Synthesis and Characterization of Chimeric Proteins Based on Cellulase and Xylanase from an Insect Gut Bacterium^,

Abstract

Insects living on wood and plants harbor a large variety of bacterial flora in their guts for degrading biomass. We isolated a Paenibacillus strain, designated ICGEB2008, from the gut of a cotton bollworm on the basis of its ability to secrete a variety of plant-hydrolyzing enzymes. In this study, we cloned, expressed, and characterized two enzymes, β-1,4-endoglucanase (Endo5A) and β-1,4-endoxylanase (Xyl11D), from the ICGEB2008 strain and synthesized recombinant bifunctional enzymes based on Endo5A and Xyl11D. The gene encoding Endo5A was obtained from the genome of the ICGEB2008 strain by shotgun cloning. The gene encoding Xyl11D was obtained using primers for conserved xylanase sequences, which were identified by aligning xylanase sequences in other species of Paenibacillus. Endo5A and Xyl11D were overexpressed in Escherichia coli, and their optimal activities were characterized. Both Endo5A and Xyl11D exhibited maximum specific activity at 50°C and pH 6 to 7. To take advantage of this feature, we constructed four bifunctional chimeric models of Endo5A and Xyl11D by fusing the encoding genes either end to end or through a glycine-serine (GS) linker. We predicted three-dimensional structures of the four models using the I-TASSER server and analyzed their secondary structures using circular dichroism (CD) spectroscopy. The chimeric model Endo5A-GS-Xyl11D, in which a linker separated the two enzymes, yielded the highest C-score on the I-TASSER server, exhibited secondary structure properties closest to the native enzymes, and demonstrated 1.6-fold and 2.3-fold higher enzyme activity than Endo5A and Xyl11D, respectively. This bifunctional enzyme could be effective for hydrolyzing plant biomass owing to its broad substrate range.

INTRODUCTION

Cellulose and hemicellulose constitute ∼70 to 80% of wood and agricultural biomass and can serve as an abundant and inexpensive source of fermentable sugar for producing various chemicals and biofuels (34, 35). Three enzymes hydrolyze the β-1,4 glycosidic linkages that are present in cellulose. β-1,4-Endoglucanases cleave within the chain, cellobiohydrolases cleave at either end of the chain, and β-glucosidases break oligomers into monomers (22). Another class of enzymes, which includes xylanase and xylosidase, hydrolyzes β-1,4-xylan into xylose and is involved in the breakdown of hemicellulose (2).

Although a process to convert lignocellulosic biomass into ethanol has been developed by various research groups, cost-effective production of lignocellulosic ethanol is still a major issue, largely because of the high cost associated with the enzymes (20, 22, 34). Fungi such as Trichoderma reesei and Aspergillus niger are common sources of plant-hydrolyzing enzymes (6, 23, 32); however, these enzymes require acidic pH (30) and have a short half-life at high temperatures (12) and a low specific activity (10), which are bottlenecks to using fungal enzymes. Bacterial plant-hydrolyzing enzymes have been reported to work at a wide range of pHs and temperatures (10, 21) and, therefore, are potential catalysts for biomass hydrolysis (3, 14, 15, 16, 18). Bacteria living in the guts of organisms are of special interest owing to their ability to degrade complex substrates, such as lignocellulosic biomass, at rapid rates (8, 35).

Hydrolysis of lignocellulosic biomass requires the coordinated action of multiple enzymes. Reducing the number of polypeptides representing these enzymes in the hydrolysis reaction mixture will reduce the cost of the process. Multidomain hydrolytic enzymes in nature (9, 29) or that have been laboratory engineered (1, 7, 13, 19) are capable of catalyzing two or more reactions and, therefore, will be useful for developing an economical process for saccharification.

We identified, expressed, and characterized a potential cellulase and a hemicellulase from a bacterium isolated from the mid-gut of an insect living on plants. We constructed bifunctional chimeras based on these two hydrolytic enzymes and evaluated their structural and functional properties.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The Escherichia coli DH5α strain (Invitrogen) was used for cloning and expression work. Plasmids pUC18 (Invitrogen) and pQE30 (Qiagen) were used for cloning and expressing recombinant enzymes, respectively. Tryptic soy broth (TSB) was used to grow the microbes isolated from the insect gut, and LB medium was used to culture E. coli. Components of the culture media were purchased from either Difco Laboratories or Sigma-Aldrich. All enzymes for molecular biology work were purchased from New England BioLabs (NEB). The PCR was performed using proofreading enzyme Phusion DNA polymerase (Finnzymes). Primers for PCR and reagents for screening and assaying the hydrolytic enzymes were purchased from Sigma-Aldrich.

