ABSTRACT
We have identified a novel Drosophila homeodomain gene, unplugged (unp), whose function is required for formation of the tracheal branches that penetrate the CNS. In unp mutant embryos the segmentally repeated ganglionic branches stall and fail to penetrate the CNS and the segment-specific cerebral branch and associated cerebral anastomosis fail to form. Expression of unp in the founder cells for the cerebral branch within the first tracheal metamere is repressed in posterior segments by Ubx and other bithorax complex genes. This pattern of expression and homeotic gene regulation is reproduced by an unusual 2.7 kb cis-regulatory sequence located downstream of the unp transcription unit. Since the unp protein is localized to the nucleus of tracheal precursor cells as they migrate and extend, unp protein appears to play a regulatory role in neural branching of the tracheae, and the segment-specific aspects of these neural branching patterns appear to be generated by homeotic regulation of unp expression.
INTRODUCTION
A key turning point in the early understanding of Drosophila homeotic gene function was the shift from analysis of partial loss-of-function mutations in adults to analysis of null mutations associated with embryonic or larval lethality. In the first major description of such lethal homeotic phenotypes, Lewis (1978) relied heavily upon branching patterns within the tracheal system as segment-specific landmarks that allowed him to formulate an insightful model relating the functions of homeotic genes to their evolution and chromosomal organization. Much has been learned in the interim about homeotic selector (HOM) gene function. HOM genes thus have been shown to contain the homeodomain, a conserved 60 residue motif that mediates sequence-specific DNA binding, and HOM genes in Drosophila and in other animals are thought to specify the characteristics of spatial units along the anteroposterior axis via the transcriptional regulation of target genes (for review see McGinnis and Krumlauf, 1992). Despite these advances and despite the identification of several potential HOM gene targets in Drosophila, however, no downstream targets have been described that are specifically involved in the segmentspecific branching of the tracheal system.
Although many genes are known to be expressed within the developing tracheal system (reviewed by Manning and Krasnow, 1993), most of these genes are also expressed in a variety of other tissues, thus complicating the analysis of their function and regulation. Two genes whose functions appear to be specifically concerned with tracheal development are breathless (btl) and pointed (pnt). The pnt gene encodes an ets domain-containing transcription factor and is expressed in the tracheal placodes and in the developing tracheal branches (Klambt, 1993). The btl gene encodes the Drosophila homologue of the FGF receptor and also is expressed in tracheal precursors during invagination and branching. Although mutations in both of these genes block migration and branching of the tracheal precursors, these effects appear to be general and are difficult to relate to the segment-specific aspects of tracheal morphogenesis that are presumed to be under homeotic gene control.
We present here the identification and molecular and genetic characterization of unplugged (unp), a gene required for formation of specific tracheal branches including some segment-specific branches. The unp locus was identified on the basis of the segment-specific expression of β-galactosidase from a lacZ enhancer detector P element. The unp gene encodes a novel homeodomain protein required for the development of the segmentally reiterated ganglionic branches, which fail to penetrate the CNS in the absence of unp function. In addition, unp function is specifically required for development of the cerebral branch, a tracheal branch uniquely derived from the first thoracic segment. Our study suggests that Ultrabithorax (Ubx) and other homeotic genes restrict the formation of the cerebral branch to the first thoracic segment by repressing the expression of unp in more posterior segments; in addition we have identified an unusual 2.7 kb enhancer region downstream of the unp gene that reproduces unp gene expression, including homeotic gene regulation.
MATERIALS AND METHODS
Fly stocks
The fly lines 1912, f85 and E22 were kindly provided by C. Doe, A. Spradling (Karpen and Spradling, 1992) and C. Goodman respectively. The Ubx922 and Df(Ubx109) stocks were originally from E. Lewis (Lewis, 1978), the triple mutant stock (UbxMX12abd-AM1AbdBM8) was obtained from M. Bienz (Casanova et al., 1987; Bienz and Tremml, 1988), and the pnt9J allele was obtained from the Indiana Stock Center.
Isolation of unp genomic sequences and cDNAs
Genomic DNA flanking the 5′ end of the P element in f85 line was recovered by plasmid rescue (Mlodzik and Hiromi, 1992). The resulting DNA fragment was used to isolate two overlapping genomic clones carrying 28 kb of contiguous DNA. Various restriction fragments of the genomic DNA were digoxigenin-labeled and hybridized to whole embryos. The 5 kb EcoRI fragment (−2.5 to −7.5 in Fig. 1) gave an RNA expression pattern similar to that of β-galactosidase activity. This fragment was subsequently used to screen a cDNA library from 4to 8-hour embryos (Brown and Kafatos, 1988). Eight positive clones were recovered from screening about 106 colonies. The largest of these (punp-1), 1.7 kb in size, was chosen for sequence determination on both strands by the Sanger dideoxy method (Sanger et al., 1977) using Sequenase v2.0 (US Biochemicals).
Genomic map of the unplugged locus. Genomic DNA at the unplugged (unp) locus is represented by the thick line with the 0 coordinate (in kilobases) corresponding to the site of insertion for the P element f85. The insertion sites of the two other P elements, E22 and 1912, are also indicated. Only one relevant SpeI restriction site is shown (asterisk). The exons of unp are represented by the filled boxes, with the direction of transcription indicated by an arrow. The location of the 1360 element is indicated by an open box. The extent of DNA lesions for the four deletion mutations, which were obtained from imprecise excision of the 1912 P element, are represented by parentheses with the uncertainty of the boundaries indicated by the dotted lines. The six thin lines below the unp genomic map represent the DNA fragments used in reporter constructs to define enhancer region of the unp gene (see text). Abbreviations: H, HindIII; R, EcoRI; S, SacI; Sp, SpeI; Xb, XbaI; Xh, XhoI.
Genomic map of the unplugged locus. Genomic DNA at the unplugged (unp) locus is represented by the thick line with the 0 coordinate (in kilobases) corresponding to the site of insertion for the P element f85. The insertion sites of the two other P elements, E22 and 1912, are also indicated. Only one relevant SpeI restriction site is shown (asterisk). The exons of unp are represented by the filled boxes, with the direction of transcription indicated by an arrow. The location of the 1360 element is indicated by an open box. The extent of DNA lesions for the four deletion mutations, which were obtained from imprecise excision of the 1912 P element, are represented by parentheses with the uncertainty of the boundaries indicated by the dotted lines. The six thin lines below the unp genomic map represent the DNA fragments used in reporter constructs to define enhancer region of the unp gene (see text). Abbreviations: H, HindIII; R, EcoRI; S, SacI; Sp, SpeI; Xb, XbaI; Xh, XhoI.
