Cloning, expression and characterization of a family-74 xyloglucanase from Thermobifida fusca Diana C. Irwin, Mark Cheng*, Bosong Xiang†, Jocelyn K. C. Rose‡ and David B. Wilson Department of Molecular Biology and Genetics, Cornell University, Ithaca, New York, USA Thermobifida fusca xyloglucan-specific endo-b-1,4-gluca- nase (Xeg)74 and the Xeg74 catalytic domain (CD) were cloned, expressed in Escherichia coli, purified and charac- terized. This enzyme has a glycohydrolase family-74 CD that is a specific xyloglucanase followed by a family-2 carbo- hydrate binding module at the C terminus. The Michaelis constant (K m ) and maximal rate (V max ) values for hydrolysis of tamarind seed xyloglucan (tamXG) are 2.4 l M and 966 lmol xyloglucan oligosaccharides (XGOs) min )1 Ælmol protein )1 . More than 75% of the activity was retained after a 16-h incubation at temperatures up to 60 °C. The enzyme was most active at pH 6.0–9.4. NMR analysis showed that its catalytic mechanism is inverting. The oligosaccharide products from hydrolysis of tamXG were determined by MS analysis. Cel9B, an active carboxymethylcellulose (CMC)ase from T. fusca, was also found to have activity on xyloglucan (XG) at 49 lmolÆmin )1 Ælmol protein )1 , but it could not hydrolyze XG units containing galactose. An XG/cellulose composite was prepared by growing Gluconacetobacter xylinus on glucose with tamXG in the medium. Although a mixture of purified cellulases was unable to degrade this material, the composite material was fully hydrolyzed when Xeg74 was added. T. fusca was not able to grow on tamXG, but Xeg74 was found in the culture supernatant at the same level as was found in cultures grown on Solka Floc. The function of this enzyme appears to be to break down the XG surrounding cellulose fibrils found in biomass so that T. fusca can utilize the cellulose as a carbon source. Keywords: xyloglucanase; cellulase; inverting; regulation; plant cell walls. Converting plant biomass into ethanol for use as fuel has been a long-term goal of scientists studying cellulases and related glycosyl hydrolases. Much progress has been made in identifying, cloning, expressing and characterizing cellu- lases from both aerobic and anerobic bacteria and from fungi. Hundreds of such enzymes have been identified and a list of glycohydrolase families can be found at http:// afmb.cnrs-mrs.fr/CAZY/. Lynd et al. [1] have written a comprehensive review of the current information on pos- sible future strategies for biomass hydrolysis. A natural biomass substrate, such as corn fiber [2], is structurally complex and many other enzymes besides cellulases are needed for efficient degradation of the polysaccharides to monosaccharides. The load-bearing structure of primary plant cell walls comprises a network of cellulose fibrils complexed through noncovalent associations with hemicelluloses such as xylo- glucan (XG) and arabinoxylan (AXG) [3]. An additional network consists of pectic polysaccharides as well as other, less abundant, wall components, including structural pro- teins, proteoglycans and hydrophobic compounds [4]. Cellulose microfibrils are composed of noncovalently, but tightly associated, linear chains of b-1,4-linked D -gluco- pyranosyl residues. XG is the predominant hemicellulose in dicotyledon type I cell walls and has a b-1,4-linked glucopyranosyl backbone of repeating cellotetraose units with a- D -xylosyl residues attached to C6 of two or more of the first three residues. In addition, some of the xylosyl residues are substituted to form oligomeric side-chains containing galactosyl, arabinose or fucosyl residues [3]. The nomenclature for XG subunits is given in Fig. 1 [5]. XG is proposed to form a monolayer coating the surface of cellulose microfibrils and to penetrate the cellulose in amorphous areas [6]. Experiments using differential extrac- tion of etiolated pea stems with a family-12 xyloglucanase, KOH and a cellulase, suggested that some of the XG is entrapped within or between cellulose microfibrils [3]. Individual XG polymers are also thought to cross-link adjacent microfibrils, forming a complex three-dimensional lattice [4,6], underscoring the potentially important struc- tural role of XG. Thermobifida fusca YX is a thermophilic actinomycete that was originally isolated by Dexter Bellamy [7]. It grows well at 50 °C in minimal medium with carbohydrate Correspondence to D. B. Wilson, 458 Biotechnology Building, Cornell University, Ithaca, New York, USA. Fax: 1607 255 6249, Tel.: 1607 255 5706, E-mail: dbw3@cornell.edu Abbreviations: AXG, arabinoxylan; BMCC, bacterial microcrystalline cellulose (Cellulon TM ); CBM, cellulose-binding module; CD, catalytic domain; CMC, carboxymethylcellulose; GBG, bacterial cellulose; GBX, bacterial cellulose/xyloglucan composite; PAHBAH, p-hydroxybenzoic acid hydrazide; SC, phosphoric acid swollen cellu- lose; tamXG, tamarind seed xyloglucan; TFSF, Thermobifida fusca crude supernatant enzymes; Xeg, xyloglucan-specific endo-b-1,4- glucanase; XG, xyloglucan; XGO, xyloglucan oligosaccharide. Present addresses: *Beth Israel Hospital, New York City, NY, USA. US Plant, Soil and Nutrition Laboratory, Tower Road., Ithaca, NY 14853, USA. àDepartment of Plant Biology, Cornell University, Ithaca, New York, USA. (Received 3 April 2003, revised 19 May 2003, accepted 30 May 2003) Eur. J. Biochem. 