The conserved noncatalytic 40-residue sequence in cellulases and hemicellulases from anaerobic fungi functions as a protein docking domain.

Two cDNAs, designated xynA and manA, encoding xylanase A (XYLA) and mannanase A (MANA), respectively, were isolated from a cDNA library derived from mRNA extracted from the anaerobic fungus, Piromyces. XYLA and MANA displayed properties typical of endo-beta 1,4-xylanases and mannanases, respectively. Neither enzyme hydrolyzed cellulosic substrates. The nucleotide sequences of xynA and manA revealed open reading frames of 1875 and 1818 base pairs, respectively, coding for proteins of M(r) 68,049 (XYLA) and 68,055 (MANA). The deduced primary structure of MANA revealed a 458-amino acid sequence that exhibited identity with Bacillus and Pseudomonas fluorescens subsp. cellulosa mannanases belonging to glycosyl hydrolase Family 26. A 40-residue reiterated sequence, which was homologous to duplicated noncatalytic domains previously observed in Neocallimastix patriciarum xylanase A and endoglucanase B, was located at the C terminus of MANA. XYLA contained two regions that exhibited sequence identity with the catalytic domains of glycosyl hydrolase Family 11 xylanases and were separated by a duplicated 40-residue sequence that exhibited strong homology to the C terminus of MANA. Analysis of truncated derivatives of MANA confirmed that the N-terminal 458-residue sequence constituted the catalytic domain, while the C-terminal domain was not essential for the retention of catalytic activity. Similar deletion analysis of XYLA showed that the C-terminal catalytic domain homologue exhibited catalytic activity, but the corresponding putative N-terminal catalytic domain did not function as a xylanase. Fusion of the reiterated noncatalytic 40-residue sequence conserved in XYLA and MANA to glutathione S-transferase, generated a hybrid protein that did not associate with cellulose, but bound to 97- and 116-kDa polypeptides that are components of the multienzyme cellulase-hemicellulase complexes of Piromyces and Neocallimastix patriciarum, respectively. The role of this domain in the assembly of the enzyme complex is discussed.

A 40-residue reiterated sequence, which was homologous to duplicated noncatalytic domains previously observed in Neocallimastix patriciarum xylanase A and endoglucanase B, was located at the C terminus of MANA. XYLA contained two regions that exhibited sequence identity with the catalytic domains of glycosyl hydrolase Family 11 xylanases and were separated by a duplicated 40residue sequence that exhibited strong homology to the C terminus of MANA. Analysis of truncated derivatives of MANA confirmed that the N-terminal 458-residue sequence constituted the catalytic domain, while the Cterminal domain was not essential for the retention of catalytic activity. Similar deletion analysis of XYLA showed that the C-terminal catalytic domain homologue exhibited catalytic activity, but the corresponding putative N-terminal catalytic domain did not function as a xylanase. Fusion of the reiterated noncatalytic 40-residue sequence conserved in XYLA and MANA to glutathione S-transferase, generated a hybrid protein that did not associate with cellulose, but bound to 97-and 116-kDa polypeptides that are components of the multienzyme cellulase-hemicellulase complexes of Piromyces and Neocallimastix patriciarum, respectively. The role of this domain in the assembly of the enzyme complex is discussed.
