Abstract
The decapentaplegic (dpp) gene product, a member of the transforming growth factor-β family, is required in Drosophila embryos for normal gastrulation and the establishment of dorsal-ventral polarity in the embryo. dpp is also expressed at specific positions in the visceral mesoderm along the developing midgut. We find that mutations that eliminate the visceral mesoderm expression of dpp lead to defects in midgut morphogenesis and alter the spatially localized expression of the homeotic genes Sex combs reduced (Scr), Ultrabithorax (Ubx), and Antennapedia (Antp) in the visceral mesoderm. The extracellular dpp protein migrates from the visceral mesoderm across the apposing endodermal cell layer in a region of the endoderm that expresses the homeotic gene labial (lab). Mesodermal expression of dpp is required for the expression of lab in these endodermal cells indicating that dpp mediates an inductive interaction between the two germ layers. We propose that extracellular dpp protein regulates gut morphogenesis, in part, by regulating homeotic gene expression in the visceral mesoderm and endoderm of the developing midgut.
Introduction
The final stages of midgut development in the Drosophila embryo involve the migration and stretching of cells in two distinct germ layers to enclose the yolk. The inner endodermal layer is composed of large polyploid cells, while the outer visceral mesoderm layer is composed of small euploid cells. The double-walled gut tube is then compartmentalized into four morphologically distinct chambers and four caeca, long appendages extending from the anterior end of the midgut tube. The four chambers of the midgut are created when three constrictions form at specific positions along the length of the midgut. The caeca are formed as buds off the anterior end of the midgut which elongate and extend into the head regions of the embryo. In the final stages of embryogenesis, the compartmentalized midgut stretches out to form a long convoluted tube (Poulson, 1950; Campos-Ortega and Hartenstein, 1985; Reuter and Scott, 1990). The specific roles in gut function of the four midgut chambers are not known.
Four homeotic genes expressed in the visceral mesoderm of the midgut are crucial for gut morphogenesis (Tremml and Bienz, 1989a; Reuter and Scott, 1990). Scr is expressed in the visceral mesoderm cells at the anterior end of the midgut immediately posterior to the budding gastric caeca. In the absence of Scr expression, the gastric caeca do not develop properly (Reuter and Scott, 1990). Adjacent and posterior to the domain of Scr expression is a domain of Antp expression. Antp is expressed by visceral mesoderm cells surrounding the first constriction and, in the absence of Antp expression, the first constriction does not form (Tremml and Bienz, 1989a; Reuter and Scott, 1990). Immediately posterior to the Antp domain is a domain of Ubx expression. Ubx is expressed in the visceral mesoderm cells of the anterior half of the second constriction with expression extending anteriorly toward the first constriction. In the absence of Ubx expression, the second constriction does not form (Bienz and Tremml, 1988; Tremml and Bienz, 1989a; Reuter and Scott, 1990). Immediately posterior to the Ubx domain, abdominal A (abdA) is expressed in the visceral mesoderm cells of the posterior midgut. The anterior limit of the abdA domain is the posterior half of the second constriction and includes the region around the third constriction. Without abdA function, neither the second nor the third constrictions form (Tremml and Bienz, 1989a). The mechanisms by which these four homeotic genes influence the formation of the constrictions are not known, nor is it clear whether constriction formation is driven by the mesoderm or formed through the actions of both germlayers. Constriction formation is associated with the elaboration of extensive microtubular arrrays in the visceral mesoderm cells at the site of the constriction (Reuter and Scott, 1990).
The molecular mechanisms regulating the domains of homeotic gene expression in the visceral mesoderm are not known. The expression domains of the four homeotic genes in the visceral mesoderm of the developing midgut are non-overlapping (Tremml and Bienz, 1989a). This is in contrast to the expression patterns in the embryonic ectoderm where the Antp domain extends posteriorly through those of Ubx and abdA, and the Ubx domain extends posteriorly through that of abdA (Carroll et al. 1988; Karch et al. 1990). The regulatory interactions between the homeotic genes in the visceral mesoderm differ from the regulatory interactions observed between homeotic genes in the ectoderm. For example, Antp negatively regulates Scr in the ectoderm (Riley et al. 1987), but positively regulates Scr in the visceral mesoderm (Reuter and Scott, 1990). The pair rule gene ftz is required for activation of Scr, Antp, and Ubx expression in the ectoderm (Ingham and Martinez-Arias, 1986) but has little effect on the expression of these genes in the visceral mesoderm (Tremml and Bienz, 19896). It seems likely that the regulation of homeotic gene expression in the visceral mesoderm involves genes other than those used in the ectoderm. Genes that are expressed in specific patterns in the visceral mesoderm are good candidates for genes that may participate in the regulation of homeotic gene expression in the visceral mesoderm.
