A major new component in the cellulosome of Clostridium thermocellum is a processive endo-β-1,4-glucanase producing cellotetraose


  • Vladimir V. Zverlov,

    1. Institute for Microbiology, Technische Universität München, Am Hochanger 4, 85350 Freising-Weihenstepan, Germany
    2. Institute of Molecular Genetics, Russian Academy of Sciences, Kurchatov Sq., 123182 Moscow, Russia
    Search for more papers by this author
  • Nikolaus Schantz,

    1. Institute for Microbiology, Technische Universität München, Am Hochanger 4, 85350 Freising-Weihenstepan, Germany
    Search for more papers by this author
  • Wolfgang H. Schwarz

    Corresponding author
    1. Institute for Microbiology, Technische Universität München, Am Hochanger 4, 85350 Freising-Weihenstepan, Germany
    Search for more papers by this author

  • Edited by W.J. Mitchell

*Corresponding author. Tel.: +49 8161 71 5445; fax: +49 8161 71 5475. schwarz@mikro.biologie.tu-muenchen.de, wschwarz@wzw.tum.de


Cel9R, a major component in the cellulosome of Clostridium thermocellum, is one of the most prevalent β-glucanases in the complex after Cel48S and Cel8A. The recombinant product of gene celR is optimally active at 78.5°C on amorphous cellulose, carboxymethyl-cellulose, and barley β-1,3–1,4-glucan. From amorphous cellulose it produces initially cellotetraose which is slowly degraded to glucose, cellobiose and cellotriose. This product pattern indicates a processive endoglucanase-mode which was corroborated by the initial and simultaneous production of new reducing ends in the soluble as well as in the insoluble fraction of amorphous cellulose. p NP-Cellopentaoside is degraded to cellotetraose and p NP-glucoside, suggesting cellotetraose release from the non-reducing end. The newly discovered Cel9R thus is a novel type of cellulase in the cellulosome of C. thermocellum: a processive endo-β-1,4-glucanase producing cellotetraose as the primary hydrolysis product. The presence in the cellulosome and the hydrolytic mode of this cellotetraohydrolase has implications for our understanding of the in vivo conversion of cellulose by bacteria.


The strictly anaerobic, thermophilic bacterium Clostridium thermocellum possesses a particularly efficient enzyme system for the hydrolysis of crystalline cellulose: an extracellular enzyme complex called the cellulosome [1,2]. Although all β-glucanases in the complex hydrolyze the same β-1,4-glucosidic bond, their mode of action is different, either randomly attacking the cellulose molecule internally (endo-glucanase) or degrading it processively from one of the two ends (exo-glucanase). The difference between the two types of enzymes is often marginal at the protein sequence level and depends on the secondary structure of the active site pocket. Even within a structural family with high sequence homology (such as glycosyl hydrolase family GHF9) non-processive and processive endoglucanases as well as cellobiohydrolases exist [1]. The mode of activity is further modified by non-catalytic modules such as substrate binding modules (CBMs) [3,4].

Recently, on the basis of the C. thermocellum whole draft genomic sequence, 71 genes for hydrolytic cellulosomal components have been identified, 24 of them β-glucanases [5]. The composition of active cellulosomes was investigated by proteomic methods to determine the major proteins actually present in the cellulosome. Thirteen major protein components were identified; three of them were hitherto undetected: the putative β-glucanase Cel9R, xylanase Xyn10D and xyloglucanase Xgh74A [5].

Cel9R appeared as the third-most abundant β-glucanase in the cellulosome and the most abundant enzyme of glycosyl hydrolase family 9 [5]. The GHF9 components are processive and non-processive endoglucanases, or cellobiohydrolases, and are generally regarded as the major cellulose degrading factors in bacterial cellulase systems besides the GHF48 exoglucanase component, which is the most abundant catalytic protein in bacterial true cellulase systems [6–9]. Enzymes of these two glucanase families are major components in every bacterial cellulase system described so far [1,4,6]. Synergism between the GHF48 and GHF9 enzymes, and multifunctional enzymes containing the two catalytic modules have been described [8,10,11]. A major role of Cel9R in the hydrolysis of crystalline cellulose can thus be surmised.

