A Drosophila protein related to the human zinc finger transcription factor PRDII/MBPI/HIV-EP1 is required for dpp signaling

The transforming growth factor beta (TGF-β) superfamily consists of more than 24 secreted factors that play key roles in cellular differentiation and proliferation (Kingsley, 1994). The high degree of evolutionary conservation of this family of proteins is indicated by the identification of three members in Drosophila, decapentaplegic (dpp), 60A and screw (Padgett et al., 1987; Doctor et al., 1992; Wharton et al., 1991; Arora et al., 1994). Genetic analysis has implicated these factors in a variety of developmental processes. For example, embryos that lack dpp are completely ventralized and fail to form dorsally derived structures (Ferguson and Anderson, 1992; Irish and Gelbart, 1987). Injection of increasing amounts of dpp mRNA causes progressive increase in dorsal cell fates, suggesting that an activity gradient of dpp instructs different dorsal/ventral ectodermal cell fates along the gradient (Ferguson and Anderson, 1992).

Recently, it has been reported that dorsal ectodermal expression of dpp is also critical for the induction of specific cell fates in the underlying mesodermal cell layer that gives rise to the visceral mesoderm (Frasch, 1995; Staehling-Hampton et al., 1994). Later in embryogenesis, dpp is involved in maintaining patterns of homeotic gene expression in the midgut visceral mesoderm and in the induction of the homeotic gene labial across germ layers in the developing midgut endoderm (Immerglück et al., 1990; Panganiban et al., 1990; Reuter et al., 1990; Bienz, 1994; Staehling-Hampton and Hoffmann, 1994). During the larval stages, dpp expression is activated by the product of the hedgehog gene in specific sets of cells in the imaginal disks (Basler and Struhl, 1994). dpp is required for the proliferation, differentiation and proximal-distal patterning of the imaginal disks and the migration of the morphogenetic furrow across the eye imaginal disk (Heberlein et al., 1993; Masucci et al., 1990; Bryant, 1988).

Cells respond to dpp or other TGF-β-related factors through the activation of transmembrane serine/threonine kinase receptors (Massagué et al., 1994; Miyazono et al., 1994). In Drosophila, four transmembrane serine/threonine kinases that are similar to mammalian TGF-β receptors have been identi-fied (Childs et al., 1993; Penton et al., 1994; Nellen et al., 1994; Brummel et al., 1994; Xie et al., 1994; Letsou et al., 1995; Ruberte et al., 1995). Three of these, thick veins (tkv), saxophone (sax) and Atr I, have sequence characteristics of type I receptors. One of the Drosophila receptors, punt, is similar in sequence to other type II receptors. Molecular evidence indicates that the tkv, sax and punt receptors can bind dpp or its closest mammalian relative BMP2 (Penton et al., 1994; Brummel et al., 1994; Letsou et al., 1995). Atr I and punt also bind to mammalian activin (Childs et al., 1993; Wrana et al., 1994). Analysis of phenotypes caused by mutations in tkv, sax and punt has implicated tkv and punt in virtually all dpp responses and sax in a subset of dpp responses. Mutations in Atr I have not been reported. By analogy to the mechanism of ligand activation of the mammalian TGF-β receptors, it is believed that ligand-induced oligomerization of the Drosophila type II and type I receptors results in the phosphorylation and activation of type I receptors by type II receptors (Wrana et al., 1992). However, it is not currently known what proteins signal downstream of the activated type I receptors.