Endo-␤1,4-xylanase (xylanase; EC 3.2.1.8) and endo-␤1,4-mannanase (mannanase; EC 3.2.1.78) hydrolyze, respectively, the ␤1,4-linked polysaccharide backbones of xylans and mannans, which form the two major hemicellulose components of hardwoods and softwoods (1). Recent studies on the structure of xylanases have revealed that some enzymes are comprised of single catalytic domains while other xylanases are modular, consisting of single or multiple catalytic domains fused via linker sequences to noncatalytic sequences, some of which constitute cellulose binding domains (CBD 1 ; Refs. 2 and 3). Hemicellulases derived from aerobic microorganisms do not appear to associate, while anaeorbic organisms often synthesize multienzyme cellulase-hemicellulase complexes (2). Many xylanases have been analyzed, but their catalytic domains appear to have evolved from only two progenitor sequences. In contrast, the primary structures of only seven mannanases have been determined, to date. Based on sequence alignments, two of the enzymes belong to glycosyl hydrolase Family 5 (4,5), three of the mannanases are located in Family 26 (6 -8), while the two enzymes from aerobic fungi exhibit significant sequence identity, although they have not been assigned to a specific family (9). The complexity of the molecular architecture of these enzymes is variable; the Streptomyces lividans, Bacillus spp., and Pseudomonas fluorescens subsp. cellulosa (4,6,8) mannanases consist of single catalytic domains, while the corresponding Caldocellum saccharolyticum enzyme (5) is comprised of two catalytic domains, that exhibit mannanase and endo-␤1,4-glucanase activity, respectively, and are separated by a duplicated CBD homologue.
Recent studies in our laboratories have focussed on plant cell wall-degrading enzymes of anaerobic fungi that are particularly active against the more recalcitrant plant structural polysaccharides. These organisms produce cellulases and hemicellulases that associate into large molecular weight multienzyme complexes and bind tightly to cellulose (10,11). Recently, the molecular architecture of cellulases and xylanases from Neocallimastix patriciarum has been analyzed. Two of the enzymes contain catalytic domains that are linked to a duplicated 40 residue noncatalytic C-terminal domain (12,13) of unknown function. In general, there is a paucity of information on the structure/function relationship of plant cell wall hydrolases of rumen chytridiomycetes. Specifically, it remains to be established whether the 40-residue noncatalytic domain is conserved between different species of fungi and if so, whether the formation of cellulase/hemicellulase complexes in these organisms is mediated by a common mechanism involving this sequence; in addition, the molecular architecture of rumen fungal plant cell wall hydrolases, other than cellulases and xylanases, remains to be elucidated, and it is unclear to what extent the plant cell wall degrading systems in these organisms arose through horizontal gene transfer between rumen fungi. To address these questions we have isolated cDNAs encoding plant cell wall hydrolases from another rumen fungus, Piromyces, and we have used the cloned sequences to characterize the encoded enzymes. In this report we describe the molecular architecture and biochemical properties of a xylanase and mannanase from Piromyces. Both enzymes contain the same noncatalytic reiterated 40-residue sequence previously observed in Neocallimastix enzymes. The mannanase exhibits homology with the mannanases from Bacillus spp. and from P. fluorescens subsp. cellulosa. The xylanase contains a large duplicated domain which shows sequence identity with glycosyl hydrolase Family 11 xylanases, although only one of these domains is functional. The conserved noncatalytic 40residue reiterated sequence, which does not bind to cellulose, binds specifically to polypeptides of 97 and 116 kDa that are components of the multienzyme cellulose-binding complexes produced by Piromyces and N. patriciarum, respectively. The possible role of this sequence as a protein docking domain that mediates assembly of the anaerobic fungal multienzyme complex is discussed.
Recombinant DNA Methodology-Plasmid DNA was purified using Qiagen columns (Hybaid) following the protocol recommended by the manufacturer. Transformation of E. coli, agarose-gel electrophoresis, Northern and Southern hybridizations, and the general use of nucleic acid modifying enzymes were as described by Sambrook et al. (16). DNA probes were labeled with [␣-32 P]dCTP by random priming. The cDNA, inserts expressing hemicellulase activity, isolated previously in recombinants of ZAPII (10), were excised into pBluescript SK Ϫ using VCS-M13 helper phage. The polymerase chain reaction was used to amplify the region of xynA encoding the 40-residue duplicated sequence. The primers used were as follows: 5Ј-CTCGGATCCACTACCGGTACTAC-TACTCG-3Ј and 5Ј-CTCGAATTCACTGCTTAGAAACCACAACC-3Ј, and the conditions for polymerase chain reaction amplification were as described by Fontes et al. (17). The amplified DNA was cloned into pCRII, its sequence verified, and then cloned as a BamHI/EcoRI restriction fragment into suitably digested pGEX-2T and pGEX-2TK to generate pCF1 and pCF2, respectively.
