Involvement of Both Dockerin Subdomains in Assembly of the Clostridium thermocellum Cellulosome

Abstract

Clostridium thermocellum produces an extracellular cellulase complex termed the cellulosome. It consists of a scaffolding protein, CipA, containing nine cohesin domains and a cellulose-binding domain, and at least 14 different enzymatic subunits, each containing a conserved duplicated sequence, or dockerin domain. The cohesin-dockerin interaction is responsible for the assembly of the catalytic subunits into the cellulosome structure. Each duplicated sequence of the dockerin domain contains a region bearing homology to the EF-hand calcium-binding motif. Two subdomains, each containing a putative calcium-binding motif, were constructed from the dockerin domain of CelS, a major cellulosomal catalytic subunit. These subdomains, called DS1 and DS2, were cloned by PCR and expressed in Escherichia coli. The binding of DS1 and DS2 to R3, the third cohesin domain of CipA, was analyzed by nondenaturing gel electrophoresis. A stable complex was formed only when R3 was combined with both DS1 and DS2, indicating that the two halves of the dockerin domain interact with each other and such interaction is required for effective binding of the dockerin domain to the cohesin domain.

Clostridium thermocellum is an anaerobic, thermophilic, and cellulolytic bacterium. Ever since the discovery that the C. thermocellum cellulase system exists as a multiprotein complex called the cellulosome (11), it has been recognized that its activity against crystalline cellulose is governed by its unique quaternary structure. This extracellular complex consists of nearly 20 subunits, which may include endoglucanases, xylanases, lichenases, and at least one exoglucanase (2). The core of the cellulosome is a 250-kDa noncatalytic polypeptide, CipA, which binds to cellulose and serves as a scaffold for the catalytic subunits. CipA contains a series of nine highly homologous domains, termed cohesin domains, which serve as receptors for the catalytic subunits (1, 7, 17). Binding to the cohesin domains is mediated by a highly conserved duplicated sequence of 22 amino acid residues called the dockerin domain (23). The dockerin domain is found almost exclusively at the C terminus of each cellulosomal catalytic subunit.

Because of its critical role in cellulosome assembly, it is important to understand the nature of the cohesin-dockerin interaction. It has been demonstrated that dockerin binding appears to be nonselective among the various cohesin domains of a given species (13, 26). However, it was recently shown that the interaction can be species specific (14).

Another important feature of the cohesin-dockerin interaction is that it is calcium dependent (26). Chauvaux et al. (5) first noted that a conserved region of the dockerin domain has some homology to the EF-hand calcium-binding site, consisting of a helix-loop-helix motif. Secondary structure predictions indicate that each duplicated sequence contains the equivalent of a calcium-binding loop and an F helix; a corresponding E helix is not found by the prediction (14).

The three-dimensional structure of the cohesin domain has been determined (20, 22). The domain forms a nine-stranded β sandwich with an overall “jelly roll” topology. In contrast, the structure of the dockerin domain has yet to be solved. Consequently, the details of the cohesin-dockerin interaction remain unknown. In particular, the roles of the individual halves of the dockerin domain in binding to the cohesin domain are unclear. It has been suggested that both halves are involved in the cohesin-dockerin interaction (12, 14), but there has been no direct experimental evidence to support this hypothesis.

We have previously cloned and expressed the celS gene (25), which encodes an exoglucanase (10) and the most abundant catalytic subunit of the cellulosome. In this work we cloned and expressed the individual duplicated sequences from the CelS dockerin domain by using an Escherichia coli expression system. The ability of these dockerin subdomains to form a complex with a cohesin domain was examined by using a polyacrylamide gel shift assay. Our results show that both subdomains are required for the cohesin-dockerin interaction.

MATERIALS AND METHODS

Bacterial strains and vectors.

The plasmids used in this work are listed in Table 1. E. coli JM109 (Promega) served as a cloning host for the vector pCYB2 (New England Biolabs). E. coli JM109(DE3) (Promega) served as the host for pR3 and all derivatives of pT7-INTCHI. The vector pT7-INTCHI was constructed from pCYB2 by inserting the NdeI-PstI fragment of pCYB2 into the same sites in pRSETB, resulting in the replacement of the P_tac promoter of pCYB2 with the T7 promoter.

TABLE 1.