Identification and cultivation of a cellulolytic microorganism.

The mid-gut of Helicoverpa armigera, also known as cotton bollworm, was screened for cellulolytic microbes as described elsewhere (N. G. Priya, A. Ojha, M. K. Kajla, and R. Rajagopal, submitted for publication). Briefly, an extract of the mid-gut was serially diluted in 1× phosphate-buffered saline (PBS), plated onto tryptic soy agar (TSA), and incubated at 30°C for 72 h. Unique colonies were picked and restreaked on TSA (TSB with 1.5% agar) plates to obtain a pure culture of each isolate. To identify cellulolytic microbes, individual colonies were grown on TSA plates containing 0.5% carboxymethyl cellulose (CMC) at 30°C for 48 h, and clearance zones were observed by staining with 0.1% Congo red solution (Fig. 1A) (4). The microbe that exhibited the highest clearance zone was used for phylogenetic analysis and was identified as a species of the bacterium Paenibacillus (see Fig. S1 in the supplemental material) and is henceforth referred to as Paenibacillus strain ICGEB2008.

Fig. 1.

(A) CMCase activity of Paenibacillus ICGEB2008 on a CMC-containing agar plate after 48 h of growth. (B) Zymogram analysis of Paenibacillus ICGEB2008 culture supernatant. The arrows indicate the endocellulase and xylanase protein bands with maximum hydrolyzing activity.

Cloning of β-1,4-endoglucanase and β-1,4-endoxylanase genes.

Genomic DNA from Paenibacillus ICGEB2008 was partially digested with HindIII to obtain an ∼ 3- to 6-kb fragment, which was ligated into HindIII-digested pUC18. The ligation mixture was used to transform chemically competent E. coli DH5α, and transformants were screened on LB agar plates containing 100 μg/ml ampicillin, 0.5% CMC, and 0.1% trypan blue. Trypan blue was used instead of Congo red for screening the positive clones because it allowed screening of the transformed colonies directly on agar plates without additional processing (11). The colony that exhibited a translucent clearance zone was selected for further analysis. A plasmid was isolated from the selected colony, and the insert was sequenced from both ends using vector-specific M13-pUC forward (5′-GTT TTC CCA GTC ACG AC-3′) and M13-pUC reverse (5′-CAG GAA ACA GCT ATG AC-3′) primers. With the sequence information obtained from M13-pUC primers, the internal primers were designed to obtain the complete sequence of the insert. Endo5A forward (5′-ACT GGA TCC ATG GGC CTT ACA CTG TAT GGG-3′) and Endo5A reverse (5′-TAC AGT CGA CCT ATT CGG CGC TTG CTT TCG-3′) primers were used to amplify the β-1,4-endoglucanase gene (containing amino acids 18 to 380) without the signal sequence. The gene was cloned into the pQE30 plasmid at BamHI and SalI sites to obtain pQE-Endo5A, and a superfamily 5 β-1,4-endoglucanase (i.e., Endo5A) with a 6-histidine tag at the N terminus was expressed.

Phylogenetic analysis of the Endo5A sequence showed a high level of similarity to β-1,4-endoglucanase from Paenibacillus KCTC 8848P (GenBank accession no. AF345984) and Paenibacillus polymyxa (GenBank accession no. M33791). Therefore, xylanase genes from both species (GenBank accession no. AF195421 and DQ100299, respectively) were selected and aligned using ClustalW. Homologous regions from 22 to 40 and 619 to 639 nucleotides, with respect to GenBank accession no. AF195421, were used to design forward and reverse primers, respectively. The resultant Xyl11D-I forward (5′-CTG GAT CC TGT TAA CGG TAG TTC TTG C-3′) and Xyl11D reverse (5′-ATT CAG TCG ACT TAC CAN ACC GTT ACG TTA GA-3′) primers were used to amplify the β-1,4-endoxylanase gene from genomic DNA from Paenibacillus ICGEB2008. The amplified product was cloned into the BamHI and SalI sites of the pUC18 plasmid and sequenced from both ends using the M13-pUC primers. A new set of primers, Xyl11D-II forward (5′-AGA GCT CTT TGC AAC AAC CTC AAG TG-3′) and Xyl11D reverse (5′-ATT CAG TCG ACT TAC CAN ACC GTT ACG TTA GA-3′), was used to amplify the gene encoding β-1,4-endoxylanase without the signal sequence from the genome of Paenibacillus ICGEB2008, and this gene was cloned into the SacI and SalI sites of the pQE-30 plasmid, to obtain pQE-Xyl11D, from which a superfamily 11 β-1,4-endoxylanase (i.e., Xyl11D) with an N-terminal 6-histidine tag was expressed.

