|Home | About | Journals | Submit | Contact Us | Français|
The Drosophila trachea is a branched tubular epithelia that transports oxygen and other gases. trachealess (trh), which encodes a bHLH-PAS transcription factor, is among the first genes to be expressed in the cells that will form the trachea. In the absence of trh, tracheal cells fail to invaginate to form tubes and remain on the embryo surface. Expression of many tracheal-specific genes depends on trh, but all of the known targets have relatively minor phenotypes compared to loss of trh, suggesting that there are additional targets. To identify uncharacterized transcriptional targets of Trh and to further understand the role of Trh in embryonic tracheal formation, we performed an in situ hybridization screen using a library of ~100 tracheal-expressed genes identified by the Berkeley Drosophila Genome Project (BDGP). Surprisingly, expression of every tracheal gene we tested was dependent on Trh, suggesting a major role for Trh in activation and maintenance of tracheal gene expression. A re-examination of the interdependence of the known early-expressed transcription factors, including trh, ventral veinless (vvl) and knirps/knirps-related (kni/knrl), suggests a new model for how gene expression is controlled in the trachea, with trh regulating expression of vvl and kni, but not vice versa. A pilot screen for the targets of Vvl and Kni/Knrl revealed that Vvl and Kni have only minor roles compared to Trh. Finally, genome-wide microarray experiments identified additional Trh targets and revealed that a variety of biological processes are affected by the loss of trh.
Tubular organs are required in multicellular organisms to transport essential substances into and out of the body and from one part of the body to another. The Drosophila trachea is a branched network of tubular epithelia that transports oxygen and other gases, and is an excellent model for discovering the cellular and molecular events underlying epithelial tube formation. The trachea develops from twenty placodes, or plates, of polarized ectodermal cells, ten on each side of the embryo from the second thoracic segment (T2) to the eighth abdominal segment (A8). Through a series of coordinated cell shape changes, tracheal cells invaginate to form small, internalized epithelial sacs, undergoing their final round of mitotic division during this process. Once the tracheal cells are internalized, small groups of cells in distinct positions within each tracheal sac begin to migrate in stereotypical patterns to form the primary branches. Dorsally-positioned branches migrate anteriorly and posteriorly and fuse to the equivalent branches in adjacent segments to create a continuous dorsal trunk that runs along the anterior-posterior axis of the embryo. The remaining branches either extend to target organs or fuse with either adjacent or contra-lateral tracheal branches to ultimately form a contiguous tubular network (Manning and Krasnow, 1993).
Much is known about specification of the Drosophila trachea. The earliest events are controlled by global patterning genes, including the dorsal-ventral, homeotic, and segment polarity genes (Kerman et al., 2006). Patterning genes both limit the number of segments in which trachea form and determine the anterior-posterior and dorsal-ventral location of the primordia within each segment. spalt (sal), which encodes a zinc finger transcription factor initially expressed in broad domains in both anterior and posterior regions of the embryo, limits tracheal formation to the ten segments from T2 through A8 (Kühnlein and Schuh, 1996). The dorsal limit on trachea formation is set by the dorsal patterning gene decapentaplegic (dpp) (Isaac and Andrew, 1996; Wilk et al., 1996). The segment polarity gene wingless (wg) limits formation of tracheal primordia to the cells between its stripes of expression within each segment (de Celis et al., 1995; Wilk et al., 1996), whereas STAT92E, the single Drosophila STAT, which also functions as a segment polarity gene in early development, is essential for tracheal formation (Brown et al., 2001; Li et al., 2003; Sotillos et al., 2010).
Three known transcription factors are specifically expressed early in the cells that will develop as trachea: trachealess (trh), ventral veinless (vvl) and knirps/knirps-related (kni/knrl). trh encodes a basic helix-loop-helix (bHLH)-PAS transcription factor, which is essential for trachea formation; in the absence of trh function, the precursor cells fail to undergo any of the morphogenetic events of tube formation and remain at their site of origin (Isaac and Andrew, 1996; Wilk et al., 1996). Interestingly, its mammalian orthologue NPAS-3 is required for lung development and homeostasis (Zhou et al., 2009). vvl encodes a POU-domain containing transcription factor also required for tracheal development; in the absence of vvl, tracheal cells invaginate but fail to migrate appropriately (Anderson et al., 1995; de Celis et al., 1995). knirps (kni) and knirps-related (knrl) are two closely linked homologous genes that encode zinc finger proteins (Chen et al., 1998). Both genes are expressed throughout the early tracheal primordia, but their expression becomes restricted to specific branches at later stages. Loss of kni and knrl disrupts migration of a subset of tracheal branches within each segment (Chen et al., 1998).
Previous studies have revealed that the initial expression of vvl and trh is independent of each other, and that early tracheal expression of kni is also independent of trh (Boube et al., 2000). Indeed, more recent findings reveal that the initial expression of both trh and vvl is directly and independently activated by STAT92E (Sotillos et al., 2010). vvl has been reported to be required for the maintenance of trh expression, suggesting that although the initial expression of early tracheal transcription factor genes is independent of the others, these genes subsequently become dependent on each other for their continued expression (Zelzer and Shilo, 2000).