In this report, we describe the molecular and genetic characterization of schnurri (shn). shn is a putative zinc finger tran-scription factor related to human PRDII/MBP1/HIV-EP1 (Fan and Maniatis, 1990; Baldwin et al., 1990; Maekawa et al., 1989). Analysis of shn mRNA expression and mutant phenotypes indicates that it is expressed in tissues that respond to dpp signaling and that its expression is required for these responses to dpp. shn mutations cause a zygotic embryonic phenotype similar to that caused by mutations in the dpp receptors tkv and punt, absence of dorsal hypoderm resulting in a dorsal hole in the cuticle. shn mutant alleles interact genet-ically during wing formation with dpp alleles and with alleles of tkv. Furthermore, dpp-induced mesoderm gene expression is blocked in shn mutants. We propose that the signal trans-duction pathway(s) which are initiated when dpp activates its receptors require shn function to effect dpp-induced cell fate changes.

shn^IB, tkv¹, Df(2L)tkv2 and dpp^s5 are described in Lindsley and Zimm (1992). twist-GAL4 and UAS-dpp flies are described in Staehling-Hampton et al. (1994) and Staehling-Hampton and Hoffmann (1994). P. element stock P{ry[+], P\T:lacZ:Hsp70 Km[r] ori=PZ} I(2)04738, cn/CyO; ry[506] was obtained from the Indiana stock center (Bloomington, IN).

Genomic DNA from adult flies with the P. element insertion was prepared according to Ashburner (1989) except that a different lysis buffer (0.1 M Tris-HCL pH 9.0; 0.1 M EDTA; 1% SDS) was used. Plasmid rescue was done using the procedure of Karpen and Spradling, Carnegie Institute (personal communication). The rescued DNA was transformed into maximum efficiency DH5α cells (GIBCO BRL). A 4 kb HindIII fragment of plasmid-rescued genomic DNA was labeled with digoxigenin and used as a probe for embryo and polytene chromosome in situ hybridizations. The 4 kb genomic fragment hybridized to the shn genomic region (47E-48B) and detected a dorsal specific message in blastoderm embryos. The same genomic fragment was then used as a probe to isolate cDNAs from a 4-8 hour embryonic cDNA library (gift of Nick Brown). The shn cDNAs were sequenced by standard dideoxy chain termination protocols (Sequenase, US Biochemicals). The longest cDNA recovered was 6.5 kb and was truncated at the 5′ end. To obtain more 5′ sequence, RACE PCR was done using the Marathon amplification cDNA kit (Clontech) according to the manufacturer’s protocol. The RACE products ranged in size from 3.5 to 2.0 kb and were identical except for length. With the RACE sequence, the total length of the cDNA was 9 kb. Sequence alignments were done using the Bestfit program developed by the Genetics Computer Group (Madison, WI).

The anti-fasIII monoclonal was a gift of M. Bate, University of Cambridge and was originally described in Patel et al. (1987). snail antibody made in rabbits was a gift of Sean Carroll, University of Wisconsin, and was originally described in Alberga et al. (1991). fasIII (1:100) and snail (1:500) antibody stainings were done according to the methods of Michael Bate, Uni-versity of Cambridge (personal communication); Horse radish peroxidase (HRP) detection was done using the Vectastain Elite ABC kit developed by Vector laboratories, Inc.

anti-shn3 antibody stainings were done according to the methods of Carroll and Scott (1985) with the following modifications. The shn primary antibody was preabsorbed overnight at 4°C in a large volume of wild-type embryos in PBS + 0.1% Triton X-100 and 0.2% BSA. To block background staining from endogenous peroxi-dases, the embryos were treated for 10 minutes in 3% H₂O₂ after fixation. The shn3 antibody was used at a concentration of 1 μg/ml.

In situ hybridization to mRNA in embryos for RNA utilized digoxigenin-labeled cDNA or RNA probes (made using the Genius kit (Boehringer-Mannheim)) according to the methods of Tautz and Pfeiffle (1989).

A 39×10³M_r C-terminal segment containing the last three shn zinc fingers was expressed and purified as a histidine-tagged fusion. A DNA fragment extending from ScaI at 6737 to BstBI at 8300 was inserted into pRSETA (Invitrogen) at the PvuII and BstBI sites. The resulting 43×10³M_r fusion protein was extracted from inclusion bodies in 6 M guanidine/20 mM NaPO₄, pH 7.8 and affinity purified on a Probond Ni²⁺ column (Invit-rogen) by adsorption, washing in 8 M urea/500 mM NaCl/20 mM NaPO₄, pH 7.8, and elution in wash buffer adjusted to pH 4. Eluted protein was dialysed against PBS (5 mM NaPO₄, pH 7.5/137 mM NaCl) plus 10 μM ZnSO₄ and used to immunize rabbits.

