Using plants as biofactories for industrial enzymes is a developing technology. The application of this technology to plant biomass conversion for biofuels and biobased products has potential for significantly lowering the cost of these products because of lower enzyme production costs. However, the concentration of the enzymes in plant tissue must be high to realize this goal. We describe the enhancement of the accumulation of cellulases in transgenic maize seed as a part of the process to lower the cost of these dominant enzymes for the bioconversion process. We have used breeding to move these genes into elite and high oil germplasm to enhance protein accumulation in grain. We have also explored processing of the grain to isolate the germ, which preferentially contains the enzymes, to further enhance recovery of enzyme on a dry weight basis of raw materials. The enzymes are active on microcrystalline cellulose to release glucose and cellobiose.
The heavy reliance of western economies on fossil fuels has given rise to energy security concerns. These concerns together with limited petroleum reserves, negative environmental impacts and the rising cost of petroleum have prompted the development of viable alternatives that are cleaner and environmentally neutral. Biofuels and biobased products have emerged as one of the alternatives to address these concerns. They are obtained from renewable biomass, a key long-term component of a sustainable industry. Sustainable, renewable resources are those derived primarily from plant biomass, re-produced with minimal inputs using energy from the sun. Biomass for biofuels and biobased products can include many sources of material: agricultural harvests such as grains, agricultural residues such as stalks and leaves, perennial crops such as hay and trees, building waste wood, municipal solid waste such as paper, and various food industry processing wastes.
Ethanol from corn is the first-generation biofuel produced in the United States and Europe. However, grain-based fuels alone will not be sufficient to address the nation’s energy needs. Biofuels must also be produced from renewable cellulosic feedstock materials to reach federally mandated levels of alternative fuels by 2022 (i.e. RFS2). The ‘Billion Ton Study’ published by the US Department of Energy and United States Department of Agriculture (USDA) estimated that sufficient renewable cellulosic resources are available from forest and agricultural lands to sustainably produce enough biofuel to displace 30% or more of US transportation fuel needs annually (Perlack et al., 2005). This could significantly reduce the US dependence on imported oil.
However, plant biomass is a very complex matrix of polymers comprising the polysaccharides cellulose and hemicellulose in addition to lignin as the major structural components. Although cellulose is a simple, linear polymer of glucose, its semi-crystalline structure is notoriously resistant to hydrolysis by both enzymatic and chemical means. Thus, a strategy designed to use cellulosic material for biobased products must include the ability to efficiently and inexpensively convert the polysaccharide components of plant cell walls into simple sugars.
The preferred method for deconstructing plant cell walls is to pretreat the biomass with steam pressure often in conjunction with an acid or base catalyst (Saville, 2011). The cellulose in this pretreated biomass is then digested with enzymes (cellulases) to release the glucose for fermentation. Pretreatment and saccharification technologies are two of the three cost-limiting factors for cellulosic ethanol industry deployment, the third being delivered cost of feedstock materials. A number of enzymes are required to break down cellulose and work synergistically including endocellulase, exocellulase and glucosidase (Wood and Ingram, 1992). Other proteins, such as expansin, may also be helpful in facilitating this process (Cosgrove, 1999).
Cellulases are a subset of the glycosyl hydrolase superfamily of enzymes that have been grouped into at least 115 families based on amino acid sequence similarity, enzyme reaction mechanism and protein fold motif (Beguin and Aubert, 1994; Hayashi et al., 2005). Two specific cellulases, endocellulase E1, from Acidothermus cellulolyticus and cellobiohydrolase I, or CBH I, from Trichoderma reesei have been shown to be effective and synergistic partners in the hydrolysis of cellulose from biomass (Baker et al., 1998).
The current trend to use biofuels to offset the dependency on petroleum has catalysed intensive efforts to find new enzymes, feedstocks and microorganisms that will be more efficient and cost-effective in the breakdown of cellulose. This is challenging because the complex nature of crystalline cellulose, as well as the particular source of feedstock, offers unique structures and contaminants to the enzymes, affecting their efficiency (Beguin and Aubert, 1994; Kabel et al., 2007). An alternative approach to finding more unique enzymes is to find an inexpensive way to produce the huge volumes of enzymes required for the industrialization of biomass conversion.