C. thermocellum growing on cellulose has recently been shown to realize bioenergetic benefits that exceed the substantial cost of cellulosome synthesis [12,13]. These benefits result from taking up cellodextrins from the growth medium. Independent lines of evidence support the uptake of hydrolysis products with a mean chain length of about 4 during growth of C. thermocellum on cellulose [13]. However, the product of cellulosome-mediated cellulose breakdown is widely thought to be cellobiose, and longer cellodextrins have not been detected as primary hydrolysis products in in vitro assays of cellulosomal activity [1,6,14]. This discrepancy provides impetus to take a closer look at the C. thermocellum cellulase system in order to determine whether biochemical studies can be reconciled with data obtained using whole cells.

We describe the cloning and biochemical characterization of Cel9R from C. thermocellum, a newly detected major component of the cellulosome. Among the cellulosomal components this endoglucanase has a new and unusual mode of hydrolysis, which could explain the occurrence of larger cellodextrins found in C. thermocellum cells upon cellulose hydrolysis [13].

2Materials and methods

2.1Strains and media

C. thermocellum F7 was obtained from the All-Russian Collection of Micro-organisms (VKMB 2203). It was grown at 60°C in rubber-stoppered glass bottles with pre-reduced, anaerobic GS-2 medium containing 2% (w/v) cellobiose [15]. E. coli cultures were shaken at 37°C in Luria–Bertani broth with or without ampicillin (100 μg ml−1 w/v).

2.2Recombinant DNA techniques

Preparation of chromosomal and plasmid DNA, endonuclease digestion, ligation and transformation were carried out by standard procedures or according to supplier protocols. Plasmid DNA was prepared with the QIAprep Spin Miniprep Kit (Qiagen). Restriction digests of DNA were done as recommended by the manufacturer (MBI Fermentas or Boehringer Ingelheim Bioproducts). E. coli cells were transformed with plasmid DNA by electroporation (BioRad Gene Pulser™).

PCR of the celR gene was carried out using the synthetic oligonucleotide primers h19f (5′-caggatcctg tttttgcagca gactataac t-3′) and h19r (5′-tgatgtcgac gtaaagaata tccgatgttg g-3′) with chromosomal DNA from C. thermocellum F7 as template and the Expand-High-Fidelity PCR-System (Roche Diagnostics). The PCR amplified material was cloned in vector pQE31 (Qiagen) by digestion with restriction enzymes Bam HI and Sal I and sequenced from supercoiled double-stranded plasmid DNA on both strands using a LICOR automated sequencer (MWG Biotech) with the Thermo Sequenase fluorescent-labelling primer sequencing kit (Amersham Pharmacia Biotech).

2.3Enzymatic assays

Recombinant Cel9R protein was purified from a 400 ml E. coli (pQE31-CelR) culture with 3 ml Ni-NTA Superflow Agarose columns (Qiagen). Recombinant Cel8A protein was prepared from E. coli cells containing plasmid pWS11 [16]. Enzyme aliquots in standard assays were incubated in MES buffer (50 mM) at the optimum pH and temperature. The substrate concentration was 1% for soluble and 2% (w/v) for insoluble polysaccharides. Reducing sugars released from polymeric substrates were quantitatively detected by the 3,5-dinitrosalicylic acid (DNSA) method [17] assuming that one unit of enzyme liberates 1 μmol of glucose equivalent per minute and mg of protein. Specific activities were determined in the linear range of the reaction. All determinations were performed in triplicate.

The optimum pH was determined by measuring the specific activity of the enzyme at a given pH. Buffers were MES, sodium phosphate and Tris–HCl (50 mM) in their respective pH ranges. The optimum temperature was the temperature with the highest activity of the enzyme during 30 min incubation. Protein concentration was determined with Coomassie brilliant blue G250 [18].

p-Nitrophenol liberated from p NP-glycosides was measured photometrically by its absorption in alkaline solution (0.6 M Na2CO3) at 395 nm. One unit of activity is defined as the amount of enzyme producing 1 μmol p-nitrophenol min−1 (0.013 ΔOD395= 1 nmol).

2.4Assay of soluble vs. insoluble product release

The release of reducing sugars in the soluble and the insoluble fraction was performed with the insoluble substrate PASC. Enzyme reactions were set up as described above and incubated (pH 6.0, T= 78°C). Aliquots were taken at different time points and centrifuged to separate the insoluble and the soluble fraction. The pellet was washed twice with MES buffer (50 mM, pH 6.0) and resuspended in 1 vol of the same buffer. Reducing sugars were determined with the DNSA reagent in the supernatant of the enzyme reaction and the washed pellet separately. The pellet samples were centrifuged again after boiling and the optical density of the clear supernatant was determined.

