related Jun and Fos family members has complicated analysis of AP-1 function in vivo. c-fos mutant mice are viable and possess defects in bone formation, whereas c-jun mutant mice die at mid-gestation with liver defects (2). In Drosophila, single homologs of c-Jun and c-Fos, DJun and DFra (here referred to as DFos), have been identified (3, 4). Like c-Jun, DJun forms homodimers and heterodimers with DFos and binds to AP-1 DNA binding sites. In contrast to mammalian c-Fos, however, DFos also forms homodimers, suggesting DJun-independent functions (3). DJun mutant embryos die as a result of failed dorsal closure (5-7). Dorsal closure occurs during mid-embryogenesis and results in stretching of the lateral epithelia over the extraembryonic membrane and fusion at the dorsal mid-line. In addition to DJun, mutations in genes encoding Jun NH2-terminal kinase kinase (DJNKK), hep, and Jun NH2-terminal kinase (DJNK), bsk, also block this cell shape change, and thus, dorsal closure (8-10). Expression of dpp in the most dorsal row of cells, the leading edge, is dependent on DJun, DJNK, and DJNKK function (5, 6, 11). Mutations in genes encoding the Dpp receptors Tkv and Put result in a similar phenotype (12). Expression of activated Tkv receptors in the embryonic ectoderm partially rescues the dorsal closure defect of DJNK mutants. We proposed that activation of the DJNK pathway by an unknown signal results in stretching of the leading edge cells and that Dpp acts as a relay signal to induce stretching of more ventrally located ectodermal cells (8).

We identified a genetic locus, kayak (kay), that has an embryonic lethal phenotype similar to DJun mutants (13). The kay 1 allele contains a restriction-site polymorphism in the DFos coding region caused by a single point mutation resulting in an in-frame stop codon at position 1045 of the DFos sequence (Fig. 1, A and B). Expression of the DFos cDNA in the ectoderm completely rescued the dorsal closure defect of homozygouskay 1 andkay 1/kay 2 mutants (Fig.1D) (14). Thus, DFos is encoded by kay, and loss of DFos function results in a phenotype similar to that of DJun mutants. A fraction of rescuedkay 1/kay 2 mutants survived to become adults with split thoraxes (Fig. 1E). The dorsal thorax is made during metamorphosis from the proximal portions of the left and right wing imaginal discs by fusion of these primordia on the dorsal side. The split-thorax phenotype is reminiscent of phenotypes described for rare homozygous hep mutant flies (9). This similarity suggested that DJNKK and DFos are not only required for dorsal closure of embryonic epithelia but also for the dorsal joining of imaginal epithelia during pupal development.

Figure 1

DFos is encoded by thekay locus. (A) Southern blot analysis ofkay 1/TM3 balancer (lane 1) and control (lane 2) genomic DNA digested with Hinc II. The probe used was the DFos cDNA. (B) Depiction of the mutation inkay 1 resulting in the change of a Leu to a stop codon within a Hinc II site at position 1045 of the published sequence (3). The mutation is a T to A transition in the following sequence: CCC ACG TT(A)G ACG (mutant in parentheses). Polymerase chain reaction amplification of mutant and control DNA and sequencing were done as in (5). Control DNA was from thekay 2 mutation, because it was induced in the same background and recovered in the same mutagenesis. In the regions sequenced, the kay 2 sequences coded for the same amino acids as the published DFos sequence (3). (C) Embryonic cuticular phenotype of akay 1/kay 2trans-heterozygote. The dorsal cuticle is partially missing (white arrowheads), and the embryo is open dorsally. In this and subsequent figures, dorsal is up and anterior is to the left unless otherwise noted; in all cuticle preparations, white arrowheads mark the extent of the dorsal hole in the cuticle. All embryos are shown at the same magnification. Cuticle preparations were done as in (8). (D) The embryonic phenotype of kay mutants is rescued by ectodermal expression of DFos. Mutant transheterozygous kay embryos are completely closed dorsally when DFos is expressed in the ectoderm under the control of the 69B Gal4 line. (E) Expression ofDFos can partially rescue kay mutants to pharate adults. Dorsal aspect of a rescued kay mutant heteroallelic combination. All rescued adults had a split thorax (arrow), indicating a failure of the dorsal portion of the imaginal epithelia to fuse properly, and frequently lacked one wing because of improper eversion of the wing blade (large arrowhead). (F) DFos is required in lateral ectodermal cells ventral to the leading edge cells during closure. Transheterozygous mutant kayembryos where DFos was expressed in the leading edge cells and three to four rows of lateral ectodermal cells ventral to the leading edge cells by means of thepannier MD237 Gal4 line (17) were not rescued. (G)kay 1 behaves genetically as a null allele: Thekay 1/Df (3R)01215 embryo is completely open dorsally [compare with the partially closed embryo in (C) and the kay 1 homozygote in Fig. 2A]. (H) The kay locus shows transvection:kay 2/Df (3R)01215 fails to complement only in a In(1)z ae(bx)homozygous mutant background, known to inhibit transvection. The noncomplementing embryos show mild dorsal-open phenotypes. All mutants and Drosophila stocks not described in (5) are described in (16).