Animals mutant for the dpp receptors, tkv or punt, lack embryonic dorsal hypoderm and have a hole in the dorsal surface of the cuticle (Penton et al., 1994; Nellen et al., 1994; Brummel et al., 1994). Other mutations that have a ‘dorsal-open’ phenotype have been isolated in genetic screens for embryonic lethal mutations in Drosophila (Jürgens et al., 1984). To investigate whether some of these genes encode products that function on the dpp pathway, we examined the genetic interactions between nine different dorsal-open mutations and alleles of dpp or tkv (Table 1). We also examined the dorsal-open mutant embryos for changes in the expression of the homeotic gene labial, a target of the dpp signaling pathway in the embryonic midgut (Staehling-Hampton and Hoffmann, 1994; Panganiban et al., 1990; Immerglück et al., 1990; Reuter et al., 1990).

One of the dorsal-open mutations, schnurri (shn), interacted genetically with dpp alleles and with tkv alleles and showed reduced midgut expression of labial. 91% of shn^IB/dpp^s5 heterozygotes had mutant wings with a shortened L4 wing vein (Fig. 1B). The shortened wing vein phenotype was rescued by addition of a dpp transgene into the background (Fig. 1A) and was indistinguishable from wild type. Flies with one copy of an adult viable tkv allele, tkv¹, over a deficiency of tkv, Df(2L)tkv2, have wing veins that are slightly thickened at the tip (Fig. 1C; Penton et al., 1994). Introducing one copy of shn^IB into a tkv¹/Df(2L)tkv2 background greatly enhanced the thickened vein phenotype and led to terminal gaps in veins L2, L3, L4 and L5 (Fig. 1D). Some of the tkv¹/Df(2L)tkv2, shn^IB flies were also missing terminal leg claws (data not shown), a phenotype associated with reduced dpp expression in the imaginal disks (Spencer et al., 1982). The expression of the homeotic gene labial was reduced in the midgut endoderm of shn^IB homozygotes (data not shown), indicating that expression of a specific dpp target gene was affected in shn mutants.

A P element insertion line was obtained that failed to complement the embryonic dorsal-open phenotype of shn^IB (data not shown). The P element insertion allele also showed heterozygous genetic interactions with dpp^s5 (data not shown). A 4 kb piece of genomic DNA flanking the transposon insertion was recovered by plasmid rescue of the P element and used to identify shn cDNAs (see Materials and Methods). The shn cDNA has an open reading frame of 7584 base pairs with approximately 0.2 kb of 5′ and 1.2 kb of 3′ untranslated sequence and is predicted to encode a 272×10³M_r (2528 amino acid) protein (Fig. 2).

The predicted shn protein sequence includes 8 regions with sequence similarity to C₂-H₂-type zinc fingers (Klug and Rhodes, 1987) (Fig. 2). The Zn fingers are clustered in two pairs and one triplet. There is also a region similar in sequence to acidic transcriptional activation domains (Fig. 2). Between the first and second predicted Zn finger pairs, there is a region, C-C-X₁₃-H-C, that may also form a single Zn finger and seven repeats of the amino acid sequence S/TPXK/R (Fig. 2). S/TPXK/R motifs are putative DNA-binding domains and may be regulated by phosphorylation (Hill et al., 1990; Suzuki, 1989; Churchill and Suzuki, 1989).

