Two adjacent protein binding motifs in the cbh2 (cellobiohydrolase II-encoding) promoter of the fungus Hypocrea jecorina (Trichoderma reesei) cooperate in the induction by cellulose.

The cellulase system of the filamentous fungus Hypocrea jecorina (Trichoderma reesei) consists of several cellobiohydrolases, endoglucanases, and beta-glucosidases, encoded by separate genes, which are coordinately expressed in the presence of cellulose or the disaccharide sophorose. Using cell-free extracts from sophorose-induced and noninduced mycelia and various fragments of the cbh2 promoter of H. jecorina in electrophoretic mobility shift assay (EMSA) analysis and performing in vitro and in vivo footprinting analysis, we detected the nucleotide sequence 5'-ATTGGGTAATA-3' (consequently named cbh2-activating element (CAE)) to bind a protein complex with different migration in EMSA of induced and noninduced cell-free extracts. EMSA analysis, employing oligonucleotide fragments containing specifically mutated versions of CAE, revealed that protein binding requires the presence of an intact copy of either one of two adjacent motifs: a CCAAT (=ATTGG) box on the template strand and a GTAATA box on the coding strand, whereas a simultaneous mutation in both completely abolished binding. H. jecorina transformants, containing correspondingly mutated versions of the cbh2 promoter fused to the Escherichia coli hph gene as a reporter, expressed hph in a manner paralleling the efficacy of CAE-protein complex formation in EMSA, suggesting that the presence of either of both motifs is required for induction of cbh2 gene transcription. Antibody supershift experiments with anti-HapC antiserum as well as EMSA competition experiments with CCAAT binding promoter fragments of the Aspergillus nidulans amdS promoter suggest that the H. jecorina CCAAT box binding complex contains a homologue of HapC. The nature of the adjacent, GTAATA-binding protein(s) and its cooperation with the HapC homologue in cbh2 gene induction is discussed.

The cellulase system of the filamentous fungus Hypocrea jecorina (Trichoderma reesei) consists of several cellobiohydrolases, endoglucanases, and ␤-glucosidases, encoded by separate genes, which are coordinately expressed in the presence of cellulose or the disaccharide sophorose. Using cell-free extracts from sophorose-induced and noninduced mycelia and various fragments of the cbh2 promoter of H. jecorina in electrophoretic mobility shift assay (EMSA) analysis and performing in vitro and in vivo footprinting analysis, we detected the nucleotide sequence 5-ATTGGGTAATA-3 (consequently named cbh2-activating element (CAE)) to bind a protein complex with different migration in EMSA of induced and noninduced cell-free extracts. EMSA analysis, employing oligonucleotide fragments containing specifically mutated versions of CAE, revealed that protein binding requires the presence of an intact copy of either one of two adjacent motifs: a CCAAT (‫؍‬ATTGG) box on the template strand and a GTAATA box on the coding strand, whereas a simultaneous mutation in both completely abolished binding. H. jecorina transformants, containing correspondingly mutated versions of the cbh2 promoter fused to the Escherichia coli hph gene as a reporter, expressed hph in a manner paralleling the efficacy of CAE-protein complex formation in EMSA, suggesting that the presence of either of both motifs is required for induction of cbh2 gene transcription. Antibody supershift experiments with anti-HapC antiserum as well as EMSA competition experiments with CCAAT binding promoter fragments of the Aspergillus nidulans amdS promoter suggest that the H. jecorina CCAAT box binding complex contains a homologue of HapC. The nature of the adjacent, GTAATA-binding protein(s) and its cooperation with the HapC homologue in cbh2 gene induction is discussed.
Cellulose is the most abundant renewable carbon source on earth. Its recycling in nature occurs in a variety of habitats by the action of various pro-and eukaryotic microorganisms, which hydrolyse this homopolymer with extracellular enzyme systems (1). Among the best characterized of these is the cellulase system of the saprophytic Ascomycete Hypocrea jecorina (ϭTrichoderma reesei, because the teleomorphic form of T. reesei has now been identified by Kuhls et al. (2), we prefer to name the organism accordingly), which contains three classes of enzymes (3). 1,4-␤-D-Glucan cellobiohydrolases (CBH I and II; EC 3.2.1.91), which cleave cellobiosyl units from the nonreducing end of cellulose chains; endo-␤-1,4-glucanases (EG I, EG II, EG III, and EG V; EC 3.2.1.4), which cleave internal glucosidic bonds; and 1,4-␤-D-glucosidases (BG I, BG II; EC 3.2.1.21), which cleave cellooligosaccharides to produce glucose are the most prominent members.
