Swollenin, a Trichoderma reesei protein with sequence similarity to the plant expansins, exhibits disruption activity on cellulosic materials


  • Enzymes: endo-1,4-glucanase (EC, cellobiohydrolase (EC

  • Note: the complete swo1 sequence data has been submitted to the EMBL database under the accession no. AJ245918.

M. Saloheimo, VTT Biotechnology, PO Box 1500, 02044 VTT, Finland. Fax: + 358 9 455 2103, Tel.: + 358 9 456 5820, E-mail: Markku.Saloheimo@vtt.fi


Plant cell wall proteins called expansins are thought to disrupt hydrogen bonding between cell wall polysaccharides without hydrolyzing them. We describe here a novel gene with sequence similarity to plant expansins, isolated from the cellulolytic fungus Trichoderma reesei. The protein named swollenin has an N-terminal fungal type cellulose binding domain connected by a linker region to the expansin-like domain. The protein also contains regions similar to mammalian fibronectin type III repeats, found for the first time in a fungal protein. The swollenin gene is regulated in a largely similar manner as the T. reesei cellulase genes. The biological role of SWOI was studied by disrupting the swo1 gene from T. reesei. The disruption had no apparent effect on the growth rate on glucose or on different cellulosic carbon sources. Non-stringent Southern hybridization of Trichoderma genomic DNA with swo1 showed the presence of other swollenin-like genes, which could substitute for the loss of SWOI in the disruptant. The swollenin gene was expressed in yeast and Aspergillus niger var. awamori. Activity assays on cotton fibers and filter paper were performed with concentrated SWOI-containing yeast supernatant that disrupted the structure of the cotton fibers without detectable formation of reducing sugars. It also weakened filter paper as assayed by an extensometer. The SWOI protein was purified from A. niger var. awamori culture supernatant and used in an activity assay with Valonia cell walls. It disrupted the structure of the cell walls without producing detectable amounts of reducing sugars.


cellulose binding domain


CBHII, T. reesei cellobiohydrolases


regulatory protein involved incatabolite repression in T. reesei


T. reesei endoglucanases

FnIII repeats

fibronectin III type repeats


T. reesei swollenin I


gene encoding SWOI



In the last few years, a new class of proteins called expansins has been discovered in plants (reviewed in [1–3]). A number of expansin genes have been identified from a wide variety of plant species, including cucumber, Arabidopsis, rice [4] and tomato [5]. The expansins were first implicated in loosening the cell wall structure during plant cell growth (the acid-growth response), and the proteins forming a distinct family with high sequence identity and having this type of activity are now classified as α-expansins [6]. The group 1 pollen allergens have approximately 25% amino-acid identity with α-expansins and have been shown to be active in an acid-growth assay on grass cell walls. Along with their vegetative homologues they are designated as β-expansins [6].

Expansins have been proposed to disrupt hydrogen bonding between cellulose microfibrils or between cellulose and other cell wall polysaccharides without having hydrolytic activity [7,8]. In this way they are thought to allow the sliding of cellulose fibers and enlargement of the cell wall. Purified cucumber expansins have been shown to catalyze extension of isolated plant cell walls such as cucumber hypocotyl walls when assayed using a constant load extensometer [9]. These cucumber expansins have also been shown to weaken filter paper without producing reducing sugar [7]. Some of the α-expansins are functional duringfruit ripening, possibly aiding the action of hydrolytic enzymes that degrade the cell wall polymers [5]. Experiments have been made where expansin activity has been over expressed, or reduced by antisense strategy in Arabidopsis thaliana. The results suggest a role for these proteins in the control of plant growth and morphogenesis [10].

A number of saprophytic and pathogenic fungi and bacteria produce a wide range of enzymes designed to break down plant biomass. These enzymes include cellulases that break down cellulose to glucose, and hemicellulases that degrade the different hemicelluloses to monomeric sugars. For the degradation of the insoluble and complex plant cell wall the microbes produce multiple enzyme forms belonging to various enzyme categories. For example, from the fungus Trichoderma reesei, one of the best known saprophytic microbes, genes have been cloned that encode two exo-acting cellulases liberating mainly cellobiose from cellulose chain ends, five endo-acting cellulases hydrolyzing internal linkages of cellulose chains and 10 hemicellulases representing different enzyme activities [11,12]. Most of the cellulases and some of the hemicellulases of this fungus have a modular structure consisting of a cellulose binding domain (CBD) at either end of the polypeptide chain, connected to the catalytic domain with a linker region. The role of the CBD is to mediate binding of the enzyme to the insoluble substrate.

In addition to plants, a protein with an endoglucanase domain and a domain with sequence similarity to expansins has been reported in the plant pathogen Clavibacter michiganensis ssp. sepedonicus[13]. In this paper, we report the discovery of a novel fungal protein having significant sequence identity to plant expansins. Unlike plant expansins, this protein has a modular structure with an N-terminal CBD. The protein was named swollenin due to its ability to swell cotton fibers without producing detectable amounts of reducing sugars.