The similarity of the DJun and DFos mutant phenotypes and the genetic interactions between DFos, DJun, andDJNK mutants suggested that these factors cooperate in vivo. Like DJun mutations, DFos null alleles completely blocked shape changes that normally occur in the leading edge cells and in more ventrally located epithelial cells during dorsal closure (Fig.2, A and B) (5). Furthermore, the DFos mutant phenotype (Fig. 2, A through E) was enhanced by removal of one copy of DJNK (Fig. 2G) or DJun (Fig. 2F).DFos mutations also dominantly enhanced DJNKmutants (Fig. 2H). As in DJun, DJNK, andDJNKK mutant embryos, dpp expression in the leading edge but not in other tissues at this stage was abolished inDFos mutant embryos (Fig. 2, I and J). Therefore, DJun and DFos control, probably as heterodimers, dpp expression in the leading edge cells.

Figure 2

DFos acts downstream of the DJNK pathway during dorsal closure. (A) A kay 1homozygous mutant embryo, and (B) an anti-Coracle antibody staining of a kay 1 homozygote. The lateral epithelial cells fail to stretch; the leading edge cells are marked by an arrow. Coracle is a band 4.1 homolog and marks the cell membranes (25). The antibody staining were done as in (5). (C) kay 2homozygotes show partial dorsal closure; (D) the corresponding Coracle antibody staining shows partial stretching of leading edge cells (marked by an arrow). (E)kay heteroallelic mutant combination. (F) Mutations in DJun act as a dominant enhancer of a hypomorphickay mutant. The embryo is completely open dorsally [compare with (C)] (26). (G) DJNK,basket (bsk), acts as a dominant enhancer ofkay. The embryo is also completely open dorsally [compare with (E)]. (H) kay mutations dominantly enhance a DJNK (bsk) hypomorphic mutant. (I and J) dpp is expressed in the leading edge cells but is missing in kay 1mutants: (I) wild-type control embryo; (J) kay 1mutant embryos. The arrows point to the leading edge cells. The embryo in (I) is at stage 13; the middle one in (J), at stage 11; and the right one, at stage 12. Stages according to (27).

In vertebrates, c-Jun and c-Fos activities are regulated at various levels. Whereas c-Jun is widely expressed at low levels and activated primarily by NH2-terminal phosphorylation by JNKs, c-fos expression is dynamic and activated in response to various extracellular stimuli (1). A similar dichotomy of DJun andDFos regulation occurs in Drosophila: DJun is widely expressed during embryogenesis and is phosphorylated by DJNK (3-5, 10), whereas DFosexpression is dynamic (Fig. 3, A through H) (3). There is strong expression of DFos in leading edge cells and cells of the lateral epithelium (Fig. 3, E and F).