The early-expressed tracheal transcription factors not only regulate each other but also regulate key downstream targets required to form trachea. For example, Trh and Vvl work together to control breathless (btl), which encodes an FGF receptor essential for tracheal migration (Anderson et al., 1996; Klämbt et al., 1992; Ohshiro and Saigo, 1997; Wilk et al., 1996), and rhomboid (rho), which encodes a transmembrane protease that processes the EGF ligand Spitz and is required for complete internalization of tracheal precursors (Lee et al., 2001; Llimargas and Casanova, 1999; Nishimura et al., 2007). Trh also independently regulates tracheal defective (tdf) and pebble (peb, also known as hindsight [hnt]), which encode two putative transcription factors (Boube et al., 2000; Eulenberg and Schuh, 1997; Yip et al., 1997), and Vvl is thought to independently regulate thick veins (tkv), a Dpp receptor, which is required in the subset of tracheal cells that migrate to form specific dorsal-ventral branches (Llimargas and Casanova, 1997). Kni/Knrl are expressed in the entire trachea initially, then become restricted to specific branches where they are required to repress expression of sal, their only known tracheal target gene (Chen et al., 1998). Altogether, the studies of gene regulation in the trachea suggest a model wherein the early-expressed transcription factors control both unique and overlapping sets of targets genes to drive tracheal formation (Boube et al., 2000).
Since the loss of known targets of Trh results in phenotypes that are relatively mild compared to loss of trh, additional downstream targets of this key regulator must exist. To find such targets, we performed an in situ hybridization screen and carried out a genome-wide microarray analysis. To our surprise, we discovered that expression of nearly every gene expressed in the trachea requires trh. A re-examination of the regulatory relationships among the early expressed tracheal transcription factors leads to a new model for how gene expression is controlled in this tissue.
For the in situ hybridization screen, two EMS-induced trh alleles, trh1 (trh5D55) and trh2 (trh7J83), and the trh8 protein-null P-element excision allele were used. For the microarray experiments, the trh8 allele was used. vvl6A3 (a kind gift from Dr. Jordi Casanova) and Df(3L)ri-79c (a deficiency deleting both kni and knrl; Bloomington stock center) were used for the pilot screen for Vvl and Kni/Knrl downstream genes. Df(3L)BSC247, Df(3L)BSC362 (deficiencies deleting trh) and Df(3L)Exel6109 (a deficiency deleting vvl) were obtained from the Bloomington stock center.
The trh mutant alleles were balanced either over TM3 carrying a ftz-lacZ transgene marker or over TM3 carrying a twi-GFP transgene marker. cDNA clones for the tracheal-expressed genes were obtained from the laboratories of Allan Spradling and Philip Beachy. In situ hybridization with gene specific probes was performed as previously described (Lehmann and Tautz, 1994) either in combination with a lacZ or GFP probe to distinguish the homozygous mutant embryos from their heterozygous siblings or by pre-sorting GFP-negative trh2 homozygous embryos using a COPAS embryo sorter (Union Biometrica). vvl6A3 and Df(3L)ri79C lines were balanced over TM3, twi-GFP balancer and homozygous embryos were pre-sorted before in situ hybridization. Images were obtained using an Axiophot compound microscope (Carl Zeiss, Inc) configured with a digital camera (Coolpix 4500; Nikon).
Fluorescent in situ hybridization with mRNA probes and antibody staining with α-Trh was performed as described (Knirr et al., 1999). Images were obtained under identical settings for each probe using a confocal LSM 510 Meta microscope (Carl Zeiss, Inc.).
Embryo fixation and staining were performed as described (Reuter et al., 1990). The primary antibodies used were rat α-Trh (Ward et al., 1998), rat α-Vvl (a gift from S. Certel; 1:50), guinea pig α-Kni (a gift from S. Small; 1:20), mouse α-βgal (Promega; 1:500 for fluorescence; 1:10,000 for HRP), rabbit α-GFP (Molecular Probes; 1:40,000), and mouse α-Tgo (Developmental Studies Hybridoma Bank; 1:2). For HRP staining, Biotin-labeled secondary antibodies were used at 1:500 dilution (Molecular Probes). HRP images were taken with an Axiophot microscope (Carl Zeiss, Inc.) and Coolpix 4500 camera (Nikon). For fluorescent staining, Alexa-488- or Alexa-568-labeled secondary antibodies were used at a 1:500 dilution (Molecular Probes). Confocal images were obtained using a LSM 510 Meta confocal microscope (Carl Zeiss, Inc.).
