Drosophila MBF1 is a co-activator for Tracheae Defective and contributes to the formation of tracheal and nervous systems

Transcriptional regulation plays a major role in the expression of the genomic information during complex biological processes such as differentiation and development. The effect of binding of transcription factors to cis-acting DNA sequences must somehow be transmitted to RNA polymerase to ensure the initiation and maintenance of active transcription. A class of proteins called co-activators serve this role by recruiting the general transcription machinery and by modifying the chromatin structure(Näär et al., 2001;Roeder, 1991). Although sequence-specific DNA-binding transcription factors provide most of the spatial and temporal specificity of expression, the expressional and functional properties of co-activators are also likely to participate in determining the kinetics and efficiency of transcription.

Two classes of transcriptional co-activators have been described to date. The first class comprises proteins that possess or recruit enzymatic activities to modify chromatin proteins, e.g. by acetylation of histones. Resulting alteration in the chromatin structure causes a switch in the `state'of chromatin between transcriptionally inactive and active. Co-activators of the second class act more directly to recruit the general transcription machinery to a promoter where a transcription factor is bound. Among the latter class are TATA element-binding protein (TBP)-associated factors (TAFs)that are subunits of TFIID, and others that serve as adaptors to mediate the contact between transcription factors and the basal transcriptional complex. Although the importance of the first type co-activators for gene expression has been demonstrated (Akimaru et al.,1997; Brownell et al.,1996; Chakravarti et al.,1996; Grant et al.,1997; Kamei et al.,1996; Ogryzko et al.,1996; Waltzer and Bienz,1999), little is known about how the second class co-activators function in vivo, or how the two types interact to achieve elaborate regulation of gene expression.

Multiprotein bridging factor 1 (MBF1) was first identified from the silkmoth as a co-factor necessary for transcriptional activation in vitro by a nuclear receptor FTZ-F1 (Li et al.,1994; Takemaru et al.,1997). The ability of MBF1 to bind both FTZ-F1 and TBP suggested a mechanistic model in which MBF1 recruits TBP to a promoter carrying the FTZ-F1-binding site by interconnecting FTZ-F1 and TBP. The MBF1 sequence is highly conserved across species from yeast to human(Takemaru et al., 1997). In the yeast Saccharomyces cerevisiae, MBF1 functions as a co-activator of a bZIP transcription factor GCN4, by bridging between GCN4 and TBP in response to amino acid starvation(Takemaru et al., 1998).

To analyze the biological role of MBF1 in a multicellular organism, we characterized its ortholog in Drosophila. The Drosophila mbf1 gene partially rescued the phenotype of a S. cerevisiae mbf1^- mutant, establishing that both the structure and the function of MBF1 remained well conserved during evolution. In nuclear extracts of Drosophila embryos, we identified a bZIP protein Tracheae Defective (TDF; APT — FlyBase)(Gellon et al., 1997;Su et al., 1999;Eulenberg and Schuh, 1997) as a new partner of MBF1. We show that MBF1 mediates transcriptional activation by TDF through directly binding both TDF and TBP. Although mbf1 null mutants can survive to adulthood, tdf becomes haploinsufficient inmbf1^- background, resulting in severe tracheal and central nervous system (CNS) defects. These results show that MBF1 acts as a crucial bridging co-activator during tracheal and CNS development.

mbf1 cDNA was cloned from a Drosophila cDNA library prepared from the 3rd instar larval CNS in the lambda gt10 vector (M. J.,unpublished). Partial cDNA was initially isolated by screening this library using nested PCR with degenerate antisense primers, corresponding to amino acids 123-129 (ERAIGIK) and 107-114 (YEAGRGIP) in silkmoth (Bombyx mori) MBF1 (Takemaru et al.,1997) (Fig. 1A) in combination with a primer derived from the vector. Hybridization screening of the same library with this PCR product was then used to isolate the full-length cDNA. The longest clone was 1562 bp, containing the entire MBF1 open reading frame plus a ∼1 kb long 3′UTR, which ends with a poly(A) tail. Subclones of this cDNA were used to construct expression vectors for producing bacterial fusion proteins and FLAG-MBF1 in transgenic flies.

