In Vitro Reconstitution of the Complete Clostridium thermocellum Cellulosome and Synergistic Activity on Crystalline Cellulose

Abstract

Artificial cellulase complexes active on crystalline cellulose were reconstituted in vitro from a native mix of cellulosomal enzymes and CipA scaffoldin. Enzymes containing dockerin modules for binding to the corresponding cohesin modules were prepared from culture supernatants of a C. thermocellum cipA mutant. They were reassociated to cellulosomes via dockerin-cohesin interaction. Recombinantly produced mini-CipA proteins with one to three cohesins either with or without the carbohydrate-binding module (CBM) and the complete CipA protein were used as the cellulosomal backbone. The binding between cohesins and dockerins occurred spontaneously. The hydrolytic activity against soluble and crystalline cellulosic compounds showed that the composition of the complex does not seem to be dependent on which CipA-derived cohesin was used for reconstitution. Binding did not seem to have an obvious local preference (equal binding to Coh1 and Coh6). The synergism on crystalline cellulose increased with an increasing number of cohesins in the scaffoldin. The in vitro-formed complex showed a 12-fold synergism on the crystalline substrate (compared to the uncomplexed components). The activity of reconstituted cellulosomes with full-size CipA reached 80% of that of native cellulosomes. Complexation on the surface of nanoparticles retained the activity of protein complexes and enhanced their stability. Partial supplementation of the native cellulosome components with three selected recombinant cellulases enhanced the activity on crystalline cellulose and reached that of the native cellulosome. This opens possibilities for in vitro complex reconstitution, which is an important step toward the creation of highly efficient engineered cellulases.

INTRODUCTION

Biomass, mainly consisting of plant cell wall polysaccharides, is the most abundant source of organic carbon compounds on earth and constitutes the richest renewable source of energy due to the sugars it contains. However, plants are built to resist degradation and hence are recalcitrant to enzymatic digestion. The basic fiber molecules in the plant cell wall consist of cellulose, hemicellulose, and lignin, a composite material which only a limited number of microorganisms can completely degrade. Most efficient among them is the bacterium Clostridium thermocellum, which exhibits one of the highest growth rates known for microorganisms on cellulose (19). Although restricted to anaerobic and thermophilic growth conditions, it is ubiquitously found in places with rotting biomass, regardless of the temperature (44, 45). The efficiency of its cellulolytic enzyme system on native cellulose is based on the formation of a huge enzyme complex, the so-called “cellulosome.” This enzyme system is organized on the outer surface of the bacterium (1, 2, 17, 19, 20, 21, 31, 32). It consists of a complex integrating protein (CipA) or scaffoldin protein composed of 9 type I cohesin modules to which enzymes are specifically docked with their attached type I dockerin modules (22, 39).

At present, not much is known how the complex is assembled on the bacterial cell surface, except that cohesin-dockerin association may form spontaneously (9). Analysis of the genomic sequence of C. thermocellum ATCC 27405 (GenBank no. CP000568; DOE Joint Genome Institute [JGI]) revealed 72 dockerin-encoding cellulosomal genes (43). More than half of them have been identified as encoding proteins in the complex (12, 29, 30, 42). Enzymes responsible for the degradation of cellulose, xylan, pectin, chitin, mannan, and other plant polysaccharides were identified, but quite a number of the genes have no cellulolytic function or even no known function (42). These cellulosomal components are multimodular proteins themselves, consisting of one or two catalytic modules and noncatalytic modules such as carbohydrate-binding modules (CBM) in addition to the obligatory dockerin module (32).

The components minimally required for crystalline cellulose degradation, their role in complex formation, and their synergistic interaction for cellulose breakdown have not yet been uncovered. However, proteomic analysis of isolated cellulosomes and transcriptomics revealed the most prevalent cellulosomal components and their differential regulation on substrate change (12, 29, 30, 42).

Various attempts have been made to disassemble the cellulosome and to reassemble it in vitro from defined components (26). Although this has not been successful so far, most of the components have been produced in recombinant form, and their structure and enzymatic function have been studied. This has enabled a functional analysis of the components. The reconstitution of small complexes composed of a recombinant miniscaffoldin combined with up to four recombinantly produced enzyme components was successful (10, 11, 28). Recombinant cellulases have also been complexed by attaching them to a protein particle or by surface display on bacteria and yeasts (25, 34, 37). The resulting activity was significantly lower than the activity of native cellulosomes. However, the data shed the first light on the nature of synergism on crystalline cellulose by complex formation. Much greater synergism was seen when mutants of C. thermocellum defective in the production of the scaffoldin protein CipA were utilized (44). The wild-type cellulosome and the same amount of cellulosomal components in uncomplexed form had about the same activity on soluble barley β-glucan; on crystalline cellulose, however, the cellulosome was about 15 times more active.

