Multimeric Hemicellulases Facilitate Biomass Conversion

Two highly active trifunctional hemicellulases were constructedby linking the catalytic portion of a xylanase with an arabinofuranosidaseand a xylosidase, using either flexible peptide linkers or linkerscontaining a cellulose-binding domain. The multifunctional enzymesretain the parental enzyme properties and exhibit synergisticeffects in hydrolysis of natural xylans and corn stover.

INTRODUCTION

Top

INTRODUCTION

Bio-depolymerization of lignocellulosic biomass requires a largenumber of enzymes, many of which work synergistically to degradecomplex polysaccharides. For example, the hydrolysis of cornstover cellulose by cellobiohydrolase I is significantly enhancedby the presence of small quantities of hemicellulases (15).Important hemicellulases for biomass hydrolysis include xylanases,xylosidases, and arabinofuranosidases (14). These enzymes commonlywork in concert, and their synergistic effects in xylan degradationhave been studied (4, 7, 17). We recently constructed two chimericxylan-degrading enzymes, demonstrating that engineering bifunctionalhemicellulases is a feasible strategy for reducing the numberof proteins required for biomass conversion (5). In this study,we focused on producing a trifunctional hemicellulase and exploredthe feasibility of using a cellulose binding domain (CBD) asboth a spacer and a functional module.

Recombinant multimeric hemicellulases are produced in Escherichia coli.

Top

Recombinant multimeric...

Overlapping PCR (4) was used to construct two in-frame fusiongenes: (i) the gene encoding arabinofuranosidase-xylanase-xylosidase(the AXX gene), consisting of an ${alpha}$ -arabinofuranosidase gene (deAFc;GenBank accession no. DQ284779 [18]), the Clostridium thermocellumxylanase gene (xynZ; GenBank accession no. M22624), and theThermoanaerobacterium sp. strain JW/SL YS485 xylosidase gene(GenBank accession no. AF001926); and (ii) a gene encoding aCBD-containing AXX protein (the CBD-AXX gene), in which the471-bp C. cellulovorans CBD sequence (cbpA; GenBank accessionno. M73817) was inserted into the two peptide linkers in AXX(Fig. 1). The soluble parental and multimeric enzymes were producedin E. coli and purified by nickel affinity chromatography tohigh homogeneity as determined by 10% sodium dodecyl sulfate-polyacrylamidegel electrophoresis (data not shown). The recombinant productionlevels of AXX ( $~$ 165 kDa) and CBD-AXX ( $~$ 200 kDa) were estimatedto be approximately 15 and 10 mg/liter of cell culture, respectively.

View larger version (13K):
[in this window]
[in a new window]

FIG. 1. Schematic representation of the organization of AXX and CBD-AXX. The amino acid sequences of the two peptide linkers are shown. PCR primers used in the construction are listed in Table S1 in the supplemental material. The residues corresponding to a BamHI site (ADP) and a HindIII site (KL) where the CBDs are inserted are underlined. Xyln, xylanase; Xylo, xylosidase.

The multimeric enzymes have pH profiles and temperature profiles similar to those of the parental enzymes.

Top

The multimeric enzymes have...

When assayed using Remazol brilliant blue-xylan as a substrate,AXX, CBD-AXX, and a mixture of the three parental enzymes generatedsimilar pH profiles, although CBD-AXX showed an apparent shifttoward pH 6 (Fig. 2A). A shift in pH optimum was observed previouslyfor the xylanase-arabinofuranosidase and xylanase-xylosidasechimeras (5). We attribute this shift in pH optimum to eitherthe change of pH in the microenvironment generated by the polymericproteins or altered tertiary structure. Similar pH profilesfor arabinofuranosidase activity were also observed for themultimeric enzymes and DeAFc in an assay using 4-nitrophenyl- ${alpha}$ -L-arabinofuranoside(4NPA) as a substrate (Fig. 2B). When assayed using 2-nitrophenyl-xylopyranoside (2NPX) as a substrate, the xylosidase activityprofiles were essentially identical for AXX, CBD-AXX, and xylosidase(Fig. 2C). In all three assays, CBD-AXX is consistently moreactive than AXX. One explanation is that the positions of theCBDs between the enzyme subunits provide optimal spacing forthe three catalytic subunits and thereby improve the overallactivities of the enzymes.