A 4574-bp genomic EcoRI fragment encompassing the entirembf1 gene was subcloned from the P1 clone DS05624 into pBluescript II(Stratagene) and sequenced on both strands in its entirety (DDBJ/EMBL/GenBank Accession Number, AB031273). Comparison of this genomic sequence with thembf1 cDNA reveals that the gene consists of four exons, the first one being non-coding. The clone contains 2.3 kb of DNA upstream of thembf1 transcription unit and 54 bp downstream of the mbf1mRNA polyadenylation site.

Using P-element-mediated germline transformation(Rubin and Spradling, 1982),we made a transgenic Drosophila line expressing a FLAG epitope-tagged MBF1 (FLAG-MBF1) protein under the control of the constitutive hsp83promoter. Expression levels of the tagged MBF1 protein were comparable with those of endogenous MBF1 during embryogenesis and also in the salivary glands. An mbf1-null mutant was generated by isolating a P-element insertion in the mbf1 locus and its subsequent imprecise excision (M. J., M. Okabe, Y. H. and S. H., unpublished). A rescue constructP[MBF1⁺], mbf1 was made by P-element-mediated insertion of the 4574 bp mbf1 genomic sequence to the mbf1 mutant chromosome. tdf^P2 allele corresponds to the P-element insertion line l(2)k15608 (Torok et al., 1993). tdf^PΔ3 has a deletion that includes the first exon and 1 kb of the first intron of the tdfgene (Eulenberg and Schuh,1997). The sgP[Gal4] line has been described by Brand and Perrimon (Brand and Perrimon,1993). Transgenic lines harboring fPE-lacZ or its mutant derivatives have been described elsewhere(Han et al., 1998). Other stocks have been described previously(Lindsley and Zimm, 1992).

Embryos 0-20 hours after egg laying (AEL) were collected from a population of the hsp83-FLAG-MBF1 transgenic line reared at 25°C. Nuclear extracts were prepared from the embryos as described previously(Ueda et al., 1990). For isolation of FLAG-MBF1-associated proteins, nuclear extracts (15 ml) were loaded onto a 2 ml column of anti-FLAG M2-Agarose affinity gel (Sigma). The column was washed extensively with TBS (50 mM Tris-HCl, 150 mM NaCl, pH 7.4),until no proteins were detectable in the wash fraction. Bound proteins were then eluted with the FLAG peptide (50 μg/ml) (Sigma). Chromatography was carried out at 4°C.

To identify FLAG-MBF1 associated proteins, the eluted proteins were concentrated by acetone precipitation and then resolved on a preparative 10%SDS-PAGE. The proteins were transferred to a PVDF membrane (Boehringer Mannheim) on a Trans-Blot Semi-Dry apparatus (BioRad). The membranes were subsequently stained with Coommassie Brilliant Blue and selected bands were subjected to N-terminal sequencing. Peptide sequence of 63 kDa band VNKQYSATDLEAFMKIAANWQNSN showed an unambiguous match with that of TDF.

For the production of anti-TDF serum, a 1455 bp cDNA fragment encoding the entire TDF was inserted into a pET-28a vector (Novagen) and the polyhistidine-tagged protein was expressed in Escherichia coli BL21(DE3). The His-TDF protein was purified by Ni²⁺ affinity chromatography and used to generate polyclonal antibodies in rabbit. The serum was used directly for immunostaining at a 1000-fold dilution. For western analyses, it was used at a 10,000-fold dilution. The production of the MBF1 antibody will be described elsewhere (M. J., M. O., Y. H. and S. H.,unpublished). This antibody detects a single band of 16 kDa in normal embryos,but not in mbf1^- embryos(Fig. 3).

Other antibodies include a rabbit anti-Drosophila TBP antibody (a gift from Yoshihiro Nakatani); mouse anti-β-galactosidase mAb 40-1a;rabbit anti-β-galactosidase antibody (Cappel); mouse anti-FLAG M2(Kodak); mouse mAb BP102; and mouse mAb 2A12 that recognizes an unknown luminal component of the trachea (a gift from Shigeo Hayashi).

The secondary antibodies used were as follows: biotinylated anti-mouse IgG;Cy3-conjugated anti-rabbit IgG; Cy3-conjugated anti-mouse IgG (Jackson Laboratory); and Alexa 488 goat anti-rabbit IgG conjugate (Molecular Probes). When biotinylated anti-mouse IgG was used for the 2A12 staining, signals were amplified by adding biotinylated tyramide (NEN Life Science Products) as a substrate, followed by application of the ABC kit (Vector Lab), and finally visualized by Cy3-conjugated streptavidin (Amersham Biosciences).

