INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The cellulosome is believed to be one of the principle mechanisms by which cellulolytic microorganisms achieve efficient breakdown of the recalcitrant polysaccharides present in plant cell walls, particularly among anaerobic microorganisms (2, 7). The cellulosome has been described as a multiprotein complex, existing in cell-associated and/or extracellular forms in a wide range of cellulolytic organisms (5, 45). A typical cellulosome was first described in Clostridium thermocellum (6, 27, 29) which includes a noncatalytic subunit, called scaffoldin, and dozens of different enzymes.

The scaffoldin subunit plays a critical role in cellulosome assembly by its repetitive cohesin domains, each of which interacts with the dockerin domain of the individual cellulosomal enzymes. A cellulose-binding domain (CBD) of scaffoldin is responsible for mediating the binding of the cellulosome to the substrate. In C. thermocellum, the scaffoldin is believed to bind to another type of cellulosome-related protein (i.e., cell surface anchoring proteins) that carry C-terminal surface-layer homology domains (19, 32). The binding is mediated via a second type (type II) of cohesin-dockerin interaction, involving a C-terminal dockerin of scaffoldin and one or more cohesin-like domains in the anchoring proteins (30, 31).

Until recently, detailed molecular support for cellulosome organization was available only for the cellulolytic Clostridium species, C. thermocellum, Clostridium cellulovorans, Clostridium cellulolyticum, and Clostridium josui. In addition, a number of cellulosome-related signature sequences (i.e., cohesins and dockerins) have been discovered in different bacteria and fungi, although most of them are dockerin-tagged enzymes. The four clostridial scaffoldin genes that have been fully sequenced (20, 24, 37, 46) all contained a single CBD and a multiplicity of cohesin modules. Of these scaffoldins, only that of C. thermocellum exhibited a dockerin domain. Recently a scaffoldin from Acetivibrio cellulolyticus was described that also exhibited a C-terminal dockerin domain, similar to that of C. thermocellum, but differed from all known clostridial scaffoldins in that the gene included a catalytic module at its N terminus (13). Finally, another scaffoldin has been described from Bacteroides cellulosolvens that contains a similar arrangement of CBD, multiple cohesin modules (but of type II), and a single C-terminal dockerin domain (12). The growing evidence suggests a broad species-specific diversity in the sequence content and modular arrangement of the cellulosomal scaffoldins among the cellulolytic bacteria.

The rumen is a highly anaerobic environment that represents one of the most active sites for breakdown of plant cell wall material in nature (22). Genes encoding glycoside hydrolases have been characterized from the major cellulolytic ruminal bacteria, ruminal fungi (15, 16, 44), and ruminal protozoa (11). Nevertheless, the organization of enzyme systems that allow ruminal microorganisms to degrade lignocellulosic material is not well understood. In Ruminococcus species, which are among the most important ruminal cellulolytic bacteria in the rumen, previous biochemical and ultrastructural evidence indicated that protuberances are present on the cell surface that resemble those identified with cellulosomes in C. thermocellum (28). More recently dockerin sequences resembling those in Clostridium spp. were found in enzymes from ruminococci. These include xylanases, cellulases, and esterases from Ruminococcus flavefaciens 17 (1, 25) and cellulases from Ruminococcus albus (35, 36). The detection of dockerins in such enzymes is strongly suggestive of cellulosome organization in Ruminococcus spp., implying the presence of a scaffoldin-like protein and/or anchoring proteins that may organize enzyme subunits into such complexes. We report here the identification of two linked genes in R. flavefaciens 17 that appear to encode key structural components of a cellulosome complex in this species and present evidence for specific dockerin-cohesin interactions between the gene products.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and growth conditions. R. flavefaciens 17 was grown anaerobically at 39°C overnight according to the method of Bryant (8) in medium M2GSC (21) as modified by Miyazaki et al. (33). The culture was then used to inoculate Hungate-Stack medium (23), containing 0.2% Avicel (E. Merck, Darmstadt, Germany) as an energy source, and allowed to grow for up to 10 days in this medium. Escherichia coli strains used for cloning (XL10-gold and XL1-blue) and for subsequent expression of pET constructs [BL21(DE3) pLySs gold] were obtained from Stratagene. The vector pET28a was obtained from Novagen (Madison, Wis.). E. coli strains were routinely grown on LB medium with appropriate antibiotic selection at 37°C; conditions used for expression of cloned products are given below.