View larger version (28K):
[in this window]
[in a new window]

FIG. 2. pH and temperature profiles of the multimeric and parental enzymes. Enzyme activities were determined at different pHs at 45°C. (A) Xylanase (Xyln) activities on substrate Remazol brilliant blue-xylan (1 enzyme unit is defined as 1 µmol dye released/nmol enzyme/min). (B) Arabinofuranosidase (Ara) activities on substrate 4-nitrophenyl- ${alpha}$ -L-arabinofuranoside (4NPA; 1 enzyme unit is defined as 1 nmol 4NP released/nmol enzyme/min). (C) Xylosidase (Xylo) activities on substrate 2-nitrophenyl-β-D-xylopyranoside (2NPX; 1 enzyme unit is defined as 1 nmol 2NP released/nmol enzyme/min). Enzyme activities were measured at different temperatures at pH 6.0. (D) Xylanase activity. (E) Arabinofuranosidase activity. (F) Xylosidase activity. Triplicate assays were performed using purified xylanase ( ${diamondsuit}$ ), DeAFc ( ${blacksquare}$ ), xylosidase (•), enzyme mixture ( ${square}$ ), AXX ( ${blacktriangleup}$ ), and CBD-AXX ( ${triangleup}$ ).

AXX and CDB-AXX have temperature profiles for xylanase, arabinofuranosidase,and xylosidase similar to those for their respective parentalenzymes when assayed at pH 6.0 (Fig. 2D to F). These results,together with results from the pH study, indicate that the enzymaticcomponents of the two multimeric enzymes retain their parentalproperties.

The multimeric enzymes are kinetically comparable to the parental enzymes.

Top

The multimeric enzymes are...

To compare the kinetic parameters of the multimeric enzymeswith those of the parental enzymes, the chromophore-tagged syntheticsubstrate 4NPA was used for kinetic analysis of arabinofuranosidaseactivity, following the procedures described elsewhere (19).The 4NPA substrate hydrolysis data were subjected to nonlinearcurve fitting to the Michaelis-Menten equation by using GraphPadPrism 5 software (GraphPad Software, CA). When assayed at 45°Cwith 50 mM phosphate buffer (pH 6.0) containing 100 µg/mlbovine serum albumin, the K_m values for DeAFc, AXX, and CBD-AXXwere 188 ± 12 µM, 200 ± 29 µM, and194 ± 12 µM, respectively; the k_cat values forDeAFc, AXX, and CBD-AXX were determined to be 1.1 ± 0.02s^–1, 0.9 ± 0.03 s^–1, and 1.0 ± 0.01s^–1, respectively. The similar K_m and k_cat values forDeAFc and the multimeric enzymes with 4NPA indicate that theDeAFc component of the multimeric enzymes maintains its parentalcatalytic efficiency.

The multimeric enzymes are highly active on natural arabinoxylans.

Top

An enzyme-coupled assay (19) was used to measure xylose releaseby the enzymes, using water-soluble wheat and rye arabinoxylanas substrates. Similar to the parental enzyme mixture, AXX andCBD-AXX were highly active in hydrolyzing natural xylans (Fig.3). The specific activities of CBD-AXX are approximately 5%and 12% higher than those of AXX for wheat and rye arabinoxylan,respectively, which is consistent with results obtained usingsynthetic substrates. Because the cbpA CBD used in CBD-AXX doesnot bind xylan (6), the increased activity of CBD-AXX on arabinoxylansis likely due to optimization of the overall structure by theCBDs rather than from enhanced substrate binding.

View larger version (33K):
[in this window]
[in a new window]

FIG. 3. Enzyme activities toward natural arabinoxylans. Equal molar concentrations (25 nM) of the parental xylanase (xyln), mixtures of two or three parental enzymes (xyln+xylo and xyln+ xylo+DeAFc, respectively), and AXX or CBD-AXX were assayed for their activities on water-soluble wheat arabinoxylan and rye arabinoxylan. An enzyme-coupled assay (19) was used to measure xylose release in 50 mM phosphate buffer (pH 6) containing 50 µg/ml bovine serum albumin at 45°C.

CBD-AXX is more efficient in hydrolysis of corn stover.

Top

CBD-AXX is more efficient...

The broad-substrate specificities of the parental enzymes enablethe liberation of sugars from corn stover by the enzyme mixtureand the multimeric enzymes (Fig. 4A), as determined using acolorimetric assay described previously (1). The combined effectsof the multimeric enzymes in the presence of cellulases werealso determined. In the presence of purified recombinant cellulaseE1 of Acidothermus cellulolyticus (GenBank accession U33212;purified in our laboratory), considerably more sugars were releasedby both the enzyme mixture and the multimeric enzymes (Fig.4B). A similar result was also observed when a commercial cellulasepreparation (Alltech, Nicholasville, KY) was used (Fig. 4C).

