Biochemical Analysis of Recombinant Fungal Mutanases

Nucleotide sequence analysis shows thatTrichoderma harzianum and Penicillium purpurogenum α1,3-glucanases (mutanases) have homologous primary structures (53% amino acid sequence identity), and are composed of two distinct domains: a NH2-terminal catalytic domain and a putative COOH-terminal polysaccharide-binding domain separated by a O-glycosylated Pro-Ser-Thr-rich linker peptide. Each mutanase was expressed in Aspergillus oryzae host under the transcriptional control of a strong α-amylase gene promoter. The purified recombinant mutanases show a pH optimum in the range from pH 3.5 to 4.5 and a temperature optimum around 50–55 °C at pH 5.5. Also, they exhibit strong binding to insoluble mutan with K D around 0.11 and 0.13 μm at pH 7 for the P. purpurogenum andT. harzianum mutanases, respectively. Partial hydrolysis showed that the COOH-terminal domain of the T. harzianum mutanase binds to mutan. The catalytic domains and the binding domains were assigned to a new family of glycoside hydrolases and to a new family of carbohydrate-binding domains, respectively.

Extracellular polysaccharides produced by microbial flora in the human oral cavity are believed to play an important role in the adherence and proliferation of bacterial aggregates on the surface of teeth (1). Consequently, these polysaccharides might have significance in the development of tartar, plaque, and possibly dental carries (2). Mutan is a major component of exopolysaccharides produced by tooth colonizing streptococci such as Streptococcus mutans (3). Mutan is composed of ␣1,3glucan with some ␣1,6-glucan (dextran) side chains. Mutanase (␣1,3-glucanase, EC 3.2.1.59) and dextranase (␣1,6-glucanase, EC 3.2.1.11) enzymes could be beneficial additives to dentifrice preparations as it has been shown that these enzymes are capable of removing biofilms created by oral bacteria in vitro (4) and reducing plaque formation in vivo (5,6). Mutanase activity from the filamentous fungus Trichoderma harzianum was first described by Guggenheim and Haller (7). However, only a limited number of reports are available on the characterization of fungal mutanases. Here we describe the cloning, expression, and subsequent characterization of two fungal mutanases representing a new family of fungal endoglucanases with their unique mutan-binding domains.

EXPERIMENTAL PROCEDURES
Fungal Strains-T. harzianum strain CBS 243.71 and Penicillium purpurogenum CBS 238.95 were used as the sources of genomic DNA. Aspergillus oryzae JaL142 and JaL125 (obtained from J. Lehmbeck, Novo Nordisk A/S), were alkaline protease-deficient strains used for heterologous expression of cloned mutanases. The A. oryzae strains show no detectable level of background mutanase activity in the assay described.
Purification and Charcaterization of the Wild-type Mutanase from T. harzianum-100 g of SP234 (Novo-Nordisk A/S, batch number PPM 3897) were dissolved in 1 liter 10 mM sodium acetate, pH 5.2. Contaminant proteins were removed by batch adsorption on DEAE-Sephadex, then by batch adsorption on S-Sepharose (Amersham Pharmacia Biotech). After concentration on a Filtron concentrator equipped with a 10-kDa cut-off membrane, the unbound material was applied to a S-Sepharose (Amersham Pharmacia Biotech) column (180 ml, 2.6 ϫ 33 cm) equilibrated with 10 mM sodium acetate, pH 4.7. The mutanase was eluted with a 0 -20 mM linear gradient of NaCl in the same buffer (3 column volumes). The residual protein was eluted with the same buffer containing 1 M NaCl. Fractions with high mutanase activity were pooled and concentrated. After the procedure was repeated 12 times, the pooled fractions were concentrated and placed in 10 mM Tris-HCl, pH 8.0. The mutanase was further purified on a HiLoad Q-Sepharose column (50 ml, 2.6 ϫ 10 cm) equilibrated with 10 mM Tris-HCl, pH 8.0, and eluted with a linear gradient from 0 to 50 mM NaCl in 12 column volumes. Fractions with high mutanase activity were pooled and concentrated in an Amicon cell equipped with a 10-kDa cut-off membrane. Finally, the mutanase preparation was dialyzed extensively against 10 mM sodium phosphate, pH 7.0. SDS-PAGE 1 gave one single band at 75 kDa (data not shown).
Carbohydrate composition analysis was performed on lyophilized samples which were hydrolyzed in vacuo in sealed glass tubes using 100 l of 2 M trifluoroacetic acid for 1 h and 4 h at 100°C. Monosaccharides were separated by high performance anion exchange chromatography using a Dionex Carbopac PA1 column eluted with 16 mM NaOH and detected by pulsed amperometric detection.
