Abstract
During Drosophila embryogenesis homeotic genes control the developmental diversification of body structures. The genes probably coordinate the expression of as yet unidentified target genes that carry out cell differentiation processes. At least four homeotic genes expressed in the visceral mesoderm are required for midgut morphogenesis. In addition, two growth factor homologs are expressed in specific regions of the visceral mesoderm surrounding the midgut epithelium. One of these, decapentaplegic (dpp), is a member of the transforming growth factor β (TGF-β) family; the other, wingless (wg), is a relative of the mammalian protooncogene int-1. Here we show that the spatially restricted expression of dpp in the visceral mesoderm is regulated by the homeotic genes Ubx and abd-A. Ubx is required for the expression of dpp while abd-A represses dpp. One consequence of dpp expression is the induction of labial (lab) in the underlying endoderm cells. In addition, abd-A function is required for the expression of wg in the visceral mesoderm posterior to the dpp- expressing cells. The two growth factor genes therefore are excellent candidates for target genes that are directly regulated by the homeotic genes.
Introduction
Mutations in Drosophila homeotic genes transform one part of the fly into another. The best studied phenotypes involve changes in the formation of the appendages and cuticle of the adult and of the larval cuticle, but the genes affect internal structures as well. We have used the development of the midgut to analyze the roles of homeotic genes in forming a normal embryo, to characterize interactions among the homeotic genes, and to identify genes that are regulated by the homeotic genes.
Many of the Drosophila homeotic genes are grouped in two clusters known as the Antennapedia complex (ANT-C) and the bithorax complex (BX-C) (Lewis, 1978; Kaufman et al. 1980; reviewed in Akam, 1987; Ingham, 1988). Apparently homologous gene clusters exist in vertebrates (Gaunt et al. 1988; Duboule and Dolle, 1989; Graham et al. 1989; reviewed in Akam, 1989). Five homeotic genes are in the ANT-C, labial (lab), proboscipedia (pb), Deformed (Dfd), Sex combs reduced (Scr), and Antennapedia (Antp), and three in the BX-C, Ultrabithorax (Ubx), abdominal-A (abd-A), and Abdominal-B (Abd-B) (reviewed in Mahaffey and Kaufman, 1988). The lab, pb, Dfd, and Scr genes act in anterior parts of the fly, while Antp is required in the thoracic segments and to some extent in the abdomen. Ubx is required for the proper formation of the posterior thorax and the anterior abdomen, abd-A for the central abdomen, and Abd-B for the posterior abdomen.
The mechanisms by which homeotic genes influence morphogenesis remain largely unknown. All of the ANT-C and BX-C homeotic genes encode proteins with homeodomains, sequence-specific DNA-binding domains (reviewed in Scott et al. 1989), and several of the protein products of the homeotic genes have been shown to function as transcription factors (Jaynes and O’Farrell, 1988; Han et al. 1989; Driever and Niisslein-Volhard, 1989; Hanes and Brent, 1989; Krasnow et al. 1989; Struhl et al. 1989; Winslow et al. 1989). These observations strongly suggest that the genes control morphology by coordinating the spatial and temporal transcription patterns of target or ‘realizator’ genes (Garcia-Bellido, 1977). In the epidermis, the nervous system, and the visceral mesoderm, cross-regulatory interactions have been observed among the homeotic genes; thus the first target genes to have been identified are homeotic genes (Hafen et al. 1984; Struhl and White, 1985; Carroll et al. 1986; Casanova and White, 1989). In addition some of the homeotic genes have been found to autoregulate, including one, Ubx, that activates its own expression in the visceral mesoderm (Bienz and Tremml, 1988). There must, however, be other genes that are controlled by homeotic genes but are more directly linked to the processes of cell differentiation and morphogenesis. In an attempt to identify such target genes we have looked for genes that are transcribed in restricted regions of the embryo overlapping with domains of homeotic gene expression. Two genes, wingless (wg) and decapentaplegic (dpp), that encode putative growth factors are expressed in the visceral mesoderm of the developing midgut in patterns that suggest they may be regulated by homeotic genes.
Several of the homeotic genes, including lab, Scr, Antp, Ubx, andabd-A, are expressed in discrete regions in the developing midgut. The midgut is made up of two cell layers: an inner layer of large polyploid endoderm cells encased in a sheath of smaller, euploid visceral mesoderm cells. The four homeotic genes Scr, Antp, Ubx, and abd-A are expressed in non-overlapping domains in the visceral mesoderm cells in an order along the anterior-posterior axis similar to that in the epidermis, i.e. Scr is active in the most anterior region of the mesoderm, then Antp, then Ubx, and abd-A in the most posterior domain (LeMotte et al. 1989; Tremml and Bienz, 1989; Reuter and Scott, 1990).
Defects in gut morphogenesis are seen when any of these four homeotic genes is defective. The morphological alterations seen in the gut in the homeotic mutants generally correspond well with where the products are found in wild-type embryos. The gastric caeca, four tentacle-like protrusions of the midgut at its anterior end, do not form if Scr function is lacking (Reuter and Scott, 1990). Constrictions that divide the gut into sections normally form at three positions along the anterior-posterior axis of the gut tube, and each constriction is dependent on homeotic gene function (Tremml and Bienz, 1989; Reuter and Scott, 1990). The anterior constriction that underlies the mesoderm cells that express Antp does not form in mutant embryos that lack Antp function. The middle constriction that normally forms near the interface between Ubx and abd-A expression fails to develop in either Ubx or abdA mutants, and the posterior constriction that underlies abd-A expression does not form if abd-A function is lacking. There is evidence that these morphological features are formed by the imposition of the structure upon the endoderm by the mesoderm (Reuter and Scott, 1990); however, it is possible that both cell layers are active participants in the formation of the constrictions. The lab gene is expressed only in a subset of the endoderm cells, and defects in the formation of the second midgut constriction have been reported in lab mutants (Immerglück et al. 1990). Thus, a key element of the control of differentiation by homeotic genes in the midgut, as well as in the epidermis, is the differential spatial expression pattern of each gene.
The sequencing of wg revealed that its product is closely related to the product of the mammalian gene int-1 (Rijsewijk et al. 1987). The wg gene is expressed in early embryos in stripes corresponding to the posterior part of each parasegment (PS; Baker, 1987) and is one of several Drosophila genes involved in establishing polarity of the embryo (Ingham, 1988). The protein is made in about one-quarter of the cells in each segment primordium, where the RNA is seen, and then some of the protein appears to move into adjacent cells (van den Heuvel et al. 1989). Studies of mosaic adult flies in which some of the cells lack all wg function while other surrounding cells have one functional copy of the gene have shown that mutant cells can be rescued by the neighboring wild-type cells (Morata and Lawrence, 1977). These observations, as well as the change in the polarity of cell structures in parts of segments in wg mutant embryos where its RNA is not transcribed, suggest that the protein is involved in cell-cell communication processes. The wg products are also detected in the developing midgut, in a narrow band of visceral mesoderm cells (van den Heuvel et al. 1989) suggesting that it also has a function in the differentiation of the gut. The location of wg products in the midgut led us to examine whether wg is regulated by homeotic genes of the BX-C. We demonstrate that wg transcription is activated by abd-A, but only in a discrete subset of the visceral mesoderm cells where abd-A is expressed, indicating that abd-A is necessary, but not sufficient, for wg expression.
