The gene for a new type of pullulan hydrolase from the hyperthermophilic archaeon Thermococcus aggregans was cloned and expressed in Escherichia coli. The 2181-bp open reading frame encodes a protein of 727 amino acids. A hypothetical membrane linker region was found to be cleaved during processing in E. coli. The recombinant enzyme was purified 70-fold by heat treatment, affinity and anion exchange chromatography. Optimal activity was detected at 95°C at a broad pH range from 3.5 to 8.5 with an optimum at pH 6.5. More than 35% of enzymatic activity was detected even at 120°C. The enzyme was stable at 90°C for several hours and exhibited a half-life of 2.5 h at 100°C. Unlike all pullulan-hydrolysing enzymes described to date, the enzyme is able to attack α-1,6- as well as α-1,4-glycosidic linkages in pullulan leading to the formation of a mixture of maltotriose, panose, maltose and glucose. The enzyme is also able to degrade starch, amylose and amylopectin forming maltotriose and maltose as main products.
Four types of pullulan-hydrolysing enzymes, based on substrate specificities and reaction products, have been described in the literature: (a) pullulan hydrolase type I (neopullulanase) attacks α-1,4-linkages in pullulan forming panose ; (b) pullulan hydrolase type II (isopullulanase) attacks α-1,4-linkages in pullulan forming isopanose ; (c) pullulanase type I specifically hydrolyses α-1,6-glycosidic linkages in pullulan or branched substrates such as amylopectin forming maltotriose or longer linear products ; and (d) pullulanase type II attacks α-1,6-linkages in pullulan and branched substrates in addition to α-1,4 links in polysaccharides other than pullulan [4,5].
Until now no enzyme system has been detected which is able to attack both α-1,4- and α-1,6-linkages in pullulan forming a mixture of panose and maltotriose. Except the pullulan hydrolase type II all other pullulan-hydrolysing enzymes belong to the glycohydrolases of the α-amylase family showing four characteristic conserved regions . In the case of pullulan hydrolase type I it has been shown that the ability of the enzyme to cleave α-1,6-glycosidic linkages in addition to α-1,4-linkages in polysaccharides other than pullulan is located at the same active site , with two Asp and one Glu being involved in the catalytic process .
In this study we describe the cloning and characterisation of a new type of pullulan hydrolase from the anaerobic, hyperthermophilic archaeon Thermococcus aggregans. To date all described archaeal pullulanases are able to degrade only α-1,6-linkages in pullulan. The newly described extremely thermoactive archaeal enzyme is unique since it is the only enzyme presently known that attacks α-1,4- as well as α-1,6-glycosidic linkages in pullulan and is active above the boiling point of water.
2Materials and methods
2.1Strains and growth conditions
T. aggregans (DSM 10597) was grown under strict anaerobic conditions at 88°C on mineral medium supplemented with starch as described previously . All Escherichia coli strains were grown at 37°C on Luria–Bertani medium with the appropriate antibiotics and expression strains were induced with IPTG. For the preparation of cell mass E. coli was grown in a 7-l membrane fermenter (Bioengineering, Switzerland).
2.2Construction of a T. aggregansλ-phage library
Chromosomal DNA of T. aggregans was isolated according to the method of Ramakrishnan and Adams  and DNA was partially digested with Sau3A. DNA fragments were separated by agarose gel electrophoresis and 4–7-kb fragments were isolated by electroelution and phenol/chloroform extraction. A λ-phage library was constructed using the λZAP express kit (Stratagene, USA). T. aggregans DNA fragments were ligated into the λZAP express vector containing the phagemid pBK-CMV. After in vitro packaging of the ligation products the primary phage library was amplified in E. coli XL1-Blue MRF′. After amplification phagemid pBK-CMV harbouring the insert DNA was excised using helper phage f1. The excised f1 population was transfected and stably established as plasmids in E. coli XLOLR cells.