Mapping and mutagenesis of the unp gene
Hybridization to polytene chromosome squashes was as described by Langer-Safer et al. (1982) using the alkaline phosphatase-based DNA detection system (GIBCO/BRL). A probe was generated from 1.7 kb unp cDNA using the random hexamer priming kit (Boehringer Mannheim) in the presence of biotinylated dUTP (ENZO Biochemicals). The unp sequences hybridized near polytene division 45C.
The precise locations of the three P element lines were determined by isolation of sequences flanking the f85, E22 and 1912 inserts using inverse PCR (Ochman et al., 1990). Briefly, two primers, GTATACTTCGGTAAGCTTCGGCTAT and CGAAATGCGTCGTTTA-
GAGCAGCAG, which hybridize to the 5′ and 3′ portions of a region near the 5′ end of the P element, were used as primers in a PCR reaction with 1.5 μg of DNA. Prior to PCR, this DNA was digested with Sau3A, ligated overnight in a 300 μl reaction at 4°C, and digested with AseI.
The P element insert in 1912 was mobilized by exposure to transposase from the P[Δ2-3 ry+] Sb chromosome. Excision events were identified by loss of the w+ marker present in the 1912 line. Twelve homozygous lethal lines were recovered from approximately 230 excision event; eleven excisions contain DNA lesions associated with large deletion beginning in the element and extending to the left or right as represented by the alleles unpr1 and unpr221 respectively in Fig. 1. One allele, unpr37, deleted about a 1.5 kb of sequence that extends from the 5′ end of the P element to a region just beyond the SpeI restriction site in the third exon. The extent of the deletions were inferred by Southern hybridization to DNA digested with restriction endonucleases from heterozygous unp individuals using various unp cDNA and genomic probes flanking the P element.
Analysis of unp RNA expression
For northern analysis, 5 μg of poly(A)+ RNA isolated from different developmental stage embryos was electrophoresed on 1.5% formaldehyde gels and transferred to nitrocellulose membranes (S&S or Costa). Filters were hybridized with radiolabeled unp cDNA and washed according to standard procedures (Sambrook et al., 1989).
Production of unp-specific antiserum
The unp-specific antibody was generated by immunization of mice with the recombinant protein containing a glutathione-S-transferase fusion to residues 26–476 of the unp protein. The induction and purification of the glutathione-S-transferase fusion protein were carried out according to the instructions provided by the pGex vector supplier (Amrad, Melbourne, Australia).
P element-mediated transformation
Genomic fragments (see Fig. 1) were cloned into the pCasperAUG-β-gal vector (Thummel et al. 1988) and injected into w1118 embryos as described (Rubin and Spradling, 1982).
In situ hybridization and immunohistochemistry
In situ hybridization to whole embryos was performed as described by Tautz and Pfeifle (1989). Probes were generated by random hexamer priming using digoxigenin-dUTP and detected with the Genius Kit (Boehringer Mannheim).
Embryos and larvae were fixed and immunostained as described (Patel, 1994), using a 1:300 dilution of secondary antibody conjugated to horseradish peroxidase (Jackson Laboratory). The signal for the antibody reaction was intensified with 0.03% NiCl2 except for double staining, where NiCl2 was omitted in the second reaction. Primary antibodies were used at the following dilution: mouse anti-β-galactosidase (Promega), 1:1000; anti-engrailed mAb 4D9, 1:3; mouse anti-unp (preabsorbed with embryos), 1:400. The stained embryos were cleared in 50% glycerol followed by 70% glycerol. In some cases, the embryos were flattened prior to photomicroscopy by dissection on a clean slide using a tungsten needle. Photography was with a Zeiss Axiophot microscope.
RESULTS
Isolation and organization of the unplugged gene
We identified the unplugged (unp) gene by systematic screening of P element enhancer detector lines for β-galactosidase expression patterns suggestive of regulation by the homeotic gene Ultrabithorax (Ubx). Line f85, which carries an insertion in the unp locus, was found to have β-galactosidase expression in a pattern restricted to the lateral ectoderm of the first thoracic segment, suggesting that Ubx may negatively regulate expression in more posterior segments. The f85 line also displayed a segmentally repeated pattern of expression in the neuroblasts and the neurons (see below).
To initiate the molecular characterization of unp, DNA flanking the 5′ end of the P element was recovered by plasmid rescue from the f85 line, and subsequently was used to isolate genomic DNA from a phage library (see Materials and Methods). The presence of the unp transcription unit in the isolated genomic DNA was confirmed by RNA in situ hybridization using various digoxigenin-labeled restriction fragments (Fig. 1). The 5 kb EcoRI fragment (Fig. 1; −2.5 to −7.5) that contains part of the unp transcription unit was used to isolate six cDNAs 1.7 kb in size, in reasonable agreement with a 1.8 transcript size as determined by northern blot hybridization (see below). Direct comparison between the unp cDNA and genomic sequences revealed that the transcription unit of unp is organized in 3 exons and separated by two introns with sizes 3.5 kb and 62 bp respectively (Fig. 1). Sequence analysis of the first intron of unp also revealed that it contains a transposable element; this 1360 element is moderately repetitive and has copy numbers ranging from 25 to 30 within the Drosophila genome (Kholodilov et al., 1988). Whether presence of the 1360 element in the unp locus has any functional significance is unknown.
The unp cDNA also permitted us to localize the unp gene to polytene chromosome 45C by in situ hybridization (data not shown). We have since obtained two additional enhancer detector lines with similar polytene localizations, 1912 and E22, and have mapped the P element insertion sites of all three lines by sequence analysis of rescued genomic DNA (see Materials and Methods). Interestingly, all three P elements are inserted within the 1360 element at intervals of 5 bp and 52 bp, respectively, between F85, E22 and 1912 (Fig. 1).
unp encodes a novel homeobox protein
The longest methionine-initiated open reading frame (ORF) within the unp cDNA extends from positions 126 to 1586 and has the capacity to encode a protein of 486 amino acid residues (Fig. 2). The sequence surrounding the first methionine in the ORF is a reasonable match with the Drosophila translational initiation consensus sequence C/A-A-A-A/C (Cavener and Ray, 1991), and this methionine codon is preceded by termination codons in all three frames. A stretch of 24 adenines at the 3′ end of the cDNA is preceded by a perfect match to the consensus polyadenylation signal AAUAAA (Proudfoot, 1991).