The expression pattern of the shn message in embryos is similar to the pattern of mRNA for the dpp receptor tkv (Penton et al., 1994; Brummel et al., 1994). Like tkv, the shn message is maternal and was expressed faintly but uniformly in early embryos (Fig. 3B). In contrast to the antisense probe (Fig. 3B), a sense probe made to shn did not detect a message in early embryos (Fig. 3A). During the blastoderm stage, shn mRNA was restricted to the dorsal half of the embryo similar to the pattern of dpp (St Johnston and Gelbart, 1987) and tkv mRNAs (Fig. 3C). As embryogenesis proceeded, shn was detected in the presumptive mesoderm of the invaginating ventral furrow (Fig. 3D, arrowhead) and was expressed in the mesoderm throughout germ band extension (Fig. 3E, solid arrow). tkv mRNA is also detected in the mesoderm throughout gastrulation (Penton et al., 1994; Brummel et al., 1994). In germ band retracted embryos, shn was expressed in the endoderm (Fig. 3F, arrows). After fusion of the posterior and anterior midgut, shn was detected in two domains corresponding to parasegment 4 and 7 (Fig. 3G, arrowheads). The shn midgut domains are transient and are only detected prior to the formation of the midgut constrictions. In dpp null embryos shn mRNA is detected in the mesoderm (Fig. 3H, arrows) and dorsal ectoderm (Fig. 3H, arrowheads), showing that dpp is not required for shn expression. The specificity of the probe for shn mRNA was confirmed by the absence of detectable expression in shn mutant embryos (Fig. 3E, open arrow).

To demonstrate that the cDNA we cloned is shn, wild-type and shn mutant embryos were stained with antibodies made to the third zinc finger set of the putative shn cDNA. In wild-type germ band retracted embryos, shn protein was detected in the endoderm (Fig. 4A, arrowheads) and in the foregut (Fig. 4A, arrow). In embryos homozygous for the shn P element insertion no protein was detected in either the foregut or the endoderm (Fig. 4B). This was consistent with the absence of shn mRNA in shn P element homozygotes (Fig. 3E). In embryos homozygous for the chemically induced shn^IB allele, no staining for shn protein was observed (Fig. 4C). Therefore, two different mutant alleles cause the absence of detectable shn product. This evidence, together with the isolation of the DNA using the shn P element insertion and the localization of the cloned DNA on polytene chromosomes to the shn region, all support the conclusion that the cloned cDNA is shn. A more complete description of the shn protein expression pattern will be presented in a future manuscript.

The dorsally restricted expression pattern of shn suggested that shn might be required for the dorsal blastoderm expression of either dpp or tkv. To test this possibility, we examined the expression of dpp or tkv mRNA in a population of embryos from a cross of shn^IB/CyO males and females. 25% of the embryos would have homozygous shn mutations. 90% of cellular blastoderm embryos (n=132) showed normal dorsal expression of dpp and 96% (n=67) showed normal dorsal expression of tkv. The small percentage of embryos not staining is likely do to problems with fixation or probe permeability. We conclude that the absence of zygotic shn product does not alter early expression patterns of dpp or tkv.

The later shn mRNA expression pattern suggested that shn might be involved in mesodermal gene expression and in the regulation of midgut gene expression. In the visceral mesoderm of the midgut, dpp regulates its own expression in the gastric caeca primordia (parasegment 4) and parasegment 7 of the visceral mesoderm (Staehling-Hampton and Hoffmann, 1994; Hursh et al., 1993). Expression of dpp in the gastric caeca primordia and parasegment 7 visceral mesoderm was absent in shn mutants but dpp expression in other regions of the alimentary canal (pharynx, esophagus and hindgut) was unaffected (Fig. 5A,B). Note, however, that the intensity of the narrow lateral stripe of ectodermal dpp expression also was reduced in shn mutants during germ band retraction (Fig. 5A,B arrowheads).

The morphology of the midgut in shn mutants is abnormal and the gut constrictions do not form. Round midguts without constrictions are characteristic of mutants that fail to differen-tiate visceral mesoderm such as bagpipe (bap) and tinman (Bodmer, 1993; Azpiazu and Frasch, 1993). To determine if visceral mesoderm differentiation is abnormal in shn mutants, we examined the expression of two visceral mesoderm markers bap (Azpiazu and Frasch, 1993) and fasicilin III (fasIII) (Bate, 1993; Patel et al., 1987). Visceral mesoderm bap mRNA expression was reduced in shn mutants (Fig. 5C-F) and the uniform expression pattern of fasIII protein in the visceral mesoderm was broken up into discrete pieces (Fig. 5G,H).

