• Open Access

Genetic modification in cellulose-synthase reduces crystallinity and improves biochemical conversion to fermentable sugar



    1. Department of Horticulture, University of Kentucky, N-318 Agricultural Science Center, North, Lexington, KY 40506, USA
    Search for more papers by this author
    • 1Equal contributing authors.


    1. Department of Horticulture, University of Kentucky, N-318 Agricultural Science Center, North, Lexington, KY 40506, USA
    Search for more papers by this author
    • 1Equal contributing authors.


    1. Department of Horticulture, University of Kentucky, N-318 Agricultural Science Center, North, Lexington, KY 40506, USA
    Search for more papers by this author

Seth DeBolt, e-mail: sdebo2@email.uky.edu


The cellulose synthase (CESA) membrane complex synthesizes microfibrils of cellulose that surround all plant cells. Cellulose is made of sugar (β,1-4 glucan) and accessing the sugar in cellulose for biofuels is of critical importance to stem the use of fossil fuels and avoid competition with food crops and pristine lands associated with starch-based biofuel production. The recalcitrance of cellulose to enzymatic conversion to a fermentable form of sugar is related to the degree of hydrogen bonding or crystallization of the glucan chain. Herein, we isolate the first viable low biomass-crystallinity mutant by screening for altered cell wall structure using X-ray scattering as well as screening for enzymatic conversion efficiency on a range of cell wall mutants in the model plant Arabidopsis thaliana (L.) Heynh. Through detailed analysis of the kinetics of bioconversion we identified a mutant that met both selection criteria. This mutant is ixr1-2, which contains a mutation in a highly conserved consensus sequence among the C-terminal transmembrane regions within CESA3. A 34% lower biomass crystallization index and 151% improvement in the efficiency of conversion from raw biomass to fermentable sugars was measured relative to that of wild type (Col-0). Recognizing the inherent ambiguities with an insoluble complex substrate like cellulose and how little is still understood regarding the regulation of CESA we propose a general model for how to manipulate CESA enzymes to improve the recalcitrance of cellulose to enzymatic hydrolysis. This study also raises intriguing possibilities as to the functional importance of transmembrane anchoring in CESA complex and microfibril formation.


cellulose synthase


high-performance liquid chromatography


filter paper units


A sustainable human economy will require the implementation of renewable forms of energy to reduce the impacts of climate change associated with continued fossil fuel consumption (Searchinger et al., 2008). One renewable form of bioenergy that has gained considerable interest is to exploit the capacity of land plants to fix atmospheric carbon using energy from the sun to make sugar polymers (photosynthesis) (Farquhar et al., 2001). The most plentiful of these plant sugar polymers is cellulose, and it is noteworthy that plants proliferate in nearly every biome on the planet (Wright et al., 2004) and cellulose is the most abundant biopolymer on Earth (Sticklen, 2008), which add to its attractiveness as a source of renewable energy. Unfortunately, cellulose polymers, referred to as microfibrils, are crystalline and highly recalcitrant to enyzmatic breakdown to form fermentable sugars that can be fermented to make alcohol for biofuel. This crystallinity presents a major technical hurdle to overcome if cost-effective biochemical conversion of lignocellulosic biomass into biofuels is to be realized.

Providing a fundamental understanding of cellulose biosynthesis may also assist in understanding how it could be more efficiently broken down to biofuel. In elongating plant tissue, cellulose deposition is generally considered to occur perpendicular to the axis of elongation, constraining lateral swelling and allowing longitudinal, or anisotropic, cell expansion (Brown, 1996). The cellulose polymer is arranged in 3-nm-thick microfibrils (Delmer & Haigler, 2002). These microfibrils are believed to consist of 8000 (primary cell wall) to 15 000 (secondary cell wall) glucose molecules (Somerville et al., 2004). Cellulose is synthesized by plasma membrane-localized proteins containing several structurally similar cellulose synthase (CESA) subunits (Arioli et al., 1998) that can be visualized as symmetrical rosettes of six globular complexes approximately 25–30 nm in diameter (Herth, 1983; Brown, 2004). The only known components of the CESA complex in higher plants are the CESA proteins, 10 genes of which have been identified in the sequenced genome of Arabidopsis thaliana (Richmond & Somerville, 2000). Advances in our understanding of the required stoichiometry of CESA subunits within the CESA complex have also recently been made demonstrating that the hexameric model for CESA complex formation requires three functional CESA subunits, with CESA1 and CESA3 compulsory, whereas CESA2, 5, 6, 9 and 10 are interchangeable (Desprez et al., 2007). It has also recently been found that the plant may have a cell wall sensing mechanism, requiring THESEUS1, to provide transcriptional feedback on the integrity of the cell wall (Hematy et al., 2007; Hematy & Hofte, 2008). Chemical and genetic screens for swollen organ morphology have also identified a handful of genes that participate in cellulose biosynthesis, such as the COBRA (Schindelman et al., 2001) and KORRIGAN (Lane et al., 2001; Paredez et al., 2008). However, the complexity of events contributing to activation of the CESA at the plasma membrane and its motility suggests that the list of players is far from complete.