In this study, we investigated the reconstitution of the cellulosome on recombinant scaffoldin proteins. The investigation was initiated to show if the natural cellulosomes could be surpassed in activity by constructing artificial complexes, either on the scaffoldin protein CipA or on nanoparticles.

MATERIALS AND METHODS

Strains and media.

Clostridium thermocellum strain DSM 1237 (corresponding to ATCC 27405, JCM 12338, or NCIB 10682) was grown at 60°C in prereduced GS-2 medium (14) for liquid cultures or CM3 medium (36) for agar plates (with 1.5% agar [wt/vol]) containing 1.0% (wt/vol) cellobiose or 0.5% microcrystalline cellulose (MN300; Macherey & Nagel, Dueren, Germany). Escherichia coli Rosetta-gami B(DE3)pLysS was obtained from Novagen (Darmstadt, Germany).

Recombinant DNA techniques.

Preparation of chromosomal and plasmid DNA, endonuclease digestion, and ligation were carried out by standard procedures or in accordance with supplier protocols. Plasmid DNA was prepared with a QIAprep spin miniprep kit (Qiagen, Hilden, Germany). Restriction digests of DNA were done as recommended by the manufacturer of the restriction endonucleases (MBI Fermentas, St. Leon-Rot, Germany). Chemically competent E. coli cells were used for transformation.

Vectors pQE-30 to pQE-32 (Qiagen) or pET21a(+) (Invitrogen, Karlsruhe, Germany) were used for cloning truncated cipA and cellulase genes in the E. coli host strain TOP10 F′, BL21-Star (Invitrogen), or M15 (Qiagen). The complete cipA gene was cloned with the vector pET101/D TOPO in E. coli Rosetta-gami B(DE3)pLysS competent cells (Novagen) to compensate for underrepresented aminoacyl tRNAs in E. coli, which are necessary for expression of clostridial genes. The cipA gene fragments were amplified by PCR using the synthetic oligonucleotide primers Cip1f (5′-TATCAGCATGCTCTTAGTTGTGGCTAT), Cip1r (5′-ATAAAGAAGCTTTGCCAATTTCTACTACCAC), Cip2f (5′-AAATGCAGCGGATCCGATTACTTTGCTTGAAGTAGG), Cip3f (5′-AAAGGATCCGGTTGGCAGTGTAGTACC), Cip4r (5′-TTTAAAGCTTACTGCATCCAGAT), Cip6f (5′-AACAGAGCATGCAACACCTACAACACCTG), Cip6r (5′-AAGTGTAAGCTTGTTCGGAGTTATCGTCGG), CipAf (5′-CACCAAAAGTCATCAGTGAGCTCTTAGTT), CipAr (5′-CAGGTCGACGTAATCTCTTGATGT), pCel9Jf (5′-ATATCTGCATGCGCCGAAACAG), pCel9Jr (5′-TTTGCCCGGGCTTATAACTTGC), pCel9Rf (5′-CAGGATCCTGTTTTTGCAGCAGACTATAAC), pCel9Rr (5′-TAGCTTGAGCTCTTTGTTTTAAAGAATACG), pCel48Sf (5′-AACTGCATGCGCAGGTCCTTACAAAGGC), pCel48Sr (5′-AAAAGACCTGCAGAAGCCGTCC), pCel8Af (5′-AGCGGCAGGTGAGCTCTTTAA), and Cel8Ar (5′-GTAGGTGGTCGACGCTCTTTAT) with chromosomal DNA from C. thermocellum ATCC 27405 as the template and the KOD XL DNA polymerase (Novagen). For PCRs, the following conditions were used: 25 cycles of 30 s at 94°C, 5 s at 55°C, and 1 to 6 min at 72°C. The reaction was completed by incubation for 10 min at 74°C. Plasmids were sequenced from double-stranded DNA of selected clones.

Production and purification of recombinant protein.

