Biological conversion assay using Clostridium phytofermentans to estimate plant feedstock quality

Lignocellulosic plant biomass is comprised mostly of cell walls, which are a complex composite of proteins, lignin, and polysaccharides; the latter holds promise as raw material for biofuel production. The most abundant polysaccharide in the majority of tissues is cellulose, which exists as unbranched chains containing up to 15,000 β-(1,4)-linked glucose molecules [1]. By contrast, the shorter hemicelluloses are chemically and physically more complex [2]. The most abundant forms exist as glucan chains much shorter than cellulose or β-(1,4)-linked xylose, both with diverse side-chain substitutions of arabinose, galactose, fucose, xylose, or glucuronic acid. Biological conversion relies on an organism, such as a unicellular fungus or bacterium, which will convert these simple sugars to high-energy chemicals such as ethanol or butanol. Unlike seed starch or the soluble sugars found in phloem sap, the fermentable sugars found in cell walls are recalcitrant to extraction. The composition and interaction between the polysaccharides and lignin strongly influence their amenability for conversion to renewable fuels. Whereas lignification has extensive merits for the plant, it has adverse effects on the digestibility by ruminant and biofuel-generating microbes. For example, up to 50% of the variation in in vitro digestibility of commercial maize hybrids can be attributed to differences in their lignin content [3]. Lignin is composed ofthree monolignols: p-coumaryl, coniferyl, and sinapyl alcohols, which polymerize to form p-hydroxyphenyl, guaiacyl, and syringyl phenylpropanoid units, respectively [4]. The biosynthesis of alcohol monomers occurs in a specialized branch of phenylpropanoid metabolism, through which successive reductions, hydroxylations and methylations can occur. Crosslinking lignin with polysaccharides in the secondary cell walls of vascular tissue increases hydrophobicity, and thus gives these functional tissues the capacity to efficiently conduct water [4]. Concurrently, the polysaccharides are less accessible to enzymatic digestion or mechanical penetration by potential pathogens [5]. The pathway for lignin biosynthesis is well conserved among vascular plants, and involves at least 10 gene families, including CAD (cinnamyl alcohol dehydrogenase) and COMT[6]. Each step in the lignin pathway has been perturbed in various species, resulting in changes in lignin content, composition, and, in many cases, digestibility [7, 8].

Genetic diversity of plant cell-wall properties within species is evident in the decades of plant breeding for improved feed and forage quality in crops such as maize, sorghum, and alfalfa [9, 10]. The merits of animal feed have been tested frequently in vivo, either by evaluating animal performance in response to a particular feeding regimen, or by estimating digestibility in vivo using livestock with fistulae [11]. With the latter approach, the gastrointestinal tract of a surgically prepared animal is sampled to measure the remaining biomass. An equivalent in vitro method was developed using rumen fluid inoculum from fistulated cows [12]. Digestibility is estimated through analysis of the organic matter lost from the simulated ruminant gut conditions after 4 days of incubation. The throughput of this approach is considerably higher than in vivo, and begins to meet the needs of traditional plant-breeding efforts and genetics research. High-throughput assays to estimate feed and forage quality parameters also include compositional measurements of cellulose [13], total lignin and monomer content [14, 15], and hemicellulose content and composition [16, 17], which also serve as valid measurements of biofuel feedstock quality. Although the parallels between digestibility and amenability to conversion to biofuels might be apparent, industry standards for biofuel feedstock quality are still needed.

Regardless of species, all new crop varieties must meet certain standards for industrial-processing efficiency and consumer-market quality. Beyond the expectation of high biomass yield with few inputs on marginal land, conversion quality standards for energy crops have yet to be defined by the biofuels industry. Recently, several methods, including some high-throughout platforms, have been established that treat plant samples with hydrolytic enzyme cocktails, such as fungal cellulase and xylanase/xylosidase, then assay for total sugars as a measurement of digestibility [18–22]. This approach can be taken one step further, using translational assays that mimic industrial simultaneous saccharification and fermentation (SSF) paradigms, in which the liberated sugars from the polymers can then be fermented by Saccharomyces cerevisiae to measure total ethanol yield [23, 24].