The decapentaplegic (dpp) gene was first recognized for its effects on the development of adult structures derived from imaginai discs (Spencer et al. 1982; St Johnston et al. 1990), and was subsequently found to be involved in other processes such as the development of proper dorsal-ventral patterning (Irish and Gelbart, 1987). It encodes a product that is a member of the TGF-/3 family of growth factors (Padgett et al. 1987). The protein, synthesized in cultured cells, has been found to be cleaved and secreted, suggesting that it is involved in carrying information between cells (Panga-niban et al. 1990a). The pattern of expression of the gene is complex, in keeping with the gene’s functions in many different cells and tissues (St Johnston and Gelbart, 1987; Masucci et al. 1990; Jackson and Hoffmann, in preparation). One of the places that dpp RNA and protein is found is in the developing midgut. The RNA is detectable only in the visceral mesoderm cells in particular positions, while the protein is found in the same cells as the RNA and additionally in some of the underlying endoderm cells (Panganiban et al. 19906).
We have explored the relationship between Ubx and dpp further by studying the induction of dpp by ectopic expression of Ubx. We demonstrate that Ubx induces the expression of dpp in visceral mesoderm cells of the anterior half of the midgut. Induction of dpp by Ubx is restricted to the mesoderm cells suggesting that Ubx must act together with other factor(s) limited to the visceral mesoderm to induce dpp, or that the action of Ubx upon dpp is blocked in all tissues but the visceral mesoderm, dpp expression in the posterior of the midgut is inhibited by abd-A, even in the presence of Ubx protein. The action of abd-A upon dpp is therefore not due to repression of Ubx by abd-A. The ectopic production of Ubx protein, and hence dpp protein, in the visceral mesoderm leads to ectopic expression of lab in many of the underlying endoderm cells.
Materials and methods
Fly strains
Stocks are described in Lindsley and Grell (1968) and additional references as follows. The mutant stocks used were st Ubx6.28e ca shd/JM\ [(Kerridge and Morata, 1982), no detectable Ubx protein due to the deletion of 32 bp in the 5 ′ exon which causes a frameshift at codon 27 and premature stop of translation (Beachy et al. 1985; Weinzierl et al. 1987)], ahd4MX1/TM6B [(Sanchez-Herrero et al. 1985) no detectable abd-A protein (Karch et al. 1990)], Df(3R)bxd100/TM1 [deficient for the Ubx transcription unit (Lewis, 1978; Bender et al. 1983)], Df(3R)Ubx109 gl es/Dp (3;3) P5, Sb [(Lewis, 1978), deficient for the Ubx and abd-A transcription units (Karch et al. 1985)], Df(3R)P9/Dp(3;3)P5, Sb [(Lewis, 1978), deficient for the complete bithorax complex (Karch et al. 1985) ], ScrwL7Ki/TM3 Sb and Scrk6cu pp/TMS. Sb [(Waki-moto and Kaufman, 1981), no detectable Scr protein (Riley et al. 1987)], Antpw10red e/TM3, Sb Ser [(Wakimoto and Kaufman, 1981), no detectable Antp protein (Carroll et al. 1986) ], Df(3R)AntpNs+RC3th st cu e11/TM3, Sb Ser [Deficiency which takes out the Antp protein-coding region (Struhl, 1981; Garber et al. 1983)].
The transformant line hsUbx which carries an Ubx cDNA under the control of the hsp70 promoter on a rosy− chromosome was provided by G. Struhl (Gonzalez-Reyes et al. 1990); a similar construct was provided by R. Mann and D. Hogness. The transformant line HSU-109 which carries a heat shock promoter-controlled Ubx cDNA on a Df(3R)t/bx109 chromosome was provided by A. Gonzalez-Reyes (Gonzalez-Reyes et al. 1990).
Antibodies
The origins of the antibodies used in this study are as follows. The anti-dpp antibody was generated as described by Panganiban et al. (1990a). The hybridoma line FP3.38 (anti-Ubx) was kindly provided by White and Wilcox (1984), the hybridoma line 4C3 (anti-Antp) by D. Brower (J. Condie and D. Brower, unpublished results) and the hybridoma line 6H4 (anti-Scr) by Glicksman and Brower (1988). The anti-lab antibody was kindly provided by T. Kaufman (Diederich et al. 1989), the anti-abd-A antibody by W. Bender (Karch et al. 1990) and the anti-twtg/ess antibody by R. Nusse (van den Heuvel et al. 1989). The antibodies were used as hybridoma supernatants diluted 1:10 (anti-Ubx) or 1:3 (anti-Scr and anti-Antp) in PBT (see below) or as affinity-purified antibodies diluted 1:200 (anti-abd-A), 1:500 (anti-iv/ng/ess) or to 1 μgml −1 (anti-dpp) in PBT.
Heat-shock expression of the Ultrabithorax protein
Eggs were collected for 3h or overnight at 25 °C and the embryos were heat-shocked at 37 °C for two times 20 min with a 20 min period at 18 °C in between. Subsequently, the embryos were kept for three hours or the time period given in the text at 25 °C for recovery.
Immunostaining of embryos
Embryos were fixed as described by Karr and Alberts (1986) and were immunostained as described by Reuter and Scott (1990). Staging was according to Campos-Ortega and Harten-stein, 1985. Antibodies raised in rabbits were immunolocalized using biotinylated secondary antibodies (Vector Lab.) and the ELITE peroxidase detection kit from Vector Lab. For the double label immunostaining the procedure was modified as follows. Fixed and PBT-saturated embryos were first simultaneously incubated with two primary antibodies from different species, for example anti-Ubx and anti-dpp, and then, after extensive washing, with a dilution of goat antimouse IgG peroxidase conjugate (Biorad) and of biotinylated goat anti-rabbit IgG (Vector Lab.). The embryos were washed again, and the bound murine antibodies were detected by oxidation of diaminobenzidine (DAB) in the presence of nickel and cobalt ions which gives a dark gray stain (Lawrence et al. 1987). The DAB solution was thoroughly removed before peroxidase activity was destroyed by the addition of hydrogen peroxide to 3 %. Subsequently, the rabbit antibody was detected after incubation with an avidin-peroxidase complex (Vector Lab) by the oxidation of DAB in the absence of heavy metal ions which gives a brown stain.