270, 3083–3091 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03695.x polymers such as cellulose, starch, xylan, or mannan as the sole carbon source and it is thought to be an important organism in the degradation of biomass. Six T. fusca cellulase genes and a xylanase gene have been cloned, expressed and characterized [8–11]. The genome of this organism has been sequenced (http://genome.ornl.gov/ microbial/tfus/) and includes many genes coding for a variety of glycosyl hydrolases. One such gene codes for a glycosyl family-74 hydrolase with a typical bacterial family-2 cellulose-binding module (CBM) at the C terminus. In the present study we show that this enzyme is a xyloglucanase and investigate its properties and its possible role in the degradation of plant biomass. Materials and methods Protein production and purification Clone KPR17269, containing the gene for T. fusca xylo- glucan-specific endo-b-1,4-glucanase (Xeg)74, was obtained from Stephanie Stilwagen Malfatti (DOE Genome Institute, Walnut Creek, CA, USA). A PCR product of the gene was made using the forward primer, 5¢-CGTCCACTCC GC GGCCGCCCCCGCCTC, which creates a NotI site (under- lined) near the predicted mature N-terminal codon (in italics), and the reverse primer, 5¢-CCCTCGTGCG CTCG AGGTACCAGGGCTTTGC, which has an XhoI site after the C-terminal codon. Plasmid pD1164 was created by ligating gel-purified fragments of pET26B+ (Novagen) digested with NdeIandXhoI, a NdeI–NotI fragment containing the T. fusca Cel6A signal peptide and the above PCR fragment digested with NotIandXhoI. This ligation mixture was transformed into Escherichia coli DH5 alpha, and plasmid minipreps (Qiagen) were used to identify a transformant containing the desired plasmid which was then transformed into E. coli BL21-DE3 (Novagen) and also into BL21 (DE3)-RP codon plus (Stratagene). An Xeg74 catalytic domain (CD) expression plasmid was created by ligating the 1.2-kb NotI–SphI fragment of pD1164, the 5.4-kb SpeI–NotI fragment of pD1164 and a 1.0-kb SphI–SpeI PCR fragment, which created a stop codon after the amino acid sequence GDLDG (mature amino acid 736). The ligation mix was transformed into BL21 gold (DE3) (Stratagene) and the correct strain (pMC1, D1212) was identified, as described above. The portion of each plasmid created by PCR was sequenced and no unwanted mutations were detected. D1170 (Xeg74) or D1212 (Xeg74CD) were cultured, from frozen stocks, overnight at 30 °C in Luria–Bertani (LB) broth containing 0.5% glucose and 60 lgÆmL )1 kanamycin. Thirty milliliters of culture was used to ino- culate 1 L of M9 containing 0.5% glucose and 60 lgÆmL )1 kanamycin. These cultures were grown to a D 600 of 0.9, isopropyl thio-b- D -galactoside was added to 0.8 m M ,and the cultures were allowed to grow overnight at 30 °C. The supernatant was clarified by centrifugation, NH 2 SO 4 was addedto1.2 M , and the supernatant was further clarified by depth microfiltration using a CUNO Beta Pure polyolefin #46368-02L cartridge. This material was loaded onto a phenyl sepharose CL-4B (Sigma) column (10 mLÆL )1 of supernatant) and the column was washed with two volumes of 0.8 M NH 2 SO 4 +10m M NaCl + 5 m M Kpi, pH 6, followed by three volumes of 0.4 M NH 2 SO 4 +5m M NaCl + 5 m M Kpi, pH 6, and eluted with 5 m M Kpi, pH 6. The purest fractions, as determined by SDS/PAGE, were combined and loaded on a Q-Sepharose Fast Flow (Pharmacia Biotech) column (2 mg of proteinÆmL )1 of column). The protein was eluted with a gradient (20· column volume) of 0–0.7 M NaCl in 10 m M Bis/Tris, pH 5.6, + 10% glycerol. The purest fractions were com- bined and concentrated in a stirred cell using a PTGC 10 000 MWCO ultrafiltration membrane (Millipore). The proteins were stored at )70 °Cin5m M NaOAc, pH 5.5, containing 10% glycerol. The final yield of purified protein was 22 mgÆL )1 Xeg74 and 14 mgÆL )1 Xeg74CD. T. fusca crude cellulase was prepared by concentration of T. fusca culture supernatant after growth on Solka Floc powdered cellulose (James River Corporation, Berlin, New Hamp- shire, USA), as previously described [2]. Cel9B was expressed in the supernatant of a Strepto- myces lividans clone, pSHE1, and purified as previously described [12]. The molecular masses of the purified proteins were determined using a Bruker Biflex III MALDI-TOF spectrometer instrument at the Cornell University Bio- resources Center. Assay methods Tamarind seed xyloglucan (arabinose/galactose/xylose/glu- cose; 3 : 16 : 36 : 45) (tamXG) was obtained commercially (Megazyme, Ireland) or extracted as described previously [13]; XG-bean was isolated from the media of bean (Phaseolus vulgaris) suspension cell cultures, as described previously [14,15]. Except where stated otherwise, xylo- glucanase activity was assayed in 500 lL microcentrifuge tubes by incubating 200 lL of samples containing 2–2.5 mgÆmL )1 tamXG, 0.05- M NaH 2 PO 4 /K 2 HPO 4 (pH 7.5) buffer, and enzyme for 30 min at the desired temperature. Twenty-microliter samples (in triplicate) were removed and added to 1.5 mL of p-hydroxybenzoic acid hydrazide (PAHBAH) reagent [16] and boiled for 6 min according to the published procedure. The absorbance (A) at 410 nm was read and a reference curve was prepared using glucose (0–25 lg). Fig. 1. Example of a xyloglucan (XG) oligosaccharide structure (XXLG) and nomenclature [5]. XXXG, XXLG, XLXG, and XLLG are known subunits of tamarind seed xyloglucan (tamXG) [15]. XGs from other sources vary in composition and the XG from some dicotyledonous plants has a- L -fucose (1 fi 2) added to some of the galactose residues, while solanaceous plants produce more complex arabinoxyloglucans (AXG) [23,26,28,34]. In addition, residues may contain either one or two O-acetyl groups on C-2, C-3 or C-6. 3084 D. C. Irwin et al. (Eur. J. Biochem. 