For staining of embryos, 0-12 hours or 12-15 hours AEL embryos were collected and fixed as described (Tautz and Pfeifle, 1989). After blocking with 5% normal goat serum/PBS-0.1% Triton X-100, the embryos were incubated with a primary antibody, followed by a secondary antibody.

Staining of larval tissues was performed as described previously(Liu et al., 2000). Larvae were dissected in PBS, fixed in 25 mM PIPES-KOH (pH 7.0), 0.5 mM EDTA, 0.25 mM MgSO₄ and 4% formaldehyde for 40 minutes on ice and then permeabilized for 15 minutes at room temperature in PBS containing 0.5% NP-40. In the case of salivary glands, the dissected tissues were first treated with PBS-0.5% NP-40 for 8 minutes before the fixation and the permeabilization step after the fixation was omitted.

Staining of polytene chromosomes using indirect immunofluorescence was performed as described (Shopland and Lis,1996) except that salivary glands were treated with PBS, 0.5%NP-40 for 5 minutes before fixation. FLAG-MBF1, TBP and TDF were detected with anti-FLAG M2, anti-TBP and anti-TDF antibodies, respectively, and visualized with the mouse Cy3- or rabbit Alexa 488-conjugated secondary antibodies. After staining, samples were washed with PBS, 0.2% NP-40, 0.2% Tween 20 and containing subsequently 300 mM NaCl and 400 mM NaCl.

GST fusion proteins were purified on Glutathione Sepharose 4B beads(Amersham Biosciences). GST pull-down assays were performed as described(Takemaru et al., 1998). Bound proteins were detected on western blots using either the anti-MBF1 or the anti-TBP antibody.

For co-immunoprecipitation assays, 20 μl of IgG-conjugated Dynabeads(Dynal, Oslo, Norway) containing 2 μg of IgG were incubated at 4°C for 2 hours with 40 μl of a primary antibody or the corresponding preimmune serum and then washed with TBS [50 mM Tris-HCl (pH 7.4), 60 mM NaCl]. Nuclear extracts were prepared from yw or mbf1 embryos (0-20 hours AEL) as described (Ueda et al.,1990) and 30 μl of the nuclear extract was added to the antibody-loaded beads. After incubation at 4°C for 2 hours, the beads were washed with TBS containing 0.01% NP-40. Bound proteins were eluted from the beads with SDS loading buffer, resolved by SDS-PAGE (8% for TBP and TDF, 12.5%for MBF1) and analyzed on a western blot.

The TDF-binding consensus was obtained from a pool of oligonucleotides containing 16-nucleotide random sequences in the middle as described(Gogos et al., 1992). Binding reactions were performed as described elsewhere(Pollock and Treisman, 1990). The binding mixture (25 μl) contained buffer E [20 mM Hepes (pH 7.9), 100 mM KCl, 20% glycerol, 0.2 mM EDTA, 0.5 mM PMSF, 0.5 mM DTT, 0.1% NP-40, 5μg/ml leupeptin, 5 μg/ml pepstatin and 0.75 μg/ml aprotinin], 1 μl anti-TDF, 2 μl nuclear extract (0-20 hours AEL), 0.1 pmole of the³²P-labeled oligonucleotide probe, 200 ng poly(dIdC)-poly(dIdC) and 200 ng bovine serum albumin (Sigma). The mixture was incubated on ice for 30 minutes and oligomers bound by TDF were collected on protein A-Sepharose beads. After washing the beads with buffer E, the oligomers were recovered by phenol treatment, amplified using PCR and then used for the next selection cycle. After the fifth round of selection, DNA fragments were subcloned into a T-vector (Novagen) and sequenced. The consensus sequence was deduced from sequences of 54 clones.

Gel mobility retardation assays were performed as described previously(Ueda and Hirose, 1991),except that electrophoresis was carried out on a 4% polyacrylamide gel in 50 mM Tris-HCl (pH 7.5), 2 mM EDTA, 380 mM glycine and 5% glycerol. The double-stranded oligomers used for the mobility retardation assay are described below. The probe or functional competitor was 5′-GAGCTCGGATCCGAATTCCAATTGGAATCTCGAGAAGCTTGATCA G-3′.

Mutant competitor was 5′-GAGCTCGGATCCGAATTCCCCTTGTGATCTCGAGAAGCTTGATCAG-3′. Underlined regions indicate the recognition sequence of TDF and its mutant derivative respectively.

