Genetic Analysis of the Bone Morphogenetic Protein-Related Gene, gbb, Identifies Multiple Requirements During Drosophila Development

Corresponding author: Kristi A. Wharton, Division of Biology and Medicine–MCB, Brown University, Box G-J160, Providence, RI 02912., kristi_wharton{at}brown.edu (E-mail)

Communicating editor: V. G. FINNERTY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have isolated mutations in the Drosophila melanogaster gene glass bottom boat (gbb), which encodes a TGF-ß signaling molecule (formerly referred to as 60A) with highest sequence similarity to members of the bone morphogenetic protein (BMP) subgroup including vertebrate BMPs 5–8. Genetic analysis of both null and hypomorphic gbb alleles indicates that the gene is required in many developmental processes, including embryonic midgut morphogenesis, patterning of the larval cuticle, fat body morphology, and development and patterning of the imaginal discs. In the embryonic midgut, we show that gbb is required for the formation of the anterior constriction and for maintenance of the homeotic gene Antennapedia in the visceral mesoderm. In addition, we show a requirement for gbb in the anterior and posterior cells of the underlying endoderm and in the formation and extension of the gastric caecae. gbb is required in all the imaginal discs for proper disc growth and for specification of veins in the wing and of macrochaete in the notum. Significantly, some of these tissues have been shown to also require the Drosophila BMP2/4 homolog decapentaplegic (dpp), while others do not. These results indicate that signaling by both gbb and dpp may contribute to the development of some tissues, while in others, gbb may signal independently of dpp.

THE TGF-ß superfamily of secreted signaling molecules consists of three evolutionarily related groups, the TGF-ßs, bone morphogenetic proteins (BMPs), and activins, which have been shown to regulate many developmental events, from cell proliferation to cell fate specification and apoptosis (for review see KINGSLEY 1994 ; MASSAGUE et al. 1994 ; HOGAN 1996 ). These structurally similar molecules are synthesized as proproteins, and they are processed to release the mature ligand. The ligand binds to a heterotetrameric receptor complex made up of type I and type II integral membrane proteins, each with a cytoplasmic serine/threonine kinase domain (reviewed in TEN DIJKE et al. 1996 ; MASSAGUE 1998 ). The signal is transduced to the nucleus through the action of a complex of proteins belonging to the Smad family of transcriptional regulators (reviewed in MASSAGUE et al. 1997 ; CHO and BLITZ 1998 ; PADGETT et al. 1998 ).

TGF-ß ligands can exist as hetero- or homodimers, but the extent to which heterodimers vs. homodimers form and function in vivo is poorly understood. In the few cases analyzed, heterodimers have been shown to have a qualitatively different function or activity. In one case, the activin homodimer and the inhibin heterodimer have been shown to be mutually antagonistic (reviewed in SPORN and ROBERTS 1990 ), and, in hepatoma cells, this antagonism appears to occur at the level of receptor binding (XU et al. 1995 ). In other cases, heterodimers behave cooperatively, as has been shown with Xenopus BMP4 and BMP7, where the BMP4/7 heterodimer has a higher activity than either homodimer (AONO et al. 1995 ; SUZUKI et al. 1997 ). The qualitatively different responses elicited by heterodimers vs. homodimers most certainly contribute, at least partially, to the diveristy of biological processes attributed to the TGF-ß/BMP family members (MASSAGUE et al. 1994 ; SIMIN et al. 1998 ).

In recent years, many different TGF-ß superfamily members have been identified and their expression patterns defined. While genetic approaches have begun to address the function of these ligands as well as identify and establish relationships between potential components of their signaling pathways, our understanding of the role of these molecules in development and in the progression of disease is still incomplete. Studies of TGF-ß superfamily members in vertebrates have revealed that, in many tissues, more than one TGF-ß- or BMP-type ligand are often expressed, and that while their overall patterns of expression are usually distinct, in some cells they often overlap whereby two ligands may be coexpressed (e.g., LYONS et al. 1995 ; DUDLEY and ROBERTSON 1997 ). These types of data suggest that the development of some tissues may be influenced by the action of multiple ligands. Currently, we understand very little about the potential combinatorial action of different ligands. Does each ligand have a unique or similar function? If the functions are the same, do multiple ligands work together to amplify the signal? Given that different ligands are often expressed in overlapping patterns during development, it is likely that a single cell may be exposed to multiple ligands simultaneously and must respond accordingly. To fully appreciate the impact of signaling by multiple ligands, it is essential that we have a complete understanding of the contribution of individual ligands and the interplay between their different signaling pathways. As a first step toward understanding the importance of signaling by multiple related ligands, we identified a new Drosophila BMP, the 60A gene (WHARTON et al. 1991 ), to characterize its function and examine potential interactions between 60A signaling and that of other Drosophila BMPs. Here we report the isolation of both null and hypomorphic alleles of 60A and describe the mutant phenotypes associated with these mutations.

To date, three TGF-ß superfamily members have been identified in Drosophila, decapentaplegic (dpp), 60A, and screw (scw; PADGETT et al. 1987 ; WHARTON et al. 1991 ; DOCTOR et al. 1992 ; ARORA et al. 1994 ), all of which belong to the BMP group. The dpp gene is the best characterized of the three, and it has been shown to function in a number of different developmental processes throughout the life cycle of Drosophila (SPENCER et al. 1982 ; SEGAL and GELBART 1985 ; GELBART 1989 ; TWOMBLY et al. 1996 ). The requirement for scw appears to be limited to embryogenesis, where it acts in combination with dpp to specify dorsal cell fates (ARORA et al. 1994 ). The 60A gene has been renamed glass bottom boat (gbb) to reflect the null phenotype (see also KHALSA et al. 1998 ). Our phenotypic analysis of gbb mutants indicates that gbb, like dpp, is required throughout development for a number of diverse developmental processes. Many of the gbb mutant phenotypes resemble those displayed by dpp alleles, suggesting that for at least some developmental processes, both gbb and dpp signaling are required. Consistent with this common requirement in the development of certain tissues or structures, we have recently shown that gbb and dpp signal together to pattern the wing (KHALSA et al. 1998 ) and that these signals are mediated by tkv and sax type I receptors (HAERRY et al. 1998 ; KHALSA et al. 1998 ). Other gbb mutant phenotypes have not been reported for alleles of dpp, suggesting that in these cells or tissues, gbb may act independently of dpp or that gbb in some way elicits a qualitatively different response. Future studies detailing gbb signaling in specific developmental processes will reveal the nature of its relationship to signaling by other members of the TGF-ß superfamily and delineate the contribution made by each ligand to that process as a whole.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fly strains and culture conditions:
All mutations and chromosomes are described in LINDSLEY and ZIMM 1992 or FLYBASE 1996 , except where noted. Enhancer trap line P-1 (SUN et al. 1995 ) was kindly provided by H. Sun. P[ry⁺; lacZ]0331 was obtained from S. Wasserman. Flies were reared on standard Drosophila cornmeal/sucrose/yeast medium at 25° unless otherwise noted.

