CrebA regulates secretory activity in the Drosophila salivary gland and epidermis

As tissues differentiate, they express a unique combination of genes encoding proteins controlling tissue structure and physiology. Deciphering how the genes are transcriptionally regulated is key to understanding the final stages of organogenesis and to providing molecular insight into how an organ becomes equipped to perform its specialized functions. For example,transcriptional regulation studies of genes crucial for liver function suggest that their regulation is relatively complex. In this tissue, the winged-helix transcription factors HNF3α and HNF3β, and the nuclear receptor superfamily member HNF4 converge on the regulation of the homeodomain-containing transcription factor HNF1(Cereghini, 1996). HNF1 is directly involved in the transcriptional activation of terminal genes that are essential to liver function (Courtois et al., 1987; Cereghini et al.,1988). Functional binding sites for the transcription factors upstream of HNF1 have also been identified in the enhancer/promoter regions of the terminal genes. Furthermore, HNF3β and HNF3α (indirectly through HNF3β) are themselves regulated by the leucine-zipper transcription factor C/EBPα(Samadani et al., 1995). Thus,the regulatory mechanisms of liver gene expression appear to be both hierarchical and non-hierarchical, and include converging regulatory pathways.

In the case of pharyngeal development in the nematode C. elegans,transcriptional regulation appears to be simpler. In this organ, PHA-4, a winged-helix transcription factor homologous to mammalian HNF3β and Drosophila Fork head, is crucial for both pharynx specification and subsequent differentiation. The pharynx completely fails to form in pha-4 mutants and ectopic expression of pha-4 can activate at least one pharynx-specific gene in new places(Kalb et al., 1998). PHA-4-binding sites are present upstream of all tested pharyngeal genes, and temporal expression correlates with PHA-4-binding site affinity(Gaudet and Mango, 2002). In general, early genes have relatively strong binding sites for PHA-4, whereas later genes have relatively weak sites. Correspondingly, early in organ development, low levels of PHA-4 are present and are sufficient to activate the early-expressed genes. As levels of PHA-4 increase temporally, late genes are expressed. By using a temperature sensitive allele of pha-4, the same group confirmed that the late expression of pha-4 is indeed important for proper organ development(Gaudet and Mango, 2002). Thus, the role of pha-4 is not only to initiate a downstream regulatory cascade but also to function throughout the development of the C. elegans pharynx by directly regulating genes pivotal to its final form and function. Whether this single-tier mode of regulation is a recurring theme in organ development remains to be determined.

The Drosophila salivary gland is an excellent model for investigating how a tissue acquires specialized function, as much is already known with respect to how the gland is specified (reviewed by Andrew et al., 2000; Abrams et al., 2003). In addition, large-scale expression studies suggest that a significant number of genes are expressed to high levels in the salivary gland, and therefore access to a large pool of potential downstream targets is available(http://www.fruitfly.org/EST/index.shtml). Salivary gland formation requires the homeotic transcription factors, Sex Combs Reduced (Scr), Extradenticle (Exd) and Homothorax (Hth); in loss-of-function mutants of any of these genes, salivary glands fail to form. Moreover, ectopic expression of Scr, the one component of the complex with limited spatial expression, leads to the formation of additional salivary glands (Panzer et al., 1992; Andrew et al., 1994). Although the Scr/Exd/Hth complex is required for the expression of every tested salivary gland gene, expression of Scr and hth and nuclear localization of Exd disappear shortly after the salivary gland initiates morphogenesis (Henderson and Andrew,2000), suggesting that, unlike in the C. elegans pharynx,the genes that specify the Drosophila salivary gland cell fate are not involved in its terminal differentiation. Genes encoding three early transcription factors are known to be expressed in the early salivary gland under the control of Scr/Exd/Hth (Panzer et al., 1992; Andrew et al.,1994): fork head (fkh), which encodes the winged-helix transcription factor homologous to C. elegans PHA-4; CrebA, which encodes the Cyclic AMP Response Element Binding protein A; and huckebein (hkb), which encodes an Sp1/egr-like transcription factor. In the embryo, Fkh controls apical constriction of the salivary cells as they invaginate and promotes salivary cell survival by inhibiting apoptosis (Myat and Andrew,2000b). During larval development, Fkh is required to activate expression of the sgs glue genes(Lehmann and Korge, 1996; Mach et al., 1996). The SGS glue proteins play a role in the adherence of the pupal case to a substratum. Thus, Fkh is similar to C. elegans PHA-4 in as far as it is important throughout salivary gland development and function. However, a significant number of salivary gland markers are still expressed in the uninvaginated salivary cells of fkh mutant embryos(Myat and Andrew, 2000b; Bradley and Andrew, 2001) (E. Grevengoed, U. Ng and D.J.A., unpublished), suggesting that Fkh is not an organ-specifying gene. CrebA has been shown to be important in cuticle patterning and in promoting the integrity of the larval cuticle(Andrew et al., 1997). Its role in the salivary gland, however, has been less clear, although a low percentage of CrebA embryos have crooked salivary glands(Andrew et al., 1997). Hkb is required for proper morphogenesis of the secretory tube. In hkbmutants, dome-shaped salivary glands form instead of elongated tubes because of a failure to generate and deliver sufficient apical membrane(Myat and Andrew, 2000a; Myat and Andrew, 2002). Thus, fkh and hkb have genetically defined roles in the morphogenesis of the salivary gland, whereas the role of CrebA in salivary gland development remains elusive.