270) Ó FEBS 2003 The pH optimum assays used buffers prepared by mixing 30-m M solutions of citric acid, NaH 2 PO 4 , boric acid, and barbital to obtain the desired pH. The final concentration of buffer in the assays was 15 m M . The temperature stability of the enzyme was tested by incubation of 2.7 l M Xeg74 for 16 h at 0–75 °C followed by performing the XG activity assay at 50 °C. Kinetic constants were determined with substrate con- centrations from 2 to 60 lgÆmL )1 and 0.375 pmolÆmL )1 enzyme. The bicinchoninate assay [17] was used to quantify the reducing sugars produced. Michaelis constant (K m )and maximal rate (V max ) values were calculated from a plot of the initial reaction rates vs. substrate concentration using Kaleidagraph ( SYNERGY Software) to fit the data to the Michaelis-Menten equation. All reducing sugar assays included a glucose standard curve. The average molecular mass of the XG oligosaccha- rides (XGOs) was calculated from the manufacturer’s data and was found to be 1293 Da. This value agreed with the MS analysis of the tamXG digestion products. Exhaustive digestion of tamXG by Xeg74 created a set of XGOs from known amounts of substrate (0–1500 lg), and the approxi- mate relationship between the glucose standard curve and the nmoles of XGOs produced was determined for both the PAHBAH and the bicinchoninic acid reducing sugar methods. For the PAHBAH method, the nmol of XGOs ¼ nmol of glucose equivalents ·0.481; for the bicinchoninic acid method, the nmol of XGOs ¼ nmol of glucose equivalents ·0.686. NMR experiments were performed at the Chemistry and Chemical Biology NMR facility, Cornell University, by the method of Pauly et al. [18]. TamXG (5 mgÆmL )1 )was addedto10m M NaCl in D 2 Oat80°C and stirred overnight. Xeg74 was prepared for NMR experiments by resuspending and concentrating it six times with 10-m M NaCl in D 2 O using a Centricon 30 (Millipore). Three- hundred micrograms of Xeg74 was added to 0.7 mL of tamXG in an NMR tube and spectra were taken over a 100- min time course. Samples were prepared for MS by reacting 0.4 nmol Xeg74 or Cel9B with 5 mg of tamXG in 1 mL of 5 m M NaOAc (pH 5.5) buffer containing 0.02% NaN 3 ,at50°C for 48 h on a rotator. Positive-mode electrospray MS was performed on a Bruker Esquire LC ion-trap mass spectro- meter. TLC of hydrolysis products was performed using What- man LK5D 150-A silica gel thin-layer plates with two ascents of the solvent, ethyl acetate/water/MeOH (40 : 15 : 20). Plates were stained (100 mL of acetic acid, 1 mL of p-anisaldehyde, 2 mL of concentrated sulfuric acid) and then heated for 1 h at 95 °C, as described by Chirco & Brown [19] and Jung et al. [12]. Glucose and xylose oligomer standards were obtained from Seikagaku America or Sigma. Preparation of GBG, GBX and tomato cell walls Bacterial cellulose (GBG) and bacterial cellulose/xyloglucan composite (GBX) were produced from Gluconaceto- bacter xylinus (formerly Acetobacter xylinus) ATCC 53524 according to a procedure published previously [6]. The resultant gel-like material was harvested, rinsed six times on afilterwithdH 2 O and stored at 4 °C in 0.04% NaN 3 . Culture reducing sugars were removed by rotating a solution of the coarsely chopped GBG or GBX in 0.05- M NaKPi, pH 7.4, for at least 2 h at 50 °C followed by further washing with 0.04% NaN 3 . Pieces of the gels were rinsed with water, blotted on filter paper and chopped finely with a razor blade. The dry mass of the material was approxi- mately 1% of the blotted gel and overdigestion of GBX with Xeg74 gave an XG content of 16%. In order to equalize the amount of cellulose to be digested, 42 mg of GBG or 50 mg of GBX were weighed into 0.5-mL screw-cap microfuge tubes. Buffer (0.05- M NaKPi, pH 7.4) and enzymes were added to achive a final volume of 0.5 mL and the microfuge tubes were then incubated on a rotator for 16 h at 50 °C. Reducing sugars in the supernatant were measured by the PAHBAH method, as described above, using glucose as a standard. Each data point represents the average of three separate digestions and the PAHBAH measurements for each digestion were run in triplicate. Tomato cell walls were isolated as follows [14,20]: 50 g of outer pericarp tissue from green tomatoes was finely diced, frozen in liquid nitrogen and ground into a fine powder. This material was boiled in 500 mL of 95% EtOH for 40 min, filtered on Miracloth TM (Calbiochem), resuspended in boiling EtOH, and refiltered. The solid material was resuspended in 500 mL of CHCl 3 /MeOH (1 : 1), filtered on a glass frit and washed with 500 mL of acetone. These steps inactivate endogenous cell-wall enzymes and extract low- molecular-mass solutes. To remove starch, the solid mater- ial was resuspended in 15 mL of dimethylsulfoxide/H 2 O (9 : 1) and stirred for 1 h, centrifuged and washed three times with dimethylsulfoxide/H 2 O (9 : 1). The solid mater- ial was resuspended in trans-1,2-diaminocyclohexane- N,N,N¢,N¢-tetraacetic acid (CDTA)/bicarb buffer (50 m M CDTA, 50 m M Na 2 CO 3 ) containing 20-m M NaBH 4 (pH 6.5) to a final volume of 125 mL and stirred overnight. (The CDTA chelates Ca 2+ , facilitating the release of pectin, and the NaBH 4 removes reducing groups.) This material was filtered on Miracloth TM (Calbiochem), rinsed several times with CDTA/bicarb buffer, stirred overnight at room temperature, homogenized, filtered again, and washed with CDTA/bicarb buffer until the filtrate was clear. The solid material was resuspended and dialyzed against 0.01 M Tris (pH 7.0) containing 0.02% NaN 3 ,at4°C. Assays were carried out, as for GBG and GBX, using 100 lL (6.8 mgÆmL )1 dry mass) of the tomato cell wall preparation as substrate in a total volume of 500 lL. The tomato XG was from Bree Urbanovicz and prepared by extraction of tomato cell walls with 1 M KOHfor1handthenwith4 M KOH at room temperature with stirring overnight. The supernatants were filtered through nylon mesh and neut- ralized, on ice, to pH 7.0 with glacial acetic acid. XG was precipitated from the 4 M extraction with two volumes of ethanol, cooled on ice, centrifuged, washed three times with cold ethanol, resuspended in water and then freeze dried. Western blots Culture supernatants were analyzed for the presence of Xeg74 by separation on SDS/polyacrylamide gels [21] followed by transfer to Immobilon-P poly(vinylidene diflu- oride) membranes (Millipore). Rabbit polyclonal antisera Ó FEBS 2003 Thermobifida fusca xyloglucanase Xeg74 (Eur. J. Biochem. 