Results
Ultrabithorax expression is required for the expression of decapentaplegic in the visceral mesoderm
dpp is expressed in three regions of the visceral mesoderm surrounding the midgut: anteriorly, in the mesoderm covering the developing gastric caeca, centrally, close to the position where the middle constriction of the midgut forms, and posteriorly, at the site of the third midgut constriction (St Johnston and Gelbart, 1987; Panganiban et al. 1990b; Fig. 1A). The third constriction expression is relatively weak and usually not detected by the dpp antibody. The homeotic gene Ubx was also found to be expressed in the central region of the visceral mesoderm, in PS 7 (White and Wilcox, 1984; Bienz and Tremml, 1988). Indeed, when dpp and Ubx protein are simultaneously detected in embryos by a double label experiment, they are expressed in overlapping domains (Fig. 1C). The posterior borders of their expression coincide precisely; anteriorly dpp protein is detected half a parasegment anterior of the Ubx protein.
Expression of decapentaplegic protein in the visceral mesoderm of wild-type and Ubx null mutant embryos. Optical sections of whole mount preparations viewed by differential interference contrast microscopy. (A,B) dpp protein in the visceral mesoderm of wild-type (A) or homozygous Ubx6,28 embryos (B) at stage 13, shown in a dorsolateral view (between arrowheads: dpp expression domains; arrows: endodermal cell cluster; ed: endoderm; vm: visceral mesoderm; hg: hindgut). Ubx and dpp protein (C) or Ubx, dpp, and wg proteins (D) have been simultaneously detected in wild-type or homozygous Ubx6 28 embryos (D) by double-label immunolocalization. Ubx protein is visualized in black and indicated between arrowheads in the visceral mesoderm, dpp (C,D) and wg (D) protein are visualized in brown and pointed out between arrows. No Ubx protein is visible in D due to the Ubx mutation.
Expression of decapentaplegic protein in the visceral mesoderm of wild-type and Ubx null mutant embryos. Optical sections of whole mount preparations viewed by differential interference contrast microscopy. (A,B) dpp protein in the visceral mesoderm of wild-type (A) or homozygous Ubx6,28 embryos (B) at stage 13, shown in a dorsolateral view (between arrowheads: dpp expression domains; arrows: endodermal cell cluster; ed: endoderm; vm: visceral mesoderm; hg: hindgut). Ubx and dpp protein (C) or Ubx, dpp, and wg proteins (D) have been simultaneously detected in wild-type or homozygous Ubx6 28 embryos (D) by double-label immunolocalization. Ubx protein is visualized in black and indicated between arrowheads in the visceral mesoderm, dpp (C,D) and wg (D) protein are visualized in brown and pointed out between arrows. No Ubx protein is visible in D due to the Ubx mutation.
Because Ubx is known to regulate transcription (Krasnow et al. 1989), the overlapping domains of expression of dpp and Ubx led us to investigate whether Ubx controls dpp. In embryos that lack Ubx protein, dpp protein is virtually absent from the visceral mesoderm in PS 6 and 7 (Fig. 1B). Only a faint band of dpp protein remains detectable around the midgut. This is located in the posterior part of the normal dpp expression domain (Fig. IB), directly adjacent to the region where wg and abd-A are expressed in the visceral mesoderm (Fig. 1D).
The second domain of dpp expression in the visceral mesoderm (Fig. 1A), in cells that later cover the budding gastric caeca, is less than half a parasegment wide and is located anterior to the domain of Scr protein expression (data not shown), dpp expression here is not influenced in embryos by the lack of Ubx function (Fig. IB) and also persists in embryos mutant for either of the homeotic genes Scr or Antp, both of which are normally expressed in the visceral mesoderm (data not shown).
Ultrabithorax expression is sufficient for the expression of decapentaplegic in the visceral mesoderm
To more conclusively test whether or not dpp is activated by Ubx we examined the expression of dpp in embryos that carry an Ubx cDNA under the control of the heat shock hsp70 promoter (HSU) (Mann and Hogness, 1990; Gonzalez-Reyes et al. 1990). One hour after a heat shock Ubx protein can be detected in every cell nucleus of the embryo at high levels (data not shown). In these embryos dpp is ectopically expressed in the visceral mesoderm from the anterior end of the midgut to PS 7 (Fig. 2C). In other tissues, the expression of dpp protein changes little if at all.
Ectopic expression of decapentaplegic in various mutant and transformant embryos, dpp protein in the visceral mesoderm of (A) a wild-type embryo, (B) an abd-AMX1, and (D) a Df(3R)Ubx109 homozygous mutant embryo or (C) a HSU and (E) a HSU-109 embryo three hours after a heat shock. In HSU-109/TM1 progeny about three-quarters of the embryos in the collection make Ubx protein ubiquitously after a heat shock treatment; the others are TM1/TM1. Of those carrying HSU about two-thirds have ectopic dpp expression similar to that of the HSU embryo in C. About one-third of the embryos, presumably the HSU-109 embryos that lack Ubx and abd-A, have ectopic dpp protein throughout the entire visceral mesoderm of the midgut as in E. The homozygous mutant embryos have been identified among their siblings due to their lack of Ubx (D) or of abd-A (B,E) protein expression in double staining experiments. The arrows point to the cluster of large endodermal cells. The arrowheads (A,C) indicate the approximate anterior end of the visceral mesoderm surrounding the developing midgut.
Ectopic expression of decapentaplegic in various mutant and transformant embryos, dpp protein in the visceral mesoderm of (A) a wild-type embryo, (B) an abd-AMX1, and (D) a Df(3R)Ubx109 homozygous mutant embryo or (C) a HSU and (E) a HSU-109 embryo three hours after a heat shock. In HSU-109/TM1 progeny about three-quarters of the embryos in the collection make Ubx protein ubiquitously after a heat shock treatment; the others are TM1/TM1. Of those carrying HSU about two-thirds have ectopic dpp expression similar to that of the HSU embryo in C. About one-third of the embryos, presumably the HSU-109 embryos that lack Ubx and abd-A, have ectopic dpp protein throughout the entire visceral mesoderm of the midgut as in E. The homozygous mutant embryos have been identified among their siblings due to their lack of Ubx (D) or of abd-A (B,E) protein expression in double staining experiments. The arrows point to the cluster of large endodermal cells. The arrowheads (A,C) indicate the approximate anterior end of the visceral mesoderm surrounding the developing midgut.