H. jecorina cellulases are not formed during growth on easily metabolizable carbon sources but induced in the presence of cellulose (4). The mechanism by which this extracellular, insoluble polysaccharide triggers the biosynthesis of cellulases has been investigated for decades. Some workers claim a role for low constitutive levels of cellulases in the initial attack on cellulose, thereby releasing the inducer (5,6), whereas others presented evidence for an involvement of carbon catabolite derepression (7) or of conidial-bound cellulases (8) in this process. The nature of the actual inducer is still unknown, but depending on the conditions, either the cellulolytic end product cellobiose itself or its transglycosylation product sophorose induce cellulase formation in pregrown mycelia (3,9,7). Despite the progress that has been made in the cloning of genes encoding several H. jecorina cellulases (10 -14), little information is available about the transcriptional regulation of their induction. More recently, two laboratories described a nucleotide region, which is located closely upstream of the TATA box, to be necessary to trigger sophorose induction of cbh1 (cellobiohydrolase I-encoding) gene transcription (15,16), but neither the binding motif nor the proteins binding to it have as yet been characterized.
We have previously identified the area between Ϫ361 and Ϫ170 in the 5Ј regulatory sequences of the cbh2 (cellobiohydrolase II-encoding) gene to be able in vitro to form a sophorosespecific DNA-protein complex (17). Starting from these findings, we will show in this paper that within this area, a 5Ј-ATTGGGTAATA-3Ј undecamer (that we will call "cbh2activating element" (CAE) 1 ) is responsible for the observed complex formation in vitro and essential for cbh2 gene expression in vivo. Evidence will also be presented that a CCAAT box-binding protein (complex) and another yet unknown protein bind to adjacent sites within this motif.

EXPERIMENTAL PROCEDURES
Microbial Strains and Plasmids-T. reesei (H. jecorina) QM9414 (ATCC 26921) was used throughout this study. T. reesei (H. jecorina) TU-6 (18), a pyr4 negative mutant of QM9414, was used as a recipient strain for pyr4-mediated cotransformation experiments. The strains were maintained on malt agar (containing 5 mM uridine in the case of * This study was supported by a grant of the Austrian Science Fund (project P11701-MOB) (to C. P. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Fungal Growth and Induction of Cellulases-H. jecorina was grown in 250 ml of the medium described by Mandels and Andreotti (21) in 1-liter Erlenmeyer flasks on a rotary shaker (250 rpm) at 28°C using 108 conidia/liter (final concentration) as inoculum. Carbon sources were added as indicated for the respective experiments. For experiments with soluble inducers, the replacement technique described by Sternberg and Mandels (4) was used. Mycelia, pregrown for 24 h with 1% glycerol as the carbon source were washed and resuspended in minimal medium lacking carbon source to give a final density of 0.7-1.5 g dry weight/liter. Either sophorose (final concentration 2 mM) or glucose (final concentration 1%, w/v) were added, and incubation continued for a further 3 or 5 h, respectively.
Construction of pSMZ Reporter Plasmids-The pSMZ reporter plasmids were developed from plasmid pLMRex30 and contain the E. coli hph (hygromycin B-phosphotransferase-encoding) gene fused to the H. jecorina cbh2 5Ј-and 3Ј-noncoding regions. Primers used are specified in Table I. To construct pSMZ1 (bearing the cbh2 wild-type promoter), primers CKT 006 and P 91635 were used to amplify a 613-bp fragment from the cbh2 5Ј-noncoding sequences, thereby also generating additional XhoI and XbaI terminal sites. This polymerase chain reaction (PCR) fragment was used to replace a XhoI/XbaI fragment of the pki1 promoter in pRLMex30, thereby fusing the cbh2 promoter fragment to the E. coli hph gene, followed by the H. jecorina cbh2 terminator. To yield pSMZ2, pSMZ4, and pSMZ5 (bearing point-mutated cbh2 promoter sequences as shown in Fig. 4A), recombinant PCR as described previously by Higuchi (22) was applied. For pSMZ2, primer pairs CKT 006/Pcbh2mutR2, and P 91635/Pcbh2mutF2, respectively, were used to amplify the primary PCR products from plasmid pSMZ1 as template. To construct pSMZ4, primer pairs CKT 006/Pcbh2mutR4 and P91635/ Pcbh2mutF4, respectively, were used for the primary PCR, and for construction of pSMZ5, primers CKT 006/Pcbh2mutR5 and P916357/ Pcbh2mutF5, respectively, were used. For construction of all plasmids (pSMZ2, pSMZ4, and pSMZ5) the secondary PCRs were performed with primers CKT 006 and P 91635, resulting in 613-bp cbh2 promoter fragments bearing point mutations as shown in Fig. 4A, which were directly cloned in vector pCR 2.1 using the TA Cloning Kit (Invitrogen). From the resulting plasmids, the cbh2 promoter fragments were excised using restriction enzymes XhoI/XbaI and inserted into pSMZ1, thereby replacing the cbh2 wild-type promoter sequences. All vectors were verified by double strand sequencing by the dideoxynucleotide chain termination method (23), and the restriction enzyme sites generated were proven via restriction enzyme analysis.