To learn how the salivary gland is programmed for its primary function,secretion, we focused on the regulation of genes encoding early secretory pathway components. The secretory components crucial for targeting proteins to the endoplasmic reticulum (ER), for signal peptide processing and for vesicular trafficking between the ER and Golgi have been studied extensively in yeast and in mammalian tissue culture cells (reviewed by Harter, 1995; Kalies and Hartmann, 1996; Romisch, 1999; Wild et al., 2002; Barlowe,2003a). Clear homologs of most of these proteins exist in flies based on sequence comparisons (Adams et al.,2000), although very few have been characterized to any extent(Valcarcel et al., 1999). Here, we show that genes encoding the Drosophila homologs to the yeast and/or mammalian early secretory pathway proteins are expressed at high levels in the Drosophila embryonic salivary gland and other secretory tissues, and that CrebA is essential for this high level salivary gland expression. Fkh is required for late expression of these genes and functions indirectly through maintenance of CrebA expression. Correspondingly, CrebA is required for enhanced secretory activity in the salivary gland. CrebA also activates SPCG expression in the embryonic epidermis, explaining the cuticle defects observed in CrebA mutant larvae.

cDNA clones were obtained either directly from BDGP (CK clones) or from Invitrogen (Huntsville, AL). CK cDNA clones expressed in the salivary gland were identified at the CK expression database at BDGP (Kopcyznski et al.,1998). Salivary gland expression of the following clones were identified through the BDGP embryonic expression study(http://www.fruitfly.org/EST/index.shtml):LD25651, RE14391, RE02772, GH04989, RE24638 and LD45288. The expression patterns of the remaining genes were determined in this study: Srp complex(RE23260, LD41750, LP10092 and AT23778); ER translocon (RE04612, RH61539,RE69515, RE23984, LD29171, LD27659 and GM12291); SPC (RH08585 and LD42119);COPII components (GH19061, RE35250, SD04292 and LD39266); and COPI components(GH18123, RE62270, RE38606 and LD24904).

For fkh mutants, we used the null fkh⁶ allele in an H99 background to prevent secretory cell apoptosis and to thus simplify the analysis (Myat and Andrew,2000b). Accordingly, we performed the same experiments in H99 only lines as a control. For CrebA mutants, we used the protein null allele CrebA^wR23(Andrew et al., 1997); for hkb mutants, we used hkb²(Bronner et al., 1994). l(3)01031 (sec13), l(3)j13C8 (sec23), l(3)rK561 (SrpRβ) and l(3)05712(sar1) are all lethal P-element lines; both stocks and plasmid rescue data were obtained from FlyBase. CrebA^wR23, fkh⁶H99 and hkb² were balanced over TM6B-Ubx lacZ. l(3)j13C8 was balanced over TM3-ftz lacZ. The lacZ-containing balancers were used to distinguish the homozygous mutant embryos from siblings by staining with a lacZ probe.

A 2.8 kb HindIII fragment that maps ∼3.5 kb upstream of the CrebA transcription start site and drives β-gal expression in the salivary gland and amnioserosa has been identified previously (K. D. Henderson, PhD Thesis, Johns Hopkins University School of Medicine, 2000). The following primer pairs were used to further isolate the CrebA salivary gland enhancers: CrebA 1100 (5′GAATTCCTTCGCTGTCATGC and 3′GGATCCAGCGTCTTCTAGAGATAC), which amplify an 1100 bp sub-fragment of HindIII 2.8 containing six potential Fkh binding sites(Fig. 3); and CrebA 770(5′GAATTCTAGAAGACGCTGGCGAATG and 3′GGATCCCATCTTCGATCTGGC), which amplify a 770 bp sub-fragment of HindIII 2.8 containing four potential Fkh-binding sites (Fig. 3).

Six candidate SPCG genes, representing members of distinct pathway components, were selected and from 1.3 to 2.6 kb of upstream genomic sequence of each gene was amplified. The length was limited by the exclusion of BamHI and EcoRI sites, which were needed for subcloning the resulting PCR fragments into the Casper β-Gal reporter plasmid(Thummel et al., 1988). The following genes and corresponding primer sets were used: srp68,5′GAATTCGTTGTGTGCTTCTCCAGATTGC and 3′AGTGTACGGCTATGGCGAAT; p24, 5′CAGGAACGTAAGCAGATCTCG and 3′TCCACGTAGACGATCTTGTGC; srpRα, 5′GAATTCTGAAGACAGGCTAGGCTGG and 3′GGATCCGCGATCGAAGTCGTAGTCAT; ξ-cop,5′GAATTCGGAGTTGAGCACCTCGTTG and 3′GGATCCAGGCGCTTCTCAACGTTC; spase25, 5′GAATTCACACCATACACCGTACACC and 3′GTGCTTCACTGCGGATCCAT; and sec61β,5′GAATTCCACCGAATCGAACGACTC and 3′CACACCAATGCCTCAACAAG.

All of the resulting PCR fragments were subcloned into the Casperβ-Gal reporter plasmid (Thummel et al., 1988) and used to transform w¹¹¹⁸ flies through standard methods (Spradling and Rubin, 1982). Individual lines were assayed for enhancer activity by staining with anti-β-Gal antibody.

