For 10 different genes involved in lignin biosynthesis, two Arabidopsis thaliana (L.) Heyhn. mutant alleles were used in this study, including pal1-2, pal1-3, pal2-2, pal2-3, c4h-2, c4h-3, 4cl1-1, 4cl1-2, 4cl2-1, 4cl2-3, ccoaomt1-3, ccoaomt1-5, ccr1-3, ccr1-6, f5h1-2, f5h1-4, comt-1, comt-4, cad6-1, and cad6-4. For a schematic presentation of the 20 mutants and their residual expression, see Vanholme et al.
. Sixteen biological replicates of each homozygous mutant and 32 biological replicates for the wild-type were grown simultaneously in a random block design, spread over different trays, in the same environment. Because of their delayed development, c4h-2, ccr1-3, and ccr1-6 were planted 2 weeks in advance to allow simultaneous bolting. Plants were grown first under short-day conditions (8 h light, 21°C, and 55% humidity) during 6 weeks and then transferred to long-day conditions (16 h light, 21°C, and 55% humidity) to allow the development of a single tall inflorescent stem. For all biological repeats, the main stem was harvested just above the rosette when the plant was completely senesced and dry. Once harvested, rosette leaves, axillary inflorescences, and siliques were removed. The main stem was weighed and the bottom 1 cm was removed. The lowest 10 cm of the remaining stem was chopped in 2-mm pieces. Biological repeats were pooled two by two to obtain 8 biological replicates for the mutant alleles and 16 repeats for the wild-type. These pooled samples were used for wet-chemistry cell wall analyses and saccharification assays.
Aliquots of 5 mg stem pieces were subjected to a sequential extraction to obtain a purified CWR. The extractions were done in 2-ml vials, each time for 30 min, at near boiling temperatures for water (98°C), ethanol (76°C), chloroform (59°C), and acetone (54°C). The remaining CWR was dried under vacuum. Lignin was quantified according to a modified version of the acetyl bromide method
, optimized for small amounts of plant tissue. The dried CWR was dissolved in 0.1 ml freshly made 25% acetyl bromide in glacial acetic acid and 4 μl 60% perchloric acid. The solution was incubated for 30 min at 70°C while shaking (850 rpm). After incubation, the slurry was centrifuged at 23,477 g for 15 min. To the supernatant, 0.2 ml of 2 M sodium hydroxide and 0.5 ml glacial acetic acid were added. The pellet was washed with 0.5 ml glacial acetic acid. The supernatant and the washing phase were combined and the final volume was adjusted to 2 ml with glacial acetic acid. After 20 min at room temperature, the absorbance at 280 nm was measured with a NanoDrop® ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA). The lignin concentrations were calculated by means of the Bouguer-Lambert-Beer law: A = ϵ × l × c, with ϵ = 23.35 l g-1 cm-1 and l = 0.1 cm.
The lignin composition was investigated with thioacidolysis as previously described
. The monomers involved in β–O–4-ether bonds, released upon thioacidolysis, were detected with gas chromatography (GC) as their trimethylsilyl (TMS) ether derivatives on a Hewlett-Packard HP 6890 Series GC system (Agilent, Santa Clara, CA, USA) coupled with a HP-5973 mass-selective detector. The GC conditions were as described
. The quantitative evaluation was carried out based on the specific prominent ions for each compound. A summary of the specific ions for each identified compound can be found in Additional file
7. Response factors for H, G, and S units were taken from
. Because we had no standards for the minor lignin units, a response factor of 0.47 was used, which is the average of the three response factors for the major lignin units.
Aliquots of 4 mg dry stem pieces were sequentially extracted to obtain a purified CWR, as described above. To estimate the amount of cellulose, we used a colorimetric method (based on
[45, 46]). The CWR was incubated with 2 M TFA and 20 μl inositol (5 mg ml-1) for 2 h at 99°C while shaking (750 rpm). This TFA extract was used for the determination of the sugar composition of matrix polysaccharides (see below). After incubation, the remaining pellet was washed three times with water and twice with acetone and dried under vacuum. Concentrated sulfuric acid (150 μl) and 30 μl 5% (w/v) phenol (freshly made in water) were added to the dried pellet and incubated for 1 h at 90°C with gentle shaking (500 rpm). After centrifugation for 3 min at 23,477 g, a 50 μl aliquot of the supernatant was diluted 20 times with MilliQ water (Millipore, Billerica, MA, USA) to measure the absorbance at 493 nm. The amount of cellulose was calculated back from a standard curve of Avicel® PH-101 (FMC BioPolymer, Philadelphia, PA, USA).