Two hundred and thirty-five (235) million dry tonnes of biomass would be needed to produce 20 billion gallons of ethanol, assuming 40% cellulose and 85 gallons per tonne (considering the glucose fraction only) at 80% cellulose conversion into glucose (Howard et al., 2011). The amount of enzyme needed per gallon of ethanol to convert a feedstock is a moving target based on the continued improvement in enzyme mixes and the types of feedstocks. Early estimates by the National Renewable Energy Lab (NREL; 119 g/gallon with a predicted threefold increase in activity) and current enzyme manufacturing recommendations indicate that a reasonable assumption would be in the range of 30 g/gallon. With this assumption, the total amount of enzyme required for this volume of ethanol would be approximately 0.6 million tonnes. If these enzymes were to be produced by conventional fungal fermentation, the tanks and instrumentation alone for this much enzyme would cost nearly $30 billion before enzyme production processes even begin (based on calculations in Howard et al., 2011). The agricultural bioproduction system offers the potential for a viable and scalable alternative to lower the cost of enzymes for biomass deconstruction.
A plant-based system for the production of industrial enzymes has numerous advantages such as low capital investment and easily scalable large volume production. The ability to achieve high levels of accumulation of the enzymes in grain allows for their production at extremely low cost in a form that is easily harvested, stored and transported (Howard et al., 2011). Plant production systems that accumulate cellulase in the normally unused or low value portion of the plant are now approaching competitive cost structure with microbial systems. As expression technology improves and cellulase reaches 4% of the dry weight of the plant organ of choice, direct delivery of plant tissue can be the enzyme delivery system of choice and will easily reach cost targets set by NREL (Howard et al., 2011).
The practical aspects of producing enzymes in commodity crops and the associated logistics have also been demonstrated, particularly as they apply to cellulosic ethanol (Howard and Hood, 2005, 2007). We demonstrate in this study the use of germplasm and genetics to increase cellulase expression to generate very high accumulation levels of enzyme production in the field and the utility of these enzymes in biomass deconstruction experiments. Increases in target protein accumulation on a dry weight basis decrease the cost of production of enzymes for industrial applications. In order for the plant production system to be competitive with other production systems, the cost on a weight basis must be low, and the way to achieve this is to create plant lines with very high protein accumulation amounts. Thus, the selection for increased enzyme amount in the seed of these transgenic corn plants is critical to commercialization success. These plant-made enzymes can be a major enabler for the biomass to bioproducts industry.
Enhancement of cellulase activity using germplasm
An independent transgenic event for endocellulase E1, referred to here as BCH01, and an independent transgenic event for cellobiohydrolase, or CBH I, referred to here as BCC02, were backcrossed for six generations into each of two elite inbred germplasm lines, a Lancaster (SP122) and a Stiff Stalk (SP114), to improve the agronomic performance of the transgenic lines in the field. The progeny of these crosses were concomitantly selected for increased target protein accumulation in seed at each generation.
The breeding scheme is illustrated in Figure 1. Tissue culture-derived plants are termed T0 or transformed first-generation nonseed-derived plants. Up to 10 plants per independent transgenic event were recovered, grown in the greenhouse, and pollinated with elite inbred pollen of the Lancaster parent (SP122) only. Six individual seeds from the ears of those plants were analysed for cellulase content by enzyme assay (Hood et al., 2007). The three highest-expressing ears per event, based on the individual seed assays, were chosen for continued propagation. Approximately 60 of the remaining seeds of these high-expressing individuals, 20 from each ear, were planted in the greenhouse and leaf-painted with Liberty™ herbicide to identify the segregating transgenic and nontransgenic progeny. The plants showed 1 : 1 segregation when treated with glufosinate ammonium, or Liberty™. Resistance to this herbicide is provided through the phosphinothricin acetyl transferase (pat) gene (Hood et al., 2007) that is linked to the cellulase gene cassette.
A fifty seed pool from each ear harvested from the greenhouse-grown plants was ground and analysed for total extractable protein and enzyme activity level. Enzyme values are expressed per unit dry weight of grain flour. Based on this analysis, up to 15 high-expressing T2 ears [nomenclature described in (Meksem and Kahl, 2010)] were selected from progeny derived from T1 plants pollinated with each inbred parent (SP122 or SP114) to continue in the backcross programme in the field. At this point, SP122 (Lancaster) lines had two doses of inbred, and SP114 (Stiff Stalk) lines had one dose of SP122 and one of SP114.