Point mutations known to affect the binding of Fkh to defined binding sites(E.W.A. and D.J.A., unpublished) were introduced into each of the six potential Fkh-binding sites contained within the CrebA 1100 enhancer using the QuikChange Multi Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA). The following primers were used (bold nucleotides indicate changes): site 1,ACTTCCTCCTATGAATGAATGCCTGTTATTGCGGGTTCCT; site 2,CGAACACATAAAGACTATAGGTATCTGCTACCACGGCTTAGAGCCG; site 3,CGCATTGAACAAAATTCTCTTGCCTAATGATATGCTATTATTA; site 4,CCCAGTTTCATCATTATCATCACCCTGTCAAAAGCAGACTT; site 5,GCGACAGCAAAGTCAGACAGGTAAGTCAGCACTACGTTCT; and site 6,TAAATGATCTCATTGTTGACTGCCTGTTCTCCCTCAGGAC. The resulting mutant constructs were used to transform w¹¹¹⁸ flies and individual lines were assayed for enhancer activity by anti-β-Gal staining.

In situ hybridization and antibody staining were performed as previously described (Lehmann and Tautz,1994; Reuter et al.,1990). In each case, we compared the levels of expression in the homozygotes to the levels observed in their heterozygous siblings to control for experimental variation in staining levels. Antibody dilutions used in this study are as follows: α-β-galactosidase (1:5000), α-DSC73(1:800), α-Pasilla (1:1000), α-PH4α-SG1 (1:5000) andα-En (1:200). α-PH4α-SG1 (E.W.A. and D.J.A., unpublished),α-Pasilla (Seshaiah et al.,2001) and α-DSC73 (D.J.A., unpublished) are polyclonal antisera made in rat. α-En is a polyclonal antibody made in rabbit that cross-reacts with unknown secretory products in the salivary gland(Myat and Andrew, 2002). Approximately 40 pairs of salivary glands were analyzed for each mutant in the En staining experiments to gauge changes in secretory granule levels. All biotin-conjugated secondary antibodies were used at a dilution of 1:500(Vector Labs; Burlingame, CA) and all fluorescent secondary antibodies were used at a dilution of 1:400 (Molecular Probes; Eugene, OR). Confocal images were obtained with the Ultraview Confocal Microscope (Perkin Elmer) at the Johns Hopkins Microscope Facility. All other images were taken on a Zeiss Axiophot microscope with a Nikon Coolpix 4500 digital camera.

Homology searches of the Drosophila SPCGs to identify human proteins were performed using BLAST(http://www.ncbi.nlm.nih.gov). Identities/similarities were calculated using CLUSTALW(Combet et al., 2000). To identify conserved sequences upstream of the SPCGs, 500 nucleotides immediately upstream of the translation start sites of all 34 SPCGs was analyzed by MEME(http://meme.sdsc.edu/meme/website/)(Bailey and Elkan, 1994). The most conserved sequence (other than the runs of PolyA+ DNA) was a good match for the mammalian Creb-binding site consensus(http://www.cbrc.jp/research/dp/TFSEARCH.html/;threshold score 86.1). We subsequently looked for this site in the 2 kb region immediately upstream of the translation start sites of all of the SPCGs.

A survey of the EST expression databases at the Berkeley Drosophila Genome Project (BDGP) revealed that several genes encoding components of the early secretory pathway are upregulated in the embryonic salivary gland(http://www.fruitfly.org/EST/index.shtml)(Kopcyznski et al., 1998). We identified Drosophila homologs to the remaining known components of the early secretory pathway through a search of the Drosophila genome database(Adams et al., 2000). The early secretory pathway components include the machinery that targets and translocates proteins into the ER, the complex that cleaves the N-terminal signal sequence, and the proteins involved in both anterograde and retrograde vesicle transport between the ER and Golgi(Fig. 1). We refer to these genes collectively as secretory pathway component genes (SPCGs; Table 1). All of these genes contain open reading frames with significant identity/similarity to their putative human counterparts (Table 1) and, with few exceptions, are expressed at much higher levels in the salivary gland than in other embryonic tissues(Fig. 2; see Figs S1, S2 in the supplementary material). Expression was observed from early stage 11(Campos-Ortega and Hartenstein,1997), just after the salivary cells begin to internalize(Myat and Andrew, 2000a), and continued throughout embryogenesis. In addition to high-level salivary gland expression, most of the SPCGs are also expressed in other embryonic tissues,including the epidermis, proventriculus, anterior and posterior midgut primordia, the hindgut and trachea (data not shown). The SPCGs that are not expressed to relatively high levels in the embryonic salivary gland are expressed to high levels in all tissues, including the salivary gland. These findings suggest that at least one aspect of salivary gland specialization into a secretory organ is mediated through transcriptional activation of genes encoding components of the early secretory machinery.

Salivary gland formation requires the function of the homeotic gene Scr and its co-factor genes exd and hth(Henderson and Andrew, 2000). Correspondingly, we failed to detect expression of the small number of SPCGs we tested in Scr mutants in the cells that would normally form salivary glands in wild-type embryos (data not shown). As Scr and hth disappear in the salivary gland as soon as cells begin to internalize (Henderson and Andrew,2000), just prior to when SPCG expression initiates, the Scr/Exd/Hth complex is unlikely to be controlling SPCG expression directly. Instead, SPCG expression is more likely to be controlled by the early transcription factors under the control of Scr such as Fkh, CrebA and/or Hkb. Thus, we examined expression of the SPCGs in embryos homozygous for null mutations of each of these transcription factor genes. An example of salivary gland morphology in fkh, CrebA and hkb mutants is shown in Fig. 2A using a salivary gland-specific marker, α-Pasilla (Ps), expression of which is not affected by these mutations (Myat and Andrew, 2000b).