Plasmids used in this study

Plasmid	Description	Source or reference
pRSETB	T7 expression vector with an N-terminal His6 fusion tag	Invitrogen
pCYB2	Ptac expression vector with a C-terminal intein-chitin-binding domain	New England Biolabs
pT7-INTCHI	T7 expression vector with a C-terminal intein-chitin-binding domain	This work
pBL123	pT7-INTCHI derivative encoding DS2, the C-terminal half of DSa	This work
pBL126	pT7-INTCHI derivative encoding DS	This work
pBL127	pT7-INTCHI derivative encoding DS1, the N-terminal half of DS	This work
pR3	pRSETB derivative encoding the third cohesin domain (R3) of CipA	9

^aDS, the dockerin domain of CelS (24). 

DNA manipulations.

DNA was manipulated by standard procedures (18), and DNA transformation was performed by electroporation. PCR was performed as described previously (15) by using Pfu DNA polymerase (Stratagene). A plasmid containing the DNA fragment encoding the CelS dockerin domain (DSCelS), subcloned from the celS gene (25), was used as a template. (DSCelS will be referred to hereafter as DS for brevity.)

Cloning the CelS dockerin constructs.

DNA fragments encoding intact DS and its subdomains were synthesized by PCR and cloned into the expression vector described above, pT7-INTCHI. An NdeI site was created upstream of each sequence, and the amplified fragments were inserted between the NdeI and SmaI sites of pT7-INTCHI. The resulting constructs contain the insert sequence fused to the 5′ end of the gene encoding intein, a protein splicing element, followed by the gene for a chitin-binding domain. The clone producing intact DS, pBL126, contains a 207-bp DNA fragment encoding amino acids 673 (Ser) to 741 (Asn) of CelS (numbering from reference 24) (Fig. 1). The N-terminal half of DS (DS1) is produced by plasmid pBL127, which contains a 96-bp DNA fragment encoding amino acids 673 (Ser) to 700 (Arg). The C-terminal half of DS (DS2) is produced by plasmid pBL123, which contains a 135-bp DNA fragment encoding amino acids 701 (Ser) to 741 (Asn).

FIG. 1.

Amino acid sequences of DS and its two subdomains (DS1 and DS2) used in this work. The numbering of the amino acids refers to the CelS sequence (24). The putative EF-hand calcium-binding loops are boxed, and the calcium-binding residues are numbered.

Expression of the recombinant proteins.

The fusion protein His₆-R3 was produced in E. coli JM109(DE3), harboring pR3 (Table 1), grown at 37°C in SOB medium (18) supplemented with ampicillin (100 μg/ml). The culture was grown to an A₆₀₀ of 0.6, and 1 mM isopropyl β-d-thiogalactoside (IPTG) was added. The cells were grown for an additional 4 h at 37°C and then harvested by centrifugation.

The fusion proteins DS-INTCHI, DS1-INTCHI, and DS2-INTCHI were produced in E. coli JM109(DE3) grown in M9 minimal medium supplemented with glucose (5 g/liter), thiamine (5 mM), and ampicillin (100 μg/ml). The cultures were grown at 37°C to an A₆₀₀ of 0.5 to 0.7. The temperature was then lowered to 30°C, 0.5 mM IPTG was added, and the cells were grown for an additional 4 to 6 h.

Protein purification.

To purify His₆-R3, cell extracts were produced by resuspending the cell pellet (10 g [wet weight] of cells) in 60 ml of buffer A (50 mM sodium phosphate [pH 8.0], 300 mM NaCl) containing 10 mM imidazole and sonicating until the suspension was no longer viscous. Cellular debris was removed by centrifugation at 15,000 × g for 15 min. Then a heat treatment step was carried out to take advantage of the native thermostability of the R3 protein. In this step, the clarified extract was heated for 20 min at 70°C, and the precipitated proteins were removed by centrifugation. The supernatant was combined with 3 ml of 50% slurry containing Ni²⁺-nitrilotriacetic acid agarose (Qiagen) and incubated for 1 h at room temperature with gentle rocking. The resin was then washed two to three times with buffer A containing 10 mM imidazole before packing it into an Econo-Pac column (1.5 by 12 cm; Bio-Rad). Elution was performed with 100 mM imidazole in buffer A. The protein was concentrated with a Centriprep centrifugal filtration device (10,000 molecular weight cutoff [MWCO]; Amicon) or by ultrafiltration with a YM membrane (3,000 MWCO; Amicon). The His₆ tag was cleaved with enterokinase (Invitrogen) according to the manufacturer’s instructions. After changing the buffer to 20 mM Tris-HCl (pH 7.5) by gel filtration with Hi-Trap Desalting columns (Amersham Pharmacia Biotech), the protein was applied to Source 15Q anion exchange resin (Amersham Pharmacia) packed in a Bio-Scale MT2 column (Bio-Rad). The protein was eluted with a linear NaCl gradient (0 to 0.4 M NaCl) in the same buffer. The gel filtration and ion exchange steps were performed with a ConSep LC100 medium-pressure chromatography system (Millipore).