2.5Rapid sampling and thin layer chromatography

For rapid sampling the reactions were stopped by submersing the reaction tubes in boiling water. Hydrolysis products from oligo- and polysaccharides were separated on 0.2 mm silica gel 60 aluminium plates (VWR, Darmstadt, D) with acetonitrile/water (80/20 v/v) as eluent. Sugars were detected by spraying the plates with a freshly prepared mixture of 10 ml stock solution (1 g diphenylamine and 1 ml aniline dissolved in 100 ml acetone) with 1 ml ortho-phosphoric acid, followed by heating the plates at 120°C until colour developed.


Avicel CF1, carboxymethyl-cellulose (CMC, low viscosity), and p-nitrophenyl-glycosides were obtained from Sigma–Aldrich (Deisenhofen, D), pachyman, barley β-glucan and tamarind xyloglucan from Megazyme International Ireland Ltd. (Bray, Ir), and cellodextrins from VWR International (Darmstadt, D). Phosphoric acid swollen cellulose (PASC) was prepared from Avicel CF1 according to [17].

2.7Nucleotide sequence accession number

The nucleotide sequence of the celR gene is deposited under GenBank Accession No.AJ585346.

3Results and discussion

The newly identified major cellulosome component Cel9R of C. thermocellum F7 was cloned in E. coli by amplifying the celR gene from genomic clostridial DNA with specific oligonucleotide primers based on the genomic sequence of strain ATCC 27405. The insert of the recombinant plasmid pQE31-CelR was sequenced and found to contain the complete amplicon (Accession No. AJ585346). The predicted amino acid sequences of the CelR proteins in the two strains (both 736 aa) differ in only three positions.

The catalytic module of Cel9R belongs to the large enzyme family GHF9, members of which are present in all cellulolytic bacteria, specifically in all cellulosomes found so far. Catalytic modules of this family have a sequence identity of >48% (similarity >62%) with CelR, often over more than 620 amino acid residues (GHF9 plus a C-terminal CBM3c module). The Cel9R protein has a modular architecture: an N-terminal leader peptide is followed by a catalytic GHF9 module, a CBM3c binding module and a dockerin module as is essential for cellulosomal components (Fig. 1). The modules are separated by short linkers rich in hydroxy-amino acid residues (“PTS-linkers”). The catalytic GHF9 module in Cel9R is not attached to an Ig-like module as in the cellulosomal cellobiohydrolases Cbh9A and Cel9K. Cel9R thus belongs to the structural theme B of the GHF9 hydrolases according to Bayer et al. [1]. Similar to Cel9R, other endo-glucanases of C. thermocellum, like Cel9N and the non-cellulosomal processive endoglucanase Cel9I, have an attached C-terminal CBM3c module which has, besides the preferential binding to non-crystalline forms of cellulose, a thermostabilizing function at least in some enzymes (Fig. 1) [1,19,20].

Figure 1.

Schematical structures of the major GHF9 β-glucanases present in the cellulosome. The non-cellulosomal protein Cel9I is added for comparison. Protein modules are drawn approximately to scale as boxes: LP, leader peptide; GHF, glycosyl hydrolase family; CBM, carbohydrate binding module; Doc, dockerin module; Ig, immunoglobulin homologous module; Fn3, fibronectin 3 homologous module. The last column designates the activity mode of the enzymes: pe, processive endo-glucanase; eg, endoglucanase; cbh, cellobiohydrolase.

The recombinant Cel9R protein was purified by His-tag affinity chromatography. The protein was homogeneous showing one protein band of 75 kDa in denaturing SDS–PAGE, close to the predicted value (74.4 kDa). The enzyme is a highly thermostable β-glucanase with optimal activity at 78.5°C and pH 6.0 on amorphous cellulose (PASC), carboxymethyl cellulose (CMC), Tamarind xyloglucan, and barley β-1,3–1,4-β-glucan. After 30 min incubation at 90°C half of its total activity was still present (data not shown). All these substrates contain β-1,4-glucosidic bonds (Table 1 The β-1,3-glucan pachyman was not degraded, which suggests a specificity for polymeric substrates with β-1,4-linkages. The aglycon is cleaved very slowly from p NP-cellodextrins with an activity about 4–6 orders of magnitude lower than the cleavage of the glycosidic bonds.