Sequence of the unp gene and its deduced protein product. The unp transcription unit includes two introns at the locations indicated by arrowheads within the open reading frame. The glutamineand proline-rich region at the amino terminus is indicated by the parentheses. Homeodomain residues near the carboxy terminal half of the protein are underlined. The consensus polyadenylation signal is represented by the bold letters. The accession number for unp sequence is U35427.
Sequence of the unp gene and its deduced protein product. The unp transcription unit includes two introns at the locations indicated by arrowheads within the open reading frame. The glutamineand proline-rich region at the amino terminus is indicated by the parentheses. Homeodomain residues near the carboxy terminal half of the protein are underlined. The consensus polyadenylation signal is represented by the bold letters. The accession number for unp sequence is U35427.
A protein data base search revealed that the unp ORF contains a homeodomain (residues 319-378; see Fig. 2) belonging to a family that includes several vertebrate homeobox genes (Fig. 3). The unp homeodomain shares amino acid identities ranging between 90 and 93% with homeodomains of the CHox7 gene in chicken (Fainsod and Greunbaum, 1989), the HOX7Q and GBX2 genes in human (Matsui et al., 1993), partial sequence of the MMoxA gene in mouse (Murtha et al., 1991), the XlHox7a and XlHox7b gene in Xenopus (King and Moore, 1994), the G9 gene in goldfish (Levine and Schechter, 1993), and the Hrox7 gene in abalone (Degnan and Morse, 1993). Little is known about the expression and function of these vertebrate homologs; however, the MMoxA and G9 genes were initially isolated from brain libraries and thus may be involved in brain development or function.
The unp homeodomain is a member of a divergent homeodomain family. Amino acid sequence of the unp homeodomain is aligned with those of the chicken CHox7 gene (Fainsod and Greunbaum, 1989), the human HOX7Q and GBX2 genes (Matsui et al., 1993), the mouse MMoxA (Murtha et al., 1991) and Dlx (Price et al., 1991) genes, the Xenopus XlHox7a and XlHox7b genes (King and Moore, 1994), the goldfish G9 gene (Levine and Schechter, 1993), the abalone Hrox7 gene (Degnan and Morse, 1993), the acscidian Ahox1 gene (Saiga et al., 1991) and the Drosophila Antp, pb, NK-1, lab, Dll, and en genes (Cribbs et al., 1992). A dash indicates residues identical to those in unp. Three regions corresponding to the α-helical regions from the crystallographically derived structure of the engrailed homeodomain are indicated above (Kissinger et al., 1990). The arrows denote the position of introns residing between residues 44 and 45 of the third helix or within the codon for amino acid 17 (en) of the first helix. The top alignment includes closely related homeodomains from a variety of species; the bottom alignment includes homeodomains whose coding sequences are interrupted by introns. Note that coding sequences for the en homeodomain are interrupted at a different location.
The unp homeodomain is a member of a divergent homeodomain family. Amino acid sequence of the unp homeodomain is aligned with those of the chicken CHox7 gene (Fainsod and Greunbaum, 1989), the human HOX7Q and GBX2 genes (Matsui et al., 1993), the mouse MMoxA (Murtha et al., 1991) and Dlx (Price et al., 1991) genes, the Xenopus XlHox7a and XlHox7b genes (King and Moore, 1994), the goldfish G9 gene (Levine and Schechter, 1993), the abalone Hrox7 gene (Degnan and Morse, 1993), the acscidian Ahox1 gene (Saiga et al., 1991) and the Drosophila Antp, pb, NK-1, lab, Dll, and en genes (Cribbs et al., 1992). A dash indicates residues identical to those in unp. Three regions corresponding to the α-helical regions from the crystallographically derived structure of the engrailed homeodomain are indicated above (Kissinger et al., 1990). The arrows denote the position of introns residing between residues 44 and 45 of the third helix or within the codon for amino acid 17 (en) of the first helix. The top alignment includes closely related homeodomains from a variety of species; the bottom alignment includes homeodomains whose coding sequences are interrupted by introns. Note that coding sequences for the en homeodomain are interrupted at a different location.
One of two introns in the unp transcription unit interrupts homeodomain coding sequences at a location first noted in labial-class homeobox genes (Fig. 2; Mlodzik et al., 1988; Diederich et al., 1989). This location, between Gln 44 and Val 45, is conserved for introns of many other homeodomain genes of different species (Fig. 3), but is distinct from the location of the intron that interrupts homeodomain coding sequences in engrailed and related genes.
Outside of the homeodomain, unp shares no significant homology with other known proteins in the data base. However, Pro/Gln-rich regions are found amino terminal to the homeodomain, with one particular region from residues 111 to 142 comprising 52% proline and glutamine residues (Fig. 2). Pro/Gln-rich regions are found in many proteins that are capable of transcriptional activation (for review see Mitchell and Tjian, 1989; Gerber et al., 1994).
Embryonic expression of unp
Northern hybridization using unp cDNA as a probe detected a 1.8 kb transcript size that correlates reasonably well with the size of the cDNA (Fig. 4). The unp transcript is first detectable in 4to 8-hour embryos and expression is maintained throughout embryogenesis. Following a transient decrease in transcript levels during larval periods, levels increase during the pupal and adult stages. In addition to the expected transcript, a 2.8 kb band is also observed at the third instar larval stage. We cannot distinguish between the possibilities that this novel band represents a product of alternative splicing or a crosshybridizing transcript.
Temporal profile of the unp transcript. Each lane contains 5 μg of poly(A)+ RNA derived from several embryonic stages (stage given in hours after egg-laying), in two larval stages, and in the pupal and adult stages, as indicated. A single 1.8 kb band was detected throughout development except in the 0to 4-hour stage, with peak expression observed in 4to 8-hour embryos. Lower-level expression is detected during late embryonic to larval stages. A 2.5 kb band is also detected during third instar larval period (open triangle). The lower panel shows the same blot subsequently hybridized with the Drosophila rp49 gene.