Experimental procedures

Strains, vectors and growth conditions

The T. reesei cDNA library in the vector pAJ401 [14] was screened in the yeast strain H1152 (a, sso2-1, leu2-3, trp1-1, ura3-1, sso1::HIS3, M. Aalto, unpublished results) on SC-Ura plates with 2% galactose as the carbon source [15] at the restrictive temperature 31 °C. The yeast strain DBY746 (α, his3Δ, leu2-3, ura3-52, trp1-289, Cyhr) was used for swollenin production. The T. reesei strain QM9414 [16] was used in Northern studies and for detection of the swollenin protein. For the Northern studies the strain was cultivated in shake flasks (28 °C, 200 r.p.m.) in minimal media [17] containing 5% glucose, 2% sorbitol, 2% cellobiose, 2% lactose or 2% Solka floc cellulose for 3 days. Alternatively, the strain was grown on 2% glycerol for 72 h, followed by addition of sophorose (1 mm final concentration). After further 15 h the culture was harvested. A culture grown in a minimal medium with 2% glycerol for 87 h was used as a control.

Nucleic acid methods

Yeast was transformed with the LiAc method [18] or by electroporation (Bio-Rad). Plasmid constructs were made using standard methodology [19]. Total T. reesei RNA was isolated as described [20]. The RNA samples (5 µg) were treated with glyoxal and analyzed in a 1% agarose gel. Northern blotting and hybridization were performed on a Hybond-N nylon membrane (Amersham). T. reesei DNA was isolated as described [21]. Stringent Southern hybridization was performed as described [19]. Nonstringent Southern hybridization was performed in a hybridization mixture without formamide [19] at 48 °C and the filter was washed in 2 × NaCl/Cit, 0.1% SDS for 10 min at room temperature and for 30 min at 48 °C.

Antibodies and Westerns

Swollenin antibodies were generated in rabbits by immunizing with the peptide CDPNYTSSRPQERYGS (amino acids 422–437 in the swollenin sequence). SDS gel electrophoresis was performed as described [22] and Western blotting was performed on a nitrocellulose membrane (Schleicher & Schull) and detection of the swollenin with asecondary antibody-alkaline phosphatase conjugate (Bio-Rad). Samples of yeast or T. reesei supernatants and purified SWOI were denatured for Endoglycosidase H treatment by heating 10 min at 100 °C in 0.5% SDS, 1% 2-mercaptoethanol and subsequently treated with Endo Hf (New England Biolabs) for 3 h at 37 °C in the G5 buffer (50 mm Na-citrate, pH 5.5). 1000 units of EndoHf was used per 20 µg of purified SWOI and per 20 µL of the T. reesei or concentrated yeast supernatants.

Swo1 gene disruption

The swo1 gene was disrupted from the T. reesei strain QM9414 [16] by replacing it with a hygromycin resistance cassette. The genomic swo1 gene was first subcloned from a cosmid library clone into pBluescript SK as a 5.5 kb EcoRV fragment to obtain the plasmid pSH1. Most of the swollenin-coding region was replaced from pSH1 by digesting it with NarI and BstEII and ligating with the hygromycin resistance cassette consisting of the Aspergillus nidulans gpdA promoter and trpC terminator and the E. coli hygromycin resistance gene derived from the plasmid pBluekan (from P.J. Punt, TNO Nutrition and Food Research, Zeist, the Netherlands). The resulting plasmid, pSH9 was digested with EcoRV and transformed into QM9414 as described [17], and transformants were selected on 100 µg·mL−1 hygromycin and purified to uninuclear clones by plating single spores on selective medium. Disruptants of the swo1 gene were screened among the transformants by Southern hybridization performed as described [19]. Two disruptants obtained were also examined by growing them in shake flasks (28 °C, 200 r.p.m., 5 days) in a medium with 3% whey and 1.5% complex grain-based nitrogen source [23] and performing Western analysis from their culture supernatants as described above. The phenotype of the disruption was studied by plating single spores on plates with minimal medium [17] supplemented with 0, 0.1 or 0.2% proteose peptone and either 2% glucose, 2% Solka floc cellulose, 2% Avicel cellulose or 2% complex grain-based carbon/nitrogen source [23]. An additional test was made on plates where the minimal medium with or without peptone and without any carbon source had been overlaid by a Whatman 1 filter paper disc. Growth of colonies of the swo1 disruptants and the parental strain was followed daily.