Table 1.  Hydrolytic activity of Cel9R on different polysaccharides
SubstrateGlycosidic bondSpecific activity (U μmol−1)
  1. Specific activity of the purified recombinant enzyme was determined in triplicate at optimal pH (6.0) and temperature (78°C). PASC, phosphoric acid swollen cellulose; p NP-G3, p-nitrophenyl-β-cellotriose; -G4 and -G5, -cellotetraose and -cellopentaose, respectively. Activity on p NP-glucosides only takes into account the liberation of p-nitrophenol, not the internal cleavage of the cellodextrins. The error margin of the activity determinations is indicated.

PASCβ-1,4193 ± 2
CMCβ-1,41250 ± 15
Tamarind xyloglucanβ-1,4149 ± 1.6
Barley β-glucanβ-1,3/β-1,438,990 ± 350
p NP-G3β-1,40.09 ± 0.01
p NP-G4β-1,42.53 ± 0.2
p NP-G5β-1,40.06 ± 0.01

The reaction products were separated after increasing incubation times by rapid sampling and thin layer chromatography. Interestingly, the amorphous, polymeric cellulose PASC was initially degraded to G4 without any detectable amounts of larger cellodextrins (Fig. 2(m) and (n)). Accumulated G4 was further hydrolysed to comparable amounts of the end products G2, and G1 plus G3 (Fig. 2(q)). A similar product pattern was observed for the non-cellulosomal processive endoglucanase Cel9I of C. thermocellum and the cellulase Cel9M of C. cellulolyticum[19,21]. Although the processivity of Cel9M was not investigated, its high activity on CMC suggests a predominant endo-mode of activity probably due to lacking a CBM.

Figure 2.

Thin layer chromatography of the reaction products from enzymatic activity. Substrates were incubated with enzyme and the reaction products after certain time intervals were subjected to rapid sampling and thin layer chromatography. Lane Mc shows markers: glucose (G1), cellobiose (G2), cellotriose (G3) and cellotetraose (G4) (from top down). Cel9R aliquots were incubated for 0, 5, 30 and 120 min with cellotetraose (lanes a–d), cellopentaose (e–h) and cellohexaose (i–l), and for 0, 5, 30, 120 min and 18 h with phosphoric acid swollen cellulose (PASC) (lanes m–q).

To analyse the degradation pattern of Cel9R further, the digestion of cellodextrins was investigated: G5 and G6 were degraded initially to G4 and G1 or G2, respectively, indicating that cellotetraose is cleaved off the substrate (Fig. 2(e)–(h) and (i)–(l), respectively) and corroborating the results with PASC. Hydrolysis of the G4 to G1, G2 and G3 was observed only after long incubation times: it is delayed and obviously slow (see also Fig. 2(a)–(d)).

To investigate the directionality of the cleavage, the product release from cellodextrins end-labelled with p-nitrophenyl residues was investigated: although p NP-G5 is hardly soluble (faint spot in Fig. 3) it was completely hydrolyzed by Cel9R within only 5 min to G4 +p NP-G1 showing fast cleavage of the glucosidic bond in contrast to the slow cleavage of the aglycon in p NP-G4 (which could not be shown in the TLC due to G4 contaminating the p NP-G4 preparation); p NP-G1 is identified as the expected corresponding product from p NP-G5 (Fig. 3(c)). This explains that the low chromogenic activity on p NP-G5 (Table 1) is because the colourless p NP-G1 is not processed further. Side products, obviously produced in a slower reaction, are G3 +p NP-G2 indicating some non-specificity.

Figure 3.

Hydrolysis of p NP-cellopentaoside. (a) p NP-G1 (p NP-glucoside); (b) p NP-G5 (p NP-cellopentaoside), no enzyme; (c) p NP-G5 incubated 5 min with Cel9R; (d) marker (abbreviations as in Fig. 2).

The much more efficient aglycon release from p NP-G4 in contrast to p NP-G5 (Table 1) and the preferred cleavage of G4 instead of aglycon from p NP-G5 (Fig. 3) indicate a specific cleavage of G4 from the non-reducing end of the substrate. Taken together with the data shown in Fig. 2 this is an indication for a processive mode and is in concordance with the results obtained with the non-cellulosomal enzyme Cel9I from C. thermocellum and the Thermobifida fuscaβ-glucanase E4 which have a similar module architecture [3,19,22].