A 20-ml portion of an overnight culture of the expression host were inoculated in 50 to 250 ml fresh LB medium. Cells were grown until the optical density at 600 nm (OD₆₀₀) reached 0.6. Expression was induced by adding isopropyl-β-d-thiogalactopyranoside (IPTG) at a final concentration of 1 mM (for truncated parts of CipA) or 0.1 mM (for complete CipA protein). Cultures were incubated for 3 to 6 h at 37°C (overnight at 30°C for complete CipA protein). Cells were harvested by centrifugation (8,000 rpm, 20 min, 4°C), and the resulting pellet was resuspended in 15 ml binding buffer (50 mM MOPS [morpholinopropanesulfonic acid], 0.1 M NaCl, 5 mM CaCl₂, 30 mM imidazole). After treatment with 1 mg/ml lysozyme for 30 min on ice, cells were sonicated, and soluble proteins were collected by centrifugation (15,000 rpm, 20 min, 4°C). The supernatant was applied to a HisTrap column (GE Healthcare, Frankfurt, Germany), and recombinant proteins were eluted with elution buffer (50 mM MOPS, 0.1 M NaCl, 5 mM CaCl₂, 0.5 M imidazole [pH 7.5]) in accordance with the manufacturer's protocols. Cel48S was reconstituted from inclusion bodies by washing sonicated cell lysates with 150 mM NaCl, dissolving the pellet in 5 M urea, 100 mM Tris-HCl (pH 8.5), and dialyzing against 20 mM Tris-HCl (pH 7.0), 1.5 mM cellobiose with a Slide-A-Lyzer cassette (30,000-kDa cutoff) with four buffer changes (35). The identity of all clones was verified by DNA sequencing, and the integrity of the purified miniscaffoldins was confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and staining with Coomassie brilliant blue G-250. The enzymes were tested for molecular size in SDS-PAGE and for activity with barley β-glucan.

Preparation of soluble cellulosomal proteins from cipA mutant SM1 (SCP_SM1).

Mutant SM1 (44) was grown in GS-2 medium containing 1% cellobiose and centrifuged (1,000 × g, 20 min, 4°C). Proteins were precipitated from cleared supernatants by dropwise addition of saturated ammonium sulfate solution until 60% (vol/vol) was reached and incubation over night at 4°C. The protein pellet was resuspended with a buffer of 50 mM MOPS, 0.1 M NaCl, 5 mM CaCl₂, and 0.02% NaN₃. Protein concentration was determined with the Bradford reagent (Bio-Rad Laboratories, Munich, Germany). Proteins were ultrafiltered with spin columns with an exclusion size of 10,000 Da (Vivaspin 500; Vivascience AG, Aubagne, France) to remove salts and impurities. Protein concentration was determined with Coomassie brilliant blue G-250 (Bradford reagent).

Binding to miniscaffoldins.

The stoichiometry of binding to the miniscaffoldin CBM-Coh3-Coh4 and CipA was tested with the purified recombinant cellulosomal cellulase Cel8A protein-bearing a dockerin module. Based on this test, the scaffoldins were mixed with Cel8A in molar ratios of 1:2 and 1:9 (1 dockerin to 1 cohesin) in calcium-containing buffer. A gel shift experiment with nondenaturing PAGE showed complete binding of the two components into a single band, indicating binding of one cohesin to one dockerin module (Fig. 1).

Fig 1.

Electrophoretic mobility shift of cohesin-dockerin complexes in native polyacrylamide gel electrophoresis. Lane 1, CipA protein (34 μg); lane 2, cellulase Cel8A (10.4 μg); lane 3, CipA-Cel8A complex; lane 4, soluble cellulosomal proteins from the cipA mutant SM1 (SCP_SM1) (32 μg); lane 5, CipA-SCP_SM1 complex (molar ratio of 1:1, or 34:32 μg); lane 6, CipA SCP_SM1 complex (molar ratio, 1:9, or 3.8:32 μg). Samples were assayed by 6% PAGE and stained with Coomassie blue R-250.

Cellulosome preparation.

Cellulosomes were prepared from 0.5 liter of cellulose-grown cultures of C. thermocellum DSM 1237 at late logarithmic growth stage in a manner similar to the affinity digestion method (27, 40). Culture supernatant was incubated with 100 mg/liter phosphoric acid-swollen, amorphous cellulose (PASC) overnight at 4°C and then centrifuged at 13,000 rpm for 20 min at 4°C. The pellet was resuspended in 50 mM Tris, 5 mM CaCl₂, 5 mM dithiothreitol (pH 7.0) and then dialyzed for a minimum of 6 h at 60°C in a Slide-A-Lyzer cassette (molecular weight, 10,000) (Pierce) against 2 liters of distilled water with a water exchange every 2 h until PASC was dissolved completely. The supernatant after centrifugation contained the cellulosomes. Cellulosome preparation was analyzed by exclusion chromatography on a Superose 6 HR 10/30 column (GE Healthcare) with an exclusion size of 46,104 kDa.