Finally, of considerably greater potential benefit, and accordingly greater difficulty, is the possibility of changing the nature of cellulose itself. Could the cellulose synthase complex be altered to produce ‘wounded’ (in terms of either degree of crystallization or polymerization) cellulose more amenable to deconstruction? Would such a plant survive and thrive?

(Himmel et al., 2007)

The tight bonding capacity of the hydroxyl groups via hydrogen bonding are critical to determining how the crystal structure of cellulose forms and also in directing important physical properties of cellulose materials (Nishiyama et al., 2002, 2003). It is postulated that the chains of glucosyl residues in the fibril periodically fail to coalesce into an ordered crystalline structure; these amorphous zones along the fibril length are recognized as possibly facilitating the association between hemicellulose and cellulose fibrils (Gomez et al., 2008). Indeed, the regulation of amorphous to crystalline zones in the cellulose microfibril and the potential for biological regulation has important connotations for plant design and cellulose bioconversion, but as yet this area of research is poorly understood. Herein, we demonstrate the first example of a viable low-crystallinity biomass-crystallinity mutant in Arabidopsis. The mutant plant contains a point mutation in the transmembrane region of CESA3 in the previously identified mutant ixr1-2 (Heim et al., 1989, 1990). Specifically, the resulting amino acid residue change is from threonine 942 to an isoleucine. Through combined genetic and biochemical screening analysis, with the use of X-ray diffraction studies to allocate structural information about cellulose crystallinity, we demonstrate the capacity to modulate cellulose digestibility via this straightforward genetic modification of CESA.

Methods and materials


All chemicals and reagents used were of analytical grade or higher. Authentic samples of organic acids and their salts were obtained from Sigma-Aldrich, FMC-BioPolymer, Fisher Scientific, Riedel den Haan and BDH as applicable.

Plant material and sample collection

Mutants used in this study had either been previously published as, or were isolated as, homozygous T-DNA insertional alleles or point mutations screened from the ABRC stock center (Table 1). These plants were grown at 22 °C under a 16 h light 8 h dark regime in crops of 50 plants and harvested as aerial plant biomass. Four independent growing cycles and analysis were performed for top candidates.

Table 1.   Cell wall mutations in Arabidopsis thaliana and their source
GeneMutationGene IDSource
At4g32410Arioli et al. (1998)
CESA2T-DNA insertionAt4g39350SALK 096542 (ABRC)
At5g44030Scheible et al. (2001)
At5g64740Desprez et al. (2007)
CESA6T-DNA insertionAt5g64740SALK 004587 (ABRC)
CESA9T-DNA insertionAt2g21770SALK 046455 (ABRC)
CESA10T-DNA insertionAt2g25540SALK 052533 (Staffan Persson)
CESA2/PRC1-1  Persson et al. (2007)
T-DNA insertion
At5g54380Hematy et al. (2007)
Theseus5′UTR2569-2789Hematy et al. (2007)
At1g05850Zhong et al. (2002)
At5g60920Schindelman et al. (2001)
Cobra1-5cob1-5At5g60920Schindelman et al. (2001)
T-DNA insertion
At5g54690Persson et al. (2005)
At5g49720Lane et al. (2001)
T-DNA insertion
At5g49720Nicol et al. (1998)
CSLC4cslc4At3g28180Cocuron et al. (2007)
CSLC4/CSLC5 At3g28180/4g31590Double mutant courtesy of J. Milne