Production and characterization of the swollenin preparations

The yeast strain DBY746 harbouring a plasmid with swo1 in the vector pAJ401 [14] or the vector alone were grown in Chemap CMF mini 1 L or Biolafitte 14 L bioreactors. The bioreactor medium was SC-Ura with 2% glucose as the carbon source [15]. The yeast supernatants were concentrated 20 times with a Centiprep concentrator (Amicon) for the treatments of cotton fibers and filter paper. The amount of SWOI was estimated by comparing signals obtained in Westerns from Endo-H-treated yeast supernatants with signals obtained with known amounts of purified Endo-H treated SWOI from A. niger var. awamori.

The 1.5 kb coding region of the swollenin cDNA clone was amplified by PCR using the following primers which were designed to add BglII and XbaI restriction endonuclease sites to the 5′ and 3′ ends, respectively. Primer ExAspBgl2: 5′-CATTAGATCTCAGCAATGGCTGGTAAGCTTATCCTC-3′. Primer ExAspXba1:

5′-CGACTCTAGAAGGATTAGTTCTGGCTAAACTGCACACC-3′. The DNA sequence of the amplified product was verified and the swollenin coding region was inserted into the BglI and XbaI sites between the glaA promoter and terminator of an Aspergillus expression vector (pGAPT-PG) to produce pGAPT-exp. The pGAPT-PG vector consists of pUC18 containing the A. nidulans pyrG gene as selectable marker and a 1.1 kb fragment of the A. niger var. awamori glaA promoter and a 0.2 kb fragment of an A. niger glaA terminator.

The expression plasmid pGAPT-exp was transformed into A. niger var. awamori strain dgr246 P2 as described [24]. Transformants were selected for their ability to grow on minimal medium lacking uridine. For swollenin protein production the transformants were grown in liquid medium as described [25]. Cells were removed and the culture supernatants were equilibrated with 1 m ammonium sulphate, 100 mm Tris pH 7. The supernantant was then applied to a cellulose (Sigma, St. Louis, MO, USA) affinity column and washed with 1 m ammonium sulphate, 100 mm Tris pH 7 to remove unbound proteins. The purified swollenin was eluted as a single peak in water.

The purified swollenin was tested for activity against hydroxyethylcellulose (HEC), β-glucan, xylan and mannan. The enzymatic activities against HEC (Fluka) and barley β-glucan (Biocon) were determined according to IUPAC [26] and against birch xylan (Roth) as presented [27]. Mannanase activity was assayed according to the procedure of IUPAC (1987) but using 0.5% locust bean gum (Sigma) as a substrate.

Action on solid substrates

Yeast supernatants.  Cotton fibers were mercerized by treating them with 25% NaOH for 15 min at 5 °C and washing several times with distilled water. The cotton fibers were suspended in the concentration of 0.5 g·mL−1 in 50 mm sodium acetate, pH 5.0 containing 1/4 of the concentrated yeast culture media from the swollenin producing yeast and control strain. Additionally, the purified T. reesei EG II, CBH I and cellulose binding domain (CBD) of CBH I at a concentration of 5 µg·mL−1 were used as controls for the swollenin [28,29]. After incubation for 4 h at 25 °C, the suspended fibers were filtered off and the amount of reducing sugars released into the filtrates was determined as described in [30]. The fibers were rinsed once with buffer and then suspended in distilled water with glass beads prior to sonication for one minute using a probe tip sonicator (Vibra Cell Sonics and Materials Inc.) The fibers were then stained and visualized by light microscopy to determine gross effects on their structure.

For the paper strength test, Whatman no. 3 filter paper was cut into strips measuring 7 × 2 cm. Sodium acetate buffer (50 mm, pH 5) was used for all of these experiments. The concentrated yeast samples were sometimes first desalted by passage through a Bio-Rad Econo-Pac10 DG column with a molecular mass cut-off of 6000 Da. After desalting, 5 mL of the yeast samples were added to 4 mL of buffer in 50 mL disposable conical tubes and the Whatman strips were added. At the same time, strips were added to buffer alone and 8 m urea in buffer. After incubating at room temperature for 15 min the strips were measured for their wet tensile strength. The assay was performed by placing each wet strip of paper between the Thwing–Albert tensile tester (Model 5564 from Instron Corporation, Canton, MA, USA) clamps spaced 4.5 cm apart. A 250 lb load cell was used. Test speed was 0.1 cm·min−1, and the peak load was measured before breaking; it typically only took a minute to reach the paper breaking load.

The purified swollenin.  The action of the purified swollenin preparation on plant cell wall material was followed using Valonia cell wall fragments as substrate. Vesicles of Valonia macrophysa were purified as described previously [31] and cut to small pieces, 4–5 mg each (dry weight). These cell wall fragments were suspended in 50 mm acetate buffer (pH 5.0) and swollenin was added at the dosages of 10 µg·mg−1 and 100 µg·mg−1. Treatments with the purified T. reesei CBH I (also known as Cel7A) and EG II (also known as Cel5A) were performed as comparison for swollenin. The samples were incubated at +45 °C under stirring for 48 h and thereafter examined under a stereo microscope (Leica, Wild M10). The control sample was treated alike but omitting the enzymes and swollenin. In addition the filtrates were analysed for solubilized sugars by HPLC.