In embryos homozygous for mutations in the abd-A gene, Ubx is expressed at high levels in posterior regions where abd-A would normally be expressed (Struhl and White, 1985; Bienz and Tremml, 1988). In such embryos dpp is also ectopically expressed in the posterior visceral mesoderm, but not in other tissues (Fig. 2B; Immerglück et al. 1990; Panganiban et al. 19906). Embryos that lack both Ubx and abd-A expression (Df(3R)Ubx109) or the entire bithorax complex (Df(3R)P9) express dpp in a pattern that is different from any dpp pattern seen in wild-type embryos (Fig. 2D). The dpp expression that normally overlaps with Ubx is gone completely, but dpp protein is found in a segmentally modulated pattern, in five faint stripes, in the region of the visceral mesoderm where abd-A is normally expressed. This observation indicates that abd-A negatively regulates dpp in the absence of Ubx. dpp expression in the posterior part of the midgut that is not dependent on the homeotic genes Ubx and abd-A might result from influences of earlier acting genes such as segmentation genes.
abdominal-A negatively regulates decapentaplegic in the visceral mesoderm, but not via Ultrabithorax
In spite of the ubiquitous nuclear Ubx protein in the HSU embryos after a heat shock treatment, dpp protein is not found in the posterior part of the visceral mesoderm where the abd-A protein is present (Fig. 2C). The posterior border of dpp expression is not changed by the heat shock-driven ectopic Ubx expression (compare embryos in Figs 2A and C). In order to test whether abd-A is the negative regulator that prevents the ectopic Ubx from activating dpp, dpp expression in HSU embryos that lack abd-A function was examined. For this purpose embryos were used that carry HSU on a third chromosome that lacks both Ubx and abd-A function (Df(3R)Ubx109) This chromosome will hereafter be referred to as HSU-109 (Gonzalez-Reyes et al. 1990). The absence of the endogenous Ubx gene is required for the experiment because otherwise dpp protein would be ectopically expressed in the posterior part of the visceral mesoderm even without the heat shock-driven expression of Ubx, just due to the derepression of Ubx in abd-A mutant embryos (Fig. 2B). The genotype of the embryos that have dpp protein all through the midgut was established by using an anti-a6d-A antibody (Karch et al. 1990) to show that they lack the abd-A protein (data not shown).
About one-quarter of the heat-shocked embryos obtained from HSU-109 heterozygous parents, those that lack Ubx and abd-A, have ectopic dpp protein throughout the entire visceral mesoderm of the midgut (Fig. 2E). Therefore the removal of a single gene function, abd-A, allows HSU, or the ectopic Ubx in abd-A mutants, to induce dpp expression in the posterior visceral mesoderm. Consequently, dpp is a likely candidate for a target gene directly negatively regulated by abd-A.
Induction of labial requires direct contact with visceral mesoderm cells during early midgut development
The study described in the accompanying paper (Panganiban et al. 1990b) shows that the endoderm receives dpp as a signal from the mesoderm, dpp protein can be detected at the apical side of the endoderm; it moves from one germ layer to the other, and it is required for labial (lab) expression in the central part of the midgut endoderm (Panganiban et al. 1990b). A close inspection of the lab expression in wildtype embryos revealed some interesting details of the signal transduction from visceral mesoderm to endoderm. In stage 13 embryos the endoderm forms a multilayered cluster of cells beneath the expression domain of Ubx and dpp. Of these cells only the outer layer, which is in direct contact with the visceral mesoderm, accumulates lab protein; the adjacent cells of the inner layers do not (Fig. 3A,F). Later, from stage 15 on, the inner endodermal cells become integrated into the single cell layer of the endoderm (Fig. 3E; diagrammed in Fig. 3F). In stage 17 endodermal cells within the epithelial sheet that do not express lab can be clearly distinguished from cells that do express it (Fig. 3B,D,E). Thus even though at stages 15 –17 dpp protein is made and is moving between germ layers, some cells still do not produce lab protein in response.
Expression of labial protein in the endodermal cells of embryos after germ band retraction, lab protein in the endoderm of (A,C) a stage 14 and (B,D,E) a stage 17 embryo. (C) A close view of the middle region of the midgut of a stage 14 embryo (cells in the same region delimited by arrowheads in panel A, but from a different embryo). Arrows in C point to the nuclei of the large endodermal cells that are located beneath the lab expressing cells and do not express lab. The large cells have a remarkably pronounced nucleoli (yk: yolk; ed: endoderm; vnc: ventral nerve chord). (D) Close tangential view of a stage 17 embryo similar to the one in panel B. Cells of the lab-expressing midgut loop viewed with a low depth of field. Arrows indicate non-lab-expressing cells that probably had been part of the endodermal cluster and are now integrated into the single-layered endoderm. (E) Optical section of a similar midgut loop. Arrow has same meaning as in D. (F) Schematic summary of the proposed integration of the cluster of endodermal cells (◍) into the midgut epithelium within the lab expressing domain (⚉).
Expression of labial protein in the endodermal cells of embryos after germ band retraction, lab protein in the endoderm of (A,C) a stage 14 and (B,D,E) a stage 17 embryo. (C) A close view of the middle region of the midgut of a stage 14 embryo (cells in the same region delimited by arrowheads in panel A, but from a different embryo). Arrows in C point to the nuclei of the large endodermal cells that are located beneath the lab expressing cells and do not express lab. The large cells have a remarkably pronounced nucleoli (yk: yolk; ed: endoderm; vnc: ventral nerve chord). (D) Close tangential view of a stage 17 embryo similar to the one in panel B. Cells of the lab-expressing midgut loop viewed with a low depth of field. Arrows indicate non-lab-expressing cells that probably had been part of the endodermal cluster and are now integrated into the single-layered endoderm. (E) Optical section of a similar midgut loop. Arrow has same meaning as in D. (F) Schematic summary of the proposed integration of the cluster of endodermal cells (◍) into the midgut epithelium within the lab expressing domain (⚉).
Ectopic decapentaplegic protein activates labial expression in the endoderm
Here we provide further evidence that dpp protein spatially controls lab expression in the endoderm. In HSU embryos lab is detectable three hours after heat shock in most of the anterior endoderm of the midgut, underlying the cells where dpp is ectopically activated in the visceral mesoderm by Ubx. (Fig. 4A). The ectopic lab expression is transient; five hours after the heat shock the amount of lab protein in the anterior part of the midgut has decreased (Fig. 4B). Even in HSU embryos after a heat shock treatment the lab protein is restricted to the outer endodermal layer, where the cells are in direct contact with cells that are producing dpp protein-(Fig. 4E,F). The anterior constriction of the midgut and the constrictions separating the gastric caeca are not formed in these embryos, confirming that the appropriate spatial distributions of homeotic and dpp gene products are required for midgut morphogenesis.
Expression of labial protein in HSU embryos after heat shock. Expression of lab protein in a HSU embryo of approximately stage 13 three hours after heat shock (A) and a HSU embryo of approximately stage 16 five hours after heat shock (B). The arrowheads indicate the relatively sharp anterior border of the ectopic lab domain within the midgut endoderm. (C) High magnification view of a HSU embryo three hours after heat shock at approximately stage 13. The arrows point to the cells of an endodermal cluster which, in the heat-shocked embryos as in normal embryos, do not express lab protein. (D) Lateral view of the anterior midgut of a stage 14 HSU embryo after heat shock. (E,F) Close lateral view in two focal planes of the midgut of a stage 14 HSU embryo after heat shock which reveals the large endodermal cells that do not express lab (F, arrows) beneath the endodermal cells that contain large amounts of lab protein.