One gene expressed in specific regions of the developing midgut is the Drosophila decapentaplegic (dpp) gene, a member of the transforming growth factor-β (TGF-β) family (Padgett et al. 1987; Reviewed by Hoffmann, 1990). The visceral mesoderm expression of dpp was originally noted in the initial characterization of dp p’s embryonic expression (St. Johnston and Gelbart, 1987). During our studies on dpp expression in wild-type and mutant embryos using in situ hybridization to whole embryos (Tautz and Pfeiffle, 1989), we noted that one domain of dpp expression in the midgut visceral mesoderm appears to be coextensive with the reported parasegment 7 domain of Ubx expression in the midgut visceral mesoderm (Jackson and Hoffmann, in preparation). Our subsequent investigations are the subject of this report and a companion article (Reuter et al. 1990).
dpp is required for normal gastrulation and dorsalventral polarity in the Drosophila embryo (Irish and Gelbart, 1988). Its patterns of embryonic expression bear little resemblance to the anterior to posterior organization of the major patterns of homeotic gene expression. Rather, the ectodermal expression of dpp is regulated along the dorsal to ventral axis of the embryo beginning with a dorsally restricted stripe of dpp mRNA at cellular blastoderm (St. Johnston and Gelbart, 1987; Jackson and Hoffmann, in preparation). Midway through embryogenesis, dorsal and lateral stripes of dpp mRNA are detected which extend along the anterior-posterior axis through all the thoracic and abdominal segments, dpp null mutant embryos exhibit a transformation of dorsal and lateral hypoderm to ventral hypoderm, the most completely ventralized phenotype observed to date in the Drosophila embryo. Mutations that affect the 5 ′ regulatory sequences, a region of five alternate dpp promoters called the short vein region, give rise to embryos with normal dorsal-ventral polarity, but the animals die as larvae with some visible defects in internal organs (Segal and Gelbart, 1985). dpp mutations affecting 3 ′ regulatory sequences, called the disk region, undergo normal embryogenesis and larval development but display an array of defects in the adult cuticular structures derived from the larval imaginai disks (Spencer et al. 1982; St. Johnston et al. 1990; Masucci et al. 1990).
Biochemical analysis of dpp protein produced in cultured Drosophila cells has indicated that dpp, like the other members of the TGF-β family, forms a dimer that is proteolytically processed and secreted from cells (Panganiban et al. 1990). Members of the TGF-β family have been shown to exert a wide variety of effects on cell proliferation, cell differentiation, extracellular matrix elaboration and hormone secretion (Roberts and Sporn, 1988). The effects of any specific factor in the family can vary depending on the target tissue and the dose. While the importance of TGF-β to wound healing has been well established, the roles of these factors in development are less well characterized. Several of the factors are expressed in a tissue specific fashion in developing mouse skin and bone, leading to the hypothesis that members of the TGF-β family establish regulatory cascades of interactions during sequential steps in the development of these tissues (Lyons et al. 1989a). Probably the most pertinent observation about the developmental role of a member of the TGF-β family is the regulation of the Xenopus homeotic gene xhox-3 by a Xenopus TGF-β homolog, XTC-MIF (Ruiz i Altaba and Melton, 1989). XTC-MIF induces animal cap cells to become mesoderm and induces xhox-3 expression to a level that determines an anterior mesodermal fate. A different growth factor, fibroblast growth factor, also induces mesoderm, but establishes higher levels of xhox-3 expression which gives rise to posterior mesoderm. Therefore the presence of a growth factor can specify the positional fate of cells through the regulation of homeotic genes.
We find that a Drosophila member of the TGF-β family, dpp, also regulates homeotic gene expression. dpp negatively regulates the homeotic gene Scr and positively regulates the homeotic gene Ubx in the visceral mesoderm cells of the midgut. Limited migration of dpp protein from the visceral mesoderm to the endoderm is required to induce expression of a third homeotic gene, labial (lab), in the endodermal cells. The expression of dpp in the visceral mesoderm is required for proper gut morphogenesis, specifically for the formation of the gastric caeca and the second midgut constriction. These same two features of gut morphogenesis require expression of Scr and Ubx, respectively.
Materials and methods
The Antp, Ubx and Scr mutant embryos used in these experiments were generated as described by Reuter and Scott (1990). abdA mutant embryos were generated by homozygos-ing the abdAMI allele (Sanchez-Herrero et al. 1985). The dppshv alleles used in these experiments were isolated by Segal and Gelbart (1985). The molecular lesions associated with these mutations are summarized in Fig. 1C and were described by St. Johnston et al. (1990). To generate mutant embryos, balanced stocks of the various dppshv alleles were outcrossed to wild-type Canton S females, dppshv/Canton S virgins were then mated similarly outcrossed males carrying a different dppshv allele. The embryos from these crosses were fixed and stained as described below. To avoid the possibilities that any defects observed were due to mutations other than dpp or to the genetic background in which a given dppshv allele had been generated, each cross was performed with two dppshv different alleles which had been induced on unrelated starting chromosomes.
dpp RNA expression in stage 14 wild-type (A) and dpps21/dppS4 (B) embryos. Lateral views with anterior left and dorsal up photographed at 20 ×. The wild-type embryo expresses dpp in the clypeolabrum (cl), the hindgut (hg), and in the pharynx (ph) and the esophagus (e) of the foregut and the gastric caeca (c) at the anterior end of the midgut and at the anlage of the second midgut constriction (mg). The dppshv mutant embryo lacks expression in the midgut and the foregut expression is greatly reduced. Panel C contains a map (modified from St. Johnston et al. 1990) of the dpp gene with the breakpoints of the dpp,shv mutants used in these experiments indicated. Note that dppshv mutations map within the transcribed region of the gene, but are 5 ′ to the coding exons which are in the Hin (haploinsufficient) region of the gene.