Preparation of protein samples from R. flavefaciens 17. R. flavefaciens cells and culture supernatant fluids were harvested after 10 days of incubation, by centrifuging (13,000 × g, for 10 min) the medium at room temperature. The pellet, containing residual Avicel plus cells, was resuspended in 1/40 of the original culture volume in 50 mM sodium phosphate buffer (pH 6.5) and 2 mM dithiothreitol and freeze-thawed twice, and the suspension was aliquoted and recentrifuged. The pellet from a 1-ml aliquot was resuspended in 250 µl of sample buffer containing 2% sodium dodecyl sulfate (SDS) (at 100°C for 5 min), and the proteins were separated by SDS-polyacrylamide gel electrophoresis (PAGE). Proteins were either visualized using Coomassie brilliant blue R250 or studied further by blot transfer or specific staining methods as described below.

PCR sequencing. PCRs were performed using a Mastercycler Personal device (Eppendorf, Hamburg, Germany). Samples (50 µl) contained 100 ng of template DNA, 10 pmol of each primer, 2.5 U of Super Taq (HT Biotechnology Ltd., Cambridge, England), 10 mM Tris-HCl (pH 9.0), 1.5 mM MgCl₂, 50 mM KCl, 0.1% Triton X-100, and 0.01% (wt/vol) stabilizer. Cycling was programmed as follows: a 3-min predenaturation step at 95°C was followed by 30 cycles comprising a 30-s denaturation step at 94°C, 55°C for 45 s, and an extension step at 72°C for 10 min. PCR products were separated on a 1% agarose gel and purified using a Spin-X centrifuge tube filter (Corning Inc., Corning, N.Y.). When possible, purified PCR products were cloned into the pGEM-T easy vector system (Promega, Madison, Wis.) and sequenced. Otherwise, selected fragments were sequenced directly using PCR primers on a model 377 DNA sequencer (PE Biosystems Foster City, Calif.). For longer PCR fragments, additional primers were designed from previously sequenced portions of the fragment in order to further extend the sequence.

Cloning and overexpression of recombinant proteins. Cohesin 2 and dockerin from ScaA and cohesin 4/5 from ScaB were amplified using primers designed to contain XhoI and NcoI sites (see Fig. 2 for details). The product was cloned into an NcoI/XhoI-treated pET28a(+) vector, such that a six-histidine stretch was fused at the C terminus of each recombinant module. Clones were expressed in E. coli Solopack Gold BL21 (DE3) pLysS (Stratagene, La Jolla, Calif.), at either 37 or 16°C as follows: single colonies were transferred into 50 ml of Luria-Bertani (LB) medium (30 µg of kanamycin per ml) and grown overnight at 37°C. A flask, containing 1 liter of LB medium, with added kanamycin (30 µg/ml), was inoculated with the 50-ml overnight culture and allowed to grow until an optical density at 600 nm of 0.8 to 1.0 was reached. Isopropyl--D-thiogalactopyranoside (IPTG) (1 mM) was then added, and the culture was incubated further for 4 h at 37°C. For expression at 16°C, 50 ml of an overnight culture was introduced into 1 liter of LB medium containing 1.2% glycerol and kanamycin (30 µg/ml), and the culture was incubated at 37°C until an optical density at 600 nm of 0.8 to 1.0 was reached. The culture was maintained at 4°C for 1 h, and 1 mM IPTG was added. The flask was incubated at 16°C without shaking for 1 h and then incubated with shaking (200 rpm) at 16°C overnight.

Purification of His-tagged proteins. The supernatant fluids from sonicated cell samples were incubated with Ni-NTA resin (Novagen) at a ratio of 1 ml of resin per 5 ml of supernatant fluids, and the suspension was incubated for 1 h at 4°C with gentle rocking. The resin was loaded in a plastic chromatographic column and washed with 10 volumes of wash buffer (50 mM sodium phosphate buffer [pH 8.0], 1 M NaCl, 20 mM imidazole, and 10% [vol/vol] glycerol). The His-tagged protein was then eluted using elution buffer (50 mM sodium phosphate buffer [pH 8.0], 0.3 M NaCl, 250 mM imidazole). The purified fraction was concentrated (10,000-Da molecular cutoff; VivaScience, Lincoln, United Kingdom) and stored at 4°C until further use.