The major protein peak immediately following the bed volume of a gel filtration run contained cellulolytic activity and was collected.

Coupling to nanoparticles.

Nanoparticles (20 mg; Estapor microspheres; Merck, Darmstadt, Germany) with a diameter of 0.110 ± 0.007 μm, a solid content of 9%, and 497 μeq g⁻¹ of carboxylic surface groups were washed three times in 2 ml activation buffer (50 mM MES, 0.5 M NaCl [pH 6.0]) by separation with a strong NdFeB disc magnet (1.43 ± 0.2 T). The modified particle surface was activated by adding fresh EDC solution [water-soluble carbodiimide 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride; Pierce] and sulfo-NHS solution (N-hydroxysulfosuccinimide; Pierce) to final concentrations of 2 mM and 5 mM, respectively, for 15 min at room temperature (15, 18, 33). Particles were magnetically separated and washed twice with 2 ml reaction buffer (0.1 M NaH₂PO₄, 0.5 M NaCl [pH 7.2]). Five milligrams of O-(2-aminoethyl)-O-(2-carboxyethyl)-polyethylene glycol (PEG) 3000 hydrochloride was dissolved in 100 μl reaction buffer under a nitrogen atmosphere and added to the activated particles. The activated particles and the amino groups of the PEG-based linkers were covalently linked within 3 h. The buffer was changed to 2 ml activation buffer. The carboxylate group at the end of the covalently bound linker was activated with EDC and sulfo-NHS. After two washing steps with 2 ml reaction buffer, 10 mg N_α,N_α-bis(carboxymethyl)-l-lysine hydrate (NTA) was added and kept at room temperature for 3 h. Particles were washed three times with 2 ml distilled water. One milliliter of 1 M NiSO₄ was added to complex the carboxylate groups of NTA (15). After 5 min, the particles were washed twice with 2 ml distilled water, followed by two washes with 2 ml 50 mM MOPS, 0.1 M NaCl, 5 mM CaCl₂ (pH 6.0). The activated nanoparticles were complexed with 1 to 1.5 mg His-tagged protein by overnight incubation in 2 ml 50 mM MOPS, 0.1 M NaCl, 5 mM CaCl₂ (pH 6.0). Coupling efficiency was determined by spectrophotometric measurement of the optical adsorption (595 nm, Bradford assay) of protein content in the supernatant before and after coupling.

Enzymatic assays.

Enzyme samples were incubated in MES buffer (50 mM) at pH 6.0 and 60°C. Reducing sugars were quantified in the linear range of the reaction by the 3,5-dinitrosalicylic acid method with glucose as the standard (38). One unit of enzyme is defined as 1 μmol of glucose equivalent liberated per minute. The following soluble and insoluble substrates were used at 0.5% and 0.67% (wt/vol), respectively: soluble barley β-glucan (Megazyme International, Bry, Ireland) and low-viscosity carboxymethyl cellulose (CMC) (Sigma-Aldrich, Deisenhofen, Germany) and insoluble cellulose powder MN300 (Serva, Heidelberg, Germany), Avicel microcrystalline cellulose (Serva 14201), and bacterial cellulose (Cellulon microfibrous cellulose; CP Kelco, Atlanta, GA). All enzymatic estimations were performed at least in triplicate.

RESULTS

Free native cellulosomal proteins, i.e., supernatant cellulosomal proteins of a C. thermocellum mutant deficient in the scaffoldin protein CipA (SCP_SM1), were used in experiments on reconstitution of artificial cellulosomes (44). Cellulosome-like particles were assembled on six different (mini)scaffoldins. These were constructed by PCR amplification of various fragments of the cellulosomal scaffoldin gene cipA from the wild-type C. thermocellum strain DSM 1237. They contain one, two, or three cohesins for binding cellulosomal components, some together with the carbohydrate-binding module CBM3 from CipA to investigate the influence of cellulose binding on degradation (Fig. 2). In addition to these miniscaffoldins, the complete CipA gene was cloned and expressed in E. coli. All recombinant proteins were purified by affinity chromatography (immobilized-metal-ion affinity chromatography).

Fig 2.

Recombinant scaffoldin constructs. CipA and its derivatives from C. thermocellum are shown. Numerals (1, 2, 3, etc.) indicate Coh1, Coh2, Coh3, etc. CBM, carbohydrate-binding module.