The signal recognition particle (Srp) interacts with the N-terminal signal peptide as it emerges from the ribosome. Subsequently, the Srp interacts with the Srp Receptor (SR) α-subunit, which is anchored to the cytoplasmic face of the ER through the transmembrane SR β-subunit(Fig. 1)(Legate et al., 2000; Song et al., 2000). The Srp complex is composed of a 7S RNA and six protein subunits, which are referred to as the 9, 14, 19, 54, 68 and 72 kDa Srp subunits(Fig. 1)(Wild et al., 2002). All of the known protein subunits of the Srp and the SR are expressed to very high levels in the embryonic salivary gland(Fig. 2B; see Fig. S1B,C in the supplementary material) and the levels are much higher than in other embryonic tissues, with the exception of Srp14 and Srp19, which are expressed to similarly high levels throughout the embryo (data not shown). Salivary gland expression of the Srp and SR genes is unaltered in hkb mutants, but is significantly diminished in later stages in both fkh and CrebA mutants (stage 13 and beyond; Fig. 2B; see Fig. S1B,C in the supplementary material). Interestingly, there are differences in the early expression of Srp and SR genes in fkh and CrebA mutants. Whereas in the early fkh mutants, expression levels are comparable with wild type, in the early CrebA mutants, expression is diminished to levels seen in surrounding, non-salivary gland tissues or was barely visible (Fig. 2D; Table 2; see Fig. S1B,C in the supplementary material). Thus, both fkh and CrebA are required to achieve persistent high-level expression of Srp and SR genes in the embryonic salivary gland; however, Fkh is required only to maintain and not initiate expression. CrebA mutants also had diminished Srp and SR gene expression in the late epidermis (where CrebA is also normally expressed), indicating a more general requirement for CrebA in the expression of these genes.

Translocation of nascent polypeptides into the ER is mediated by the sec61 complex, which is composed of sec61α, sec61β and sec61γsubunits (Fig. 1)(Schnell and Hebert, 2003). sec61α is expressed ubiquitously in the embryo and increased levels were detectable only in the salivary gland prior to internalization (data not shown). As the cells internalized, elevated expression of sec61α was restricted to the salivary gland cells that were still on the surface of the embryo (data not shown). In contrast to sec61α, high-level salivary gland expression of both sec61β and sec61γ persists throughout the remainder of embryogenesis. As with Srp and SR, salivary gland expression of sec61β and sec61γ is unaffected by mutations in hkb and is dramatically reduced in CrebA mutants at all embryonic stages(Fig. 2B; see Fig. S1D in the supplementary material). Similarly, mutations in fkh affect later but not earlier expression of both sec61β and sec61γ(Fig. 2D; data not shown). Epidermal expression of both genes is also diminished in CrebAmutants (Fig. 2B; data not shown).

The translocation-associating membrane protein (TRAM) is required for efficient co-translational translocation of proteins into the ER and is important for the incorporation of transmembrane proteins into the lipid bilayer (Schnell and Hebert,2003). TRAM is highly expressed in the Drosophilasalivary gland and epidermis. This expression appears normal in early fkh mutants and is attenuated but not completely removed in late fkh embryos (see Fig. S1D in the supplementary material). TRAM expression in both the salivary gland and epidermis is significantly reduced at all stages in CrebA mutants and is unaffected by hkb (see Fig. S1D in the supplementary material; data not shown).

sec62, sec63 and sec71 encode three ER transmembrane proteins implicated in mediating post-translational translocation of proteins into the ER(Romisch, 1999)(Fig. 1). These genes are highly expressed in the salivary gland as well as in the epidermis (see Fig. S1D in the supplementary material). In contrast to the Srp and sec61 genes,but similar to TRAM, expression of sec62 and sec63 can still be detected in the salivary gland cells of fkh embryos beyond stage 13, albeit at lower than wild-type levels (Fig. 2B; see Fig. S1D in the supplementary material). As in the previous cases, the salivary gland and epidermal expression is greatly diminished in both early and late CrebA mutants (see Fig. S1D in the supplementary material; data not shown). With sec62, the segmental muscle staining, which in wild-type embryos is obscured by the epidermal staining, is now obvious (Fig. 2B). sec71 expression is affected similarly to TRAM, sec62 and sec63 in fkh,CrebA and hkb mutants.

The translocation-associated protein (TRAP) complex is composed of four subunits, α, β, δ and γ, and is thought to be associated with the translocon machinery. Recently, the TRAP complex has been shown to be required for the translocation of proteins in a substrate-specific manner. For example, the secretory protein Prolactin is efficiently translocated into the ER in the absence of functional TRAP. However, prion protein (PrP), which enters the secretory pathway in three different topographical configurations, requires TRAP(Fons et al., 2003). Homologs to all four TRAP subunits exist in Drosophila, although the expression of only TRAP δ was analyzed in this study. TRAPδ is expressed at high levels in the salivary gland and epidermis. The salivary gland expression requires fkh late and CrebA throughout embryogenesis for wild-type expression in the salivary gland and epidermis(Fig. 2B,D).