GST-XynD fusion construct. A HindIII fragment of 1,423 kb that specifies residues 337 to 802 of the R. flavefaciens xynD product (i.e., a bifunctional xylanase-lichenase enzyme) was excised from the clone L956 (17). This fragment encodes a dockerin domain, a T-rich region, and the C-terminal catalytic domain, the product of which exhibits lichenase activity. The fragment was first cloned into the Pinpoint AX3 vector (Promega Corp.) and then excised using the enzymes NruI and SmaI, whose cleavage sites flank the insert. The resulting blunt-ended fragment was ligated into the vector pGEX2TK that had been cut with SmaI and phosphatase treated (Amersham Pharmacia Biotech Ltd., Little Chalfont, Bucks, United Kingdom), so as to create an in-frame fusion with a 235-amino-acid vector-determined polypeptide fragment, carrying glutathione-S-transferase (GST) at the amino terminus. Ampicillin-resistant transformant colonies of E. coli XL1-blue were isolated that expressed lichenase activity, which was detected by Congo red staining of agar overlays containing lichenan (17). The isolated transformants were shown to carry the predicted insert by sequencing across the fusion site and by PCR and restriction enzyme mapping.

Purification of GST-tagged proteins. The clarified supernatant fluids from sonicated host cells were incubated with shaking for 1 h at 4°C with glutathione-Sepharose 4B (Pharmacia Biotech, Uppsala, Sweden), previously equilibrated with phosphate-buffered saline (140 mM NaCl, 2.7 mM KCl, 10 mM N₂HPO₄, 1.8 mM KH₂PO₄ [pH 7.2]) at a ratio of 1 ml of slurry per 5 ml of clarified supernatant fluids. The resin was then loaded into a plastic chromatographic column and washed with 15 volumes of the same buffer. The bound protein was then eluted using 5 bed volumes of elution buffer (10 mM reduced glutathione in 50 mM Tris-HCl [pH 8.0]). The purified fraction was concentrated (molecular cutoff weight 10,000; Vivaspin 4), 0.05% (wt/vol) sodium azide was added, and the preparation was stored at 4°C until further use.

Immunoblotting. Concentrated, cell R. flavefaciens 17 supernatant fluids, cell-associated proteins, or the partially purified GST-XynD construct was subjected to SDS-PAGE (7% polyacrylamide). Separated proteins were then transferred onto a polyvinylidene difluoride (PVDF) transfer membrane (Immobilon-P; Millipore, Bedford, United Kingdom) and rinsed with wash buffer (50 mM Tris-HCl buffer [pH 7.5] containing 150 mM NaCl and 0.01% Tween 20). The membrane was then incubated for 1 h with blocking buffer (3% bovine serum albumin in wash buffer) and rinsed three times with wash buffer. The membrane was then incubated with the recombinant His-tagged protein (5 µg of protein/ml in blocking buffer, containing 5 mM CaCl₂) for 1 h at room temperature and washed three times with wash buffer. The membrane was incubated for 1 h at room temperature with peroxidase-conjugated antibody [Anti-His(C-term)-HRP mouse antibody; (Invitrogen, Groningen, The Netherlands), 0.24 µg/ml in blocking buffer] and washed five times with wash buffer. The bands were visualized using an appropriate chemiluminescent substrate (SuperSignal Substrate, Western Blotting; Pierce, Rockford, Ill.) following the manufacturer's instructions.

Glycoprotein staining. SDS-PAGE-separated proteins were transferred to a PVDF membrane in a semidry system (LKB 2117 Multiphor II electrophoresis unit). The membrane was stained for glycoproteins by periodate-induced biotinylation and visualization using streptavidin-peroxidase, according to the manufacturer's instructions (Glycotrack carbohydrate detection kit; Oxford GlycoSystems, Abingdon, United Kingdom).

Phylogenetic analysis. Phylogenetic trees were generated using the ClustalW program (http://www2.ebi.ac.uk/clustalw/). Protein sequences were obtained from the GenBank website (http://www.ncbi.nlm.nih.gov/) or via the Carbohydrate-Active Enzymes server (CAZy and CAZyModO websites, http://afmb.cnrs-mrs.fr/~pedro/CAZY/db.html and http://afmb.cnrs-mrs.fr/~pedro/DB/db.html, respectively), designed by Coutinho and Henrissat (10). See also the phylogenetic treatment of cellulosomal components in previous publications (4, 12, 13).