270) 3085 raised against recombinant Xeg74 was used as the first antibody and goat anti-rabbit Ig alkaline phosphatase conjugate (Bio-Rad) was used as the second antibody. Immunodetection was performed using nitro-blue tetra- zolium and 5-bromo-4-chloro-3-indolyl phosphate, accord- ing to the Bio-Rad protocol. The GenBank accession number for the T. fusca genome is NZ_AAAQ00000000. The file containing the Xeg74 gene is NZ_AAAQ01000018 and the gene is Tfus0318, CDS 39974.42751. The protein identification is ZP_00056977. The accession numbers for T. fusca cellulases are: Cel5A, Q01786; Cel6A, P26222; Cel6B, Q60029; Cel9A, P26221; Cel9B, Q08166; and Cel48A, AAD39947. T. fusca YX (BAA-629) and T. fusca ER1 (BAA-630) have been depo- sited in the American Type Culture Collection. Results Cloning and purification The gene for Xeg74 was cloned into pET26b using the T. fusca Cel6A signal sequence (MRMSPRPLRALL GAAAAALVSAAALAFPSQAA) in place of its native signal sequence. Cel6A, Cel6Acd, and Cel48A have been expressed and secreted successfully by E. coli [11,22] using this signal sequence and it has a convenient NotIsiteatits C terminus, which allows cloning in-frame with alanine as the first amino acid of the mature protein. An alignment of the amino acid sequences of 11 family-74 catalytic domains shows that their amino acid identity to T. fusca Xeg74 ranges from 29 to 63% with the alignment starting at GYTWR. Xeg74 was predicted by SignalP (http:// www.cbs.dtu.dk/services/SignalP-2.0/#submission) to have the mature N-terminal sequence, APASATTGYTWR, with the start of the conserved sequence at amino acid 8. The catalytic domain was produced from the same vector after inserting a stop codon after amino acid 736 ending with VGDLDG. This C-terminal sequence agrees well with that of other family-74 catalytic domains and, in the native protein, is followed by a linker region, (737) PPPQPTEEP…, which is similar to the linker region in T. fusca Cel9A [12]. Of the expressed protein, about 90% was secreted to the culture supernatant and about 10% was in the shock fluid, as determined by SDS gels (data not shown). The molecular mass of purified Xeg74CD, as determined by MS, was 79 443 Da (expected mass 79 480 Da). However, the mass spectrum of Xeg74 showed a broad set of peaks with mass values from 96 200 to 94 718 (expected mass 94 705 Da), indicating that the signal peptide was being cut in different places, resulting in the addition of up to 17 extra amino acids at the N terminus of the protein. It is not clear why the Xeg74CD signal peptide cuts cleanly and Xeg74 does not, as both genes are cloned in the same way except for the presence of the linker and the family-2 CBM at the C terminus of XEG74. The molecular masses of Cel9A-68 (CD + family3c CBM) and Cel48ACD, both cloned with the Cel6A signal peptide, were also determined by MS and the peaks also had several small shoulders at higher molecular masses although they were much sharper than the peak for Xeg74. Possibly the more complicated domain structure of Xeg74 (a catalytic domain, a linker, and a CBM) results in nonspecific cleavage of the signal sequence. Characterization Xeg74 had very low activity on swollen cellulose (SC) or carboxymethyl cellulose, and assays using tamXG, barley b-glucan, Avicel, locust bean gum (galactomannan), soluble starch, xylan, pectin, or corn fiber as substrates showed that Xeg74 had significant activity only on tamXG. The products of Xeg74 hydrolysis of tamXG were analyzed by TLC (Fig. 2A). Digestion of tamXG was complete within 1.5 h and produced three main bands with no carbohydrate remaining at the origin. In contrast, Xeg74 digestion of SC and carboxymethylcellulose (CMC), after overnight incu- bation, produced only very faint bands of cellotriose (G3), cellotetraose (G4) and cellopentaose (G5). Each of the six purified T. fusca cellulases was tested for activity on tamXG and only Cel9B was active. However, it produced only two of the three product bands and there was a large quantity of undigested material remaining at the origin (Fig. 2B). Xeg74 had very limited activity on G4, G5 and G6 and only faint product bands were produced after overnight digestion (Fig. 2C). Xeg74 was active on both bean and tomato XG and inactive on boiled barley b-glucan (Fig. 2D,E). The molecular masses of the digestion products of tamXG by Xeg74 and Cel9B were determined by MS (Fig. 3). Using the known composition of tamXG [15,23], the possible products were determined, as shown in the figure. Xeg74 cleaved the backbone of XG to give products decorated with two or three xylose units and further decoratedwithuptotwogalactosemolecules.XXGGwas not reported by York et al. [15] to be a component of tamXG, but there is a peak at mass 953.8, which appears to be XXGG in Fig. 3A. Cel9B only cleaved a portion of the tamXG (Fig. 2B, lane 5) and did not produce any products containing galactose, XXGG appears to represent the major product of hydrolysis by this enzyme and this suggests that Cel9B prefers to cleave where there are two undecorated glucoses, although it can also cleave slowly to produce XXXG. The K m wasdeterminedtobe3.2and3.9lgÆmL )1 for Xeg74 and Xeg74CD, respectively. This corresponds to 2.4 and 3.0 l M using a molecular mass of 1293 for the average XGO hydrolyzable unit. The V