Three independent samples of stage 11-16 trh8 homozygous mutant embryos and three of wild type (Oregon R) embryos were isolated using a COPAS Select embryo sorter (Union Biometrica). Total RNA was isolated using Trizol (Invitrogen) and cleaned up with the RNeasy kit (Qiagen). Total RNA (100ng) was labeled according to standard Affymetrix protocols and hybridized to the Drosophila genome 2.0 chip. After scanning, intensity values were normalized by RMA (Irizarry et al., 2003a; Irizarry et al., 2003b) and statistical analysis was performed using Spotfire software (TIBCO). Trh target genes were identified based on a 1.4 fold change in gene expression, with a p-value ≤0.05. This fold change was selected because many known confirmed Trh target genes fall into the range of a 1.4 fold or more change.
All microarray data have been deposited in Gene Expression Omnibus (GEO), accession no. GSE28780.
Total RNA from three independent samples each of wild-type and trh8 homozygous mutant embryos (stage 11-16) was reverse transcribed using SuperScript III Reverse Transcriptase (Invitrogen) with oligo(dT20V) primers for the two-step RT-qPCR assays. Gene-specific primers were used to amplify the corresponding cDNA by RT-qPCR on a CFX96 Real-Time PCR Detection System (BioRad) using the SYBR Green Supermix (BioRad). Primer sets for each gene were designed against the complete nucleotide sequence, as deposited on FlyBase, using PerlPrimer. The specificity of each primer set and molecular weight of the amplicon were monitored by dissociation curve analysis and verified by running the PCR products against size standards on a 1.5% agarose gel. A sample volume of 20 μl was used for all assays, which contained a 1× final concentration of SYBR Green mix, 200nM gene specific primers, and 4μl of diluted template. All samples were run in triplicate. Genes for the RT-qPCR were chosen based on their fold change in the microarray analysis (−72.8 to −1.4 fold for downregulated genes and 1.4 to 10.5 fold change for upregulated genes) and the Gene Ontology (GO) cluster into which they have been categorized.
trh encodes three alternatively spliced transcripts, designated RA-RC (BDGP; Fig.1A). trh-RA and trh-RB transcripts share all exons except a small additional exon (exon 6) found in the RA form. The trh-RC transcript includes all but the first exon of trh-RB but also contains several unique 5′ protein coding exons. All three splice forms include the bHLH, PAS-A and PAS-B domains (Fig.1A). As previously reported, trh mRNA is expressed in the tube-forming cells of the salivary duct, trachea and filzkörper, and in a subset of cells in the central nervous system (CNS) (Isaac and Andrew, 1996; Wilk et al., 1996). In situ hybridization with all of the splice form-specific probes (exon 6 of RA form, exon 1 for RA/RB-specific form, and exon 2 of RC form) recapitulated the previously published trh mRNA expression pattern (Isaac and Andrew, 1996), suggesting that all three splice forms are expressed similarly (Fig.S1).
Before initiating our screens for transcriptional targets of Trh, we first characterized the three trh alleles with the most severe and consistent tracheal invagination defects, two EMS-induced alleles, trh1 and trh2, and a P-element excision allele, trh8 (Isaac and Andrew, 1996; Jürgens et al., 1984; Wilk et al., 1996). For the EMS alleles, we generated and sequenced PCR fragments spanning the entire trh coding region from homozygous mutant embryos. We discovered several nucleotide substitutions. Most are silent mutations that do not affect the amino acid sequence, but one of the substitutions (ATC to AAC) changed an Ile to Asn in the PAS-A domain in both trh1 and trh2 (Fig.1A). This residue is conserved in all Drosophila species and in some insects, including the silkworm (Fig.1B). A conserved Val or Leu residue is found in the Trh orthologs in other insects and crustaceans, and the orthologous mouse and human NPAS3 proteins have a conserved Met in this position (Fig.1B). Previous studies showed that the PAS domain of Trh confers target gene specificity (Zelzer et al., 1997), and our molecular characterization of trh mutants further confirms that the PAS domain is critical to Trh function. The trh8 allele is a rosy - excision derivative of the trh3 enhancer trap line (also known as trh10512) in which a 14.5 kb PZ element had inserted right after nucleotide 14 of the 5′ UTR of the trh-RA and trh-RB transcripts (Isaac and Andrew, 1996). Genomic PCR and sequence analysis of trh8 revealed an 8949 bp internal deletion of the PZ element (2740-11687 nt), leaving the flanking genomic region intact (Fig.1A).