Microscale enzymatic saccharification

Celluclast 1.5 L (cellulase from Trichoderma reesei) and Novozyme 188 (cellobiase from Aspergillus niger) were purchased from Sigma-Aldrich (St Louis, MO, USA). The enzyme cocktail was obtained by mixing equal volumes of the two enzymes that contained an enzymatic activity of 486 endoglucanase units ( mL−1 for cellulase [45.6 filter paper units (FPU) – as standardized in LAP-009] and 134 CBU mL−1 for cellobiase. Enzymatic saccharification of lignocellulosic material was according to the laboratory analytical procedure #009 (LAP-009) of the National Renewable Energy Laboratory (NREL). The cellulose content of mutant and wild-type Arabidposis plants was measured spectrophotometrically (ThermoFischer Biomate3, Madison, WI, USA) on homogeneous samples of 150 ground whole Arabidopsis plants and sugar release values are a percentage of this total (described in detail below). Modification for the microscale experiment was made using dry biomass samples equivalent to 0.02 g cellulose. In addition, the total reaction volume was reduced to 2 mL and a range of enzyme concentrations based on cellulase activity were used including 60, 30, 15 and 7.5 FPU. The samples were positioned horizontally in an Innova 4300 incubator/shaker (New Brunswick Scientific, New Brunswick, NJ, USA) at 50 °C while shaking at 100 rpm using a 1-inch orbit. The progress of the reaction was measured by taking representative 100 μL aliquots at 2, 4, 6, 12, 24, 72 and 168 h. Enzyme blanks and Whatman #1 filter paper were digested alongside the samples.

X-ray scattering

Arabidopsis material was grown under both greenhouse and growth chamber for analysis. Plant samples were prepared by oven drying biomass at 60 °C for 36 h. Alternative temperatures for the drying regime were used, such as 37 °C for 7 days or 80 °C for 12 h followed by 110 °C for 2 h, neither of which altered the measured values in Arabidopsis tissue (similar to Harris & DeBolt, 2008). Tissue was then homogenized using an Arthur H Thomas Co Scientific grinder (Philadelphia, PA, USA) equipped with a 1 mm sieve. Biomass samples were then contained in a custom built sample holder of pressed boric acid. In brief, plant material was placed into a mold, containing a sleeve and hand pressed with a solid metal plug forming a disk shape. The sleeve and plug were removed and a boric acid (Fischer, Madison, WI) base was then formed by pouring the boric acid over the bottom and sides of the sample and applying 40 000 psi of pressure to the 40 × 40 mm mold using a Carver Autopellet Press (Wabash, IN, USA). Samples were pressed to create an even horizontal surface. A Bruker-AXS Discover D8 Diffractometer (Bruker-AXS USA, Madison, WI, USA) was used for wide angle X-ray diffraction with Cu Kα radiation generated at 30 mA and 40 kV. The experiments were carried out using Bragg-Brentano geometries (symmetrical reflection). Diffractograms were collected between 2° and 70° or 2° and 40° (for samples with little baseline drift), with 0.02° resolution and 2 s exposure time interval for each step. Sample rotation to redirect the X-ray beam diffraction site was achieved per replicate. The data analysis was carried out using the calculation for relative crystallinity index (RCI)=I002Iam/I002× 100, where I002 is the maximum peak height above baseline at approximately 22.5° and Iam is the minimum peak height above the baseline at ∼18° (Weimer et al., 1995). Peaks at the 22.5° and 18° were consistent with a control of synthetic crystalline cellulose (Avicel®, FMC-Biopolymer, Philadelphia, PA, USA) (Andersson et al., 2003). For assessment of experimental accuracy, the pressed samples were examined using reflective geometries at 22.5° 2θ with the sample scanned rotationally (360°) and in an arc (90°) to obtain an intensity/spatial orientation plot of a sample for which the RCI had already been established. The range of reflective intensities was then used to estimate the accuracy of the RCI determination using a 95% cutoff across the plot range (Harris & DeBolt, 2008). Three experimental replicates for mutant and wild-type biomass samples were performed. Diffractograms were collected in diffrac-plus-xrd commander software (Bruker-AXS, Karlsruhe, Germany) and minimally processed (baseline identification, noise correction, 3D display and cropping of RCI signature region) using the eva and texeval (Bruker-AXS) software.

Enzyme kinetics

The initial rate of sugar release using identical enzyme cocktail loadings (Celluclast 1.5 and Novozyme 188) as a function of substrate concentration was obtained by performing a similar microscale experiment as that conducted for the saccharification analysis. Dry biomass samples equivalent to sequentially increasing amounts of mutant and wild-type plant derived cellulose were mixed with enzyme (7.5 FPU) and incubated for 2 h. Similar to the NREL LAP-009 saccharification experiments and using the same enzyme buffer solution, the samples were incubated in an Innova 4300 incubator/shaker (New Brunswick Scientific) at 50 °C while shaking in a horizontal position at 100 rpm. The progress of the reaction was measured by taking individual aliquots at 2 h and determining the glucose concentration using a Yellow Springs Instruments (YSI)–glucose analyzer standardized for glucose determination using YSI buffer and membranes purchased from YSI (Yellow Springs, OH, USA). Enzyme blanks and Whatman #1 filter paper were digested alongside the samples. The inability to exactly calculate the number of catalytic ends in the complex mixture of cell wall biomass allowed only the calculation of a relative estimation, expressed as apparent kinetics values. Hence, classical Michaelis–Menten kinetics are not determined and the Km and Vmax values are apparent Km and Vmax relative estimates. The enzyme kinetics experiment varied the concentration of substrate (cellulose) and measured enzyme velocity to determine the difference in enzyme velocities between the two substrates. These values were generated using the statistical graphing program graphpad prism-4 (Graphpad, La Jolla, CA, USA).