Design and construction of chimeric proteins.

Four chimeric protein models were designed based on Endo5A and Xyl11D and were used to predict three-dimensional structures using the web-based I-TASSER server, available at http://zhanglab.ccmb.med.umich.edu/I-TASSER/ (37). A benchmark scoring system that included the confidence score (C-score) and the estimated template modeling score (TM-score) was used for in silico quantitative assessments of I-TASSER models. The models tested were as follows: model 1, Endo5A at the N terminus and Xyl11D at the C terminus (Endo5A-Xyl11D); model 2, Xyl11D at the N terminus and Endo5A at the C terminus (Xyl11D-Endo5A); model 3, Endo5A at the N terminus and Xyl11D at the C terminus separated by a linker (Endo5A-GS-Xyl11D); and model 4, Xyl11D at the N terminus and Endo5A at the C terminus separated by a linker (Xyl11D-GS-Endo5A). GS represents a glycine-serine linker (GGGGSGGGGS), which was introduced in the model to keep the catalytic domains of Endo5A and Xyl11D separate (19). All four fusion proteins were synthesized in E. coli with an N-terminal 6-histidine tag, as described in the supplemental material.

Heterologous expression and purification of recombinant enzymes.

A recombinant E. coli DH5α strain transformed with plasmid constructs was used for expression and purification of recombinant Endo5A, Xyl11D, or the fusion proteins. Cells were grown in LB medium (1 liter) containing 100 mg liter⁻¹ ampicillin and induced with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) at an optical density at 600 nm (OD₆₀₀) of 0.6. The cultures were then grown for an additional 4 h and harvested by centrifugation. The cells were lysed by sonication in lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole [pH 8.0]), which was then clarified by centrifugation. The resultant supernatant was used to purify recombinant protein by using immobilized metal affinity chromatography (IMAC). The clarified cell lysate was loaded onto a Ni-nitrilotriacetic acid (NTA) agarose matrix, washed with buffer containing 20 mM imidazole, and eluted with buffer containing 250 mM imidazole. The recombinant proteins were analyzed for purity and activity as described in the “Analytical methods” section. The recombinant chimeric proteins for analysis by circular dichroism (CD) spectroscopy and specific activity comparison were further purified by gel permeation chromatography using a Superdex 200 HR10/300 column (GE Healthcare) in 50 mM sodium phosphate buffer (pH 7.0).

Purification of Endo5A and Xyl11D from a native source.

Paenibacillus ICGEB2008 was grown in TSB medium (6 liters) containing 0.5% CMC at 37°C and harvested after 48 h. Endo5A and Xyl11D were purified from the cell culture supernatant via various chromatography methods, by selecting fractions based on high CMCase activity and xylanase activity, respectively. Culture supernatants were concentrated to 1.5 liters and loaded onto Streamline phenyl resin (GE Healthcare) for purification by hydrophobic interaction chromatography (HIC) after addition of 1 M ammonium sulfate. The resin was washed with wash buffer (20 mM sodium phosphate [pH 7.5] and 1 M ammonium sulfate), and the proteins bound to the resin were eluted using a linear gradient of wash buffer and elution buffer (20 mM sodium phosphate [pH 7.5] and 10% ethylene glycol). The eluted fractions were pooled for further purification by gel permeation chromatography (GPC), using an XK16/60 column packed with Superdex 75 resin (GE Healthcare) that was equilibrated with buffer containing 20 mM sodium phosphate (pH 7.2) and 150 mM NaCl. The fractions with maximum CMCase activity were pooled, diluted 10-fold with 20 mM sodium phosphate (pH 8.0), and applied to a 5-ml Hi-Trap Q Sepharose HP column (GE Healthcare) for anion-exchange chromatography (IEX). Proteins were eluted with a linear gradient developed with buffer A (20 mM sodium phosphate [pH 8.0]) and buffer B (20 mM sodium phosphate [pH 8.0] and 500 mM NaCl) and used for biochemical analysis. The GPC fractions that exhibited high xylanase activity were diluted 10-fold with 20 mM sodium acetate buffer (pH 4.5) and loaded on SP-Sepharose FF resin (GE Healthcare) for cation-exchange chromatography. The Xyl11D was eluted using a linear gradient between buffer A (20 mM sodium acetate buffer [pH 4.5]) and buffer B (20 mM sodium acetate buffer [pH 4.5] and 250 mM NaCl) and used for biochemical analysis.

Analytical methods.