Next, we examined trh mRNA and protein expression in the mutant alleles. trh is reported to regulate its own expression in the trachea, but not in the salivary duct (Haberman et al., 2003; Wilk et al., 1996). Indeed, although trh mRNA was detected in the early trachea in all three trh alleles, the number of trh-expressing cells decreased over time and trh mRNA was not detected after early st.14 (Fig.1C). Failure of the tracheal cells to express trh mRNA is not due to cell death; we and others did not detect any significant increase of apoptosis in the tracheal cells of these mutants (Zelzer and Shilo, 2000). Moreover, expression of three pro-apoptic genes, hid, reaper and grim, was unaffected by loss of trh (data not shown). Finally, tracheal cells persist through at least embryonic stage 15 in trh mutants based on the expression patterns of a lacZ reporter gene driven by two independent vvl enhancer fragments; β-gal protein expression (which persists longer than lacZ RNA or the endogenous vvl RNA/protein) from these enhancers is maintained in the uninvaginated tracheal cells of trh mutant embryos through stage 15 (Sotillos et al., 2010). Expression of trh mRNA in the salivary duct and filzkörper was at approximately wild-type levels with all three alleles, suggesting that trh does not autoregulate in either tissue (Haberman et al., 2003). Interestingly, trh expression in the CNS was unaltered in trh1 and trh2, but was not detected in trh8 mutant embryos (Fig.1C). In trh1 and trh2, α-Trh staining showed a very similar pattern to that of the mRNA; protein was detected in tracheal cells until early st.14 and absent at later stages (Fig.1D). These data suggest that the trh1 and trh2 alleles make non-functional Trh protein that fails to maintain its own expression. In the trh8 mutants, no protein expression was detected at any stage, revealing that trh8 is protein null (Fig.1D). The strong α-Trh signals detected in the secretory portion of the salivary glands at st.13 and later are nonspecific; the same staining was also detected in embryos homozygous for a deficiency that deletes trh and several flanking genes (Fig.1D; data not shown).
To identify additional transcriptional targets of Trh required for tracheal tube formation, we performed an in situ hybridization analysis with cDNA clones for ~100 genes expressed in the trachea based on findings from the Berkeley Drosophila Genome Project (BDGP; http://insitu.fruitfly.org/cgi-bin/ex/insitu.pl). We examined the expression of each gene in both wild type and trh mutant embryos. Homozygous trh embryos were unambiguously distinguished from their heterozygous siblings by including probes specific for a marker on the balancer chromosome (lacZ or GFP) in the in situ hybridization mix. For the genes whose expression overlapped significantly with the balancer chromosome marker, we presorted the trh null homozygous embryos from their non-mutant siblings, which carried a GFP marker on the balancer chromosome, using a COPAS embryo sorter. The genes, cDNAs, and the trh allele(s) used for the in situ screen are included in Table 1.
Surprisingly, we found that tracheal expression of all the genes we examined was completely absent in trh mutants at late stage 12 and later, suggesting that trh may be required for maintaining expression of all tracheal genes. The affected genes include those expressed in the entire trachea (Fig.2A-E) as well as genes expressed in only a subset of branches or in limited tracheal cell types (Fig.2F-H). The absence of tracheal gene expression is not due to a loss of tracheal cells since we have shown that the cells are present and express trh mRNA (and Trh protein) until embryonic stage 14 in the trh mutants (Fig.1C, D). Some of these genes were also expressed in other tissues, but only their tracheal expression was affected by trh loss (Fig.2D, H). The absence of tracheal gene expression in the trh1 and trh2 mutants confirms that the Trh proteins made in the EMS-induced mutants are non-functional.
We then asked if trh is also required for the initial expression of tracheal genes. Out of the 96 genes we have examined, 38 genes begin to be expressed at embryonic stages 10 and 11 (referred to as early-expressed genes hereafter). We found that expression of 20 of the 38 early-expressed genes absolutely required trh at all stages. Previous studies had suggested that Trh is not required for the initial expression of all tracheal genes; specifically, the early expression of tkv was shown to not change in trh mutants (Boube et al., 2000). To carefully examine the expression of tkv and other early-expressed genes, we performed fluorescent in situ hybridizations with tkv and three other early-expressed tracheal genes (ft, CG13333 and CG8312). Indeed, their mRNA expression was detected in tracheal cells at stage 10/11, albeit at somewhat reduced levels (Fig.2I, J, L, M and data not shown). Thus, although trh is required for the expression of all tracheal genes by late stage 12, the initial expression of a small subset of early expressed genes occurs independently of trh.