Figure 3

DFosis expressed dynamically during embryogenesis. DFos RNA in situ hybridization of wild-type embryos. Part of the pattern ofDFos expression has been described in (3). The RNA hybridizations were done as in (8). (A) Stage 2 embryo, showing homogeneous maternally deposited DFos RNA. (B) Stage 6 embryo showing DFos expression in the most dorsal cells, the anlagen of the amnioserosa, and lateral epithelial cells. Residual maternal mRNA is still present at this stage. (C) Stage 7 embryo, showing DFosexpression in the amnioserosa anlagen and lateral epithelial cells. (D) Stage 10 embryo showing DFos staining in the amnioserosa (arrow) and head region. (E) Stage 11 embryo (germband extended stage) showing strong DFos expression in the leading edge cells (arrow) and weaker expression in lateral epithelial cells (arrowhead), but no expression in the amnioserosa, as well as the head region and muscle attachment sites. Some cells of the peripheral nervous system are also labeled. (F) Stage 12 embryo (germband retraction stage) showing prominent leading edge and strong lateral epithelial cells staining (arrow); rest of the pattern as in (E). DFos staining in lateral epithelial cells develops from stage 10 [compare (D) with (F)]. (G) Stage 15 embryo showing staining in a portion of the endoderm (arrow). (H) Stage 16/17 embryo, showing prominent perinuclear staining in the hindgut (arrow) and the Malpighian tubules (one of them is in the plane of focus, arrowhead). (I)DFos expression in lateral epithelial cells ventral to the leading edge cells is reduced in thetkvstrII mutant embryo. A stage 12 embryo [compare with (F)], showing staining only in the leading edge cells and very little, if any, in the lateral epithelial cells (24). Similar results were also obtained with anotherdpp receptor mutant, put135.tkvstrII is a loss of function mutation in tkv, and put135 is an amorphic allele of put (16). (J)DFos expression is reduced in late embryos in the endoderm of dppS4mutants. Arrow points to residual expression; compare with (G). Mutant dpps4embryos lack dpp expression in the visceral mesoderm (16) and were recognized by the lack of the second midgut constriction. (K) race expression in the amnioserosa is missing in kay mutants. The embryo in (K) is heterozygous for the null kay1mutation and shows race expression in the anterior and posterior midgut and in the cells of the amnioserosa (arrow). The homozygous kay1 mutant embryo in (L) is at a slightly later stage (note the position of the anterior and posterior gut endoderm, stained positive forrace in both embryos). This embryo completely lacks expression in the amnioserosa cells (small arrow). This lack of amnioserosa race expression has also been documented fordpp mutant embryos (18, 19). DJunand bsk mutant embryos, including embryos deprived of maternal and zygotic expression, show wild-type raceexpression (21). (M) labialexpression in kay mutants is normal. The embryo is homozygous for kay1; note cuticule formation in the ventral part of the embryo, as evidenced by nonspecific cuticular staining in a striped-like pattern (and its absence dorsally), and the graded nuclear Labial expression in the midgut (arrow). Antibody stainings were done with a Labial polyclonal antibody as in (28). (N) DFos is also downstream of thedpp pathway. The embryo shows strong ectopic DFos expression in stripes in the engrailed (en) expression domain. We expressed the activated Dpp receptor TkvQD in an en pattern by means of an en Gal4 driver. The ectopic expression of DFos is superimposed on the normal DFos expression pattern [compare with (D)].

Expression of DFos in the lateral epithelia is reduced in Tkv and Put mutant embryos but is still detectable in the leading edge (Fig.3I) (15). Therefore, not only does DFos control expression of dpp in the leading edge, but in a reciprocal manner,DFos expression is dependent on Dpp function in cells of the lateral epithelium. Similarly in late embryos, DFosexpression in the endoderm depends on dpp expression in the overlaying visceral mesoderm. In dpp S4 mutant embryos, where dpp is expressed in the ectoderm but not in the visceral mesoderm (16), DFos expression in the underlying endoderm is greatly reduced (Fig. 3J). Activation of the Dpp signaling pathway is indeed sufficient to activate DFosexpression. Expression of an activated Dpp receptor,Tkv QD, in 14 stripes in the ectoderm of developing embryos with the use of an engrailed Gal4 driver (16) results in a corresponding pattern of DFosexpression (Fig. 3N). During dorsal closure, DFos appears to be required in all epithelial cells, because restricted expression ofDFos in the dorsal-most lateral ectoderm with apannier GAL4 line (17) was not sufficient to rescue the dorsal-open phenotype ofkay 1/kay 2 mutant embryos (Fig. 1F). Thus, DFos may be required in all ectodermal cells in order to activate target genes required for cell shape changes (Fig.4).