The imaginal disk primordia form in the lateral ectoderm at sites approximating the intersection of dpp expression and expression of wingless, the wnt1 growth factor homologue in Drosophila (Cohen et al., 1993). In shn mutant embryos, the dorsal/ventral patterning of the cuticle is largely normal with the exception of the most dorsal cuticular structures. However, when shn^IB mutant embryos were stained with antibodies to snail (Alberga et al., 1991), a marker of the wing and haltere primordia, no expression in the imaginal disk precursors was seen (Fig. 5I,J). The absence of snail expression and the reduction of the lateral stripe of dpp expression (Fig. 5B) in this region of the embryo indicates a defect in the cell fates at this lateral position along the dorsal/ventral axis in shn mutant embryos.

The effects of shn mutations on the mesoderm and lateral ectoderm implicate it in dpp functions. Although the mRNA expression pattern was most consistent with shn being involved in responding to the dpp signal rather than regulation of dpp expression, we tested the effects of shn mutations on dpp responses caused by ectopic expression of dpp to confirm its involvement in response to or ‘downstream of’ dpp. In wild-type embryos bap is expressed in dorsal/lateral patches on both sides of the ventral midline (Fig. 6A). When dpp is expressed ectopically throughout the mesoderm by expressing UAS-dpp in the twist-GAL4 pattern (see Brand and Perrimon (1993) for a description of the GAL4/UAS system), the expression of bap spreads and extends across the midline (Fig. 6B and Staehling-Hampton et al., 1994). In a shn mutant background, however, the expansion of bap by ectopic dpp was blocked (Fig. 6C) suggesting that shn functions downstream of dpp in mesodermal induction.

In Drosophila, the identification of gene products acting on signaling pathways has occurred by analysis of genes with similar null phenotypes, e.g., the cascade of maternal effect genes responsible for dorsal-ventral axis formation (Chasan and Anderson, 1994), or by second-site modifier screens, e.g. the signal transduction pathway downstream of the sevenless receptor tyrosine kinase (Simon, 1994). To identify other gene products on the dpp signal transduction pathway, we examined genes whose mutant phenotypes were similar to the dorsal-open cuticle caused by mutations in the dpp receptor tkv. Mutations in one of these genes, shn, interacted with alleles of dpp and alleles of tkv. The expression of dpp target genes was reduced in shn mutants suggesting that shn is required for the expression of some dpp target genes. Mutations in other genes causing a dorsal-open phenotype did not exhibit additional interactions with dpp mutations. Several of these other genes have also recently been cloned including pannier (Ramain et al., 1993; Winick et al., 1993) and canoe (Miyamoto et al., 1995). pannier encodes a GATA transcription factor that regulates achaete and scute expression and functions during neural precursor development (Ramain et al., 1993). pannier mRNA is expressed dorsally in blastoderm embryos and its expression is absent in dpp mutant embryos (Winick et al., 1993). This is in contrast to shn mRNA expression in the early embryo which is unaffected in dpp null mutant embryos (Fig. 3H). Thus, pannier may be directly or indirectly a target of dpp signaling whereas shn protein is more likely to participate in the dpp signaling cascade. The product of the canoe gene is a novel protein with a GLGF/DHR motif that is thought to function in the same developmental pathways as Notch and scabrous (Miyamoto et al., 1995). Thus, we expect that inter-ference with dpp signaling is only one of several causes for the dorsal-open phenotype.