Protein purity at various steps of chromatography was analyzed by SDS-PAGE. The protein samples were separated on a 12% SDS-PAGE gel and stained with Coomassie blue to visualize the bands. To assess the CMCase or xylanase activity of the protein resolved on the gel by zymogram, samples were denatured by heating at 95°C for 5 min in sample buffer containing β-mercaptoethanol and then applied to a 12% SDS-PAGE gel containing either 0.5% (wt/vol) CMC or 0.5% (wt/vol) xylan. After electrophoresis, the gel was washed once with 20% (vol/vol) isopropanol in PBS for 1 min followed by three washes of 20 min each in PBS. The gel was incubated in PBS at 37°C for 1 h, stained with 0.1% (wt/vol) Congo red for 30 min, and destained with 1 M NaCl. CMCase or xylanase activity on the gel was visible as a clear band against a red background. Protein concentration was estimated with the bicinchinonic acid (BCA) protein assay kit (Pierce) using bovine serum albumin as a standard.

Protein identification and carbohydrate product profiles by mass spectrometry.

The protein of interest exhibiting cellulase activity was separated by SDS-PAGE, silver stained, and subjected to in-gel trypsin digestion with tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-treated bovine trypsin (Roche) at a final concentration of 0.01 mg/ml. The peptides were loaded onto a matrix-assisted laser desorption ionization (MALDI) target plate and measured using an Ultraflex III MALDI-tandem time of flight (TOF/TOF) instrument (Bruker Daltonics, Germany). For peptide mass fingerprinting-based protein identification, the tryptic peptide mass maps (monoisotopic) were searched with MASCOT Peptide Mass Fingerprint using Biotools version 3.2 (build 1.31) (Bruker Daltonics, Germany).

To determine the profile of enzyme-hydrolyzed products of CMC and xylan, the samples were processed as follows. The enzymes with their respective substrates were incubated at 37°C for 6 h. The samples containing sugars were desalted and purified using C₁₈ reverse-phase spin columns (Sigma) and then vacuum concentrated. The concentrated samples were redissolved in a methanol-1 mM sodium acetate solution (1:1 [vol/vol]), spotted onto a MALDI target plate, and assessed using the Ultraflex III MALDI-TOF/TOF instrument. The spectral maps were annotated by adding the residue mass of the standard sugars to the mass of sodium (M + Na⁺) (24). The spectral maps of the samples were compared to those of the standards to identify probable sugars obtained upon enzymatic digestion.

Circular dichroism spectroscopy.

CD spectroscopy was performed on a J-810 spectropolarimeter (Jasco, Easton, MD) using a rectangular quartz cell with a 0.1-cm path length. Spectra were acquired using an 8-s time response and a 50-nm/s scan speed; the spectra were averaged for five acquisitions. Individual proteins or protein mixtures were analyzed at 1.78 μM in 50 mM sodium phosphate buffer (pH 8.0). The percentages of alpha-helices and beta-pleated sheets were calculated according to the method of Raussens et al. (31).

Enzyme activities.

Enzyme activity in the culture supernatant of Paenibacillus ICGEB2008 was determined by measuring the amount of reducing sugar released during incubation with various substrates such as CMC, Avicel, Sigmacel101, Sigmacel50, Sigmacel20, oat spelt xylan, barley β-d-glucan, and locust bean gum. The enzyme activities of native and recombinant endoglucanase and endoxylanase were determined by measuring the amount of reducing sugars released during incubation with CMC and oat spelt xylan, respectively. The protein sample (250 μl) was added to 250 μl of 1% (wt/vol) substrate in 50 mM sodium-phosphate buffer (pH 7.0) and incubated at 37°C for 30 min. The amount of reducing sugar produced was measured by the 3,5-dinitrosalicylic acid (DNSA) method. One unit of enzyme activity was defined as the amount of enzyme that released 1 μmol of reducing sugar min⁻¹ using various substrates (5). The effect of temperature on enzyme activity was determined by incubating samples from 20 to 80°C at pH 7.0. Similarly, the effect of pH on enzyme activity was determined by incubating samples in the pH range of 3 to10 at 37°C. The buffers used to maintain the pH of the solution were citrate (pH 3 to 4), acetate (pH 4 to 6), phosphate (pH 6 to 8), and carbonate (pH 8 to 10).

Nucleotide sequence accession numbers.

Novel nucleotide sequences obtained in this study have been deposited in the NCBI gene bank. The GenBank accession number for the 16S rRNA gene of Paenibacillus ICGEB2008 is JF714415. The GenBank accession numbers for the genes encoding Endo5A, Xyl11D, and Endo5A-GS-Xyl11D are HQ657203, HQ657204, and HQ657205, respectively.