dpp RNA expression in stage 14 wild-type (A) and dpps21/dppS4 (B) embryos. Lateral views with anterior left and dorsal up photographed at 20 ×. The wild-type embryo expresses dpp in the clypeolabrum (cl), the hindgut (hg), and in the pharynx (ph) and the esophagus (e) of the foregut and the gastric caeca (c) at the anterior end of the midgut and at the anlage of the second midgut constriction (mg). The dppshv mutant embryo lacks expression in the midgut and the foregut expression is greatly reduced. Panel C contains a map (modified from St. Johnston et al. 1990) of the dpp gene with the breakpoints of the dpp,shv mutants used in these experiments indicated. Note that dppshv mutations map within the transcribed region of the gene, but are 5 ′ to the coding exons which are in the Hin (haploinsufficient) region of the gene.
Scr, Ubx and Antp antibodies used are described in Reuter and Scott (1990); lab antibodies are described in Diederich et al. (1989). Antibodies to dpp were generated as described by Panganiban et al. (1990). To affinity purify the resulting polyclonal antibodies, sera were first passed over Affigel-protein A columns (BioRad) to purify the IgG. Sera were diluted with an equal volume of 1.5 M glycine; 3 M NaCl pH 8.9 prior to loading on the columns. The columns were washed with 5 volumes of this buffer prior to elution with 0.1M glycine; 20 HIM NaCl pH 3.2. IgG containing fractions were cut with 50% (final concentration) saturated ammonium sulfate, and centrifuged at 10000g to pellet the precipitate. The IgG was then dialyzed against PBS (10HIM KPO4; 140 mM NaCl pH 7.2) prior to loading on a dpp affinity column, dpp affinity columns were generated by coupling 5 – 10 mg of SDS-PAG purified and electroeluted pG17 inclusion bodies (described in Panganiban et al. 1990) in 0.1M MOPS pH7.5 to Affigel 10 (BioRad) according to manufacturers instructions. IgG fractions containing antibodies to dpp were loaded on the affinity columns at 5 – 10ml h − 1. The columns were washed extensively with 10mM NaPO4 pH 7.0 to reduce the buffer concentration prior to elution with 0.1M glycine pH 2.5. 1ml fractions were collected directly into 100 μ l of 1 M NaPO4 pH 8 to neutralize the glycine (Harlow and Lane, 1988). Antibody concentration was determined by absorbance at 280 nm (A280 of l=0.56mgml − 1). BSA (Sigma) was added to a final concentration of 1mgml − 1 and sodium azide was added to 0.02% prior to storage at 4 or − 20 ° C.
Embryos for immunohistochemistry were dechorionated, fixed and devitellinized using a modification of the methods of Mitchison and Sedat (1983). Briefly, embryos were dechorionated in 50 % bleach for two minutes, rinsed with distilled water and fixed for 20 min in a two phase system composed of 1 part heptane and 1 part 4 % paraformaldehyde in PBS (PF-PBS). The PF-PBS was then replaced with methanol, leaving the embryos at the interface. To devitellinize the embryos, the tubes were shaken gently until the majority of the embryos had settled through the methanol. The heptane and methanol were then removed with a pipette. Embryos were rinsed 3 times with absolute ethanol and treated with 3 % hydrogen peroxide (in ethanol) for 1 min to inactivate endogenous peroxidases. For optimum staining with the dpp antibody, it was necessary to stain the embryos immediately following fixation. All other antibodies gave excellent staining with embryos fixed 3 – 6 months previously and stored at − 20 ° C in ethanol. Antibody incubations and washes were modified from the methods of Carroll and Scott (1985). Embryos were rehydrated through a series of ethanol: PT (PBS plus 0.1% Triton X-100) 50:50; 25:75; 0:100 and blocked at least 3h in PBT (PBS plus 0.2 % BSA plus 0.1 % Triton X-100) with 2 % normal goat serum (Sigma). All subsequent washes and incubations were in PBT unless otherwise specified.
Primary antibodies were incubated with embryos overnight at 4 ° C. The dpp antibody was used at 1 μ gml − 1; lab was used at 1:150; and Scr, Antp and Ubx monoclonals were used at 1:3, 1:5 and 1:5, respectively. At least 10 1ml washes were carried out over 4 – 8 h after each incubation with antibody or with streptavidin-HRP. Secondary antibodies, biotinylated goat anti-rabbit IgG (Vector Laboratories) or HRP-conjugated goat anti-mouse IgG (BioRad), were preadsorbed against embryos and incubated with the embryos for 2h at room temperature. Final dilutions in PBT plus 2 % NGS were 1:600 for the anti-rabbit and 1:400 for the anti-mouse. Washes were as described above. Embryos probed with the monoclonals were then developed in PBT; 0.25 mg ml − 1 diaminobenzidine (Sigma); and 0.03% hydrogen peroxide (Sigma). Embryos probed with the lab and dpp antibodies were incubated for 1 h at room temperature with a 1:300 dilution of streptavidin-HRP (BRL) which had been preadsorbed against embryos. Embryos were then washed and developed as described above. Stained embryos were washed extensively with PBT and PBS, dehydrated through an ethanol series and cleared in methyl salicylate (Sigma). The embryos were then either mounted in salicylate or embedded in Epon 812-Aral-dite prior to observation by differential interference contrast optics on a Zeiss Axiophot. 6 μ m sections were cut from embedded embryos using a Reichert-Jung Ultracut E.
