Cellulosic substrates and cellulases
The cellulosic substrates Whatman filter paper No.1, α-cellulose, Avicel PH101 and Sigmacell 101 were purchased from Sigma-Aldrich (MO, USA). Physical properties and product information have been summarized by various authors [10, 64]. Agglomerates of Whatman filter paper No.1 were prepared by using a hole-punch and quartering the resulting filter paper discs. The final filter paper agglomerates had an average diameter of approximately 3 mm. The cellulase preparation Celluclast® 1.5 L (Novozymes, Bagsværd, DK) - a filtrated culture supernatant of T. reesei  - was used for the hydrolysis of the pretreated cellulosic substrates. According to various authors, Celluclast® contains CBHs (Cel7A and Cel6A), EGs (for example, Cel7B and Cel5A) as well as β-glucosidases [81, 82]. To remove salts, sugars and other interfering components, Celluclast® was previously rebuffered with an Äkta FPLC (GE Healthcare, Little Chalfont, UK). Celluclast® was loaded on Sephadex G-25 Fine (2.6 cm × 10 cm, GE Healthcare), and 0.05 M sodium acetate (pH 4.8) was used as a running buffer at 110 cm/h. Sephadex G-25 Fine exhibits an exclusion limit of 1-5 kDa which is comparable to the molecular mass cut-off of dialysis membranes for protein desalting. Since cellulases have a molecular mass of > 25 kDa [62, 83, 84], the mixture of cellulases was not changed during rebuffering. Chromatography was conducted at room temperature, and the automatically collected fractions were directly cooled at 4°C. To determine specific filter paper activities, different dilutions of Celluclast® and the rebuffered Celluclast® - applied for all hydrolysis experiments - were tested according to Ghose . Here, the following specific filter paper activities (per g protein) were measured: 201 U/g (Celluclast®) and 279 U/g (rebuffered Celluclast®).
Genetic engineering for recombinant swollenin
The below-mentioned cloning procedure was designed for secreted protein expression according to the K. lactis Protein Expression Kit (New England Biolabs, MA, USA). The cDNA of the swollenin-coding region was synthesized by reverse-transcription PCR using mRNA isolated from T. reesei QM9414 (swo1 gene [GenBank: AJ245918], protein sequence [GenBank: CAB92328]) and reverse transcriptase (M-MLV, Promega, WI, USA) according to the manufacturer's protocol. Specific primers were applied to synthesize a cDNA starting from the 19th codon of the swollenin-coding region and, therefore, missing the secretion signal sequence of T. reesei . By using the aforementioned primers, SalI and SpeI restriction sites were added upstream and downstream of the swollenin-coding region, respectively. The amplified cDNA was cloned into the pCR2.1-TOPO vector (Invitrogen, CA, USA) according to the manufacturer's protocol. After DNA sequencing and isolation of a correct clone, the DNA was excised from pCR2.1-TOPO and cloned into the pKLAC1-H vector using XhoI and SpeI restriction enzymes (New England Biolabs, MA, USA) according to the manufacturer's protocol. The pKLAC1-H is a modified version of the integrative pKLAC1 vector (New England Biolabs; [GenBank: AY968582]). The pKLAC1 - developed by Colussi and Taron  - exhibits the α-mating factor signal sequence and can be used for the expression and secretion of recombinant proteins in K. lactis . The pKLAC1-H was constructed by including an additional SpeI restriction site directly followed by a His-tag coding sequence (6xHis) between the XhoI and AvrII restriction sites of pKLAC1. The DNA sequence of the final pKLAC1-H construct (containing the DNA coding for recombinant swollenin) is shown in Additional file 1. Moreover, the final amino acid sequence of recombinant swollenin (without the α-mating factor signal sequence) is given in Figure 1C.