A single copy of the insert is present in each transgenic line (Hood et al., 2007). Thus, dosage of the linked pat and cellulase genes described above is one copy in half the seeds (see Table 1) because nontransgenic inbred pollen is used to produce seed on the hemizygous transgenic plants. This condition remains throughout the backcross programme until the plants are self-pollinated in the later generations. When homozygous plants are recovered in T8, they contain a pair of transgenic loci in every individual, and thus, the gene dosage increases from the original hemizygous ½ N to a final 2N. The relative expression could be as much as four times greater than the hemizygous ears assuming a linear correlation between the gene dosage and cellulase levels.
Table 1. Increase in transgene copy number with self-pollination of backcrossed inbreds and hybrid seed for production
*Individuals in the population either have 1 copy of the gene (hemizygous) or are null.
†In the first selfed populations, ½ are still hemizygous, ¼ are homozygous and ¼ are null, generating the apparent equivalent of 1 copy of the transgene in each plant (or N). In the second selfed populations, the hemizygous plants will produce the same ratio of progeny, but the homozygous plants will produce 2N progeny. Homozygous individuals are selected by herbicide screening for the inclusion in hybrid production.
2 to 4
Hybrids for production
T8 and beyond
T8 and beyond
During each planting season, one to several rows of 20 seeds from each selected ear was planted, and all were treated with Liberty™ herbicide. Surviving plants were manually crossed with the appropriate recurrent parent and individually harvested, yielding from 75 to 200 ears per inbred parent. After analysis, ranking and selection based on high protein accumulation, seed from chosen individuals was then replanted in the following season’s nursery (Figure 1).
Data for all ears from a single original plant were compared across generations to assess the overall increase in target protein accumulation. Using this approach, the total amount of target enzyme recovered per ear per generation can be ranked, and the highest-expressing transgenic lines selected. After three backcross generations, progeny from one specific plant line derived from a single transgenic event expressing each enzyme, BCH0101 (E1) or BCC0206 (CBH I), were continued in the breeding programme because they showed the largest increases in expression and the best agronomic phenotype. Although abiotic factors likely influence protein accumulation on an individual plant basis, cumulative data for progeny from a single original line show statistically significant increases in protein accumulation over generations.
For E1-expressing lines using elite germplasm as the recurrent male parent, cellulase increased to approximately 0.10% of dry weight in the highest lines harvested (Figure 2). Although significant differences in target gene expression level were not observed in the entire population of ears in generations T4-T6, they were all significantly different than the T1 generation at the 95% confidence level. The highest-expressing ears were sevenfold higher than the average protein accumulation level of 0.014% dry weight in first-generation (T1) seed. Because this T6 seed was still segregating 1 : 1, when the lines were self-pollinated to complete the breeding programme, up to four times the gene dose would be expected in the final homozygous hybrid, which should promote enzyme accumulation even further (Table 1). We have not observed gene silencing in seed-expressed transgenes in homozygous condition previously or in this project. Indeed, the first selfed generation (T7) of the Lancaster (SP122) parent increased the gene dosage to ∼1 N and the amount of E1 enzyme to 0.15–0.2% DW of grain (Figure 2), significantly different (95% confidence level) than the hemizygous seed in the T6 generation.
Similar experiments were conducted with the CBH I lines. The parental inbreds each achieved accumulation of enzyme to approximately 0.2% DW of grain flour, a 20-fold increase over the original T1 amount of approximately 0.01% DW (Figure 3). T5 and T6 generations were not significantly different from one another, but in both inbred backgrounds, were significantly different than T1. The Lancaster (SP122) inbred is one breeding generation ahead of the Stiff Stalk (SP114) lines because it was the original male parent of greenhouse plants from tissue culture. T7 plants are the progeny of selfs of T6 plants (Figure 3a), and the enzyme accumulation levels were as high as 0.45% of dry weight of grain flour or twice the levels observed in the previous generation and significantly different from the T6 generation at the 95% confidence level. This is expected because the gene dosage is increasing (Table 1).