Our unexpected finding that Trh is required to maintain expression of all tracheal genes led us to more carefully investigate the interdependence of the early-expressed tracheal transcription factors. trh and vvl are among the earliest-expressed tracheal genes (Anderson et al., 1995; Isaac and Andrew, 1996; Wilk et al., 1996) and both genes are directly regulated by STAT92E, which functions in early embryonic patterning (Sotillos et al., 2010). Two other early expressed tracheal transcription factor genes include kni, which also functions earlier as a zygotic gap gene, and knrl, which has a redundant role with kni in the trachea (Chen et al., 1998; Rothe et al., 1989). The initial expression of trh, vvl and kni is reported to be unaffected by the loss of expression of the other two genes (Boube et al., 2000). We hypothesized that these early-expressed transcription factors might become interdependent at later stages, which could explain why the expression all tracheal genes is gone in trh mutants by embryonic stage 12. Thus, we examined both the early and late expression of trh, vvl, and kni in embryos mutant for the other genes. First, we examined trh mRNA and protein expression in vvl and kni mutants, using the null allele vvl6A3 and a deficiency that deletes both kni and knrl, Df(3L)ri-79C. Consistent with previous findings (Zelzer and Shilo, 2000), the initial expression of trh mRNA was unaffected by the absence of vvl or of kni (Fig.3B, C). Because of the role of kni as a zygotic gap gene, kni deficient embryos develop with two relatively normal anterior metameres, one relatively normal posterior metamere and several fused metameres in between; nonetheless, trh mRNA was expressed to WT levels in all tracheal cells in these embryos (Fig.3C). Similarly, Trh protein levels appeared normal in both vvl and kni deficient trachea at early stages (Fig.3E, F). Previous studies reported that although the initial expression of Trh is not affected in vvl null mutants, the levels of Trh are reduced over time and become almost undetectable by stage 15 (Zelzer and Shilo, 2000), suggesting a requirement for Vvl in the regulation of Trh. We find that, although the number of trh-positive cells in vvl null mutants or vvl deficiency embryos (Df(3L)Exel6109) may be fewer than in wild type, trh mRNA and protein expression levels in these cells are comparable to wild-type levels throughout embryogenesis (Fig.3H, K; data not shown). Thus, Vvl plays only a minor role, if any, in maintaining trh expression. Similarly, trh mRNA and protein levels are comparable to WT at all stages in kni deficient embryos, indicating that kni is not required for trh expression (Fig.3I, L).
The initial early expression of Vvl was not significantly affected in either the trh or kni mutant backgrounds, consistent with previous findings (Boube et al., 2000; Fig.4B, C). Expression of Vvl was reduced in later stage trh mutants and was completely gone in all tracheal cells by embryonic stage 15 (Fig.4E). In contrast, loss of kni did not affect Vvl expression at any stage (Fig.4C, F).
Previous studies indicated that the early expression of kni is independent of trh, suggesting that kni may be one of the primary tracheal genes (Boube et al., 2000). Inconsistent with these findings, we observed that the tracheal expression of kni was completely absent in trh mutants at all stages (Fig.4H, H’). In wild-type embryos, Kni is expressed in a subset of tracheal cells (Fig.4G, G’). Double staining with Trh and Kni antibodies revealed that whereas Trh protein is easily detected in the tracheal cells in trh2 mutants at early stages, the Trh-positive cells do not express Kni (Fig.4H, H’). By stage 15, neither Trh nor Kni tracheal expression can be detected (Fig. 4K, K’). In contrast, Kni is expressed to wild-type levels in the tracheal cells of vvl mutants at all stages (Fig.4I, I’, L, L’).
Taken together, this analysis reveals that expression of both trh and vvl is initiated independently of other tracheal genes and that maintenance of their expression absolutely requires Trh. As observed with most tracheal genes analyzed in our in situ analysis, kni requires Trh to both initiate and maintain its tracheal expression. A small number of tracheal genes behave like trh and vvl, where initial expression is independent of other tracheal genes. Nonetheless, every gene expressed in the late trachea required trh, regardless of whether initial expression was Trh-dependent or not.
Interestingly, both trh and vvl null embryos appear to have fewer tracheal primordial cells, based on the early expression of Trh, Vvl and/or Tango (Fig. 2I-M; data not shown). This observation suggests that activation of both Trh and Vvl by the upstream patterning genes (JAK/STAT signaling in the absence of Sal, Wg and Dpp signaling) may be required for committing cells to a tracheal lineage. This idea is consistent with the finding that ectopic expression of trh can induce formation of additional trachea, but only in cells that also express vvl (Boube et al., 2000; Wilk et al., 1996; Zelzer and Shilo, 2000).
Our findings that vvl is not required for trh expression led us to ask how many tracheal genes require Vvl. A pilot in situ analysis revealed that only six out of 21 tested tracheal genes were affected by loss of vvl. Three genes, mipp1, CG4726 and CG3777, were completely absent at all stages (Table 2). Expression of serrano (sano), which encodes an apically-enriched cytoplasmic protein that controls tracheal tube length (Chung et al., 2009), was significantly reduced at early stages and disappeared completely at later stages (Table 2; Fig.5A). Levels of CG33006 and CG32499 were strongly reduced at all stages (Table 2; Fig.5B, D). The remaining 16 genes were expressed to wild-type levels in the vvl mutants (Table 2; Fig.5C, E, F). Thus, we estimate that Vvl contributes to the expression of approximately 25 - 30% of all tracheal genes. We also performed fluorescent in situ with tkv probe in vvl mutants to examine it more carefully since tkv was the only known gene whose early expression is regulated by Vvl but not by Trh (Boube et al., 2000; Llimargas and Casanova, 1997). Inconsistent with the previous studies, however, levels of tkv mRNAs in the vvl mutant tracheal cells were as high as levels in trh mutants (Fig.2J, K), suggesting that the initial expression of tkv is not dependent on trh or vvl.