Reporter constructs were fusions of the lacZ gene with three tandemly repeated copies of either the functional TDF-binding site(TDS-lacZ) or the mutated site (TDMS-lacZ)(Fig. 4A), inserted into the hsCaSpeR-AUG-β-gal vector. The functional and mutant binding sites were those used in the DNA binding assays (above). These constructs were introduced into the yw host by the P-element transformation method of Rubin and Spradling (Rubin and Spradling,1982). A line carrying the TDS-lacZ reporter at the cytological position 34F on the second chromosome was used for most experiments. The position of the reporter gene was determined by circularization of Sau3AI-digested genomic DNA with DNA ligase,followed by sequencing of PCR-amplified fragment with P-element primers.

To examine the effect of the loss of tdf function on the reporter expression, the TDS-lacZ transgene was introduced into thetdf^P2 chromosome. The TDS-lacZ expression inmbf1^- and heterozygous tdf^P2background were examined in TDS-lacZ/CyO; mbf1 and TDS-lacZ,tdf^P2/CyO; mbf1, respectively. β-Galactosidase activities in embryonic extracts were measured using a Galacto-Light Plus System (Applied Biosystems) and a luminometer LUMAT LB9507 (Berthold Technologies). For scoring of X-gal-stained embryos, we excluded embryos that died earlier and only counted embryos that developed to stage 16.

To misexpress TDF in the salivary glands, a UAS-tdf transgene was controlled by a Gal4 driver sgP[Gal4] on the X chromosome.sgP[Gal4] is one of the enhancer trap lines isolated by Brand and Perrimon (Brand and Perrimon,1993) and is active in the salivary gland cells from embryonic through larval stages. To localize TDF on its binding site in vivo, theTDS-lacZ reporter was introduced into the UAS-tdf chromosome and the y⁺sgP[Gal4]/Binscinsy females were crossed with the yw:UAS-tdf, TDS-lacZ males. After culturing at 18°C, y⁺ third-instar larvae were dissected and stained with anti-TDF serum and the anti-β-gal mAb 40-1a.

To analyze genetic interactions between mbf1 and tdf, we made two double mutant stocks, homozygous for mbf1 deletion:tdf^PΔ3/CyOP[ftz-lacC]; mbf1 andtdf^P2/CyOP[ftz-lacC]; mbf1. To compensate for thembf1 deletion, the rescue construct P[MBF1⁺] was introduced into the mbf1 mutant chromosome. A double mutanttdf^PΔ3/CyO; P[MBF1⁺], mbf1 was also prepared and used for rescue experiments. To rescue the CNS phenotype of thetdf mutant, UAS-tdf was introduced into thetdf^PΔ3 chromosome. elav-Gal4;tdf^PΔ3/CyO females were crossed withUAS-tdf, tdf^PΔ3/CyO males, then embryos were collected at 12-15 hours AEL and stained with the mAb 2A12 and mAb BP102 antibodies. The ability of Drosophila MBF1 to rescue the yeastmbf1 disruptant was tested using aminotriazole sensitivity assay. Yeast MBF1 mediates transcriptional activation by GCN4 that governs induction of many genes in the amino acid synthetic pathways; hence, a yeastmbf1 disruptant is sensitive to aminotriazole, an inhibitor of theHis3 gene product (Takemaru et al., 1998).

A cDNA encoding Drosophila MBF1 was cloned from a larval CNS library. The predicted protein of 145 amino acids had 44, 64 and 83% identity to MBF1 from yeast, human and silkmoth, respectively(Fig. 1A). MBF1 consists of two structural domains: a well-structured C-terminal half that binds the general transcription factor TBP; and a flexible N-terminal half that participates in binding to various activators (Ozaki et al., 1999). The region conserved in Drosophila MBF1 includes both of these functional domains(Fig. 1A)(Ozaki et al., 1999). Expression of Drosophila mbf1 cDNA partially rescued the yeastmbf1 mutant phenotype upon amino acid starvation(Fig. 1B), indicating that the ability to bind partner transcription factors is also conserved between yeast and Drosophila.

In situ hybridization revealed that a large amount of maternalmbf1 mRNA was deposited to the egg (data not shown). Likewise, MBF1 protein was present in preblastoderm embryos and was later expressed in many tissues, including the CNS and the trachea(Fig. 1C). Widespread expression of MBF1 was also seen in post-embryonic stages, with particularly high levels in the larval salivary glands, gonads(Fig. 1C) and adult gonads(data not shown).