Alternative germplasm can also increase yields of protein from transgenic maize, as has been demonstrated for high oil corn (HOC) (Hood et al., 2002). Therefore, we tested the effect of high oil germplasm on the accumulation of cellulases in our transgenic lines. Each of the four transgene/inbred combinations was crossed to the HOC line—those yielding the most ears are shown (Figure 4). The effect of high oil germplasm was dramatic in a single cross using the high oil line as the male parent for both cellulase enzymes—E1 and CBH I (Figure 4). E1 accumulated to approximately 50% higher levels, and CBH I accumulated to nearly 2.5- to 5-fold higher levels compared with elite crosses of the same generation (significantly different at the 95% confidence level). High oil germplasm generally produces seed with large embryos that have increased oil content. Because our transgene product accumulates in embryos, one might argue that the increased size alone may account for the increase in protein. However, in previous experiments, we have determined that factors other than embryo size alone account for this phenomenon (Hood et al., 2002).
Hybrid production and fractionation of grain
The purpose of the backcross programme is to produce transgenic inbred lines that can be used to generate transgenic hybrids with good agronomic performance that yield at parity with current hybrids. These hybrids should also contain high concentrations of the target proteins in grain. A hybrid was prepared at the T5 generation for each transgene [BCC0206 (CBH I) Stiff Stalk by Lancaster; BCH0101 (E1) Stiff Stalk by Lancaster] and used for production on an Arkansas farm in summer 2009. The amount of enzyme recovered on a grain dry weight basis in this trial production was extremely encouraging (Table 2)—at this rate, E1 (0.086% DW) would be recovered at approximately 1 kg per tonne of grain and CBH I (0.176% DW) would be nearly 1.8 kg per tonne. These hybrid production levels are approximately equivalent to those seen in the highest T5 ears as shown in Figures 2 and 3, clearly showing that the expression levels can be maintained under nonselective production conditions. Also, from the point of view of land-use efficiency, assuming 150 bu/acre grain yield, these expression levels translate to approximately 3.2 kg/ac for E1 and 6.6 kg/ac for CBH I.
Table 2. Yields of E1 and CBH I in hybrid germplasm and fractions therefrom
Transgenic grain fraction
CBH I exocellulase
Mean % DW ± standard deviation
*The milled fraction from Satake maize de-germinator prior to separation of germ.
0.086 ± 0.001
0.176 ± 0.05
Endosperm flour (grits)
0.01 ± 0.002
0 ± 0
Mixed embryo fractions*
0.175 ± 0.02
1.143 ± 0.143
0.463 ± 0.021
1.5 ± 0.576
Because the enzymes are predominantly confined to the embryo (germ) of the kernel, we tested dry mill fractionation of the grain to enrich the concentration of the enzymes in dry matter. Approximately 500 kg of grain for each hybrid was processed at 20% moisture content through a Satake De-Germinator at their facility in Stafford, TX. Two fractions were recovered—one with primarily endosperm (grits) that is mostly starch and a second that contains the protein-rich germ/oil fraction also containing small endosperm pieces and powdered starch (mixed embryo fractions; Table 2). Each fraction was analysed for extractable cellulase activity. The endosperm had no appreciable activity of either enzyme. The mixed embryo fractions from the de-germer amounted to only 0.25 tonnes each and thus were not a large enough volume to process on a gravity table for germ separation. Therefore, as a first step to assess germ-specific enzyme activity, embryos were hand-picked from the mixed fractions and analysed. On a dry weight basis, E1 embryos contained approximately five times more activity than the whole grain (Table 2), whereas CBH I germ exhibited a ninefold increase in activity over the whole grain. It is not clear whether these differences in germ enrichment of enzyme are because of differences in handling during processing or to impurities in the germ that may differentially impact enzyme activity assays.
Quantity of enzyme in grain
Total enzyme available in the grain has an impact on commercial value of the material. To determine the absolute quantity of the enzymes in grain, an exhaustive sequential extraction experiment was conducted. CBH I and E1 from corn meal were extracted five successive times, and protein detection was performed by enzyme assay and Western blot analysis on each separate extract (Figure 5). Sixty-four per cent of the CBH I activity was recovered in the first extraction and subsequent extractions resulted in 26%, 8% and 2% of the total (Figure 3a). The fifth extraction did not show any CBH I enzyme activity. A similar trend was observed in the case of E1, 63% of the E1 activity was recovered in the first extraction, and subsequent extractions resulted in 23%, 9%, 4% and 1% of the total E1 activity. Results of Western blot analysis showed a strong correlation with the enzyme activity results (Figure 5b).