Nucleotide sequence accession number. The DNA sequence for the fragment and the complete scaB gene has been deposited in the GenBank database under accession number AJ278969.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of a putative R. flavefaciens cellulosomal protein. Cell-bound proteins from the culture fluids of R. flavefaciens 17 were separated by SDS-PAGE. A well-separated high-molecular-mass (~350-kDa) band was identified as a glycosylated protein (Fig. 1). The designated band was extracted from the gel and subjected to proteolysis, and the amino acid sequence of selected peptides was determined. Four degenerate primers (1F, 1B, 2F, and 2B [Table 1]) were designed based on two peptide sequences (Seq-1, DSIAWVVDTVAAYPGDEVTL; Seq-2, VVDLDNSQLPIAGAQFNIV [amino acids used for primer design are underlined]).

View larger version (85K): [in this window] [in a new window]   FIG. 1.   Identification of putative cellulosome-related glycoproteins from R. flavefaciens 17. A cellulose-grown culture was centrifuged, and both pellet (lane A) and supernatant fluids (lane B) were analyzed by SDS-PAGE on the same gel (7.5% polyacrylamide separating gel). A duplicate sample of the supernatant fluids was blotted onto a PVDF membrane and stained for the presence of glycosylated proteins (lane C). Molecular mass markers (in kilodaltons) are shown to the left of the gel.

Gene cloning and sequencing. Several fragments were amplified from the R. flavefaciens 17 genomic DNA, using the degenerate primers given above. Three major products (1.1, 1.7, and 2.8 kb) were purified. The 1.7-kb fragment was amplified, using primers 2F and 1B, and sequenced by PCR sequencing using primers 1F, 2B, and 16-5C. The deduced polypeptide contained three cohesin-like domains and a T-rich linker. The other two fragments were both amplified in the same PCR, using primers 1F and 2B. The 1.1-kb fragment was cloned and sequenced. The 2.8-kb fragment was sequenced using primers 1F, 2B, 28-3N, DEN, and 11-C. The resultant sequences indicated that the 1.1-kb fragment was part of the 2.8-kb fragment. The sequence of the 2.8-kb PCR fragment revealed the presence of two tandem open reading frames (ORFs). The first ORF, later termed scaA, represented an incomplete portion of a gene, coding for a protein that contains several cohesin-like domains and a dockerin-like domain at its C terminus. The remaining downstream portion of the 2.8-kb fragment included a second ORF, later termed scaB, which contained a series of cohesin-like domains. Two primers were designed from the two larger fragments (DEN from the 2.8-kb fragment and DEC from the 1.7-kb fragment) and used to determine whether these fragments are linked. Indeed, the resultant 0.9-kb PCR product was found to contain two complete cohesin-like domains and the linker segment between them. This short fragment, in fact, proved to be of added importance, since it verified the actual sequence of the site masked by the degenerate primer. The remainder of the scaB gene and an additional portion of scaA were obtained by library screening. For this purpose, the 0.9-kb PCR product was labeled with ³²P and employed as a probe to screen a Lambda-ZAPII EcoRI* R. flavefaciens 17 genomic library. Five positive clones were selected from 20,000 phage plaques. The positive clones were verified by PCR using primers DEN and DEC. Primers M13/F, M13/R, 16C-5, 20N-5, and 20C-3 were applied to amplify and sequence the inserted EcoRI fragment from the positive phage clones. In this manner, a 7.3-kb EcoRI fragment from R. flavefaciens was completely sequenced (Fig. 2). All portions of the 7.3-kb fragment were sequenced at least twice from different PCR products, except for the first T-rich linker of scaB, for which unique primers could not be designed in view of the numerous internal repeats. ScaB contains the originally sequenced peptide (Seq-1 and Seq-2), which indicated that ScaB indeed represents the ~350-kDa glycosylated protein initially extracted and isolated from the R. flavefaciens cells.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Overview of the 7.3-kb fragment selected from an R. flavefaciens EcoRI* library. (A) DNA fragments crucial for solving the sequence. (B) Domain organization of the C-terminal portion of scaA and the complete scaB gene. Both genes contain multiple copies of cohesin domains (numbered), and scaA contains a C-terminal dockerin domain (Doc). All of the domains are separated by short linker sequences (black); ScaB also includes two long T-rich linking segments (T-r1 and T-r2). The sequence of scaB shows a typical signal peptide at its N terminus and a C-terminal module (X) of unknown function. (C) Overexpression of selected modules. The C-terminal cohesin 3 and dockerin domains from ScaA were expressed separately, together with a His tag, in the pET-28a vector. A double-cohesin segment (cohesins 4 and 5) from ScaB was similarly expressed.