To investigate the processivity of Cel9R, the hydrolysis pattern of the insoluble substrate PASC was compared to that produced by endoglucanase Cel8A [16,23]. PASC was used because it is a polymeric cellulosic substrate for testing endoglucanolytic as well as processive activity. The occurrence of reducing ends in the soluble and the insoluble fraction after incubation with enzyme was determined to evaluate the activity mode. A processive enzyme would exclusively produce short cellodextrins which initially appear in the soluble fraction, whereas an endo-glucanase should – at least initially – produce new reducing ends exclusively in the insoluble fraction. Cel9R produced almost equal amounts of reducing residues in the soluble as well as in the insoluble phase, whereas with Cel8A additional reducing ends were initially found only in the insoluble phase (Fig. 4). Cel8A thus behaves as expected for a non-processive endo-glucanase [16]. In contrast, the simultaneous appearance of additional reducing power in the soluble as well as in the insoluble fraction of PASC, as well as the relatively high activity on CMC (Table 1), where purely processive action is blocked by chemical modifications, indicate that Cel9R behaves as a processive endoglucanase.

Figure 4.

Estimation of the increase in reducing power. PASC was incubated with Cel9R (closed circles) and Cel8A (open circles). Glucose equivalents in the soluble (vertical axis) and insoluble fraction (horizontal axis) following incubation with increasing amounts of enzyme were estimated by using the dinitrosalicylic acid method. Soluble and insoluble fractions were separated by centrifugation, and assayed as described in Section 2.

Irwin et al. [3] suggested an accessory role for the CBM3c module in holding a single cellulose chain and feeding the non-reducing end into the active site pocket of cellulase E4, after the endo-glucanolytic activity of the catalytic centre has produced a nick. An analogous mode of action can be surmised for Cel9R. However, not all GH9-CBM3c enzymes are processive β-glucanases: despite high sequence identity and an identical GH9-CBM3c architecture, the highly homologous enzyme Cel9N was found to hydrolyse cellulose randomly and non-processively (endo-mode) (Fig. 1) [19]. Nevertheless, cellodextrins with a degree of polymerization of 4 (±1) were produced as end products from amorphous, insoluble cellulose by digestion with Cel9N, too, presumably because these cellodextrins are not suitable substrates for further degradation [19]. Other CBM3c containing cellulosomal GH9-components like Cbh9A and Cel9K produce cellobiose exclusively, but have different module architecture with an Ig-like module and an additional N-terminal CBM4 module (Fig. 1) [1,24–26].

This is the first description of a cellotetraose-producing processive enzyme in the cellulosome of C. thermocellum, Cel9R; a thermostable, processive endo-β-glucanase with unique features. It is active from the non-reducing end of the cellulose molecule and can be called a cellotetraohydrolase in analogy to the term cellobiohydrolase. The analysis of its mode of action was only possible through a kinetic investigation with rapid sampling of the initial hydrolysis products, a method which was previously also used to characterize the processive β-glucanase Cel9I [19]. Only with these methods can longer cellodextrins be detected which are formed relatively slowly but are further processed – they do not appear in significant concentrations in cell-free experiments but are available for immediate microbial uptake. We thus present supporting evidence for a distinctive new cellulose hydrolysis mechanism that has beneficial implications for the feasibility of microbial (as opposed to purely enzymatic) conversion of cellulosic biomass [2,13,27].

With the exception of Xgh74A and Xyn10D [paper submitted for publication] all major cellulosomal components are now characterized as recombinant proteins [5]. These are the components which are most probably the enzymes responsible for microbial hydrolysis of cellulose. Understanding their biochemistry and mechanistic relationships is the basis for the future reconstruction of a recombinant cellulase system which is efficient in vitro.


This work was supported by grants from the Deutsche Forschungsgemeinschaft DFG (SCHW489/7–1 and 436RUS17/113/04) and the Leonhard-Lorenz-Foundation to WHS, and from the A.-v.-Humboldt Foundation to VVZ (RUS1071807STP). We are very grateful to to W.L. Staudenbauer and L. Lynd for stimulating discussions and suggestions.