F₂ lethal screen and complementation analysis:
Males isogenic for the second chromosome bearing the markers dp cn and bw were mutagenized with ethyl methanesulfonate (Sigma, St. Louis) as described in LEWIS and BACHER 1968 . Mutagenized males were crossed to dpp^d-ho Bc Elp/CyO females. Male progeny of the genotypes dp cn bw/dpp^d-ho Bc Elp or dp cn bw/CyO were mated individually to Df(2R)bw^S46/SM6a females. A total of 7000 fertile crosses were scored for the presence of Cy⁺ progeny. Any line identified as lethal in trans to Df(2R)bw^S46 was retested against Df(2R)G10-7-5 and Df(2R)HB-132. In addition to the 100 lethals recovered from our screen, 66 lethals were isolated in an independent screen (REED 1992 ) and mapped to the interval between the proximal breakpoint of Df(2R)G10-7-5 and the distal breakpoint of Df(2R)bw^S46.

Genomic walk and deficiency breakpoint mapping:
Genomic clones corresponding to the 60A chromosomal region were isolated from a DASH II (Stratagene, La Jolla, CA) genomic library constructed from a strain isogenic for dp cl cn bw (from R. W. Padgett). A genomic walk was constructed, and phage subclones were used to identify deficiency breakpoints. Genomic DNA was isolated from various mutant strains by grinding 50–100 adult flies in 0.1 M Tris-HCl, pH 9.0, 0.1 M EDTA, 1% SDS, and 1% DEPC. After a 30-min incubation at 70°, the samples were made 1 M potassium acetate and incubated on ice for 30 min. After centrifuging for 15 min at 4°, the DNA was precipitated from the supernatant by adding 0.5 volumes of isopropanol at room temperature. The DNA isolated from deficiency strains was analyzed by Southern analysis with specific subclones from the genomic walk as probes. Restriction-digested genomic DNA isolated from wild-type flies (Oregon-R) and flies heterozygous for Df(2R)HB132, Df(2R)b23, and Df(2R)egl² were probed with the phage from our genomic walk.

Quantitative Southern analysis:
To determine which deficiencies deleted the gbb gene, we quantitated the signal produced by a gbb coding region probe on total genomic blots of DNA from wild type (+/+) and flies heterozygous for a deficiency (Df/+). Genomic DNA was isolated as described above from stocks heterozygous for Df(2R)bw^S46, Df(2R) HB132, Df(2R)egl², Df(2R)b23, Df(2R)bw^DRa, Df(2R)106, Df(2R)G10-CD14, Df(2R)G10-7-5, and Oregon-R (OR). The DNA was digested with EcoRI, and the amount of DNA in each lane was quantitated by hybridizing each blot with a control probe. The blot was also hybridized with a 9.5-kb EcoRI fragment derived from the gbb gene. The intensity of each band was measured using a densitometer (LKB, Piscataway, NJ). The data for each deficiency line (Df/+), normalized for the amount of DNA loaded in each lane, were then compared to those of the wild type (+/+).

Constructs, P-element transformation, and rescue analysis:
A 6.8-kb genomic SalI fragment from T3-6a was subcloned into pCasper 2 (PIROTTA 1988 ). This P[w⁺; 60A S6.8, gbb⁺] construct was transformed into w¹¹¹⁸, and eight independent insertions were isolated. Lines Tn6.6 and Tn55.2 on the X chromosome and Tn1.2, Tn6.3, and Tn55.4 on the third chromosome were used in this study. The gbb gene does not contain intronic sequences (WHARTON et al. 1991 ); therefore, this rescue construct contains 3.7 kb of genomic DNA 5' to the start of gbb transcription and 1.5 kb 3' of the polyadenylation site. Representative alleles of five different complementation groups (J, F, K, L, or M) were tested for rescue of lethality in trans to Df(2R)b23 by crossing each stock (lethal/SM6a) to Df(2R)b23 If /+; P[w⁺; 60A S6.8, gbb⁺]/+ and scoring for the presence of If Cy⁺ flies.

The gbb knockout construct was made by inserting a linker into the gbb coding region, generating a stop codon at residue 38 (WHARTON et al. 1991 ). Two 17-mer oligonucleotides, 623: CTAGTCTAGACTAGTTG and 622: CTAGTCTAGACTAGCAA, were annealed and ligated into the SfiI site (nucleotide 510) of the 6.8-kb SalI genomic subclone (see Figure 2). Clones were identified by the loss of the SfiI site and the presence of the XbaI site introduced by the linker (in boldface above). The insertion of a stop codon in the correct translational frame was verified by sequencing. The gbb knockout (KO) fragment was cloned into pCasper 2, and a single transformant, P[w⁺; 60A S6.8, gbb^KO], which was inserted on the second chromosome, was obtained. Attempts to transpose this insertion to another chromosome were unsuccessful, so it was recombined onto a Df(2R)b23 chromosome marked with the dominant marker Irregular facets (If). Males of the genotype w/Y; P[w⁺; 60A S6.8, gbb^KO] Df(2R)b23 If/CyO were crossed to alleles of the F, J, and M complementation groups (*/SM6a). The ability of P[w⁺; 60A S6.8, gbb^KO] to rescue the lethality associated with these alleles was determined by scoring the progeny for the presence of Cy⁺ If flies. The percentage rescue was calculated as the number of Cy⁺ If/half the number of Cy x 100. Pupal lethality was scored by the presence of uneclosed, desiccated, or black pupae 5–6 days after the normal time of eclosion.

View larger version (40K): In this window In a new window Download PPT slide   Figure 1. Genetic map of the 60A chromosomal region. The extents of deficiencies used to define intervals in the 60A region are indicated by thick black lines. Eight chromosomal intervals (1–8) are defined by the distal breakpoints of the overlapping deficiencies. Complementation groups (A–O) identified in the screen are indicated below the interval to which they map. The polytene chromosome hybridization site for cDNA16, the original cDNA clone described in WHARTON et al. 1991 , is shown at the top of the figure.

View larger version (12K): In this window In a new window Download PPT slide   Figure 2. Molecular map of the 60A chromosomal region. At the top of the figure, black lines denote the regions deleted by Df(2R)b23 and Df(2R)egl2, and the vertically hatched bars indicate regions of uncertainty around the distal breakpoints. Molecular coordinates are indicated in 10-kb intervals above a restriction map of the gbb region. Lines extending above the restriction map correspond to EcoRI sites, those below to SalI sites. The position of the gbb transcription unit is denoted by a thick arrow below the restriction map. The 6.8-kb SalI genomic fragment used for constructs to generate P[w+; 60A S6.8, gbb+] and P[w+; 60A S6.8, gbbKO] is denoted by the stippled gray bar below the restriction map. Probes A and B used in the transcriptional analysis are indicated. The positions of phage and cosmid clones along the molecular map are shown at the bottom of the figure.