Only a single target of Kni has ever been identified; Kni represses sal expression in dorsal tracheal cells (Chen et al., 1998). The observation that kni does not affect the expression of any other known downstream gene suggests that it might act in parallel in early tracheal development (Boube et al., 2000). A pilot in situ screen of 32 tracheal genes, including all 21 of the genes we tested in vvl mutants, revealed that not a single tested gene was affected by loss of kni (Table 2). This finding suggests that Kni has a very minor role in this tissue compared to Trh or even to Vvl. Consistent with our failure to find additional targets of Kni, genetic studies support a model wherein sal may well be the only transcriptional target of Kni in this tissue (Ribeiro et al., 2004).
Our in situ hybridization screen identified about 100 tracheal-expressed genes regulated by Trh, but many Trh targets will have been missed since the BDGP expression datasets are incomplete and cover only about 1/3 of the genome. To find additional genes regulated by Trh, we performed microarray analysis comparing RNA from wild type and trh mutant embryos. 983 genes had expression levels reduced by 1.4-fold or greater; a number chosen because many of the Trh target genes identified in our in situ screen were included in this set (35 out of the 98 known genes; Fig.6A and Table S1). The expression patterns for about 1/3 of the target genes identified by microarrays are available in the BDGP dataset (356 genes), with 26.1% of them being expressed in the trachea (93 genes) (Table S1). In addition to tracheal-expressed targets of Trh, the microarray analysis also identified many novel targets expressed in the other tissues where Trh is normally expressed, such as the filzkörper, central nervous system (CNS), salivary duct and early salivary gland, suggesting that they are regulated by Trh in those tissues (19.9%; 71 genes; Table S1). To validate the microarray data, we chose 11 novel target genes (one gene without a known expression pattern and ten genes with known expression patterns in the trachea, filzkörper or salivary duct), and performed in situ hybridization in both wild type and trh mutant embryos. All eleven genes were affected by loss of trh (Fig. 6B and Table S1). Spn43Aa, which encodes a serine protease inhibitor, is expressed in several tissues including the trachea, and its tracheal expression requires trh at all stages (Fig. 6B). Osi19, the gene with greatest reduction in expression levels in trh mutants relative to WT, is specifically expressed in the late trachea, and as observed with all of the tracheal genes we have examined, this expression is gone in trh mutants (Fig. 6B). CG12009, which encodes a tracheal-specific chitin-binding protein, and CG13640, a novel filzkörper-specific gene, also absolutely require trh for their tracheal expression (Fig. 6B).
We also performed real-time quantitative PCR (RT-qPCR) with a dozen downregulated Trh target genes to validate microarray data. The set included four genes with known expression in the salivary duct, CNS or epidermis, and eight genes without any known expression patterns (Table S1). Their fold change ranges from −72.8 to −1.4 (Table S1). All but one of the genes showed reduced levels of expression in trh mutants compared to WT and seven showed significant reduction by this analysis, including Glucose dehydrogenase (Gld), which is expressed in the salivary duct and other tissues, Synapse protein 25 (Snap25), which is expressed in the CNS midline, and several novel genes such as CG14934, whose expression pattern is not known (Fig. 6C and Table S1). Four other genes, including Glutamate receptor I (Glu-RI) and CG31955, also showed reduced expression, but not to statistically significant levels (Fig. 6C and Table S1).
The microarray experiments not only identified downregulated genes but also a number of genes whose expression went up in trh mutants. RT-qPCR with ten of these genes with expression levels increased by 1.4 to 10.5 fold, revealed that genes with four fold or greater change by microarray analysis also showed significant increases by this assay, suggesting that a four fold change is a reasonable cut-off for upregulated genes (Fig. 6D and Table S2). The 26 genes with four-fold or greater increases in expression levels in trh mutants include Rieske iron-sulfur protein (RFeSP), which encodes a Ubiquinol-cytochrome-c reductase, GstD4 and GstD7, both of which encode Glutathione S transferases, and Cyp12a4, which encodes a protein with putative electron carrier activity that also has been shown to be upregulated in response to insecticides (Fig.6D and Table S2; King-Jones et al., 2006). Although we do not yet understand the roles of all the genes whose expression is affected by loss of trh, our validation of the microarray data by in situ hybridization and RT-qPCR suggest that the genes identified by this method are indeed regulated by Trh.
To ask which biological processes are affected by Trh, gene ontology (GO) clustering was performed using DAVID, a program that identifies enriched biological clusters in a given dataset relative to their representation in the entire genome (Dennis et al., 2003; Hosack et al., 2003; Huang da et al., 2009). This analysis revealed that a range of biological processes are affected by the absence of trh, which is not surprising since all tracheal-expressed genes are downregulated in trh mutants. The affected biological processes include “glucose metabolic process”, “proteolysis”, “chitin metabolic process” and “nucleotide catabolic process”, with an enrichment of the associated GO terms by at least twofold compared to the Drosophila genome (Table 3). Genes with function linked to the tricarboxylic acid cycle (TCA)/aerobic respiration were highly enriched, which makes sense since the trachea is a respiratory organ (Table 3). Many genes involved in synapse transmission were also enriched, perhaps revealing a role for Trh in the CNS, where it is also expressed (Table 3).