Figure 4

Model of DFos function during dorsal closure. DFos is required both upstream and downstream of dpp during dorsal closure. Upon activation of the DJNK cascade in the leading edge cells by an unknown signal, activated DJNK translocates to the nucleus, where it phosphorylates the E26-specific (ETS) domain repressor Aop (the gene aop, or anterior open, is also known as yan and also acts in the rl/MAPK pathway) and DJun (5). Aop is inactivated by phosphorylation, whereas phosphorylation of DJun activates DJun. Both activated DJun and DFos are required for target gene expression. dpp is one of these target genes in the leading edge cells. Dpp then diffuses to ectodermal cells ventral to the leading edge cells, where it signals by means of its receptors Tkv and Put. This action then results in DFostranscription, upon which these lateral epithelial cells change shape. AP-1 activity is required for this last process, because DFos is also required in these cells for the change in cell shape. Thus, DFos is required in all lateral epithelial cells for stretching.

In the endoderm of late embryos, Dpp-induced expression ofDFos correlates with induction of the homeotic genelabial (lab) in endodermal cells underlyingdpp-expressing cells (15). Although ubiquitous expression of a dominant negative form of DFos,DFos bZip, blocked lab expression in the endoderm (15), we observed normal levels oflab expression in kay mutants (Fig. 3L). This result indicates that in the endoderm DFos is not essential for Dpp-induced lab expression.

Independent of DJun, DFos also controls the expression of Dpp target genes during early embryonic development. Early dppexpression on the dorsal side of embryos induces expression of several genes, including race, which encodes a protein with homology to angiotensin-converting enzyme in the amnioserosa (18-20) (Fig. 3K). The race cis-acting sequences required fordpp-mediated expression contain AP-1 binding sites (18). Consistent with DFos-mediated direct activation ofrace through these AP-1 sites, race expression in the amnioserosa was abolished in kay mutant embryos (Fig.3L). In contrast, race expression was normal inDJun or bsk mutant embryos (21). Thus, in addition to its role in controlling dpp expression in the leading edge, DFos also performs an essential function as mediator of Dpp during dorsal closure and during early embryogenesis, controlling Dpp target genes such as race. This early DJun-independent function of DFos may be mediated by DFos homodimers.

The joint requirement of DJun and DFos during dorsal closure provides the first genetic evidence for cooperation of these transcription factors during development. Defects in the dorsal joining of embryonic and adult epidermis observed in DFos and hepmutants suggest a common role of the DJNK pathway in joining of epithelial sheets. Control by AP-1 of epithelial cell morphology and possibly the composition of the extracellular matrix may be a universal phenomenon: During wound healing in vertebrates, a process that exhibits parallels with dorsal closure (5), TGF-β induces c-fos expression and AP-1 activity. The reciprocal regulatory relation between DFos and dpp inDrosophila appears to be conserved in mammalian cells. In mammalian myeloid cells, induction of c-jun and c-fos by serum or oncogenic v-src results in expression of TGF-β1 by direct activation ofTFG-β1 transcription by AP-1 (22). Furthermore, TGF-β induces AP-1 activity in keratinocytes during wound healing (23). The parallels between TGF-β and AP-1 signaling in mammalian cells and Dpp signaling and DFos and DJun signaling in Drosophila provide a striking example of evolutionary conservation of a complex regulatory pathway in epithelial cell morphogenesis.

  • * To whom correspondence should be addressed. E-mail: hafen{at}


View Abstract

Navigate This Article