Four-cutter analysis and sequencing:
Genomic DNA isolated from stocks heterozygous for alleles of the F, M, and J complementation groups was digested with a number of four- and five-cutter restriction enzymes, and Southern blots of digested DNA were probed with a genomic fragment containing the gbb gene. To verify the P6-103 restriction-site polymorphism, P6-103/+ genomic DNA containing the aberrant restriction site was amplified by PCR from a single adult and sequenced. The lesions associated with the other three alleles of the J complementation group (ac-17, Ab-4, and An-4) were also determined by sequence analysis. As expected, due to the isogenic nature of the mutagenized chromosomes, no other changes or polymorphisms were detected.

Lethality studies:
More than 1000 embryos from the crosses gbb^*/SM6a x Df(2R)b23/+ and gbb^*/SM6a x gbb^*/+ (^* = allele 1, 2, 3 or 4) were collected on apple juice plates. The number of individuals that hatched, pupated, and eclosed was compared to the number of individuals that survived the previous developmental stage and served as a measure of embryonic, larval, or pupal lethality, respectively.

Germ-line clones:
Germ-line clones were produced in females as described in CHOU et al. 1993 . Cy⁺ females produced in the cross w P[ry⁺; FLP]/Y; P[w⁺; FRT]G13 P[w⁺; ovo^D1]32X9 x +/+; P[w⁺; FRT]G13 L bw gbb¹/CyO, S cn bw were tested for their ability to lay viable eggs. A total of 10 different FRTG13 gbb¹ recombinant chromosomes were tested in this way. Fertile females were recovered from all 10 lines and, in all cases, each produced >90% viable eggs.

Immunohistochemistry and microscopy:
For antibody incubations, dechorionated embryos were fixed in 1:1 heptane:4% formaldehyde (in 100 mM Pipes, 2 mM MgSO₄, 1 mM EGTA) for 17 min. Fixed embryos were devitellinized in 1:1 heptane:methanol, washed in methanol + 0.3% hydrogen peroxide for 2 min, 1:1 methanol:PBT (PBS + 0.1% Triton-X) and PBT + 0.2% BSA (PBTB) for 3 hr. Embryos were blocked in PBTB + 5% normal goat serum for 30 min and incubated with preadsorbed primary antibody overnight at 4°. Horseradish peroxidase (HRP)-conjugated secondary antibodies (Jackson Immunological Research) were used at 1:500 for 2–3 hr at room temperature. HRP was detected with 0.5 mg/ml diaminobenzidine and 0.03% hydrogen peroxide. Stained embryos were mounted in 2 parts Permount (Fisher, Pittsburgh):1 part methyl salicylate (Sigma, St. Louis). Rabbit anti-ß-galactosidase (Cappel) was used at 1:1000. Rabbit anti-Antp (T. Kaufman) was used at 1:100 after X-Gal staining of gbb¹/CyO, ftz lacZ embryos. Wings were mounted as described previously (KHALSA 1998). Images were obtained with a photomicroscope (FXA; Nikon, Garden City, NY). For scanning electron microscopy, adult flies were dehydrated through an ethanol series and immersed in hexamethyldisilazane (Electron Microscopy Sciences, Fort Washington, PA) that was then allowed to sublime at room temperature. Flies were sputter coated and examined on a Hitachi 2700 SEM microscope.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

F₂ lethal screen in 60A chromosomal region:
To identify mutations in the TGF-ß/BMP family member gbb, we carried out an F₂ lethal screen. In situ hybridizations to polytene chromosomes indicated that Df(2R)bw^S46 deletes the gbb gene while Df(2R)HB132 does not (data not shown). Thus, Df(2R)bw^S46/SM6a was used as a tester strain for our screen, as described in MATERIALS AND METHODS. A total of 175 chromosomes that failed to complement Df(2R)bw^S46 were recovered; 100 of these also failed to complement Df(2R)G10-7-5, but complemented Df(2R)HB132. Thus, the lethal in each of the 175 lines mapped to the 59F8;60A7 chromosomal interval. Of the 66 lethal lines obtained from B. Reed and M. Ashburner, 57 were assigned to the 59F8;60A7 interval on the basis of their ability to complement Df(2R)-HB132 and failure to complement Df(2R)G10-7-5. In total, 157 mutagenized chromosomes were isolated that could potentially contain a mutation in the gbb gene.

Deficiency mapping and complementation analysis:
Each mutagenized chromosome was tested in trans to six different deficiencies [Df(2R)G10-CD14, Df(2R)106, Df(2R)bw^DRa, Df(2R)egl², Df(2R)b23, and Df(2R)bw^DRj] that break within the 59F8;60A7 interval (Table 1) and form a series of nested deletion breakpoints defining eight chromosomal intervals (Figure 1). Each lethal was thus assigned to a specific interval (Figure 1). Inter se crosses were performed with approximately one-third of the total 175 lines, and out of these, at least 15 complementation groups within the 59F6-8;60A7 interval were established (Table 2).

A number of deficiencies were tested for the presence of the gbb locus by quantitative Southern analysis. This analysis corroborated our polytene in situ hybridization data that Df(2R)bw^S46 deletes the gbb gene while Df(2R)HB-132 does not. In addition, it demonstrated that Df(2R)-G10-CD14, Df(2R)106, Df(2R)bw^DRa, Df(2R)egl², and Df(2R)-b23 also delete the gbb gene, thus placing gbb within the region defined by the distal breaks of Df(2R)b23 and Df(2R)HB132 (Figure 1, intervals 3 and 4). We mapped the distal breakpoints of Df(2R)b23 and Df(2R)egl² on our genomic walk, further verifying our results from the quantitative Southern analysis (Figure 2). A total of 21 different lines have lethal mutations that map within the interval defined by the distal break of Df(2R)b23 and Df(2R)HB132 and that constitute five separate complementation groups (K, L, F, M, or J; Table 2). Subsequently, we have shown that alleles of complementation groups K and L fail to complement Df(2R)OV1 and thus map to interval 3, while groups F, M, and J map to interval 4.

Genomic rescue and functional identification of gbb alleles:
A 6.8-kb genomic SalI fragment containing the 1.67-kb gbb transcription unit was used to make a gbb rescue construct (Figure 2). Five transgenic lines with insertions of the 6.8-kb genomic fragment were tested for their ability to rescue the lethality associated with alleles of complementation groups F, J, K, L, and M in trans to Df(2R)b23 (Table 3). All five transgenic lines rescued alleles from complementation groups F, M, and J. We did not observe consistent rescue of the lethality associated with alleles in the F complementation group. However, as some F alleles that are not rescued with a single copy of P[w⁺; 60A S6.8, gbb⁺] are rescued with two copies, we presume that this variability is caused by position effect. A transcriptional analysis of the genomic region surrounding gbb provided an explanation for the rescue of all three complementation groups, F, M, and J. At least four distinct RNAs (2.4, 1.55, 1.4, and 1.2 kb) are derived from sequences immediately proximal to gbb (Figure 2, fragment A), while a 6-kb RNA is derived from sequences immediately distal (Figure 2, fragment B; data not shown). Thus, gbb is one of at most six genes that map to this 6.8-kb genomic fragment should each transcript represent one gene.