Temporal profile of the unp transcript. Each lane contains 5 μg of poly(A)+ RNA derived from several embryonic stages (stage given in hours after egg-laying), in two larval stages, and in the pupal and adult stages, as indicated. A single 1.8 kb band was detected throughout development except in the 0to 4-hour stage, with peak expression observed in 4to 8-hour embryos. Lower-level expression is detected during late embryonic to larval stages. A 2.5 kb band is also detected during third instar larval period (open triangle). The lower panel shows the same blot subsequently hybridized with the Drosophila rp49 gene.
To learn more about spatial expression of unp, a digoxigenin-labeled cDNA probe was used to perform whole-mount in situ hybridization. The unp expression pattern was also analyzed with transgenic embryos carrying unp enhancer/reporter constructs (see below) and by immunolocalization using unp-specific antiserum. All three expression patterns gave indistinguishable results, although expression in the CNS is relatively weaker in the transgenic embryos.
unp expression first appears at stage 8 (3–3.5 hours of development; Campos-Ortega and Hartenstein, 1985) in the midline of the central nervous system (CNS), at a location corresponding to a subset of neuroblasts (Fig. 5A). These neuroblasts divide during germband extension to generate sibling neuroblasts and neurons that largely correspond to engrailedexpressing cells within the CNS (Fig. 5B-F; Doe 1992). As the germband retracts, midline CNS expression begins to fade (Fig. 5I and J), and by stage 14 the CNS expression is restricted to a few cells in each segment (Fig. 5M-O).
Embyonic patterns of unp gene expression. The localization of unp transcripts at various stages was visualized by in situ hybridization to whole embryos using digoxigenin-labeled unp cDNA as a probe, except for F, J, L and N, which show embryos immunostained with an antibody against unp protein (F,J) or against β-galactosidase (L,N; embryos are carrying the unpU6-lacZ reporter construct described in the text). (A) Lateral view of a stage 8 embryo showing the expression of unp in midline neuroblasts. (B,C) Lateral and ventral views of a stage 9 embryos. Arrowheads mark the two groups of unp-expressing ectodermal cells in the labial and first thoracic segments. (D) Lateral view of a stage 10 embryo. Note the increase in the expression in T1 and a corresponding loss of expression in the labial segment. (E,F) Ventral views of stage 11 embryos showing that the distributions of unp RNA (E) and protein (F) are identical. Note that lateral expression at this stage is restricted to cells surrounding the anterior half of the first tracheal pit (arrowhead). (G,H) Ventral and ventrolateral (H) views of stage 12 embryos during germband retraction. The lateral cells in T1 begin to migrate anteriorly and dorsally (arrowheads). A new ventral expression domain that flanks the CNS is indicated by arrows. (I-J) Ventral view of stage 13 embryos showing that the RNA expression pattern (I) is identical to the protein distribution pattern (J). Note the extension of the ventral cells toward the CNS (arrows). (K) Dorsolateral view of a stage 13 embryo illustrating the migration of lateral cells from both sides to the dorsal midline (arrowheads; embryos shown with dorsal side down).(L) Higher magnification view of the dorsal cells originating from both lateral T1 regions (arrowheads). (M,N) Ventral views of stage 14 embryos. The 7–9 ventral cells extending into the CNS are clearly visible. (O) Ventral view of a stage 16 embryo showing persistent expression of unp in the CNS. All embryos are oriented with anterior to the left and dorsal up (except K).
Embyonic patterns of unp gene expression. The localization of unp transcripts at various stages was visualized by in situ hybridization to whole embryos using digoxigenin-labeled unp cDNA as a probe, except for F, J, L and N, which show embryos immunostained with an antibody against unp protein (F,J) or against β-galactosidase (L,N; embryos are carrying the unpU6-lacZ reporter construct described in the text). (A) Lateral view of a stage 8 embryo showing the expression of unp in midline neuroblasts. (B,C) Lateral and ventral views of a stage 9 embryos. Arrowheads mark the two groups of unp-expressing ectodermal cells in the labial and first thoracic segments. (D) Lateral view of a stage 10 embryo. Note the increase in the expression in T1 and a corresponding loss of expression in the labial segment. (E,F) Ventral views of stage 11 embryos showing that the distributions of unp RNA (E) and protein (F) are identical. Note that lateral expression at this stage is restricted to cells surrounding the anterior half of the first tracheal pit (arrowhead). (G,H) Ventral and ventrolateral (H) views of stage 12 embryos during germband retraction. The lateral cells in T1 begin to migrate anteriorly and dorsally (arrowheads). A new ventral expression domain that flanks the CNS is indicated by arrows. (I-J) Ventral view of stage 13 embryos showing that the RNA expression pattern (I) is identical to the protein distribution pattern (J). Note the extension of the ventral cells toward the CNS (arrows). (K) Dorsolateral view of a stage 13 embryo illustrating the migration of lateral cells from both sides to the dorsal midline (arrowheads; embryos shown with dorsal side down).(L) Higher magnification view of the dorsal cells originating from both lateral T1 regions (arrowheads). (M,N) Ventral views of stage 14 embryos. The 7–9 ventral cells extending into the CNS are clearly visible. (O) Ventral view of a stage 16 embryo showing persistent expression of unp in the CNS. All embryos are oriented with anterior to the left and dorsal up (except K).
Outside the CNS, unp expression is first observed in two clusters of ectodermal cells located laterally within the labial and first thoracic (T1) segments of stage 9 embryos (Fig. 5B). During germband extension unp expression continues in T1 and rapidly diminishes in the labial segment (Fig. 5C,D). By stage 11, the lateral cells are recognizable as 15–20 unp-expressing cells around the anterior part of the first tracheal pit (Fig. 5E,F). As the germband retracts, these cells begin to migrate anterodorsally with expression restricted to 5-6 cells (Fig. 5G,H). By stage 13, the expression is detected in a few cells close to the dorsal midline of the embryos (Fig. 5K,L); these cells appear to form long cytoplasmic connections that prefigure the cerebral branches of the tracheal system (Fig. 5L; see below).