Isolation and Manipulation of Nucleic Acids-Genomic DNA and mRNA were isolated as described by Gruber et al. (18) and Chomczynski and Sacchi (24), respectively. After electrophoretic separation, RNA was blotted onto nylon membranes (Hybond-N, Amersham Pharmacia Biotech) and hybridized according to standard protocols (25) at 42°C for 20 h. Washing was performed with 2ϫ SSC(1ϫ SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate) ϩ 0.1% SDS at 42°C(2 ϫ 15 min). Standard methods were used for plasmid isolation, restriction enzyme digestion, random priming and Southern analysis (25). All PCR amplifications were performed by the aid of Taq polymerase (Promega, Madison, WI) in a Biometra thermocycler.
DNA Transformations-E. coli transformations were carried out according to standard techniques (25). Transformation of T. reesei (H. jecorina) TU-6 was carried out as described by Gruber et al. (26) using cotransformation of pFG1 with linear 2.6-kb HindIII/EcoRI fragments of the pSMZ vectors (Fig. 4A).
Determination of Gene Copy Number and Integration Locus in Transformants-Southern hybridization was carried out as described by Sambrook et al. (25). Chromosomal DNA was digested with EcoRI, and hybridization was performed using a 862-bp EcoRI fragment of the vector pSMZ1 bearing 613 bp of the cbh2 promoter and 249 bp of the hph gene labeled with [␣-32 P]dCTP. Signals observed upon fluorography for the single-copy gene cbh2 and for intact, integrated vector copies, respectively, were calculated by imager analysis. Values were corrected by including the length of the labeled probe. The copy numbers were calculated by dividing the values observed by imager analyses through the length of the labeled probe. To confirm the point mutations in the cbh2 promoter (generating a DraI restriction enzyme site) in transformants bearing the pSMZ2 and pSMZ4 plasmids, chromosomal DNA was digested with DraI/XhoII/ClaI, and hybridization

5Ј-TGA AGC ATC ATC ACG ATG TGG TAT T-3Ј
(from Ϫ124 to Ϫ148) P2c 5Ј-GTA TTC GGG AGC ACA GGA TAA AGC C-3Ј was performed with a 2.6-kb XhoI/HindIII fragment of the vector pSMZ1 bearing the cbh2 promoter and terminator sequences and the hph structural gene labeled with [␣-32 P]dCTP.
Hygromycin B Resistance Assay-Transformants were analyzed for hygromycin B resistance by means of a plate assay essentially as described previously (27).
Electrophoretic Mobility Shift Assay (EMSA)-DNA fragments used for EMSA were generated as follows. A cbh2 fragment containing the regulatory nucleotide regions from Ϫ263 to Ϫ170 was prepared by PCR amplification of a 613-bp fragment of the cbh2 upstream regulatory sequences using primers pcbh2F and pcbh2R and subsequent digestion with DdeI, HhaI, and HpaII. A 94-bp fragment was isolated and endlabeled with [␣-32 P]dCTP using Sequenase version 2.0 (Amersham Pharmacia Biotech). After purification by nondenaturing polyacrylamide gel electrophoresis, binding was achieved by incubating 100 g of cell-free extracts with 5 ng of labeled fragment on ice for 10 min. Preparation of cell-free H. jecorina extracts following growth on various carbon sources was carried out as described previously (17). Because maximal accumulation of cbh1-and cbh2-mRNA was observed after 5 h of incubation with sophorose, we prepared protein extracts from mycelia harvested after 3 h. The binding assay and acrylamide gel electrophoresis have been described previously (17). For competition experiments, synthetic oligonucleotides were used in a 10-, 50-, or 150-fold molar excess. The oligonucleotides were annealed with the complementary synthetic oligonucleotide as described by Strauss et al. (28). After annealing, double strands were filled in using Sequenase version 2.0 (Amersham Pharmacia Biotech). Oligonucleotides used for EMSA were also annealed with their complementary oligonucleotides and end-labeled using Sequenase version 2.0 and [␣-32 P]dCTP as described above. The resulting double-stranded oligonucleotides were purified by nondenaturing polyacrylamide gel electrophoresis, and binding was achieved by incubating 100 g of cell-free extracts with 25 ng of labeled oligonucleotide. For supershift assays 5 l of polyclonal anti-HapC antiserum were added to the binding reaction.