To determine which of the three complementation groups corresponds to the gbb gene, we generated a gbb knockout construct that is derived from P[w⁺; 60A S6.8, gbb⁺]. A translational stop was introduced into the gbb gene in the context of the 6.8-kb genomic fragment, thus creating a "gbb knockout construct" (P[w⁺; 60A S6.8, gbb^KO]). A linker that introduces stop codons in all three reading frames was inserted into the SfiI site at nucleotide position 508, resulting in termination of translation 38 residues after the first AUG. A transgenic line containing P[w⁺; 60A S6.8, gbb^KO] was generated and tested for its ability to rescue alleles from complementation groups F, M, and J in trans to Df(2R)b23 (Table 4). The gbb knockout construct rescued M and F alleles but not J alleles, thus providing functional proof that the J complementation group corresponds to the gbb gene.

Identification of molecular lesions associated with gbb alleles:
To verify that the J complementation group corresponds to the gbb gene, genomic DNA isolated from each J line was tested for restriction fragment length polymorphisms (RFLPs). Genomic DNA was digested with four- and five-cutter restriction enzymes, and DNA from the P6-103 line revealed an RFLP when digested with HinI or TfiI. To verify this polymorphism, the region was PCR amplified from P6-103 genomic DNA and sequenced. In the mutant DNA, the sequence at nucleotide 1518 is altered from GGATC to GAATC, creating a novel HinfI and TfiI site at this position (Table 5). This change introduces a nonsense codon at aa 371 that results in a truncated gbb protein that lacks the majority of the ligand domain. The molecular lesions associated with the other three J alleles were also determined by sequence analysis, and all are point mutations in the gbb coding region. ac17 is a nonsense mutation at the start of the ligand domain, An-4 is a methionine-to-isoleucine change at the putative translational start of the gbb protein, and the Ab-4 mutation changes a conserved alanine in the ligand domain into a valine. We refer to lines P6-103, ac17, An-4, and Ab-4 as gbb¹, gbb², gbb³, and gbb⁴, respectively (see Table 5).

gbb is a zygotic larval lethal:
Lethal phase studies of gbb/Df or gbb trans-heterozygotes with gbb alleles gbb¹, gbb², and gbb³ indicate that the lethality occurs primarily during early larval stages. Less than 10% of the individuals die as embryos. To determine if the larval lethality results from rescue of an earlier embryonic requirement by a maternal contribution, we used the FLP/FRT system (CHOU et al. 1993 ) to generate germline clones of amorphic gbb alleles. Females with homozygous gbb¹ germline clones produced phenotypically wild-type eggs, and the embryos hatched and survived to adulthood. These results indicate that gbb is not required maternally if zygotic function is supplied.

gbb mutant larvae show a number of defects in morphology and cuticle patterning. They are lethargic and appear flaccid when compared to their gbb/+ siblings and have dramatically reduced imaginal discs. The morphology of the fat body is abnormal, and this defect is most likely responsible for the transparency of the larvae for which the gene is named glass bottom boat (gbb; KHALSA et al. 1998 ). The cuticle also exhibits a number of defects in the telson region. In the most severe cases, the posterior spiracles do not protrude from the larval body, and the stigmatophores are partially fused and more dorsally situated.

Severity of mutant alleles:
gbb¹ and gbb² are lethal when homozygous or in trans to Df(2R)b23, and they behave as genetic nulls on the basis of the fact that the mutant phenotype of gbb/Df or gbb/gbb is indistinguishable. gbb³ and gbb⁴ appear to be hypomorphs. gbb⁴ is a very weak hypomorphic allele and was isolated in our F₂ lethal screen because of a secondary mutation on the chromosome that enhanced the phenotype of the gbb⁴ mutation. When this secondary mutation was removed by recombination, homozygous gbb⁴ individuals survived to adulthood (Table 6; see also KHALSA et al. 1998 ). Crosses between gbb⁴ and other gbb alleles or Df(2R)b23 revealed that the number of progeny that eclose is variable, depending on the culture conditions (Table 6). In addition, we determined that the phenotype of gbb⁴ is temperature sensitive, and at 18°, 100% of gbb⁴ homozygotes are recovered as viable adults (Table 6). gbb³, while completely lethal, is nevertheless a weaker mutation than gbb¹ or gbb², as we see a greater percentage of gbb³/gbb⁴ survivors when compared to the number of gbb¹/gbb⁴ or gbb²/gbb⁴ survivors.

gbb mutant embryos exhibit defects in midgut morphogenesis:
While the gbb homozygous mutant embryos survive and hatch into larvae, we have determined that gbb is required during embryogenesis for the proper formation of the midgut. During wild-type midgut development, three morphological constrictions are formed during stages 15 and 16 of embryogenesis (SKAER 1993 ; Figure 3A). In gbb null mutant embryos, the anterior constriction fails to form, giving rise to a bulbous anterior midgut (Figure 3C, Figure E, and Figure G). This mutant phenotype is interesting in light of previous studies that have identified a requirement for dpp in promoting the central midgut constriction (PANGANIBAN et al. 1990 ; HURSH et al. 1993 ). While dpp mutants fail to form the central constriction, gbb mutant embryos appear to initiate the anterior constriction as a slight invagination, but the constriction fails to be maintained and is never completed. In gbb mutants the central and posterior constrictions form completely, but the position of each is somewhat abnormal. The central constriction is often shifted from its position between parasegments (PS) 7 and 8 to a more posterior position between PS 8 and 9 (Figure 3C and Figure G). At this stage in our analysis, it is impossible to determine whether the shift in position of the central and posterior constrictions reflects a change in the regional specification of the midgut, or if it is a secondary consequence of the absence of the anterior constriction resulting from physical constraints placed on the cells of the developing midgut.