As the germband retracts a new expression domain within the invaginated tracheal pits appears on each side of the CNS in segments T1-A7 (open triangles in Fig. 5G,H). Expression in this domain is restricted to a few cells per hemisegment, which may represent the precursors of the ganglionic branches of the tracheal system (see below). During germband retraction, these precursor cells extend ventrally and dorsally (Fig. 5I-K). By stage 14, the ganglionic branch in each hemisegment consists of 7–9 unp-expressing cells whose cell bodies appear to form a continuous chain that penetrates the CNS of stage 14 embryos (Fig. 5M,N). No RNA or protein expression of unp outside the CNS can be detected in later stage embryos (Fig. 5O).
unp expression marks two neural branching patterns of the tracheal system
To determine the tissue types of cells expressing unp outside the CNS, we have performed double labeling experiments using unp-specific antiserum and other antibodies that recognize different tissue types in the embryos. A β3-tubulin antibody (Kimble et al., 1990) as well as the 22C10 antibody (Zipursky et al., 1984), which highlight muscle and peripheral nervous system (PNS) cells respectively, fail to co-localize with unp-specific antiserum, indicating that cells expressing unp are not muscle or PNS (data not shown).
The elongated morphology of unp-expressing cells resembles the morphology of cells in the developing tracheal system. Indeed, double-labelling with unp-specific antiserum and 2A12, a monoclonal antibody that specifically highlights the lumen of the tracheal system, demonstrates that most unpexpressing cells outside the CNS also express the 2A12 antigen (Fig. 6). On the ventrolateral side of each hemisegment, the unp protein accumulates in the nuclei of 7–9 cells overlapping with the 2A12 antigen in the ganglionic and lateral branches of the tracheal system (Fig. 6A and B). The organization of ganglionic branches differs between thoracic and abdominal segments, and this difference is reflected by the unp expression pattern (Fig. 6A).
Expression of unp in cells of the developing cerebral and ganglionic tracheal branches. Embryos were double-stained with antibodies against unp (dark purple product) and the 2A12 monoclonal antibody that stains the lumen of the tracheae (brown product).(A) Ventral view of a stage 14 embryo. The ventral cells that accumulate unp protein (nuclear stain) overlap with the lumen of the ganglionic branches (brown), as indicated by the arrowheads.(B) Higher magnification view of a dissected embryo illustrates the unp protein accumulation in cells of the ganglionic branches that extend into the CNS (arrowhead) and in cells of the lateral branches (black arrow). (C) Dorsal view of a stage 13 embryo. The overlapping distribution of the unp protein (purple) and 2A12 antigen (brown) in the developing cerebral branch is indicated by a white arrow. (D) Dorsal view of a stage 14 embryo. The unp protein accumulates in 4-5 nuclei within the caudal end of the cerebral branch (white arrow).
Expression of unp in cells of the developing cerebral and ganglionic tracheal branches. Embryos were double-stained with antibodies against unp (dark purple product) and the 2A12 monoclonal antibody that stains the lumen of the tracheae (brown product).(A) Ventral view of a stage 14 embryo. The ventral cells that accumulate unp protein (nuclear stain) overlap with the lumen of the ganglionic branches (brown), as indicated by the arrowheads.(B) Higher magnification view of a dissected embryo illustrates the unp protein accumulation in cells of the ganglionic branches that extend into the CNS (arrowhead) and in cells of the lateral branches (black arrow). (C) Dorsal view of a stage 13 embryo. The overlapping distribution of the unp protein (purple) and 2A12 antigen (brown) in the developing cerebral branch is indicated by a white arrow. (D) Dorsal view of a stage 14 embryo. The unp protein accumulates in 4-5 nuclei within the caudal end of the cerebral branch (white arrow).
On the dorsal side of stage 13 embryos, unp protein accumulates in 5–6 nuclei overlapping with 2A12 antigen in the cerebral branch of the first tracheal metamere (Fig. 6C; Manning and Krasnow, 1993). By stage 14, the cerebral branch courses posteriorly and medially such that it lies close to the dorsal midline of T2 (Fig. 6D). Thus, unp expression outside of the CNS is restricted to cells of the cerebral and ganglionic branches of the tracheal system during embryonic development.
unp is required for the development of the tracheal branches
To study the function of the unp gene, the 1912 line carrying a P element insertion in the first intron of unp was exposed to transposase from the P[Δ2–3] element to generate mutations for phenotypic analysis. Of approximately 230 excision events, 12 were associated with homozygous lethality; the extents of chromosomal deletions in four of these excisants are as shown in Fig. 1. Since lethal alleles unpr1, unpr221 and unpr225 have DNA lesions extending beyond the unp transcription unit, we concentrated our phenotypic analysis on the unpr37 allele. The DNA lesion associated with this allele begins in the 5′ end of the P element and extends to the region close to a SpeI restriction site in the third exon (see Materials and Methods). Thus, the unpr37 deletion removes all of exon 2 and part of exon 3, including the entire homeodomain sequence. Interestingly, the mutation still retains lacZ expression in embryos, consistent with our findings that the major regulatory sequences for unp expression are located downstream of the unp transcription unit (see below).
Specific expression of unp in neural branches of the tracheal system suggests that it may play a role in tracheal development. Indeed, tracheal staining of unpr37 homozygous mutant embryos (identified by the absence of a marked balancer chromosome; see Materials and Methods) with antibody 2A12 revealed the absence of the entire cerebral branch (Fig. 7A,B), with occasional ectopic branches in the first tracheal metamere (Fig. 7C,D; arrowhead). In addition, the cerebral anastomosis, which normally is associated with the cerebral branch is also absent (compare Fig. 7E,F). A specific defect is also observed in the ganglionic branches, which in most cases extend only partially and fail to penetrate the CNS (compare Fig. 7G and H). Similar effects on the cerebral branch and anastomosis and on ganglionic branches were observed with the unpr225 and unpr1 alleles (data not shown). The specific defects observed in the unp mutants are consistent with the unp protein distribution and suggest a specific role for unp in the formation of tracheal branches that penetrate the CNS. Despite these tracheal defects, about 3-5% of homozygous unpr37 flies, under uncrowded culture conditions, eclose to adulthood; these escapers exhibit an upheld wing phenotype (data not shown).