Reassociation experiments.

The different cohesin-bearing recombinant miniscaffoldins and the complete CipA protein were complexed with the dockerin-bearing cellulosomal components SCP_SM1 in the appropriate stoichiometric ratio. Successful complex formation was verified using native PAGE (Fig. 1). The hydrolytic activities of the assembled cellulosome particles and the nonassembled mixtures were determined on soluble, amorphous, and crystalline cellulosic substrates. The release of reducing sugars after 24 h incubation at 60°C was approximately 8.0, 1.1, and 2.7 U/mg protein on barley β-glucan, CMC, and PASC, respectively, regardless of the presence of a scaffoldin and the number of its cohesins (Table 1). In contrast, bringing two cellulolytic components into close proximity in a complex (Coh1-Coh2) enhances the catalytic efficiency on crystalline cellulose about 2-fold. The activity of enzymes bound to a miniscaffoldin containing a CBM and three cohesins (Coh1-Coh2-CBM-Coh3) or to recombinant CipA was about 5 and 13 times higher, respectively (Table 1). The cellulolytic activity of the enzyme complexes was further improved about 2-fold when the scaffoldin contained a CBM (Coh1 versus Coh2-CBM, or Coh1-Coh2 versus Coh2-CBM-Coh3 or CBM-Coh3-Coh4). The position of the CBM in relation to the cohesins does not seem to be of great importance for activity (Coh2-CBM-Coh3 versus CBM-Coh3-Coh4).

Table 1.

Enzymatic activities of free cellulosomal enzymes

Substrate	Activity (mU/mg protein) of SCP from C. thermocellum SM1a
	Free enzyme	Coh1	Coh2-CBM	Coh1-Coh2	Coh2-CBM-Coh3	CBM-Coh3-Coh4	Coh1-Coh2-CBM-Coh3	rCipA	Cellulosome
β-Glucan	7,912 ± 582	7,634 ± 489	8,014 ± 561	8,045 ± 680	7,845 ± 597	7,792 ± 488	8,056 ± 690	8,178 ± 609	8,267 ± 573
CMC	1,134 ± 95	1,065 ± 106	1,129 ± 86	1,178 ± 113	1,067 ± 105	1,272 ± 157	1,108 ± 98	1,201 ± 124	1,267 ± 125
PASC	2,731 ± 257	2,847 ± 386	2,785 ± 206	2,629 ± 308	2,756 ± 193	2,683 ± 247	2,741 ± 313	2,850 ± 200	2,782 ± 306
MN300	30 ± 2.3	32 ± 2 0.4	67 ± 5.1	63 ± 5.8	112 ± 9.3	115 ± 10.2	145 ± 11.8	373 ± 47.6	423 ± 42.5
Avicel	12 ± 1.5	13 ± 1.5	25 ± 1.9	19 ± 2.5	39 ± 4.6	40 ± 4.7	57 ± 4.9	150 ± 22.3	190 ± 20.0

^aActivity of free enzymes from C. thermocellum mutant SM1, incubated with miniscaffoldins and recombinant CipA protein on barley β-glucan, CMC (soluble), MN300 cellulose, or Avicel (crystalline) under standard conditions (60°C, pH 6.0). The specific activity of native purified C. thermocellum cellulosome is shown in the last column. CBM, carbohydrate-binding module. Values are means ± standard deviations.

The complexed miniscaffoldin Coh2-CBM-Coh3-Coh4 showed about 2.6 times less activity than the full-size CipA protein on insoluble substrates. The reconstituted full-size cellulosome is only 12% and 21% less active than the native cellulosome on MN300 and Avicel, respectively.

A possible selectivity of the cohesin binding sites for the binding of different enzyme components was tested with two cohesin modules, the cohesins Coh1 and Coh6. These showed the greatest sequence difference in a comparison of the nine cipA cohesins (3) and were thus chosen to compare their specific interaction with the dockerin-bearing hydrolases in SCP_SM1. A single cohesin (Coh1 or Coh6) was immobilized, and SCP_SM1 was applied to the column in molar excess. After thorough washing of the column, the bound dockerin-bearing hydrolases were eluted by imidazole-mediated release of the cohesin-dockerin-enzyme complexes and subjected to SDS-PAGE. The profile of the SCP_SM1 bound to the cohesins was indistinguishable. This suggests a certain nonspecificity of binding of all cellulosomal dockerins to the different CipA cohesins when they are in competition with each other (Fig. 3).

Fig 3.