View larger version (123K): In this window In a new window Download PPT slide   Figure 3. gbb mutant embryos are defective in midgut morphogenesis. (A) A stage 16 wild-type embryo stained for expression of the enhancer trap line P-1. The three midgut constrictions, anterior (a), central (c), and posterior (p) are shown, as well as the buds of the gastric caecae (gc) forming in the ventriculus. The P-1 enhancer trap is expressed in endodermal cells throughout the midgut, with the exception of a narrow band of 8–10 cells between the ventriculus and the anterior constriction. (B–E) A comparison of P-1 expression in wild-type and gbb mutant embryos. (B) A stage 15 wild-type embryo in which only the central constriction has formed. P-1 expression is evident in the midgut and the ventriculus. (C) A gbb mutant embryo of similar stage to that in B showing loss of P-1 expression from the ventriculus and the anterior and posterior midgut. The central constriction has formed, but is shifted posteriorly with respect to its position in wild type. Also, the buds of the gastric caecae (vertical arrow) have not formed in this embryo. (D) A stage 17 wild-type embryo. At this stage, the midgut is highly convoluted, and the gastric caecae have begun to extend. (E) A gbb mutant embryo of similar stage to that shown in D. P-1 expression is not detected in the anterior and posterior regions of the midgut, and the gastric caecae have not formed (vertical arrow). (F and G) Expression of Antp in wild-type and gbb mutant embryos. (F) A stage 16 wild-type embryo showing the expression of Antp protein in the visceral mesoderm cells overlying the anterior constriction. (G) A gbb mutant embryo of similar stage to that in F. Antp expression is for the most part absent; however, in some cases, a few cells can be found that express Antp (arrow). Genotypes are as follows: (A, B, and D) gbb1/SM6a; P-1(lacZ)/+, (C, E, and G) gbb1/gbb1; P-1(lacZ)/+ and (F) gbb1/CyO ftz-lacZ.

The regional specification of the midgut is initiated by the action of the three homeotic genes, Antennapedia (Antp), Ubx, and abd-A (reviewed in SKAER 1993 ). In the visceral mesoderm cells immediately anterior to the site of the central constriction, Ubx has been shown to activate dpp expression (IMMERGLUCK et al. 1990 ; PANGANIBAN et al. 1990 ; HURSH et al. 1993 ). In turn, dpp positively regulates Ubx to maintain its expression in these cells. In the absence of either Ubx or dpp expression, the central constriction fails to form. Given these observations and the result that Antp mutants fail to form the anterior constriction (REUTER and SCOTT 1990 ), we investigated the possibility that gbb and Antp may be components in a similar regulatory loop. In gbb mutants, the expression of Antp protein is altered in the visceral mesoderm overlying the site where the anterior constriction should form (Figure 3G). In the majority of embryos examined, Antp protein expression was absent; however, in some stage 15 gbb mutant embryos a few cells expressing Antp remain (Figure 3G). The presence of these cells suggests that Antp expression may be initiated, but not maintained, in gbb mutant embryos. It was not possible for us to establish whether Antp activates gbb expression in the visceral mesoderm, for unlike dpp, gbb is expressed broadly in the developing midgut (K. A. WHARTON, unpublished data; DOCTOR et al. 1992 ). We were unable to detect a localized change of gbb expression in the thin, squamous cells of the visceral mesoderm of Antp mutants.

In response to Ubx in PS 7, dpp is secreted from the visceral mesoderm cells and induces the expression of labial in the underlying endoderm (IMMERGLUCK et al. 1990 ; REUTER et al. 1990 ; TREMML and BIENZ 1992 ). As there is no endodermal marker analogous to labial that is specific to the anterior midgut, it was not possible to directly assess the effect of gbb on such an endodermal target. However, in gbb mutants we observe a change in the endodermal expression pattern of the enhancer trap P-1 (SUN et al. 1995 ), indicating that gbb is involved in pattern specification in the embryonic endoderm. In wild-type embryos, P-1 is expressed in the nuclei of endodermal cells within a large region of the midgut, extending from PS 6 through anterior PS 11 (Figure 3A and Figure B). In a gbb null mutant, P-1 expression is restricted to PS 7 through PS 9 (Figure 3C). These results indicate that gbb signaling is required in the endoderm of both the anterior and posterior midgut. Furthermore, the expression of P-1 in the endodermal cells at the very anterior of the wild-type midgut (Figure 3B and Figure D) is absent in gbb mutants (Figure 3C and Figure E). These cells of the ventriculus will evaginate during stage 17 and give rise to the gastric caecae. Consistent with this result, gbb mutants fail to extend the gastric caecae by stage 17 (Figure 3E).

gbb is required in imaginal disc development:
Using the hypomorphic gbb allele gbb⁴ we identified requirements for gbb during imaginal development. gbb⁴ homozygous mutant adults and gbb⁴/gbb^null viable adults exhibit a number of defects. We have previously shown that gbb is required for wing morphogenesis (KHALSA et al. 1998 ). Wings of gbb mutants are smaller and more pointed than wild type and lack the posterior cross vein (Figure 4B). Regions of longitudinal veins 4 and 5 (L4 and L5) are lost, as is the posterior portion of the anterior cross vein (ACV). The extent of longitudinal vein and ACV loss is dependent on the severity of the allelic combination. In general, vein material is lost from the distal margin, but gaps in L4 and L5 are also observed. In addition to vein loss, some gbb mutants also exhibit a slight thickening of distal L2 and/or ectopic vein material flanking L2 (data not shown). In addition to abnormalities in wing morphology, the eyes of gbb¹/gbb⁴ individuals are reduced in size. These flies have 10–20% fewer ommatidia than observed in wild-type flies. The loss of ommatidia appears to be limited to the ventral portion of the eye (data not shown).

View larger version (102K): In this window In a new window Download PPT slide   Figure 4. gbb is required for wing morphogenesis. (A) Wild-type wing on which the longitudinal veins 2–5 (L2–L5) and the anterior and posterior cross veins (ACV, PCV) are noted. (B) Wings from gbb4/gbb2 adults are narrower and more pointed, with a complete loss of the PCV and distal portions of L5 and L4. The ACV is also often incomplete. This wing exhibits the most severe phenotype observed in viable gbb transheterozygotes.

gbb⁴/gbb⁴ and gbb¹/gbb⁴ individuals both exhibit ectopic scutellar and dorsocentral bristles (Table 6; Figure 5). The ectopic scutellar bristles most often occur in close proximity to the endogenous bristle; however, in some cases, extra bristles are observed between the anterior and posterior scutellars (Figure 5B and Figure C). Ectopic bristles are also evident in individuals raised at 18°, but are observed at a lower frequency (Table 6). The absence of bristle defects in gbb¹/gbb⁴ individuals recovered at 25° (Table 6) reflects the fact that in the few individuals recovered in that particular experiment (3% of 380), none had ectopic bristles. Ectopic scutellar or dorsocentral bristles were observed in <3% of the gbb/+ sibs.

View larger version (114K): In this window In a new window Download PPT slide   Figure 5. gbb mutants exhibit ectopic macrochaete. (A) A wild-type scutellum showing the pattern of the four scutellar bristles consisting of a pair of anterior and a pair of posterior macrochaete. In gbb mutants (B and C), ectopic macrochaete are observed. Scutellar bristles are often "twinned" (*, B), and, in some cases, an ectopic bristle (arrow) is found between the anterior and posterior scutellars (C). We do not observe any correlation between the position of ectopic bristles and the genotype. (B) gbb4/gbb1; (C) gbb4/gbb4.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Identification of gbb mutations:
Alleles of the gbb gene were isolated in an F₂ lethal screen for mutations in the 60A chromosomal region. Two null alleles, gbb¹ and gbb², were recovered, as well as two alleles that retain partial gbb function, gbb³ and gbb⁴ (Table 5). The mutations associated with gbb¹ and gbb² each introduce a nonsense codon that would result in a truncated gbb protein that lacks the majority of the ligand domain. The gbb³ mutation changes the putative initiator methionine and, thus, we would expect this mutation to affect the initiation of gbb translation. However, our genetic analysis indicates that the gbb³ allele is not null, and, therefore, it is unlikely that this lesion completely eliminates translation. While the first of four methionines found within the first 52 aa of the gbb ORF conforms best to the Drosophila translational consensus sequence (C/A AA C/A ATG; CAVENER 1987 ), it is possible that in the gbb³ mutant, translation begins at one of the three downstream methionines. Two of these downstream methionines (aa 21 and aa 25) fall within the signal sequence, and should translation initiate at either of these two residues, it is conceivable that some functional gbb protein could be produced.