To determine the different monosaccharides present in the TFA extract, 800 μl TFA extract was dried under vacuum and further converted to the corresponding alditol acetates as described
. The GC-MS analyses were carried out with a mass-selective detector (HP 5973 model; Agilent), interfaced to a GC (HP 6890 model; Agilent) equipped with an automated sample injector and an VF-5 ms capillary column (30 m × 0.25 mm). The GC conditions were as described
: the oven was kept at 100°C for 1 min, increasing the temperature to 245°C at a rate of 20°C min-1, held at 245°C for 30 min, and decreasing the temperature to a final temperature of 100°C at a rate of 30°C min-1. Peak areas of the different sugars were normalized with the peak area of the internal standard inositol (20 μl, 5 mg ml-1). Response factors were determined based on standard curves for each of the different sugars: rhamnose (2.01), fucose (2.05), arabinose (1.35), xylose (1.35), mannose (1.45), glucose (1.59), and galactose (1.55).
Analysis of xylan structure
Senesced stems (10 mg) from five biological replicates of each genotype were incubated at 70°C in 96% ethanol for 20 min and then homogenized using a ball mixer mill (Glen Creston, London, UK). The insoluble material was washed with 100% ethanol, twice with chloroform:methanol (2:1), and then successively washed with 65%, 80%, and 100% ethanol prior to air drying to produce an alcohol-insoluble residue (AIR). AIR (100 μg) was pretreated with 4 M NaOH (20 μl) for 1 h at 21°C to make the xylan enzyme accessible, neutralized, and then resuspended in 500 μl 0.1 M ammonium acetate buffer (pH 5.5). Samples were incubated overnight at 21°C with an excess of the xylanase NpXyn11A (a kind gift of Harry Gilbert, University of Newcastle, UK) to ensure complete digestion.
The samples were dried under vacuum and analyzed by DNA sequencer Assisted Saccharide analysis in High throughput (DASH). The xylanase-released oligosaccharides or quantitation standards and appropriate controls were derivatized with 8-aminopyrene-1,3,6-trisulfonic acid (APTS; Biotium, Hayward, CA, USA). The dried oligosaccharide sample was combined with 10 μl APTS (0.02 M in 1.2 M citric acid) and 10 μl 0.1 M NaCNBH3 solutions. Following overnight incubation (30°C), the samples were diluted to 1 μg ml-1 initial AIR, of which 10 to 30 μl were loaded into a 96-well microtiter plate, and analyzed by capillary electrophoresis with a laser-induced fluorescence (CE-LIF) on an Applied Biosystems 3730xl DNA Analyzer (Life Technologies, Carlsbad, CA, USA). Peaks were identified by co-migration with known standards and quantified based on peak area using quantitation standards analyzed in parallel. Oligosaccharides Xyl, Xyl2, GlcA Xyl4, and MeGlcA Xyl4 were used to calculate the degree of xylan branching and the proportion of GlcA methylation, as described previously with the polysaccharide analysis using the carbohydrate gel electrophoresis technique
The protocol for saccharification of senesced Arabidopsis inflorescences was as follows. Aliquots of 10 mg of dry 2-mm stem segments were used. The biomass was pretreated with 1 ml of 1 N HCl at 80°C for 2 h, while shaking (850 rpm). The acid extract was removed and the pretreated material was washed three times with 1 ml water to obtain neutral pH. Subsequently, the material was incubated with 1 ml 70% (v/v) ethanol overnight at 55°C. The remaining biomass was washed three times with 1 ml 70% (v/v) ethanol, once with 1 ml acetone, and dried under vacuum for 45 min. For the saccharifications without pretreatment, 10 mg aliquots of dry stem segments were immediately incubated with 1 ml 70% (v/v) ethanol overnight at 55°C. The extracted material was washed three times with 1 ml 70% (v/v) ethanol, once with 1 ml acetone, dried under vacuum for 45 min, and weighed. As this overnight ethanol extraction was an alternative procedure for cell wall preparation, the percentage of CWR in dry matter, specific for saccharification, could be calculated based on the weights before and after overnight ethanol extraction. As inclusion of the acid pretreatment prior to the overnight ethanol extraction removed some cell wall components, weighing the material after acid pretreatment and ethanol extraction overnight underestimated the amount of CWR. Therefore, the CWR data from the untreated samples were taken to calculate the glucose release with acid pretreatment per CWR.