RESULTS

Identification, cloning, and heterologous expression of a novel β-1,4-endoglucanase.

Mid-gut fluid from Helicoverpa armigera, also known as cotton bollworm, was used to identify cellulolytic microorganisms. The microorganism that exhibited the most CMCase activity on the agar plate containing CMC was characterized as a Paenibacillus sp. (Paenibacillus ICGEB2008) by phylogenetic analysis of its 16S rRNA gene (Fig. 1A). The Paenibacillus strain ICGEB2008 was cultured with agitation and analyzed for its ability to secrete various hydrolytic enzymes (Table 1). Significant endoglucanse and xylanase activities were detected in the culture supernatant, which were further confirmed by zymogram analysis (Fig. 1B). The most prominent clearance zones detected by the zymogram were at ∼41 kDa and ∼20 kDa, corresponding to endoglucanase and xylanase, respectively.

Table 1.

Substrate specificity of enzymes secreted in the culture medium of Paenibacillus ICGEB2008

Substrate	Sp act (U/108 cells)a
CMC	0.074 ± 0.0081
Avicel	0.020 ± 0.0069
Sigmacel101	0.042 ± 0.0030
Sigmacel50	0.013 ± 0.0030
Sigmacel20	0.006 ± 0.0014
Oat spelt xylan	0.348 ± 0.0253
Barley β-d-glucan	0.141 ± 0.0061
Locust bean gum	0.140 ± 0.0030

^aData represent average ± SD of specific activities with respect to cell density from three independent experiments.

A shotgun cloning approach was used to identify the gene encoding endoglucanase. A positive clone of E. coli was selected for its CMCase activity on a CMC-agar plate and was analyzed for the presence of an endoglucanase gene in the plasmid insert. An insert of 3,241 bp was detected in the plasmid, and this insert contained an open reading frame of 1,194 bp (GenBank accession no. HQ657203), which encoded a polypeptide with a conserved domain belonging to cellulase superfamily 5. The encoded endoglucanase, named Endo5A, was 397 amino acids long with a theoretical molecular mass of 44.4 kDa. The SignalP 3.0 program predicted the initial 32 amino acids as the signal sequence. The BLAST search using the amino acid sequence of Endo5A indicated that the reported endoglucanase of Paenibacillus KCTC8848P (GenBank accession no. AAL83749) had the closest identity, at 97% (see Fig. S2A in the supplemental material).

We expressed Endo5A with an N-terminal histidine tag intracellularly in E. coli. The recombinant Endo5A was purified by metal affinity chromatography from the E. coli lysate (Fig. 2A) and then characterized. To ensure that the heterologous expression of Endo5A did not affect its catalytic activity, we purified native Endo5A from the culture supernatant of Paenibacillus ICGEB2008, as described in Materials and Methods, and compared its activity with that of the recombinant form (Fig. 2B and C; see Table S1 in the supplemental material). The purified native endoglucanase was in-gel digested with trypsin, and peptide mass fingerprinting-based identification was performed using MALDI-TOF/TOF to confirm that it was β-1,4-endoglucanase. Two peptide sequences of 32 (ions score, 56) and 14 (ions score, 143) amino acids each were obtained that had sequence similarity to the reported β-1,4-endoglucanase (GenBank accession no. AAL83749) in the regions of 180 to 211 amino acids and 308 to 321 amino acids, respectively. These two regions matched 100% with the predicted amino acid sequence of recombinant Endo5A, confirming that the endoglucanase purified from the native source was Endo5A.

Fig. 2.

Purification and activity profiles of recombinant and native Endo5A. (A) Purification profile of recombinant Endo5A on an SDS-PAGE gel. Recombinant Endo5A was expressed in E. coli and purified by immobilized metal affinity chromatography (IMAC). Lane 1, molecular mass marker (in kDa); lane 2, uninduced sample; lane 3, induced sample; lane 4, insoluble cell lysate fraction; lane 5, soluble cell lysate fraction; lane 6, flowthrough; lane 7, wash; lanes 8 to 10, chromatography eluates from IMAC columns. (B and C) Purification profiles of native Endo5A on an SDS-PAGE gel. Culture supernatant from Paenibacillus ICGEB2008 was used for purification by ultrafiltration (lane 1), hydrophobic interaction chromatography (lane 2), gel permeation chromatography (lane 3), and ion-exchange chromatography (lane 4) and then was separated by SDS-PAGE in a gel containing CMC. The gels were stained with either Coomassie blue (B) to visualize the proteins or Congo red (C) to visualize the CMCase clearance zone. A molecular mass marker (in kDa) was loaded in lane M. The arrow indicates the location of Endo5A. (D and E) Temperature (D) and pH (E) optima for native Endo5A, recombinant Endo5A, and the recombinant Endo5A-GS-Xyl11D fusion protein. All three enzymes were purified and tested for activity at various temperatures and pHs. The results indicate that CMCase activities for all three enzymes are similar at the different temperatures and pH conditions.