Expression and purification of recombinant swollenin
All below-mentioned transformation, selection and precultivation procedures - developed by Colussi and Taron  - were performed according to the manufacturer's protocol (K. lactis Protein Expression Kit, New England Biolabs). After cloning, K. lactis GG799 cells were transformed with pKLAC1-H (containing the DNA coding for recombinant swollenin), and transformed clones were selected (acetamide selection). One clone was precultivated in YPGal (yeast extract, peptone and galactose) medium, consisting of 20 g/L galactose, 20 g/L peptone and 10 g/L yeast extract - all media components were purchased from Carl Roth (Karlsruhe, Germany). After inoculation with 2.5 mL of the preculture, the main culture was cultivated in triplicates in 2 L Erlenmeyer flasks with YPGal medium under the following constant conditions: temperature T = 30°C, total filling volume V
= 250 mL, shaking diameter d
= 50 mm, shaking frequency n = 200 rpm. Additionally, a non-transformed K. lactis wild type was cultivated as a reference. After incubation for 72 h, the main cultures were centrifuged (6000 g, 20 min, 4°C), and the pooled supernatants of the triplicates were treated with endoglycosidase Hf by using 20 U per μg protein for 12 h  according to the manufacturer's protocol without denaturation (New England Biolabs). Afterwards, the protein solution was concentrated 100-fold at 4°C using a Vivacell 100 ultrafiltration system with a molecular mass cut-off of 10 kDa (Sartorius Stedim Biotech, Göttingen, Germany). For affinity chromatography, the recombinant swollenin was previously rebuffered using Sephadex G-25 Fine (2.6 cm × 10 cm, GE Healthcare) at 110 cm/h with a running buffer (pH 7.4) consisting of 0.05 M sodium dihydrogen phosphate, 0.3 M sodium chloride and 0.01 M imidazole. The rebuffered sample was loaded on Ni Sepharose 6 Fast Flow (1.6 cm × 10 cm; GE Healthcare) at 120 cm/h. The bound swollenin was eluted with the aforementioned running buffer, containing 0.25 M imidazole.
SDS-PAGE and Western blot analysis
SDS-PAGE and Western blot analysis were applied to analyze the purity and to identify the recombinant swollenin. Novex 12% polyacrylamide Tris-Glycine gels (Invitrogen), and samples were prepared according to the manufacturer's protocol. The Plus Prestained Protein Ladder (Fermentas, Burlington, CA, USA) was used as a molecular mass marker. Finally, the proteins were stained with Coomassie Brilliant Blue and analyzed densitometrically  using the scanner Perfection V700 (Epson, Suwa, Japan). The molecular mass and purity of swollenin was determined using the software TotalLab TL100 (Nonlinear Dynamics, Newcastle, UK). For Western blot analysis, gels were blotted onto a nitrocellulose membrane (Whatman, Springfield Mill, UK) according to the manufacturer's protocol (Invitrogen). The membranes were blocked at room temperature with 50 g/L skim milk dissolved in phosphate buffered saline containing 0.5 g/L Tween-20 (PBST) for 30 min. To detect the recombinant swollenin, the membranes were incubated at room temperature for 1.5 h with a rabbit polyclonal antibody against His-tag (Dianova, Hamburg, Germany) diluted 1:10,000 in PBST. After the membrane was washed thrice with PBST, it was incubated with alkaline phosphatase conjugated goat anti-rabbit IgG (Dianova) diluted 1:5,000 in PBST at room temperature for 1 h. Finally, bound antibodies were visualized by incubating the membrane for 5 min with nitro blue tetrazolium/5-Bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) diluted 1:100 in phosphatase buffer (100 mM Tris-HCl, 100 mM NaCl, 5 mM MgCl2, pH 9.6).
Measurement of protein concentration
Protein concentrations were analyzed with the bicinchoninic acid assay  using the BCA Protein Assay Kit (Thermo Fisher Scientific, MA, USA) and BSA as a standard. Depending on the protein concentration of the samples, the standard procedure (working range: 0.02 to 2 g/L) or the enhanced procedure (working range: 0.005 to 0.25 g/L) was performed according to the manufacturer's protocol. The absorbance at 562 nm was measured with a Synergy 4 microtiter plate reader (BioTek Instruments, VT, USA). To quantify swollenin in the culture supernatant of K. lactis, the bicinchoninic acid assay was combined with the aforementioned SDS-PAGE (including densitometric analysis). Here, total protein concentrations were determined and multiplied with the ratio of swollenin to total protein (purity).
Mass spectrometry and glycosylation analysis
Mass spectrometry was applied to identify the expressed and purified recombinant swollenin. The protein band (approximately 80 kDa) was excised from the SDS-polyacrylamide gel, washed in water, reduced with dithiothreitol, alkylated with iodoacetamide, and digested with trypsin . Peptide analysis was carried out using a nanoHPLC (Dionex, Germering, Germany) coupled to an ESI-QUAD-TOF-2 mass spectrometer (Waters Micromass, Eschborn, Germany) as previously described . The Mascot algorithm (Matrix Science, London, UK) was used to correlate the mass spectrometry data with amino acid sequences in the Swissprot database. Thereby, the sequences of the analyzed peptides could be identified, and, ultimately, protein matches could be determined. The Mascot score is derived from the ions scores of the detected peptides matching the peptides in the database and reflects a non-probabilistic basis for ranking protein hits . By using this database, the peptide mass tolerance was set at ± 0.3 Da. Additionally, the following modifications to the amino acids in brackets were allowed: carbamidomethyl (C), carboxymethyl (C), oxidation (M), propionamide (C). Moreover, potential areas for N-glycosylation and O-glycosylation were identified by using the NetNGlyc 1.0 and NetOGlyc 3.1 servers http://www.cbs.dtu.dk/services/.