Cell extracts of 1-liter cultures of E. coli JM109(DE3) containing DS-INTCHI (pBL126), DS1-INTCHI (pBL127), or DS2-INTCHI (pBL123) were prepared in the same manner as for R3 except that buffer B (20 mM Na-HEPES [pH 8.0], 500 mM NaCl, 0.1 mM EDTA, 0.1% Triton X-100) was used. The extract was loaded into an Econo-Pac column after mixing with approximately 2.5 ml of chitin beads (New England Biolabs) equilibrated with buffer B. The column was washed with approximately 10 volumes of buffer B and then 2 volumes of buffer C (20 mM Na-HEPES [pH 8.6], 50 mM NaCl, 0.1 mM EDTA). On-column cleavage was carried out by flushing the column with about three volumes of buffer C containing 100 mM dithiothreitol (DTT) and then incubating the column overnight at 4°C. The cleaved protein was eluted with buffer B. It was concentrated by ultrafiltration using a YM 3,000 MWCO membrane (DS) or a PLAC 1,000 MWCO membrane (DS1 and DS2) (Millipore) and applied onto 27 ml of Superdex 30 prep grade gel filtration medium (Amersham Pharmacia) in a column (1 by 50 cm) for final purification. The column was eluted with 50 mM Tris-HCl (pH 7.0) at a flow rate of 0.3 to 0.5 ml/min, and the protein-containing fractions were concentrated by ultrafiltration.

Protein assay.

Protein concentrations were measured by the Bradford method (4) with a Bio-Rad protein assay kit with bovine serum albumin as a standard.

Protein electrophoresis.

Protein purity was assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) with 16% Tris–tricine gels (NOVEX) (19). Nondenaturing electrophoresis was performed as described previously (9) with 10 to 20% Tris–glycine gels (NOVEX). Proteins were stained with Coomassie brilliant blue R-250.

Analysis of complex formation.

Complex formation between R3 and the various DS constructs was carried out by combining the proteins in 50 mM Tris-HCl (pH 7.0) containing 5 mM DTT. In some cases, 10 mM CaCl₂ was added. The samples were incubated at room temperature for 15 to 30 min and then analyzed by nondenaturing electrophoresis as described above.

RESULTS

Cloning, expression, and purification of DS and its two subdomains.

The amino acid sequence of DS, separated into subdomains DS1 and DS2, is shown in Fig. 1. The two putative calcium-binding loops of DS, each containing 12 amino acid residues, are also shown in Fig. 1. The subdomains were cloned by PCR with primers designed to split the gene fragment encoding DS evenly in half between each putative calcium-binding loop. Thus, DS1 (amino acid residues 673 to 700) has 10 amino acid residues after residue 12 (Asp [D]) of the first loop, and DS2 (amino acid residues 701 to 741) has 10 amino acid residues before the first residue (Asp) of the second loop (Fig. 1). DS1 is smaller than DS2 by 13 amino acid residues. All three recombinant proteins contain a methionine at the N terminus for translation initiation. In addition, since the proteins were fused to the intein-chitin-binding domain at their C termini, the final residue of the cleaved fusion proteins is glycine, from the SmaI cloning site.

Expression of DS (pBL126), DS1 (pBL127), and DS2 (pBL123) as chimeric proteins containing a protein splicing element, called intein, fused to a chitin-binding domain greatly facilitated the purification of the target protein. A prior DS construct (TRX-DSCelS [13]) required enterokinase cleavage to remove the fusion protein, whereas the intein fusion protein can be cleaved on the chitin column with DTT. The expression of DS, DS1, and DS2 was optimized in order to minimize inclusion body formation. Variables investigated included induction temperature, IPTG concentration, and length of induction. The most critical factor was temperature (data not shown). Soluble protein production was optimal when the temperature was switched from 37 to 30°C upon induction.