The mutanase mass was measured using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-MS) (VG Analytical). Typically 2 l of sample were mixed with 2 l of saturated matrix solution (␣-cyano-4-hydroxycinnamic acid in 0.1% trifluoroacetic acid:acetonitrile (70:30)) and 2 l of the mixture were deposited on the target plate. After evaporation of the solvent, the samples were introduced in the spectrometer. They were desorbed and ionized by 4-ns laser pulses (337 nm) and subjected to an accelerating voltage of 25 kV. Ions were detected by a microchannel plate set at 1850 V.
Generation of a cDNA Probe for the T. harzianum Mutanase Using Reverse Transcriptase PCR-T. harzianum was cultivated as described (8). A 2-liter sample was taken after 4 days of growth at 30°C, and the * 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF214480 (T. harzianum mutanase gene sequence previously listed as Geneseq™ accession number V12368) and AF214481 (P. purpurogenum mutanase gene sequence; previously listed as Geneseq™ accession number V81911 mycelium was collected, frozen in liquid N 2 , and stored at Ϫ80°C. First-strand cDNA was synthesized from 5 g of T. harzianum poly(A) ϩ RNA as described earlier (9). A 387-bp fragment of the T. harzianum mutanase cDNA (10) was amplified using two mutanase-specific primers (100 pmol each): forward (5Ј-ACTAAGCTTTATGTTCAAAAT-GAGCA-3Ј) and reverse (5Ј-ACACTCTAGAACATATGGGTTGAAGT-TGT-3Ј), a DNA thermal cycler (Landgraf, Germany) and 2.5 units of Taq polymerase (Perkin-Elmer Cetus). Initially, two cycles of PCR were done using a cycle profile of denaturation at 94°C for 1 min, annealing at 45°C for 2 min, and extension at 72°C for 3 min, then the annealing temperature was increased to 55°C and 30 additional cycles were performed. The PCR fragment of interest was subcloned into pUC18 vector and sequenced as described previously (9) Construction and Screening of the T. harzianum cDNA Library-Total RNA was prepared from frozen, powdered mycelium of T. harzianum by extraction with guanidinium thiocyanate followed by ultracentrifugation through a 5.7 M CsCl cushion (11). The poly(A) ϩ RNA was isolated by oligo(dT)-cellulose affinity chromatography (12). Doublestranded cDNA was synthesized from 5 g of T. harzianum poly(A) ϩ RNA as described earlier (13), except that 25 ng of random hexanucleotide primers (Life Technologies, Inc.) were included in the first strand synthesis. A cDNA library, consisting of 1.5 ϫ 10 6 independent clones was constructed in the yeast expression vector pYES 2.0 (Invitrogen) as described (13), and screened by colony hybridization (14) using a random-primed (15) 32 P-labeled (Ͼ1 ϫ 10 9 cpm/g) mutanase cDNA fragment as a probe. The hybridizations were carried out in 2 ϫ SSC, 5 ϫ Denhardt's solution (14), 0.5% (w/v) SDS, 100 g/ml denatured salmon sperm DNA for 24 h at 65°C followed by washes in 2 ϫ SSC (2 ϫ 15 min), 2 ϫ SSC, 0.5% SDS (15 min), 0.2 ϫ SSC, 0.5% SDS (15 min), and finally in 2 ϫ SSC (2 ϫ 15 min) at 65°C.
Cloning of P. purpurogenum Mutanase Gene-Total cellular DNA was isolated from P. purpurogenum cells by a previously described method (16), and used for construction of genomic DNA libraries in the bacteriophage -ZipLox cloning system (Life Technologies Inc., Gaithersburg, MD) (17). Approximately 45,000 plaques from the library were screened by plaque hybridization (18) with a radiolabeled T. harzianum mutanase probe fragment using moderate stringency conditions (5 ϫ SSPE, 35% formamide (v/v), 0.3% SDS, 200 g/ml denatured and sheared salmon testes DNA; hybridization temperature 45°C. Membranes were washed once in 0.2 ϫ SSPE with 0.1% SDS at 45°C followed by two washes in 0.2 ϫ SSPE (no SDS) at the same temperature.) Plaques which gave hybridization signals were purified twice on Escherichia coli Y1090ZL cells, and the mutanase clones were subsequently excised from the -ZipLox vector as pZL1-derivatives (19). One such clone, designated pZL-Pp6A, was selected for further study.
DNA Sequence Analysis-DNA sequencing was done with an Applied Biosystems Model 373A Automated DNA Sequencer (Applied Biosystems, Inc., Foster City, CA) using a combination of shotgun DNA sequencing (20) and the primer walking technique with dye-terminator chemistry (21).