Expression of labial protein in HSU embryos after heat shock. Expression of lab protein in a HSU embryo of approximately stage 13 three hours after heat shock (A) and a HSU embryo of approximately stage 16 five hours after heat shock (B). The arrowheads indicate the relatively sharp anterior border of the ectopic lab domain within the midgut endoderm. (C) High magnification view of a HSU embryo three hours after heat shock at approximately stage 13. The arrows point to the cells of an endodermal cluster which, in the heat-shocked embryos as in normal embryos, do not express lab protein. (D) Lateral view of the anterior midgut of a stage 14 HSU embryo after heat shock. (E,F) Close lateral view in two focal planes of the midgut of a stage 14 HSU embryo after heat shock which reveals the large endodermal cells that do not express lab (F, arrows) beneath the endodermal cells that contain large amounts of lab protein.
In the part of the midgut endoderm posterior to the developing gastric caeca, ectopic dpp protein is sufficient to induce lab expression in the neighboring germ layer (Fig. 4D). However, the ectopic domain of lab protein does not extend as far to the anterior as the ectopic dpp in the heat-shocked HSU embryos (Fig. 4A,C). lab is not expressed in the endoderm of the budding caeca or anterior to them.
abdominal-A expression is required for the expression of wingless in the visceral mesoderm
In regions of the midgut where cells do not make dpp protein, communication between germ layers is most likely to be mediated by other growth factors. The only other Drosophila growth factor known to be produced in the midgut is the protein encoded by the wg gene. The segment polarity gene wg, the Drosophila homolog of the int-1 oncogene, was found to be expressed in PS 8 of the visceral mesoderm (van den Heuvel et al. 1989). abd-A is required for wg expression in the visceral mesoderm. Embryos homozygous for a null mutation of abd-A (Fig. 5C), or lacking part of (not shown) or the entire bithorax complex (Fig. 5D), do not accumulate wg protein in the visceral mesoderm. Thus wg looks like a good candidate for a target gene positively regulated by abd-A. However, abd-A cannot be sufficient to control the spatial distribution of wg because wg is only expressed in the most anterior part of the broad visceral mesoderm region where abd-A protein is made (Fig. 5A). This function is probably not Ubx as Ubx628 mutants retain wild-type wg expression (Fig. 5B). However, at least one other BX-C function is required for wg expression because Df(3R) bxdlOO embryos lack wg products in PS 8 (Fig. 5E). In contrast to another report (Immerglück et al. 1990), we find that the wg domain overlaps those of Ubx and dpp (Fig. 6A, B, C). Some visceral mesoderm cells at the anterior rim of the wg domain have Ubx protein in their nuclei and wg protein at the cell periphery (Fig. 6B). This could be due to movement of the wg protein rather than to transcription of both genes in some of the same cells. We have observed movement of wg protein from the visceral mesoderm to the apical side of the endoderm in older embryos (Fig. 6C).
wingless protein expression in the visceral mesoderm depends on abdominal-A function, wg protein in the visceral mesoderm of (A) a wild-type embryo, (B) a Ubx6 28, (C) a abd-A™1, and (D) a Df(3R)P9 homozygous mutant embryo. The tvg-expressing visceral mesoderm cells are indicated by arrowheads. The arrows point at the endodermal clusters. While the wg expression in the visceral mesoderm of the Ubx6 28 embryo is indistinguishable from the wild-type pattern, wg protein is absent from this tissue in all mutant embryos that lack abd-A function (C,D). (E) In embryos deficient for abx, bx, Ubx, pbx, and bxd functions, i.e. Df (3R)bxd100 homozygotes, wg expression is absent from the visceral mesoderm. The embryos are between stages 13 and 14 of development; anterior is oriented to the left.
wingless protein expression in the visceral mesoderm depends on abdominal-A function, wg protein in the visceral mesoderm of (A) a wild-type embryo, (B) a Ubx6 28, (C) a abd-A™1, and (D) a Df(3R)P9 homozygous mutant embryo. The tvg-expressing visceral mesoderm cells are indicated by arrowheads. The arrows point at the endodermal clusters. While the wg expression in the visceral mesoderm of the Ubx6 28 embryo is indistinguishable from the wild-type pattern, wg protein is absent from this tissue in all mutant embryos that lack abd-A function (C,D). (E) In embryos deficient for abx, bx, Ubx, pbx, and bxd functions, i.e. Df (3R)bxd100 homozygotes, wg expression is absent from the visceral mesoderm. The embryos are between stages 13 and 14 of development; anterior is oriented to the left.
wg protein migrates to the endoderm. (A) In stage 17 embryos wg protein (brown stain) is detectable at the apical side of the endoderm (arrowheads), i.e. the side towards the lumen of the midgut, indicating that it has moved from the visceral mesoderm where wg protein is seen at an earlier stage (Fig. 3B in van den Heuvel et al. 1989). The adjacent visceral mesoderm cells express Ubx (black staining; arrows) The wg expressing midgut loop is visible to the posterior of the Ubx domain, to the left in the photograph, because the loop has inverted its position during the convolution of the midgut. (B) High magnification view of the central region of the midgut of a stage 15 embryo. Ubx protein in the visceral mesoderm nuclei is stained in black (arrowheads); wg protein is visualized in brown (arrows). The more posteriorly located Ubx protein-containing cells (to the right) also display wg protein on their surfaces. (C) Central region of the visceral mesoderm of a stage 13 embryo which has been stained for both dpp and wg protein (in brown) and Ubx protein (in black).
wg protein migrates to the endoderm. (A) In stage 17 embryos wg protein (brown stain) is detectable at the apical side of the endoderm (arrowheads), i.e. the side towards the lumen of the midgut, indicating that it has moved from the visceral mesoderm where wg protein is seen at an earlier stage (Fig. 3B in van den Heuvel et al. 1989). The adjacent visceral mesoderm cells express Ubx (black staining; arrows) The wg expressing midgut loop is visible to the posterior of the Ubx domain, to the left in the photograph, because the loop has inverted its position during the convolution of the midgut. (B) High magnification view of the central region of the midgut of a stage 15 embryo. Ubx protein in the visceral mesoderm nuclei is stained in black (arrowheads); wg protein is visualized in brown (arrows). The more posteriorly located Ubx protein-containing cells (to the right) also display wg protein on their surfaces. (C) Central region of the visceral mesoderm of a stage 13 embryo which has been stained for both dpp and wg protein (in brown) and Ubx protein (in black).