In Vitro Footprinting Procedure-In vitro methylation protection and hydrazine interference footprinting was performed with a 191-bp DNA fragment containing upstream regulatory sequences of the cbh2 gene from Ϫ361 to Ϫ171. The fragment was generated by digestion of a 613-bp PCR product with HpaII, and the resulting 233-bp fragment was isolated and end-labeled on the coding strand using Sequenase version 2.0 (Amersham Pharmacia Biotech) and [␣-32 P]dCTP. After labeling, the probe was digested with DdeI, and the resulting 191-bp fragment was purified by nondenaturing polyacrylamide gel electrophoresis. Labeling on the noncoding strand was performed by cleaving the 233-bp HpaII fragment with DdeI and labeling the digestion reaction with [␣-32 P]dATP using Sequenase version 2.0, and purification of the resulting 191-bp fragment by nondenaturing gel electrophoresis.
In vitro methylation protection footprinting was performed by incubating 3 ϫ 10 6 counts of end-labeled DNA fragment with 100 g of cell-free extract as described above, methylating with dimethylsulfate, and subjecting the mixture to EMSA analysis. After separation by native gel electrophoresis, complexes were visualized by autoradiography, excised, eluted, purified, and cleaved with piperidine (G) or 0.5 M HCl, 0.1 M NaOH, 1 mM EDTA (A/G) as described previously (29). After resuspension in loading buffer, the samples were applied to denaturing polyacrylamide gel electrophoresis on a 6% sequencing gel as described by Strauss et al. (28).
In vitro hydrazine interference footprinting was performed by incubating 300,000 counts of end-labeled, hydrazine-treated DNA fragment (25) with 100 g of cell-free extract and separating the bound and free probes by native polyacrylamide gel electrophoresis. After elution and purification, the samples were cleaved with piperidine, suspended in loading buffer, and loaded on a 6% sequencing gel.
In Vivo Genomic Footprinting via Ligation-mediated PCR-T. reesei (H. jecorina) QM9414 was cultivated as described above. Methylation of genomic DNA was performed at 30°C in a shaking water bath by transferring 18-ml aliquots of the cultures to 100-ml Erlenmeyer flasks at the indicated time points and incubating them with 40 l of dimethyl sulfate in 2 ml of 200 mM MES buffer (pH 5.5) for 2 min. Methylation was stopped by addition of 50 ml of ice-cold TLE␤ buffer (10 mM Tris (pH ϭ 8), 1 mM EDTA, 300 mM LiCl, 2% (v/v) ␤-mercaptoethanol). Mycelial samples were filtered and washed twice with 50 ml of TLE␤ buffer, and genomic DNA was extracted following a standard protocol. The extracted methylated DNA was cleaved at guanine bases by incubation with 20 l of piperidine for 30 min at 90°C followed by three sequential ethanol precipitations and was finally dissolved in Tris-EDTA at a concentration of 0.4 mg/ml.
To cleave guanine as well as adenine bases, extracted methylated DNA was incubated on ice for 2 h with 25 l of 1 M HCl. After precipitation with ethanol, the DNA was dissolved in 180 l of bidistilled water and incubated at 90°C for 30 min with 20 l of 1 M NaOH, followed by a second precipitation with ethanol. Finally, DNA was dissolved in Tris-EDTA at a concentration of 0.4 mg/ml. In vitro methylation and cleavage of genomic DNA was performed as described by Mueller and Wold (30). Methylated and cleaved DNA was analyzed by ligation-mediated PCR as described by Garrity and Wold (31) and modified by Wolschek et al. (32) using Vent polymerase (New England Biolabs, Beverly, MA) and primer sets P1c, P2c, and P3c for visualizing the coding strand and P1n, P2n, and P3n for visualizing the noncoding strand (cf. Table I).