To address the role of MBF1 in vivo, we generated a null mutant ofmbf1 (see Materials and Methods). This allele lacks a 2082 bp DNA segment encompassing the entire coding region and most of the 3′-noncoding region of mbf1 (from 211 bp upstream of the initiation codon to 857 bp downstream of the stop codon).mbf1^- animals derived from mbf1^-/+parents were viable, with no detactable loss in viability. mbf1mutant females, however, had a maternal effect phenotype with incomplete penetrance; a variable fraction of progeny from mbf1^-homozygous females died before stage 5, regardless of whether they were mated to mbf1⁺ or mbf1^- males. As escapers developed to adults with no obvious morphological defects, a stablembf1 homozygous line could be maintained under laboratory conditions. No MBF1 protein was detectable in the mbf1 mutant by western analyses(Fig. 3A, lane 2;Fig. 3B, lane 2). Based on the requirement of MBF1 as an co-activator for FTZ-F1-dependent transcription in vitro, we had expected to see a ftz-f1-like segmentation defect among the dead mbf1 homozygous embryos. However, no such pair-rule phenotype was observed. Furthermore, we were unable to detect any decrease in the expression of a FTZ-F1-dependent reporter gene inmbf1^- mutant embryos(Fig. 1D). These results showed that MBF1 is not a crucial co-activator for FTZ-F1-dependent transcription in vivo. This led us to search for new partners of MBF1.

To isolate new partners of MBF1 in vivo, we used the transgenic line expressing FLAG-MBF1 and pulled out MBF1-associated factors from embryonic nuclear extracts. Complexes including MBF1 were captured on anti-FLAG antibody beads, eluted with the FLAG peptide and resolved by SDS-PAGE. While the nuclear extract of a strain lacking FLAG-MBF1 yielded no specific protein bands (Fig. 2A, lane 1), many proteins were recovered from the FLAG-MBF1 transgenic embryos(Fig. 2A, lane 2). As anticipated, TBP was identified among the MBF1-associated proteins(Fig. 2B, lane 4). The above procedure is also expected to pull down TAFs; indeed some of the bands had sizes corresponding to Drosophila TAFII32, TAFII60, TAFII110 and TAFII150, but we did not examine them further.

We focused on the most prominent protein at 63 kDa that could not be accounted for by any TAF. Its N-terminal amino acid sequence identified it as the bZIP protein TDF (Gellon et al.,1997; Eulenberg and Schuh,1997). This was supported by western analysis with an anti-TDF antibody (Fig. 2B, lane 6). As yeast MBF1 has been known to interact with the bZIP protein GCN4(Takemaru et al., 1998), TDF appeared as a good candidate for a new functional partner of MBF1. Indeed, GST pull-down assays using bacterially expressed purified proteins showed that MBF1 bound directly to TDF (Fig. 2C) and to TBP (Fig. 2D). Yeast GCN4 binds MBF1 with its bZIP domain(Takemaru et al., 1998). Similarly, the bZIP domain of TDF was sufficient for the binding ofDrosophila MBF1 (Fig. 2C).

If MBF1 bridges TDF and TBP, association between these two proteins should depend on the presence of MBF1. To examine the requirement of MBF1 for tethering TBP to TDF in vivo, we carried out co-immunoprecipitation using nuclear extracts prepared from mbf1⁺ or mbf1-null mutant embryos. From the mbf1⁺ nuclear extract, MBF1 and TDF were co-immunoprecipitated with the anti-TBP antibody, but not with the pre-immune serum (Fig. 3A,lanes 3 and 5). By contrast, when we started from the extract frommbf1^- embryos, TDF was not co-immunoprecipitated with the anti-TBP antibody (Fig. 3A,lanes 2 and 6). In a reciprocal test, MBF1 and TBP were co-immunoprecipitated with the anti-TDF antibody from the mbf1⁺, but not from the mbf1^- nuclear extracts, or from thembf1⁺ extract using the pre-immune serum(Fig. 3B, lanes 3, 5 and 6). These results demonstrate that MBF1 forms a bridge between TDF and TBP and that in its absence, a stable association of TDF and TBP does not take place.