max of Xeg74 was 966 lmol XGOÆmin )1 Ælmol )1 protein, while that of Xeg74CD was somewhat higher, at 1257 XGOÆmin )1 Ælmol )1 protein. The specific activities were also determined using a similar assay to that used for cellulases [8] with a substrate concentration of 2.5 mgÆmL )1 and increasing amounts of enzyme. The data shows a linear relationship up to 50% digestion and the specific activities were 578 and 875 lmol XGOÆmin )1 Æ lmol )1 enzyme for Xeg74 and Xeg74CD, respectively. The extra amino acids left on the N terminus from the signal peptide may have caused the whole protein to have lower specific activity than the CD, or the smaller size of the CD may enable it to fit into the xyloglucan/cellulose matrix more easily, increasing the availability of the insoluble xyloglucan substrate to the enzyme. The specific activity of Cel9B was also measured in this way and was found to be 46 lmol XGOÆmin )1 Ælmol enzyme at 15% digestion. 3086 D. C. Irwin et al. (Eur. J. Biochem. 270) Ó FEBS 2003 However, the maximum digestion obtained was about 30% of the total substrate and the curve was not linear. In contrast, Cel9B has activities of 5410 and 363 lmol cellobioseÆmin )1 Ælmol )1 , respectively, on CMC and SC while Xeg74 activity on CMC and SC was 1.1 and 1.9 lmol cellobioseÆmin )1 Ælmol )1 , respectively. Xeg74 retained more than 60% activity at pH 6.0–9.4 and 90% activity at pH 6.9–8.6 (Fig. 4A). When assayed with 0.025 M (pH 7.5) NaH 2 PO 4 /K 2 HPO 4 , Hepes, Tris/ HCl, or NaOAc buffers, the enzyme had 100%, 83%, 73%, and 92% activity, respectively. Temperature stability tests showed that the enzyme retained full activity after 16 h of incubation at 55 °C and 76% activity after incubation at 60 °C (Fig. 4B). Thirty-minute activity assays showed that the activity increased with temperature up to 76 °C (Fig. 4B). Enzymatic hydrolysis of b-glycosidic bonds occurs by two general mechanisms giving rise to either retention or inversion of the anomeric configuration [24]. Figure 5 shows the [ 1 H]-NMR spectra of tamXG after reaction with Xeg74for0,5,20and100min.Thea-anomeric proton appears at 5 min, indicating an inverting enzyme, and after 20 min the b-anomeric proton appears as the result of mutarotation. The NMR assignments relied on the work of Pauly et al. [18] for the family-12 Aspergillus aculeatus xyloglucanase, a retaining enzyme, in which the b-anomeric proton appeared first followed by the subsequent appear- ance of the a-anomeric proton. Concentrated supernatants from T. fusca cultures grown on Solka Floc, xylan, or corn fiber, all had activity on tamXG and revealed the same product pattern on TLC as observed for Xeg74 (Fig. 2B). The xylan-grown supernatant produced four additional brown-colored bands of lower molecular mass; although the composition of these bands is not known, the brown color indicates that they contain xylose. A Western blot with polyclonal rabbit antiserum prepared against purified Xeg74 showed that the level of this protein was highest in the super- natants of the Solka Floc-, xylan-, or corn fiber-grown cultures and was present at low levels when T. fusca was grown on glucose, cellobiose, xylose, or bacterial micro- crystalline cellulose (BMCC) (Fig. 6). T. fusca was not able to grow with tamarind, tomato, or bean XG as the sole carbon source. When glucose was added to the culture, growth resumed, ruling out the presence of an inhibitor. Although attempts to culture T. fusca on tamXG produced little or no growth, high levels of Xeg74 were induced (Fig. 6, lane 11) and TLC analysis of the culture supernatant showed that the XG had been degraded to the expected products (data not shown). The Xeg74 antiserum also reacted with T. fusca Cel6A, but not with the Cel6A catalytic domain (Fig. 6, lanes 9 and 10), which indicates that the CBM of Cel6A cross-reacts with the Xeg74 antiserum. The family-2 CBMs of all six cellulases show 33–52% amino acid identity with the Xeg74 CBM (43% identical to Cel6A CBM). Presumably these two CBMs have antigenic epitopes in common while the others do not. This cross-reaction is useful for comparing the differences in induction of the two enzymes. Cel6A is induced by growth on cellobiose, Solka Floc, BMCC, and corn fiber, while Xeg74 is present in all of the supernatants but is found at a higher level in cultures grown on XG, xylan, Solka Floc, and corn fiber (Fig. 6). The low level of Xeg74 found in BMCC cultures Fig. 2. TLC analyses of reaction products (A–E). All reactions were run at 50 °C for 16 h except for lane 2, which was incubated for 1.5 h. The enzymes and substrates used are noted on the figure under each lane using the following abbreviations: 74, xyloglucan-specific endo-b-1,4-glucanase (Xeg)74; and 9B, Cel9B. CF, XY, and SF indicate hydrolysis by enzymes in concentrated crude supernatants from cultures of Thermobifida fusca ER1 grown on corn fiber, xylan, or Solka Floc, respectively. Abbreviations: tom, tomato; tam, tamarind; and G6, cellohexose. Standards for each TLC analysis were glucose, cellobiose, cellotriose, cellotetraose, cellopentaose (G1–G5) and xylose and xylobiose (X1–X2). Ó FEBS 2003 Thermobifida fusca xyloglucanase Xeg74 (Eur. J. Biochem. 