Domain structure of cellulosomal proteins from R. flavefaciens. The overall modular architecture of the C-terminal portion of ScaA and the complete ScaB is shown in Fig. 2B. ScaA harbors three successive cohesins and a dockerin at its C terminus. The linkers separating the ScaA modules are about 20 residues in length, most of which are threonines. The first four cohesins of ScaB are very closely linked. The two segments linking cohesin 4 to 5 and cohesin 5 to 6 are very similar to the threonine-rich linkers of ScaA. However, the two linking segments that flank cohesin 7 are special. They are particularly long (275 and 156 residues, respectively) and also contain a high content of threonine. Finally, ScaB includes at its C terminus an X module of unknown function.

Expression and function of selected ScaA and ScaB modules. In order to test the possible interactions between ScaA and ScaB, several of their key modules were subcloned and expressed with a C-terminal His-Tag in E. coli (Fig. 2C). These include the second cohesin and C-terminal dockerin from ScaA (ScaA-Coh2 and ScaA-Doc, respectively) and the double cohesin from ScaB (ScaB-Coh4/5). The expressed proteins were isolated on a Ni-nitrilotriacetic acid column and each preparation exhibited a major SDS-PAGE band, consistent with the estimated molecular weight according to the deduced sequence (Fig. 3). The calculated molecular weights (including the His tag) of ScaA-Coh2, ScaA-Doc, and ScaB-Coh4/5 are 17,784, 10,712, and 32,047, respectively.

View larger version (44K): [in this window] [in a new window]   FIG. 3.   SDS-PAGE of purified recombinant proteins, derived from R. flavefaciens ScaA and ScaB. Cohesin 2, the dockerin from ScaA, and the double cohesin (cohesins 4 and 5) from ScaB were subcloned and expressed as indicated in Fig. 2C. The His-tagged proteins were isolated on an Ni-nitrilotriacetic acid column; the eluted proteins were subjected to SDS-PAGE and stained with Coomassie brilliant blue. Molecular mass markers (in kilodaltons) are shown to the left of the gel.

View larger version (58K): [in this window] [in a new window]   FIG. 4.   Affinity blotting of cell-derived material from R. flavefaciens, using selected recombinant protein domains from ScaA and ScaB. Samples containing the pelleted, cellulose-grown culture (Fig. 1) were subjected to SDS-PAGE (Gel), and blotted onto PVDF membranes (Blots). The blots were probed with the indicated recombinant protein sample, and labeled bands (labeled at left [in kilodaltons]) were detected by chemiluminescence using peroxidase-conjugated, anti-His tag antibody.

Interaction of ScaA cohesin with an expressed R. flavefaciens glycanase. In order to further investigate the possible role of ScaA as a scaffoldin that integrates relevant enzymes into a cellulosome complex, we examined the interaction of the ScaA cohesin construct with an expressed recombinant form of a known R. flavefaciens dockerin-containing enzyme, XynD. The enzyme construct represented a 701-residue GST-XynD fusion protein that comprised a 1,423-bp HindIII fragment of the XynD gene (encoding for the dockerin domain) and T-rich linker and C-terminal lichenase domains of XynD, fused to the C-terminal end of GST. The estimated molecular weight of the construct was 78,953. The expressed protein was partially purified on a glutathione-Sepharose column. SDS-PAGE of the preparation exhibited a major ~80-kDa band (consistent with the predicted size of the intact construct) plus a series of lower-molecular-mass contaminating bands and breakdown products (Fig. 5). Affinity blotting of the contents of the gel, probed with the recombinant ScaA-Coh2, revealed a strong signal with the ~80-kDa band and two additional low-molecular-mass bands, presumably representing dockerin-containing proteolytic fragments of the intact construct. Such fragmentation has been observed previously for expressed preparations of this enzyme (17).