Cellulose content measurement

Crude cell walls were prepared as published previously (Reiter et al., 1993). Briefly the samples for the measurement were homogenized using an Arthur H Thomas Co Scientific grinder (Philadelphia, PA, USA) equipped with a 1 mm sieve. Twenty-five milligram plant material were incubated in 1 mL 70% ethanol overnight at 65 °C, and washed twice with 1 mL 70% ethanol for 1 h and once with 1 mL acetone for 5 min. After washing solutions were extracted using a 1 mL pipette the samples were dried under vacuum. Cellulose content determination was essentially achieved using the methods described by Updegraff (1969): briefly, 5 mg of the dry biomass extract were weighed out in triplicates from either wild type or mutant plants and boiled in acetic-nitric reagent (acetic acid : nitric acid : water 8 : 1 : 2) for 30 min to remove lignin and hemicellulose. The samples were allowed to cool down to room temperature and the reagents were carefully removed. The plant cell wall material was washed twice with 8 mL MQ-water and 4 mL acetone, and dried under vacuum. The cellulose samples were then hydrolyzed in 67% sulfuric acid for 1 h. The glucose content of the samples was determined by the anthrone method (Scott & Melvin, 1953). Exactly 25 μL of the sulfuric acid hydrolyzed samples were mixed with 475 μL water and 1 mL 0.3% anthrone in concentrated sulfuric acid on ice. The samples were boiled for 5 min then placed immediately back on ice. The absorbance of the samples was measured using a Bio-Mate thermo scientific spectrophotometer (Thermo Fischer, Waltham, MA, USA) set at Abs620 and compared with a standard curve obtained from known (10–50 mM) concentrations of glucose (the standard curve was set each time together with the reaction). The cellulose content was calculated by multiplying the measured glucose concentration of each sample by the total volume of the assay and then by the hydration correction factor of 0.9 (to correct for the water molecule added upon hydrolytic release of each glucose residue from the cellulose polymer).

High-performance liquid chromatography (HPLC) analysis of fermentable sugars

Aliquots of the fermentable sugars released by enzymatic hydrolysis were isolated and their sugar composition quantified by HPLC according to methods of (Zhao et al., 2004). The enzymatic hydrolysates of 20 μL were injected into an eluent of 19 mM NaOH introduced at 1 mL min−1 using a Bio-LC HPLC (Dionex Corp., Palo Alto, CA, USA) and separated through anion exchange using a Carbo-Pac PA1 with guard column (Dionex Corp.). Signal strength from a pulsed electrochemical cell monitoring eluting sugars in column effluent amounts was estimated by numerical integration of the pulsed electrical cell monitor signal (chromeleon software version 6.80, Dionex Corp.). Sugars were identified and quantified by comparing their retention times and peak areas with that of known standards for glucose, xylose, galactose, fucose and rhanmose. Glucose was quantified by comparing the sample peak areas to that of known concentration standards.

Statistical analysis

Analysis of variance was conducted using the freeware statistical software package r (Auckland, NZ, USA) to test the null hypothesis of no statistical differences in RCI values between the mutant plants and wild type. The null hypothesis was rejected at the 0.05 level. Nonlinear regression analysis was performed using the statistics program built into graphpad prism.

Sequence alignment and analysis

The protein sequence for Arabidopsis was blast against all plant protein sequence data using the NCBI pblast format (Altschul et al., 1997). Sequences with significant homology and computational annotation as putatively encoding a CESA were aligned manually.