To identify downstream targets of Trh, we performed two large-scale screens, including an in situ hybridization screen of approximately 100 tracheal-expressed genes and a genome-wide microarray analysis. Surprisingly, our in situ screen revealed that trh is required for maintaining expression of all tested tracheal-expressed genes, suggesting that trh is required for tracheal cell identity, a role similar to that of the C. elegans pharyngeal regulator pha-4 (Gaudet et al., 2004) and the mouse pancreatic regulator Pdx1 (Jonsson et al., 1994; Svensson et al., 2007). A re-examination of the regulatory hierarchy among the early-expressed transcription factors and the pilot in situ screen for Vvl and Kni targets during tracheal formation suggest a new model wherein Trh plays a far greater role than these other transcription factors in regulating the expression of tracheal genes. Indeed, loss of vvl, the gene previously thought to function in parallel with Trh (Boube et al., 2000), affects expression of only an estimated 25-30% of tracheal genes, whereas loss of kni/knrl affects expression of only a single known tracheal gene. A role for Trh in tracheal cell identity and in regulating all tracheal genes is consistent with both loss-of-function and overexpression phenotypes, wherein all of the morphological events of trachea formation completely fail to occur in trh null embryos and formation of extra tracheal segments can be induced by global expression of a trh cDNA (Isaac and Andrew, 1996; Wilk et al., 1996).
The genes affected by loss of trh have variable expression patterns both in terms of when and where they are expressed in the trachea (Fig.2). Presumably, many of the earliest-expressed genes could be direct targets of Trh, whereas the genes expressed only at late stages could include both direct and indirect targets since trh continues to be expressed in all tracheal cells through the end of larval life. The tracheal genes whose expression begins only late in embryogenesis or whose expression is limited to only a subset of tracheal cells must also be regulated by other tracheal-expressed transcription factors. Indeed, many transcription factors were identified as Trh targets in our screen. Examples include Xbp1 and CrebA, two bZip transcription factors that regulate secretory function in multiple secretory organs (Abrams and Andrew, 2005; Fox et al., 2010; Lee et al., 2005; Shaffer et al., 2004), No Ocelli (Noc), a zinc finger transcription factor controlling morphogenesis of specific tracheal branches (Dorfman et al., 2002), Molting defective (Mld), another Zinc finger transcription factor involved in ecdysone biosynthesis (Neubueser et al., 2005), and Limpet (Lmpt), a zinc finger protein of unknown function (Table 1). Our microarray experiments also identified a number of completely uncharacterized putative transcription factors downregulated in trh mutants (Table S1). A cascade of transcriptional regulation could explain how so many genes are affected by the loss of only a single transcription factor and could explain the range in temporal expression observed with the different tracheal genes. The genes downstream of Trh also have a range of spatial expression patterns; some genes are expressed in the entire trachea, some are expressed in only specific branches, and others are expressed in only specific cell types, such as in the fusion cells (Fig. 2). The differential expression of genes within the trachea suggests that many Trh targets also require input from branch-specific or cell-type specific pathways, many of which have already been described, including FGF, Wg, Dpp and Notch signaling (Chiang et al., 1995; Chihara and Hayashi, 2000; Franch-Marro and Casanova, 2002; Ikeya and Hayashi, 1999; Llimargas, 1999, 2000; Steneberg et al., 1999; Vincent et al., 1997). Nonetheless, our finding that Trh is required for the expression all tracheal genes, regardless of when and where they are expressed, suggests a role for Trh in both specifying and maintaining the tracheal cell fate.
Although many of the early-expressed (stage 10/11) tracheal genes are completely dependent on trh at all stages, some tracheal genes, including four genes we have examined by fluorescence in situ analysis (tkv, ft, CG13333 and CG8312), are expressed in trh mutant tracheal cells at stage 10/11, suggesting that another factor may be required for regulating the initial expression of a subset of tracheal genes. It is unlikely, however, that trh and vvl independently activate expression of these early genes, because all of the four genes whose early expression is either unaffected or only slightly downregulated by trh were still expressed in trh vvl double mutant trachea at early stages (data not shown). Another possibility is that expression of a few early-expressed tracheal genes could be activated by the same factors that initiate trh and vvl expression, such as STAT92E (Fig. 7). A third possible explanation is based on the expression pattern of many of these early genes. Three of the four genes, including tkv, ft and CG13333, are not exclusively expressed in the tracheal cells. Instead, they are expressed in a broad ectodermal domain that overlaps the early trachea (data not shown). Gene expression in this broader domain may not necessarily be regulated by tracheal-specific transcription factors.
trh, vvl and kni/knrl are among the earliest-expressed transcription factors in tracheal formation and their initial expression has been reported to be independent of one another (Boube et al., 2000; Zelzer and Shilo, 2000). However, our re-examination of the expression of each of the early-expressed transcription factors in the mutant background of the other two genes revealed that not all of the reported changes in gene expression are correct. Consistent with previous findings, the early expression of trh and vvl are independent of each other; vvl is dispensable for the maintenance of trh, however, whereas trh is absolutely required for the maintenance of vvl. Also, kni tracheal expression is completely dependent on trh, suggesting that kni is just one of many transcriptional targets of trh, rather than being among the earliest-expressed transcription factors whose activation is directly dependent on the early patterning genes (Fig. 7).