Swollenin has sequence similarity with plant expansins. The swollenin cDNA was isolated in a screening where components of the Trichoderma secretory pathway were searched for by yeast complementation. The sso2 temperature-sensitive S. cerevisiae strain was transformed with a T. reesei cDNA expression library, and cDNAs derived from clones able to grow at the restrictive temperature were sequenced. One of them encoded a protein predicted to have an N-terminal signal sequence followed by a cellulose binding domain. A major part of the remaining sequence was found to have sequence similarity with plant expansins in a BLAST database search (Fig. 1C). Based on its swelling activity on cotton fibers (see below) the protein was named swollenin (SWOI) and the gene swo1.

Figure 1.

The basic structure of the swollenin protein(A), alignment of the swollenin CBD with the CBD sequences of T. reesei cellulases(B), Alignment of the swollenin with two α-expansins(C) and The alignment of swollenin(SWOI) with FnIII repeat sequences(D). In (A), vertical arrows indicate the intron positions. SS, signal sequence; CBD, cellulose binding domain, QQ, two glutamines suggested to form the N-terminus of the mature swollenin. The numbering refers to amino-acid positions in the mature protein. In (B), the invariant amino acids are boxed and amino acids forming the flat surface interacting with cellulose are shaded. Lines below the sequences show the proposed disulphide bridges of the SWOI CBD. (C) Alignment of the swollenin with two α-expansins, LeEx1 of tomato (5) and CuExS2 of cucumber (4). Invariant amino acids are shown by asterisks and conservative substitutions by dots. The regions best conserved between α- and β-expansins are underlined in CuExS2. The conserved cysteines are shown by arrows and positions with conserved aromatic amino acids in all three sequences by + 0. The regions with homology to the FnIII repeats in titin are in bold in SWOI. (D) The alignment of swollenin (SWOI) with FnIII repeat sequences of human titin and a consensus sequence of the bacterial FnIII repeats (BACT). The amino acids strictly conserved between the bacterial and mammalian FnIII repeats are shown by asterisks.

The genomic copy of the swo1 cDNA was isolated from a cosmid library, subcloned and sequenced. The gene contains five short introns (Fig. 1A). The promoter contains a putative TATA box 90 bp and a putative binding sequence of the glucose repressor protein CREI [32] 117 bp upstream from the translation start codon (data not shown).

The putative swollenin protein starts by a typical signal sequence. It is followed by two glutamic acids. In the T. reesei cellulases EGI (also known as Cel7B) [33] and EGII [34] there are two glutamic acids at the N-terminus and in CBHI there is one [35], and the N-termini of these enzymes are blocked by a pyroglutamic acid residue. By analogy, it is suggested that the swollenin signal sequence would be 18 amino acids in length and be cleaved before the two glutamines in the sequence (Fig. 1b). The SWOI has three potential N-glycosylation sites at positions 160, 336 and 406.

The swollenin cellulose binding domain (CBD) has the typical sequence features of fungal CBDs. The amino acids invariant in the CBDs of the T. reesei cellulases are conserved in swollenin (Fig. 1B). In the NMR structure solved from the CBHI CBD there are two disulphide bridges [36]. Based on a close proximity of an additional cysteine pair in the modelled structures of the CBHII and EGI CBDs it has been suggested that they would have a third disulphide bond [37]. The swollenin CBD has six cysteines at positions conserved with those of the CBHII (also known as Cel6A) CBD and thus it probably has three disulphide bridges as well. The residues forming the flat surface binding to cellulose are conserved in the swollenin CBD with one exception (Fig. 1B). Position 8 in the swollenin sequence has a phenylalanine while the other cellulases have tyrosine or tryptophan. The linker region of the swollenin is rich in serines and threonines and is expected to be heavily O-glycosylated [38]. Without protein structure data the length of the linker cannot be determined unambiguously. However, the region rich in serines and threonines in SWOI apears to be among the longest found in T. reesei enzymes, approximately 50 amino acids. The region of SWOI between the putative linker and the expansin-like area shown in Fig. 1c does not match with any sequences in databases.

