Background

Multienzyme pathways in living systems comprise cascades in which enzymes are tethered together into assemblies that facilitate substrate flow between components, limit the diffusion of intermediate metabolic products, and increase the yield from sequential reactions [1,2]. There have been various attempts to produce multienzyme assemblies in vitro, including by gene fusion, protein or DNA scaffold construction, and chemical modification [3]. Although the simplest way is by enzyme fusion, this often results in loss of enzymatic activity or formation of inclusion bodies [4], while chemical modification can also impair enzymatic activity [5]. Moreover, the high cost of generating DNA scaffolds makes this approach infeasible on a large scale [6,7].

Protein scaffolds are an attractive strategy for bringing together enzymes. A 77-fold enhancement in product concentration was observed in an assembly of three mevalonate biosynthetic enzymes with a protein scaffold composed of metazoan signaling proteins [8]. In another study, a self-assembled enzyme complex using cellulosome achieved a 13.4-fold increase in reaction rate [9], while a proliferating cell nuclear antigen-based assembly of P450 with ferredoxin and ferredoxin reductase showed high catalytic activity [10]. The protein scaffolds used in these studies were limited to metazoan signaling proteins and cellulosome components [11-14], which may not be amenable to all types of enzyme cascades. As such, there is an ongoing need for novel and different types of protein scaffolds.

Cellulose is the most abundant renewable resource on Earth and plays a significant role in biofuel production [15]. Cellulose is broken down into oligosaccharides by endoglucanase (EG) and exoglucanase (CBH) before β-glucosidase (BGL) hydrolyzes cellobiose into glucose [16]. EG, CBH, and BGL are enzymes in the cellulose degradation pathway that act synergistically; coexpressing these enzymes improves the efficiency of cellulose degradation [17]. It was hypothesized that a highly ordered assembly containing the three cellulases would enhance their activities and thereby increase cellulose hydrolysis. Therefore, CelccA (EG), CelccE (CBH), and Cel2454 (BGL) were selected as a model system for protein scaffold-mediated assembly strategy.

In this study, considering the correct expression and assembly of proteins, three pairs of interacting proteins (IPA/IPa, IPB/IPb, and IPC/IPc) from E. coli K12 [18] were selected to construct a protein scaffold for three cellulases - namely, CelccA (EG), CelccE (CBH), and Cel2454 (BGL) (Figure 1). The enzymes were arranged in different orders and the effect on activity was investigated. The multienzyme complex in best order was characterized in detail and compared with the unassembled components.

Figure 1
figure 1

Schematic representation for assembly of cellulases on a protein scaffold. Three cellulases (CBH, EG, BGL) and three pairs of interaction proteins (IPA-IPa, IPB-IPb, IPC-IPc) were used for the assembly.

Full size image

Methods

Strains and medium

Escherichia coli DH5α was used for cloning, and E. coli BL21 (DE3) was used for protein expression. Cells were cultured in Luria-Bertani (LB) medium (10.0 g/l tryptone, 5.0 g/l yeast extract, 10.0 g/l NaCl) supplemented with either 100 mg/l ampicillin or 50 mg/l kanamycin.

Plasmid construction

PCR primer sequences are listed in Table 1. All plasmid sequences were validated by sequencing. To generate the pET21a-IPA plasmid, the IPA gene (GenBank accession no. EU899769.1) was amplified from E. coli K12 genomic DNA using the IPA-F/IPA-R primer pair. The PCR product was digested with NdeI/XhoI and ligated into pET21a digested with the same restriction enzymes. Similarly, plasmids pET21a-IPa, pET21a-IPB, pET21a-IPb, pET21a-IPC, and pET21a-IPc were constructed by amplifying the DNA sequences of IPa, IPB, IPb, IPC, and IPc (GenBank accession nos. EU895384.1, EU8971IPC.1, AB9163IPA.1, EU893769.1, and EU891369.1, respectively) using the primer pairs IPa-F/IPa-R, IPB-F/IPB-R, IPb-F/IPb-R, IPC-F/IPC-R, and IPc-F/IPc-R, respectively. Plasmids pET28a-celCCA, pET28a-celCCE, and pET28a-cel2454 were constructed by amplifying the DNA sequences of celCCA, celCCE, and cel2454, respectively (GenBank accession nos. M93096.1, M87018.2, and ACL76783.1, respectively), from the genomic DNA of Clostridium cellulolyticum DSM 5812 using the primer pairs A-F/A-R, E-F/E-R, and 4-F/4-R, respectively [19,20].