270) 3087 shows that it is not the cellulose in Solka Floc which is inducing the higher level of Xeg74 but some minor component. Role of Xeg74 G. xylinus synthesizes cellulose I chains that extrude parallel to the bacterial wall and which coalesce into bundles of highly crystalline microfibrils [25]. When G. xylinus was grown in the presence of glucose and tamXG, 38% of the added XG was incorporated into the cellulose pellicle and microscopic analysis showed the formation of cross-linkages between the ribbons [6]. The level of incorporated XG was similar to that found in primary cell walls and much of the XG was thought to be intimately bound to the surface or woven into the cellulose fibers [6]. The role of Xeg74 in plant biomass degradation was studied using the GBX composite produced by G. xylinus when grown on glucose plus tamXG and, as a control, GBG was prepared by growth on glucose. A mixture of purified T. fusca cellulases (cel mix) consisted of Cel5A (an endocellulase), Cel6A (a nonreducing end-directed exocellulase), Cel9A (a processive endocellulase), and Cel48A (a reducing end-directed exocellulase). Under the assay conditions, Xeg74 enzyme alone (0.05 nmol, 4.7 lg) produced 58 lg of reducing sugar from GBX and 6.9 lg from GBG. The cel mix alone was not able to degrade GBX to any appreciable extent; however, when Xeg74 was added, the activity was very similar to that of concentrated T. fusca crude supernatant enzymes (TFSF) (Fig. 7A). The reactions of the cel mix and the cel mix + Xeg74 on GBG were very similar to that of TFSF on GBG (data not shown). XG protects the cellulose microfibrils from degradation by cellulases. Xeg74CD was foundtohave 86% of the activity of whole Xeg74 when combined with the cel mix on GBX, implying that the CBM contributes to the degradation, but is not essential. Tomato cell walls that had been processed to inactivate endogenous enzyme activity and remove low-molecular- mass solutes and starch, were used as a substrate in a similar assay. Figure 7B shows that the amount of reducing sugar produced by the cel mix, plus and minus Xeg74, was about the same. However, TFSF was able to produce four times as much reducing sugar as the cel mix. Analysis of the hydrolysate by TLC showed glucose and cellobiose to be the Fig. 3. MS analysis of the products of tamarind seed xyloglucan (tam- XG) hydrolysis by xyloglucan-specific endo-b-1,4-glucanase (Xeg)74 (A) and Cel9B (B). The theoretical mass values are given in parentheses for possible products. Fig. 4. Temperature and pH optima for Xeg74 activity. (A) Percentage xyloglucanase activity at various pH values. (B) Effect of temperature on the activity and stability of xyloglucan-specific endo-b-1,4-gluca- nase (Xeg)74. Stability was tested by incubating xyloglucan-specific endo-b-1,4-glucanase (Xeg)74 at the indicated temperatures for 16 h followedbydilutionanda30-minassayat50°C. 3088 D. C. Irwin et al. (Eur. J. Biochem. 270) Ó FEBS 2003 dominant products, with only faint bands at the higher molecular massess expected for XGOs (Fig. 2D, lane 19). The products from Xeg74 hydrolysis of the 4 M KOH- extracted tomato XG are shown in Fig. 2E, lane 22. XGO bands are produced, but much of the substrate remains undegraded, at the origin. An activity assay measuring reducing sugars showed that a maximum of  26% of this tomato XG extract could be degraded by Xeg74. This may reflect the presence of small amounts of polysaccharides other than XG in the extract and/or variability in the structure of tomato XG. For example, the AXG of solanaceous plants, such as tomato, is characterized by the presence of arabinose-containing side-chains, such as b-ara-(1fi 3)-a- L -ara-(1fi 2)-a- D -xyl [26]. It is possible that theseresiduesdonotfitintheactivesiteofXeg74. Discussion In this study, results were obtained which showed that XG protects the cellulose in the cellulose/xyloglucan composite and that the activity of Xeg74 with a synergistic mixture of cellulases can easily degrade this material. However, tomato cell walls have a much more complex structure, and additional enzymes in the TFSF crude mixture are essential for tomato cell-wall digestion. Similar results were seen in studies by Vincken et al. [27,28], which showed good synergistic activity on water-unextractable solids from apples (about 24% xyloglucan and 33% cellulose) when Trichoderma viride Endo IV (a family-12 CMCase/xyloglu- canase) was added to a mixture of EXOIII and EndoI. The amount of cellobiose released was twice as high as the amounts released by EXOIII and EndoI alone; however, even the most effective combination was not able to solubilize all of the cellulose. The results of the NMR experiment show that Xeg74 and presumably all family-74 enzymes catalyze hydrolysis with inversion of the anomeric configuration. This is in contrast to the family-12 A. aculeatus xyloglucanase, which is a retaining enzyme [18]. The Xeg74 gene does not have an upstream 14-bp DNA-binding site for the regulatory protein, CelR, which is found in the six T. fusca cellulase genes. This is consistant with our observation that Xeg74 and Cel6A are regulated differently. It is interesting that T. fusca produces Xeg74 during incubation with tamXG, even though it cannot grow on the XGOs released from XG hydrolysis. The amount of enzyme produced in a nongrow- ing culture appears to be equal to or even a little higher than the amount produced in growing cultures, using either Solka Floc or corn fiber as the sole carbon source. Another example of an enzyme produced in a nongrowing culture is a polyester-degrading extracellular hydrolase from T. fusca DSM43793 [29], which is induced in cultures containing Ecoflex TM (BASF AG, Germany), a random co-polyester of 1,4-butanediol, terephthalic acid and adipic acid, as the sole carbon source. It is curious that T. fusca grows on xylose and yet does not metabolize the a- D -xylose in the XGOs produced by Xeg74. There are open-reading frames in the T. fusca genome which are homologous to a-xylosidases; however, neither extracts nor concentrated supernatants Fig. 6. Western blot of an SDS/polyacrylamide gel using rabbit polyclonal antiserum against xyloglucan-specific endo-b-1,4-glucanase (Xeg)74. Supernatants (10 lL) from cultures of Thermobifida fusca grown on glucose (lane 1), cellobiose (lane 2), xylan (lane 3), Solka Floc (lane 4), and corn fiber (lane 5); lane 6, Benchmark molecular mass standards; lane 7, Xeg74 (0.05 lg); lane 8, Xeg74CD (0.05 lg); lane 9, Cel6ACD (0.5 lg); lane 10, Cel6A (0.05 lg); supernatants (10 lL) from cultures of T. fusca grownontamXG,nogrowthwasapparent (lane 11), bacterial microcrystalline cellulose (BMCC) (lane 12), Solka Floc (lane 13). Fig. 5. [ 1 H] NMR spectra of tamarind seed xyloglucan (tamXG) reac- ted with xyloglucan-specific endo-b-1,4-glucanase (Xeg)74 at 50 °C. The a-anomeric proton appears rapidly, indicating an inverting enzyme, while the b-anomeric proton appears later as a result of mutarotation. Ó FEBS 2003 Thermobifida fusca xyloglucanase Xeg74 (Eur. J. Biochem. 