The C-terminal two-thirds of the swollenin show clear amino-acid similarity with plant expansins (Fig. 1C). The identity between swollenin and individual α- or β-expansins in pairwise comparisons is about 25% over an area of about 200 amino acids. The alignment of the swollenin with two expansin sequences (Fig. 1C) suggests that two large insertions have occurred in the swollenin gene in the N-terminal half of its expansin-like domain. The identity between the α- and β-expansins is 20–25% and there are five sequence elements that are well conserved between the expansin categories [6]. Four of the elements form the best conserved parts between swollenin and the expansins and thus they are probably functionally important. The N-terminal half of the expansins contains eight conserved cysteines with a spacing similar to that of cysteines in the chitin binding domain of wheat germ agglutinin [4]. Seven of these cysteines are conserved in swollenin. Aromatic amino acids are often important in the interaction of enzymes and their carbohydrate substrates. In the alignment between swollenin and expansins there are eight positions where an aromatic amino acid is conserved (Fig. 1C). Alignments of swollenin with individual expansins suggestthat it is better conserved with β-expansins than α-expansins.

There are two short sequences in swollenin that show relatively strong conservation with fibronectin type III (FnIII) repeats of mammalian titin proteins (Fig. 1D). Interestingly, such repeats have been found in prokaryotic hydrolases such as cellulases, chitinases and amylases [39,40] but thus far not in fungal enzymes. The amino acids invariant between the bacterial hydrolases and mammalian FnIII repeats are conserved in swollenin. Unlike the continuous FnIII repeats in the bacterial enzymes, the region with similarity to titin in swollenin is divided into two parts about 170 amino acids apart.

Regulation of the swollenin gene

The cellulase and hemicellulase genes of Trichoderma reesei are regulated by the carbon source [12,41] and thus it was of interest to analyze if the swollenin gene is regulated in a similar manner. The T. reesei strain QM 9414 was grown in shake flasks on different carbon sources and Northern hybridization was performed. The role of sophorose, a strong cellulase inducer, in the swollenin gene regulation was studied by adding it to a Trichoderma culture grown on the neutral carbon source glycerol. The swo1 mRNA level was undetectable in the glucose culture sample in a short exposure (Fig. 2, lane 1), but in a long exposure a very low level was observed (lane 9). In a late stage of a glucose cultivation the swo1 gene was derepressed (lane 2). In sorbitol (lane 3) and glycerol samples (lane 4) a low mRNA level was present, and when sophorose was added to the glycerol culture, strong induction of swo1 occurred (lane 5). In media with lactose and cellobiose the swo1 mRNA level was moderate and in a medium with cellulose it was at its highest.

Figure 2.

Regulation of the swo1 gene according to the carbon source. Northern analysis of RNA samples isolated from mycelia grown in a medium with (1) glucose (2) glucose, sample taken after glucose depletion (3) sorbitol (4) glycerol (5) glycerol, induced by sophorose (6) cellobiose (7) lactose (8) Solka floc cellulose (9) glucose, 7 times longer film exposure. The gpd1 probing serves as a loading control.

Production of SWOI by T. reesei

Polyclonal antibodies against SWOI were obtained by immunizing rabbits with a synthetic peptide designed based on the swollenin sequence. The expected molecular mass of the deduced SWOI protein is 49 kDa, but the antibodies recognize in a Western a protein of approximately 75 kDa in a T. reesei supernatant from a cellulose-based culture (Fig. 3, lanes 1 and 3). The difference between the calculated and observed molecular masses can not be explained by N-glycosylation, because endoglycosidase H that removes N-glycans changes the apparent molecular mass of SWOI only slightly (Fig. 3, lane 4). Also the SWOI produced in yeast and A. niger var. awamori gained a molecular mass close to 75 kDa when they were treated withendoglycosidase H (see below). This band was also absent from the supernatants of the swo1 disruptants (Fig. 3). These data show that the 75 kDa band is indeed derived from the SWOI protein. The75 kDa band could not be observed in a T. reesei culture filtrate from a culture grown on glucose (Fig. 3, lane 2). Thus the very low basal expression detected at the mRNA level (Fig. 2) was undetectable in the Western analysis. As estimated from Western blotting, the production level of SWOI in the T. reesei culture analysed was about 1 mg·L−1. This is by far less than the production levels of the major cellulases.

Figure 3.

Western detection of SWOI by polyclonal antibodies from culture supernatants of T. reesei and yeast. Lane 1, T. reesei culture performed in a medium inducing cellulase production (1 µL of culture medium loaded); lane 2, T. reesei culture performed in a medium with glucose as the carbon source (15 µL; lane 3, T. reesei culture supernatant (10 µg of total protein); lane 4, T. reesei culture supernatant treated with endoglycosidase H (3 µg of total protein); lane 5, yeast strain expressing swo1 cDNA (the sample corresponds to 0.5 mL of supernatant); lane 6, yeast strain expressing swo1 cDNA, treated with endoglycosidase H (0.2 mL of original supernatant); lane 7, yeast strain carrying the expression vector alone (0.5 mL of original supernatant); lanes 8, 9, swo1 disruptant T. reesei strains (15 µL of culture supernatant); lane 10, the parental strain of the swo1 disruptants (15 µL of culture supernatant); lane 11, Western detection of SWOI purified from A. niger var. awamori (100 ng); lane 12, Western detection of purified SWOI, endoglycosidase H treated (30 ng); lane 13, Coomassie-stained SWOI purified from A. niger var. awamori (3 µg).