Elution profile of two different cohesins interacting with soluble proteins (SCP_SM1). Isolated Coh1 and Coh6 were bound to a HisTrap column and loaded with SCP_SM1. Lane 1, SCP_SM1; lane 2, proteins retained by Coh1; lane 3, proteins retained by Coh6. Molecular masses (kDa) are indicated on the left, proteins are identified on the right.

Addition of recombinant enzymes.

To improve the cellulolytic activity, the proportion of cellulases was enhanced by addition of recombinant cellulases. This reduces the fraction of, e.g., xylanases and a chitinase. Recombinantly produced cellulases Cel48S, Cel9J, and Cel9R were added before complex reconstitution (molar ratio, 1:1:1; together, 30 mol% [wt/wt] of all enzyme components, assuming an average molecular mass of 80 kDa for the SCP_SM1). The stoichiometry of all enzymes added to the complex was calculated to be one dockerin per cohesin. The added cellulases are major exo- and endoglucanase components in the cellulosome (12, 42). The raised stoichiometry for cellulases increased the average activity of the synthetic complex on cellulose and came close to that of the native cellulosome (Table 2).

Table 2.

Hydrolytic activity of complexes with added recombinant cellulases^a

Complex	Sp act (mU/mg)
SCPSM1	12 ± 1.5
SCPSM1 + CipA	150 ± 22.3
70% SCPSM1 + 30% rEnz	30 ± 2.8
70% SCPSM1 + 30% rEnz + CipA	165 ± 21.4
Cellulosome	190 ± 20.0

^aHydrolytic activity on 0.5% (wt/vol) Avicel of CipA with SCP_SM1 (100 mol%) and 70 mol% SCP_SM1 with 30 mol% recombinant cellulases on 0.5% (wt/vol) Avicel. Activity was determined from the released reducing sugars per mg of enzyme; the nonenzymatic CipA component was not taken into account. The cellulosome is the native preparation from wild-type C. thermocellum. rEnz, 10 mol% each of Cel48S, Cel9R, and Cel9J.

Cellulases on nanoparticles.

To enlarge the complexes beyond the size of a single CipA scaffoldin, spherical nanoparticles were used to reconstitute cellulosome-like particles on an inorganic basis. To achieve this, Coh3-Coh4 miniscaffoldins were chemically bound to the surfaces of 0.11-μm ferromagnetic nanoparticles by peptide-coupling chemistry (15, 18). The maximum coupling efficiency of protein (79.6 ± 4.5 μg mg⁻¹ of nanoparticles) was achieved at an EDC/sulfo-NHS ratio of 2/5. This equates to approximately 1,300 miniscaffoldins per nanoparticle, corresponding to 2,600 binding sites for cellulosomal components. The loading of the COOH-modified nanoparticle surface (9% solid content, 497 μeq g⁻¹) was better than that of NH₂-modified surfaces (5% solid content, 11 μeq g⁻¹), as was the coupling efficiency (61.7 ± 3.8 versus 79.6 ± 4.5 μg mg⁻¹). The scaffoldin-loaded nanoparticles were complexed with the SCP_SM1.

Enzyme-loaded nanoparticles had a specific activity that was within the assay error identical to that of the cellulosome on soluble barley β-glucan and CMC and on PASC (Table 3). This indicates that the binding of miniscaffoldins to nanoparticles has no inhibitory effect on the enzymatic activity itself. Identical complexes not bound to nanoparticles had comparable activity on crystalline cellulose, and neither stimulation nor inhibition could be shown. The inflexibility created by the binding to the nanoparticle surface seems to have been neutralized by neighboring effects and does not seem to be an obstacle for enzymatic activity on the substrate surface (Tables 1 and 3).

Table 3.

Cellulosomal particles on nanoparticles

Substrate	Sp act (mU/mg protein) of SCPSM1a
	Free enzyme	NP + Coh1	NP + Coh1-Coh2	NP + Coh2-CBM-Coh3	NP + CBM-Coh3-Coh4	NP + Coh1-Coh2-CBM-Coh3	Cellulosome
β-Glucan	7,912 ± 894	7,576 ± 931	8,045 ± 734	7,845 ± 732	7,926 ± 801	8,132 ± 836	8,267 ± 1,021
CMC	1,134 ± 214	1,165 ± 214	1,098 ± 154	1,253 ± 136	1,184 ± 107	1,190 ± 122	1,267 ± 186
PASC	2,731 ± 325	2,785 ± 432	2,764 ± 384	2,698 ± 467	2,651 ± 307	2,703 ± 294	2,782 ± 498
MN300	30 ± 8	31 ± 9	62 ± 17	107 ± 21	108 ± 22	147 ± 31	423 ± 48
Avicel	12 ± 1.5	13 ± 1.2	26 ± 1.9	45 ± 3.1	44 ± 3.7	59 ± 5.2	190 ± 20.0

^aSpecific activity of SCP_SM1 in complexed form on nanoparticles with different miniscaffoldins and native purified cellulosome on MN300 and Avicel (0.5% [wt/vol]) at 60°C. CBM, carbohydrate-binding module; NP, nanoparticles. Each value is the average of triplicate measurements ± standard deviations.

