Enzyme preparation
The enzymes used in this study were CBH I (TrCel7A), CBH II (TrCel6A), EG I (TrCel7B), EG II (TrCel5A), EG IV (TrCel61A), Xyn III, β-xylosidase (TrXyl3A, BXL), BGL I, and Xyn10 (PspXyn10). CBH I, CBH II, EG I, EG II, EG IV, Xyn III and BXL were derived from the T. reesei strain PC-3-7 [19] and BGL I was derived from Aspergillus aculeatus [20]. These eight enzymes were heterologously expressed in A. oryzae (Ozeki Co. Ltd., Hyogo, Japan) following the method reported by Kawai et al. [21]. Each enzyme was purified from the culture supernatant of A. oryzae cells by hydrophobic chromatography (TOYOPEARL®Butyl-650) followed by anion exchange chromatography (TOYOPEARL® DEAE-650).
To prepare Xyn10 from Penicillium sp., the pspxyn10 gene was amplified from pUC-Pcbh1-pspxyn10-amdS plasmid [22] by a polymerase chain reaction and expressed in A. oryzae cells under the control of an improved enoA promoter (PenoA142f) [23] that harbored 12 tandem repeats of the cis-acting element (region III) of the agdA promoter [24]. A. oryzae cells expressing the pspxyn10 gene were cultured in a DP medium (2% dextrin hydrate, 1% peptone) containing 0.5% potassium dihydrogen phosphate, 0.05% magnesium sulfate, 0.187% l-glutamic acid monosodium salt, and 0.003% l-methionine at 30 °C, 105 rpm for 3 days. After cultivation, the A. oryzae cells were removed by filtration (0.45 μm) and Xyn10 was purified as described above.
Enzyme labeling with a fluorescent dye
Herein, CBH I, CBH II, EG I, EG II, Xyn III, and Xyn10 were labeled with a fluorescent dye as theses enzymes are the main components of a wild-type cellulase and have important functions.
Each enzyme was labeled with Alexa Fluor® 546 NHS Ester (Invitrogen, California, USA) in accordance with an attached instruction. The fluorescent molecule forms a covalent bond with a primary amine group in enzyme. Determination of the protein concentration was performed using a Quick Start™ Bradford protein assay (Bio-Rad, California, USA), and a gamma-globulin standard was used throughout this study. Bio-Gel® P-4Gel fine (Bio-Rad, wet bead size 45–90 µm, molecular weight exclusion limit of > 4000) was used to separate the labeled enzyme from the free dye. The absorbance at 554 nm of the labeled enzyme solution was measured to calculate the degree of labeling (DL), using the following formula:
$${\text{DL}} = \, \left( {A_{ 5 5 4} \times k} \right) \, / \, \left( {\mu_{\text{ext}} \times C_{\text{protein}} } \right),$$
where DL is the amount of moles of dye per mole of protein, A554 is the absorbance at 554 nm, k is a dilution factor, µext is the molar extinction coefficient of Alexa Fluor® 546 NHS Ester at 554 nm (104,000 cm−1M−1), and Cprotein is the protein concentration (M).
Assessment of labeled xylanases
To compare the enzymatic activity of two xylanases, a saccharification test was investigated in the presence of cellulase mixture including Xyn III or Xyn10. The substrate, sugarcane (Saccharum officinarum) bagasse powder, was sieved through a 1-mm mesh and pretreated by autoclaving in 1% sodium hydroxide solution at 120 °C for 20 min. The composition was estimated as: cellulose (63%), hemicellulose (18%), lignin (7.6%), and ash (4.0%). The bagasse (10 mg on a dry matter basis) was hydrolyzed in 0.1 M acetate buffer (pH 5.0) at 50 °C, shaking at 150 rpm. The total solution volume was 1 mL and the enzyme concentration was 3 mg/g of substrate. The enzyme composition is the same as that described later in the microscopy section. The experiments were performed for both nonlabeled and labeled xylanases. The supernatant was collected at 5, 24, 48, and 96 h to measure d-glucose yield using a CII Test Wako kit (Wako, Osaka, Japan), and d-xylose yield using a d-xylose kit (Megazyme, Wicklow, Ireland).