Disruption of the swo1 gene

The swo1 gene was disrupted from the genome of T. reesei by replacing it with a hygromycin resistance cassette. Two disruptants were shown by Southern analysis to be single-copy transformants where the gene replacement had occurred (data not shown). Western analysis of their culture supernatants further confirmed that they do not produce the SWOI protein (Fig. 3, lanes 8 and 9).

We attempted to demonstrate the phenotype of the swo1 disruption, its effect either on the formation of the T. reesei cell wall and growth of the fungal mycelium or on the degradation of cellulosic carbon sources by the fungus. This was performed by comparing the growth rates of the disruptants and the parental strain on plates having glucose or different cellulosic compounds as carbon sources. The compounds tested were two commercial celluloses, filter paper and a complex grain-based carbon/nitrogen source [23]. No significant differences in the growth rates could be observed between the strains on any of the carbon sources and thus swo1 disruption had no apparent phenotype in our experiments.

Non-stringent hybridization of T. reesei genomic DNA was performed with a swo1 gene fragment encoding the expansin-like domain as a probe. Hybridization at 48 °C revealed several other bands in addition to the ones originating from swo1, suggesting that there are other genes having expansin-like domains present in the T. reesei genome in addition to swo1(Fig. 4). The presence of these genes could compensate the lack of swo1 in the disruptants and thus explain the result of the disruption experiment.

Figure 4.

Southern hybridization of T. reesei genomic DNA with the region encoding the expansin-like domain from the swo1 gene, performed at stringent(68 °C) and nonstringent(48 °C) conditions. The restriction enzymes used are indicated.

Characterization of the swollenin preparations

When the swollenin cDNA was expressed in S. cerevisiae under the PGK1 promoter in a multicopy plasmid, Western analysis of bioreactor culture supernatants showed that a heterogeneous high molecular mass protein reacting with the swollenin antibodies was produced by the yeast (Fig. 3, lane 5). In many instances it has been shown that yeast tends to overglycosylate heterologous proteins, e.g. the T. reesei cellulases CBHI and CBHII [42]. When the swollenin produced in yeast was treated by endoglycosidase H to remove N-glycans, it gained an apparent molecular mass close to the swollenin produced by Trichoderma (Fig. 3, lane 6). The production level of SWOI in yeast wasapproximately 25 µg·L−1 as estimated from Western blotting experiments.

Swollenin was also produced in A. niger var. awamori and after a single step purification procedure the purified swollenin protein was obtained for biochemical characterization. The SWOI expressed in this host migrated as two relatively diffuse bands with apparent molecular masses between 80 and 95 kDa (Fig. 3, lanes 11 and 13), and Endo-H treatment reduced the molecular mass to the same level as SWOI produced by T. reesei (Fig. 3, lane 12). Thus A. niger var. awamori slightly overglycosylated SWOI. Activities of purified swollenin against hydroxyethyl cellulose (HEC), β-glucan, xylan and mannan were measured and the results are shown in Table 1. Minor hydrolytic activity on β-glucan, xylan and mannan, but not on HEC, wasobserved for the purified swollenin protein expressed in A. niger.

Table 1. Characteristics of the swollenin preparation purified from A. niger var. awamori.
 Protein (mg·mL−1)aActivity (nkat·mL−1)
a Lowry protein.

Demonstration of the swollenin activity on solid substrates.

The swollenin expressed in yeast.  The activity of the swollenin produced in yeast towards cellulosic materials was shown by treatments of cotton fibers and filter paper. Cotton fibers were incubated with concentrated yeast supernatants from bioreactor cultivations of the SWOI-producing yeast (approximately 0.125 µg·mL−1 of SWOI) and the control strain with vector alone or, as controls, with the T. reesei cellulases EGII, CBHI (5 µg·ml−1) and the cellulose binding domain of CBHI. After the treatment the fibers were removed from the reaction mixture by filtering, rinsed, sonicated with glass beads, stained and analyzed by light microscopy. Soluble reducing sugars were measured from the reaction mixture. Treatment with the control yeast supernatant did not change the fiber structure (Fig. 5A). The supernatant of the yeast strain producing swollenin caused local disruption of the fiber structure that became visible only after sonication. This was seen as swollen areas occurring along the fibers (Fig. 5B). CBH I caused some light fibrillation of the fibers (Fig. 5C), whereas the treatment with EG II resulted in damaged and rugged outlook of the fibers accompanied by fiber cutting (Fig. 5E). Din and coworkers [43] have reported on disruption of cellulosic fibers by a bacterial cellulose binding domain. No modification of fiber surface could be detected in our experiment with the fungal (CBHI) CBD by the light microscopical method used (Fig. 5D), suggesting that the effect of SWOI on the fibers is not caused by its CBD. The treatment of the cotton fibers with the yeast supernatants or with the CBD did not release detectable amounts of reducing sugars. In contrast, the filtrates from the CBHI- and EGII-treated fibers contained 0.08% and 1.61% reducing sugars of the original dry mass, respectively.