Summary of the expression patterns of dpp, wg, Ubx, abd-A, and lab proteins in the visceral mesoderm of various mutant or transformant embryos. The graph summarizes the expression patterns of Ubx, abd-A, dpp, wg, and lab (see Panganiban et al. 1990b) in the visceral mesoderm or the endoderm of the midgut in wild-type (wt), in Ubx null mutant embryos (Ubx628,Ubx−), in abd-A null mutant embryos (abd-AMx1,abd-A−) or in embryos (Df(3R)Ubx109 lacking both Ubx and abd-A function (Ubx-109), and in heat-shocked HSU or HSU-109 embryos. The approximate parasegmental positions of expression were determined relative to the segmental grooves of the ectoderm during germ band retraction. The expression patterns of lab in HSU-109 embryos, and those of wg in Ubx109 and HSU-109 embryos were not determined.
Summary of the expression patterns of dpp, wg, Ubx, abd-A, and lab proteins in the visceral mesoderm of various mutant or transformant embryos. The graph summarizes the expression patterns of Ubx, abd-A, dpp, wg, and lab (see Panganiban et al. 1990b) in the visceral mesoderm or the endoderm of the midgut in wild-type (wt), in Ubx null mutant embryos (Ubx628,Ubx−), in abd-A null mutant embryos (abd-AMx1,abd-A−) or in embryos (Df(3R)Ubx109 lacking both Ubx and abd-A function (Ubx-109), and in heat-shocked HSU or HSU-109 embryos. The approximate parasegmental positions of expression were determined relative to the segmental grooves of the ectoderm during germ band retraction. The expression patterns of lab in HSU-109 embryos, and those of wg in Ubx109 and HSU-109 embryos were not determined.
Ubx may not be required for spatially restricted expression of wg in the visceral mesoderm despite the (limited) physical overlap of Ubx and wg products. Embryos homozygous for the Ubx6 28 allele, which has 32 base pair deletion at codon 27 (Beachy et al. 1985; Weinzierl et al. 1987) and therefore probably has little or no Ubx function, have wg protein in an unchanged pattern (Fig. 5B) as do HSU embryos in which ectopic Ubx expression has been induced (not shown). However, a function within the bithorax complex other than abd-A is required for normal wg expression. In embryos homozygous for Df(3R)bxd100, which removes the Ubx and the bxd transcription units from the third chromosome, or in embryos bearing Df(3R)bxd100 in trans to Df(3R)P9, which removes all of the BX-C, not only are Ubx and dpp expression gone from the visceral mesoderm, but wg expression is also severely reduced (Fig. 5E).
Discussion
The requirement for homeotic gene function in establishing a wild-type body plan is well documented (reviewed in Mahaffey and Kaufman, 1988); however, the mechanism by which these genes act is just beginning to be elucidated. Part of this mechanism clearly involves inter-regulation among the homeotic genes (Hafen et al. 1984; Struhl and White, 1985; Carroll et al. 1986; Riley et al. 1987; Casanova and White, 1987). Another part of this mechanism is likely to involve the regulation of target genes which implement the program set up by the homeotic genes. These target genes could mediate intercellular interactions as well as intracellular differentiative events. Recently, two putative targets that are not homeotic genes, dpp and wg, have been postulated based on their altered expression patterns in the visceral mesoderm of embryos mutant for Ubx or abd-A (Immerglück et al. 1990; Panganiban et al. 1990b). In this report, we examine the effect of ectopic expression of Ubx during gastrulation on the expression of dpp. We conclude that Ubx and abd-A are sufficient to delimit the spatial expression of dpp in the middle part of the visceral mesoderm. Although we have not proven that dpp is directly regulated by the Ubx protein, this possibility is the simplest explanation for our observations. Further evidence consistent with a direct interaction model comes from experiments in which binding sites for the Ubx protein were found in or near the dpp gene (P.A. Beachy; J. Botas and D.S. Hogness, personal communications).
Ubx is required for dpp expression
In embryos bearing the putative Ubx protein null mutation, Ubx6 28, the expression of dpp is reduced in the visceral mesoderm of PS 6 and 7 to a faint band in posterior PS 7. In embryos homozygous for deficiencies of the bithorax complex that remove the Ubx transcript, e.g. Df(3R)bxd100, dpp expression is completely eliminated in the visceral mesoderm of PS 6 and 7. These data indicate that Ubx is required for dpp expression in the central region of the midgut, and are essentially in agreement with the report of Immerglück et al. (1990). They reported that in embryos bearing a Ubxx1 mutation in trans to a triple mutation of the bithorax complex that eliminates all its known protein-coding functions (Casanova et al. 1987) there is no dpp expression in the middle of the midgut visceral mesoderm. We are not yet sure why in embryos homozygous for Ubx6 28 there is some spatially limited dpp expression, while in the other cases mentioned dpp expression in the midgut has vanished completely, t/bx6 28 has a deletion of 32 bp which creates a stop codon after only 27 codons and therefore is likely to be a null mutation in Ubx (Weinzierl et al. 1987), but perhaps it has residual function below detectable levels. Df(3R)bxd100 removes functions in addition to Ubx such as abx, bx, pbx, or bxd; so perhaps one or more of these can affect dpp expression and is also defective in the Ubx1/triple mutant. Another mystery is why dpp expression in the posterior part of PS 6, where no Ubx protein is detected, is dependent upon Ubx function. One possibility is that dpp expression in posterior PS 6 is due to limited paracrine autoactivation by the dpp protein coming from PS 7. Such a localized effect would be prevented in more posterior regions by the negative influence of abd-A on dpp.
Ubx is sufficient for dpp expression
When the synthesis of Antp or Ubx products is induced throughout the embryo using a heat shock promoter, the effects of the proteins are limited to specific tissues and to specific regions of the embryo (Gibson and Gehring, 1988; Kuziora and McGinnis, 1988; Mann and Hogness, 1990; Gonzalez-Reyes et al. 1990). For example, only the cuticle of an embryo anterior to the normal region where Ubx is expressed is affected by expression of Ubx in all cells. Ectopic Ubx protein in the posterior abdomen has no detectable effect on cuticle patterning. Thus cells in certain positions are permissive for the actions of the regulator while others are not. The restriction of the effect is at least in part due to competition between homeotic proteins. The presence of abd-A products prevents Ubx protein from having an effect in the posterior abdomen (Gonzalez-Reyes et al. 1990; Mann and Hogness, 1990). Thus it is not simply the negative regulation of Ubx by abd-A (Struhl and White, 1985) that prevents Ubx from deciding the fate of posterior abdominal cells. Rather, groups of cells that contain both abd-A and Ubx protein form structures characteristic of abd-A function, apparently ignoring the presence of the Ubx protein.