The ethanol-extracted residue, after acid pretreatment or no pretreatment, was dissolved in 1 ml acetic acid buffer solution (pH 4.8) and incubated at 50°C. The enzyme mix added to the dissolved material contained cellulase from Trichoderma reseei ATCC 26921 and β-glucosidase (Novozyme, Bagsvaerd, Denmark) in a 5:3 ratio. Both enzymes were first desalted over an Econo-Pac 10DG column (Bio-Rad, Hercules, CA, USA), stacked with Bio-gel® P-6DG gel (Bio-Rad) according to the manufacturer’s guidelines. The desalted β-glucosidase was 350-fold diluted prior to mixing with desalted cellulase. The enzyme mix was further diluted 10-fold and the activity of the diluted enzyme mix was measured with a filter paper assay
. To each biological sample, dissolved in acetic acid buffer (pH 4.8), the enzyme mix with an activity of 0.06 filter paper units was added. After a short spinning to remove droplets from the lid of the reaction tubes, 20 μl aliquots of the supernatant were taken after 48 h of incubation at 50°C and 30-fold diluted with acetic acid buffer (pH 4.8). The concentration of glucose in these diluted samples was measured indirectly with a spectrophotometric color reaction (glucose oxidase-peroxidase; GOD-POD). A 100 ml aliquot of the reaction mix from this color reaction contained 50 mg 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid), 44.83 mg GOD (Sigma-Aldrich, St. Louis, MO, USA), and 173 μl of 4% (w/v) POD (Roche Diagnostics, Brussels, Belgium) in acetic acid buffer (pH 4.5). To measure the glucose concentration, 50 μl of the diluted samples was added to 150 μl GOD-POD solution and incubated for 30 min at 37°C. The absorbance was measured spectrophotometrically at a wavelength of 405 nm. The concentration in the original sample was calculated with a standard curve based on known d-glucose concentrations (Sigma-Aldrich).
Descriptive statistics and significance tests
All statistical analyses were performed with SAS® 9.2 (SAS Institute Inc., 2008, Cary, North Carolina). Mixed model analysis was performed for each variable to test whether there was a significant line effect. Tray was put as a random effect in the model. The significance of the tray effect was assessed with a likelihood ratio test, using a mixture distribution of χ1
. When model assumptions were not met, box-cox powertransformations were applied to ensure the validity of the model. All variables were ≥0. To variables that also contained values between 0 and 1, the value 1 was added prior to transformation. Osborne
] pointed out that for some transformations numbers between 0 and 1 are treated differently than numbers above 1. In cases where no suitable powertransformation was possible, the nonparametric Friedman test was used. The applied transformations were as follows:
H = square root of (H+1);
bis-β–O–4-FA = (bis-β–O–4-FA +1)0.2;
β–O–4-FA = loge (β–O–4-FA +1);
G aldehyde = 1/(Gald+1);
S aldehyde = 1/(Sald+1);
glucose = 1/square root (glucose);
galactose= loge (galactose)
A nonparametric analysis was applied to % H, % G, % S. Post-hoc Dunnett’s tests (2-sided) were performed to test for significant differences between a particular line and wild-type. Differences with a Dunnett adjusted p-values < 0.01 were considered significant. Differences are reported on their original scale, however the null hypotheses are valid on the transformed scale only. For the nonparametric test approximations to the rank-sum multiple comparisons were obtained.
Saccharification yield (mg glucose/mg CWR) was measured at different timepoints. A repeated measurements analysis was performed using a linear spline model with knots at timepoints 3 h, 8 h and 24 h based on the line plots. Several covariance structures were modelled. The model with the lowest AIC value was the model with an unstructured covariance. Tray was also included in the model as a random effect. The significance of the tray effect was assessed with a likelihood ratio test, using a mixture distribution of χ1
. The full model was:
Saccharification yield = intercept+tray+line+time+time*line+time3+time3*line+time8+time8*line+time24+time24*line.
One-sided post-hoc Dunnett’s tests were performed at 48h to test for significant increases in saccharification yield in a particular line compared to wild-type. Separate analyses were performed for the experiments with and without pretreatment.
Statistical modeling of saccharification yield
To understand which factors had the most significant influence, the saccharification yield in mg per mg dry weight at 48 h was modeled separately for the analysis with and without pretreatment by means of multiple linear regression. Data from the two ccr1 and the c4h-2 mutants were not taken into account. The factors that were considered to influence the glucose release were lignin (mg), S/G, ferulic acid (bis-β–O–4-FA, μmol), the three most abundant sugars (arabinose (μg), galactose (μg), and xylose (μg)), cellulose (mg), and matrix polysaccharides (mg). All factors were expressed on a mg dry weight basis. In a first step, all correlations between the different independent variables were investigated to avoid multicollinearity. As the correlation between arabinose and galactose (r = 0.94, p < .0001) was very high, arabinose was eliminated from the model. Multicollinearity was monitored with the variance inflation factor (VIF) of which the square root indicates how much larger the standard error is compared with what it would be if the variables were uncorrelated with the other independent variables in the equation. Models were built by manual backward selection with removal of one severe outlier. The R2 of the final model for the saccharification yield was 0.70 and 0.71 without and with pretreatment, respectively. The effect of pretreatment was also modeled with the same covariates; the R2 of the final model for this pretreatment effect was 0.58.