2.3Screening of the phagemid clones for amylolytic and pullulytic activity
E. coli clones of the plasmid library were plated on LB/Km/IPTG agar plates and incubated overnight. The plates were overlaid with 50 mM Tris buffer pH 6.5 containing dyed substrate (amylopectin or pullulan). Plasmid libraries were incubated for up to 3 days at 80°C and screened for clearing zones around the colonies.
2.4Methodology for subcloning and expression of the ORF2 encoding the pullulan-hydrolysing enzyme
Plasmids of active clones were isolated and the size of the inserts was determined by restriction analysis followed by genotypic determination with the Southern blot technique. DNA sequence was analysed on an ABI automatic sequencer according to the Sanger method. PCR fragments were generated using the Expand HiFi PCR Kit (Roche Diagnostics, Germany).
A PCR fragment encoding a hypothetical pullulan-hydrolysing enzyme was subcloned in the NdeI/BamHI multiple cloning site of the expression vector pET-15b and transformed into E. coli BL21(DE3) pLysS (Novagen). The plasmid harbours an IPTG-inducible T7 promoter/terminator and expresses a fusion protein with an N-terminal hexahistidine tag for affinity purification on Ni-chelate columns.
2.5Enzyme assay and chromatography
Enzyme activity was determined by measuring the amount of reducing sugars liberated when the enzyme was incubated in 50 mM acetate with different polysaccharides . One unit is defined as the amount of enzyme that releases 1 μmol min−1 of reducing sugars under assay conditions. Maltose was used as a standard. Zymogram staining for pullulytic and amylolytic activity was performed according to the method of Furegon et al. . Maltotriose Sepharose affinity matrix was prepared by coupling maltotriose to epoxy-activated Sepharose 6B following the instructions of the manufacturer (Affinity chromatography, Pharmacia Fine Chemicals, Sweden).
2.6Purification of the recombinant enzyme
Purification of the pullulan-hydrolysing enzyme was performed at room temperature. Recombinant cells (7 g) were washed and resuspended in Na-acetate buffer pH 6.5. After sonication the cell debris was removed by centrifugation (30 0000×g, 15 min). The supernatant was heat-treated (80°C, 10 min) and the denatured host proteins were pelleted again by centrifugation. The recombinant protein remained in the supernatant.
The sample was applied to a maltotriose Sepharose column equilibrated with 50 mM Na-acetate buffer pH 6.5. The column was washed in the same buffer and proteins were eluted with 1% pullulan in Na-acetate buffer. Active fractions were pooled and pullulan was removed by dialysis against 50 mM Na-acetate pH 6.5. The protein solution was then applied to a UnoQ anion exchange column (Bio-Rad, USA) equilibrated with Na-acetate buffer. Elution was carried out using the same buffer with a NaCl gradient (10 mM–1 M). Active fractions were pooled and dialysed against the same buffer without salt.
2.7Characterisation of hydrolysis products
The hydrolysis products of various polysaccharides were monitored by high performance liquid chromatography (HPLC) with an Aminex-HPX-42 A column (Bio-Rad, USA). Distilled water was used as mobile phase at a flow rate of 0.3 ml min−1. The purified pullulan-hydrolysing enzyme was incubated in 50 mM acetate buffer pH 6.5 with various substrates (starch, amylose, amylopectin, pullulan, oligosaccharides DP1–DP7, α-, β-, γ-cyclodextrin) at 90°C, aliquots were withdrawn at different time intervals and the reaction was stopped by incubation of the mixture on ice.
Thin layer chromatography (TLC) of mono- and oligosaccharides was performed by an ascending technique on 0.2 mm silica gel-coated aluminium sheets (type, Merck, Darmstadt, Germany) using 1 M DL-lactic acid/acetone/isopropanol (20:40:40, v/v) at room temperature. Carbohydrate spots were visualised by spraying the chromatograms with diphenylamine–aniline reagent (1% (w/v) diphenylamine and 1% (v/v) aniline in acetone, mixed with 0.1 vol. 85% phosphoric acid just before use) and incubating the plates at 140°C for 10 min.