Soluble nascent polypeptides have their N-terminal signal peptides cleaved off by the signal peptidase complex (SPC). The SPC is made up of five transmembrane subunits (SPases) with the following masses: 12, 18, 21, 22/23 and 25 kDa (Fig. 1)(Kalies and Hartmann, 1996). The gene encoding the 18 kDa homolog is not present in the Drosophilagenome and is most related to the 21 kDa subunit by sequence comparison. The four recognizable Drosophila SPase genes require fkh for late but not early expression (Fig. 2B; see Fig. S1E in the supplementary material). CrebA is required for wild-type levels of SPase expression in the salivary gland at all stages. As with all the other genes discussed so far, hkb mutants have levels of SPase expression comparable with those in wild type(Fig. 2B; data not shown).

Soluble proteins that are fully translocated, processed (e.g. signal peptide cleaved, glycosylated) and destined to proceed through the secretory pathway interact with proteins that recruit them to specialized domains (exit sites) in the ER (Aridor et al.,1998; Tang et al.,2001). For soluble proteins, this recruitment is thought to occur,in part, through interactions with cargo receptors of the p24 single transmembrane family of proteins (Fiedler et al., 1996; Dominguez et al., 1998). In Drosophila, expression of three p24 family members is upregulated in the salivary gland; this upregulated expression requires CrebA at all embryonic stages and fkh from stage 13 onwards (Fig. 2C; see Fig. S2A in the supplementary material; data not shown).

Once recruited to export sites in the ER, cargo molecules are packaged into transport vesicles. COPII coatomer molecules assemble onto vesicles involved in the anterograde movement of secretory products from the ER to the cis Golgi(Fig. 1)(Ellgaard et al., 1999). Transmembrane proteins destined to exit the ER contain cytoplasmic tails with exit signals, which interact directly with COPII subunits(Tang et al., 2001). The assembly of the coat from a cytoplasmic pool of soluble COPII subunits is thought to drive the actual budding of vesicles. This process is regulated by the small G protein Sarl, where hydrolysis of GTP-Sar1 to GDP-Sarl leads to coat disassembly shortly after vesicle formation(Barlowe, 2003b). In Drosophila, the genes encoding the COPII components Sec13, Sec31,Sec23, Sec24 and Sar1 are conserved and require fkh and CrebA for high levels of salivary gland expression(Fig. 2C; see Fig S2B in the supplementary material). However, as with the aforementioned SPCGs, expression of the COPII subunits appears to be relatively independent of fkhprior to stage 13 (Fig. 2D;data not shown).

COPI coats are typically involved in retrograde movement of transport vesicles, although in certain contexts they have been shown to also be involved in anterograde movement (Orci et al., 1997). The COPI coatomer component genes, α, β,β′, δ, ϵ, γ, and ζ-cop(Rothman and Orci, 1992) are all conserved in Drosophila (Adams et al., 2000). As with COPII, COPI coat formation is regulated by a small G-protein, ARF-1. Although each of the five Drosophila ARF-1 homologs is highly conserved (data not shown), ARF79F is the most homologous to human ARF1 (Table 1). Finally, the KDEL receptor (KDEL-R) has been shown in yeast and mammals to be important in the retrieval of escaped resident proteins back to the ER(Lewis et al., 1990; Lewis and Pelham, 1992).δ, ϵ, γ, ζ-cop, ARF79F and the KDEL-R genes are upregulated in the salivary gland and require fkh for their elevated expression from stage 13 and later (Fig. 2C; see Fig. S2C in the supplementary material). Although the expression of these genes is reduced in the salivary glands of CrebAmutants, both early and late expression of δ-cop, γ-cop and the KDEL-R gene are not as reduced as the other SPCGs(Table 2; Fig. S2C in the supplementary material; data not shown).

In summary, our expression studies reveal that the genes encoding components of the early secretory pathway are transcriptionally upregulated in the Drosophila embryonic salivary glands beginning shortly after the gland is specified and continuing, for the most part, throughout embryogenesis. This high level expression in the salivary gland requires CrebA function throughout embryogenesis and fkh function only after embryonic stage 13. CrebA is also required for the elevated expression of the early secretory pathway genes in the embryonic epidermis at late stages. These data suggest that Fkh maintains salivary gland expression of the SPCGs indirectly by maintaining expression of CrebA.

To test whether regulation of CrebA by Fkh is direct, we identified a 2.8 kb fragment upstream of the CrebA transcription unit that could drive salivary gland expression of a lacZ reporter gene(Fig. 3) (K. D. Henderson, PhD Thesis, Johns Hopkins University School of Medicine, 2000). Two smaller fragments from this enhancer resulted in salivary gland expression of the lacZ reporter gene either only after invagination had begun and later(CrebA-1100) or prior to invagination and later (CrebA-770)(Fig. 3; top two rows of embryos). As the later expression pattern fitted the timeframe for Fkh-dependent salivary gland expression of CrebA, we further characterized the CrebA-1100 construct, which contains six consensus Fkh-binding sites (Kaufmann et al., 1994; Lehmann and Korge, 1996; Mach et al., 1996; Takiya et al., 2003).β-Gal expression in the salivary glands with the CrebA-1100 construct was significantly reduced in fkh homozygotes although expression in the amnioserosa was unaffected, indicating that we had identified a Fkh-dependent salivary gland enhancer of CrebA(Fig. 3, third row of embryos). We next transformed flies with a CrebA-1100 reporter construct in which all six consensus Fkh-binding sites were mutated (CrebA-1100 fkh1-6 lacZ). Both lines carrying the mutated construct had significantly diminished salivary gland expression of β-Gal, although βGal expression in other tissues, including the amnioserosa and hemocytes was unaffected (Fig. 3, bottom row). We conclude that Fkh functions directly to maintain late high-level expression of CrebA in the salivary gland.