A transposon containing the Ubx cDNA under the control of the heat shock promoter was used to ectopically express Ubx throughout the embryo and the pattern of dpp expression was assayed. Within 3h after the heat shock, dpp expression in the HSU embryos extends along the entire anterior end of the midgut in the visceral mesoderm. The rapid time course of dpp induction is consistent with a direct interaction. We observe ectopic dpp and ectopic labial expression in stage 15 and stage 16 embryos. At the time of heat shock, these embryos would have been older than 9 h, a stage which is refractory to HSU induced segmental transformations (Gonzalez-Reyes et al. 1990). Thus the response of dpp to Ubx probably occurs in the absence of general changes in segment identity.
The cellular context affects whether a homeodomain protein is capable of activating a particular target gene. It is striking that Ubx does not govern dpp expression in embryonic tissues outside of the anterior half of the midgut visceral mesoderm. The restriction of the activating effect of Ubx could be due to a transcriptional cofactor present only in the visceral mesoderm cells that is required for the Ubx protein to turn on dpp. This cofactor could be another DNA-binding protein, a protein that affects the alternative RNA splicing events that produce different Ubx proteins, a protein that modifies the Ubx protein, or an adaptor protein that interfaces between Ubx and the basic transcription machinery. Alternatively, there could be a block to Ubx activation of dpp in all tissues except the visceral mesoderm. The block could take the form of a repressor protein that competes with Ubx for binding to the dpp gene, a repressor that competes with Ubx protein for contacting the basic transcription apparatus, an enzyme that modifies the dpp gene or the Ubx protein, or a chromatin component that keeps the dpp gene in an inactive or inaccessible state.
abd-A represses dpp via a Ubx-independent mechanism
We have previously shown that in the absence of abd-A function dpp is derepressed in visceral mesoderm of the posterior midgut (Panganiban et al. 1990b). In this report, we demonstrate that the repression of dpp by abd-A is independent of Ubx. In the heat-shocked HSU embryos, dpp is ectopically expressed only in the anterior midgut and not in the visceral mesoderm of the posterior midgut, where abd-A is expressed. This indicates that the repression of dpp by abd-A can take place even in cells that are also expressing Ubx. That abd-A can repress dpp even in the presence of large amounts of Ubx protein indicates that the mechanism by which abd-A negatively regulates dpp is not by negatively regulating Ubx. In heat-shocked HSU-109 embryos, which are deficient for Ubx and abd-A in addition to harboring the HSU transposon, ectopic dpp expression extends over the posterior as well as the anterior midgut, confirming that abd-A represses Ubx induction of dpp in the posterior half of the midgut. The domination of abd-A over Ubx in the control of dpp provides a specific example of target gene regulation that parallels the observations on the effects of ectopic Ubx on cuticle pattern discussed above. It will be interesting to determine whether Ubx protein does in fact bind to dpp DNA in vivo and whether the presence of abd-A interferes with this interaction.
A further indication of the negative regulation of dpp by abd-A is the appearance of a striped pattern of dpp in Df(3R)Ubx109 embryos which are deficient for Ubx and abd-A. This pattern is reminiscent of segment polarity gene expression patterns and suggests that, in the absence of normal negative regulation of dpp, the gene exhibits an aberrant response to one or more segment polarity genes. In the lateral ectoderm, dpp exhibits a segmentally repeated pattern; thus the gene does possess mechanisms for responding to this sort of pattern information. Presumably, this response is normally overridden in the visceral mesoderm by the action of the homeotic regulatory proteins. In addition, weak dpp expression is normally present in the visceral mesoderm of PS 9 (Jackson and Hoffmann, in preparation). The PS 9 expression of dpp in the visceral mesoderm is regulated by a completely different set of regulatory elements than those required for dpp expression in PS 7 and the gastric caeca (Jackson and Hoffmann, in preparation). It is not known what regulates dpp expression in PS 9, but in the absence of regulatory input from the homeotics, these unknown factors may be able to generate the striped pattern of dpp expression observed in the Df(3R)Ubx™ mutants.
That abd-A can have positive as well as negative effects on target genes is indicated by the observation that embryos null for abd-A, or for the entire BX-C, do not express wg in PS 8 of the visceral mesoderm. However, at least one other function within the BX-C is also required for wg expression: Embryos that are homozygous for Df(3R)bxd100 or trans-heterozygous for Df(3R)bxd100/Df(3R)Ubx109 and therefore still have abd-A function, also lack wg expression. This function may not be Ubx because, in Ubx6 28 mutant embryos, wg expression in PS 8 of the visceral mesoderm is normal. However, as is discussed above, t/6x628 may not be null for all Ubx functions. The absence of ectopic wg products in regions of HSU embryos where abd-A, Ubx, and dpp are all expressed is consistent with the hypothesis that Ubx is not the BX-C function required for wg expression. Immerglück et al. (1990) have reported that dpp expression in PS 7 is required for wg expression in PS 8. Perhaps there is sufficient movement of dpp protein, at levels undetected in the present studies, from PS 7 into the visceral mesoderm of PS 8, to induce wg expression in cells expressing abd-A and not expressing Ubx. The overlap we have observed in the Ubx and wg protein domains might be due to an anterior movement of the wg protein from its site of synthesis into the most posterior part of PS 7. Experiments using heat shock promoter driven ectopic expression of dpp could be used to test this hypothesis.
Reciprocal regulation of dpp and Ubx
Although Ubx is responsible for localized expression of dpp in posterior PS 6 and PS 7 of the visceral mesoderm, we have also observed an effect of dpp on Ubx in the visceral mesoderm (Panganiban et al. 19906). We have not established whether Ubx protein or dpp protein is present first in the visceral mesoderm of PS 7. dpp RNA is clearly detected in the visceral mesoderm by stage 11 as germ band shortening begins (Jackson and Hoffmann, in preparation). We think it most likely that Ubx initiates dpp expression, and that dpp is required to maintain Ubx expression in PS 7. Secreted factors, like dpp, with a limited capacity for moving from their site of synthesis, may provide an ideal means of mediating intercellular interactions in a specific region of the embryo. The intercellular signals may be necessary to maintain particular differentiated cell states among the population of cells, possibly by causing maintenance of homeotic gene expression.
Requirements for lab expression in the midgut endoderm
Expression of dpp in the midgut visceral mesoderm is required for the expression of lab in the underlying endoderm (Immerglück et al. 1990; Panganiban et al. 1990b). We have extended this analysis and find that in order for endoderm cells to express lab, they must have been in contact with the visceral mesoderm cells which express dpp relatively early (i.e. stage 13) during midgut development. The apparent temporal requirement for early exposure to dpp for lab expression may reflect a subsequent commitment by the endoderm cells to specific patterns of differentiation, for which lab expression is a marker.