Cell wall mutant selection

To establish whether genetic mutations in genes central to cell wall synthesis could result in improved enzymatic saccharification due to a wounded or flawed cellulose fibril, previously generated Arabidopsis plants containing homozygous T-DNA insertions or point mutations in different genes potentially involved in cellulose biosynthesis as well as plants with mutations conferring resistance to cell wall synthesis inhibitors (Table 1) were selected. Selections were made based on both the type of mutation, being either a T-DNA mutation or a point mutation, and the coverage of the different CESA genes in the Arabidopsis genome as well as several mutants that we term ‘outliers’. These outliers are thought to be involved in cell wall synthesis but not directly in catalysis of UDP-glucose to cellulose (Table 1). The study was performed in Arabidopsis because identifying a gene mutation that results in a wounded cellulose fibril is very fundamental in nature and has been identified as a major challenge (Somerville et al., 2004; Himmel et al., 2007; Gomez et al., 2008; Harris & DeBolt, 2008). Moreover, the genetic resources that have been compiled over the past two decades provided a base of mutants with which to work. A vigorous and upright growth phenotype of the mutant plant was used as additional selection criteria. Mutants such as eli1-1 or cev1 summarized in (Robert et al., 2004), which have severe growth phenotypes due to impairment of CESA1 or CESA3, were therefore eliminated from the study. Severe phenotypic differences in growth habit were also avoided to the best of our ability. In order to control for differences in the stage of growth that might affect the results, all plants were harvested at maturity defined as the onset of senescence after seed maturity and maximal plant biomass have been reached. Harvesting at maturity was also an attempt to nullify any influence of spatial or temporal gene expression and focus on mutations that alter cellulose in the whole plant. To further reduce any bias resulting from phenotype each plant replicate was a crop of 50 plants grown under identical conditions that were dried and homogenized to create a pool.

Screening for altered digestibility

Based on the cellulose content determined for each mutant, identical cellulose loadings were analyzed for recalcitrance to saccharification after 24 h of enzymatic digestion using a commercial cellulase cocktail from T. reesei (Sigma-Aldrich). The results for each mutant were then compared as the percentage of cellulose converted into fermentable sugars (Fig. 1a). In order to confirm that cellulose was being converted, the fermentable sugars released by enzymatic hydrolysis were analyzed by HPLC, which indicated that >90% of sugar was glucose. The highest saccharification efficiency was 51.2% (±0.6) measured using ixr1-2 biomass relative to wild type, followed by cesa10-salk, cesa2-salk and ixr2-1, which displayed 14.8% (±1.2), 11.4% (±1.4) and 6.3% (±0.8) higher saccharification efficiency than wild type, respectively (Table 1 and Fig. 1a). The ixr1-2 mutant displayed greater than threefold improvement in saccharification than the next closest candidate (CESA10). A large majority of mutants analyzed displayed very similar and/or negative conversion potential relative to wild-type biomass (Fig. 1a). Lower saccharification efficiency relative to wild type was evident for cobra1-1, cobra1-5, the1-3, CESA6-salk, ctl1, rsw2-1, irx8, korrigan16-2 and clsC4 as well as the double mutants clsC4/cslC5 and cesa2/prc1-1 (Nicol et al., 1998; Lane et al., 2001; Schindelman et al., 2001; Zhong et al., 2002; Persson et al., 2005, 2007; Cocuron et al., 2007; Hematy et al., 2007) (Fig. 1a, Table 1). Batch comparisons from four independent plantings of mutant ixr1-2 vs. wild-type plants indicated that improved conversion efficiency remained at approximately 151% (Fig. 1b). From this saccharification efficiency screen of Arabidopsis cell wall mutants, we identified the missense mutant ixr1-2 as the top candidate for further investigation.

Figure 1.

 Improved biochemical conversion of dry biomass into fermentable sugars in the cell wall mutants. (a) Analysis of various Arabidopsis plants carrying mutations in a number of genes critical to cellulose biosynthesis by saccharification efficiency using hydrolytic enzymes. The graph plots fermentable sugars released relative to the conversion rate of wild-type biomass as a percentage (n=3) (b) Batch comparison between wild types and ixr1-2 mutant biomass over three generations of plants (B stands for batch)