A second prediction of our model is that SPCG expression is controlled directly by CrebA. As a first step toward testing this possibility, we built and injected lacZ reporter constructs for six of the 34 SPCGs analyzed in this study. Each SPCG enhancer fragment spanned the 5′ end of the most 5′ cDNA for each gene and included ∼1-2 kb of DNA further upstream (Fig. 4). Transformant lines generated from five of the six constructs resulted in embryonic salivary gland expression. The srp68 lacZ enhancer construct did not express in the embryonic salivary gland(Fig. 4A). Salivary gland lacZ expression from the spase25 and sec61βenhancer constructs was detected from early stage 12 and throughout embryogenesis (Fig. 4E,F). Salivary gland lacZ expression from the p24-1, zCop and SrpRa enhancer constructs was first detected during stage 13 and later (Fig. 4B-D). Expression of three of the constructs were examined in CrebA mutants; expression of β-Gal from both the zCop-lacZ and sec61β-lacZ constructs was completely absent in the salivary glands (Fig. 4D,E),whereas salivary gland β-Gal expression from the spase25-lacZ construct was significantly reduced (Fig. 4F). Thus, we have identified CrebA-dependent salivary gland enhancers for at least three of the SPCGs.

A search of the regions immediately upstream of the translation start sites of the SPCGs using MEME(http://meme.sdsc.edu/meme/website/)(Bailey and Elkan, 1994)revealed a motif that is an excellent match for a mammalian Creb-binding site(http://www.cbrc.jp/research/dp/TFSEARCH.html/;threshold score 86.1) and that is present within 2 kb upstream of 32 of the 34 SPCGs (Table 3). (The translation start site is used as a reference point since transcription start sites have not been mapped for any of the SPCGs.) Interestingly, of the two SPCGs that do not contain this consensus, one (sec62) is among the least affected by mutations in CrebA(Fig. 2B) and the other, srp19, is one of only two genes we examined that had ubiquitously high levels of expression in all tissues, including the salivary gland(Table 1). Even more compelling is the finding that 13/32 have the site within 100 bp, another 7/32 have the site within 200 bp and another 5/32 have the site within 500 bp of the translation start site. All of the SPCG reporter gene constructs we built contain this consensus site. Thus, not only do we predict that the site is important for salivary gland expression of the SPCGs, but that this could be the site through which CrebA acts to elevate transcription. The proximal location of these putative binding sites with respect to the start site of translation is consistent with the finding that mammalian Creb proteins bind close to the start of transcription (Mayr and Montminy, 2001). Also of relevance to these studies was the failure to discover consensus Fkh-binding sites conserved among the SPCGs through MEME analysis, further supporting an indirect role for Fkh in SPCG regulation.

Our discovery that CrebA is required to achieve high-level expression of all early secretory pathway genes in the salivary gland suggests a predominantly physiological rather than morphogenetic role for this transcription factor. Such a role in secretion is consistent with the previously reported salivary gland defects in CrebA mutants, where a relatively mild defect was observed: the salivary glands of stage 13 to 15 CrebA mutants were, on average, slightly more crooked than those of their wild-type siblings (Andrew et al.,1997). To determine directly if defects in salivary gland secretion occur with the loss of CrebA function, we stained collections of CrebA mutant embryos with a polyclonal Engrailed (En)antiserum that crossreacts with material found in both late secretory vesicles and the salivary gland lumen (Myat and Andrew, 2002). These embryos were simultaneously stained with an antibody to PH4αSG1, a resident ER protein (E.W.A. and D.J.A.,unpublished). At early stages, the localization of these markers in the CrebA mutants was indistinguishable from wild type. By embryonic stage 15, however, far fewer En-positive vesicles were observed in the salivary gland cells of CrebA mutants compared with their wild-type and heterozygous siblings (Fig. 5A, arrows). The difference in vesicle accumulation in salivary gland cells was even more pronounced at embryonic stage 17, where the number of En-positive vesicles in wild-type embryos increased substantially relative to earlier stages, yet few could be detected in the stage 17 CrebAmutants (Fig. 5B, arrows and arrowheads). These results indicate that loss of CrebA function is associated specifically with a decrease in the secretory vesicle population in late embryos, supporting a role for CrebA in the increased secretory function associated with the salivary gland.

We also tried to determine whether zygotic loss-of-function mutations in individual secretory pathway component genes had secretion defects similar to those observed with CrebA mutants. For these experiments, we obtained lethal P-element insertions in several early secretory genes: srpRβ, sec13, sec23 and sar1. The P-elements had inserted in the coding region (sec13), the 5′ UTRs(srpRβ and sec23) or the first intron (sar1),and thus, would be expected to either eliminate or severely attenuate gene function. The accumulation of En-positive secretory vesicles was not altered detectably in any of the single secretory pathway mutants(Fig. 5C; data not shown). These observations suggest that residual zygotic function in combination with maternally supplied mRNAs are sufficient to support salivary gland secretory function of these single secretory pathway components during embryogenesis.