Table 1 Primers used in this work
Full size table

Plasmid pET28a-IPaA had an expression cassette containing IPa and celCCA at the N and C termini, respectively. A DNA fragment encoding IPa was amplified with the primer pair IPaA-F/IPaA-R and cloned into pET28a-celCCA by Seamless Cloning [21]. Plasmid pET28a-AIPa containing celCCA and IPa at the N and C termini, respectively, was generated with the primer pair AIPa-F/AIPa-R. Plasmid pET28a-IPbE containing IPb and celCCE at the N and C termini, respectively, was generated with primer pair IPbE-F/IPbE-R. Plasmid pET28a-EIPb containing celCCE and IPb at the N and C termini, respectively, was constructed with primer pair EIPb-F/EIPb-R. Plasmid pET28a-IPc4 containing IPc and cel2454 at the N and C termini, respectively, was generated with primer pair IPc4-F/IPc4-R. Plasmid pET28a-4IPc containing cel2454 and IPc at the N and C termini, respectively, was constructed with primer pair 4IPc-F/4IPc-R.

Plasmid pET21a-ScafBAC had an expression cassette containing IPA flanked by IPB and IPC at the N and C termini, respectively. The DNA fragments encoding IPB, IPA, and IPC were amplified with the primer pairs ScafBAC-F1/IPB-R, ScafBAC-F2/IPA-R, and ScafBAC-F3/ScafBAC-R, respectively. The three fragments were cloned into pET21a by Seamless Cloning. Plasmid pET21a-ScafABC was constructed using primer pairs ScafABC-F1/IPA-R, ScafABC-F2/IPB-R, and ScafABC-F3/ScafABC-R, and plasmid pET21a-ScafBCA was generated using primer pairs ScafBAC-F1/IPB-R, ScafABC-F3/IPC-R, and ScafBCA-F/ScafBCA-R.

Protein expression and purification

Recombinant proteins were precultured overnight at 37°C in LB medium supplemented with appropriate antibiotics. The cultures were inoculated in fresh LB medium containing antibiotics and incubated at 37°C until the optical density at 600 nm reached 0.8. The cultures were then cooled to 18°C, and isopropyl-β-d-thiogalactopyranoside was added to a final concentration of 0.1 mM. After 20 h, cells were harvested by centrifugation for 10 min at 8,000 rpm and 4°C, resuspended in 20 mM phosphate-buffered saline (PBS; pH 7.0), and disrupted by sonication on ice. Cellular debris was removed by centrifugation for 40 min at 11,000 rpm. Proteins were purified using a HisTrapFF column (GE Healthcare, Waukesha, WI, USA), and protein concentration was determined by the Bradford method.

Biolayer interferometry

Binding affinities between proteins were measured by biolayer interferometry (Octet QKe; Fortebio, Menlo Park, CA, USA), which detects changes in mass (protein density) on a biosensor; changes in the reflected interference wave pattern between the sample and an internal reference layer result in a phase shift that can be followed in real time in both kinetic and quantitative modes [22]. All experiments were performed in kinetic buffer (20 mM PBS, pH 7.0; 1 mg/ml bovine serum albumin (BSA), and 0.02% Tween 20). One of the proteins (1 μM) was biotinylated by incubating with 2 μl biotinyl N-hydroxysuccinimide ester for 1 h at room temperature, with excess biotin removed using a desalting column. The biotinylated protein was loaded onto the streptavidin biosensor by incubating for 240 s. The immobilized protein was equilibrated with kinetic buffer for 180 s, and the corresponding protein (1 μM) was associated to the biotinylated protein by incubating for 800 s. Dissociation was measured for 800 s in kinetic buffer. For each assay, a control experiment was carried out using BSA. Binding affinity was independent of which protein was loaded onto the streptavidin biosensor [23-26].