The physical interaction between MBF1 and TDF, as well as the presence of bZIP domain in TDF suggests that TDF is a transcription factor that uses MBF1 for transcriptional regulation of its target genes. To test this possibility,we first determined the binding sequence of TDF. The full-length TDF protein or its C-terminal 56 amino acids harboring the bZIP domain were expressed in bacteria and purified. In a preliminary gel mobility retardation experiment,both the full-length and the C-terminal proteins showed DNA-binding activities(data not shown). The optimal binding sequences were selected from a pool of random oligonucleotides (see Materials and Methods). The consensus sequence deduced from the selected oligonucleotides was (A/G) TTC (C/T)(A/T) AT (T/A)(G/A) GA (A/T)(T/C) (Fig. 4A).

The specificity of TDF binding to the consensus DNA sequence was verified by a competition assay. Incubation of ³²P-labeled double-stranded DNA carrying the sequence 5′-ATTCCAATTGGAAT-3′ with the full-length TDF protein yielded a slowly migrating complex. The complex formation was efficiently competed with the unlabeled oligonucleotide of the same sequence (Fig. 4B, lanes 2-5) but not with a mutant oligonucleotide carrying four base substitutions(Fig. 4B, lanes 6-8). Essentially the same results were obtained with the C-terminal 56 amino acid fragment of TDF (data not shown). These data indicate that TDF is a sequence-specific DNA binding protein that recognizes a 14 bp palindrome (A/G)TTC (C/T) (A/T) AT (T/A) (G/A) GA (A/T) (T/C).

We then tested whether TDF regulates transcription from a promoter carrying its binding site. A reporter gene was constructed in which three tandemly repeated TDF-binding sites (TDS) were placed upstream of a hsp70basal promoter (-40 ∼ +90)-lacZ fusion gene (TDS-lacZ;Fig. 5A). We also made another reporter gene containing three tandemly repeated mutant binding sites(TDMS-lacZ). Transgenic fly lines carrying these reporter genes were analyzed for lacZ expression by staining with anti-β-galactosidase antibody. TDS-lacZ was expressed in the tracheae, heart precursor cells, head and CNS of stage 16 embryos(Fig. 5B,C), i.e. in the pattern of the endogenous TDF protein(Eulenberg and Schuh, 1997;Gellon et al., 1997;Su et al., 1999) (Q.-X. L., M. J., H. U., Y. H. and S. H., unpublished). By contrast, the TDMS-lacZreporter with mutated binding sites showed no activity(Fig. 5D). Expression ofTDS-lacZ was not detectable in the tdf loss-of-function mutants tdf^P2 (Fig. 5E) and tdf^PΔ3 (data not shown). Conversely, when TDF was ectopically expressed in the posterior salivary gland cells, it induced expression of the TDS-lacZ reporter in these cells(Fig. 5I). Such induction oflacZ was not observed when we used the line carryingTDMS-lacZ (data not shown). These results clearly show that TDF is required and sufficient to activate transcription by binding to its specific recognition sequence TDS.

If MBF1 and TDF form a complex to activate the reporter gene, both proteins should localize to TDS. When TDF was ectopically expressed in the salivary glands of a transgenic line harboring a TDS-lacZ insertion at the polytene chromosome position 34F, TDF was detected at the TDS-lacZinsertion site. Visualization of MBF1 using FLAG-MBF1 strain revealed that MBF1 also accumulates at 34F, colocalizing with TDF(Fig. 6A). The localization of MBF1 to the TDS site at the position 34F occurred only when TDF was expressed in the salivary gland. These data support the idea that MBF1 is recruited to TDS owing to the sequence-specific DNA-binding activity of TDF.

As animals that lack MBF1 can survive to adulthood while null mutants of TDF are embryonic lethal, TDF must be able to carry out its function even when no stable association with TBP occurs. This suggests that, in the absence of MBF1, transcriptional activation by TDF may be reduced, but not abolished. To measure the transcriptional activity of TDF, we quantitated the expression of the TDS-lacZ reporter in normal and MBF1-deficient embryos(Fig. 6B). We found thatβ-galactosidase expression from TDS-lacZ was reduced by 80% when MBF1 was absent. This effect was reverted by expressing a transgenic MBF1(data not shown). The reduction was even more pronounced when only one copy of the tdf⁺ gene was present, while removing one copy of thetdf⁺ gene in mbf1⁺ background had no effect on TDS-lacZ reporter activity. This indicates that the loss of MBF1 has a quantitative effect on TDF-mediated transcriptional activation. Interestingly, when a population of stage 16 mbf1^- embryos carrying TDS-lacZ gene was stained for β-galactosidase expression, there was a considerable individual variation in the expression levels; about 15% of embryos had no detectable levels of β-galactosidase expression, while the rest exhibited varying levels. Such variation was not observed in mbf1⁺ embryos. It is possible that MBF1 function becomes essential for TDF-mediated transcription under conditions that are not completely controlled in our experiments.