RESULTS
Delimitation of the cbh2 Promoter Area, Which Binds a Protein Complex Unique for Sophorose Induction-When H. jecorina is pregrown on a carbon source not inducing cellulase formation (glycerol) and then replaced to a medium containing the inducer sophorose, a transient accumulation of cbh1 and cbh2 mRNA is triggered (Fig. 1A). This pattern of expression is also paralleled by that of transcript formation of other cellulase-encoding genes (e.g. egl1 and egl2; data not shown). Fusion of 613 bp of the upstream regulatory region of the cbh2 gene to the E. coli hygromycin B phosphotransferase (hph) gene as a reporter mimicked this induction (Fig. 1A), thus providing evidence for a sophorose-responding element within these 613 bp. Within these 613 bp, a 191-bp fragment has previously been identified to form a unique complex with proteins isolated from sophorose-induced or glucose-grown mycelia (17). To further delimit this area, we prepared extracts from cultures replaced on glucose or sophorose and subjected them to EMSA with three smaller DNA fragments spanning this area. Hereby, the binding ability was retained by a 94-bp fragment, spanning from Ϫ263 to Ϫ170 (Fig. 1B and C). Consequently, this fragment was further analyzed by competition experiments using three synthetic oligonucleotides spanning overlapping parts of this region (Fig. 1, B and C). Competition was only achieved with CKT 057, whereas oligonucleotides CKT 067 and CKT 069 were unable to specifically compete (Fig. 1, B and C), suggesting that the area between Ϫ250 and Ϫ243 was essential for complex formation. Hence sophorose-specific protein binding seems to be confined to a relatively short motif within the 613 bp.
Identification of the CAE-To precisely identify the protein binding sequence, we performed in vitro footprinting with cellfree extracts of H. jecorina: methylation protection demonstrated protein contact with G Ϫ241 of the coding strand (Fig.  2C). In addition, the involvement of T Ϫ238 , T Ϫ239 , and T Ϫ246 on the template strand was shown by hydrazine interference (Fig.  2, A and C). Interestingly, despite the different mobility of the protein-DNA complexes from induced and noninduced mycelia, essentially the same protection/interference pattern was observed under both conditions.
To investigate whether the observed binding in vitro actually reflects the binding conditions in vivo, we performed in vivo genomic footprinting by the aid of ligation-mediated PCR of the respective area of the cbh2 upstream regulatory region. To this end, methylated DNA was isolated from mycelia, treated with HCl, and cleaved with NaOH (29) before the respective PCR amplification steps. Thereby, several purine bases on the coding strand within and in the direct neighborhood of the region previously identified in vitro were identified to be involved in protein-DNA contact (Fig. 2B); on the template strand, no protection of As within the identified motif could be detected. Consistent with the results from in vitro footprinting, essentially the same protection pattern was observed under all conditions tested.
Summarizing the in vitro and in vivo data, we therefore propose that the nucleotide area between Ϫ246 and Ϫ236 is responsible for the binding of the protein complex.
CAE Consists of Two Adjacent, Cooperating Protein Binding Motifs-Visual inspection of the identified nucleotide region revealed no major similarities to known protein binding motifs with two exceptions: one is a CCAAT box on the template strand (ϭATTGG, at Ϫ246 to Ϫ242); the other is the sequence GGGTAAT (at Ϫ243 to Ϫ237), which bears some similarity to the 5Ј-GGCTAAA-3Ј motif recently shown to bind the xylanase regulator XlnR of Aspergillus niger (33). To test whether any of these two motifs may be involved in protein binding, we introduced point mutations into these two motifs and compared the effect of these mutations on protein binding by EMSA (Fig. 3,  A-D). To this end, we replaced the two As in the CCAAT box by Ts, because this mutation has been shown to impair binding of the Hap complex to the CCAAT motif in the A. nidulans amdS promoter (34). The GGGTAAT motif, on the other hand, was mutated to GGGTTTT because of the strong protection of the two As in footprinting experiments. Interestingly, competition experiments with either of the single mutated oligonucleotides alone (CKT 063 and CKT 065) still exerted a very strong effect on binding of cell-free extracts to the wild-type 94-bp promoter fragment. A significant reduction in competition could only be observed when both motifs were simultaneously mutated (CKT 083; Fig. 3A). Consistent results were also obtained when oligonucleotide CKT 057 was used as a labeled probe in EMSA with the same competitors as in Fig. 3A (Fig. 3B). To directly compare the effect of these mutations on protein binding, all four oligonucleotides were labeled and directly used as probes for EMSA (Fig. 3C). In these experiments, two specific DNAprotein complexes were detected with the wild-type promoter (CKT 057). Substitution of CCTTT for CCAAT (CKT 065) or of GGTTTT for GGTAAT (CKT 063) completely eliminated one of the two protein-DNA complexes, and a mutation in both motifs (CKT 083) resulted in a complete loss of specific protein-DNA complex formation (Fig. 3C). It is intriguing that the separation into two defined complexes, which is because of the shorter DNA fragment used, was not observed with the 94-bp promoter fragment. Combining the results from Fig. 3, A, B, and C, we conclude that CAE is simultaneously contacted by several DNA-binding proteins binding to two different cooperating motifs.