At late germ band extension, when the mesoderm differen-tiates into subtypes of mesoderm, shn and tkv are expressed in the mesoderm, whereas dpp expression at this stage is confined to the overlying ectoderm. This expression pattern suggested that shn might function in tissues responding to dpp rather than in tissues producing dpp. Expression of dpp in the apposing ectoderm is both necessary and sufficient to induce the expression of visceral mesoderm-specific genes such as bap (Frasch, 1995; Staehling-Hampton et al., 1994). In shn mutants, bap expression is reduced and visceral mesoderm development is abnormal as revealed by fasIII protein expression. In order to test whether shn might be exerting its effect through the regulation of dpp expression, we expressed the dpp cDNA under the control of an artificial promoter responsive to the yeast GAL4 upstream activation sequence (UAS). The expansion of bap expression by ectopic dpp was blocked in shn mutants supporting the model that shn acts downstream of dpp during mesoderm induction. It will be important to determine whether shn protein binds to target sites in bap and activates bap transcription directly.

dpp and labial expression are reduced in the midgut of shn mutants, indicating that shn expression in the midgut (Fig. 3G) may be necessary for the induction of labial in the endoderm by dpp in the visceral mesoderm and/or for the autoregulation of dpp in the visceral mesoderm. However, the general defectiveness of the midgut revealed by fas III staining suggests that the absence of normal patterns of dpp and labial expression in the midgut should be interpreted with caution as they may be an indirect consequence of the incomplete differentiation of the midgut. To determine if shn functions in the induction of labial or in the autoregulation of dpp, mutations that remove or overexpress shn specifically at later embryonic stages, e. g., the germ band retracted stage, are needed. If a role for shn can be established in the visceral mesoderm responses to dpp, it will be of interest to determine whether the dpp-responsive regulatory regions identified in Ultrabithorax (Ubx) (Thüringer et al., 1993), labial (Tremml and Bienz, 1992) and dpp (Hursch et al., 1993) have binding sites for shn protein.

The requirement for shn in the embryonic ectoderm is revealed by the defects in dorsal hypoderm as well as the defects in the lateral ectoderm. The narrow lateral stripe of dpp expression in the lateral ectoderm is greatly dimin-ished and the expression of snail in the wing and haltere disk primordia is not detectable. The defects in dorsal hypoderm and the lateral ectoderm correspond to two regions of late dpp expression, the narrow stripe at the dorsal edge of the ectoderm and the narrow stripe in the lateral ectoderm (St Johnston and Gelbart, 1988; Jackson and Hoffmann, 1994). The function of dpp expression in these two stripes is not known as mutations that specifically remove these aspects of the dpp expression pattern have not been recovered. Further analysis of the shn mutant phenotype may provide new insights into the functions of the mid-embryonic lateral stripe of ecto-dermal dpp expression.

Computer database searches revealed that shn is related to PRDII/MBPI/HIV-EP1, a mammalian transcription factor (Fan and Maniatis, 1990; Baldwin et al., 1990; Maekawa et al., 1989). PRDII/MBPI/HIV-EP1 is one member of a family of related zinc finger transcription factors that bind to nuclear factor κB (NF-κB)-related sites (Fan and Maniatis, 1990; Baldwin et al., 1990; Maekawa et al., 1989). PRDII/MBPI/HIV-EP1 related proteins bind to sites in the major histocompatability complex class I genes (Van ‘T Veer et al., 1992; Baldwin et al., 1990; Mitchelmore et al., 1990; Rustgi et al., 1990), the Kappa immunoglobulin genes (Baldwin et al., 1990; Rustgi et al., 1990), human interferon-β gene (Fan and Maniatis, 1990), β2-microglobulin gene (Baldwin et al., 1990), rat angiotensinogen gene (Ron et al., 1991), the αA-crystalin gene (Nakamura et al., 1990), α₁-anti-trypsin gene (Mitchelmore et al., 1990), the recombination signal sequences (Rss) of immunoglobulin and T-cell receptor genes (Wu et al., 1993) and the HIV long terminal repeat (LTR) (Nomura et al., 1991; Maekawa et al., 1989; Seeler et al., 1994). PRDII/MBPI/HIV-EP1 binding to NF-κB sites in the HIV-1 LTR can activate HIV-1 gene expression (Seeler et al., 1994), however, no other functional role for PRDII/MBPI/HIV-EP1 related factors has been demonstrated. PRDII/MBPI/HIV-EP1 mRNA is induced by viral infection and by growth factors (Fan and Maniatis, 1990; Baldwin et al., 1990) suggesting that the proteins have the potential to be regulated by growth factor or viral induced signaling.