The involvement of MBF1 in TDF-mediated transcription suggests that loss ofmbf1 may also affect a tdf mutant phenotype. In addition to the tracheal system (Eulenberg and Schuh,1997; Gellon et al.,1997), TDF is strongly expressed in the CNS(Eulenberg and Schuh, 1997). Although a defect in neural function has been reported in a tdfmutant (Takasu-Ishikawa et al.,2001), the role of TDF in the CNS development remained to be investigated. We found that tdf^PΔ3 embryos had breakages in the CNS axon tracts and tracheal networks(Fig. 7C,D). CNS and tracheal phenotypes were individually rescued when TDF was expressed in all neurons or in all tracheal cells, respectively, indicating that these phenotypes represent independent requirement for TDF in two tissue types, rather than one being a secondary consequence of another(Fig. 7E,F)(Eulenberg and Schuh, 1997)(Q.-X. L., M. J., H. U., Y. H. and S. H., unpublished). The breakages in the CNS and tracheal networks were also observed in the mbf1 mutant,although the penetrance of these phenotypes was only 2-3%(Fig. 7G,H).

Although the developmental defect caused by the lack of MBF1 was rather subtle, mbf1 mutation exhibited a strong genetic interaction withtdf mutants. Animals that were heterozygous for thetdf^PΔ3 allele had no detectable defects in either the CNS or tracheae (Fig. 7A,B). However, when MBF1 was removed from thetdf^PΔ3 heterozygotes, we observed a significant increase in the number of defective embryos (17-18% inFig. 7I,J). Moreover, the degree of defects was also enhanced: increase in the number of breakages and in their size (Fig. 7I,J versusFig. 7G,H). This effect was completely reverted by expressing MBF1 from a transgene carrying the genomicmbf1 locus (Fig. 7K-N). These genetic interactions between mbf1 andtdf strongly suggest that MBF1 participates in TDF-mediated transcription during normal development of the tracheae and CNS.

We have analyzed the biological function of an evolutionarily conserved transcription co-activator MBF1 through identifying its partner protein TDF. MBF1 bound directly to TBP and TDF and colocalized with TDF on a TDF-binding site experimentally placed on a polytene chromosomal locus. Co-immunoprecipitation experiments demonstrated that a TDF-MBF1-TBP ternary complex occurred and that MBF1 was required for bridging between TDF and TBP. In the absence of MBF1, the association between TDF and TBP was weakened and TDF-dependent expression of a reporter gene was reduced by 80%. Loss ofmbf1 function resulted in developmental defects similar to those seen in tdf mutants and affected the proper formation of the tracheae and CNS. These phenotypes were enhanced when the gene dose of tdf was halved. Based on these biochemical and genetic data, we conclude thatDrosophila MBF1 functions as a co-activator of TDF during the development of the tracheae and CNS.

The relationship between Drosophila MBF1 and TDF is similar to that between yeast MBF1 and its partner transcription factor GCN4(Takemaru et al., 1998). Just as yeast MBF1 contacts GCN4 through its bZIP domain, Drosophila MBF1 binds the bZIP domain of TDF. Moreover, the lack of GCN4-dependent activation in yeast mbf1 mutant can be partially restored by expressingDrosophila MBF1. The sequence and functional conservation between yeast and Drosophila MBF1 indicates that the interaction with bZIP proteins is a conserved feature of the bridging factor MBF1.