Construction of T. harzianum Mutanase Expression Vector-The T. harzianum mutanase cDNA fragment was inserted in a two-step cloning procedure into an A. oryzae expression vector, pMHan37 (kindly provided by I. G. Clausen, Novo Nordisk A/S), which contains the A. nidulans amdS gene as a selectable marker, pUC plasmid sequences for replication in E. coli, an ␣-amylase gene promoter from A. oryzae, and the A. niger glucoamylase (glaA) terminator (22). In the first step, pMHan37 was linearizd with the restriction enzymes EcoRI and XhoI. This fragment was ligated with the following three segments: 1) a 618-nt fragment of the ␣-amylase promoter sequence bordered by an EcoRI site at the 5Ј end and a BamHI site at the 3Ј end; 2) linker number 1 listed below which has a BamHI site at the 5Ј end and a NarI site at the 3Ј end. This linker includes the Met start codon and 12 amino acids of the mutanase signal sequence; and 3) a 68-nt NarI/XhoI frag-ment from the mutanase cDNA clone containing amino acids 12 to 34 of the mutanase gene. The resulting plasmid pJW99 contains the ␣-amylase promoter immediately upstream from the first 34 amino acids of the mutanase gene, followed by the A. niger glucoamylase terminator. To complete the expression vector the mutanase cDNA fragment was cleaved with XhoI and SphI giving a 1790-nt fragment encoding amino acids 35-598. This fragment was ligated with pJW99 that had been linearized with XhoI plus XbaI and linker number 2, yielding the vector pMT1802, which contains the entire mutanase coding region under the transcriptional control of the A. oryzae ␣-amylase promoter and A. niger glucoamylase terminator. Plasmid pMT1796 is identical to pMT1802 except that Glu-35 of the mutanase protein has been changed to Lys-35 by replacing the XhoI/KpnI fragment of pMT1802 with a PCR-amplified fragment containing this mutation. This PCR fragment was created in a two-step procedure as reported in Ref. 23 using the following primers: Primer 1 (nt 2761, 5Ј-CAGCGTCCACATCACGAGC, nt 2779) and Primer 2 (nt 3306, 5Ј-CAAGAAGCACGTTTCTCAGAGACCG, nt 3281); Primer 3 (nt 3281 5Ј-CGGTCTCTGAGAAACGTGCTTCTTC, nt 3306) and Primer 4 (nt 4276, 5Ј-GCCACTTCCGTTATTAGCC, nt 4257); nucleotide numbers refer to the pMT1802 plasmid.

Expression of Recombinant T. harzianum Mutanase in A. oryzae-
The A. oryzae host strain JaL125 was transformed using a polyethylene glycol-mediated protocol (24) and a DNA mixture containing 0.5 g of a plasmid encoding the gene that confers resistance to the herbicide Basta (25) and 8.0 g of the expression vector pMT1796. Transformants were selected on minimal plates containing 0.5% Basta and 50 mM urea as a nitrogen source. Each transformant was purified twice on selection media and conidia were harvested. Universal containers (20 ml, Nunc, catalog number 364211) containing 10 ml of YPM (2% maltose, 1% bactopeptone bactopeptone, and 0.5% yeast extract) were inoculated with spores from the transformants and incubated 5 days with shaking at 30°C. Culture supernatants were harvested after 5 days growth and assayed for the recombinant mutanase.