Figure 5.

Light microscopy of cotton fibers treated with culture supernatant of control yeast with the vector alone(A), supernatant of swollenin-producing yeast (approximately 0.125 µg·L−1 of SWOI)(B), and isolated T. reesei cellulases CBHI(C) and EGII(E) and the cellulose binding domain of CBHI(D)(5 µg·mL−1). The swollen areas caused by SWOI treatment are shown by arrows. Small fibrils can be seen on the surface of fibers treated by CBHI. EGII caused modification of the fibers that can be seen as rugged surface outlook.

To test the effect of swollenin on paper, filter paper strips were incubated in concentrated culture supernatants of the swollenin-producing and control yeast strains and measured for their wet tensile strength. The data shows how much load each strip of paper could hold before it broke; breakage at a lower mass indicates less tensile strength (Table 2). The average load is only slightly decreased when broth from yeast which does not contain the swollenin gene is used. However, the same amount of broth from the yeast expressing the swollenin gene results in a 15–20% decrease in the average load compared to the control broth. Incubation in 8 m urea decreases the average load the paper can bear by about 40% compared to buffer alone.

Table 2. The average peak load a strip of filter paper could bear before breakage. The filter paper strips were treated with buffer, yeast culture supernatants or urea as indicated. The results are the average of 3 or 4 readings.
SampleAverage Peak Load (g)SD
Control yeast22510.4
Swollenin-producing yeast186 5.0
m Urea13216.2

The purified swollenin.  Fragments of Valonia cell walls were used as the solid substrate for studies on the mode of action of the purified swollenin. This algal cell wall is made of highly crystalline cellulose with a layered structure as shown in Fig. 6A. Fragments of the cell wall were incubated individually with the purified cellulases, CBH I and EG II, and swollenin and alterations in the cell wall structure were followed microscopically. The cellulases modified the cell wall fragments with a concomitant release of soluble sugars (Table 3). Treatment with EG II disrupted totally the cell wall structure resulting in a milky solution whereas with CBH I disintegration of cell wall to fibrils was observed (Fig. 6). The action of swollenin resembled that of CBH I, but integrity of the cell wall was partially retained and no soluble sugars were released.

Figure 6.

Atomic force microscopy image of the structure of the Valonia cell wall(A), and light microscopy of Valonia cell wall fragments after treatment with buffer alone (B), SWOI (C, 10 µg·mg−1), CBHI (D, 100 µg·mg−1) and EGII (E, 100 µg·mg−1).

Table 3. Disintegration of Valonia cell walls by the purified swollenin and T. reesei cellulases. The treatments were performed at a consistency of 0.25%, at + 25 °C for 48 h.
TreatmentDosage (µg·g−1)Effects on cell wallSolubilized sugars (% of dw)
Swollenin10Partial disintegration to fibrils0
CBH I10Total disintegration to fibrils0.09
EG II10Total disintegration to milky solution9.2


T. reesei produces one of the most powerful mixtures of extracellular enzymes for efficient hydrolysis of the plant polysaccharides cellulose and hemicellulose. This fungus has served as a model, and extensive studies on the biochemistry, genetics, regulation, structure-function relationships and applications of T. reesei enzymes have been carried out [44]. The discovery of the expansin-like protein SWOI in T. reesei provides new insight to the mechanism of microbial lignocellulose degradation, together with the report on an endoglucanase with an expansin-like domain in a pathogenic bacterium [13] and the discovery of sequence similarity of expansins with family 45 glycosyl hydrolases (see below).