The fourth gbb allele isolated in our F₂ lethal screen, gbb⁴, results in the alteration of a conserved alanine within the ligand domain. This alanine is located within an -helical loop thought to be conserved among all members of the TGF-ß superfamily (DAOPIN et al. 1992 ; SCHLUNEGGER and GRUTTER 1992 , SCHLUNEGGER and GRUTTER 1993 ). As this region is not thought to participate in dimer formation, but rather in receptor binding (DAOPIN et al. 1992 ; GRIFFITH et al. 1996 ), it is likely that the gbb⁴ mutation affects ligand-receptor interactions or interactions with other extracellular proteins essential for gbb signaling. The gbb⁴ allele is temperature sensitive, consistent with the notion that the mutation affects protein-protein interactions. The isolation of gbb mutations not only identifies residues essential for gbb function, but it allows us to determine the role of gbb in development and to ultimately establish the relationship between gbb signaling and signaling by other BMP or TGF-ß superfamily members.

gbb is required multiple times during development:
With the isolation of both null and hypomorphic gbb alleles, we have determined that gbb has multiple requirements during development. In the embryo, gbb is required for midgut morphogenesis and proper telson formation. During larval stages, wild-type gbb function is necessary for normal fat body and imaginal disc morphology (KHALSA et al. 1998 ). gbb also plays a role during imaginal disc development in the attainment of normal eye and wing size as well as the establishment of wing veins and proper positioning of macrochaete on the notum. Another Drosophila BMP, dpp, has previously been shown to also exhibit multiple requirements during development. Interestingly, many of the tissues or structures affected by a loss in gbb function are also affected by mutations in dpp. In some cases, the resulting mutant phenotype is very similar, while in others, it is qualitatively different. For example, both gbb and dpp mutants exhibit a reduction in the size of imaginal discs and in the eyes of adults (SPENCER et al. 1982 ). The severity of tissue loss differs between dpp and gbb mutants, with the animal being most sensitive to a loss of dpp function. Alterations in wing morphology are observed in both gbb and dpp mutants, with a reduction in the size of the wing and a loss of vein material. However, in this case the specific veins that are preferentially lost in gbb mutants are different from those lost in dpp mutants (SEGAL and GELBART 1985 ; DE CELIS 1997 ).

gbb in midgut morphogenesis:
Defects in the embryonic midgut are also observed in both dpp and gbb mutants, but each BMP appears to play a different role in midgut morphogenesis. gbb is required for the formation of the anterior midgut constriction, while dpp is required for the central constriction. Previous work has indicated that the localized visceral mesoderm expression of homeotic genes Antp, Ubx, and abd-A is required for the correct positioning of the anterior, central, and posterior constrictions, respectively, in the developing midgut (TREMML and BIENZ 1989 ; reviewed in SKAER 1993 ). The homeotic genes have been shown to provide regional specification through their regulation of genes encoding secreted factors, such as dpp and wg, which subsequently act on the underlying midgut endoderm (reviewed by BIENZ 1994 ; BIENZ 1996 ). dpp is activated directly by Ubx in a discrete band of cells in PS 7 of the visceral mesoderm from which Dpp is secreted, resulting in the induction of labial expression in underlying endodermal cells. It has been shown that Ubx expression is, in turn, maintained in the visceral mesoderm via a regulatory feedback loop through the action of dpp. In a manner similar to this regulation of Ubx by dpp, we have shown that the expression of the homeotic gene Antp is regulated by gbb in the visceral mesoderm cells of PS 5 and 6 (Figure 3G). However, a reciprocal regulation of gbb by Antp, as is true of the regulation of dpp by Ubx, is unlikely. The broad expression of gbb throughout the midgut indicates that gbb cannot be regulated exclusively by Antp.

gbb is expressed in both the visceral mesoderm and endoderm (DOCTOR et al. 1992 ; K. WHARTON, data not shown), and, as indicated by the regulation of Antp, gbb signaling is required in the visceral mesoderm. gbb signaling is also required in specific regions of the endoderm. The absence of gbb function eliminates the expression of the endodermal marker P-1 from cells in both the anterior and posterior midgut, as well as from cells in the ventriculus, the site from which the gastric caecae bud. The absence of P-1 staining in the primordia of the gastric caecae in gbb mutant embryos is consistent with gastric caecae defects observed in gbb mutant first instar larvae. It appears that although no gastric caecae are evident in stage 17 gbb mutant embryos, gastric caecae do form, albeit abnormally, by the end of the first larval instar (R. RAY and K. WHARTON, data not shown). In summary, our analysis indicates, as is true for dpp, that gbb signaling is required in both the visceral mesoderm and endoderm of the Drosophila midgut. At this time, we do not know which germ layer or layers serve as the source of the gbb signal.

The specification of positional identity often arises from the localized expression of genes or factors controlling that particular process. It is of interest that although gbb does not exhibit a localized expression pattern, it is involved in regional specification of the midgut. The role of gbb in this process can be explained by two different models. In one model, gbb acts throughout the midgut, but with a partner that provides specific positional information. This partner or cofactor could be another BMP-type ligand or some other signaling component that is specifically localized. Given that the loss of gbb signaling has profound effects, for example, on the formation of the anterior midgut constriction, we would predict that a gbb partner would be localized to the anterior region of the midgut if this model were true.