In addition, the protein complexes on nanoparticles showed activity at a wider range of pHs and increased stability (Fig. 4), as expected from reports of enzymes immobilized on, for instance, polystyrene particles (5). More than 60% of the enzymatic activity was found between 42.5 and 67.8°C with the noncomplexed enzymes (SCP_SM1), versus between 37.8 and 71.0°C with the nanoparticles (data not shown), and a pH of 5.6 to 8.2 versus 4.7 to 8.8. After 40 days' storage at 60°C and pH 6.5, 60% of the activity was observed for the free enzymes, whereas 95% was observed for the particle-bound enzymes (Fig. 4).

Fig 4.

Relative activity at different time points (days) of free and immobilized nanoparticle-bound cellulosomal components (SCP_SM1). △, free SCP_SM1; ■, SCP_SM1 immobilized miniscaffoldin Coh1-bearing nanoparticles.

Due to their superparamagnetic behavior, the nanoparticles could be separated from a digested cellulose preparation in a magnetic field with a recovery rate between 93 to 97%. In contrast to the nonbound complexes, the loaded particles could be reused. Particles could also be recharged with new enzymes, but with decreasing efficiency: experiments showed 79.6% ± 4.5% recharging in the first binding cycle, 51.3% ± 3.7% in the second, and 24.6% ± 3.2% in the third.

DISCUSSION

Cellulosomes of C. thermocellum are the most effective cellulases for the degradation of crystalline cellulose (8). Here, they were reconstituted for the first time in vitro in active form from a complete set of cellulosomal components derived from the noncomplexed, soluble protein mixture from the supernatant of the cipA mutant SM1 (SCP_SM1), which is deficient for the scaffoldin (44). These were combined with externally added, recombinantly produced CipA scaffoldin backbone proteins from C. thermocellum containing 1, 2, 3, and 9 cohesins.

After initial attempts of many authors to disassociate and reconstitute the cellulosome had failed, there were many efforts to reconstitute a cellulolytic “minicellulosome” from recombinant components (4, 10, 24, 26). All of them used scaffoldin proteins with a limited number of cohesin-binding sites or a limited number of enzyme components (6, 7, 11, 25, 34, 37). In most of these complexes, a specific dockerin-binding site was assigned to a certain enzyme component by using cohesin-dockerin pairs with different specificities. The specific activities of these minicellulosomes (designer cellulosomes) were usually not compared to that of native cellulosomes. Fierobe et al. found that trifunctional minicellulosomes were about 12 times less active on microcrystalline cellulose than native cellulosomes from Clostridium cellulolyticum (11).

In contrast to the aforementioned minicellulosomes, the complexes assembled here on full-size CipA protein closely resemble the native cellulosomes of C. thermocellum. As expected, complexes containing the complete scaffoldin CipA reached a specific activity on crystalline cellulose close to the activity of native cellulosomes (79 to 88%).

From the in vitro experiments, it can be assumed for the in vivo assembly that the dockerin-bearing components of the cellulosomes that are not directly cell wall bound are secreted in a mature and correctly folded form into the medium. The results do not contradict the assumption that the assembly of the fully active cellulosomes is achieved by displaying a mature CipA protein on the outside of the cell; the independently secreted cellulosomal components seem to be assembled spontaneously by cohesin-dockerin interaction, as was shown previously with in vitro-created components (6, 9, 22).

All dockerin-bearing proteins (at least the major components) apparently have identical or nearly identical binding affinities at each cohesin position of the scaffoldin. This has already been proposed from affinity blotting experiments and from Biacore determination of the binding affinity of single defined cohesin-dockerin pairs (13, 23, 39). However, in contrast to the determination of the binding force of defined dockerin-cohesin pairs, the difference in the affinity of various dockerins for single or multiple cohesins should be much clearer if binding occurs in competition. Even slight differences in binding “on” and “off” rates should have been clearly visible in the distribution of band intensities. On the other hand, Bomble et al. (4) predicted a differential binding for proteins displaying different flexibility. Such differences could not be seen with the experimental setup used here.