unp controls the development of specific tracheal branching patterns. The tracheal branching patterns in all the panels were visualized by 2A12 antibody staining of stage 15 embryos. A-F are dorsal views and G-H are ventral views. The wild-type embryo in A is photographed in a focal plane that illustrates the normal morphology of the dorsal cephalic branch, dorsal branch 1, and the cerebral branch. Note the absence in unp mutant embryos of the cerebral branch (B, asterisk) with frequent occurrence of ectopic dorsal branch-like (C; arrowhead) and dorsal cephalic-like (D; arrowhead) tracheal outgrowths. In a more ventral plane of focus, the normal mophology of the cerebral anastomosis and pharyngeal branch are apparent (E); the cerebral anastomosis is missing in unp mutant embryos (F, asterisk). The ganglionic branches normally extend into the CNS in a stereotyped pattern (G), and this pattern is altered in unp mutant embryos (H), with branches stalling as they reach the CNS. Dotted lines indicate the midline of the CNS. Abreviations: CB, cerebral branch; DB1, dorsal branch 1; DC, dorsal cephalic branch; GB, ganglionic branch; PB, pharyngeal branch.
unp controls the development of specific tracheal branching patterns. The tracheal branching patterns in all the panels were visualized by 2A12 antibody staining of stage 15 embryos. A-F are dorsal views and G-H are ventral views. The wild-type embryo in A is photographed in a focal plane that illustrates the normal morphology of the dorsal cephalic branch, dorsal branch 1, and the cerebral branch. Note the absence in unp mutant embryos of the cerebral branch (B, asterisk) with frequent occurrence of ectopic dorsal branch-like (C; arrowhead) and dorsal cephalic-like (D; arrowhead) tracheal outgrowths. In a more ventral plane of focus, the normal mophology of the cerebral anastomosis and pharyngeal branch are apparent (E); the cerebral anastomosis is missing in unp mutant embryos (F, asterisk). The ganglionic branches normally extend into the CNS in a stereotyped pattern (G), and this pattern is altered in unp mutant embryos (H), with branches stalling as they reach the CNS. Dotted lines indicate the midline of the CNS. Abreviations: CB, cerebral branch; DB1, dorsal branch 1; DC, dorsal cephalic branch; GB, ganglionic branch; PB, pharyngeal branch.
Regulation of unp expression by genes of the bithorax complex
Lewis (1978) showed that appropriate development of segment-specific tracheal structures requires the function of the homeotic genes of the bithorax complex (BX-C). As described above, unp expression highlights portions of the tracheal system that penetrate the CNS, including the cerebral branch specific to T1 (Fig. 8A). To test the possibility that genes in the BX-C play a role in regulating unp expression, we examined the distribution of unp transcript in Ubx9.22, in Df109 (lacking Ubx and abd-A functions; Karch et al., 1985), and in UbxMx12abd-AM1Abd-BM8 triple mutants (lacking function of all three BX-C genes; Casanova et al., 1987). In Ubx mutant embryos additional unp expression is observed in cells surrounding the tracheal pits of T2 and T3, indicative of a role for Ubx in repression of unp in the posterior segments and consistent with homeotic transformation in Ubx mutants of posterior T2 and T3 towards a T1 identity (Fig. 8B). In Df109 embryos extra patches of unp-expressing cells around the tracheal pits extend posteriorly to A7, indicating a role of abdA in the repression of unp expression in the abdominal segments (data not shown). The homeotic gene abd-B probably contributes to the repression of unp expression in A7, since slightly elevated expression in A7 is observed in the triple mutants as compared to the Df109 embryos (Fig. 8C).
Regulation of unp expression by genes of the bithorax complex. Lateral views of unp expression at the retracted germband are shown in the following genotypes: wild type (A); Ubx (B); and Ubx abd-A abdB (C). Note the appearance in bithorax complex mutants of additional posterior patches of augmented expression.
Regulation of unp expression by genes of the bithorax complex. Lateral views of unp expression at the retracted germband are shown in the following genotypes: wild type (A); Ubx (B); and Ubx abd-A abdB (C). Note the appearance in bithorax complex mutants of additional posterior patches of augmented expression.
3′ region of unp confers Ubx-mediated repression of a reporter gene
To identify cis-regulatory regions that are responsible for the normal unp expression in embryos, six restriction fragments encompassing 20 kb of unp genomic sequence were tested for their ability to direct expression of a lacZ reporter gene containing a minimal hsp70 promoter (Thummel et al., 1988). The region far upstream of the unp transcription unit, represented by fragment U1 in Fig. 1, gave a restricted lacZ expression pattern in the dorsal vessel (data not shown), although this pattern does not correspond to any known unp product distribution. The U2 and U3 fragments that are located immediately downstream of the U1 fragment did not generate any detectable lacZ expression, and the promoter proximal fragment U4 (see Fig. 1 for location) gave weak but nonspecific lacZ expression in embryos (data not shown), suggesting that a general enhancer-like element may be present in the fragment. Finally, the U5 fragment in the first intron does not show any detectable lacZ expression.
The only region capable of driving a lacZ reporter gene expression similar to the unp protein distribution is located at the 3′ end of the unp transcription unit. In 10 out 10 independent transgenic lines, the 2.7 kb fragment (U6 in Fig. 1) gave characteristic unp-like expression in the CNS and the cerebral and ganglionic branches of the tracheal system during embryogenesis (see Fig. 5L,N). This fragment also gave unp-like expression in the cells around the first tracheal pit at the germband extended stage (Fig. 9A). Thus, the results indicate that a 2.7 kb fragment of unp 3′ flanking sequence contains most of the cis-regulatory elements for the normal unp expression in embryos.
A downstream regulatory region of the unp locus reproduces normal expression, including homeotic regulation. A-D show the expression of a lacZ reporter (unpU6-lacZ) carrying the U6 segment downstream of the unp locus in wild-type (A,C) and Ubx mutant embryos (B,D). β-galactosidase antibody staining patterns are shown in embryos during extended (A,B) and retracted (C,D) germband strages. Open triangles in the Ubx mutant embryos indicate additional patches of expression resulting from the appearance of ectopic cerebral branch (CB) precursor cells. E and F display the 2A12 antibody staining pattern in wild-type (E) and Ubx mutant (F) embryos at stage 14. Note the presence of ectopic cerebral branchlike structures (arrowheads) originating from T2 and T3 (asterisks) in the Ubx mutant embryos (F). Note also the the absence of dorsal trunk segments (DT) projecting anteriorly at T2 and T3, consistent with a general transformation of tracheal branches at T2 and T3 toward a T1 identity.