In the second model, gbb signaling does not specifically require a novel partner to provide positional information, but instead, cells within the midgut respond differently to varying levels of gbb and dpp signaling. This model is consistent with the paradigm we proposed for gbb and dpp signaling in the wing (KHALSA et al. 1998 ). Specification of different regions of the gut could result from the interpretation of different relative levels of gbb to dpp signaling. The total level of BMP signaling may be important, and the localized expression of dpp could provide a source of asymmetry necessary for the establishment of different positional information throughout the midgut. Low levels of signaling provided by gbb alone would specify anterior and posterior midgut vs. the high levels of signaling provided by both gbb and dpp that would specify the central domain of the midgut. Alternatively, differences in the responses elicited by a putative Gbb/Dpp heterodimer and Gbb and Dpp homodimers could be responsible for the assignment of different positional values. Other factors could certainly be involved in refining or elaborating the coarse pattern laid out by gbb and dpp. wg is an example of such a factor, as it has been shown that both wg and dpp are required to activate the expression of certain target genes in several tissues (COHEN 1990 ; TREMML and BIENZ 1992 ; CAMPBELL et al. 1993 ; THURINGER et al. 1993 ; MATHIES et al. 1994 ; BILDER et al. 1998 ). At this time, it is not possible to distinguish between these two simple models, but these models provide a framework within which to investigate further the contribution of multiple BMP signaling to a specific developmental process, midgut morphogenesis.

gbb may signal independently of dpp:
In addition to the gbb mutant phenotypes that resemble dpp mutant phenotypes or those that affect tissues also affected by dpp mutations, we have identified several phenotypes that have not been observed in dpp mutants. Defects in the development of the telson and fat body of the larva have not been reported as aspects of dpp mutants, suggesting that in some developmental processes, gbb may function independently of dpp. It is interesting to note that mutations in the Drosophila BMP signaling components Mad, Medea, and sax can produce a clear larva phenotype (RAFTERY et al. 1995 ; SEKELSKY et al. 1995 ; DAS et al. 1998 ; HUDSON et al. 1998 ; WISOTZKEY et al. 1998 ; V. TWOMBLY and W. M. GELBART, personal communication). Mad and Medea encode Smad proteins shown to mediate dpp signaling in the midgut and the wing imaginal disc (SEKELSKY et al. 1995 ; DAS et al. 1998 ; HUDSON et al. 1998 ; WISOTZKEY et al. 1998 ). sax encodes a type I receptor that has been implicated in dpp signaling (BRUMMEL et al. 1994 ; NELLEN et al. 1994 ; XIE et al. 1994 ) and in gbb signaling (HAERRY et al. 1998 ). It is possible that in the formation of the telson and the fat body, gbb may be the only BMP signal mediated by Mad, Medea, and sax, and gbb signals independently of dpp in these tissues. Alternatively, the earlier requirement for dpp in dorsal/ventral patterning of the embryo may have precluded the identification of dpp involvement in fat body or telson differentiation, and in fact, both gbb and dpp signaling are required for proper development of these structures. Without the ability to bypass the early requirement for dpp, it is not possible at this time to distinguish between these two possibilities.

Our analysis of gbb alleles has also identified a requirement for gbb in the proper specification or positioning of bristles on the notum of the adult fly. A reduction in gbb activity results in the formation of ectopic macrochaete, most frequently on the scutellum. Such a phenotype has not previously been reported for dpp mutants. However, a recent report describing the ubiquitous activation of Tkv, a proposed Dpp receptor, results in ectopic macrochaete formation within the dorsolateral region of the notum (TOMOYASU et al. 1998 ). In this case, ectopic macrochaete formation results from the proposed activation of Dpp signaling via the Tkv receptor. In contrast, we observe ectopic macrochaete with a reduction of gbb function. These opposite phenotypes could reflect a fundamental difference in the role of gbb signaling vs. dpp signaling in the formation or patterning of sensory mother cells, the precursor cells to the macrochaete. Furthermore, the appearance of ectopic macrochaete in the dorsocentral vs. scutellar regions of the notum may reflect a different positional or spatial requirement for dpp vs. gbb. Further analysis will reveal whether the requirement for gbb in scutellar macrochaete formation is independent of the potential role for dpp in the dorsocentrals.

Our phenotypic analysis indicates that gbb and dpp participate in many of the same developmental processes; in some tissues the functions of gbb and dpp appear to be the same or very similar, while in others, their functions appear to be distinct. It is clear that while both gbb and dpp signaling contribute to the proper formation of the embryonic midgut and to patterning of the wing veins in the adult, the relative contribution of each BMP must be different. It is possible that overall, gbb and dpp participate in the development of certain tissues, and this could be accomplished by both cooperative or synergistic interactions and/or antagonistic interactions. As the different mutant phenotypes indicate, the mechanism by which gbb and dpp signaling each contribute to a developmental process must differ depending on the tissue. Understanding the different mechanisms by which these signals are sent and how these differences are regulated in Drosophila will provide significant insight into signaling by multiple TGF-ß/BMP ligands in both invertebrates and vertebrates.

	ACKNOWLEDGMENTS

We are indebted to Bill Gelbart for his efforts and support during the initial stages of this work, notably during the F₂ lethal screen. We thank Bruce Reed for his generosity in providing us with lethal lines from his screen. We also thank Cissy Sun, Phil Lamberty, and Nick Priest for technical assistance, Spyros Artavanis-Tsakonas' lab for help with injections, and Henry Sun, Michael Forte, Steve Wasserman, Dennis McKearin, and the Bloomington Stock Center for generously providing stocks. We appreciate thoughtful comments by Robert Ray on the manuscript and his contributions to the project. We also thank Thom Kaufman and John Doctor for their generosity in providing antibodies. K.A.W. was supported by an Established Investigatorship from the American Heart Association (AHA), with funds contributed in part by the AHA, Maine Affiliate. This work was supported by a National Science Foundation grant and an American Cancer Society Research grant to K.A.W.

Manuscript received August 17, 1998; Accepted for publication March 11, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AONO, A., M. HAZAMA, K. NOTOYA, S. TAKETOMI, and H. YAMASAKI et al., 1995  Potent ectopic bone-inducing activity of bone morphogenetic protein-4/7 heterodimer. Biochem. Biophys. Res. Commun. 210:670-677[Medline].

ARORA, K., M. LEVINE, and M. O'CONNOR, 1994  The screw gene encodes a ubiquitously expressed member of the TGF-ß family required for specification of dorsal cell fates in the Drosophila embryo. Genes Dev. 8:2588-2601[Abstract/Free Full Text].

BIENZ, M., 1994  Homeotic genes and positional signalling in the Drosophila viscera. Trends Genet. 10:22-26[Medline].

BIENZ, M., 1996  Induction of the endoderm in Drosophila. Semin. Cell Dev. Biol. 7:113-119.

BILDER, D., Y. GRABA, and M. P. SCOTT, 1998  Wnt and TGFß signals subdivide the AbdA Hox domain during Drosophila mesoderm patterning. Development 125:1781-1790[Abstract].

BRUMMEL, T. J., V. TWOMBLY, G. MARQUÉS, J. L. WRANA, and S. J. NEWFELD et al., 1994  Characterization and relationship of dpp receptors encoded by the saxophone and thick veins genes in Drosophila. Cell 78:251-261[Medline].

CAMPBELL, G., T. WEAVER, and A. TOMLINSON, 1993  Axis specification in the developing Drosophila appendage: the role of wingless, decapentaplegic, and the homeotic gene aristaless.. Cell 74:1113-1123[Medline].

CAVENER, D. R., 1987  Comparison of the consensus sequence flanking translational start sites in Drosophila and vertebrates. Nucleic Acids Res. 15:1353-1361[Abstract/Free Full Text].