A distinct and promising approach to cellulosic biofuel production is consolidated bioprocessing (CBP) technology for conversion of biomass to fuel. CBP could lead to a significant reduction in processing costs, greater than the reductions gained from any other potential improvement, such as reducing enzyme loading, eliminating pre-treatment, or improving the processes associated with converting sugars to ethanol [25]. The recently discovered anaerobic forest soil bacterium Clostridium phytofermentans may further enhance the efficiency of CBP. This organism produces ethanol as its major fermentation byproduct during growth on all substrates tested, including cellulose, hemicellulose, pectin, and starch [26], as well as switchgrass, corn stover, and pulp wastes ([Warnick and Leschine, unpublished data). Unlike S. cerevisiae, of which only engineered strains are capable of limited pentose utilization [27],C. phytofermentans directly converts a wide array of fermentable components of biomass to ethanol, including cellulose, pectin, polygalacturonic acid, starch, xylan, arabinose, cellobiose, fructose, galactose, gentiobiose, glucose, lactose, maltose, mannose, ribose, and xylose [26]. Without the addition of exogenous cellulases and xylanases, CPB using C. phytofermentans can yield approximately 70% of the yield of SSF using engineered S. cerevisiae[28].

In this paper, we report on an assay that provides the ability to measure the influence of variation in biomass composition, pre-treatment methods, and conversion processes on digestibility, and thereby determine the potential effects of numerous variables in biofuel production. In addition to sorghum, feedstock quality was evaluated for cultivars of shrub willow (Salix spp.) and Brachypodium distachyon accessions to demonstrate the applicability of this assay for a wide range of feedstocks, from woody crops to herbaceous grasses. The C. phytofermentans bioassay provides a direct and quantitative means of assessing feedstock quality, both in terms of digestibility and conversion.

Some key crop species, such as maize, wheat, and rice, have been domesticated from about 10,000 years ago, and have been under continuous selection for important agronomic and quality traits. Whereas progenitor species of many crops are unrecognizable from their domesticated counterparts, many of our potential energy crops are at best just a few crosses away from the wild accessions. Miscanthus, switchgrass, energy cane, and willow are just beginning the process of domestication. Thus, there is equivalent potential for improvement of fuel yield per unit of land through breeding of energy crops, but this must be accomplished in an abbreviated timeframe to have an effect on global climate change [32, 33]. One trait to improve upon is the ease of biomass deconstruction to simple sugars that are suitable for conversion to a liquid fuel [34]. Many species have demonstrated genetic variation in digestibility for both forage quality and enzymatic digestion, thus traditional plant breeding facilitated by suitable phenotypic assays, such as the bioconversion approach described here, will lead to a more efficient biofuels industry [3, 35]. The overall content and composition of cell-wall lignin has proven to be a significant determining factor for lignocellulosic biomass digestibility [7, 8]. In the 1920s, agronomists identified the first lignin mutants in maize as genotypes with characteristic browning of the leaf midrib, hence the mutant name 'brown midrib' [36]. Negative perturbation of multiple points in the lignin biosynthetic pathway results in increased digestibility in numerous species including maize, sorghum, switchgrass, tall fescue, alfalfa, tobacco, and poplar [4, 7]. In both sorghum and maize, mutations in the CAD or COMT genes have been identified as causes for the brown midrib phenotype [37–39].

The bioconversion assay described here was developed using a well-characterized set of sorghum bmr NILs, the same samples previously described by Dien. [23]. In this study, the Klason lignin content was 15% lower in the bmr-6 and bmr-12 single mutants than in wild-type sorghum, and nearly twice as low (27%) in the bmr-6/bmr-12 double mutant. Interestingly, no other differences were between the NILs were seen for carbohydrate content or non-lignin-associated measures of in vitro digestibility, such as neutral or acid detergent fiber [23]. Therefore, the differential effects of C. phytofermentans inoculation on NILs is not due to the total amount of digestible sugars in the NILs, but rather the accessibility of those polysaccharides to enzymatic digestion, which is therefore strongly influenced by lignin content. Fermentation of the same material by S. cerevisiae after dilute acid pre-treatment and cellulase and β-glucosidase enzymatic digestion produced improvements in ethanol-conversion efficiency of 11% and 17% for the single and double mutants, respectively [23]. The differences in ethanol production by NILs digested by S. cerevisiae were far less pronounced than those digested with C. phytofermentans. Considering that xylose yield was roughly constant across the NILs, and that S. cerevisiae is capable of fermenting hexoses only, it is likely that the differences in ethanol yield were due to differential enzymatic release of glucose. C. phytofermentans digests plant biomass with secreted cellulolytic enzymes, and subsequently imports and metabolizes pentoses and hexoses. Thus, ethanol yield is a combined measure of the digestibility of hemicelluloses and cellulose.