A downstream regulatory region of the unp locus reproduces normal expression, including homeotic regulation. A-D show the expression of a lacZ reporter (unpU6-lacZ) carrying the U6 segment downstream of the unp locus in wild-type (A,C) and Ubx mutant embryos (B,D). β-galactosidase antibody staining patterns are shown in embryos during extended (A,B) and retracted (C,D) germband strages. Open triangles in the Ubx mutant embryos indicate additional patches of expression resulting from the appearance of ectopic cerebral branch (CB) precursor cells. E and F display the 2A12 antibody staining pattern in wild-type (E) and Ubx mutant (F) embryos at stage 14. Note the presence of ectopic cerebral branchlike structures (arrowheads) originating from T2 and T3 (asterisks) in the Ubx mutant embryos (F). Note also the the absence of dorsal trunk segments (DT) projecting anteriorly at T2 and T3, consistent with a general transformation of tracheal branches at T2 and T3 toward a T1 identity.
To ascertain that the 2.7 kb fragment contains regulatory elements that can mediate homeotic control of gene expression, we examined expression of the U6-lacZ construct in Ubx9.22 mutant embryos and found that the lacZ reporter gene is derepressed in the T2 and T3 segments in a manner similar to the endogenous gene (Fig. 9B). Expression in T2 and T3 continues as the cells extend to the dorsal side of the embryo (Fig. 9D). We note that the cells at T2 and especially at T3 do not extend as far dorsally as the cells at T1. The ultimate fate of cells in T2 and T3 in Ubx mutant embryos is uncertain since the expression diminishes in later stages; these cells nevertheless appear committed to cerebral branch fates, since ectopic cerebral-like branches in T2 and T3 can be detected with tracheal-specific 2A12 antibody in Ubx mutant embryos (Fig. 9F).
The pointed gene acts upstream of unp in regulating branching morphogenesis
The specific branching defects observed in unp mutant embryos prompted us to examine the relationship between unp and pointed (pnt), a gene encoding a member of the ets family of transcription factors; pnt is expressed in tracheal placodes and in the developing tracheal branches (Klambt, 1993). In the absence of pnt gene function, tracheal cells fail to migrate and branches do not extend to target tissues (Klambt et al., 1992). In particular, stalling of the ganglionic branches at the ventral oblique musculature in hypomorphic pnt embryos is reminiscent of the most extreme phenotype observed in unp embryos.
To determine the regulatory relationship between pnt and unp genes we immunostained pnt homozygous embryos with unp-specific antibody to follow the fate of the ganglionic branches. At stage 12.5, ganglionic branch precursor cells migrating toward the CNS are clearly visible in wild-type embryos (Fig. 10A). In pnt mutant embryos no migratory cells that are expressing unp protein can be detected (Fig. 10B); however, a few precursor cells accumulating low levels of unp protein occasionally can be identified (Fig. 10B; asterisk). By stage 14, the ganglionic branches of normal embryos are well developed, as is evident from a group of 8 to 9 unp-expressing cells along the ventrolateral region of each hemisegment (Fig. 10C; also see Fig. 5). In pnt mutant embryos, only 3 to 4 unp-expressing cells can be detected (Fig. 10D); these unpexpressing cells are clustered in a group suggesting that the precursor cells remain immobile and fail to extend from the tracheal pits. These results suggest that the ganglionic branch phenotype in pointed embryos may be in part due to a loss or reduction of unp gene expression and failure of unp-expressing cells to extend into the CNS.
Alteration of unp protein distribution in pnt mutant embryos. Wild-type (A,C) or pnt mutant (B,D) embryos were immunostained with unpspecific antiserum. During germband retraction, the unp protein normally accumulates in cells of the ganglionic branches (A; arrowheads); in pnt mutant embryos, however, the unp protein levels are barely detectable (B; asterisks). After germband retraction, each ganglionic branch consists of 7–9 cells that traverse the ventral oblique muscles and penetrate into the CNS (C; arrows). In pnt mutant embryos, only 4–5 precursor cells can be observed to accumulate low levels of unp protein (D; arrowheads), and these cells fail to extend into the CNS. Note the abnormal pattern ofunp expression in the CNS of pnt mutants (B,D), consistent with other CNS defects such as collapsed commissures seen in pnt mutant embryos. The lines in C and D indicate the midline of CNS.
Alteration of unp protein distribution in pnt mutant embryos. Wild-type (A,C) or pnt mutant (B,D) embryos were immunostained with unpspecific antiserum. During germband retraction, the unp protein normally accumulates in cells of the ganglionic branches (A; arrowheads); in pnt mutant embryos, however, the unp protein levels are barely detectable (B; asterisks). After germband retraction, each ganglionic branch consists of 7–9 cells that traverse the ventral oblique muscles and penetrate into the CNS (C; arrows). In pnt mutant embryos, only 4–5 precursor cells can be observed to accumulate low levels of unp protein (D; arrowheads), and these cells fail to extend into the CNS. Note the abnormal pattern ofunp expression in the CNS of pnt mutants (B,D), consistent with other CNS defects such as collapsed commissures seen in pnt mutant embryos. The lines in C and D indicate the midline of CNS.
DISCUSSION
The role of unp in development of the tracheal branches
The Drosophila tracheal system is organized in a network of tubular branches that meet the respiratory needs of appropriate target tissues. The branches of the tracheal system originate from ten pairs of tracheal placodes located on the lateral side of segments T2 through A8 of stage 10 embryos (Hartenstein and Jan, 1992; Manning and Krasnow, 1993). During development, founder cells of the tracheal placodes extend and migrate via stereotyped paths to establish specialized tracheal branching patterns. Cells of the cerebral and ganglionic branches that specifically accumulate unp protein are the major tracheal cells known to target the brain and ventral nerve cord of the CNS (Manning and Krasnow, 1993). The absence of the cerebral branch and associated cerebral anastomosis and the abnormal development of the ganglionic branches in unp mutant embryos indicate that unp plays an important role in the formation of all the major neural branches of the tracheal system.