A comparison of the cellulolytic activities of both native and recombinant Endo5A was performed. The specific activities of both the endoglucanases were similar, indicating that heterologous expression of endoglucanase did not affect its conformation and activity (see Table S1 in the supplemental material). The specific activities of both Endo5A enzymes were compared at temperature and pH ranges of 20 to 80°C and 3 to 10, respectively, and were found to be similar (Fig. 2D and E). The optimal temperature and pH for the endoglucanase activities were 50°C and pH 7, respectively (Fig. 2D and E).

Identification, cloning, and heterologous expression of a novel β-1,4-endoxylanase.

We also attempted to isolate the xylanase gene from Paenibacillus ICGEB2008 because the culture supernatant exhibited significant xylanase activity (Table 1). The zymogram analysis revealed a major band with xylanase activity at ∼20 kDa (Fig. 1B). Based on the nucleotide sequence of Endo5A of Paenibacillus ICGEB2008, the two closest relatives were selected from the BLAST search, and conserved regions were used for primer design to isolate the xylanase gene from Paenibacillus ICGEB2008, as described in Materials and Methods. Primary sequence analysis of the novel xylanase, named “Xyl11D” (GenBank accession no. HQ657204), revealed the presence of a xylanase domain belonging to glycosyl hydrolase family 11. A BLAST search using the amino acid sequence of Xyl11D revealed 99% similarity to the xylanase of Paenibacillus polymyxa (GenBank accession no. AAZ17384) (see Fig. S2B in the supplemental material).

The Xyl11D protein was overexpressed intracellularly in E. coli and purified from the cell lysate by metal affinity chromatography (Fig. 3A). Native Xyl11D (20 kDa) was also purified from the culture supernatant of Paenibacillus ICGEB2008 and used for characterization (Fig. 3B and C). The specific activities of native and recombinant Xyl11D were similar (see Table S2 in the supplemental material), indicating that recombinant Xyl11D folded in a similar manner to native Xyl11D. The activity of recombinant Xyl11D at different temperatures and pHs was similar to that of native Xyl11D and exhibited optimal activity at 50°C and pH 6 to 7 (Fig. 3D and E).

Fig. 3.

Purification and activity profiles of recombinant and native Xyl11D. (A) SDS-PAGE gel profile for purification of recombinant Xyl11D. Recombinant Xyl11D was expressed in E. coli and purified by IMAC. Lane 1, clarified culture lysate; lane 2, flowthrough; lanes 3 to 5, chromatography eluates of IMAC. Molecular mass markers (in kDa) are shown. (B and C) SDS-PAGE gel profiles for purification of native Xyl11D. Native Xyl11D from culture supernatant of Paenibacillus ICGEB2008 was purified by ultrafiltration (lane 1), hydrophobic interaction chromatography (lane 2), gel permeation chromatography (lane 3), and ion-exchange chromatography (lane 4) and then separated on an SDS-PAGE gel containing xylan. Gels were stained with either Coomassie blue (B) to visualize the proteins or Congo red (C) to visualize the xylanase clearance zone. A molecular mass marker (in kDa) was loaded in lane 5. (D and E) Temperature (D) and pH (E) optima for xylanase activities of native Xyl11D, recombinant Xyl11D, and the recombinant Endo5A-GS-Xyl11D fusion protein. The results indicate that xylanase activities are similar for all three enzymes over a range of temperatures and pH conditions.

Construction of chimeric proteins containing endoglucanase and xylanase domains.

Both Endo5A and Xyl11D exhibited their optimal activities at similar temperatures and pHs. To take advantage of this feature, we constructed four bifunctional chimeric proteins composed of the Endo5A and Xyl11D enzymes of Paenibacillus ICGEB2008. Two strategies were adopted to generate the chimeric models: changing the orientation and introducing a glycine-serine linker. The glycine-serine linker (GGGGSGGGGS) used in this study has been commonly used to fuse two or more domains (19). We used a structural modeling approach with the help of the I-TASSER program to predict the three-dimensional structures of four chimeric models, as described in Materials and Methods. Of the four models, model 3 exhibited the highest C-score and TM-score (Table 2).

Table 2.