Mutations Impairing Protein Binding in Vitro Lead to a Loss of cbh2 Gene Transcription in Vivo-To analyze whether the identified motifs actually confer inducibility of cbh2 transcription in vivo, a set of reporter constructs consisting of the 613-bp of the 5Ј-noncoding region of the cbh2 promoter fused to the hph structural gene from E. coli were made. The different constructs were essentially identical with the exception of the point mutations introduced, which exactly resembled those used for the analysis of protein-DNA complex formation in vitro (pSMZ2 ϭ CKT 065; pSMZ4 ϭ CKT 083; pSMZ5 ϭ CKT 063; cf. Figs. 3D and 4A). Reporter cassettes derived from plasmids pSMZ1-pSMZ5 were introduced into H. jecorina by cotransformation with pFG1 (26). For each construct, 30 pyrϩ transformants were purified to mitotic stability, and their DNA was isolated and examined for integration and copy number of the respective reporter cassette by Southern analysis. Cotransformation frequency varied between 30 to 50%, leading to 9 (pSMZ1), 10 (pSMZ2), 13 (pSMZ4), and 15 (pSMZ5) stable co-transformants per construct, respectively. All obtained transformants exhibited copy numbers from 1 to 4 copies, where the reporter cassette had integrated into ectopic loci. They were examined for the expression of hygromycin B resistance on plates containing cellulose or glucose as carbon sources, respectively. The observed results, taking into account the copy number of the integrated reporter cassette, pointed to a weak influence of mutations present in pSMZ2 and pSMZ5 (less than 20%), whereas the mutation of both motifs in pSMZ4 led to a complete loss of inducibility of the hph reporter gene (data not shown).
To directly demonstrate the effect of these mutations on the induction of transcription from the cbh2 promoter, selected transformants were induced by sophorose and subjected to Northern analysis. Consistent with the results from hygromy- cin B resistance formation, strains SMZ2 and SMZ5 exhibited only a partial reduction in hph gene transcription compared with transformants SMZ1 bearing the wild-type promoter (about 80% and 62% to 44% residual growth in relation to copy number, respectively), whereas strains SMZ4 showed no transcript formation under inducing conditions (Fig. 4, B and C).
These data are in perfect agreement with those from EMSA analysis and stress that mutations in one of the two deduced binding areas alone (strains SMZ2 and SMZ5) cause only a partial reduction in gene transcription, whereas mutation of both results in a complete loss of transcription.
The CCAAT Box in CAE Binds a Hap-like Protein Complex-In view of the CCAAT box on the template strand of CAE, we reasoned whether this box may in fact bind a H. jecorina homologue of the yeast and A. nidulans Hap complex (35)(36)(37). Because the hap genes of H. jecorina have not yet been cloned, and defective strains are therefore not yet available, we used two different indirect approaches to answer this question. First, we made use of an antibody against A. nidulans HapC (38). This antibody was shown to recognize one major protein of H. jecorina on Western blots, which exhibited a similar, yet not identical, size as A. nidulans HapC (data not shown). When this antibody was applied in EMSA analysis with cell-free extracts of induced and noninduced mycelia of H. jecorina, a supershift was observed in both cases, indicating the binding of the antibody to components of the DNA binding complex of both induced and noninduced cells (Fig. 5A).
In a second approach, we made use of a fragment of the A. nidulans amdS promoter, for which HapC binding has been demonstrated in vivo and in vitro (37,39). Except for the CCAAT box, this oligonucleotide shares no further sequence identity with the investigated area (Ϫ236 to Ϫ246) of the cbh2 upstream regulatory region. Fig. 5B shows that the amdS promoter fragment clearly competed the specific complex formed with oligonucleotide CKT 057 (Fig. 5B). From these data we conclude that a HapC homologue of H. jecorina is part of the protein complex binding to the CCAAT box in CAE of cbh2.