Screening for altered cellulose structure by X-ray diffraction

A second screen for altered cellulose structure was performed on selected mutant plants identified in the saccharification efficiency screen. X-ray scattering analysis was performed to generate a RCI (Fig. 2) that established possible changes in the orientation, size or density of the cellulose crystallites (Andersson et al., 2003; Harris & DeBolt, 2008). X-ray diffraction patterns showed consistent signature peak distribution with previous published reports (Weimer et al., 1995) and these were overlayed with synthetic crystalline cellulose (Avicel) to determine the relative crystallinity index for synthetic crystalline cellulose (Weimer et al., 1995; Harris & DeBolt, 2008). The experimental accuracy was approximated by determining the noise in the diffractogram using a Phi and Chi scan (360° rotational by 90° in an arc in the X-ray diffractometer) of the sample, creating an intensity/spatial orientation plot at 22.5° 2θ and was determined to contribute approximately 10% error, which was added variability between replicates. Experimental accuracy value does not take into account the possibility that texture of the sample influences RCI (Andersson et al., 2003). Attempts to determine sample texture by transmission geometries (Saren et al., 2001) were not successful in the Arabidopsis plant material. Most technically challenging was mounting the sample, X-ray penetration through the sample and orientation of the fiber axis, which was random in leaf cells and highly diverse in a heterogeneous total plant sample. A lack of texture analysis in the samples by transmission geometries does not allow us to determine the preferred orientation of crystallites within the samples, nor the size or density of crystallites. Rather, RCI provided a screening tool for sample crystallinity of which the crystalline polymer within the complex mixture of polysaccharides is cellulose. Examination of mutant alleles compared with wild-type parental lines were performed in triplicate. The RCI of wild-type parental lines was measured to be 48.9±4.5%. Furthermore, five out of the seven plants displayed RCI values similar to that of wild type. Mutants analyzed were the nonconditional CESA1 allele rsw1-2 that displayed an RCI of 44.82±4.9%, cesa2-salk 46.3%±5.2%, CESA3 ixr1-2 allele 31.9±3.4%, CESA6 ixr2-1 allele 38.6±3.8%, CESA9 allele cesa9-salk 47.9±4.2%, and CESA10 allele cesa10-salk 50.0±5.2%. In addition the cobra1-1 mutant was analyzed and was not different to wild type (n=3) (Fig. 2).

Figure 2.

 Plots of the relative crystallinity index for the biomass sample of seven different mutant plants measured by X-ray diffraction (n=3). Selected from the saccharification screen, four out of the seven mutants (cesa2-salk, ixr1-2, ixr2-1 and cesa10-salk) yielded significantly more fermentable sugar than wild type. As controls, two mutants which yielded comparable amounts to wild type (rsw1-2, cesa9-salk) and one mutant which yielded less (cobra1-1) are represented.

Analysis of the FTVTSKA domain in higher plant genomes

CESA orthologs were identified based on homology to the CESA3 gene from Arabidopsis and amino acid sequences examined. Orthologs were identified in many sequenced or partially sequenced plant genomes available on public databases including Betula, Zea, Arabidopsis, Solanum, Populus, Vitis, Bambusa, Acacia, Eucalyptus, Oryza and Triticum. Analysis of sequence conservation among CESA genes revealed that the threonine-942 that was mutated to an isoleucine in the ixr1-2 mutant occurred within a highly conserved FTVTSKA domain among all sequences analyzed. A putative ortholog from the sequenced moss Physcomitrella patens also contained the FTVTSKA domain (Fig. 3). In Arabidopsis CESA3, the FTVTSKA domain is located in a cluster of six C-terminal transmembrane spanning regions and is specifically located on the extracellular loop between transmembrane spanning regions 3 and 4 (Fig. 3). Furthermore, analysis of the CESA protein family showed that several other CESA proteins including CESA1, CESA4 and CESA8 contained the FTVTSKA domain (Fig. 3, data not shown).

Figure 3.

 Sequence alignment and genetic mutation. (a) The region containing the ixr1-2 mutation was compared between Arabidopsis CESA3 and homologs from Betula, Zea, Arabidopsis, Solanum, Populus, Vitis, Bambusa, Acacia, Eucalyptus, Oryza, Triticum and Physcomitrella (b) Schematic of the structure of the CESA3 gene highlighting the region of the threonine to isoleucine mutation in the C-terminal transmembrane region (adapted from Scheible et al., 2001). (ZBD=zinc binding domain, HVR=hypervariable region, TSR=transmembrane spanning region at N-terminus and C-terminus).

Kinetic analysis of cellulose bioconversion in the ixr1-2 reduced crystallinity candidate

To gain a better understanding of the increased saccharification efficiency in the ixr1-2 mutant, a more detailed saccharification experiment was conducted to determine kinetics of the reaction by measuring eight time points from zero to 168 h and using four different enzyme concentrations ranging from 7.5 to 60 FPU. The results indicated that at the 168 h time point in conditions of excessive enzyme loading (60 FPU), the fermentable sugar released was over 50% for the ixr1-2 mutant as compared with approximately 30% for wild type (Fig. 4a). In addition, the ixr1-2 mutant released a greater percentage of sugar at each time point during the reaction (Fig. 4a–d) and more sugar was released by ixr1-2 using the lowest enzyme concentration (7.5 FPU) than that of wild type using the highest enzyme concentration (60 FPU) (Fig. 4a compared with Fig. 4d). The nature of the improved conversion efficiency was examined further by determining pseudo-apparent Michaelis–Menten kinetic parameters recognizing that there are inherent ambiguities with an insoluble substrate like cellulose and a multienzyme cellulase cocktail. Nevertheless, pseudo-apparent Km (Km) and Vmax (Vmax) values were significantly different between wild-type and ixr1-2 forms of cellulose. Wild type displayed a Vmax of 4.18 × 10−6 moles min−1 unit−1 protein−1 glucose and Km of 10.26 mg cellulose whereas ixr1-2 displayed a Vmax of 7.93 × 10−6 moles min−1 unit−1 protein−1 glucose and Km of 16.55 mg cellulose. There was a 61% increase in the Km for cellulose between wild type and ixr1-2 suggesting a significant reduction in the binding affinity for ixr1-2 cellulose. However, the Vmax for ixr1-2 cellulose was 88% higher than for wild-type indicative of a higher enzymatic turnover rate as might be expected if ixr1-2 is to be considered as a preferred substrate. A relative estimation of an apparent specificity-like constant (Vmax/Km) for wild type was 40.7 × 10−3 and ixr1-2 was 47.9 × 10−3 moles min−1 unit−1 protein−1 mg−1 (Fig. 5), representing a 17% improvement for ixr1-2 in the hydrolytic efficiency of the reaction.