2.8N-terminal analysis of purified proteins and DNA sequencing
N-terminal protein sequencing was carried out by automated Edman degradation with electroblotted samples on PDVF membrane analysed on a Pulsed Liquid protein sequencer (model 473A, Applied Biosystems, USA). DNA sequencing was performed with an ABI sequencer with primer extension in both directions. DNA sequence analysis was done using the DNaid 1.8 program. Multiple alignments were carried out using the Clustal W algorithm . Database search was performed using the BLAST algorithm (http://www.bork.embl-heidelberg.de/Blast2/). The EMBL sequence accession number is AJ25153.
3Results and discussion
3.1Cloning and sequencing of the 4.1-kb insert encoding the pullulan hydrolase from T. aggregans
Thirty thousand phagemid clones obtained from a λZAP (Stratagene, USA) genome bank were screened for hydrolytic activity at 80°C as described in Section 2. Ten clones showed clearing zones on dyed starch and pullulan. Southern blot analysis revealed that all clones harboured the same gene although the fragments varied in size. A 4.1-kb insert was chosen for further analysis.
Complete sequencing revealed a GC content of 42.9% and two open reading frames (ORF, Fig. 1). The first one of 1257 bp encodes a protein of 418 aa with a predicted molecular mass of 47.9 kDa. In BLAST searches this protein showed significant homology to a molybdenum cofactor biosynthesis protein from the hyperthermophilic archaeon Pyrococcus horikoshii, followed by other proteins belonging to the archaeal order Thermococcales, indicating that the cloned fragment is not a contamination.
ORF2 is 2181 bp in length and encodes a protein of 726 aa with a predicted molecular mass of 82.7 kDa. The gene, in the following referred to as pulhA, is flanked by a hypothetical ribosome binding site and an archaeal transcription termination site  (Fig. 2). A signal sequence of 23 amino acids is predicted using the method of Nielsen et al. . The pulhA-encoded protein revealed identities to a large variety of family XIII glycosyl hydrolases . Alignments with these sequences displayed four conserved regions which are characteristic for this group of enzymes (Table 2). The overall homology at the amino acid level was not higher than 30%. The highest homology was found with Bacillus stearothermophilus pullulan hydrolase I . Because of the heterogeneity of this group of enzymes and the low sequence identity no predictions concerning the type of enzyme action could be made on the basis of the primary structure.
Table 2. Conserved regions of pullulan hydrolase type I compared to T. aggregans PulhA
Residues involved in the catalytic site are in bold letters.
3.2Overexpression and purification of pullulan hydrolase
For efficient expression in E. coli ORF2 was subcloned in an IPTG-inducible vector under the control of the T7 promoter (pET-15b; Novagen). The specific activity of the recombinant pullulan hydrolase in the crude extract was 0.85 U mg−1. Denaturation of most of the contaminating E. coli proteins was achieved by treatment of the cell extract at 80°C for 10 min. After heat treatment, affinity and anion exchange chromatography, the enzyme was purified 69.5-fold with a specific activity of 59.1 U mg−1 (Table 1). Samples from the purification steps were analysed by SDS–PAGE and the sample of the MonoQ pool revealed two active protein bands with molecular masses of 67 and 80 kDa. Both proteins co-purified and could not be separated by gel filtration or ion exchange chromatography. Analysing the N-terminal sequences of both enzyme species (VKGEQGL for the 67-kDa and QSPTTQE for the 80-kDa protein) indicated that the smaller enzyme with 67 kDa is an N-terminally truncated variant of the larger enzyme with 80 kDa (Fig. 2). No evidence for proteolysis was provided since the appearance of two molecular species was independent of the addition of various protease inhibitors during purification. It can be speculated that the E. coli system recognises two start codons leading to the translation of two active pullulanases as has been observed with the pullulanase of Fervidobacterium pennivorans.
Table 1. Purification of the recombinant PulhA from E. coli BL21 (DE3) pLysS
Total protein (mg)
Total activity (U)
Specific activity (U mg−1)
E. coli cells (7 g) were resuspended in 15 ml of Na-acetate buffer, pH 6.5 and cells were disrupted by sonication. Cell debris was pelleted by centrifugation (20 000×g, 20 min) and the supernatant was subjected to further treatment.