Embryos for RNA in situs were fixed in a formaldehyde and probed with a digoxigenin labelled dpp cDNA probe according to the procedures of Tautz and Pfeiffle (1989). Hybridized RNA was visualized via use of an alkaline phosphatase conjugated antibody to digoxigenin using the Genius kit from Boehringer Mannheim Pharmaceuticals.
Results
dpp is expressed at specific locations along the developing gut
dpp RNA is expressed in six positions along the developing larval gut (Fig. 1A). dpp expression is first detected in the midgut visceral mesoderm at the beginning of germband shortening, stage 12 (Campos-Ortega and Hartenstein, 1985). By the time germband shortening is complete, six specific regions of dpp expression are detected along the gut tube. These include expression in the ectoderm of the foregut and hindgut and in the visceral mesoderm of the midgut. The two spots of staining in the foregut ectoderm are within the anlage of the pharynx and the esophagus. The two principle regions expressing dpp RNA in the midgut are within the anlage of the gastric caeca and the site of the second midgut constriction. Low levels of dpp expression can also be detected at the site of the third midgut constriction although this expression is not visible in Fig. 1. This latter expression is under the control of different cis-regulatory elements than the elements required for the other two parts of dpp expression in the midgut (Jackson and Hoffmann, in preparation) and is not dealt with further in the present report. In addition, dpp is expressed by the ectoderm of the hindgut. The sites of dpp expression in the midgut correspond to regions that undergo morphological changes during development. For example, the caeca evaginate from the anterior end of the midgut and extend anteriorly well into the head, dpp expression in the gastric caeca is no longer detected when the caeca reach full extension at stage 17. The second midgut constriction occurs in the domain expressing dpp in the middle of the midgut. Double labelling experiments using antibodies to dpp and Ubx indicate that the dpp domain overlaps that of Ubx (Reuter et al. 1990). The domain expressing Ubx has been shown to correspond to a metameric unit of the embryo known as parasegment 7 (PS 7) (Tremml and Bienz, 1989a). dpp expression in PS 7 is maintained during the process of forming the gut tube and then declines. To determine whether dpp expression is required for the morphological events of midgut development, we examined gut morphogenesis in dpp mutant embryos.
Mutations in the dpp shortvein region eliminate dpp expression in the gut
Null mutations at dpp lead to severe defects in gastrulation and embryonic dorsoventral polarity making examination of gut morphogenesis in these embryos difficult. Mutations in the shortvein region of dpp, however, allow normal gastrulation and dorsal-ventral pattern formation to occur but lead to larval lethality associated with slow larval growth and visible defects in internal structures, especially the Malphigian tubules (Segal and Gelbart, 1985). We have examined embryos mutant for dppshortvein(dppshv) alleles for expression of dpp (Fig. 1B).
The shv region consists of more than 20 kb of DNA located 5 ′ to the coding region of the dpp gene (Fig. 1C). This region harbors four (of five known) promoters with their respective alternative 5′ noncoding exons (St. Johnston et al. 1990). dppshv mutant embryos were generated by crossing two different alleles to avoid making other undefined lesions on the chromosomes homozygous. A mutant embryo, dppS21/ dppS4, probed with a digoxigenin labelled dpp DNA probe is shown in Fig. IB. dppshv mutant embryos retain the dorsally localized blastoderm expression of dpp and the dorsal and lateral stripes of dpp in the ectoderm, but lack dpp expression in the midgut and have severely reduced foregut expression. Embryos mutant for dppshv alleles which eliminate only promoters P4 and P5 (dppS2 and dpps12) retain some dpp expression along the foregut and at the caeca, however, all dppshv mutant embryos we have examined exhibit midgut defects similar to those described below. Presumably, a minimum level of dpp product is required for wild-type development of the visceral mesoderm. Other aspects of dpp embryonic function are also sensitive to the level of dpp product as indicated by the haploinsufficient characteristics of dpp (Spencer et al. 1982; Irish and Gelbart, 1987).