Figure 4.

 Saccharification efficiency reported as % of cellulose converted to glucose at 2, 4, 6, 12, 24, 48 and 168 h using (a) 7.5, (b) 15, (c) 30 and (d) 60 FPU of enzyme from wild type (dashed line) and ixr1-2 (solid line) (error bars n=4).

Figure 5.

 Initial rate of sugar release from biomass by the enzyme mixture as a function of cellulose concentration. Wild-type biomass closed circles and dashed line; ixr1-2 closed squares and solid line (error bars n=3).


The overarching aim of this study was to establish whether mutations in genes central to cell wall synthesis could result in improved enzymatic saccharification efficiency of lignocellulosic biomass. This was possible due to the massive genetic infrastructure and research efforts invested in the model plant Arabidopsis over the past two decades. A central discovery was the identification of a missense mutant in CESA3 that resulted in marked 51% improvements in saccharification efficiency relative to wild-type biomass (Fig. 1a). Relatively minor effects on plant growth resulted from this mutation, namely ixr1-2, (Supporting Information, Fig. S1), possibly due to the mutation occurring far distal to the active site in an extracellular loop between two transmembrane spanning regions (Scheible et al., 2001). Further, ixr1-2 displayed a 34% lower relative crystallinity measurement compared with wild-type biomass; possibly providing a mechanism underlying higher saccharification efficiency. These results are discussed in detail below.

Screening for improved saccharification efficiency resulted in the identification of 4 mutants, namely ixr1-2, ixr2-1, cesa2-salk and cesa10-salk that yielded significantly more fermentable sugar than wild type (Table 1 and Fig. 1a). The ixr1-2 mutant contains a point mutation in CESA3, which is a compulsory subunit in the cellulose synthesizing machinery (Scheible et al., 2001). Null mutations in CESA3 are embryo lethal (Somerville et al., 2004) and thus it is likely that the ixr1-2 mutation is not completely deleterious to protein function, but more likely is altering a part of its normal structure or function. In contrast, the ixr2-1 mutant also confers resistance to isoxaben (Desprez et al., 2002) and contains a point mutation in the CESA6 subunit. Moreover, the CESA2 and CESA10 mutants, cesa2-salk and cesa10-salk, contained T-DNA insertions and are null mutants. These latter three plants were not selected as top candidates because CESA2 and CESA6 are partially redundant and gene expression is limited to elongating tissue (Desprez et al., 2007; Persson et al., 2007), whereas it is evident that CESA3 gene expression occurs in all tissues at all times (Persson et al., 2007). Thus, of the genetic mutations that were analyzed in this study, only one cellulose biosynthesis mutant was capable of providing a substantive improvement in the digestibility of cellulose. It is also noteworthy that a majority of the mutants analyzed displayed very similar and/or negative conversion potential relative to wild-type biomass (Fig. 1a) and that most of these mutants are knockouts beckoning the question as to the result of overexpressing these genes. Hence, from this initial screen for conversion efficiency ixr1-2 was identified as the top candidate due to its greater than threefold improvement in saccharification over the next closest candidate (cesa10-salk) and due to the intrinsic nature of CESA3 expression in all plant cells.