The expression of unp in founder cells of the cerebral branch within the first tracheal placode suggests an early role during branch development. unp function, however, is most likely not involved in the initial commitment to founder cell fates, since expression of the lacZ reporter gene by the enhancer trap is maintained in unp mutant embryos (data not shown). Instead unp appears to be involved in branching morphogenesis by regulating cell migration or extension, and in the absence of such function the founder cells either die or adopt other branch patterns. This is consistent with the observation that in unp mutant embryos the absence of the cerebral branch is occasionally accompanied by the presence of an ectopic branch in the first tracheal metamere. The appearance of this ectopic branch resembles that of the dorsal branch or of the dorsal cephalic branch, which like the cerebral branch originate from the first tracheal placode (Manning and Krasnow, 1993).
In addition to the cerebral branch, unp is also expressed in cells of the ganglionic branches, but here expression occurs much later (stage 9 vs stage 12) suggesting that unp may have secondary functions during ganglionic branch development. Consistent with this view is the observation that the ganglionic branches develop but fail to extend consistently to the CNS in unp mutant embryos (see Fig. 7F). This phenotype is reminiscent of the hypormorphic alleles of pnt and breathless mutants (Klambt et al., 1992). breathless encodes a Drosophila homologue of the fibroblast growth factor (FGF) receptor, and its expression in the developing tracheal system is required for the migration of tracheal cells. Thus, the observed unp phenotype appears to be consistent with the role of unp in the specification of tracheal cell migration or extension.
Regulation of tracheal branch development
A limited number of target genes has been characterized whose functions correlate with HOM gene regulation in Drosophila. These genes, initially identified either by their mutant phenotypes or homology to other gene families, encode distinct proteins whose functions are associated with development of many different structures. For example, repression of the homeobox gene Distal-less and activation of the helix-loophelix containing transcription factor nautilus in abdominal segments by genes of the bithorax-complex (BX-C) are required for proper development of limb and somatic muscle respectively (Paterson et al., 1991; Vachon et al., 1992; Michelson, 1994); activation of another homeobox gene empty spiracles in the 8th abdominal segment by Abdominal-B is required for development of the fillzkorper (Jones and McGinnis, 1993); and activation of the Drosophila TGF-β homologue decapentaplegic by Ultrabithorax (Ubx) is required for gut morphogenesis (Immergluck et al., 1990; Panganiban et al., 1990; Reuter et al., 1990; Capovilla et al., 1993; Hursh et al., 1993; Sun et al., 1995). Other targets of Ubx have also been isolated by either immunoprecipitation of protein-DNA complexes from native chromatin (Gould et al., 1990) or UV-crosslinked chromatin (Graba et al., 1992) using a Ubx-specific monoclonal antibody. Although these methods suggest direct interaction between Ubx protein and the cognate target sequences, it is difficult to determine the significance of the Ubx regulation due to the absence of obvious mutant phenotypes associated with these target genes (Brookman et al., 1992; Gould and White, 1992; Nose et al., 1992; Strutt and White, 1994).
Our results indicate that restricted expression of unp in the cerebral branch founder cells requires normal function of genes in the BX-C. In the absence of these homeotic genes, the expression of unp expands more posteriorly to the abdominal segments. This is consistent with the notion that Ubx controls tracheal development by regulating the expression of target genes (Lewis, 1978). Since unp encodes a transcription factor and is required for cerebral branch development, we suggest that normal restriction of cerebral branch development to T1 is mediated by Ubx repression of unp. This repression is mediated by the 2.7 kb fragment located downstream of the unp transcription unit. Further dissection of this regulatory region will be necessary to establish whether repression by Ubx protein is direct or mediated by another transcription factor.
As mentioned above unp does not appear to be required for initial commitment to a tracheal branch cell fate. Thus, initiation of unp expression must be mediated by other transcription factors. There are three observations to suggest that pnt is involved in regulating unp gene expression in the ganglionic branches. First, the branching defects observed in pnt embryos are more severe than those of unp embryos. Second, pnt is expressed in the tracheal founder cells including the cells that later give rise to the ganglionic branches, while unp expression in the ganglionic branches does not occur until founder cells have been established. pnt expression thus clearly precedes unp expression. Third, the unp protein levels and the number of unpexpressing cells in the ganglionic branches are greatly reduced in pnt mutant embryos. These observations, however, do not indicate that the pnt protein is the only regulator of unp gene expression, nor even that this regulation is direct. Indeed, several other transcription factors known to be expressed during early tracheal development are candidates for regulators of unp gene expression, although their precise role in regulating branching morphogenesis is not clear, since their complex embryonic expression patterns obscure the determination of primary versus secondary effects (Manning and Krasnow, 1993).
In conclusion our studies demonstrate that unp is essential for normal development of the tracheal branches that penetrate the central nervous system, possibly by regulating genes involved in tracheal cell migration and extension. Regulation of the unp gene requires normal function of the pnt gene in a general manner, whereas segmentally specific aspects of unp expression are established through repression by the homeotic genes of the bithorax complex. Recent elucidation of FGF receptor function in airway branching of mouse lung buds resembles that of Drosophila FGF receptor in tracheal formation, suggesting the existence of parallel molecular pathways that operate in morphogenesis of branched tubular structures in insects and mammals (Peters et al., 1994; Klambt et al., 1992). It would be interesting to know whether this parallel extends to the control of specific branch formation by homeodomain proteins, as we have found for neural branching of tracheae in Drosophila.
ACKNOWLEDGEMENTS
We thank D. von Kessler for assistance in initial screening of enhancer detector lines, J. Lee for the initial plasmid rescue and the isolation of unp genomic phages, and Dennis Wilson for generating the unp-specific antibody. We also thank C. Doe for the 1912 line, C. Goodman for the E22 line, the Indiana Stock Center for the pnt9J mutant, B. Dattman and E. Raff for the β-tubulin antibody and N. Patel for the 2A12 and the 4D9 antibodies. We are also grateful to M. Krasnow for a pre-publication copy of the Manning and Krasnow review on Drosophila tracheal development. P. A . B. is an investigator of the Howard Hughes Medical Institute.