Absence of dpp affects Scr expression and development of the gastric caeca
Scr has been localized to the visceral mesoderm cells immediately posterior to the budding gastric caeca (Fig. 2A and B) (Tremml and Bienz, 1989a; Reuter and Scott, 1990). dpp is detected anterior to the Scr expressing cells in the visceral mesoderm of the budding caeca. Double labelling experiments have confirmed that dpp and Scr domains are non-overlapping (data not shown). Expression of both dpp and Scr in the anterior midgut is initially detected at stage 13 when the germ band is fully retracted. Scr expression persists through stage 17 while dpp expression in the caeca diminishes in stage 15 and is gone by the end of stage 16. In dppshv mutants, which have reduced dpp expression at the caeca, the domain of Scr expression is expanded anteriorly to encompass those visceral mesoderm cells that in wild-type embryos express dpp (Fig. 2C and D) indicating that dpp expression represses Scr expression in the visceral mesoderm cells that give rise to the gastric caeca. Caeca development is arrested in dppshv mutants: instead of extending anteriorly to form long narrow tubes, the caeca of dppshv mutants remain short and broad (Fig. 2C and D).
Midgut morphogenesis in dppS21/dppS2 embryos stained with anti-Scr antibody. The domain of Scr expression is extended anteriorly in the dppshv mutant embryo (C,D) compared to wild-type (A,B). In addition, the gastric caeca (c) arrest growth and fail to elongate and narrow, the second midgut constriction (2) does not form, and the first midgut constriction (1) is shifted posteriorly.
Midgut morphogenesis in dppS21/dppS2 embryos stained with anti-Scr antibody. The domain of Scr expression is extended anteriorly in the dppshv mutant embryo (C,D) compared to wild-type (A,B). In addition, the gastric caeca (c) arrest growth and fail to elongate and narrow, the second midgut constriction (2) does not form, and the first midgut constriction (1) is shifted posteriorly.
Absence of dpp affects Ubx expression and eliminates the second midgut constriction
In dppshv mutants, the second midgut constriction does not form but the first and third constrictions do form (Fig. 2C,D). Antp is expressed in the visceral mesoderm cells surrounding the first constriction of the midgut (Fig. 3A) (Tremml and Bienz, 1989a; Reuter and Scott, 1990), and Ubx-is expressed in the visceral mesoderm cells of and anterior to the second constriction (Fig. 4A,4C) (Bienz and Tremml, 1988; Tremml and Bienz, 1989a; Reuter and Scott, 1990). dppshv mutant embryos exhibit a range of Ubx expression patterns. Most frequently, the domain of Ubx expression is narrower than in wild-type embryos (Fig. 4D) or the levels of Ubx protein in expressing cells are greatly reduced (Fig. 4B). We have also observed dppshv mutant embryos in which both the level of Ubx and the number of cells expressing it are reduced (data not shown). In addition, the position of the Antp domain of expression and the position of the first constriction are shifted posteriorly in dppshv mutants (Fig. 3B). These effects are identical to the effects that Ubx mutations have on the positions of Antp expression and the first constriction, indicating that the morphological defects caused by dppshv mutations are a consequence of the greatly reduced expression of Ubx in PS 7. However, we also find that dpp expression in PS 7 is reduced or absent in Ubx mutant embryos (Reuter et al. 1990), therefore the lack of either sufficient Ubx or dpp gene product could be responsible for the shift in the Antp expression domain and the position of the first constriction.
Antp in a wild-type (A) versus dpps21/dppS2 (B) embryo. In the wild-type embryo, Antp protein is found in the nuclei of the visceral mesoderm cells surrounding the first constriction (1) (A). In the dppshv mutant embryo (B), the domain of Antp expression and the first constriction are shifted posteriorly to approximately the position where the second constriction forms (2) in the wild-type embryo. In addition, the Antp domain no longer extends on either side of the first constriction but is limited to the anterior portion of the first constriction in the dppshv mutant.
Antp in a wild-type (A) versus dpps21/dppS2 (B) embryo. In the wild-type embryo, Antp protein is found in the nuclei of the visceral mesoderm cells surrounding the first constriction (1) (A). In the dppshv mutant embryo (B), the domain of Antp expression and the first constriction are shifted posteriorly to approximately the position where the second constriction forms (2) in the wild-type embryo. In addition, the Antp domain no longer extends on either side of the first constriction but is limited to the anterior portion of the first constriction in the dppshv mutant.
Ubx expression in wild-type (A and C) versus dpps21/dppS4 (B and D) embryos. The predominant alterations in Ubx expression observed in dppshv mutant embryos are a reduction in the level of Ubx protein in the cells which express it and/or a reduction in the number of cells expressing Ubx. Panels A and B show stage 15 embryos fixed and stained in the same tube. The level of Ubx protein is clearly reduced in the dppshv embryo (B) when compared to that of the wild-type embryo (A). Panels C and D show late stage 16 embryos also fixed and stained in the same tube. There are fewer cells in the dppshv embryo (D) expressing Ubx than in the wild-type control (C). Also note that the Ubx expression in the dppshv embryos is no longer restricted to the anterior part of the constriction which forms in PS 7 (A and C), but extends both anteriorly and posteriorly from the construction (B) or, more often, just posteriorly (D).
Ubx expression in wild-type (A and C) versus dpps21/dppS4 (B and D) embryos. The predominant alterations in Ubx expression observed in dppshv mutant embryos are a reduction in the level of Ubx protein in the cells which express it and/or a reduction in the number of cells expressing Ubx. Panels A and B show stage 15 embryos fixed and stained in the same tube. The level of Ubx protein is clearly reduced in the dppshv embryo (B) when compared to that of the wild-type embryo (A). Panels C and D show late stage 16 embryos also fixed and stained in the same tube. There are fewer cells in the dppshv embryo (D) expressing Ubx than in the wild-type control (C). Also note that the Ubx expression in the dppshv embryos is no longer restricted to the anterior part of the constriction which forms in PS 7 (A and C), but extends both anteriorly and posteriorly from the construction (B) or, more often, just posteriorly (D).