Like shn, PRDII/MBPI/HIV-EP1 has pairs of widely spaced Zn fingers (Fig. 7A). The shn and PRDII/MBPI/HIV-EP1 Zn fingers are remarkably conserved with amino acid identities of 67% and 78% (Fig. 7A,B). PRDII/MBPI/HIV-EP1 also has a single C-C-X₁₃-H-C zinc finger motif and a region rich in acidic amino acids. However, unlike shn, PRDII/MBPI/HIV-EP1 does not have a third triplet of Zn fingers and has only two S/TPXK/R repeats. Other PRDII/MBPI/HIV-EP1-related factors, Rc and MBP2, have 5 and 11 S/TPXK/R repeats respectably (Wu et al., 1993; Van ‘T Veer et al., 1992). We conclude that shn and PRDII/MBPI/HIV-EP1 are members of the same gene family but that there may be a closer mammalian homologue of shn.

Since the first two pairs of Zn fingers in shn and PRDII/MBPI/HIV-EP1 are highly conserved it is possible that shn, like PRDII/MBPI/HIV-EP1, binds to NF-κB-related sites. Two Drosophila proteins that bind to NF-κB-related sites are dorsal and the dorsal-related protein Dif. dorsal protein is present in the nucleus of ventral blastoderm cells where it is critical for activating expression of twist and repressing expression of zen and dpp (Roth et al., 1989; Rushlow et al., 1989; Steward, 1989). Dif functions in the Drosophila immune response (Ip et al., 1993). The analysis of shn mutant phenotypes has not yet addressed whether there are any functional interactions between shn and dorsal or shn and Dif that might indicate competition of the proteins for binding sites at target genes.

The expression pattern of shn mRNA, the effects of shn mutations on dpp-induced responses and the sequence similarities of shn to mammalian zinc-finger transcription factors suggest that shn is a critical factor on a signaling pathway activated by dpp. Genetic screens for second-site modifiers of dpp mutant phenotypes have recovered mutations in other genes whose products may be important for dpp signaling (Raftery et al., 1995). The first of these genes to be molecularly cloned, mad, encodes a novel protein whose sequence has unfortunately not provided any immediate clues to its molecular role in dpp signal transduction (Sekelsky et al., 1995). Further genetic dissection of the dpp signaling pathway will be facilitated by strategies that sensitize the pathway at different points to recover second-site modifiers. The identification of shn as a component on the dpp signaling pathway therefore provides new opportunities for sensitizing the pathway in an effort to identify additional gene products involved. The similarity of shn to known DNA-binding proteins also suggests that shn will be useful in the identification of additional target genes regulated by dpp signaling. This might be accomplished for example by procedures used recently to identify new target genes of Ultrabithorax (Mastick et al., 1995). Identification of target genes will be particularly useful in further understanding the roles of dpp in the imaginal disks since very little is known about the molecular mechanisms for dpp’s action in distal patterning of legs and wings and migration of the morphogenetic furrow in the eye. Finally, the evolutionary conservation of the TGF-β-related ligands and their receptors raises the promise that other components on the signaling pathway will also be well conserved and that a mammalian homologue of shn will be involved in some of the responses to TGF-β-related factors in mammalian cells.

We thank Drs Michael Bate and Sean Carroll for antibodies, the Indiana Stock Center for fly stocks and Patricia Rower-Nutter for assistance with antibody stainings and DNA sequencing. We thank Dr Rahul Warrior for discussions and Drs Marcus Affolter and Konrad Basler for communication of unpublished results. This work was supported by grants from the NIH: RR06610 to F. M. H., Cancer Center Core support CA07175 and Predoctoral Training Grant CA09135 support for K. S-H; F. M. H. is the recipient of a Faculty Research Award from the American Cancer Society.

The Genbank accession number for the schnurri sequence is U31368