Similarly to plant expansins, filter paper was shown to be weakened by SWOI in an extensometer assay. We also demonstrate that the structure of mercerized cotton fibers was changed upon swollenin treatment in a manner clearly visible by light microscopy. It can be assumed that the swollen areas of the cotton fibers appearing after swollenin treatment coincide with the tilt/twist areas of cotton fibers, where the structure of cellulose is less ordered and more accessible for modification than in crystalline regions [45]. Both cotton and filter paper consist of relatively pure cellulose, and therefore swollenin would appear to be able to open the crosslinking of cellulose fibers. No reducing sugar formation was detected in the cotton swelling test, which is in accordance with the plant expansin results published. Disruption activity that was detected upon treatment of Valonia cell wall frgaments with purified SWOI is in line with the results obtained with yeast supernatants containing SWOI. Although the Valonia cell wall are not representative of the higher plant cell walls, the ability of SWOI to disrupt the Valonia cell wall without producing reducing sugars is of special interest. Activity of this type has been reported for the expansins. The SWOI preparate purified from A. niger var. awamori had a slight activity towards β-glucan, mannan and xylan, but no activity towards hydroxyethyl cellulose. The detected enzyme activities were very low, e.g. the specific activity of the T. reesei endoglucanases EGI or EGII against β-glucan are thousands of nkat·mg−1, whereas the SWOI preparate had an activity of 79 nkat·mg−1. At present we can not be sure whether the activities observed in the SWOI preparation are due to trace amounts of contaminating A. niger var. awamori enzyme(s) or to a weak hydrolytic activity of the SWOI protein itself. However, the disruption ability of solid substrate structures observed in this work is most probably not due to hydrolytic activity of SWOI, as no reducing sugar release from the solid substrates was detected in these activity tests.

It has been reported that expansins have limited sequence similarity with the family 45 of glycosyl hydrolases, which includes the T. reesei endoglucanase EGV [3,14,46]. This similarity is in the same range in identity percentages as the similarity in an alignment between swollenin and individual expansins, but it is limited to a smaller area (data not shown). The sequence conservation between EGV and SWOI is hardly detectable and thus it is weaker than conservation between EGV and expansins. The sequence motif HFD forming a part of the active site of the family 45 hydrolases [47] is conserved in the expansins, and according to the alignment in Fig. 1C would appear to be replaced by HLD in SWOI. The degree of conservation between swollenin and the expansin-like domain of celA from Clavibacter michiganensis is lower than conservation between swollenin and plant expansins (data not shown).

In a recent report the β-expansin of Phleum pratense was shown to have proteinase activity and to have limited sequence similarity to papain-type proteinases [48]. The authors proposed that expansins loosen the plant cell wall structure by cleaving cell wall proteins that crosslink cellulose fibers together rather than by disrupting hydrogen bonding between fibers. The regions around the three active site residues of papain were suggested to be conserved in both α- and β-expansins. Some amino-acid similarity between papain and swollenin can be detected at one of these regions (around Cys256 of swollenin) but the others are not conserved.

An interesting feature of the T. reesei swollenin is that it has a modular structure typical of fungal cellulases and some hemicellulases. SWOI has an N-terminal cellulose binding domain (CBD) that is very well conserved with other fungal CBDs. Thus it can be expected that its function is to bind the SWOI protein to cellulosic compounds. An other interesting feature, although much less clear in its functional importance, is the sequence similarity to the fibronectin III (FnIII) type repeats of mammalian titin proteins (Fig. 1D). The FnIII repeats of titin form β sandwich domains that have been suggested to be able to unfold and refold easily [49] and this would make the protein able to stretch. The ability to stretch might be important for swollenin, if its function is to allow slippage of cellulose microfibrils in plant cell walls as suggested for expansins.

Our results suggest that swollenin is a component of the enzyme mixture produced by the fungus which is needed for degradation of plant biomass and not, e.g. in modifying the Trichoderma cell wall during the growth of the fungus. The regulation pattern of the T. reesei swo1 gene is highly reminiscent to that of the cellulase genes of this fungus [41]. The gene is induced for instance by plant materials and certain oligosaccharides. The swo1 gene has a low expression level on glucose, sorbitol and glycerol unlike the more tightly repressed major cellulases [41]. This could imply to the interesting possibility that swollenin would be among the enzymes that, before the onset of massive cellulose degradation, aid in liberating a soluble inducer when the fungus is encountering the insoluble cellulosic substrate. This soluble inducer would further induce the main cellulolytic machinery.

According to early theories on cellulose degradation, the cellulase system of fungi like T. reesei would comprise two kinds of activities. It was suggested that C1 (‘swelling factor’), a nonhydrolytic component would be needed to make the substrate more accessible to Cx, the hydrolytic component consisting of the endo- and exo-acting enzymes and β-glucosidases that degrade the substrate to glucose [50]. A large number of hydrolytic enzymes have been characterized but so far the C1 factor has remained unsolved. The Trichoderma C1 has not been well characterized, but based on gel filtration it has been reported to have a molecular mass of 61 kDa [51], not far from 75 kDa that was estimated by SDS gels to be the molecular mass of SWO1. Based on the properties of swollenin shown in this work, it provides a possible candidate for a component of C1. Our results also point towards the existence of other proteins with sequence similarity to SWOI in T. reesei (Fig. 4). Thus it is possible that there exist several swollenin-like activities as is the case with the hydrolytic enzymes, which vary somewhat in their modes of action but all contribute synergistically to the efficient hydrolysis of the plant polysaccharides.


We wish to thank Riitta Nurmi and Kati Uotila for excellent technical assistance. The work was supported by the Finnish National Technology Agency (Tekes).