Purification of enzymes
To characterize the enzyme activity on cellulosic substrates, enzyme was purified from grain samples. Purified, concentrated protein fractions were analysed by SDS–PAGE. Each protein showed >95% purity by Coomassie blue staining (Figure 6). The size of E1 (42 kDa) indicates that it is the catalytic subunit without the cellulose binding domain. CBH I at 53 kDa is intact with (apparent) multiple glycosylated forms.
Activity of enzyme on a cellulosic substrate
The data presented above assume that the activity of the enzymes on soluble substrates (MUC) is similar to the activity that catalyses cleavage of beta-1,4-glycosidic linkages in cellulose chains from cellulosic substrates. To verify this critical assumption, we first tested the ability of the plant-produced cellulases to release fermentable sugars from cellulose as the sole carbon source using the microbial growth assay as described earlier (Jimenez-Flores et al., 2010). This experiment also provides evidence that in the process of releasing free sugars, no interfering compounds are released that would significantly interfere with the growth of an ethanol-producing microbe. The release of free sugars from cellulosic material requires at least three enzymes: an endocellulase (E1), an exocellulase (CBH I) and β-glucosidase to hydrolyse cellobiose into monomer glucose molecules. Therefore, we added β-glucosidase to our purified, plant-produced E1 and CBH I to ensure that fermentable sugars were available for the yeast in the microbial assay. The results in Figure 7 illustrate that yeast can grow on the commercial preparation of cellulase (Celluclast, which is known to contain beta-glucosidase activity) as well as the plant-produced preparations. Cultures with no enzyme and cultures with β-glucosidase alone acted as controls to identify background fermentable sugars in the assay.
While the microbial growth assay confirms the ability of these enzyme preparations to generate fermentable sugars from cellulose, it does not positively identify which sugars are present. Glucose oxidase (GLOX) is a convenient assay that is specific for glucose (http://www.worthington-biochem.com/gop/default.html). Therefore, this assay was used in experiments similar to those in Figure 7 to measure the amount of glucose released from cellulose by the enzymes. The results in Figure 8 illustrate that both the plant-produced cellulase mixture and the commercial preparation of cellulase were capable of releasing glucose in amounts that were significantly greater than the controls. No attempt was made to make these samples equal in protein concentration. The combination of proteins in commercial cellulase is complex and not well-defined, and the corn-derived enzymes do not have all the factors present in the commercial cellulase.
In a final confirming experiment, free sugars released by the enzyme treatments were quantified by conventional HPLC analysis. The results in Figure 9 demonstrate that the commercial cellulase provides a significant amount of glucose (minimum detectable limit = 0.065 mg/mL) after 24 h and accumulating to over 0.1 mg/mL by 72 h and beyond. Cellobiose concentrations released by the commercial cellulase were much higher, accumulating to over 0.4 mg/mL. The plant-produced cellulase (E1 + CBHI) that also contained microbially produced β-glucosidase showed approximately the same amount of free sugar released but in this case it is almost all in the form of glucose rather than cellobiose. The abundance of additional β-glucosidase that converts cellobiose to glucose included with the plant-produced enzymes is a likely cause for this result. This result also explains why the plant-produced enzyme appears to be much more active than the commercial cellulase in the previous test because the GLOX assay only detects glucose and not cellobiose.
With the current state of technology for biomass conversion, the overwhelming enzyme requirement is for cellulases: endocellulase, exocellulase and β-glucosidase (Merino and Cherry, 2007). The specific activity of commercially available cellulases is quite low (Jorgensen et al., 2007; Sticklen, 2008), and considerable effort has focused on increasing their activity levels. However, even with improved enzymes and improved methods of production, the amount of cellulase required to deconstruct the volumes of biomass necessary for 30% replacement of transportation fuel (the 2022 RFS2 target) is in the millions of tons per year.
This is an unprecedented challenge in terms of the amount and the extremely low cost of enzymes that is required for competitively priced biofuels and biobased products. Moreover, to produce the enzymes through conventional fungal fermentation, the amount of upfront capital investment required for fermenter capacity is daunting. This situation has led many groups to investigate ways to reduce this cost burden. Reports continually appear of improvements from many groups on different enzyme cost reduction technologies. One solution described here is through plant-produced enzymes.