Expression of P. purpurogenum Mutanase in A. oryzae-Two synthetic oligonucleotide primers were designed to amplify the P. purpurogenum mutanase gene from plasmid pZL-Pp6A, 5Ј-cccatttaaatATGA-AAGTCTCCAGTGCCTTC and 5Ј-cccttaattaaTTAGCTCTCTACTTGA-CAAGC (capital letters correspond to the sequence present in the mutanase coding region). One hundred picomoles of each primer was used in a PCR reaction containing 52 ng of plasmid DNA, 1ϫ Pwo polymerase buffer (Roche Molecular Biochemicals, Indianapolis, IN), 1 mM each dATP, dTTP, dGTP, dCTP, and 2.5 units of Pwo polymerase (Roche Molecular Biochemicals). The PCR conditions were 95°C 3 min, 25ϫ (95°C 1 min, 60°C 1 min, 72°C 1.5 min), 72°C 5 min. The amplified 2.2-kilobase DNA fragment was purified by gel electrophoresis and cut with restriction endonucleases SwaI and PacI (using conditions specified by the manufacturers). The fragment was cloned into plasmid pBANe6 (26) that had been previously cut with SwaI and PacI and the resultant expression plasmid was named pJeRS35. This vector was introduced into A. oryzae host strain JaL142 using a standard protoplast transformation procedure (24) and 40 transformants were selected by their ability to grow on COVE medium using acetamide as sole nitrogen source. The transformants were grown in 20 ml of MY50N media ( Preparation of Mutan-Mutan was prepared by growing S. mutans CBS 350.71 at 37°C, pH 6.5 (kept constant at a stirring rate of 75 rpm in a medium comprised of the following components: NZ-Case, 6.5 g/liter; yeast extract, 6 g/liter; (NH 4 ) 2 SO 4 , 20 g/liter; K 3 PO 4 , 3 g/liter; glucose, 50 g/liter; pluronic PE6100, 0.1%). After 35 h, sucrose was added to a final concentration of 60 g/liter to induce glucosyltransferase. The total fermentation time was 75 h. The supernatant from this fermentation was centrifuged and filtered (sterile). Sucrose was added to the supernatant to a final concentration of 5% (pH was adjusted to pH 7.0 with acetic acid) and the solution was stirred overnight at 37°C. The solution was filtered and the insoluble mutan harvested on a Propex 23 filter (Scapa Filtration) and washed with deionized water containing 1% sodium benzoate, pH 5 (adjusted with acetic acid). Finally, the insoluble mutan was lyophilized and ground.
Enzyme Assays-The production of soluble reducing sugars released from mutan was employed as a measure of enzyme activity. First, 0.1 ml of 5% mutan in 50 mM sodium acetate (allowed to swell at least for 1 h), pH 5.5, was added to 0.3 ml of enzyme sample (diluted in water) in a round-bottomed Eppendorf vial to ensure sufficient agitation and incubated for 15 min at 40°C while shaking vigorously. The reaction was terminated by adding 0.1 ml of 0.4 M NaOH and the samples were centrifuged for 5 min at 14,000 ϫ g and filtered through 0.45-m HV-filters (Millipore). To each filtrate (100 l) in Eppendorf vials 750 l of ferricyanide reagent (0.4 g/liter K 3 Fe(CN) 6 , 20 g/liter Na 2 CO 3 ) was added and incubated 15 min at 85°C. After allowing the samples to cool, the decrease in absorbance at 420 nm was measured. A dilution series of glucose was included as a standard. Proper controls (substrate and enzyme blanks) were always included. One mutanase unit (MU) was defined as the amount of enzyme releasing 1 mol of reducing sugar per minute at pH 5.5 and 40°C. Temperature profiles were obtained by incubating the assay mixture (50 mM sodium acetate, pH 5.5) at various temperatures. The pH profiles were obtained by suspending the mutan in 50 mM buffer at various pH (glycine-HCl, pH 3-3.5; sodium acetate, pH 4 -5.5; and sodium phosphate, pH 6 -7.5).
Purification of Recombinant T. harzianum Mutanase-The fermentation broth (700 ml) containing 15.4 MU/ml was filtered using GF/A (Whatmann) and HV 0.45-m (Millipore) filters and concentrated on a Filtron concentrator equipped with a 10 kDa cut-off membrane. The pH was adjusted to 4.7 (conductivity approximately 300 microsiemens/cm), and the broth was loaded onto an S-Sepharose column (XK 50/22, Amersham Pharmacia Biotech) equilibrated in 10 mM sodium acetate, pH 4.7. The mutanase was eluted in a linear NaCl gradient. Fractions containing mutanase activity were pooled and concentrated on an Amicon Cell (YM10) and loaded onto a HiLoad Q-Sepharose column (Amersham Pharmacia Biotech) equilibrated in 10 mM Tris-HCl, pH 8.0 (approximately 600 S/cm), in three rounds. The mutanase was eluted in a linear gradient of NaCl. Pooled fractions (according to activity/purity) were concentrated and further purified by gel filtration on a Superdex 75 (16/60) column (Amersham Pharmacia Biotech) in 0.1 M sodium acetate, pH 6.0.
Purification of Recombinant P. purpurogenum Mutanase-The fermentation broth (780 ml) containing 2.2 MU/ml was filtered (0.45 m; HV Millipore) and mixed with 15.6 g of mutan, washed in 0.1 M sodium acetate, pH 5.5, to provide a 2% solution. The pH was adjusted to 5.5 and the suspension was allowed to stand at 4°C for 1 h while stirring. The suspension was then filtered on a sintered glass filter funnel and the mutan was washed four times with 0.1 M sodium acetate, pH 5.5 (total volume: 1110 ml), and then six times with Milli Q-filtered deionized water (total volume, 1250 ml); after each washing step the suspension was filtered. The mutanase eluted during the washing with water. These filtrates were pooled, filtered (0.7 m, Whatman), concentrated on a Filtron concentrator equipped with a 10 kDa cut-off membrane, and further concentrated to 25 ml on an Amicon cell (YM10 membrane).