Extracellular dpp protein moves from the visceral mesoderm across the endoderm
The similarity of dpp to the TGF-β family of growth factors (Padgett et al. 1986) and the fact that it is secreted from Drosophila cells in culture (Panganiban et al. 1990) led us to examine whether dpp might move from its sites of synthesis in the embryo. In the PS 7 region of the developing midgut, dpp RNA is detected in the visceral mesoderm cells at and anterior to the developing second constriction at stages 11-17 (Fig. 5A,5B, and data not shown) (stages according to Campos-Ortega and Hartenstein, 1985). Polyclonal antibodies to dpp (Panganiban et al. 1990) detect protein in and around these same visceral mesoderm cells at stage 14 (Fig. 5C). By stage 16, dpp protein is also detected surrounding the underlying endoderm cells (data not shown). By stage 17, most of the dpp protein is concentrated at the luminal surface of the endoderm cells (Fig. 5D), while the RNA can still be detected only in the mesoderm (Fig. 5B). This indicates that dpp protein, secreted from visceral mesoderm cells, moves from its site of synthesis across at least one cell layer. The fact that the site of synthesis is the visceral mesoderm and the adjacent cells are endoderm, raised the intriguing possibility that dpp might be involved in intercellular communication or inductive interactions between the two germ layers.
dpp RNA (A and B) versus protein (C and D) localization at and anterior to the second midgut constriction in 6 micron sagittal sections of stage 14 (A and C) and 17 (B and D) wild-type embryos photographed at 100 ×. While dpp RNA remains localized to the visceral mesoderm cells (vm), dpp protein migrates through the endoderm (ed) cells and by stage 17 is concentrated at the apical surface of the endoderm cells adjacent to the yolk (y).
dpp RNA (A and B) versus protein (C and D) localization at and anterior to the second midgut constriction in 6 micron sagittal sections of stage 14 (A and C) and 17 (B and D) wild-type embryos photographed at 100 ×. While dpp RNA remains localized to the visceral mesoderm cells (vm), dpp protein migrates through the endoderm (ed) cells and by stage 17 is concentrated at the apical surface of the endoderm cells adjacent to the yolk (y).
Absence of dpp in the visceral mesoderm eliminates lab expression in the endoderm
The homeotic gene lab is expressed in the endodermal cells between the first and second constrictions (Diederich et al. 1989; and Fig. 6A), directly beneath the visceral mesoderm cells that express Ubx and dpp. Minor defects in second constriction formation have been reported to occur in lab mutants (Immerglück et al. 1990). In dppshv mutant embryos, lab expression is absent from the endodermal cells of the midgut (Fig. 6B) indicating that the migration of the dpp protein from the mesoderm across the endoderm (Fig. 5) may induce the expression of lab in the underlying endoderm.
lab protein in a wild-type embryo (A) is present in the polyploid endoderm nuclei between the first and second constrictions. No lab protein is detected in the endoderm nuclei of the dppshv mutant embryo (B).
Mutations in abdA expand the domains of expression of Ubx, dpp, and lab
The abdA gene is expressed in the midgut visceral mesoderm in parasegments 8 – 12 (Tremml and Bienz, 1989a; Karch et al. 1990) and inhibits the expression of Ubx in the posterior midgut (Bienz and Tremml, 1988). Although wild-type expression of Ubx (Fig. 7A) and dpp (Fig. 7C) are limited to the PS 7 visceral mesoderm, their expression in an abdA mutant embryo is no longer restricted to PS 7, but extends over the entire posterior midgut (Fig. 7B,D). To determine whether the ectopic expression of dpp in the visceral mesoderm induces ectopic expression of lab in the adjacent endodermal cells, we examined the lab expression in abdA mutant embryos. In contrast to the wild-type pattern of lab expression (Fig. 7E), we find that the domain of lab expression is extended in the abdA mutants to include the endoderm of the posterior midgut directly beneath the visceral mesoderm cells that ectopically express dpp (Fig. 7F).
Ubx, dpp and lab expression in abdA mutant embryos. Ubx (A), dpp (C), and lab (E) are normally expressed in the central region of the midgut. abdA is normally expressed in the visceral mesoderm cells of the posterior midgut (Karch et al.). In abdAM1 mutant embryos, the domains of expression of both Ubx (B) and dpp (D) are expanded to include the visceral mesoderm cells in the posterior midgut (pm) as well. When dpp is ectopically expressed in the visceral mesoderm of the posterior region of the midgut due to a mutation in abdA, the expression of lab is activated in the endoderm cells of the posterior region of the midgut (F). Panels A-D are dorsal views photographed at 40 × E and F are lateral views photographed at 63 –.