Genetic studies of mbf1 in yeast and Drosophila suggest that MBF1-associated transcription factors have two pathways for activation. In addition to the MBF1-mediated recruitment of TBP via its bZIP domain(Takemaru et al., 1998), GCN4 also recruits the SAGA complex with its N-terminal activation domain and effects transcription through chromatin modification(Grant et al., 1997). Likewise, Drosophila TDF has a region similar to the glutamine-rich transactivation domain (Eulenberg and Schuh, 1997) and may employ an activation pathway independent of recruiting TBP through MBF1. Such pathway may account for the residual expression of the TDS-lacZ reporter gene in the absence of MBF1. Although MBF1 is essential for GCN4-dependent transcription of its target geneHIS3 (Takemaru et al.,1998), low level of TDF-dependent transcription of theTDS-lacZ gene can still occur in the absence of MBF1. This suggests that the relative importance of the two pathways is different between GCN4 and TDF. The DNA-binding domain of FTZ-F1 carries a basic region homologous to those in bZIP proteins and binds MBF1 through this region(Takemaru et al., 1997). However, loss of mbf1 showed no effect on FTZ-F1-dependent transcription in vivo, suggesting that the activation by FTZ-F1 relied solely on the pathway through its transactivation domain. In our in vitro transcription system, the transactivation domain does not seem to be functional because FTZ622 polypeptide bearing only the DNA-binding domain of FTZ-F1 showed the same transcriptional activity as the intact FTZ-F1(Takemaru et al., 1997). This may explain the difference in the MBF1 requirement between FTZ-F1-dependent transcription in vivo and in vitro.

It is possible that the role of MBF1 becomes more critical under certain circumstances, when rapid induction of gene expression is demanded by environmental conditions. The expression of the TDS-lacZ reporter inmbf1^- background varied considerably from embryo to embryo, suggesting that certain conditions that are uncontrolled in our experiments may render transcription particularly dependent on the MBF1-mediated pathway. In the natural environment, there are many stimuli that alter gene expression profile: UV radiation, poison agents, nutrient starvation and so on. Therefore, direct recruitment of TBP by MBF1 may become essential for rapid activation of transcription under such conditions. In agreement with this idea, the yeast mbf1 disruptant is viable under normal culture conditions, but sensitive to amino acid starvation(Takemaru et al., 1998).

Studies on MBF1 homologs also support the idea that MBF1 may function when gene expression is required in response to developmental or environmental signals. Rat MBF1 has been isolated as a calmodulin-associated peptide 19(CAP-19) (Smith et al., 1998)and human MBF1 (Kabe et al.,1999) has been identified as endothelial differentiation-related factor 1 (EDF1) (Dragoni et al.,1998). EDF1/MBF1 is downregulated when endothelial cells are induced to differentiate. Interestingly, EDF1/MBF1 binds to calmodulin in the cytoplasm under low Ca²⁺ conditions but the two proteins dissociate when intracellular Ca²⁺ is high. The released EDF1/MBF1 is then phosphorylated and shuttled into the nucleus, where it binds TBP(Mariotti et al., 2000). Nuclear translocation of MBF1 has also been observed at a specific stage of molting in the silkworm B. mori(Liu et al., 2000). Considering the Ca²⁺ elevation upon exposure to the molting hormone ecdysteroid (Biyasheva et al.,2001), these data raise an intriguing possibility that MBF1 is involved in Ca²⁺-induced gene activation. Although in this study we have analyzed only the developmental roles of MBF1 associated with TDF function, Drosophila MBF1 may also be involved in other biological processes, such as stress response, homeostasis and longevity.

Several lines of evidence suggest that Drosophila MBF1 has partners other than TDF. MBF1 is expressed in a wide spatiotemporal pattern,including tissues and stages where TDF is absent. Although TDF is not expressed in the salivary gland, immunolocalization of MBF1 on salivary gland chromosome revealed a large number of loci associated with MBF1 (Q.-X. L., M. J. and S. H., unpublished). Furthermore, FLAG-tagged MBF1 pulled down many proteins besides TDF. Although mbf1-null mutants were viable under laboratory conditions, tdf became haploinsufficient inmbf1^- genetic background, clearly indicating the importance of MBF1 in the expression of the genomic information. This finding opens a way to identify new partners of MBF1 through genetic screening for loci that exhibit dominant phenotypes in the absence of MBF1. Characterization of MBF1 partners will contribute to our knowledge on how co-activators mediate specific biological events.

We thank Shigeo Hayashi, Masataka Okabe, Reinhard Schuh, Leslie Pick and Yoshihiro Nakatani for kindly supplying flies, DNA and antibodies; and Jun-ichi Tomizawa for critical reading of the manuscript. This work was supported by Grants-in aid for Scientific Research from the Ministry of Education, Science, Sports, Culture and Technology of Japan and grants from the Naito Foundation and from the Czech Grant Agency number204/00/0811. Q.-X. L. and M. J. were supported by a Center of Excellence Program of Japan. Q.-X. L. was also supported by Japan Society for the Promotion of Science.