Enzyme assays

The BGL activity was measured by incubating 135 μl of 2.5 mM p-nitrophenyl β-d-glucopyranoside solution in 20 mM sodium phosphate buffer (pH 7.0) with 7.5 μg of pure enzyme solution at 37°C for 30 min. The reaction was terminated by adding 70 μl of 0.4 M Na2CO3 and the absorbance at 420 nm was measured. One unit of enzyme was defined as the activity producing 1 μmol of p-nitrophenol per min under the assay conditions. EG/CBH activity was measured by incubating 90 μl of 1.5% (wt/vol) carboxymethyl cellulose (CMC) in 20 mM sodium phosphate buffer (pH 7.0) with 5 μg of pure enzyme solution at 37°C for 30 min. A 100-μl volume of sample was mixed with 150 μl of 3,5-dinitrosalicylic acid reagent, and after boiling for 10 min, the absorbance at 540 nm was measured. One unit of enzymatic activity was defined as the amount of enzyme required to produce a 1 μmol reduction sugar per min under the assay conditions.

The activity of free or assembled enzymes was assayed in the presence of 0.75% (wt/vol) CMC at 37°C in 20 mM sodium phosphate buffer (pH 7.0). The reduction sugars were measured as described above. Glucose concentration was determined using an SBA biosensor analyzer (Biology Institute of Shandong Academy of Sciences, Jinan, China).

Results and discussion

Selection and characterization of interacting proteins

To synthesize the protein scaffold, three pairs of interacting proteins from E. coli K12 were expressed in E. coli BL21 (DE3). The six proteins were purified by Ni-chelating affinity chromatography (Figure 2A), and protein-protein affinity was measured by biolayer interferometry [23-26], in which the obtained values reflect the amount of protein bound to the biosensor. The response values (Figure 2B,C,D) confirmed their binding, and the KD calculated by Fortebio Data Analysis software version 7.0 for IPA-IPa (7.36 × 10−8 M), IPB-IPb (1.39 × 10−12 M), and IPC-IPc (5.27 × 10−8 M) demonstrated their high affinity, suggesting that they are suitable for constructing protein scaffolds.

Figure 2
figure 2

Purification of interaction proteins and their binding affinity. (A) SDS-PAGE analysis of purified interaction proteins. Biolayer interferometry sensorgrams of (B) IPA-IPa, (C) IPB-IPb, (D) IPC-IPc. IPA, IPB, and IPC were immobilized to the streptavidin-coated biosensors, respectively. Then IPa, IPb, and IPc were bound to the corresponding proteins, respectively. The figure shows the response of association and dissociation. All the response value was normalized by minusing the value of control.

Full size image

Fusion of cellulases and interacting proteins

The three cellulases were fused with the interacting proteins IPa, IPb, and IPc, which endowed them with an interaction interface. To minimize the effect of fusion on enzymatic activity, six fusion proteins were generated by attaching the interacting proteins to the C or N terminus of cellulases. The commonly used glycine-serine linker (G4S)2 was used to link two proteins and thereby reduce folding interference [27]. The resultant constructs were expressed in E. coli with a 6-histidine tag at the N terminus and purified by Ni-chelating affinity chromatography. IPa could only be attached to the N terminus of celCCA (IPaA) (Figure 3A), given that partial proteolysis occurred at the C terminus [19]. Similarly, the ideal celCCE and IPb fusion was obtained by attaching IPb to the N terminus of celCCE (IPbE). For the cel2454-IPc fusion, the activity of IPc4 was slightly higher than that of 4IPc (Figure 3D), and the former was therefore chosen for the assembly. The three enzyme fusions that were ultimately selected retained almost all of the specific activity of the respective unfused proteins (Figure 3B,C,D), indicating that protein structure was not appreciably compromised by the fusion.