To sequence xynA and manA, a series of nested deletions of each of the two cDNAs was constructed using the exonuclease III/S1-nuclease kit supplied by Promega. The nested deletions were sequenced using a PRISM™ ready reaction dye-deoxy terminator cycle sequencing kit supplied by Applied Biosystems. Sequences were compiled and ordered using the computer programs of Staden (18). The two cDNAs were sequenced completely in both strands.
Assays and Protein Analysis-The source of XYLA and MANA employed in the liquid enzyme assays was cell-free extract (CFE) prepared, as described by Hazlewood et al. (19), from E. coli strains harboring pGX1 and pGM1, respectively. Cellulase, xylanase, and mannanase activities were assayed at 37°C in 50 mM potassium phosphate, 12 mM citric acid buffer, pH 6.5. The substrates were incubated with the appropriate CFE at a final concentration of 0.2%, and reducing sugar released was determined using the dinitrosalicylic acid reagent (20). Hydrolysis of the dyed substrates Red debranched arabinan and AZCL-galactan (both from Megazyme Ltd.) was determined by incorporating the substrates into LB agar and monitoring the appearance of clear haloes against a red background or the solubilization of the blue AZCL dye, respectively, after inoculation and incubation. Oligosaccharides (333 g/ml) were incubated with either MANA or XYLA in a total reaction volume of 1 ml. Aliquots were removed from the enzyme reactions at regular time intervals, boiled for 5 min to inactivate the enzyme, and applied to a Dionex PA1 column. Separation of xylooligosaccharides was achieved using isocratic elution with 100 mM NaOH from 0 -5 min, a gradient of 0 -75 mM sodium acetate in 100 mM NaOH from 16 to 25 min. Separation of mannooligosaccharides was achieved using isocratic elution with 100 mM NaOH from 0 to 5 min and a gradient of 0 -50 mM sodium acetate in 100 mM NaOH from 5 to 25 min. The sugars were detected with a pulsed amperometer. Products released during the hydrolysis of native xylans and mannans were analyzed by Dionex high performance liquid chromatography and were identified by reference to standard compounds. Protein was determined by dye binding (21) using bovine serum albumin as standard. Viscometric assays were performed as described in Ref. 17 except that 2% (w/v) wheat arabinoxylan and 4% (w/v) carob galactomannan were used as substrates for XYLA and MANA, respectively. Glutathione S-transferase (GST) was assayed as described previously (22). SDS-PAGE (23) was carried out in 10% (w/v) polyacrylamide gels containing 0.2% carob galactomannan or oat spelt-soluble xylan. Renaturation of mannanase and xylanase activities and zymogram staining were carried out as described previously (24).
Western Analysis of Anaerobic Fungal Cellulase-Hemicellulase Complexes Using the GST-XYLA Fusion Protein-CFE containing the GST-XYLA fusion protein encoded by pCF1 was obtained by culturing E. coli JM83 harboring pCF1 to mid-log phase, inducing for 4 h by adding isopropyl-␤-D-thiogalactopyranoside (1 mM) and sonicating the harvested cells in phosphate-buffered saline (PBS). A similar approach was used to produce CFE containing GST-XYLA derived from pCF2 except that E. coli BL21 was used as the host strain, and induction was achieved by adding 100 M isopropyl-␤-D-thiogalactopyranoside for 4 h at 30°C. A control CFE containing GST alone was obtained from E. coli JM83 harboring pGEX-2T. Multiprotein cellulase-hemicellulase complexes were isolated from cultures of N. patriciarum, Piromyces sp. (11), and Clostridium thermocellum. The multienzyme complexes and E. coli CFE were either separated by SDS-PAGE (10% polyacrylamide) and electroblotted onto Immobilon-P transfer membrane (Millipore) or dotblotted directly onto the membranes without denaturing the samples. The membranes were then probed using two different methods. In the first method the filters were incubated at room temperature for 1 h with shaking in 5 ml of blocking buffer (PBS containing 10%, w/v, dried milk powder and 0.1% Triton X-100) to which had been added 200 l of CFE containing either the GST-XYLA fusion protein or GST alone. After washing twice for 10 min with PBS containing 0.1% Triton X-100 