Plant bioproduction of industrial enzymes offers an alternative to fungal fermentation. The advantages of the plant system include high expression levels in seed (Clough et al., 2006; Hood et al., 2007; Miles et al., 2007), established infrastructure for growing and processing the crops on a commodity scale, a stable and easily transportable and processable production package (i.e. seed), easy scaling up or down to meet market demand, and no capital investment for production. An additional advantage for corn is the ability to enhance the levels of target protein accumulation to a high degree in production lines through breeding and selection, which will help to minimize the footprint of this technology on productive land.
Maize genetics is a powerful tool with which to manipulate the expression of input and output traits engineered through biotechnology. For the transformation-competent germplasm of maize (Armstrong et al., 1991), field performance characteristics are poor. Nevertheless, grain yields similar to those of commercial hybrids are imperative for the production of genetically engineered lines and must be improved through transfer of transgenic traits into elite, high-yielding germplasm. We conducted a breeding programme to improve agronomic performance, enhance grain yields and increase cellulase protein accumulation in grain. We have also demonstrated the quality of activity of the cellulases produced in the grain. The quality and phenotype of the inbreds and hybrids produced from these crosses matched the wild-type germplasm as would be expected from an increase of 98% elite inbred genes into the transgenic lines. Moreover, parallel increases in transgenic protein accumulation could also be selected.
In this study, the increase in enzyme accumulation through backcrossing to elite inbreds was sevenfold for E1 and approximately 20-fold for CBH I. In each case, the potential increase with self-pollination of the inbred lines is an additional two to fourfold, and we have demonstrated up to a twofold with the first self-pollinated generation (Table 3; Figures 2 and 3). Thus, the potential increase over T1 expression in elite grain produced from hybrid seed is greater than 14-fold for E1 and >40-fold for CBH I. We have observed this phenomenon numerous times with various transgenes (Streatfield et al., 2002; Hood et al., 2003; Hood, 2004; Clough et al., 2006). Each of these proteins has different characteristics, and the germplasm used in the crosses vary; thus, the phenomenon appears to be a general one. Although it is empirically reproducible, the mechanism is unknown. We have constructed isogenic lines of high and low expression at advanced generations and are exploring the mechanism of this phenomenon.
Table 3. Summary of increase in protein accumulation per dry weight at various steps in breeding and processing of mature grain over T1 (first-generation seed from tissue culture plants) selected lines
Increase from T1–T6
Increase with self-pollination
Increase with hybrid grain
Increase with high oil germplasm
Potential increase in elite grain*
Processing increase in germ flour
Total possible increase with processing†
*Potential increase in elite grain is derived from multiplying the T6 increase by the self-pollination increase.
†Final number is derived from multiplying elite grain increase by processing increase.
two to fourfold
14- to 28-fold
∼70- to 100-fold
two to fourfold
40- to 80-fold
∼360- to 720-fold
Efficient recovery of enzyme from the production raw material plays an important role in reducing the cost of industrial enzymes. The bottom line for industrial enzymes is how much enzyme can be recovered from production material on a dry weight basis. Separating the germ fraction from the grain and removing the oil using standard process conditions provide a defatted germ that is often used in animal feed and in ethanol production (Duensing et al., 2003). Because the cellulase is expressed from an embryo-preferred promoter, the germ should contain as much as 10-fold more cellulase than the whole grain, which provides a potential formulation for adding concentrated enzymes to industrial processes. However, processing is not a 100% efficient procedure, and losses occur at each step. Thus, it was critical to assess the enzyme concentration and activity in the germ fractions. Thus, dry-milled, processed grain, producing the embryo (germ) and the endosperm (grits) fractions, was analysed to determine the recovery of enzyme from the germ fraction. In the recovered germ, the enzyme shows as much as a fivefold increase in accumulation on a dry weight basis for E1 and as much as a ninefold increase for CBH I (Table 3). When all breeding and processing steps available for corn production are combined, the resulting recovery of enzyme per dry weight will allow highly cost-effective production of E1 and CBH I from the corn seed biofactory.