Consistent with a new transcriptional hierarchy, expression of only a subset of tracheal genes depends on vvl. Our pilot in situ screen suggests that vvl is required for the expression of approximately 25-30% of tracheal genes and that Vvl plays a relatively minor role in the transcriptional regulation of tracheal genes compared to Trh. Indeed, vvl mutants have milder tracheal defects than trh mutants; whereas no morphogenetic events occur in trh mutants, vvl mutant tracheal cells do invaginate and form rudimentary branches (Fig. 3). Identification of the full transcriptional targets of vvl will lead to further understanding of the function of vvl during tracheal formation. Consistent with our finding that kni is just one of many trh targets, kni itself had only one previously known tracheal target, sal, which is repressed by Kni in dorsal tracheal cells (Chen et al., 1998). We failed to identify any additional transcriptional targets of kni in our pilot in situ screen.
Our microarray experiments using mRNA from stage 11-16 embryos not only identified a number of novel Trh targets, but also confirmed many of the Trh target genes identified in our in situ screen. 35 out of the 98 genes we examined in the in situ screen were reduced by 1.4 fold or more in trh mutants compared to WT. There could be several reasons why all of the in situ targets were not included in the microarray targets: Many Trh downstream genes showed less than 1.4 fold change in the microarray analysis. An example is kkv, which encodes chitin synthase that regulates tracheal tube diameter (Tonning et al., 2006); kkv was down only 1.36 fold in the trh mutants. A similarly mild change in expression levels was observed with many Trh downstream genes, including many genes that have only transient tracheal expression and genes that are expressed in multiple other tissues, where expression is unaffected by loss of trh. The change in mRNA levels in the trachea relative to the whole embryos that were used in the microarray experiments resulted in total averaged mRNA changes that are likely too small to be detected in the microarray. We also found a number of genes that showed more than 1.4 fold change, but had p-values greater than 0.05. Although the three sets of our microarray samples showed more than 98% correlation, the occasional outliers made the fold change non-significant for some known Trh targets. In summary, although the whole-genome approach was successful in revealing new Trh target genes, the sensitivity limitations of this type of study will result in many target genes being missed. Nonetheless, microarray experiments identified almost 1,000 transcriptional targets of Trh, many of which are novel targets and a subset of which we have confirmed by either in situ hybridization or RT-qPCR. GO clustering revealed that Trh downstream genes are involved in various steps in numerous biological processes, which is consistent with our findings that Trh functions very early during tracheogenesis and controls all of the genes required to form of this organ. Microarray experiments also identified a number of targets expressed in the other tissues such as CNS providing potential insight into which aspects of neuronal development are under the control of Trh.
trh encodes three splice forms that are expressed in the same spatial and temporal patterns.
trh autoregulates in the trachea, but not in the other tissues.
trh regulates expression of all tracheal genes based on in situ analysis of ~100 genes.
ventral veinless (vvl) regulates only 25-30% of tracheal genes.
knirps (kni) regulates only a single tracheal gene.
Contrary to published findings, Trh maintains vvl and activates kni, but neither kni nor vvl affect trh.
Microarray studies reveal that Trh regulates a broad range of biological processes.
Fig.S1. Alternative splice forms of trh are expressed similarly in the multiple embryonic tissues, including the trachea. (A) The region of each isoform-specific probe is indicated by red bars (trh-RA-specific probe, 70bp; trh-RA/RB-specific probe, 193bp; trh-RC-specific probe, 261bp). (B-D) In situ hybridizations of each splice form in wild-type embryos are shown. All probes specific for trh-RA (B), trh-RA/RB (C) and trh-RC (D) isoforms show the typical trh expression pattern, with trachea (arrow), salivary gland (black arrowhead), and CNS expression (open arrowhead).
We thank Dr. R. Fox and Dr. R. Maruyama for comments and suggestions on the manuscript and all members of the Andrew Lab for their input during this study. We also thank Dr. J. Casanova, Dr. L. Jiang, Dr. S. Certel, Dr. S. Small, the Bloomington stock center and the Developmental Studies Hybridoma Bank for reagents. We gratefully acknowledge Flybase for gene information and BDGP for the genome-wide in situ analysis. This work was supported by National Institutes of Health grant RO1 DE012873 to D.J.A. The COPAS embryo sorter was supported by an ARRA-supported NIH Shared Instrumentation Grant 1S10RR25477.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.