To gain a greater understanding into why ethanol yields might be greater with the bmr mutants, we assayed the polysaccharide content both before and after inoculation, using a comprehensive collection of 147 glycan-directed monoclonal antibodies to quantify changes in most major classes of cell-wall polysaccharides [29]. Considering the minute quantities of pectins in sorghum stems [40], we focused on changes in the abundance of xylan, which accounted for a substantial proportion of the overall biomass content [23]. The relative amount of extractable xylan differed between the NILs, as did the difference in extractable xylan metabolized by C. phytofermentans. Therefore, the increased quantities of ethanol produced from the bmr mutants is in part due to the increased accessibility of xylan to hydrolysis.

Although it was apparent that there are genetic differences in plant properties that influence feedstock digestibility, several analyses of sorghum, maize, and alfalfa lignin mutants revealed genotype × chemical pre-treatment interaction effects. In other words, the magnitude of the differences between genotypes or even their rank order will change depending on the method of pre-treatment. Enzymatic release of glucose from the maize lignin mutant bm1 was equivalent to that of wild type after alkali pre-treatment, but significantly greater after acid pre-treatment [19]. Relative to untreated sorghum, the difference between wild-type sorghum and the sorghum lignin mutants bmr-3 and bmr-12 diminished after dilute acid pre-treatment [41]. In the same study, mutations in other loci resulting in a bmr phenotype produced little or no change in response to pre-treatment. Similarly, the magnitude of the difference in enzymatic saccharification between wild-type alfalfa and plants downregulated for several different lignin genes varied in response to acid pre-treatment [8]. In the case of plants downregulated for CCoAMT, a significant difference from wild type was seen only in the absence of pre-treatment. By contrast, the difference in hydrolysis efficiency between C4H and HCT downregulated plants and wild type was twice as large in the absence of pre-treatment. After treatment with 15% ammonium hydroxide the hydrolysis yield potential of a large panel of sorghum accessions ranged from the same level as that of the untreated sorghum up to an approximately threefold increase over the untreated plants [42]. The goal of pre-treatment is to improve the accessibility of the polysaccharides to hydrolysis [43]. Clearly, there is a complex relationship between efficiency of feedstock utilization and pre-treatment. In this study, we attempted to maximize the ease of conducting the bioconversion quality assay and to maximize the sensitivity of the assay to detect genetic differences. Although small differences in genetic potential may not be relevant in certain industrial scenarios, detection remains important for development of energy crops. Consistent and incremental progress in modern plant breeding is the product of changing the frequencies of alleles with small additive effects. Thus, a sensitive assay is required to detect such variation, and certain pretreatments may diminish sensitivity.

We measured significant genetic variation among collections of two plant taxa considered to have excellent potential as energy crops, sorghum and shrub willow, and a model for grasses, B. distachyon[44–46]. The variation was notable considering that the samples tested by no means represent the overall genetic variation within each taxon. Thus, we would expect a wider range of phenotypic variation with an increase in sample number and diversity. The ethanol yields from sorghum were similar to those from willow, ranging from 26.5 to 39.4 mg ethanol/g feedstock. For willow, no significant correlations were seen between ethanol yield and biomass compositional data (Serapiglia and Smart, unpublished data). Most of the variation could be attributed to genetics rather than to experimental differences; consequently, routine gain from selection could be expected in the range of variation detected. At the same time, the assay could clearly identify differences between major genetic perturbations in biomass composition (typical of the bmr mutations) and continuous variation among natural and segregating populations. The most digestible genotypes from all three biofuel crop taxa yielded approximately 25% more ethanol that the most recalcitrant. This is only slightly less than the differences between wild-type sorghum and the single-gene bmr mutants. The quantity of material tested is well within the amount of stem biomass typically produced by a single plant of rapid-cycling model plants such as Arabidopsis thaliana and B. distachyon. We are currently using this system in a 96-well format, which, combined with a commercial robotics preparation [19, 22], is capable of throughput levels required by a core feedstock testing facility for plant breeding and mutant screens.

In this paper we have described a laboratory assay that quantifies the amount of ethanol produced by C. phytofermentans cultured on plant biomass. This approach may be useful to estimate the potential effects of pre-treatment, conversion methods, and microbial and plant genetic diversity on biofuel manufacturing. The assay is capable of measuring subtle genetic differences within energy crops and the research model species B. distachyon, making it particularly useful for genetic analysis of the mechanisms underlying plant feedstock amenability to biological conversion. The C. phytofermentans bioassay provides a means of assessing feedstock quality in terms of both digestibility and conversion.