Adsorption experiments were performed in 0.05 M sodium acetate buffer (pH 4.8) using 20 g/L untreated filter paper and various concentrations (0.05 to 1.25 g/L) of purified swollenin. Solutions with filter paper and solutions with swollenin were preincubated separately at 45°C for 10 min, and experiments were started by mixing both solutions. The final mixtures were incubated in 2 mL Eppendorf tubes on a thermomixer MHR23 (simultaneous shaking and temperature control; HLC Biotech, Bovenden, Germany) under the following constant conditions for 2 h: T
= 45°C, V
= 1 mL, d
3 mm, n
= 1000 rpm. The shaking frequency was chosen to ensure the complete suspension of cellulose particles [64
]. Three different blanks were incubated similarly: (i) without swollenin, (ii) without filter paper, or (iii) without filter paper and without swollenin. The incubation was stopped by centrifugation (8000 g, 1 min), and the supernatants were immediately analyzed for unbound swollenin by using the bicinchoninic acid assay. The adsorbed swollenin concentration was calculated as the difference between initial (blanks) and unbound swollenin concentration. Adsorption isotherm parameters were determined using the Langmuir isotherm [92
in which A denotes the amount of adsorbed protein per g cellulose (μmol/g), A
, the maximum protein adsorption per g cellulose at equilibrium (μmol/g), E, the free protein concentration (μmol/L), and K
, the dissociation constant (μmol/L). Within the literature , the association constant K
(L/μmol) is sometimes used instead of the dissociation constant K
To analyze the effect of swollenin pretreatment (see below) on cellulase adsorption [44, 93], the maximum cellulase adsorption was also determined by incubating various concentrations (0.7 to 2.5 g/L) of rebuffered Celluclast® with 10 g/L pretreated cellulosic substrates. Here, all incubations were conducted under the aforementioned conditions for 1 h, 1.5 h and 2 h.
Pretreatment with swollenin
Pretreatment experiments were performed with 20 g/L cellulosic substrates and various concentrations of swollenin in 0.05 M sodium acetate buffer (pH 4.8). The mixtures were incubated as triplicates in 2 mL Eppendorf tubes on a thermomixer under the following constant conditions: T = 45°C, V
= 1 mL, d
= 3 mm, n = 1000 rpm. To exclude a sole mechanical effect on cellulosic substrates due to shaking and to verify a specific effect of swollenin, blanks without swollenin (buffer) or with 0.4 g/L BSA instead of swollenin were incubated similarly. To detect a possible hydrolytic activity of recombinant swollenin, the sensitive p-hydroxy benzoic acid hydrazide assay  was applied by using glucose as a standard. After incubation for 48 h, the supernatants of the pretreatment solution were analyzed and the absorbancies were measured at 410 nm in a Synergy 4 microtiter plate reader. Subsequently, all cellulosic samples were washed to remove adsorbed proteins. Therefore, the mixtures were centrifuged (14,000 × g, 10 min, 4°C), and the cellulosic pellets were washed four times with 800 μL 0.05 M citrate buffer (pH 10) , and once with 800 μL distilled water. Finally, the triplicates were pooled. According to Zhu et al. , citrate buffer (pH 10) is an appropriate washing solution, and a single washing step with 0.05 M citrate buffer (pH 10) leads to a desorption efficiency of 61% in case of fungal cellulases and Avicel. Since no acids or bases are formed during the washing procedure, the weak buffer capacity of citrate buffer at pH 10 can be neglected. In this study, the washing procedure was conducted four times to ensure a high desorption of swollenin. The measurements of protein concentration in the washing supernatants - by applying the aforementioned bicinchoninic acid assay (working range starting from 0.005 g/L) - showed that swollenin desorbed almost completely. Already after three washing steps, a total swollenin desorption efficiency of > 90% was achieved.
Photography and microscopy
Photography and microscopy were applied to visualize the effect of swollenin pretreatment on filter paper. After pretreatment with buffer, BSA or swollenin, the different filter paper solutions were transferred into petri dishes, the particles were evenly distributed and images were taken with an Exilim EX-FH100 camera (Casio, Tokyo, Japan). Afterwards, the number of filter paper agglomerates (> 0.5 mm) was determined by image analysis using the software UTHSCSA ImageTool 3.0 (freeware) and a ruler as a reference. Light microscopic pictures were taken with an Eclipse E600 (Nikon, Tokyo, Japan). Additionally, scanning electron microscopy was performed using a Hitachi S-5500 (Hitachi, Tokyo, Japan) and a field emission of 5 kV. All washed filter paper samples were covered with a layer of carbon (3 nm) and, subsequently, with a layer of PtPd (3 nm, 80% to 20%). The images were taken by using secondary electrons.