Preparation of Binding Isotherms-Equilibrium binding was ascertained with 10 mg/ml mutan incubated at 4°C with 0.5 MU mutanase in 10 mM Britton-Robinson buffer, pH 7. At various time points, samples were taken and filtered (0.45-m HV, Millipore) prior to measuring the activity. Binding isotherms were obtained by incubating various concentrations of purified mutanase in a 0.2% suspension of mutan in 0.1 M sodium phosphate, pH 7, for 1 h at 4°C while stirring. The mutan was rinsed in buffer prior to use. Samples were then centrifuged for 10 min at 15,000 ϫ g and the amount of enzyme left in the supernatant determined by fluorescence spectrometry (Perkin-Elmer LS50) with excitation at 280 nm and emission at 345 nm. A fluorescence standard curve of the enzyme diluted in buffer was always included. Alternatively, the activity was measured in the supernatant and compared with the control. The data was fitted using the simple Langmuir theory for adsorption to a surface: A ϭ (A max ϫ E free )/(K D ϩ E free ), where A is the adsorbed protein, A max is the maximum amount of protein which can be adsorbed to the surface, E free is the free protein and K D the equilibrium constant for ES 7 E ϩ S (27).
Differential Scanning Calorimetry-Samples for DSC were desalted into the appropriate buffer using NAP-5 columns from Amersham Pharmacia Biotech. Final enzyme concentrations were in the range from 2 to 3 mg/ml. Samples were scanned from 20 to 90°C using a scan rate of 90°/h at the MC-2 (MicroCal).
Protein Sequencing-NH 2 -terminal amino acid sequencing was done using an Applied Biosystems 473A protein sequencer according to the manufacturer's instructions.
Isolation and Mutan Binding Activity of the COOH-terminal Domain from the T. harzianum Mutanase-Mutanase was incubated for 2.5 h at 30°C with chymotrypsin (Roche Molecular Biochemicals) in a ratio of 100:1 (mutanase:chymotrypsin, w/w) in 50 mM NH 4 HCO 3 . The digest was investigated on SDS-PAGE and the 41-kDa band observed was electroblotted from SDS-PAGE onto a Millipore Immobilin P SQ polyvinylidene difluoride membrane in 10 mM CAPS, 6% methanol at 175 mA for 3 h and subjected to NH 2 -terminal amino acid sequencing, revealing a sequence of SLTIGL-corresponding to proteolytic cleavage after amino acid residue Phe-473. A 50-l sample of the digest was incubated for 30 min at room temperature with 50 l of 2.5% mutan suspension. The sample was centrifuged for 2 min at 15,000 ϫ g. A 30-l volume of supernatant was then analyzed by SDS-PAGE (Novex 4 -20%). Controls without mutan were included. Isolation and Characterization of cDNA Clones Encoding the Mutanase from T. harzianum-To obtain a cDNA probe for the T. harzianum mutanase, two oligonucleotides based on a genomic mutanase clone from T. harzianum (10) were designed. These primers were used to amplify a mutanase cDNA fragment from T. harzianum first-strand cDNA employing the PCR technique (28). Sequencing of the subcloned PCR fragment revealed a 387-bp cDNA with an open reading frame of 129 amino acids. In addition to the primer-encoded residues, the ORF was identical to the corresponding region in the T. harzianum mutanase amino acid sequence (10), confirming that the PCR had specifically amplified the desired cDNA species. Approximately 10,000 colonies from a T. harzianum cDNA library in E. coli were screened using the mutanase-specific PCR product as a probe. This yielded 12 positive clones with inserts ranging from 0.8 to 2.0 kilobase. These were further analyzed by sequencing the ends of the cDNAs with forward and reverse pYES polylinker primers, and determining the nucleotide sequence of the longest cDNA from both strands with synthetic oligonucleotide primers. The nucleotide sequence and the deduced amino acid sequence of the mutanase cDNA from T. harzianum are presented in Fig. 1 Cloning of P. purpurogenum Mutanase-Southern blotting experiments indicated that the T. harzianum mutanase cDNA could be used as a probe to identify mutanase gene-specific fragments in P. purpurogenum genomic DNA (data not shown). Consequently, a genomic library was constructed from P. purpurogenum cellular DNA using the bacteriophage vector -ZipLox. Approximately 45,000 plaques from this library were screened by hybridization using a segment of the T. harzianum mutanase cDNA as the probe. Eighteen positive clones which hybridized strongly to the probe were picked and 10 were plaque-purified (18) and excised from the cloning vector using the in vivo excision protocol (19). Preliminary restriction mapping on one of the pZL1-mutanase clones (designated pZL- Pp6A) revealed that the region which hybridized to the T. harzianum mutanase cDNA was localized near one end of a 3.6-kilobase genomic DNA insert (not shown). DNA sequencing of a portion of this segment showed an open reading frame with clear homology to the T. harzianum mutanase cDNA and its deduced amino acid sequence (Fig. 2). The positions of introns and exons within the P. purpurogenum mutanase gene were assigned based on alignments of the deduced amino acid sequences to the corresponding T. harzianum mutanase gene product. On the basis of this comparison, the P. purpurogenum mutanase gene is composed of five exons (126, 532, 226, 461, and 548 bp) which are punctuated by four small introns (63, 81, 58, and 78 bp). These appear to be typical fungal introns with respect to size and composition in that all contain consensus splice donor and acceptor sequences as well as the consensus lariat sequence (PuCTPuAC) near the 3Ј end of each intervening sequence (29).