View larger version (48K): [in this window] [in a new window]   FIG. 5.   Affinity blotting of a known R. flavefaciens enzyme construct with a recombinant ScaA cohesin. A GST-XynD fusion protein (molecular weight, 78,953) was subjected to SDS-PAGE (Gel), and the proteins were transferred to a PVDF membrane (Blot), and probed with ScaA-Coh2 as described in the legend to Fig. 4. Molecular mass markers (in kilodaltons) are shown to the left of the gel.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Although indirect biochemical evidence has suggested the presence of a cellulosome in R. flavefaciens (14), direct support has only recently been reported upon detection of cellulosome signature sequences, i.e., dockerin domains from different cloned enzymes. Dockerin-containing enzymes suggest the existence of a scaffoldin-like counterpart, namely, a multiple-cohesin-containing protein, which would organize the various glycosyl hydrolases into a cellulosome complex. This hypothesis prompted the quest for scaffoldins, described in the present communication.

Preliminary studies on the binding of cloned dockerin regions from two R. flavefaciens 17 polysaccharidases had indicated recognition of an extracellular proteins(s) of approximately 130 kDa from cellulose-grown R. flavefaciens cultures (unpublished results), but this protein could not easily be recovered in a pure form. Another potential candidate for a cellulosome component, however, was a high-molecular-mass (>300-kDa) glycosylated polypeptide that was readily separable by one-dimensional SDS-PAGE. Limited peptide sequencing followed by the design of degenerate oligonucleotide primers allowed the amplification by PCR of portions of the coding sequence, following the strategy recently adopted successfully for A. cellulolyticus and B. cellulosolvens (12, 13). However, many Ruminococcus sequences are notoriously unstable in standard plasmid cloning vectors. Likewise, a similar instability has been experienced with the cloned PCR products. Consequently, the new sequence described in this work was achieved mainly through limited direct sequencing of appropriate PCR products and through analysis of a 7.3-kb fragment cloned in a phage vector. Quite fortuitously, the sequencing revealed portions of two ORFs, separated by a short segment of only 70 bp. Both of these ORFs, later designated scaA and scaB, encode polypeptides that contain cohesin-like sequences. Both of these were therefore strong candidates for components of a cellulosome complex.

Our evidence points to a specific interaction between the C-terminal dockerin region of ScaA and cohesins 4 or 5 (or both) of ScaB. This conclusion is supported by the interaction between these two domains. Moreover, the cloned ScaA dockerin recognizes selectively the large ~350-kDa polypeptide derived from R. flavefaciens cells, and the cloned cohesin 4/5 from ScaB binds specifically to the 130-kDa polypeptide from this strain. These data imply that the ScaA product is around 130 kDa, while the apparent size of ScaB is very roughly estimated at about 350 kDa. The two lengthy T-rich linkers in ScaB are probable sites for the glycosylation observed for the ~350 kDa polypeptide. Their sequences display a considerable internal homology within and between the two linker segments (Fig. 6). Secondary structure analysis (41-43) of the linker regions indicates a near-complete -sheet structure (data not shown), interspersed with distinctive loops at or near the glycine residues (notably the Gly-Pro dyads).

View larger version (31K): [in this window] [in a new window]   FIG. 6.   Alignment of internal segments within the T-rich linkers of ScaB. Note the high degree of internal repetition, the marked similarity among segments of these two linker regions, and the presence of distinctive Gly-Pro dyads (shown in boldface type). T-r1 comprises residues 939 to 1213, and T-r2 comprises residues 1356 to 1511.

The molecular mass predicted for the ScaB polypeptide is around 200 kDa, which suggests that anomalous migration and/or glycosylation would contribute to the observed low mobility in SDS gels. At this stage we can propose tentatively that the ScaA product may bind to the cell surface via the larger ScaB product. The C-terminal domain of unknown function is a possible candidate for involvement in cell surface attachment. In this context, preliminary evidence (Rinchon et al., unpublished) has shown that this domain is not involved in cellulose binding.

The cloned ScaA-Coh2 was able to interact with multiple R. flavefaciens polypeptides ranging from 60 to 100 kDa. The sizes of the polypeptides recognized by ScaA-Coh2 are consistent with the contention that they include plant cell wall-degrading enzymes, since known dockerin-containing enzymes from R. flavefaciens 17 range in size from 80 to 95 kDa (1, 18). This implies recognition of other dockerin domains present in a variety of R. flavefaciens proteins via another type of dockerin-cohesin interaction. Indeed, a recombinant dockerin-bearing construct, based on a known R. flavefaciens enzyme (17), was selectively recognized by ScaA-Coh2. If we assume that the R. flavefaciens system is analogous to those found in cellulolytic Clostridium spp., ScaA appears the more likely candidate to act as a scaffolding protein for the assembly of plant cell wall-degrading enzymes, while ScaB might play a cell wall anchoring role (Fig. 7).