Comparison of I-TASSER scores and specific activities of chimeric models

Activity and model	I-TASSER C-score/TM-score	Sp act (U mg−1)	Molar sp act (U μmol−1)	Fold change in activity
Endoglucanase
Endo5A	NDa	14.6 ± 0.63	642 ± 28	1.00
Model 1 (Endo5A-Xyl11D)	−2.21/0.45 ± 0.15	5.1 ± 0.03	329 ± 1.8	0.51
Model 2 (Xyl11D-Endo5A)	−2.79/0.39 ± 0.13	14.3 ± 0.52	937 ± 34	1.46
Model 3 (Endo5A-GS-Xyl11D)	−2.06/0.47 ± 0.15	16.3 ± 0.02	1,070 ± 1.6	1.67
Model 4 (Xyl11D-GS-Endo5A)	−2.73/0.40 ± 0.13	7.1 ± 0.01	459 ± 0.3	0.72
Xylanase
Xyl11D	ND	17.5 ± 0.77	389 ± 17	1.00
Model 1 (Endo5A-Xyl11D)	−2.21/0.45 ± 0.15	8.0 ± 0.03	521 ± 2.2	1.34
Model 2 (Xyl11D-Endo5A)	−2.79/0.39 ± 0.13	11.5 ± 0.30	754 ± 19	1.94
Model 3 (Endo5A-GS-Xyl11D)	−2.06/0.47 ± 0.15	13.7 ± 0.46	899 ± 30	2.31
Model 4 (Xyl11D-GS-Endo5A)	−2.73/0.40 ± 0.13	8.5 ± 0.23	554 ± 15	1.42

^aND, not done.

We synthesized all four fusion proteins in E. coli (Fig. 4A) and tested their structural and functional properties. CD spectroscopy was performed to compare the secondary structures of Endo5A and Xyl11D individually or in the chimeric form (Fig. 4B and C). Deconvolution of the CD spectra indicated a predominant alpha-helix for Endo5A and beta sheet for Xyl11D (Table 3). The CD spectrum of the equimolar mixture of Endo5A and Xyl11D indicated an alpha-helix/beta-sheet (α/β) ratio of 0.84. Of the four models, model 3 exhibited the closest α/β ratio to the Endo5A and Xyl11D mixture (Table 3).

Fig. 4.

CD spectroscopy of recombinant bifunctional chimeric proteins. (A) Recombinant proteins used for CD analysis. All four fusion proteins and their individual counterparts were expressed in E. coli, purified by metal affinity and gel permeation chromatography, and then used for CD analysis. Lane M, molecular mass markers; lane 1, Endo5A-Xyl11D; lane 2, Xyl11D-Endo5A; lane 3, Endo5A-GS-Xyl11D; lane 4, Xyl11D-GS-Endo5A; lane 5, Xyl11D; and lane 6, Endo5A. (B) CD spectra of Endo5A, Xyl11D, and an equimolar mixture of Endo5A and Xyl11D. (C) CD spectra of all four chimeric models.

Table 3.

CD spectral analysis of Endo5A, Xyl11D, and bifunctional chimeras

Protein(s) or model	% alpha-helix (α)	% beta-sheet (β)	% turn	% random	α/β ratio
Endo5A	28.9	14.5	12.4	34.7	1.99
Xyl11D	8.0	46.1	11.9	43.3	0.17
Endo5A-Xyl11D mixture	16.8	19.9	12.3	36.7	0.84
Model 1 (End5A-Xyl11D)	21.8	16.8	12.1	31.1	1.29
Model 2 (Xyl11D-Endo5A)	24.0	18.8	12.3	33.0	1.27
Model 3 (Endo5A-GS-Xyl11D)	18.9	20.1	11.9	31.5	0.94
Model 4 (Xyl11D-GS-Endo5A)	22.7	17.3	12.2	32.3	1.31

A quantitative measurement of the hydrolytic activity of all of the bifunctional chimeras was made to analyze the consequences of fusion. All four chimeric models resulted in different enzymatic activities, with 0.5- to 1.6-fold and 1.3- to 2.3-fold differences in activity compared with Endo5A and Xyl11D, respectively (Table 2). Here also, model 3 had the highest endoglucanase (1,070 U μmol⁻¹) and xylanase (899 U μmol⁻¹) activities of all of the models.