Wild-type T. harzianum
Comparison of Trichoderma and Penicillium Mutanase Primary Structures-The signal peptide and propeptide portions of P. purpurogenum mutanase and T. harzianum mutanase share little amino acid sequence similarity; however, the ma-ture polypeptides (after removal of signal and propeptides) are approximately 53% identical. The regions of greatest identity are located in the NH 2 -terminal portion (residues 42 through 491; T. harzianum numbering) and over approximately the last 70 residues of these two proteins where 60 and 63% identity is observed, respectively. In both mutanases the NH 2 -terminal and COOH-terminal domains are separated by a Pro-Ser-Thrrich linker region. Remarkably, the Pro-Ser-Thr-rich region of P. purpurogenum mutanase (residues 475 through 547) is composed of 69% Pro, Ser, and Thr, and is bordered roughly by Cys residues at positions 477 and 547. As the two mutanases appeared to have a modular structure, sequence comparisons using the BLAST algorithm to search the non-redundant Gen-Bank CDS translations on the NCBI server (30) were therefore conducted on each domain separately. BLAST searches using the NH 2 -terminal domains did not produce any hits with known glycosidases, however, two ORFs of unknown function in Schizosaccharomyces pombe (C 14 C4.09 and BC646.06c, Gen-Bank Z98596 and AL035216, respectively) were picked with highly significant scores (E values ranging from 6 ϫ 10 Ϫ50 to 10 Ϫ26 ) suggesting that these ORFs encode similar glycosidases. An alignment of the sequences of the catalytic domains of the two mutanases with the two ORFs of S. pombe is shown in Fig.  3a. BLAST searches conducted with the COOH-terminal domains of the mutanases also failed to produce any significant hit in GenBank. Within the COOH-terminal domains of the two mutanases, two short regions display intriguing similarity (10 residues conserved out of 15) suggesting the existence of an internal duplication (Fig. 3b).
Heterologous Expression of T. harzianum Mutanase in A. oryzae-The T. harzianum mutanase coding region was amplified by PCR, and the amplicon was inserted into an Aspergillus expression vector so that the gene was under the control of an A. orzyae ␣-amylase gene promoter and an A. niger glaA terminator. The resulting expression construct, pMTH1802, was further modified by changing aa 35 from Glu to Lys resulting in the presence of a dibasic (KEX2-type) processing site at the amino terminus of the mature mutanase protein. This new expression vector, pMT1796 (Fig. 4a), was used to transform an A. oryzae strain, and 25 independent transformants were isolated. Mycelia from each transformant was used to inoculate 20-ml culture tubes containing 10 ml of YPM media and cultures were grown with shaking for 5 days at 30°C. SDS-PAGE analysis revealed a dominant 85-90-kDa band indicating that these transformants were indeed expressing the recombinant mutanase gene.
Heterologous Expression of P. purpurogenum Mutanase-The P. purpurogenum mutanase coding region was amplified by PCR using primers that created 5Ј-and 3Ј-terminal restriction sites compatible with an Aspergillus expression vector. The amplified DNA segment was subsequently inserted into the vector which employed a strong A. oryzae ␣-amylase gene promoter. The resulting plasmid, designated pJeRS35 (Fig. 4b), was used to transform an A. oryzae recipient strain, and 40 transformants were isolated. Mycelia from each of the transformants were used to inoculate shaker flask cultures that were incubated for 3 days. Using a mutan agar plate assay, 14 of the transformants showed extracellular mutanase activity as indicated by opaque clearing zones (the control showed no clearing zone). Broth samples that were positive in the plate assay were subsequently analyzed by SDS-PAGE. These transformants showed a prominent band at approximately 90 kDa.