View larger version (17K):
[in this window]
[in a new window]

FIG. 4. Hydrolysis of corn stover by the enzyme mixture or the multimeric enzymes in the presence or absence of cellulases. Assays were performed at 45°C with 50 mM phosphate buffer (pH 6) containing 20 nM of each enzyme and 30 mg/ml of corn stover in the presence or absence of 10 nM cellulase E1 or 1.5 µg commercial cellulase. (A) Reducing sugar release by enzyme mixture ( ${blacksquare}$ ), AXX ( ${blacktriangleup}$ ), and CBD-AXX ( ${triangleup}$ ) in a course of 48 h. (B) Reducing sugar release by cellulase E1 (x), the enzyme mixture plus E1 ( ${blacksquare}$ ), AXX plus E1( ${blacktriangleup}$ ), and CBD-AXX plus E1 ( ${triangleup}$ ). (C) Reducing sugar release by the commercial cellulase (•), the enzyme mixture plus the commercial cellulase ( ${blacksquare}$ ), AXX plus the commercial cellulase ( ${blacktriangleup}$ ), and CBD-AXX plus the commercial cellulase ( ${triangleup}$ ).

In corn stover assays, AXX resembles the parental enzyme mixtureexcept when assayed with the commercial cellulase, which resultedin a relative decrease in AXX activity after 24 h (Fig. 4C).In all cases, however, CBD-AXX is distinctly more active thaneither AXX or the enzyme mixture in corn stover hydrolysis.Thermostability assays of AXX and CBD-AXX indicated little differencebetween them; after 48 h, both proteins retained 60 to 80% oftheir activities, depending on the substrates used in the assay(data not shown). Therefore, protein stability is unlikely tobe responsible for the superior performance observed for CBD-AXX.The carbohydrate-specific binding modules greatly enhance theenzyme activities against insoluble substrates, such as cellulose(2, 13). Protein engineering fusing CBDs with carbohydrate hydrolaseshas led to increased cellulosic binding efficiency (16). Ourresults are consistent with a higher affinity for binding ofCBD-AXX than non-CBD-containing enzymes to corn stover, leadingto higher hydrolysis efficiency. The observations that CBD-AXXdoes not possess significantly higher activity than AXX or theenzyme mixture toward arabinoxylans, which have very low cellulosecontent, provide further support for the specific role of CBDsin cellulosic binding. Nonetheless, it is possible that theCBDs not only enhance substrate binding but also improve theoverall structure of CBD-AXX, leading to the observed increasein corn stover hydrolysis.

Naturally occurring and artificial multidomain enzymes are usefulin metabolic engineering (3). We and others have demonstratedthat the creation of artificial, multifunctional, lignocellulosichydrolases is a realistic and practical approach for the improvementof biomass conversion (5, 8, 9, 12). These novel enzymes canbe used alone, when a 1:1:1 stoichiometry is preferred, or inmixtures with small quantities of other enzymes to create anoptimal enzyme cocktail when enzyme ratio is critical. Becauseof the potential synergy and ease for transformation, engineeredfusion enzymes can be used to improve industrial microbes (11)or transgenic plant feedstock (10). However, artificial enzymeswith three to five functional subunits are not commonly reported.The multimeric enzymes characterized in this study closely resemblethe parental enzymes or their mixtures with regard to theirenzymatic properties. CBD-AXX surpasses both AXX and the parentalenzyme mixture in hydrolysis of corn stover, providing supportfor our design strategy of using CBDs to enhance substrate binding.The multifunctional enzymes have the potential to be more cost-effectivein industrial enzyme production than mixtures of multiple singleenzymes.

ACKNOWLEDGMENTS

We thank D. Zaitlin and K. Shen for their critical review ofthe manuscript.

This work was supported by a grant from the U.S. Departmentof Agriculture to L.Y. (2006-35504-17413).

FOOTNOTES

* Corresponding author. Mailing address: Department of Plant and Soil Sciences, Kentucky Tobacco Research and Development Center, University of Kentucky, Cooper and University Drives, Lexington, KY 40546. Phone: (859) 257-4086. Fax: (589) 323-1077. E-mail: lyuan3{at}uky.edu

Published ahead of print on 16 January 2009.

Supplemental material for this article may be found at http://aem.asm.org/.

REFERENCES

Top