Fluorescence microscopy of a selectively labeled enzyme in a cellulase cocktail
Transverse sections (30-µm-thick) were cut from a stem of sugarcane harvested in Okinawa, Japan, by a microtome equipped with a freezing stage. The sections were treated in 0.5% sodium hydroxide using an oil bath at 100 °C for 1 h. After washing thoroughly, the treated sections were used as the substrate. The enzyme mixture comprised purified components of CBH I 35 wt%, CBH II 20 wt%, EG I 15 wt%, EG II 5 wt%, EG IV 5 wt%, BGL I 5 wt%, BXL 5 wt%, and Xyn III 10 wt%. One of the enzymes was replaced with a fluorescent-labeled enzyme at a fixed ratio of 5 wt% (total enzyme basis); for example, a system comprising labeled CBH I at 5 wt% and nonlabeled CBH I at 30 wt%. For comparison purpose, Xyn III was replaced with Xyn10 to visualize the functional difference between the two xylanases.
A pretreated section was mounted on a glass slide together with a 25-μL aliquot of an enzyme mixture (~ 40 mg/g biomass). A cover slip was then placed on top of the specimen and sealed with nail polish to prevent water evaporating during the reaction. The preparation was performed on a thermostage, maintained at 50 °C, with an inverted fluorescent microscope (IX71, Olympus, Tokyo, Japan) under a constant illumination flux from a super-high pressure mercury lamp and a 4× objective lens (UPlanFLN, NA: 0.13, Olympus). Images (1600 × 1200 pixels, 8 bit RGB) were recorded every 5 min for 360 min in fluorescent mode with a charge-coupled device camera having an exposure time set at 500 ms (DP 73, Olympus). As the filter set, TRIRC-B (Semrock, N.Y., USA) was used comprising a bandpass excitation filter (543 nm/22 nm), a dichroic mirror (> 562 nm), and a bandpass emission filter (593 nm/40 nm). No auto-fluorescence was detected in the presence of a nonfluorescent-labeled enzyme subjected to the same conditions.
Relationship between the number of labeled enzymes and fluorescence image intensity
Prior to image data interpretation, the relationship between the fluorescence intensity of the microscopic image and the labeled enzyme concentration was determined. A series of 2-μL enzyme-labeled solutions, at various concentrations, were placed in a circle (φ = 8 mm) surrounded by water-resistant fluororesin on a glass slide, TF0808 (MATSUNAMI, Osaka, Japan). Thereafter, a cover slip was placed on the solution and observations made using fluorescent microscopy under the same conditions as previously described. The average fluorescence intensity was calculated from five different positions using ImageJ software, and plotted against the corresponding enzyme concentration.
Time-lapse fluorescence profiles from two-dimensional images
A stack of fluorescent images (at 5-min intervals) were carefully aligned by the registration algorithm proposed by Thévenaz et al. [25] as a plugin for ImageJ. A 500 × 500 pixel region was cropped, wherein one complete vascular bundle (VB) from the inner part of the stem was recorded. After conversion to a gray-scale, and noise reduction by median filtering, the intensity profiles from each pixel in an image stack were taken as a function of time. The 250,000 time-dependent intensity profiles were then classified into eight representative profiles by the k-Means algorithm, and the corresponding regions associated with the eight profiles were contour-mapped into two-dimensional images. The number of clusters was chosen to be slightly larger than the number of cell types in the region of interest: phloem, bundle sheath, metaxylem and parenchyma, which were expected to show different susceptibilities to the enzymatic attack. All calculations were performed in python 3.6 using the scikit-learn v0.19.2 [26] data mining tool.