The purity of all recombinant proteins as analyzed by SDS-PAGE is shown in Fig. 2. As expected, DS1 is slightly smaller than DS2 (Fig. 1).

FIG. 2.

SDS-polyacrylamide gel (16% Tris–tricine) showing the purity of the recombinant proteins. Lanes: M, molecular mass markers; 1, R3 (1.1 μg); 2, DS (1.5 μg); 3, DS1 (0.5 μg); 4, DS2 (0.4 μg). The calculated molecular weights are 16,038, 7,849, 3,287, and 4,768, respectively. (Adobe Photoshop 4.0 was used to produce the graphic.)

Formation and dissociation of the R3-DS complex.

The ability of the intact DS to function as a docking domain was verified by nondenaturing gradient gel electrophoresis. When R3 was mixed with DS, a new band corresponding to the R3-DS complex could be clearly seen on the gel (Fig. 3, lane 1). This complex was dissociated upon incubation with various concentrations of EGTA (5 to 25 mM) for 1 h at 60°C. Dissociation is observed on the gel as a decrease in the intensity of the complex band with a concomitant increase in the intensities of the DS and R3 bands (Fig. 3, lanes 2 to 6). It can be seen that free DS has a lower mobility than that of the complex, suggesting that bound DS is more compact than free DS. Nearly complete dissociation occurred when the complex was incubated with 25 mM EGTA (Fig. 3, lane 6). This result indicates that the interaction between R3 and DS is calcium dependent.

FIG. 3.

Nondenaturing polyacrylamide gel (10 to 20%) showing the DS-R3 complex and its dissociation into free DS and free R3. DS and R3 were combined without the addition of exogenous calcium to form a complex. The complex was then incubated before electrophoresis for 1 h at 60°C with various concentrations of EGTA as indicated. (Adobe Photoshop 4.0 was used to produce the graphic.)

Complex formation between R3 and the DS subdomains.

Complex formation between R3 and the two DS subdomains, DS1 and DS2, was analyzed in a similar manner by nondenaturing gel electrophoresis. R3 is clearly visible when loaded to the gel alone (Fig. 4, lane 1). Without R3, no binding was observed between DS1 and DS2 (Fig. 4, lane 2). Due to its low molecular weight and basic nature, unbound DS1 is not retained well in the Coomassie blue-stained gel and is therefore not visible at the low concentrations used here (21). In contrast to the complex observed upon mixing R3 and DS, when DS1 and DS2 were individually mixed with R3, no new band corresponding to a complex was detected on the gel (Fig. 4, lanes 3 and 4, respectively). However, when R3 was combined with both DS1 and DS2, a new band corresponding to the complex is clearly seen (Fig. 4, lane 5), indicating that both DS1 and DS2 are required for complex formation with R3. This complex has a mobility somewhat higher than that of the complex of R3 and intact DS (Fig. 4, lane 6), suggesting that it has a slightly different conformation. Free R3 has completely disappeared, while an excess of unbound DS2 can be observed (Fig. 4, lane 5).

FIG. 4.

Complex formation between R3 and DS, DS1, or DS2 as analyzed by nondenaturing PAGE on a 10 to 20% polyacrylamide gel. Proteins, both individually and combined, were incubated for 15 min at room temperature with 10 mM CaCl₂ before electrophoresis. The amounts of the proteins used were as follows: 0.56 μg of R3, 0.30 μg of DS1, 1.5 μg of DS2, and 1.5 μg of DS. (Adobe Photoshop 4.0 was used to produce the graphic.)

Titration with the DS subdomains.

To further determine if both DS1 and DS2 are required for complex formation, a mixture of R3 and DS2 was titrated with increasing amounts of DS1. Unlike the experiment shown in Fig. 4, no exogenous calcium was added to the mixture of proteins; however, this did not affect complex formation. The intensity of the complex band on the nondenaturing gel increased with increasing DS1, while the unbound R3 intensity decreased (data not shown). Starting with a mixture at the equivalence point, the amount of DS2 was then decreased until the complex was no longer visible. These results indicate that complex formation requires both DS1 and DS2 in a dose-dependent manner.

Dissociation of the DS1-DS2-R3 complex.