The purified active proteins, in the following termed PulhA, displayed activity over a broad pH and temperature range with at least 50% activity at pH 3.5 and 8.5 and at 60°C and 120°C. The recombinant enzyme showed highest activity at 90°C and pH 6.5. It displays a remarkable stability with no decrease of activity while incubated for 4 h at 90°C; the half-life at 100°C is 2.5 h. A complete loss of activity was observed after 30 min at 120°C (Fig. 3). The Km value for pullulan is 2.38 mg ml−1 and the Vmax is 16.6 U mg−1; values for starch are 3.72 mg ml−1 and 22.7 U mg−1 and for β-cyclodextrin 1.62 mg ml−1 and 2.8 U mg−1, respectively. Divalent cations such as Zn and Cu inhibited the enzyme activity while Mg, Mn, and Ca had no effect (data not shown).
Using pullulan as substrate the action of the recombinant thermostable enzyme resulted in the formation of DP3, DP2 and DP1 (Fig. 4d, DP: degree of polymerisation). Incubation of this mixture with α-glucosidase from yeast (the enzyme attacks exclusively α-1,4-linkages) converted DP2 completely to DP1 (glucose) indicating that DP2 is maltose (Fig. 4e). Interestingly, only part of DP3 was converted to glucose indicating that the DP3 peak comprises not only maltotriose, which was hydrolysed by α-glucosidase, but also panose. The latter trisaccharide, which contains α-1,6- and α-1,4-linkages, cannot be hydrolysed by the α-glucosidase of yeast. The formation of both panose and maltotriose by PulhA was confirmed by an additional TLC (data not shown). The enzyme system was also able to hydrolyse starch, amylose and amylopectin forming DP1, DP2 and DP3 as major products (Fig. 4a–c). 95% of the substrates were converted within 2 h of incubation at 80°C. Complete conversion was achieved after 24 h. Maltose was formed as a major product. Linear oligosaccharides down to maltotetraose are converted mainly to maltotriose and/or maltose (data not shown). The enzyme is also able to cleave β- and γ-cyclodextrin confirming the endo-acting fashion of the enzyme. Based on the unique action of the enzyme from T. aggregans by attacking α-1,4- as well as α-1,6-glycosidic linkages in pullulan we propose to name the enzyme pullulan hydrolase type III.
3.4Sequence comparison of T. aggregans pullulan hydrolase to other pullulan-degrading enzymes
On comparison with sequence databases, the pullulan hydrolase from T. aggregans has similarities to family XIII glycosyl hydrolases. Although maximum similarity with B. stearothermophilus pullulan hydrolase type I is only 30% the four characteristic regions for this type of enzymes are present. In addition, the three acidic residues, two Asp and one Glu, which are involved in the catalytic sites, are conserved (Table 2).
Interestingly, it has been shown that the substrate specificity of the pullulan hydrolase type I (TVA I) of Thermoactinomyces vulgaris could be altered. Mutagenesis of the sequence AAQY into VANE changed the specificity of the enzyme from pullulan hydrolase type I to pullulanase type II . Comparing the sequence of TVA I of T. vulgaris to the homologous region in pulhA from T. aggregans indicates that the Thermococcus sequence is most probably a hybrid form (type I→AAQY; type III→APQE; type II→VANE). It can be speculated that the alteration of the C-terminus of region II in PulhA can be used to switch the specificity of the enzyme in one or the other direction. Although the gene product shows two active bands, an 80-kDa protein and a 67-kDa N-terminal truncated form, it is unlikely that the two forms have different specificity (one for α-1,4 and the other for α-1,6). The 351 residues long arm of the 80-kDa species does not affect the enzyme specificity since the conserved region II that is responsible for α-1,4 and α-1,6 hydrolytic activity  is located in the C-terminus and there is no conserved sequence known in any described glycosyl hydrolases that can be found in the N-terminal region.
The authors are grateful to the European Commission (EU project ‘Extremophiles as Cell Factories’) and the ‘Fonds der Chemischen Industrie’ for financial support.