and once with blocking buffer, membranes were incubated at room temperature for 1 h in 5 ml blocking buffer containing goat anti-GST antibodies (Pharmacia) diluted 1:20,000. Washing was repeated as above and membranes were incubated at room temperature for 1 h in 5 ml of blocking buffer containing anti-Goat IgG/horseradish peroxidase conjugate (Sigma) diluted 1:5000. After washing twice with PBS containing 0.1% Triton X-100 and once with 50 mM Tris/HCl, pH 7.4, binding of the horseradish peroxidase conjugate was detected by incubating the membrane at room temperature in PBS containing diaminobenzidine (0.4 mg/ml) and hydrogen peroxide. In the second method, the GST-XYLA hybrid protein, encoded by pGEX-2TK, was applied to a glutathione-Sepharose 4B column and the protein labeled by incubating the column with [ 32 P]ATP and heart muscle kinase according to the manufacturer's instructions (Pharmacia). The labeled duplicated domain was then cleaved from the GST by incubating with 20 units of thrombin for 4 h at 20°C and eluted from the column by washing with PBS. The purified probe was then incubated with the filters in blocking buffer for 5 h at room temperature. After excess probe was removed by repeated washing of the filters in blocking buffer, the filters were subjected to autoradiography at Ϫ70°C with intensifying screens.
Polysaccharide Binding Studies Using the GST-XYLA Fusion Protein-CFE (2 ml derived from 200-ml cultures) containing either GST-XYLA fusion protein or GST alone was incubated for 1 h at 0°C with 50 mg of Avicel or insoluble oat spelt xylan. The polysaccharide was recovered by centrifuging and was washed five times with 2-ml volumes of 50 mM sodium phosphate, pH 7.0. Avicel and xylan pellets, resuspended in 250 l of phosphate buffer, were assayed for bound GST activity and on SDS-PAGE.

RESULTS
Isolation of manA and xynA-In a previous study cDNAs encoding functional cellulases and hemicellulases were isolated from a cDNA library constructed using mRNA derived from Piromyces cells cultured with Avicel and soluble xylan as carbon sources (10). Restriction maps of the longest cDNAs encoding a functional mannanase and a xylanase, designated manA and xynA, respectively, are displayed in Fig. 1. Nucleic acid probes consisting of the 5Ј regions of manA and xynA hybridized to single Piromyces mRNA species. The sizes of the hybridizing transcripts suggest that the two cDNAs were almost full length (data not shown).
Characterization of XYLA and MANA-The enzymes encoded by xynA and manA were designated XYLA and MANA, respectively. XYLA produced by E. coli harboring pGX1 (Fig. 1) hydrolyzed oat spelt xylan, wheat arabinoxylan, and rye arabinoxylan (Table I). The major products generated during prolonged hydrolysis of each substrate were xylotriose and xylobiose. No arabinose was detected among the reaction products, indicating that the enzyme does not exhibit arabinofuranosidase activity. The enzyme displayed typical endo activity promoting a rapid decline in the viscosity of the substrate (data not shown). XYLA was more active against xylopentaose than xylotetraose, displayed trace activity against xylotriose, but did not hydrolyze xylobiose. The major products released from the two xylooligosaccharides, after prolonged incubation, were xy-lotriose and xylobiose (Table II). Interestingly, during the early phase of xylotetraose hydrolysis, a significant quantity of xylotriose was generated, which was not associated with the production of xylose (Table II). This pattern of product release indicates that either (i) the enzyme is releasing xylose from xylotetraose, which then participates in transglycosylation reactions or (ii) the xylobiose released from the substrate trans-  Fig. 3. The cDNA inserts containing 5Ј deletions were cloned into appropriate plasmids such that the xynA and manA derivatives were in-frame with the vectors' lacZЈ translational starts. The capacity of the plasmids to express a functional hemicellulase is expressed as a percent of the activity of the respective full-length enzymes. glycosylates with the tetramer to form a hexamer, which is then very rapidly hydrolyzed to release the triose. XYLA displayed no detectable