Like the DS-R3 complex, the complex of DS1, DS2, and R3 was dissociated upon incubation with EGTA. Complete dissociation occurred when the complex was incubated with 5 mM EGTA for 1 h at 60°C (Fig. 5). In this case, DS1 can be seen on the gel, along with DS2 and R3.

FIG. 5.

Complex formation among R3, DS1, and DS2 with and without EGTA as analyzed by nondenaturing PAGE on a 10 to 20% polyacrylamide gel. Both lanes contain 5 mM DTT, 0.56 μg of R3, 0.52 μg of DS1, and 0.37 μg of DS2. No complex is observed in the lane containing sample incubated with 5 mM EGTA for 1 h at 60°C. (Adobe Photoshop 4.0 was used to produce the graphic.)

DISCUSSION

Cloning and expression of DS and its subdomains (DS1 and DS2) allowed us to study the interactions of these domains with the cohesin domain, R3, and with each other under nondenaturing conditions. Through PCR, DS was dissected evenly between the two putative calcium-binding loops to produce as nearly symmetrical halves as possible, as well as to include the putative EF-hand helices flanking the calcium-binding loops. DS1 and DS2, as well as DS, were produced from their native sequences except for the additional N-terminal methionine and C-terminal glycine contained by each recombinant protein. Most previous studies have used either fusion proteins (13, 16) or intact cellulases containing both the catalytic domain and the dockerin domain (8, 14, 22, 26). In one study, which used the CelD dockerin domain fused to a His₆ tag, it was reported that unbound CelD dockerin is insoluble without the addition of detergents or chaotropic agents and does not enter nondenaturing polyacrylamide gels (3). In our hands, the isolated CelS dockerin can be produced as a soluble recombinant protein. Furthermore, the nondenaturing gel shift assay verified that DS is able to bind to R3 in the presence of calcium. Thus, it can be concluded that isolated DS, free from an attached protein or catalytic domain, can function as a docking domain.

It is noteworthy that the DS-R3 complex migrated with a higher mobility than that of DS on the nondenaturing gradient gel (Fig. 3). Since the mobility of proteins on a nondenaturing gradient gel is dependent on their size as well as shape, this suggests that the binding of DS to R3 causes a conformational change in DS that results in a more compact shape. This difference in mobility was also observed between CelD and the CelD-cohesin complex (22). Our results provide evidence that the conformational change occurs within the dockerin domain. As will be discussed below, this conformational change may require interactions between the two DS subdomains.

The division of DS into subdomains DS1 and DS2 enabled us to show that, under the experimental conditions, interactions between DS1 and DS2 are required for the formation of a stable complex between DS and R3, as analyzed by the nondenaturing gel shift assay. This result is consistent with the evidence suggesting that the cohesin-dockerin interaction requires the cooperative binding of two calcium ions, one in each dockerin subdomain (12). In contrast, Choi and Ljungdahl (6), on the basis of a study utilizing synthetic peptides, reported that only the first duplicated sequence is able to bind to the cohesin domain. It should be noted that the sequences of the subdomains used in their work differ from those used in this work. In particular, their second subdomain sequence begins at the first amino acid residue of the putative calcium-binding loop, whereas ours begins 10 amino acid residues before the loop. Pagès et al. (14) have suggested that the residues flanking the calcium binding loop are necessary for proper folding of the calcium-binding subdomains. In any event, the results of our gel shift assay highlight the importance of the DS1-DS2 interaction in the docking process.

The complex which was formed upon binding of both DS1 and DS2 to R3 had a slightly higher mobility than that of the complex between intact DS and R3. This difference in mobility suggests that, upon binding to R3, the separated subdomains take on a slightly different conformation than that of intact DS. This may be due to the lack of a covalent bond between the two subdomains. Alternatively, the C-terminal glycine of DS1 and the N-terminal methionine of DS2 introduced in the cloning procedure may cause a slight conformational change. Despite this potential conformational difference, the overall structural integrity of the subdomains must be retained since they are still able to bind to R3.

Based on the evidence presented here, we conclude that high-affinity binding of dockerin to cohesin requires interactions between the two dockerin subdomains, whether they are separated or covalently linked. The structural determinants for the subdomain interaction necessary for cohesin binding will need to be determined through nuclear magnetic resonance or X-ray crystallography. In addition, elucidation of the three-dimensional structures of the dockerin domain and the cohesin-dockerin complex will reveal the detailed mechanism of the docking process responsible for the molecular assembly of the cellulosome.