Ubx, dpp and lab expression in abdA mutant embryos. Ubx (A), dpp (C), and lab (E) are normally expressed in the central region of the midgut. abdA is normally expressed in the visceral mesoderm cells of the posterior midgut (Karch et al.). In abdAM1 mutant embryos, the domains of expression of both Ubx (B) and dpp (D) are expanded to include the visceral mesoderm cells in the posterior midgut (pm) as well. When dpp is ectopically expressed in the visceral mesoderm of the posterior region of the midgut due to a mutation in abdA, the expression of lab is activated in the endoderm cells of the posterior region of the midgut (F). Panels A-D are dorsal views photographed at 40 × E and F are lateral views photographed at 63 –.
Discussion
Similarities in the expression patterns of dpp and Ubx in the visceral mesoderm led us to examine the regulatory relationships between dpp and homeotic genes during midgut development. We have identified three responses to dpp in the development of the Drosophila midgut that involve homeotic gene expression. The significance of these regulatory relationships to morphogenesis is best indicated by the fact that mutations in dpp alter the same structures, the gastric caeca and the second constriction, affected by mutations in two of the homeotic genes regulated by dpp. That dpp functions as a secreted factor is best illustrated by its regulation of a third homeotic gene across germlayers.
dpp regulation of Scr expression and gastric caeca formation
dpp negatively regulates Scr expression in the visceral mesoderm cells that become the caeca (Fig. 8). In dppshv mutants, the domain of Scr expression at the gastric caeca is extended anteriorly to include those cells that normally express dpp. The ectopic expression of Scr in these cells may prevent the cells from evaginating as caeca. Alternatively, ectopic Scr expression by itself may not be sufficient to block development of the caeca but may serve as an indicator that the cell fate at the anterior end of the midgut is changed in the absence of dpp. The cell fate indicated by Scr expression must then be incompatible with the processes needed to form caeca. The mechanism by which dpp normally represses Scr expression in the anterior midgut and permits development of the caeca remains to be determined.
The interactions between dpp and homeotic genes during midgut morphogenesis. Arrows indicate positive regulation and inverse tees indicate negative regulation, dpp expression in the gastric caeca precludes Scr expression there, since the Scr domain is extended anteriorly in embryos lacking dpp expression in the caeca, dpp expression is unaffected by a lack of Scr, however, both dpp and Scr are required for appropriate development of the caeca. In the absence of either dpp or Scr, the caeca do not form properly, but the defects are distinct for each mutant (See text and Reuter and Scott, 1990 for details). Ubx and dpp are expressed in the visceral mesoderm cells between the first and second midgut constrictions. Both are required for the formation of the second constriction and stimulate expression of one another. Between the first and second constrictions, dpp migrates from the visceral mesoderm cells which synthesize it through the underlying endoderm cells, stimulating expression of the lab gene in those cells. The abdA gene is expressed by visceral mesoderm cells of the posterior region of the midgut. abdA represses the expression of Ubx and dpp in the cells where it is expressed, as in the absence of abdA function the Ubx and dpp domains are extended posteriorly, wg expression is activated in the visceral mesoderm posterior to the second constriction by abdA. Regulation of wg expression requires other factors as wg is not expressed throughout the abdA domain, dpp expression is activated in a third domain in the mesoderm of the midgut, at the third constricition, by a mechanism that does not require Ubx and that is not down-regulated by abdA.
The interactions between dpp and homeotic genes during midgut morphogenesis. Arrows indicate positive regulation and inverse tees indicate negative regulation, dpp expression in the gastric caeca precludes Scr expression there, since the Scr domain is extended anteriorly in embryos lacking dpp expression in the caeca, dpp expression is unaffected by a lack of Scr, however, both dpp and Scr are required for appropriate development of the caeca. In the absence of either dpp or Scr, the caeca do not form properly, but the defects are distinct for each mutant (See text and Reuter and Scott, 1990 for details). Ubx and dpp are expressed in the visceral mesoderm cells between the first and second midgut constrictions. Both are required for the formation of the second constriction and stimulate expression of one another. Between the first and second constrictions, dpp migrates from the visceral mesoderm cells which synthesize it through the underlying endoderm cells, stimulating expression of the lab gene in those cells. The abdA gene is expressed by visceral mesoderm cells of the posterior region of the midgut. abdA represses the expression of Ubx and dpp in the cells where it is expressed, as in the absence of abdA function the Ubx and dpp domains are extended posteriorly, wg expression is activated in the visceral mesoderm posterior to the second constriction by abdA. Regulation of wg expression requires other factors as wg is not expressed throughout the abdA domain, dpp expression is activated in a third domain in the mesoderm of the midgut, at the third constricition, by a mechanism that does not require Ubx and that is not down-regulated by abdA.