CrebA has a crucial role in larval cuticle development; CrebAmutant first instar larvae are only about 40% the length of wild-type larvae,and have a faint cuticle with frequent large dorsal holes(Andrew et al., 1997). A close examination of the ventral denticles and dorsal hairs reveal morphologies of structures typically found in more lateral positions in wild-type cuticle,suggesting a role for CrebA in controlling dorsal/ventral cuticle patterning. Our findings that CrebA is required for robust SPCG expression in the epidermis prompted us to ask if some or all of the cuticle defects observed in CrebA mutants could be explained by reduced SPCG expression. Thus, we performed cuticle preparations of larvae carrying the same P-element mutations discussed in the previous section. Mutations in srpRβ, sec13, sec23 and sar1 result in cuticle defects largely indistinguishable from those of CrebA mutant larvae. The cuticles were on average only 40-50% the length of wild-type cuticles and the cuticles were faint (Fig. 6A). The mouthparts and filzkörper were barely detectable in dark-field low magnification images of the CrebA and SPCG mutant larvae(Fig. 6A) and only faint outlines of these structures, with very little pigmentation, could be seen in higher magnification phase images (Fig. 6B). The SPCG mutant larvae exhibited defects in ventral denticle and dorsal hair morphology that were similar to those previously described for CrebA mutants (Fig. 6C). Ventral denticles were frequently missing and those that remained were smaller and less pigmented on average than wild-type denticles. Dorsal hairs were often completely absent and, when present, showed alternations between naked cuticle and thin hairs, similar to those observed in more lateral domains of the dorsal cuticle of wild-type larvae. Although all of the SPCG mutants analyzed had the same range of cuticle defects, the srpRβ and sec23 phenotypes appeared somewhat milder than those of CrebA, sec13 and sar1, perhaps reflecting either how much gene function remains in the P-element mutants or differential persistence of the maternally supplied components. Interestingly, larvae mutant for the individual SPCG genes did not show the large dorsal holes frequently observed in CrebA mutants.

The defects in the larval cuticles of CrebA and SPCG mutants support a crucial role for these genes in cuticle secretion. In wild-type animals, secretion begins at about 12 hours after egg laying (AEL)(Martinez-Arias, 1993), which roughly corresponds to embryonic stage 15(Campos-Ortega and Hartenstein,1997). To determine if secretory defects could be detected at this earlier stage, we stained CrebA and wild-type embryos with antibodies to DSC73, a secreted protein expressed to high levels in all cells that form the denticles and hairs of the larva (D.J.A., unpublished). Wild-type stage 15 and older embryos showed high-level DSC73 staining in all epidermal cells that form the denticles and hairs (Fig. 6D; top two rows). Such consistent and uniform DSC73 staining was not observed in the CrebA mutants. In the ventral cuticle precursors, CrebA mutants showed variable DSC73 staining, with levels ranging from almost wild type to barely detectable(Fig. 6D; bottom two rows). DSC73 levels in the dorsal cuticle were only slightly diminished, if at all,when compared with wild-type levels. DSC73 staining did reveal a significant lag in dorsal closure in the CrebA mutants, a defect likely to be linked to the dorsal holes frequently observed in the CrebA larval cuticles. Thus, defects in secretion in both the salivary gland and epidermis can be detected in CrebA mutants as early as 12 hours AEL, during embryonic stage 15.

This study demonstrates that the Drosophila salivary gland prepares soon after specification to generate the machinery for its high-level secretory activity. The machinery includes components of the early secretory pathway crucial for targeting and translocating proteins into the ER and for vesicle transport between the ER and Golgi. Thus, one way the gland distinguishes itself from surrounding tissues is to greatly increase the relative transcriptional levels of the secretory pathway component genes(SPCGs). The leucine zipper transcription factor CrebA has a crucial and probably direct role in activating increased levels of SPCG expression not only in the salivary gland, but also in the epidermal cells, which secrete the larval cuticle. Fkh, the Drosophila FoxA/PHA-4 homolog, is required to maintain SPCG expression in the salivary gland, but acts indirectly, by maintaining CrebA expression. Hkb, the other early transcription factor examined in this study, is not required for elevated SPCG expression.

CrebA is expressed at very high levels in the early salivary gland and this high level expression persists throughout larval life. Nonetheless,embryonic salivary glands in CrebA mutant embryos are relatively normal, showing only a mildly crooked phenotype when compared with the salivary glands of wild-type embryos(Andrew et al., 1997). This study indicates a role for CrebA in mediating salivary gland secretory function through the transcriptional upregulation of genes encoding early components of the secretory pathway (Figs 2, 4), supporting a physiological rather than morphogenetic role for this protein. Even so, in the CrebA mutants, where over 30 SPCGs are expressed at significantly reduced levels, effects on salivary gland secretion were not evident until late embryonic stages, when we observed a significant reduction of secretory vesicles (Fig. 5A,B). The late occurrence of overt defects in secretion in the CrebA mutants could reflect not only the increased secretory load on these cells that occurs only at the later embryonic stages, but also some level of maternal rescue of secretory function, as CrebA is provided maternally(Smolik et al., 1992). Interestingly, loss-of-function mutations in single secretory pathway component genes did not show the same loss of secretory activity observed in CrebA mutant salivary glands. The residual function of each of the individual SPCGs, from either maternal supplies or the remaining function of the P-element insertional alleles, appears to suffice when all other components are present at wild-type levels, at least with regards to salivary secretion during late embryonic stages.