We previously demonstrated that high oil germplasm can be used to enhance the yield of target proteins in maize (Hood, 2004). In the current study, we tested the effect of a single high oil cross on the accumulation of E1 and CBH I. The increase in CBH I was fivefold, and a 50% increase in total E1 protein was observed. These results suggest a production method—that high oil germplasm could be used as a component of a production field with out-crossing of the transgenic line pollen onto a high oil line for production of high accumulation of enzymes and high yields of grain. Thus, using a variety of methods, the quantity of enzyme in transgenic lines can be increased 70- to 100-fold or 360- to 720-fold for E1 or CBH I, respectively, over the original transgenic lines recovered from tissue culture (Table 3). This strategy is powerful for promoting industrial enzyme accumulation in maize and has impressive potential for lowering production costs associated with the system.
High accumulation is only useful if the enzymes so accumulated are active in their target applications. Thus, enzymes were analysed for their ability to release free sugars by a variety of methods all of which indicated their ability to deconstruct microcrystalline cellulose. E1 and CBH I purified from corn flour were active in deconstructing cellulose when combined with β-glucosidase. The primary product of this reaction (over 24 h) was glucose in contrast to purchased control mixtures of T. reesei enzymes for which cellobiose was the primary product. Further experiments to optimize plant-made enzyme load alone and in combination with microbial enzymes are in progress.
In the near term, the accepted production method for cellulase enzymes for biomass conversion is likely to be microbial fermentation. However, because of the potential in cost savings for large-scale production using plants, in the longer term, the plant process could enhance the use of microbial enzymes, lowering the capital investment necessary for this cost-sensitive industry.
To grow these corn lines with enzymes at the current accumulation levels, an approximately equal number of acres would be required to deconstruct the harvestable stover (Howard and Hood, 2007). However, the technology is in its infancy and will produce new lines with more than 10-fold greater amounts of enzyme than are currently available. In addition, because the system does not require capital infrastructure, the cost of production is far less than for microbial enzyme fermentation (Howard et al., 2011). Nevertheless, because they are genetically engineered, deregulation of the lines through USDA APHIS will be required for commercial production. However, even though the cost of this regulatory approval process could be as high as $30 million (McElroy, 2003), it is far less than the cost of building the fermentation infrastructure required for the microbial production systems.
The plant biofactory is thus a viable system for biomass conversion enzyme production. To scale these enzymes to industrial production, the next steps include to:
1determine the efficacy of biomass conversion using these enzymes in industry relevant conditions, i.e. biomass substrates and process conditions.
2scale-up reactions to pilot plant scale
3use enzymes in combination with microbial enzymes at pilot scale.
This excellent system can demonstrate the power of plant production to reduce the cost of effective enzymes for biomass conversion processes.
Transgenic plant lines
Each cellulase gene is driven by the embryo-preferred maize globulin-1 promoter (Belanger and Kriz, 1991) and is linked to a CaMV 35S promoter-driven maize-optimized phosphinothricin acetyl transferase (pat) gene, which confers resistance to the herbicide Bialaphos (Hood et al., 2007). Each independent transgenic event [BCC02 (CBH I) and BCH01 (E1)] has a single insertion and a single copy of the respective cellulase gene. Breeding experiments described below were carried out with individual plants derived from these two events.
Transgenic plants (from Hi II germplasm) were generated using Agrobacterium tumefaciens-mediated transformation experiments described earlier (Hood et al., 2007). The plants include two independent events expressing the E1 endocellulase (E.C. 220.127.116.11 or Cel5A; GenBank Accession #U33212) from Acidothermus cellulolyticus and the cellobiohydrolase I gene (E.C.18.104.22.168 or Cel7A; GenBank Accession #X69976) from T. reesei (Hypocrea jecorina). Each gene is expressed using the globulin I promoter from maize. Up to 10 individual plants were recovered for each independent transgenic event and pollinated with the SP122 (Lancaster) inbred in the greenhouse. The resulting ears were individually analysed, and 10–15 of the ears showing the highest-expressing seed were chosen for subsequent agronomic performance improvement through successive backcrosses to elite germplasm.
In each crossing experiment, multiple rows (typically five rows, 20 seeds per row) of transgenic seed were planted from each line described above. Nontransgenic pollen donors were planted in intervals at 5 days prior to transgenic plantings, on the same day as the transgenic plantings and at 5 days post-transgenic plantings to ensure sufficient pollen for crossing to the transgenic lines. Transgenic rows were sprayed at the 5-leaf stage with Liberty™ herbicide (1% active ingredient) to select against nontransgenic segregating plants. Selected lines were crossed separately to either an elite Stiff Stalk (SP114) or an elite Lancaster (SP122) inbred line to generate both parents for a high-yielding hybrid. In each cross, the transgenic line was used as the female parent with pollen from the elite inbred parent. For the final hybrid seed for grain production, the Stiff Stalk (SP114) line was used as the female parent. All field-based experiments were performed with APHIS permits and inspected facilities.