CHO, K. W. Y. and I. L. BLITZ, 1998  BMPs, Smads and metalloproteases: extracellular and intracellular modes of negative regulation. Curr. Opin. Genet. Dev. 8:443-449[Medline].

CHOU, T.-B., E. NOLL, and N. PERRIMON, 1993  Autosomal P[ovo^D1] dominant female-sterile insertions in Drosophila and their use in generating germ-line chimeras. Development 119:1359-1369[Abstract].

COHEN, S. M., 1990  Specification of limb development in the Drosophila embryo by positional cues from segmentation genes. Nature 343:173-177[Medline].

DAOPIN, S., K. PIEZ, Y. OGAWA, and D. DAVIES, 1992  Crystal structure of transforming growth factor-ß2: an unusual fold for the superfamily. Science 257:369-373[Abstract/Free Full Text].

DAS, P., L. L. MADUZIA, H. WANG, A. L. FINELLI, and S.-H. CHO et al., 1998  The Drosophila gene Medea demonstrates the requirement for different classes of Smads in dpp signaling. Development 125:1519-1528[Abstract].

DE CELIS, J. F., 1997  Expression and function of decapentaplegic and thick veins during the differentiation of the veins in the Drosophila wing. Development 124:1007-1018[Abstract].

DOCTOR, J. S., D. JACKSON, K. E. RASHKA, M. VISALLI, and F. M. HOFFMANN, 1992  Sequence, biochemical characterization, and developmental expression of a new member of the TGF-ß superfamily in Drosophila melanogaster.. Dev. Biol. 151:491-505[Medline].

DUDLEY, A. T. and E. J. ROBERTSON, 1997  Overlapping expression domains of bone morphogenetic protein family members potentially account for limited tissue defects in BMP7 deficient embryos. Dev. Dyn. 208:349-362[Medline].

FLYBASE,, 1996  Flybase: the Drosophila database. Nucleic Acids Res. 24:53-56[Abstract/Free Full Text].

GELBART, W. M., 1989  The decapentaplegic gene: a TGF-ß homologue controlling pattern formation in Drosophila. Development 107(Suppl.):65-74.

GRIFFITH, D., P. KECK, T. SAMPATH, D. RUEGER, and W. CARLSON, 1996  Three-dimensional structure of recombinant human osteogenic protein 1: structural paradigm for the transforming growth factor ß superfamily. Proc. Natl. Acad. Sci. USA 93:878-883[Abstract/Free Full Text].

HAERRY, T. E., O. KHALSA, M. B. O'CONNOR, and K. A. WHARTON, 1998  Synergistic signaling by two BMP ligands through the SAX and TKV receptors controls wing growth and patterning in Drosophila. Development 125:3977-3987[Abstract].

HOGAN, B. L. M., 1996  Bone morphogenetic proteins: multifunctional regulators of vertebrate development. Genes Dev. 10:1580-1594[Free Full Text].

HUDSON, J. B., S. D. PODDOS, K. KEITH, S. L. SIMPSON, and E. L. FERGUSON, 1998  The Drosophila Medea gene is required downstream of dpp and encodes a functional homolog of human Smad4. Development 125:1407-1420[Abstract].

HURSH, D. A., R. W. PADGETT, and W. M. GELBART, 1993  Cross regulation of decapentaplegic and Ultrabithorax transcription in the embryonic visceral mesoderm of Drosophila. Development 117:1211-1222[Abstract].

IMMERGLUCK, K., P. A. LAWRENCE, and M. BIENZ, 1990  Induction across germ layers in Drosophila mediated by a genetic cascade. Cell 62:261-268[Medline].

KHALSA, O., J.-W. YOON, S. SCHUMANN-TORRES, and K. WHARTON, 1998  TGF-ß/BMP superfamily members, Gbb-60A and Dpp, cooperate to provide pattern information and establish cell identity in the Drosophila wing. Development 125:2723-2734[Abstract].

KINGSLEY, D. M., 1994  The TGF-ß superfamily: new members, new receptors, and new genetic tests of function in different organisms. Genes Dev. 8:133-146[Free Full Text].

LEWIS, E. B. and F. BACHER, 1968  Method for feeding ethyl methane sulfonate (EMS) to Drosophila males. Drosophila Inf. Serv. 43:193.

LINDSLEY, D. L., and G. ZIMM, 1992 The Genome of Drosophila melanogaster. Academic Press, San Diego.

LYONS, K. M., B. L. M. HOGAN, and E. J. ROBERTSON, 1995  Colocalization of BMP7 and BMP2 RNAs suggests that these factors cooperatively mediate tissue interactions during murine development. Mech. Dev. 50:71-83[Medline].

MASSAGUÉ, J., 1998  TGF-ß signal transduction. Annu. Rev. Biochem. 67:753-791[Medline].

MASSAGUÉ, J., L. ATTISANO, and J. L. WRANA, 1994  The TGF-ß family and its composite receptors. Trends Cell Biol. 4:172-178[Medline].

MASSAGUÉ, J., A. HATA, and F. LIU, 1997  TGF-ß signalling through the Smad pathway. Trends Cell Biol. 7:187-192.

MATHIES, L. D., S. KERRIDGE, and M. P. SCOTT, 1994  Role of the teashirt gene in Drosophila midgut morphogenesis: secreted proteins mediate the action of homeotic genes. Development 120:2799-2809[Abstract].

NELLEN, D., M. AFFOLTER, and K. BASLER, 1994  Receptor serine/threonine kinase implicated in the control of Drosophila body pattern by decapentaplegic.. Cell 78:225-237[Medline].

PADGETT, R., R. ST. JOHNSTON, and W. GELBART, 1987  A transcript from a Drosophila pattern gene predicts a protein homologous to the transforming growth factor-ß family. Nature 325:81-84[Medline].

PADGETT, R. W., P. DAS, and S. KRISHNA, 1998  TGF-ß signaling, Smads, and tumor suppressors. BioEssays 20:382-391[Medline].

PANGANIBAN, G. E. F., R. REUTER, M. P. SCOTT, and F. M. HOFFMAN, 1990  A Drosophila growth factor homolog, decapentaplegic, regulates homeotic gene expression within and across germ layers during midgut morphogenesis. Development 110:1041-1050[Abstract/Free Full Text].

PIROTTA, V., 1988 Vectors for P-mediated transformation in Drosophila, pp. 437–456 in Vectors: A Survey of Molecular Cloning Vectors and Their Uses, edited by R. L. RODRIGUEZ and D. T. DENHARDT. Butterworths, Boston/London.

RAFTERY, L., V. TWOMBLY, K. WHARTON, and W. GELBART, 1995  Genetic screens to identify elements of the decapentaplegic signaling pathway in Drosophila. Genetics 139:241-254[Abstract].

REED, B., 1992 The genetic analysis of endoreduplication in Drosophila melanogaster. Ph.D. Thesis, University of Cambridge, England.

REUTER, R. and M. P. SCOTT, 1990  Expression