Figure 3
figure 3

Characterization of cellulases fused with interaction proteins. (A) SDS-PAGE analysis of purified cellulases. The activities of (B) endoglucanase, (C) exoglucanase, (D) β-glucosidase were assayed. The arrows indicate the expected fusion proteins.

Full size image

Construction of scaffolds and their effect on assembled enzyme complexes

To determine the appropriate order of cellulases within assemblies, three protein scaffolds were constructed by fusing the three interacting proteins (IPA, IPB, and IPC) in different orders and assembling three tri-enzyme complexes by mixing the scaffolds and cellulases in a 1:1:1:1 stoichiometry for 1 h. The reduction sugars from CMC hydrolysis catalyzed by the three tri-enzyme complexes were assayed. The complex assembled with ScafBAC had the highest sugar reduction (Figure 4), indicating that the synergistic effect was related to the location of the enzyme within the complex. Given that the activity of EG is much higher than that of CBH or BGL, oligosaccharides and cellobioses produced by EG can be immediately catalyzed by CBH and BGL when EG is between CBH and BGL.

Figure 4
figure 4

Reducing sugar production by free cellulases or three multienzyme complexes assembling with different scaffolds. 1.5 g/l of CMC was catalyzed by 0.6 μM of enzyme for 30 min. ‘Free’ means three fusion cellulases without scaffolds.

Full size image

Characterization of the tri-enzyme complex assembled by ScafBAC

The fused cellulases and protein scaffold were purified using a Ni-sepharose column, and the correct formation of the protein assembly as well as the affinity of each cellulase fused to the protein scaffold was verified by biolayer interferometry [28] (Figure 5). The response of biolayer interferometry is dependent on the amount of proteins bound to the biosensor which indicates the binding of proteins. Although different binding responses were observed for the three cellulase fusions, they were all attached to the scaffold. Moreover, the binding responses were similar to values obtained for the interacting proteins, indicating that the binding affinity was not affected by the fusion.

Figure 5
figure 5

Biolayer interferometry sensorgrams of cellulases binding to biotinylated scaffold on streptavidin-coated biosensors. Biotinylated scaffold was immobilized to the streptavidin-coated biosensors. Then IPc4, IPbE, and IPaA were bound to the scaffold. The figure shows the response of association with IPc4, IPbE, and IPaA. All the response value was normalized by minusing the value of control.

Full size image

The potential for synergistic effects on enzyme cascades is the most interesting property of protein assemblies. The capacity for CMC hydrolysis was compared between the tri-enzyme complex and free cellulases. Catalytic activity was evaluated by glucose production using a SBA biosensor analyzer. After a 4-h reaction, a 1.5-fold enhancement in activity was observed for the tri-enzyme complex as compared to the corresponding free enzymes (Figure 6), suggesting the channeling of intermediates within the complex; that is, cellobiose generated by EG and CBH was immediately transferred to the adjacent BGL for glucose production, thereby resulting in enhanced cellulose hydrolysis.

Figure 6
figure 6

Glucose production of catalyzed by multienzyme complex or by equal amount of free cellulases. 7.5 g/l of CMC was catalyzed by 0.6 μM of enzyme. ‘Free’ means three fusion cellulases without scaffolds.

Full size image

Conclusions

An artificial tri-enzyme complex was constructed by assembling three cellulases with a novel protein scaffold composed of interacting proteins. The effect of the order of cellulase within the scaffolds on the catalytic efficiency was determined. Moreover, the complex had higher catalytic activity than the individual components. These results suggest that this novel protein scaffold can serve as a powerful tool for facilitating multienzyme cascades.