In complexes with a small number of cohesins, e.g., with the Coh2-CBM-Coh3 miniscaffoldin, the juxtaposition of two enzymes can accordingly be thought to be purely statistical. Since not all statistically occurring enzyme pairs will show synergistic cooperation on crystalline cellulose, the chance for synergistically active combinations of enzymes is rarer in smaller complexes. This leads to a lower activity and thus a lower degree of synergy.

The binding of a cellulosomal enzyme to a single cohesin via its dockerin does not by itself stimulate its activity on crystalline cellulose (see the data for the complexes with Coh1 in Table 1). However, it was reported to increase stability, pH, and temperature range (16). When the noncatalytic CBM3 from CipA was connected to the cohesin, stimulation of the enzyme by a factor of 2 was observed on crystalline cellulose, as was shown for other cellulases (6) (Table 1). An identical increase in activity was observed for complexes with one and two cohesins, regardless of whether the CBM was in the N-terminal position or a central position (11). This is an indication that the targeting effect on the substrate would be position independent, at least in connection with a small number of cohesins. Similar results were described previously (6).

A distinct synergism was observed when the number of cohesins in the complex was increased. This increase from one cohesin module to two showed a synergism of 1.7 on both crystalline substrates (Table 1). However, the increase in activity from two to three cohesins was only a factor of 1.3 to 1.4, which indicates a higher complexity of the synergism; this should be made the subject of further research. Similar results have been shown for increasing the number of components in minicellulosomes in a nonstatistical manner (6, 7, 10, 11).

The rather inflexible immobilization of the enzyme complexes on the rigid nanoparticle surface was expected to reduce the opportunities of the enzymes to find suitable sites for activity on the similarly inflexible substrate surface. Surprisingly, the immobilization of miniscaffoldins on the surface of nanoparticles did not reduce activity on crystalline cellulose (Tables 1 and 3). There is, however, a possibility that the loss of active sites, especially on the side of the nanoparticle not facing the substrate, is outweighed by an increase in synergism through the close proximity of complexes on the particle surface. Although the constructed nanoparticles do not have the advantage of increasing the activity on crystalline cellulose, some advantages of a cellulolytic process were observed, such as increased stability and the opportunity for magnetic separation and reutilization. In addition, the reloading efficiency of the particles could be increased by improving binding chemistry and methods. Regarding the high share enzymes have in the total production cost of industrial biotechnology (24, 41), the reduction of cost brought about by recycling the enzymes could potentially compensate the additional costs of the nanoparticles and the binding chemistry.

These experiments show that the reconstitution of full-sized cellulosomes is feasible although the in vitro conditions are surely different from the in vivo situation especially regarding to the overall protein and salt composition and concentration, and the diffusibility of the single reaction components on the cell surface. However, the achievable activity on crystalline cellulose is difficult to predict, partly because it is not clear whether the added cellulosomal components, either isolated from SCP_SM1 or recombinantly produced, are as active as they are in the native cellulosome, or because the influence of the composition of cellulosomal proteins depends on the growth substrate.

Based on these results, we investigated whether partial supplementation of the native enzyme mix with three of the most prevalent cellulases (the recombinantly produced cellobiohydrolase Cel48S, the cellotetraohydrolase Cel9R, and the processive endoglucanase Cel9J) could increase the high level of activity. The addition of 30 mol% purified recombinant cellulases increased the calculated content of cellulases (within all enzymatic components) by about 12%, based on the number of cellulases which make up 60% of the enzyme components in the cellulosomes of cellobiose-grown cells (42). With this approach, an activity increase of about the same value (15%) was observed on microcrystalline cellulose, indicating that the enzymes matched the expected activity. Extrapolating these results indicates that under our assay conditions, activities of 310 and 690 mU, respectively, for the cellulose preparations Avicel and MN300 could be expected with 100% recombinant components in the complex, provided that a suitable mixture of cellulases could be identified, and that these components could be produced in active and complete form. With such an in vitro system, the composition of the enzymes can easily be adapted for optimal hydrolysis of natural cellulosic substrates, such as plant cell wall of defined origin.

This surprising result is promising for the production of a new, complexed type of cellulase preparation for the degradation of biomass-derived cellulose on the basis of the cellulosome concept. Once properly developed, it could outcompete the fungal cellulase preparations currently being produced.