Purification of and Molecular Properties of Recombinant Mutanases-Recombinant T. harzianum mutanase was purified in a three-step procedure using cation-exchange chromatography, anion-exchange chromatography followed by size exclusion chromatography resulting in a yield of around 24%. The essentially pure mutanase exhibited a molecular mass around 86 kDa (Fig. 5a). A rather broad band was observed indicating some heterogeneity and/or heavy glycosylation. The NH 2 -terminal amino acid sequence was determined by protein sequencing to be Ala-Ser-Ser-thus predicting a calculated molecular mass of 63.8 kDa for the mature enzyme (Table I). This obser-  vation suggests that the first 37 amino acid residues deduced from the gene sequence function as a secretory signal peptide and propeptide. This is supported by the fact that the NH 2 terminus of the mature mutanase is not preceded by a typical signal peptidase cleavage site (31, Fig. 1) but rather by a cleavage site for a monobasic processing enzyme. Furthermore, the mutanase cDNA encodes an apparent signal sequence of 24 amino acids, with a predicted signal peptidase cleavage site between Ala-24 and Ala-25 in the mutanase precursor (31). A simpler procedure for purification of the recombinant P. purpurogenum mutanase was established using the information that a putative COOH-terminal-binding domain is present in the enzyme. The enzyme was adsorbed to insoluble mutan and subsequently eluted in water. This procedure resulted in a 129-fold purification and a yield around 20%. The essentially pure mutanase had a molecular mass of about 90 kDa (Fig. 5b). NH 2 -terminal amino acid sequencing revealed the following sequence: Ser-Thr-Ser-Asp-Arg-. Thus, the deduced amino acid sequence of the mutanase gene product (Fig. 2) predicts an amino-terminal extension of 30 amino acids which are not present in the mature enzyme and a molecular mass for the mature enzyme of 63.6 kDa (Table I). Based on the rules of von Heijne (31), the first 20 amino acids likely comprise a secretory signal peptide, and the next 10 residues probably represent a propeptide segment which is removed by a subsequent proteolytic cleavage following the dibasic Arg-Arg sequence.
Characterization of the Purified Recombinant Mutanases-The two mutanases showed similar catalytic properties. They both exhibit slightly acidic pH optima in the range from pH 3.5 to 5.0 and pH 3.0 to 4.5 for the T. harzianum and P. purpurogenum mutanases, respectively. At pH 5.5 the two enzymes have specific activities of 16 and 12 MU/mg, respectively, on insoluble mutan at 40°C. Also, the two mutanases have virtually identical temperature optima around 50 -55°C at pH 5.5. This correlates with DSC analysis of the thermal stability of the recombinant T. harzianum mutanase which shows a midpoint denaturation temperature (T M ) around 56°C at pH 5.5 identical to that of the wild-type enzyme.
The binding properties of the two mutanases toward insoluble mutan were investigated at steady state conditions at pH 7 and 4°C in order to limit hydrolysis. The kinetics of adsorption was followed by taking samples from the supernatant of mutanase incubated with mutan. The equilibrium was reached within 5 min, and then no further net adsorption was observed (data not shown). Varying concentrations of mutanase were incubated for 1 h under the above conditions with mutan and the amount of free mutanase was determined by fluorescence spectroscopy (concentrations verified by activity analysis) and the amount of bound enzyme was calculated. Thus, binding isotherms were generated, and the data fitted using the simple Langmuir model for adsorption to a surface (Fig. 6). Rather strong binding was observed with desorption constants of 0.13 and 0.11 M for the T. harzianum and P. purpurogenum mutanases, respectively (Table II). A significant difference is observed in the maximum level of enzyme which can be adsorbed to the insoluble mutan 0.549 versus 0.244 mol of enzyme/g of mutan for the T. harzianum and P. purpurogenum mutanases, respectively (Table II).
In order to probe the hypothesis that the homologous COOH-terminal domain of the two fungal mutanases constitutes a mutant-binding domain, the T. harzianum mutanase was subjected to limited proteolysis using chymotrypsin. The protease treatment resulted in a 41-kDa band on SDS-PAGE (Fig. 7), which was NH 2 terminally sequenced after being electroblotted onto a polyvinylidene difluoride membrane revealing the sequence Ser-Leu-Thr-Ile-Gly-Leu-corresponding to proteolytic cleavage between Phe-473 and Ser-474. This strongly suggests that the 41-kDa band corresponds to the linker and the COOHterminal domain. The chymotrypsin digest of the mutanase was then incubated with 2.5% mutan before centrifuging the sample and loading the supernatant onto SDS-PAGE. From the SDS-PAGE analysis (Fig. 7) it is apparent that the 41-kDa band has been adsorbed to the insoluble mutan since it is no longer present in the supernatant.