Does a XlnR homologue Bind to the 3Ј Motif of CAE?-The demonstration of binding of a HapC homologue to the 5Ј part of CAE supports the assumption that the observed binding involves two different proteins rather than a single one contacting the whole area. As the 3Ј area, as explained before, bears some resemblance to the proposed binding sequence for the xylanase regulator XlnR (33), we also tested whether a respective H. jecorina homologue may be involved in this binding. To this end, we tested whether the binding of protein extracts to oligonucleotide CKT 057 could be competed by a respective fragment of the A. niger xlnD promoter (xlnD(wt)), for which XlnR binding has been demonstrated in vivo and in vitro (33) and which, with exception of the GGCTAAA sequence, does not contain any nucleotide similarity to CKT 057. However, even at a high excess (50-fold) of xlnD(wt), only a low degree of competition was observed (Fig. 6). To test whether this low degree of competition was specific at all, we repeated the experiment with a mutated version of the competing oligonucleotide, for which a loss of XlnR binding had been demonstrated (33). In our case, we again observed a weak competition at a 50-fold excess of the competitor xlnD(mut), (Fig. 6) and thus, the mutation has no effect. From these results we conclude that binding to the 3Ј area of CAE is not because of a protein with binding characteristics similar to XlnR.

DISCUSSION
In this study, we have used in vitro and in vivo techniques to identify the nucleotide area in the cbh2 promoter with the sequence 5Ј-ATTGGCAATTA-3Ј (CAE), which is responsible for binding of protein complexes from cellulase-forming (induced) and nonforming (not-induced) mycelia of H. jecorina and which is essential for induction of gene expression by cellulose and sophorose in vivo. Results from the application of different in vitro and in vivo techniques point to this region to be essential for this process. To the best of our knowledge, this is the first time that a cellulose-responsive nucleotide region has been verified in any cellulase-encoding gene, and these results could therefore become a basis for deeper investigations into the Results from mutational analysis of CAE are consistent with the assumption that it consists of two adjacent and functionally cooperating binding sites for transcriptional activators. Indirect evidence suggests that one of these involves a H. jecorina homologue of the CCAAT box binding protein HapC. The   FIG. 3. Characterization of protein binding to the cbh2-activating element using various mutated oligonucleotides. A, EMSA analysis with labeled 94-bp cbh2 promoter fragment (see Fig. 1) and 100 g of cell-free extracts. In competition experiments, unlabeled oligonucleotides CKT 057, CKT 063, CKT 065, and CKT 083 were added to the binding assay in a 150-fold molar excess where indicated. B, EMSA analysis using labeled oligonucleotide CKT 057 as a probe and 100 g of cell-free extract. As competitors, oligonucleotides CKT 057, CKT 083, CKT 063, and CKT 065 were applied in a 10-and 50-fold molar excess where indicated. C, EMSA analysis using labeled oligonucleotides CKT 057, CKT 063, CKT 065, and CKT 083 as probes and 100 g of cell-free extracts. In competition experiments (indicated by ϩ), a 50-fold molar excess of unlabeled CKT 057 was added as competitor. D, sequences of annealed oligonucleotides CKT 057 (bearing the wild-type CCAAT motif (boxed)), CKT 063 (bearing two point mutations downstream of the CCAAT element), CKT 065 (CCAAT motif altered to CCTTT), and CKT 083 (bearing four point mutations within and downstream of the CCAAT box), that were used in EMSA as labeled probes and competitors, are given. F, free probe; S and G, cell-free extracts derived from cultures grown in the presence of glucose (G) or Sepharose.
CCAAT motif is a common cis-acting element found in the promoter and enhancer regions of a large number of genes from eukaryotes, including filamentous fungi, and a number of different nuclear proteins have been found to interact with the CCAAT sequence (40). However, with respect to fungi, the only CCAAT box binding protein characterized in detail is the Saccharomyces cerevisiae Hap complex, which consists of at least three subunits: Hap2p, Hap3p, and Hap5p (41,42). In filamentous fungi, A. nidulans hapC and Neurospora crassa aab-1, which exhibit significant similarity to the S. cerevisiae HAP3 and HAP5 genes, respectively, are so far the only functionally homologous genes known (37,43). HapC has been shown to bind to CCAAT boxes present in the promoters of genes as amdS, gatA, taaG2, yA, pcbAB, and pcbC (37-39, 44, 45). It is interesting to note that functional inactivation of the CCAAT motif in these promoters has been leading to an altered efficacy of expression of the respective genes but not to a loss of their regulation. HapC has therefore been considered to be only responsible for basal expression (37). On a first glance, this compares well with the results from the present study, in which a mutation of the CCAAT motif alone only slightly decreased transcription from the cbh2 promoter. However, our data with H. jecorina show that the presence of the CCAAT box becomes essential for transcription if the adjacent, 3Ј-located protein binding motif is mutated, and vice versa, a mutation in the CCAAT element is balanced by binding via the adjacent protein binding motif. The fact that a mutation in the CCAAT box of H. jecorina cbh2 does not lead to a change in the size of the bound  (Table  I) derived from the promoter region of the amdS gene of A. nidulans, containing a CCAAT box that was shown to bind an Aspergillus Haplike complex (34,37), was added to the binding assay in a 50-fold molar excess.