Model 3 was selected for further characterization of its dual activity (see Fig. S3 in the supplemental material). The comparison of optimal CMCase activities at various temperatures and pHs indicated that model 3 and Endo5A shared maxima at 50°C and pH 7.0, respectively (Fig. 2D and E). Similarly, comparison of the xylanase activities indicated maximal specific activity at 50°C for both model 3 and Xyl11D (Fig. 3D). The fusion protein also exhibited 90 to 100% activity at pH 6 to 7, which was similar to that of Xyl11D (Fig. 3E). We also compared the product profiles of polysaccharides that had been incubated with the chimeric protein with those of polysaccharides incubated with the individual enzymes by using mass spectrometry. A spectral map of saccharide standards was generated (see Fig. S4 in the supplemental material) and used to compare the hydrolytic products of the reaction mixtures. Only cellobiose was detected in the hydrolysate when CMC was incubated with Endo5A (Fig. 5A); however, xylobiose, xylotriose, and xylopentaose were detected in the hydrolysate when xylan was incubated with Xyl11D (Fig. 5B). The model 3 fusion chimera incubated with CMC and xylan produced a mixture of sugars, with xylobiose, cellobiose, xylotriose, and xylotetraose as the end products, indicating optimal activity of both catalytic domains (Fig. 5C).

Fig. 5.

Positive-mode matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectra of sugars (M + Na⁺) present in the hydrolyzed products of CMC after incubation with Endo5A (A), hydrolyzed products of xylan after incubation with Xyl11D (B), and hydrolyzed products of a xylan-CMC mixture following incubation with the Endo5A-GS-Xyl11D chimeric protein (C).

DISCUSSION

Efficient hydrolysis of the plant cell wall is challenging for anyone interested in producing lignocellulosic ethanol in a commercially viable manner (20). The guts of insects living on lignocellulosic biomass contain diverse microbial flora for obtaining nutrients during biomass degradation (35). We identified a bacterium from the gut of Helicoverpa armigera larvae based on its ability to degrade plant biomass and analyzed the potential of this bacterium to produce hydrolytic enzymes.

The isolated bacterium was characterized as a Paenibacillus sp. and was found to express enzymes capable of hydrolyzing plant biomass (Table 1). Although many strains of Paenibacillus have been reported to exist in the gut of insects (27, 33, 36), in-depth characterization of the cellulolytic enzymes of these bacteria is lacking. Therefore, we performed a genetic and biochemical characterization of two cellulolytic enzymes of Paenibacillus ICGEB2008, β-1,4-endoglucanase (Endo5A) and β-1,4-endoxylanase (Xyl11D), and devised strategies to produce them in E. coli in an economically viable manner.

The Endo5A enzyme cloned from Paenibacillus ICGEB2008 contained a catalytic domain associated with glycosyl hydrolase family 5. The enzymes classified into this family are typically present in cellulolytic bacteria and fungi (http://www.cazy.org/GH5.html). A similar bacterial endoglucanase with 97% homology has been reported; however, limited characterization of this enzyme has been performed (28). Endoglucanase has been reported in several fungal species, and we found the specific activity of Endo5A to be similar (25, 26). Xyl11D of Paenibacillus ICGEB2008 had a relatively low molecular mass (∼20 kDa) and contained a catalytic domain belonging to glycosyl hydrolase family 11. A similar enzyme, XynA, with 95% similarity to Xyl11D has been reported but has not been functionally characterized (17). We performed systematic characterization of Endo5A and Xyl11D and determined that both enzymes exhibited maximum activity at 50°C and pH 6 to 7.

One way to reduce saccharification costs is by producing enzymes economically. Because saccharification of lignocellulosic biomass requires the coordinated action of multiple enzymes, the production cost should be reduced if a single polypeptide has more than one catalytic function. The most striking feature of the isolated Endo5A and Xyl11D enzymes was that they exhibited optimal activity at the same temperature and pH and were, therefore, potential candidates for constructing a bifunctional polypeptide; however, constructing a chimeric protein often reduces the activity of one or more of the catalytic domains (1, 13). One way to address this issue is to express chimeric proteins with various linkers to separate the two domains (19). We synthesized four chimeric protein models with and without linker and performed structural and functional characterizations. Of the four models, model 3, which was composed of Endo5A followed by Xyl11D separated by a glycine-serine linker, exhibited the most favorable score based on in silico three-dimension prediction by the I-TASSER server. This model also had secondary structure properties closest to those of the native enzymes. More importantly, the molar specific activity of model 3 was highest among all the models and was 1.6-fold higher than that of Endo5A and 2.3-fold higher than that of Xyl11D. The results indicated that both the orientation and the linker are important for optimal activity of both enzymes.

In conclusion, we identified and characterized two key enzymes used in the saccharification process of lignocellulosic biomass from an insect gut bacterium and overexpressed them in E. coli. We further synthesized bifunctional polypeptides and assessed their structure-function properties. We obtained a polypeptide that had higher enzyme activity than the individual native enzymes. The study detailed here can help minimize the cost of saccharification, making lignocellulosic biofuel production more commercially viable.