activity against mannan, arabinan, galactan, or cellulosic polysaccharides. MANA hydrolyzed carob and locust bean galactomannan and ivory nut mannan. The major products released from each substrate were mannose and mannobiose. The enzyme exhibited no activity against other plant structural polysaccharides (Table I). To evaluate whether MANA had a preference for polymeric substrates, the rate at which the enzyme hydrolyzed a range of mannooligosaccharides was evaluated. The data (Table II) showed that the enzyme had very high activity against mannohexaose compared with mannopentaose and smaller mannooligosaccharides. The major products generated when the reaction had gone to completion were mannose and mannobiose (data not shown). MANA displayed no detectable transglycosylating activity. These data indicate that the active site of MANA has a minimum of six sugar-binding sites and that substrates which occupy five or less of these sites exhibit low affinity for the enzyme. This is in contrast to the mannanase from P. fluorescens subsp. cellulosa which displays similar activity toward mannohexaose, mannotetraose, and mannotriose (6). In general, the biochemical properties of XYLA and MANA are typical of xylanases and mannanases, respectively.
Nucleotide Sequence of xynA and manA-The Piromyces inserts in pGX1 and pGM1 were sequenced in both strands (Fig.  2). Translation of the two nucleotide sequences revealed open reading frames for xynA and manA of 1875 and 1818 bp, respectively, encoding proteins with M r values of 68,049 (XYLA) and 68,055 (MANA). The assignment of the ATG translation initiation codons is, in each case, based on the presence of translational stop codons, in all three phases, upstream of the proposed start codon, and the absence of further ATG codons upstream of the open reading frames. In addition, zymogram analysis of XYLA and MANA, contained in CFE from E. coli JM83 harboring pGX1 or pGM1, respectively, revealed M r values of 64,000 and 65,000, respectively, for the two enzymes (Fig. 3). The codon utilization for both cDNAs was very similar; 16 codons were not utilized and there was a marked preference for T and an exclusion of G from the wobble position. The noncoding regions flanking the two open reading frames were extremely A ϩ T-rich (approximately 90% A ϩ T for both sequences), while the A ϩ T content of the protein coding sequences was 58 and 56% for xynA and manA, respectively. The codon utilization and the A ϩ T-rich character of the noncoding sequences of manA and xynA cDNAs are similar to other DNA sequences encoding cellulases, xylanases, and phosphoenolpyruvate carboxykinase from Neocallimastix (12, 13, 25, 26). It could be argued that the similarity in DNA structure between genes from the two fungi supports the view that the two organisms are closely related. However, this conclusion must be viewed with some caution as the data could be interpreted simply as evidence that gene transfer has taken place between the two organisms, which after all occupy the same ecological niche.
Primary Sequence of MANA and XYLA-Inspection of the primary sequences of XYLA and MANA revealed several interesting features. The N-terminal sequences of both enzymes conformed to a classical signal peptide, comprising of a hydrophilic basic N-terminal sequence followed by a stretch of 17-18 small hydrophobic residues. Downstream of the signal peptide, XYLA contained two domains of 223 residues that exhibited 57 and 68% sequence identity and similarity, respectively (Fig. 2). These regions showed significant sequence identity to the catalytic domains of glycosyl hydrolase Family 11 xylanases (Table III). It would appear, therefore, that XYLA arose through the duplication of an ancestral gene that originally encoded a single catalytic domain enzyme.
Between the two putative catalytic domains of XYLA is a 40-residue sequence which is tandemly repeated (Fig. 2). The reiterated sequence exhibits extensive sequence identity with duplicated noncatalytic domains located at the C terminus of CelB and XYLA of N. patriciarum (12,13) and with a triplicated sequence from Piromyces MANA (Fig. 4). The overall identity between the 40-residue sequences is 38%.