The determination of plant sample crystallinity through X-ray scattering analysis indicated that five of the seven mutant plants displayed RCI values similar to that of wild type, while two of the mutants, ixr1-2 and ixr2-1, were identified as having lower RCI values than wild type (Fig. 2). Interestingly, both of these mutants were previously identified in a forward chemical genetics screen as conferring resistance to the cellulose synthesis inhibitor isoxaben (Heim et al., 1989, 1990) and both mutant alleles displayed improved cellulose conversion efficiencies meeting the selection criteria for enhanced biofuel conversion (Fig. 1a). Why similar mutations in different but related proteins confer isoxaben resistance, increased saccharification efficiency and reduced RCI value is not clear. With respect to the latter two characteristics, perhaps these mutations create structural changes in the proteins that alter their orientation within the CESA membrane complex, giving rise to irregular angles at which they produce and incorporate their glucan chain into the growing microfibril. This repositioning could disrupt a certain percentage of hydrogen bonding within the microfibril causing an increase in amorphous zones along the fibril length which may contribute to a reduction in RCI in the biomass sample and an increase in accessibility to enzymatic hydrolysis. In addition, the partial redundancy between CESA6 and other CESA isoforms could help explain why our data indicates that ixr2-1 displays similar RCI and enzymatic hydrolysis characteristics as ixr1-2 albeit not to the same extent. At a fundamental level, the RCI determination accounts only for the volume fraction of cellulose within a complex matrix of cellular biomass. For this reason, we do not and cannot claim that absolute crystallinity of cellulose is lower, but rather that the relative crystallinity arising from the cellulose within the cell wall is lower in the ixr1-2 and ixr2-1 mutants. Consistent with RCI data (Fig. 2), evidence for altered cellulose structure in the ixr1-2 mutant was previously shown using Fourier transform infrared spectroscopy (FTIR) (Robert et al., 2004).

Kinetic analysis of the saccharification reaction indicated three areas where ixr1-2 showed marked improvement over wild type: fermentable sugar released at the 168 h time point was 151% that of wild type (Fig. 4d); at each time point a larger percent of fermentable sugar was released in ixr1-2 compared with wild type (Fig. 4a–d); more sugar released using eight fold less enzyme than wild type (Fig. 4a compared with Fig. 4d). At an application level, this gain in saccharification efficiency alone represents a dramatic improvement in bioconversion without costly chemical or heat pretreatment (Sticklen, 2008). Further analysis of the affinity of the enzyme cocktail for the cellulosic substrate was also suggestive of altered cellulose as the appVmax/Km value indicated a 17% increase in hydrolytic efficiency when using the ixr1-2 cellulose compared with wild-type cellulose (Fig. 5). Hence, analysis of pseudo apparent catalytic efficiency of cellulose breakdown provided a quantitative measure that the ixr1-2 cellulose substrate was preferred over wild-type cellulose by a cellulase cocktail.

Herein, we identify the first viable low biomass-crystallinity mutant in Arabidopsis and demonstrate the capacity to alter the efficiency at which cellulosic biomass is converted to fermentable sugars through the genetic modification of the primary cell wall CESA3. Creating plants with low lignin has also been shown to be a viable strategy to improve the efficiency of using lignocellulosic rather than starch based biomass for biofuel production (Chen & Dixon, 2007). Arabidopsis is not a biofuel crop and therefore long-term targeted mutagenesis studies to enhance biomass to fermentable sugar conversion in mainstream biofuel crops such as maize or switchgrass are needed in order to determine whether similar enhanced bioconversion can be obtained. More fundamentally, these findings raise intriguing questions regarding the point mutation in the extracellular loop between transmembrane regions 3 and 4 at the C-terminus of CESA3. Further, what, if any, is the importance of transmembrane anchoring in establishing the correct organization of the CESA subunits within the CESA complex? The ubiquitous nature of CESA3 orthologs in the primary cell wall of higher plants and conservation of the FTVTSKA domain (Fig. 3) further indicates that the outcome of this study could have high value in the development of feedstock grasses for both the biofuels and forage industry. An additional point of interest pertains to secondary cell wall cellulose, which occurs as wall thickenings of woody vascular tissue (Taylor et al., 2004). Because the ixr1-2 mutation occurs in the universally present primary cell wall CESA subunit number 3, it is quite plausible that efforts to generate the same amino acid change in the C terminal transmembrane region of the secondary cell wall CESA subunits (CESA4, 7 and 8) may provide a rational strategy to similarly improve the efficiency of biomass conversion from tree crops such as Populus.


Authors wish to kindly acknowledge the cell wall community in general, because the mutants that were analyzed herein are the collective effort of many labs over the past 2 decades. Funding was provided as startup funds to SD from The University of Kentucky, Department of Agriculture. We also thank Robert L. Houtz for critical reading of the manuscript and Randy Collins for technical assistance. Two reviewers and the editor are also acknowledged for suggestions that improved the manuscript. This article is published with approval of the Director of the Kentucky Agricultural Experiment Station as article number 08-11-118.