DISCUSSION
Nucleic acid sequences encoding extracellular mutanases from the filamentous fungi T. harzianum and P. purpurogenum were cloned and successfully expressed in A. oryzae. The primary translation products of these two DNA sequences appear to be preproenzymes, having both NH 2 -terminal signal peptides and propeptides that are removed post-translationally. The two mutanases show deduced amino acid sequence identities of 53% overall. Further analyses of the protein sequences   (33).
Experiments showing that the chymotrypsin produced fragment of the T. harzianum mutanase adsorbs to mutan gave indirect indication that the COOH-terminal domain of the two fungal mutanases is responsible for binding to insoluble mutan. As a first step in an effort to further characterize the COOH-terminal mutan-binding domain of T. harzianum mutanase, two expression plasmids were constructed harboring (i) an internal deletion of the coding region encompassing residues 32-542 (i.e. coding for the isolated COOH-terminal binding domain without any linker) and (ii) the NH 2 -terminal catalytic domain only. The transformants were tested by immunodiffusion using antibodies raised against the whole mutanase. Whereas the isolated catalytic domain was unaffected by preincubation with mutan, the first transformant became negative upon preincubation with mutan (data to be described elsewhere). The inability of the catalytic domain to bind to mutan was verified by activity anaylsis showing that no activity could be removed from the supernatant by preincubation with insoluble mutan.
Glycoside hydrolases and transglycosidases have been classified in a number of distinct families based on amino acid sequence similarities (34 -36). BLAST searches (30) conducted with the NH 2 -terminal catalytic domains of the two mutanases described here failed to display any similarity with known glycosidases from previously defined families. Families of glycoside hydrolases being defined with at least two related sequences (34), the two mutanases therefore allow the definition of a new family (designated family 71). Although, sequence similarities between the mutanase catalytic domains and the described ORFs of S. pombe are such that it is predictable that all these proteins adopt a similar fold and operate via the same catalytic mechanism using a similar catalytic machinery (37)(38)(39), the precise substrate specificity of the S. pombe proteins cannot be reliably ascertained as it has been shown that sequence-based families of glycosidases contain enzymes with sometimes widely different substrate specificity (34). Finally, it is worth mentioning that, unlike the two mutanases, none of the two S. pombe ORFs carries a COOH-terminal extension suggesting that the encoded proteins are made of a single domain.
Cellulases, xylanases, chitinases, and starch-degrading enzymes have long been recognized to have a modular structure with a catalytic domain carrying one or several ancillary modules whose function is often binding to insoluble polysaccharides (40). The best described of these ancillary modules are probably the cellulose-binding domains which have been classified in several distinct families based on sequence similarities (41,42). The lack of sequence similarity of the two COOHterminal domains of the mutanases with any known carbohydrate-binding domains together with their insoluble mutan binding activity allows the definition of a new family of carbohydrate-binding modules.
The pH optimum observed for the two mutanases is not exactly in agreement with the earlier reported pH optimum around pH 6.0 for the T. harzianum mutanase (7) but comparable to the pH optimum obtained for the Trichoderma viride (43) and the Penicillium funiculosum mutanases (3). For comparison, the bacterial mutanases from Bacillus circulans (44) and Streptomyces chartreusis (45) have slightly higher pH optima than the two fungal mutanases but similar temperature optima. Although, the pH in the oral cavity is around pH 6 -7, the slightly acidic pH profile of the two fungal mutanases may be of importance in the application for plaque removal as low pH values have been observed locally in the plaque (46).
The substrate binding constants observed for the two mutanases to insoluble mutan are, although slightly higher, in the range of reported binding constants for cellulase adsorption to insoluble cellulose (27). The difference in the maximum binding capacity observed for the two mutanases may be explained by differences in the batches of mutan used for the experiment as these have been found to vary somewhat in quality/purity. Alternatively, a possible explanation would be that the T. harzianum mutanase is capable of dispersing the insoluble mutan (in analogy to cellulose-binding domains and cellulose) to a larger extend than the P. purpurogenum mutanase and thus revealing a larger surface area onto which the enzyme can adsorb. The strong adsorption of the fungal mutanases may be beneficial for their application in dentrifice as the enzymes are expected to bind to dental plaque and thus be retained in the oral cavity where it is supposed to exhibit its action in removing the dental plaque.