The N-terminal portion of MANA comprised of a 450-residue domain that exhibits significant homology between residues 180 and 382 to the catalytic domains of two mannanases from Bacillus spp., mannanase A from P. fluorescens subsp. cellulosa and an endoglucanase from Bacteroides ruminicola (Fig. 5; Refs. 6, 8, and 27), 2 placing MANA in glycosyl hydrolase Family 26, according to the classification of Henrissat and Bairoch (7). Inspection of this sequence reveals the highly conserved and unusual protein motif of WFWWG. Upstream of this region (6 residues) is the dipeptide sequence HE which is conserved in all of the enzymes in Family 26 enzymes. Hydrophobic cluster analysis revealed two other acidic residues at positions 371 and 376 that are conserved in the Family 26 mannanases. 3 In view of the widespread participation of glutamate and aspartate residues in the acid-base catalysis of glycosidic bonds by ␤-glycanases, it is possible that the conserved glutamates and aspartates in the four Family 26 mannanases are candidate nucleophiles or proton donors in the active sites of these enzymes.   derivatives of xynA and manA were constructed and the capacity of the modifed genes to express functional enzymes was evaluated. Data presented in Fig. 1 show that a truncated form of manA, in which 176 bp of the 3Ј-coding region (pGM3), encoding the 40 amino acid C-terminal repeat, had been removed, encoded a mannanase which retained catalytic activity. In contrast, no functional mannanase was produced from a derivative of manA lacking 483 bp of 3Ј-coding sequence (pGM4). Similarly, cloning a truncated form of manA lacking 543 bp of 5Ј-coding sequence, into pMTL23p such that the mannanase gene was in-frame with the vector's lacZЈ translational start codon, did not generate a plasmid encoding a functional mannanase (pGM5). These data support the view that MANA comprises of a 450 residue N-terminal catalytic domain which is linked to a C-terminal noncatalytic reiterated sequence.
Derivatives of xynA, lacking only 48 bp of 3Ј-coding sequence ( Fig. 1; pGX10) did not encode a functional xylanase. Insertion of the 5Ј 1539 bp of xynA into pMTL22p on a SstI/EcoRI restriction fragment, such that the ATG start codon was in-frame with the vector's lacZЈ, generated a plasmid (pGX3) that also did not direct the synthesis of a functional xylanase. In contrast, cloning 1090 bp of the 3Ј-coding region of xynA into pMTL23p, such that the xylanase gene was in-frame with the vector's lacZЈ, generated a plasmid (pGX7) that encoded a functional xylanase. These data suggest that while the C-terminal catalytic domain homologue exhibited catalytic activity, the putative N-terminal catalytic domain was inactive.

Analysis of the Role of the Conserved Duplicated 40-Residue
Noncatalytic Domain-The conservation of the noncatalytic duplicated sequence within both cellulases and hemicellulases from two different species of anaerobic fungi suggests that this domain plays an important role in enzyme function. In view of the observation that the plant cell wall hydrolases of these eukaryotes associate into large multienzyme complexes that bind to cellulose (9, 10), two possible roles for the reiterated noncatalytic domain can be envisaged: (i) it could constitute a protein docking region which plays a pivotal role in the assembly of the complex; (ii) the sequence could function as a CBD. To distinguish between these two possibilities, the region of xynA encoding the 40-residue repeated sequence was amplified by polymerase chain reaction and cloned into pGEX-2T such that the inserted DNA was in-frame with the GST gene. The encoded fusion protein was used to probe Western blots of the polypeptides contained in the multienzyme cellulose-binding complexes from N. patriciarum and Piromyces, respectively. Data presented in Fig. 6 show that the GST-XYLAЈ hybrid protein bound selectively to 116-and 97-kDa polypeptides derived from Neocallimastix and Piromyces, respectively. GST did not bind to any polypeptides contained in the multienzyme cellulose-binding complexes from the two anaerobic fungi. To further investigate the protein binding specificity of the 40residue sequence, the domain was purified from the GST-XYLAЈ hybrid protein (encoded by pCF2), radiolabeled, and used to probe Western blots of proteins derived from E. coli cell-free extracts and the multienzyme cellulase complex from C. thermocellum, termed the cellulosome (3). The data, presented