Although ectopic expression of Scr blocks the development of caeca, normal expression of Scr immediately posterior to the cells forming the caeca is required for caeca development (Reuter and Scott, 1990). As dpp expression at the anterior end of the midgut is not affected in the Scr mutant embryos, dpp expression, while required, is not sufficient for caeca development. The two adjacent regions of visceral mesoderm cells at the anterior end of the midgut, one expressing dpp and the other expressing Scr, must cooperate to direct the formation of the gastric caeca.
dpp and Ubx regulate one another
Wild-type expression of both dpp and Ubx is crucial for development of the midgut (Fig. 8) because in the absence of either Ubx or dpp the second constriction does not form. These two genes positively regulate the expression of one another: in the absence of dpp function in the gut, Ubx expression is reduced, but not absent, and in the absence of Ubx function, dpp expression is reduced or absent depending on the Ubx allele (See Reuter et al. 1990 for discussion of the regulation of dpp expression by Ubx). This is in contrast to what has been reported by Immerglück et al. (1990) who report a unidirectional effect of Ubx on dpp. By analogy to the relationship between dpp and zen (Rushlow et al. 1987; Rushlow and Levine, 1990), we predict that Ubx is required for the initiation of dpp expression in PS 7, and that dpp is in turn required to maintain expression of Ubx. This prediction is consistent with the range of Ubx expression phenotypes we observe in dppshv mutant embryos.
The requirement for the coordinated activities of Ubx and dpp is illustrated by the posterior extension of Antp expression and the posterior shift of the first constriction in embryos mutant for either dpp or Ubx. As expression of both dpp and Ubx is regulated negatively by abdA (Fig. 7) it is of interest to determine whether abdA acts on both genes independently or whether, for example, the effects of abdA on Ubx expression are mediated by abdA’s negative regulation of dpp expression. When both abdA and Ubx are expressed in the same cells, abdA can still repress dpp expression (Reuter et al. 1990) but it is not yet known whether abdA can also repress Ubx in the presence of ectopic dpp expression.
dpp migration across germ layers affects lab expression
One of the most interesting results of these experiments is that the dpp protein, after migrating from the visceral mesoderm across the endoderm, may induce the expression of the lab gene in the endoderm cells. In support of this mechanism, ectopic expression of dpp in the visceral mesoderm cells in abdA mutant embryos gives rise to a parallel ectopic expression of lab in the endoderm of the posterior midgut. Because Ubx and dpp are colocalized in the part of the visceral mesoderm of wild-type embryos that overlies the tab-expressing endoderm, and because Ubx is also ectopically expressed in the posterior midgut of abdA mutants, it remains a formal possibility that Ubx acts via a dpp- independent pathway to stimulate lab expression in the adjacent tissue. We believe this is unlikely because in dppshv mutants, which retain some Ubx expression, lab is not expressed in the underlying endoderm.
The fact that dpp is a secreted protein makes it a strong candidate for regulation of gene expression between germlayers. However, the movement of dpp is quite limited. We do not observe extracellular dpp protein moving into regions of the visceral mesoderm that do not make dpp mRNA and cannot detect it migrating into or across the endoderm at the anterior end of the midgut. Perhaps dpp is associated with other extracellular proteins which limit its movement. We have observed that dpp protein produced in a Drosophila cell culture expression system associates with other proteins on the surface of the tissue culture dish (Panganiban et al. 1990). The Drosophila wingless gene product also appears to exhibit a limited potential for movement from the cells in which it is made (van den Heuvel et al. 1989). However, even the limited movement of extracellular dpp may allow it to exert an influence in a specific domain, for example, PS 7 of the visceral mesoderm, such that the intercellular communication mediated by dpp stabilizes a specific cell fate decision in that domain. This mechanism allows a population of cells to cooperate in maintaining the cell fate of the population.
dpp exerts pleiotropic effects on homeotic gene expression and midgut morphogenesis
The developmental origin or the environment of cells must determine how cells respond to dpp because, as reported here, at least three different homeotic genes can respond to the presence of dpp in the midgut. The only previous report of a change in gene expression due to dpp dealt with the failure to maintain zen expression in the amnioserosa of dpp null animals (Rushlow and Levine 1990); interestingly, zen also encodes a homeodomain containing protein (Rushlow et al. 1987). The differences in homeotic gene expression between the dpp-expressing regions of the midgut are paralleled by morphogenetic differences in these regions of the gut. In one dpp domain, the caeca are formed, while in the other a constriction is formed. Although dpp and Ubx are necessary to form the second constriction, they are not sufficient for constriction formation, as is illustrated by the absence of the second and third constrictions in abdA mutant embryos in which Ubx and dpp are expressed throughout the posterior midgut. Therefore, the effects of dpp on both gene expression and morphogenetic events are specific in different positions in the embryo. We propose that expression of dpp is necessary, but not sufficient, for specific aspects of homeotic gene expression and morphogenesis required to form the Drosophila midgut. The mechanisms by which the domains of dpp and homeotic gene expression are initially established in the visceral mesoderm remain to be determined. Clearly, once expressed, the genes regulate one another.
ACKNOWLEDGEMENTS
We thank Tom Kaufman for providing antibody to labial and Sean Carroll for critically reading the manuscript. Support for this research was provided by grants from the American Cancer Society to F.M.H., Cancer Center Core Support from NCI to H.C. Pitot, NIH Training Grant GM07215 for G.E.F.P., NIH Grant 18163 to M.P.S. M.P.S. is an Investigator of the Howard Hughes Medical Institute. This paper is dedicated to the memory of Calvin Calmon.