View larger version (19K): [in this window] [in a new window]   FIG. 7.   Schematic representation of the proposed binding specificity of cellulosomal components from R. flavefaciens.

The mechanism for attachment of the complex to the substrate has yet to be established. As for other known cellulosomes, the process might involve an appropriate carbohydrate-binding module(s) within the unsequenced portion of the scaffolding protein or within an enzyme or enzymes that belong to the complex. Another possibility would be that substrate binding would be mediated by a cellulose-binding pilus-like surface protein, reported recently in a related ruminococcal species (34, 38). A. cellulolyticus, B. cellulosolvens, and now R. flavefaciens, which synthesize cellulosomes resembling those in clostridia, all belong to the Bacillus/Clostridium subphylum of the gram-positive bacteria based on 16S rDNA sequences (9, 26, 40). However, given the special nature of the rumen environment and the genetic distance between R. flavefaciens and cellulolytic clostridia (39), it is reasonable that significant differences would exist in cellulosome organization between these bacteria.

While ScaA and ScaB from R. flavefaciens both contain cohesin domains that have been identified by sequence similarity and by binding studies, the sequence relationships indicate a high degree of divergence from previously described cohesins. The closest percent identity between the R. flavefaciens ScaA and ScaB cohesins and those in Clostridium spp. is less than 27% for certain ScaB cohesins. The ScaA cohesin sequences are very similar to each other (85% identity) but show little overall similarity (less than 25%) to the clostridial cohesins. In fact, phylogenetic comparison of the known cohesins (Fig. 8) reveals that the R. flavefaciens cohesins represent a group distinct from those previously identified either in the clostridia or more recently in A. cellulolyticus and B. cellulosolvens. The ruminococcal cohesins emanate from a separate branch of the phylogenetic tree and are therefore herein designated type III cohesins. Finally, it can be noted that the seven cohesins described here from ScaB include two (cohesins 5 and 6) that are very similar to each other, while the remaining five are relatively divergent in sequence (Fig. 8). We have yet to investigate the binding specificity of all ScaB cohesins and so cannot rule out the possibility that other ScaB cohesins differ in their binding specificity from the cohesin 4/5 doublet studied here. The possible functional significance of cohesin and dockerin (1) divergence in R. flavefaciens cellulosome components is currently under investigation.

View larger version (38K): [in this window] [in a new window]   FIG. 8.   Phylogenetic relationship of R. flavefaciens cohesin domains. The individual ScaA and ScaB cohesins are numbered as they appear in sequence from the N terminus of the gene. The type I cohesins include those from the known scaffoldins (CipA from C. thermocellum, C. cellulolyticum CipC, C. josui CipJ, C. cellulovorans CpbA, and A. cellulolyticus CipV). Type II cohesins include those from the B. cellulosolvens CipBc scaffoldin and the surface anchoring proteins (Slp's) of C. thermocellum. The sources of the sequences used in this figure are given in reference 13. The scale bar indicates percentage (0.1) of amino acid substitutions.

Initially, the cellulosome was described on the basis of biochemical characterization as a cellulose-binding, multienzyme complex. Later molecular biological data and recognition of dockerin-containing enzymes and scaffoldin-borne CBDs and cohesins appeared to support the biochemical information. The recent findings (3) of noncellulosomal cohesin- and dockerin-like sequences in the genome of the archeon Archeaeoglobus fulgidus suggested that such signature sequences cannot be used alone to identify a cellulosome in a given organism. However, the A. fulgidus genome contains no known polysaccharidase, and the arrangement of the two cohesins and single dockerin precludes the formation of a multienzyme complex. In contrast, R. flavefaciens ScaA and ScaB, together with the demonstrated intermolecular connectivities, clearly allow the formation of cellulosome-like multienzyme complex. It is hard to predict whether in the future the term cellulosome will adequately categorize all types of multiprotein polysaccharide-degrading complexes. Nevertheless, the cellulosome concept should continue to provide a framework by means of which future discoveries in this field can be compared.