Early exposure of the endoderm cells to dpp is required, but clearly not sufficient to induce lab. In heat-shocked HSU embryos, ectopic expression of dpp extends over the entire anterior midgut, while lab expression is restricted to the endoderm of the midgut between the gastric caeca and the second constriction. The distribution of other factors (e.g. dpp transport protein or receptor, signal transduction machinery, and/or specific transcription factors) may determine whether particular endoderm cells can respond to dpp. Differences in the responses of parts of the midgut endoderm to dpp were noted previously (Panganiban et al. 1990b). dpp is expressed in wild-type embryos in the visceral mesoderm of the gastric caeca as well as the visceral mesoderm of PS 6 and 7, yet lab expression is only detected in the endoderm cells of PS 6 and 7. The dpp protein has been shown to migrate from the visceral mesoderm across the endodermal cell layer in PS 6 and 7, however, no migration of dpp protein has been observed at the caeca (Panganiban et al. 1990; Panganiban and Hoffmann, unpublished observations). It may be that the dpp protein simply does not migrate to the endoderm at the anterior end of the midgut, and that if it did, lab would be induced in the underlying endoderm.
Gene interactions in the midgut
The interactions among the genes discussed above and by Panganiban et al. (1990b) are summarized in Fig. 8. Immerglück et al. (1990) have independently arrived at many of the same conclusions. At the anterior end of the midgut, both dpp and Scr expression are required for the gastric caeca to form, dpp expression prevents Scr expression in the cells that form the gastric caeca. As the cells expressing Scr act at a distance on the cells that form the gastric caeca, we expect that an as yet unidentified secreted factor, perhaps another growth factor homolog, may be made in response to Scr. Antp expression positively regulates Scr expression, also at a distance, and Antp expression is required for the anterior midgut constriction to form. Mutations in Scr or Antp do not affect the expression of the other homeotic genes, or of dpp or wg. In the adjacent region, we propose that Ubx causes dpp expression, which positively maintains Ubx expression in the visceral mesoderm cells and induces labial expression in the adjacent endodermal cells. Both dpp and Ubx expression are required for these cells to form the middle constriction. In the posterior midgut, abd-A expression negatively regulates Ubx and dpp, and positively regulates wg. The domain of wg expression is restricted to PS 8 - only a portion of the visceral mesoderm cells expressing abdA - by as yet unidentified factors. It has been reported that movement of the wg protein to the apposing endoderm is involved in maintaining lab expression in the endoderm (Immerglück et al. 1990).
Summary of gene functions in the visceral mesoderm. The diagram is a schematic representation of the visceral mesoderm and endoderm of an embryo, not to scale, with anterior to the left. Results are incorporated from the present paper, Tremml and Bienz, 1988; Bienz and Tremml, 1988; Panganiban et al. 1990b, Reuter and Scott, 1990; and Immerglück et al. 1990. In the visceral mesoderm Scr and dpp are required for formation of the gastric caeca. Scr is expressed in cells posterior to the cells that form the caeca and therefore appears to act at a distance to affect caeca formation. The target genes acted upon by Scr are unknown, dpp is active in visceral mesoderm cells that overlie the developing caeca but dpp protein has not been observed to move into the endoderm in this part of the embryo, dpp prevents Scr expression from spreading to the anterior. Antp is expressed in the visceral mesoderm, acts upon unknown targets, and is required for the formation of the anterior midgut constriction. Both Ubx and abd-A functions are required for the formation of the posterior midgut constriction. In PS 7 Ubx prevents Antp transcription and activates dpp, setting the anterior border of dpp expression, dpp protein moves across to the endoderm and activates lab. abd-A represses the expression of both Ubx and dpp and sets the posterior border of dpp and Ubx expression. In PS 8 abd-A is essential for the expression of wg, and the irgprotein moves to the endoderm. The factors controlling the posterior limit of wg expression are not known, wg may activate residual dpp expression in the visceral mesoderm in the absence of Ubx function (not shown in diagram). The regulation of the weak dpp expression in the vicinity of the third constriction has not been studied and is not shown.
Summary of gene functions in the visceral mesoderm. The diagram is a schematic representation of the visceral mesoderm and endoderm of an embryo, not to scale, with anterior to the left. Results are incorporated from the present paper, Tremml and Bienz, 1988; Bienz and Tremml, 1988; Panganiban et al. 1990b, Reuter and Scott, 1990; and Immerglück et al. 1990. In the visceral mesoderm Scr and dpp are required for formation of the gastric caeca. Scr is expressed in cells posterior to the cells that form the caeca and therefore appears to act at a distance to affect caeca formation. The target genes acted upon by Scr are unknown, dpp is active in visceral mesoderm cells that overlie the developing caeca but dpp protein has not been observed to move into the endoderm in this part of the embryo, dpp prevents Scr expression from spreading to the anterior. Antp is expressed in the visceral mesoderm, acts upon unknown targets, and is required for the formation of the anterior midgut constriction. Both Ubx and abd-A functions are required for the formation of the posterior midgut constriction. In PS 7 Ubx prevents Antp transcription and activates dpp, setting the anterior border of dpp expression, dpp protein moves across to the endoderm and activates lab. abd-A represses the expression of both Ubx and dpp and sets the posterior border of dpp and Ubx expression. In PS 8 abd-A is essential for the expression of wg, and the irgprotein moves to the endoderm. The factors controlling the posterior limit of wg expression are not known, wg may activate residual dpp expression in the visceral mesoderm in the absence of Ubx function (not shown in diagram). The regulation of the weak dpp expression in the vicinity of the third constriction has not been studied and is not shown.
The molecular basis of these interactions remains to be determined. These studies present an initial picture of the regulatory circuitry in the midgut of the embryo and establish growth factor homologs in Drosophila as target genes for the pattern information inherent in localized homeotic gene expression. It is likely that the homeotic gene products themselves also regulate the production of other molecules including cytoskeletal components, specialized enzymes, and cell surface components. Our understanding of how specific morphological events occur in response to the homeotic genes will require continued progress in identifying the target genes of the homeotics themselves and in identifying the genes regulated indirectly by molecules like dpp and wg.
ACKNOWLEDGEMENTS
We are very grateful to Welcome Bender, Susan Celniker, Thom Kaufman, Marcel van den Heuvel, Roel Nusse, and Rob White for gifts of antibodies (W.B., anti-afed-A ; S.C., anti-AM-B; T.K., anti-fab; R.N.; anti-wg; R.W., anti-t/bx) and to Gary Struhl, Gines Morata, Acaimo Gonzalez-Reyes, and Richard Mann for sending us fly stocks (HSU transformant flies, G.S. and R.M.; HSU-109 transformant flies G.M. and A.G.-R.). R.R. would like to thank Maria Leptin for space and support in her laboratory. We thank Mariann Bienz for communication of results prior to publication and for stimulating discussions. The research was supported by NIH grant no. 18163 to M.P.S. and ACS grant no. NP-612 to F.M.H. M.P.S. was an Investigator, and R.R. a Research Associate, of the Howard Hughes Medical Institute.