CrebA mutants have major defects in cuticular development(Fig. 6); the larval cuticles are smaller and weaker than the cuticles of their wild-type siblings, the mouthparts and filzkörper are poorly formed, and CrebA mutants frequently have large holes in the dorsal cuticle(Andrew et al., 1997). In addition, there appears to be a general defect in patterning of the cuticle,with dorsal and ventral structures appearing more lateralized(Andrew et al., 1997). Embryos mutant for individual SPCGs, whose epidermal expression is also dependent on CrebA, had nearly identical defects in the larval cuticle(Fig. 6). The similarity in CrebA and the individual SPCG mutant cuticles suggests that CrebA defects could be entirely due to compromised secretory function in the epidermal cells that produce the cuticle. The lateralized appearance of the denticles and hairs could simply reflect compromised secretory function,which would limit the types of cuticular structures that form to the smaller,less pigmented structures that are characteristic of the lateral cuticle.

Our expression studies of the SPCGs indicate that CrebA could directly activate their high level expression. Moreover, we have discovered a conserved motif upstream of the SPCGs that is not only a good fit with the mammalian Creb-consensus binding site (TGACGTG G/T C/A; Table 3), but that also matches the first six nucleotides of the sequence that was used to discover CrebA (TGACGTCAG)(Smolik et al., 1992). However, the experiments of Smolik et al. were designed to discover the Drosophila homolog of the cAMP-regulated Creb protein, which turns out to be what is now known as CrebB (Usui et al., 1993). Gel shift experiments (EMSAs) indicate that CrebA can bind the to the TGACGTCAG consensus but not with the same high affinity and specificity as the mammalian cAMP-regulated Creb protein(Smolik et al., 1992); thus,CrebA may bind instead with high affinity to the site discovered in our MEME motif search of the regions upstream of the SPCGs to regulate their expression. The CrebA-dependent SPCG enhancers we have characterized so far(for z-cop; sec61β and spase25) contain at least two copies of the consensus motif.

Fkh has several roles in salivary gland development and function, including mediating the cell shape changes of invagination(Myat and Andrew, 2000b),maintaining secretory cell viability (Myat and Andrew, 2000b) and transcriptional activation of the sgs genes in late larval life(Lehmann and Korge, 1996; Mach et al., 1996). In addition to these positive roles, FKH also represses the expression of salivary duct-specific genes in the secretory cells(Haberman et al., 2003). Here,we discover yet another role for fkh in the salivary gland: the maintenance of SPCG expression.

fkh is a direct transcriptional target of Scr and Exd(Ryoo and Mann, 1999) and the temporal expression of CrebA and the presence of consensus Scr/Exd-binding sites upstream of the gene suggest that CrebA may also be directly controlled by Scr and its co-factors. Late expression of CrebA, however, requires fkh(Myat and Andrew, 2000b), as does late expression of fkh itself(Zhou et al., 2001). Here, we show that Fkh functions directly to maintain CrebA expression in the salivary gland (Fig. 3). Based on the requirement for CrebA for expression of the SPCGs at all embryonic stages and the requirement for fkh only at late stages, our data support a model in which CrebA controls the expression of the SPCGs and Fkh is required only because of its role in maintaining CrebAexpression (Fig. 7). A direct test of this model would be to express CrebA in the salivary glands of embryos missing fkh function; this experiment, unfortunately,could not be carried out because Fkh-independent drivers capable of providing high-level salivary gland-specific expression of CrebA are not yet available.

As previously mentioned, a subset of the SPCGs that encode proteins required for retrograde vesicle transport from the Golgi to the ER were still expressed at low levels in CrebA mutants(Fig. 2D). However, in late but not early fkh mutants, expression of these genes was not above levels in surrounding tissues. We propose that the residual expression of the genes observed in the CrebA mutants would be controlled through other early transcription factor genes that, like CrebA, would require Scr and its co-factors for their initial expression and would require Fkh for maintaining late expression (Fig. 7). Taken together, our studies suggest that regulation of salivary gland genes does not fit the simple paradigm suggested by studies of the C. elegans pharynx. The genes that specify the salivary gland(Scr/Exd/Hth) are distinct from the genes that activate and maintain gene expression in the organ. Moreover, no single gene takes over for the organ-specifying genes as even Fkh, the PHA-4 homolog, is not required for expression of every salivary gland gene(Myat and Andrew, 2000b;Bradley et al., 2001). In cases where Fkh is required, it is often indirect,such as with the SPCGs. Fkh does appear to have direct roles, however, much later in development, as demonstrated by regulation studies involving the sgs glue genes (Lehmann and Korge, 1996; Mach et al.,1996). Thus, the involvement of Fkh in salivary gland development and function is complicated and more consistent with the complexity of gene regulation seen in the liver than that suggested for the C. eleganspharynx. The existence of a single `organ-specifying gene' may be more the exception than the rule.

Salivary gland expression of the srpRα lacZconstruct shown in Fig. 4C is completely lost in a CrebA mutant background, indicating that at least four CrebA-dependent salivary gland SPCG enhancers have been identified.

We thank C. Machamer for many helpful discussions. We thank A. Cheshire, L. Grevengoed, B. Kerman, C. Machamer, M. Vining and two anonymous reviewers for critical reading of the manuscript. We thank the Indiana Stock Center in Bloomington for the fly stocks and P. O'Farrell for the En polyclonal antiserum. This work was supported by an NIH grant to D.J.A. (NIH RO1 DE13899).