270) 3089 from T. fusca cultures grown on Solka Floc, xylan, or corn fiber had measurable activity on p-nitro-phenol a- D -xylose [2]. The first family-74 enzyme to be reported was an A. aculeatus enzyme (FIII avicelase), which produces cello- triose and cellobiitol from reduced cellopentaose, clearly showing that it releases cellobiose from the reducing end [30]. However, the enzyme was not tested on XG and the degree of hydrolysis was only about 0.7%. Thermotoga maritima Cel74 is most active on barley b-glucan, which has a mixed-linkage b-1,3/4 glucose backbone and only 23% as much activity on tamXG [31]. A. niger EglC [32] resembles Xeg74 in having the highest activity on tamXG with about 5% as much activity on CMC or b-glucan and it has a C-terminal family-1 fungal CBM. EglC is regulated by XlnR, a transcriptional activator which binds to the DNA sequence, GGCTAA. We did not find a protein homolog- ous to XlnR in the T. fusca genome, nor did we find GGCTAA upstream of the Xeg74 start codon. The Geotrichum sp. M128 family-74 enzyme is an exoglucanase, which attacks the reducing end of XG, releasing GG, XG, or LG [33]. This protein is unique in having four regions of amino acids (235–251; 310–318; 361–372; 398–413) that are not found in the other family-74 proteins. These may form loops that make the active site a tunnel rather than an open cleft, leading to exoglucanase rather than endoglucanase activity. The nature and presence of a CBM varies among the family-74 proteins: four of 13 family-74 genes code for a family-2 CBM; three have a family-1 fungal CBM; one has a family-3 CBM; and five have no CBM. The low K m of T. fusca Xeg74 shows that the binding of the catalytic site to the substrate is very tight and perhaps this makes the CBM useful but not essential for hydrolytic activity. Overall, the characterized family-74 enzymes show a variety of substrate specificities with most having a strong preference for a cellulosic backbone substituted with xylose (and further decorated) side-chains. The structure of the Xeg74 active cleft should provide interesting insights into its mechanism of action and efforts are underway to solve the structure of Xeg74CD. In summary, Xeg74 is produced at low levels on all substrates, is induced at higher levels by growth on Solka Floc, corn fiber and xylan, and is also highly induced by XG in the media, although additional growth does not take place. The purpose of the enzyme seems to be to degrade XG surrounding cellulose microfibrils to allow the organism access to a substrate which it can hydrolyze and metabolize efficiently. This function is consistent with the presence of a family-2 CBM on the enzyme. Xeg74 may also function to help T. fusca grow on other plant cell-wall polymers, such as xylan or mannan, which may be protected by xyloglucan. Acknowledgements This work was supported by DE-FG02-01ER63150, from the US Department of Energy. References 1. Lynd, L.R., Weimer, P.J., van Zyl, W.H. & Pretorius, I.S. (2002) Microbial cellulose utilization: fundamentals & biotechnology. Mol. Microbiol. Rev. 66, 506–577. 2. Irwin, D., Leathers, T.D., Greene, R.V. & Wilson, D.B. (2003) Corn fiber hydrolysis by Thermobifida fusca extracellular enzymes. Appl. Microbiol. Biotechnol. 61, 352–358. 3. Pauly, M., Albersheim, P., Darvill, A. & York, W.S. (1999) Molecular domains of the cellulose/xyloglucan network in the cell wallsofhigherplants.Plant J. 20, 629–639. 4. Carpita, N.C. & Gibeaut, D.M. (1993) Structural models of pri- mary-cell walls in flowering plants – consistency of molecular structures with the physical properties of the wall during growth. Plant J. 3, 1–30. 5. Fry, S.C., York, W.S., Albersheim, P., Darvill, A., Hayashi, T., Joseleau, J.P., Kato, Y., Lorences, E.P., Maclachlan, G.A., McNeil,M.,Mort,A.J.,Reid,J.S.G.,Seitz,H.U.,Selvendran,R.R., Voragen, A.G.J. & White, A.R. (1993) An unambiguous nomenclature for xyloglucan-derived oligosaccharides. Physio- logia Plantarum 89, 1–3. Fig. 7. Role of Xeg74 in digestion of GBX and tomato cell walls. (A) Digestion of Gluconacetobacter xylinus cellulose (GBG) and cellulose/ XG composite (GBX) pellicles. TFSF, concentrated crude supernatant from a Thermobifida fusca culture grown on Solka Floc; cel mix, Cel5A + Cel6A + Cel9A + Cel48A (molar fractions of 0.15 : 0.35 : 0.15 : 0.35, respectively); 4.7 lg of xyloglucan-specific endo-b-1,4- glucanase (Xeg)74 was added to each reaction tube containing cel mix + Xeg74. The portion of the reducing sugar produced by Xeg74 in these reactions was subtracted from the total reducing sugar produced. (B) Digestion of tomato cell walls as in (A) except that 75 lg of Xeg74 was added to the reaction tubes containing cel mix + Xeg74. 3090 D. C. Irwin et al. (Eur. J. Biochem. 270) Ó FEBS 2003 6. Whitney, S.E.C., Brigham, J.E., Darke, A.H., Reid, J.S.G. & Gidley, M.J. (1995) In vitro assembly of cellulose/xyloglucan net- works: ultrastructural and molecular aspects. Plant J. 8, 491–504. 