FIG. 6. Analysis of the involvement of a XlnR-like factor in binding to the cbh2 activating element. EMSA analysis with labeled oligonucleotide CKT 057 as a probe and 100 g of cell-free extract of H. jecorina cultures grown in the presence of glucose (G) or sophorose (S) is shown. In competition experiments a 10-or 50-fold molar excess of unlabeled oligonucleotide xlnD(wt) and xlnD(mut), respectively, was added to the binding reaction where indicated. F, free probe. proteins indicates an interaction between the proteins binding to the CCAAT box and the GTAAT motif, stressing that binding to either of these motifs is apparently sufficient for functional complex formation. Whether this mechanism may be a general feature of fungal CCAAT boxes remains a matter of speculation so far. However, the presence of a binding site for a transcriptional regulator adjacent to the CCAAT box as observed in this study has been shown in the A. nidulans amdS, gatA, and pcbAB/pcbC promoters (34,45). In higher eukaryotes, the Hap homologue NF-Y was in some cases shown to promote transcription by stabilizing or recruiting the binding of additional factors to adjacent promoter or enhancer elements. This function seems to be brought about by NF-Y-induced DNA distortions at the CCAAT box (46). Such a mechanism would also be considered possible for the cbh2 promoter.
With respect to the nature of the protein contacting the 3Ј adjacent area within CAE, a sequence comparison showed identity in 5 of the 7 nucleotides with a box identified to bind the A. niger transcriptional regulator of xylanase gene expression, XlnR (33), hence raising the possibility that a H. jecorina homologue of XlnR could bind to the 3Ј motif in CAE. However, the results obtained did not support this hypothesis. While this paper was prepared for submission, we became aware of the cloning of two putative regulatory genes (ace1 and ace2) of cbh1 from H. jecorina. 2 The ace2 gene product was thereby reported to bind to the DNA sequence 5Ј-GGCTAATAA-3Ј, which matches the motif identified by us more closely than that recognized by A. niger XlnR (5Ј-GGCTAAA-3Ј). As our footprinting results show the protection of the sequence GTAATA, we consider the involvement of ACE II not only in cbh1 but also in cbh2 regulation plausible, because both genes are coregulated by cellulose/sophorose. We therefore speculate that ACE II and a complex containing Hap3p are involved in cbh2 gene transcription.
If our hypothesis is true, how does the resulting protein complex bring about induction of cbh2 gene transcription in the presence of sophorose? Footprinting in vitro and in vivo unequivocally showed that the same bases within CAE are protected under induced as well as noninduced conditions. Hence both the Hap3p homologue as well as ACE II appear to be formed constitutively, are not inducible by cellulose, and seem to bind in a similar fashion under induced and noninduced conditions. In S. cerevisiae, Hap4p, which is not a DNA-binding protein but provides the activation domain for signaling transcription (47) by binding to the Hap2p-Hap3p-Hap5p-DNA complex, is responsible for activation of genes regulated via this system. In analogy, a mechanism could be proposed in which a cellulose-inducible protein binds to the CAE binding complex and hence promotes cbh2 transcription. However, looking at the results from EMSA analysis, we see that cell-free extracts from induced mycelia form a protein-DNA complex of increased mobility, indicating that either its size has been decreased or its charge become more negative. Unless this hypothetical protein has a very low isoelectric point, the results from EMSA therefore do not support the theory of binding of a further protein. An alternative explanation would be modification of the DNA-protein complex by phosphorylation. Phosphorylation of components of the C/EBP family of CCAAT-binding proteins is well known (48 -50) but has not yet been reported for the Hap homologues. However, we have recently observed that treatment of cell-free extracts of H. jecorina with alkaline phosphatase inhibits protein binding to CAE, 3 which adds some support to this possibility. Whether or not phosphoryla-tion indeed plays a role in CAE-protein binding and which of the protein components are involved, therefore, is the subject of further studies.