in Fig. 7, showed that although the duplicated sequence bound to the 97-kDa Piromyces polypeptide, it did not bind to any polypeptide in the E. coli extract or the C. thermocellum cellulosome. These data suggest that the capacity of the 40residue sequence to bind to the Piromyces-and Neocallimastixderived polypeptides is a specific interaction. To establish whether the duplicated domain bound to the nondenatured form of the Piromyces polypeptide, different quantities of native proteins from E. coli, the C. thermocellum cellulosome, and the Piromyces complex were dot-blotted onto Immobilon-P and probed with the 40-residue sequence. The data, presented in Fig. 7, revealed that the probe bound to polypeptides in the Piromyces complex, but not to the bacterial proteins. To establish whether the binding of the XYLAЈ sequence to the Piromyces complex can be competed out by the 40-residue sequence, increased amounts of nonradiolabeled duplicated sequence were added to incubations consisting of 32 P-labeled probe and Piromyces complex. The data (Fig. 7) showed that the nonradiolabeled C-terminal region of XYLA competed with the labeled probe for binding to the fungal complex. To establish whether the 40-residue sequence exhibited affinity for polysaccharides, the GST-XYLAЈ fusion protein and GST alone were incubated with xylan and cellulose, and protein that remained associated with the two polysaccharides was evaluated by SDS-PAGE and GST enzyme assays. The data (not shown) revealed that neither the fusion protein nor GST alone bound to xylan or crystalline cellulose (data not shown). The data described above clearly indicate that the 40-residue reiterated sequence functions as a protein docking domain, whose specificity for its target ligand is conserved between plant cell wall hydrolases from different anaerobic fungi. DISCUSSION A primary objective of this study was to evaluate the extent to which the noncatalytic reiterated 40-residue sequence is conserved in anaerobic fungi and to investigate the function of this domain in plant cell wall hydrolysis. To facilitate this objective, the primary structures of two hemicellulases from  the anaerobic fungus Piromyces were determined, and the relationship between the structure and function of these two enzymes was investigated. Data presented in this report indicate that XYLA contains two catalytic domain homologues. Although other plant cell wall hydrolases comprising of multiple catalytic domains, originating from distinct ancestral genes, have been reported from C. saccharolyticum (5) and Ruminococcus flavefaciens (29,30), the reiteration of catalytic domains in cellulases and xylanases has only been observed previously in enzymes derived from the anaerobic fungus Neocallimastix (12,31). The similarity in the molecular organization of XYLA from Piromyces and xylanase A from Neocallimastix supports the view that Neocallimastix and Piromyces are closely related or that there has been extensive gene transfer between the two organisms. In contrast to the Neocallimastix xylanase, which contained duplicated catalytic domains that were both functional, only the C-terminal domain of XYLA from Piromyces appears to exhibit catalytic activity. However, this interpretation must be viewed with some caution as: (i) it is possible that the folding of the protein in E. coli is different to that in the endogenous host, and the consequence of misfolding is an inactive N-terminal domain; (ii) a mutation in xynA, during the cloning of the cDNA, may have resulted in the inactivation of the first catalytic region. This is unlikely, because the same xylanase cDNA was isolated from a number of distinct recombinant phage, and each form of the xylanase exhibited the same properties. Inspection of the primary structure of the N-terminal catalytic domain of the Piromyces xylanase did not provide an explanation for its lack of catalytic activity; residues that are conserved in the catalytic domains of all other Family 11 xylanases were also present within XYLA. Of particular note is the conservation of Glu 133 and Glu 218 ; these active site residues function as the nucleophile and proton donor (32), respectively, in the acid-base hydrolysis of glycosidic bonds mediated by Family 11 xylanases. It is difficult, therefore, to evaluate precisely why the N-terminal catalytic domain homologue is inactive. The rationale for the evolution of a xylanase with a second, but inactive