7. Bellamy, W.D. (1977) Cellulose and lignocellulose digestion by thermophilic actinomycetes for single cell protein production. Dev. Ind. Microbiol. 18 (249–254), 249–254. 8. Irwin, D., Walker, L., Spezio, M. & Wilson, D. (1993) Activity studies of eight purified cellulases: specificity, synergism, and binding domain effects. Biotechnol. Bioengin. 42, 1002–1013. 9. Irwin, D., Jung, E. & Wilson, D.B. (1994) Characterization and sequence of a Thermomonospora fusca xylanase. Appl. Environ. Micrbiol. 60, 763–770. 10. Irwin, D., Shin, D., Zhang, S., Barr, B.K., Sakon, J., Karplus, P.A. & Wilson, D.B. (1998) Roles of the catalytic domain and two cellulose binding domains of Thermomonospora fusca E4 in cel- lulose hydrolysis. J. Bacteriol. 180, 1709–1714. 11. Irwin, D.C., Zhang, S. & Wilson, D.B. (2000) Cloning, expression and characterization of a Family 48 exocellulase, Cel48A, from Thermobifida fusca. Eur. J. Biochem. 267, 4988–4997. 12. Jung, E.D., Lao, G., Irwin, D., Barr, B.K., Benjamin, A. & Wil- son, D.B. (1993) DNA sequences and expression in Streptomyces lividans of an exoglucanase gene and an endoglucanase gene from Thermomonospora fusca. Appl. Environ. Microbiol. 59, 3032–3043. 13. Edwards, M., Dea, I.C.M., Bulpin, P.V. & Reid, J.S.G. (1985) Xyloglucan (amyloid) mobilization in the cotyledons of Tropaeolum majus L. seeds following germination. Planta 163, 133–140. 14. Maclachlan, G. & Brady, C. (1994) Endo-1,4-b-glucanase, xylo- glucanase, and xyloglucan endo-transglycosylase activities versus potential substrates in ripening tomatoes. Plant Physiol. 105, 965–974. 15. York, W.S., van Halbeek, H., Darvill, A.G. & Albersheim, P. (1990) Structural analysis of xyloglucan oligosaccharides by 1H- n.m.r. spectroscopy and fast-atom-bombardment mass spectro- metry. Carbohydr. Res. 200, 9–31. 16. Lever, M. (1972) A new reaction for colorimetric determination of carbohydrates. Anal. Biochem. 47, 273–279. 17. Doner, L.W. & Irwin, P.L. (1992) Assay of reducing end-groups in oligosaccharide homologues with 2,2¢ bicinchoninate. Anal. Biochem. 202, 50–53. 18. Pauly,M.,Andersen,L.N.,Kauppinen,S.,Kofod,L.V.,York, W.S., Albersheim, P. & Darvill, A. (1999) A xyloglucan-specific endo-beta-1,4-glucanase from Aspergillus aculeatus:expression cloning in yeast, purification and characterization of the recombinant enzyme. Glycobiology 9, 93–100. 19. Chirico, W.J. & Brown, R.D. (1985) Separation of [1- 3 H] cello- oligosaccharides by thin-layer chromatography: assay for cellu- lolytic enzymes. Anal. Biochem. 150, 264–272. 20. Rose, J.K., Hadfield, K.A., Labavitch, J.M. & Bennett, A.B. (1998) Temporal sequence of cell wall disassembly in rapidly ripening melon fruit. Plant Physiol. 117, 345–361. 21. Laemmli, U.K. (1970) Cleavage of structural proteins during assembly of the head of bacteriophage T4. Nature (London) 227, 680–685. 22. Zhang, S., Barr, B.K. & Wilson, D.B. (2000) Effects of non- catalytic residue mutations on substrate specificity and ligand binding of Thermobifida fusca endocellulase Cel6A. Eur. J. Bio- chem. 267, 244–252. 23. Vincken, J.P., York, W.S., Beldman, G. & Voragen, A.G. (1997) Two general branching patterns of xyloglucan, XXXG and XXGG. Plant Physiol. 114, 9–13. 24. Davies, G. & Henrissat, B. (1995) Structures and mechanisms of glycosyl hydrolases. Biochem. J. 3, 853–859. 25.Haigler,C.H.,White,A.R.,Brown,J.R.M.&Cooper,K.M. (1982) Alteration of in vivo ribbon assembly by carboxymethyl cellulose and other cellulose derivatives. J. Cell Biol. 94, 64–69. 26. York,W.S.,Kolli,V.S.K.,Orlando,R.,Albersheim,P.&Darvill, A.G. (1996) The structures of arabinoxyloglucans produced by solanaceous plants. Carbohydr. Res. 285, 99–128. 27. Vincken, J.P., Beldman, G. & Voragen, A.G.J. (1994) The effect of xyloglucans on the degradation of cell-wall-embedded cellulose by the combined action of cellobiohydrolase and endoglucanases from Trichoderma viride. Plant Physiol. 104, 99–107. 28. Vincken, J.P., Beldman, G. & Voragen, A.G.J. (1997) Substrate specificity of endoglucanases: what determines xyloglucanase activity?. Carbohydr. Res. 298, 299–310. 29. Gouda, M.K., Kleeberg, I., van den Heuvel, J., Mu ¨ ller, R.J. & Deckwer, W.D. (2002) Production of a polyester degrading extracellular hydrolase from Thermomonospora fusca. Biotechnol. Prog. 18, 927–934. 30. Arai, M., Sakamoto, R. & Murao, S. (1989) Different action by two avicelases from Aspergillus aculeatus. Agric. Biol. Chem. 53, 1411–1412. 31. Chhabra, S.R. & Kelly, R.M. (2002) Biochemical characterization of Thermotoga maritima endoglucanase Cel74 with and without a carbohydrate binding module (CBM). FEBS Lett. 531, 375–380. 32. Hasper, A.A., Dekkers, E., van Mil, M., van de Vondervoort, P.J.I. & de Graaff, L.H. (2002) EglC, a new endoglucanase from Aspergillus niger with major activity towards xyloglucan. Appl. Environ. Microbiol. 68, 1556–1560. 33. Yaoi, K. & Yasushi, M. (2002) Purification, characterization, cloning, and expression of a novel xyloglucan-specific glycosidase, oligoxyloglucan reducing end-specific cellobiohydrolase. J. Biol. Chem. 277, 48276–48281. 34.Vierhuis,E.,York,W.S.,Kolli,V.S.K.,Vincken,J.P.,Schols, H.A., Van Alebeek, G.J.W.M. & Voragen, A.G.J. (2001) Struc- tural analyses of two arabinose containing oligosaccharides derived from olive fruit xyloglucan: XXSG and XLSG. Carbo- hydr. Res. 332, 285–297. Ó FEBS 2003 Thermobifida fusca xyloglucanase Xeg74 (Eur. J. Biochem. 270) 3091 . Cloning, expression and characterization of a family-74 xyloglucanase from Thermobifida fusca Diana C. Irwin, Mark Cheng*, Bosong Xiang†, Jocelyn. sequence (MRMSPRPLRALL GAAAAALVSAAALAFPSQAA) in place of its native signal sequence. Cel 6A, Cel6Acd, and Cel4 8A have been expressed and secreted successfully