Laser diffraction and X-ray diffraction
The particle-size distributions of all pretreated cellulosic substrates were measured by laser diffraction [96
] using a LS13320 (Beckman Coulter, CA, USA). In the case of filter paper, particles with an average diameter of greater than 0.75 mm were manually removed before laser diffraction to exclude a disturbance of measurement signals. The geometric mean particle size was calculated using the software LS 5.01 (Beckman Coulter). Moreover, the CrI
was determined by powder XRD. XRD patterns were obtained using a STOE STADI P transmission diffractometer (STOE & Cie GmbH, Darmstadt, Germany) in Debye-Scherrer geometry (CuKα
= 1.54060 Å) with a primary monochromator and a position-sensitive detector. Thereby, XRD patterns were collected with a diffraction angle 2θ
from 10° to 30° (increments of 0.01°) and a counting time of 6 s per increment. The sample was adhered to a polyester foil (biaxially-oriented polyethylene terephthalate) by using a dilute solution of glue. After drying the sample in open-air, the sample was covered with a second polyester foil. This set was then fixed in a sample holder. To improve statistics and level out sample orientation effects, the sample was rotated at around 2 Hz during XRD measurement. The CrI
was calculated using the peak height method [28
] and the corresponding equation:
is the maximum intensity of the crystalline plane (002) reflection (2θ = 22.5°) and I
is the intensity of the scattering for the amorphous component at about 18° in cellulose-I . Here, it should be noted that there are several methods for calculating CrI from XRD data and these methods can provide significantly different results [28, 70]. Although the applied peak height method produces CrI values that are higher than those of other methods, it is still the most commonly used method and ranks CrI values in the same order as the other methods .
Hydrolysis experiments and dinitrosalicylic acid assay
Hydrolysis experiments with 10 g/L pretreated cellulosic substrate and 1 g/L rebuffered Celluclast® were conducted in 0.05 M sodium acetate buffer (pH 4.8). The mixtures were incubated as triplicates in 2 mL Eppendorf tubes on a thermomixer under the following constant conditions: T = 45°C, total filling V
= 1 mL, d
= 3 mm, n = 1000 rpm. In general, attention has to be paid to cellulase inactivation, which would reduce the final yield of cellulose hydrolysis . In this current study, however, a shaken system with relatively low shear forces was applied. According to Engel et al. , rebuffered Celluclast® is stable under the applied incubation conditions, so that cellulase inactivation could be neglected. The shaking frequency was chosen to ensure the complete suspension of cellulose particles [64, 91]. Thus, mass transfer limitations are excluded, and the whole cellulose particle surface becomes accessible to the cellulases, thereby optimizing cellulase adsorption and activity . Three different blanks were incubated similarly: (i) without cellulase, (ii) without substrate, or (iii) without substrate and without cellulase. The dinitrosalicylic acid assay  was applied to quantify the reducing sugars released during hydrolysis by using glucose as a standard. After defined time intervals, samples were taken, and the hydrolysis was stopped (10 min, 100°C). According to Wood and Bhat , low reducing sugar concentrations were quantified by adding 1.25 g/L glucose to the samples. The absorbancies were measured at 540 nm in a Synergy 4 microtiter plate reader. Since the dinitrosalicylic acid assay exhibits a lower sensitivity towards cellobiose than glucose, reducing sugar concentrations may be underestimated when glucose is used as a standard and β-glucosidase is not in excess . However, under the applied hydrolysis conditions, cellobiose did not accumulate (the highest cellobiose to glucose ratio was measured in the case of Sigmacell after 10 h at 0.12) and, therefore, this underestimation was minimal and the addition of β-glucosidase was not needed. Initial hydrolysis rates (g/(L*h)) were calculated by applying a linear fit to the reducing sugar concentration data from 0 to 6 h.
Parameters (including standard deviations) of the adsorption model were calculated by nonlinear, least squares regression analysis using MATLAB R2010 (The MathWorks, Natick, USA). TableCurve 3D 4.0 (Systat Software, San Jose, CA, USA) was used to empirically correlate CrI
and mean particle size with initial hydrolysis rates via the non-linear Gaussian cumulative function:
in which a, b, c, d, e, f and g denote the various fitting parameters of the non-linear Gaussian cumulative function (-).