catalytic domain is also not readily apparent. It is possible that early in evolution xynA encoded an enzyme with two active catalytic domains, and although a mutation occurred in the 5Ј region of the gene resulting in the inactivation of the N-terminal domain, the xylanase activity of the C-terminal region of XYLA was sufficient selective pressure to ensure the retention of xynA within the Piromyces genome. This paper describes, for the first time, the primary structure and molecular architecture of a mannanase from an anaerobic fungus. The location of the fungal enzyme in a glycosyl hydrolase Family (Family 26) that also contains bacterial mannanases, provides evidence for a common evolutionary origin for the rumen fungal and bacterial mannanases, and raises the possibility that Piromyces manA arose through the lateral transfer of a prokaryotic gene to the eukaryotic microorganism. Inspection of the enzymes belonging to Family 26 reveals both mannanases and endoglucanases, suggesting that these two enzyme species evolved from a common progenitor sequence. Whether the ancestral sequence was an endoglucanase, mannanase, or ␤-glycanase with a broad substrate specificity remains to be elucidated. However, it is interesting to note that, while some endoglucanases display xylanase activity (33,34), the endoglucanases and mannanases in Family 26 exhibit no detectable cross-specificity, even though the two enzyme species have evolved from the same ancestral sequences (7).
A further major objective of this study was to evaluate the function of the highly conserved noncatalytic 40-residue sequence located in anaerobic fungal cellulases and hemicellulases. Noncatalytic repeated sequences have been observed in a number of cellulases and xylanases (3,35), and the role of these reiterated sequences has been defined for certain enzymes. For example, duplicated noncatalytic CBDs have been identified in cellulases from C. saccharolyticum, Cellulomonas fimi, Clostridium cellulolyticum, and Clostridium stercorarium (3). In addition, the duplicated 95-residue sequences, which exhibited 65% sequence identity, and are located in xylanase D from C. fimi, constitute a CBD and noncatalytic xylan-binding domain, respectively (36,37). In addition to repeated CBDs, a 23-residue duplicated sequence has been observed in cellulases and xylanases from C. thermocellum and C. cellulolyticum (3). This sequence plays an important role in the docking of the cellulases and xylanases to a noncatalytic scaffolding protein to form a multienzyme cellulase complex referred to as the "cellulosome" (28,38). Data presented in this study clearly showed that the reiterated noncatalytic domain observed in anaerobic fungal plant cell wall hydrolases is not a polysaccharide binding domain, but functions as a protein docking sequence that interacts with polypeptides of 116 and 97 kDa that are present in the multienzyme cellulase-hemicellulase complexes of Neocallimastix and Piromyces, respectively. Although the role of this protein docking domain has not been defined, we suggest that the polypeptides to which the conserved noncatalytic domain binds are the scaffolding proteins and that this proteinprotein interaction mediates the assembly of the multienzyme cellulase-hemicellulase complexes synthesized by these eukaryotic microorganisms. It should be emphasized, however, that other mechanisms could also contribute to the formation of the eukaryotic enzyme complexes. For example, it is possible that, in addition to the binding of the protein docking sequence to a scaffolding protein, direct interactions between the catalytic domains could also exist. were fractionated by SDS-PAGE, electroblotted onto Immobilon-P membrane, and probed with the 32 P-labeled 40-residue noncatalytic domain. The size of the hybridizing polypeptide is shown in kilodaltons. In B, the same three samples (native, nondenatured) were dot-blotted onto Immobilon-P at protein loadings of 10, 5, and 2.5 g and probed with the duplicated domain. In C, cellulase complex from Piromyces sp. (20 g) was dot-blotted onto Immobilon-P and probed with radiolabeled probe alone (lane 1) or with labeled probe that had been mixed 1:1 (lane 2), 1:5 (lane 3), or 1:10 (lane 4) with unlabeled probe. The duplicated 40residue noncatalytic fungal domain was prepared and labeled with 32 P as described under "Materials and Methods." Autoradiography was performed at Ϫ70°C for 24 h with intensifying screens.