In high oil germplasm (HO-703) crosses, either the high oil or the transgenic line was used as the female parent in the single cross. Plants from each transgenic line/inbred parent combination were crossed to the HO-703 line. Data are reported for the lines producing the most progeny.
Cellulase assays were performed as described earlier (Hood et al., 2007). Corn meal (from coffee grinder) from fifty seeds was ground in the presence of liquid nitrogen, and 0.1 g of sample was extracted in 1 mL of 50 mm sodium acetate buffer, pH 5.0. The homogenate was vortexed and centrifuged at 9.3 g for 10 min or 0.4 g for 15 min at 4 °C. Supernatants from two independent extractions and two or three samples from each extraction were used for enzyme activity assays. Reaction mixtures contained 90 μL of 50 mm sodium acetate buffer, pH 5.0, and 10 μL of extract (approximately 500 ng total protein) with 25 μL of 5 mm methylumbelliferyl β-D-cellobioside (MUC; # M6018 Sigma Chemical Co., St. Louis, MO). A 96-well plate containing the reaction mixtures was incubated at 50 °C. For CBH I and E1 activity assays, incubation time was 120 or 30 min, respectively. Highly purified CBH I and E1 (provided by NREL) were used initially as the standard. As the amount of purified material was limiting and we had a desire to run a standard with every assay (thousands of samples over several years), we calibrated a commercial preparation of cellulase from T. reesei (E1 and CBH I mixture; Sigma # C8546) to use as a relative standard for every assay. Trichoderma reesei from Sigma showed 3.6-fold more activity on a weight basis than purified CBH I and 10-fold less activity than purified E1; therefore, these correction values were used when individual assay plates were run for corn extracts. The amount of protein in each sample was estimated using the Bradford method (Bradford, 1976).
Polyclonal antibodies were prepared by Cocalico Labs (Reamstown, PA) against purified CBH I and E1 proteins that were individually expressed in the E. coli Gateway™ vector system (Invitrogen, Carlsbad, CA). One-half microgram of total protein from a crude protein extract of corn flour was size separated using 12% SDS–PAGE (Invitrogen). Western blots were prepared from gels by transferring proteins to a PVDF membrane (Millipore, Billerica, MA) and blocking with 5% BSA (Fisher Scientific, Atlanta, GA) in 1× Tris-buffered saline buffer (25 mm Tris, 0.8% NaCl, pH 7.4). The blocked membrane was incubated with primary antibody, anti-CBH I or anti-E1 (1 : 2500), for 2 h. After incubation, membrane was washed three times for 5 min each with 1× TBS buffer and then incubated with secondary antibody, goat-anti-rabbit-alkaline phosphatase conjugate (1 : 5000), for 2 h. Colour detection was carried out using the NBT/BCIP reagent per manufacturer’s instructions (Promega Corporation, Madison, WI).
A 100-g sample of corn meal of each CBH I and E1 transgenic lines was separately soaked and agitated in 500 mL of 50 mm sodium acetate buffer pH 5.0 for 1 h at 4 °C. The suspension was filtered through four layers of cheese cloth and centrifuged at 10 000 rpm (Eppendorf centrifuge 5810R and Eppendorf fixed –angle rotor F-34-6-38) for 10 min at 4 °C. The pellet was resuspended in 500 mL of extraction buffer, thus repeating the enzyme extraction. Supernatants from the two extractions were pooled and used for ammonium sulphate precipitation. For CBH I, a 50%–90% ammonium sulfate pellet contained the protein, whereas a 0%–50% ammonium sulfate pellet contained the E1 protein. Pellets obtained after ammonium sulfate precipitation were dissolved in 15 mL of 20 mm Tris-HCl pH 7.0 buffer containing 150 mm NaCl.
Protein purification was carried out using the Bio-CAD HPLC system (Perceptive Biosystems; Global Medical Instrumentation, Inc., Ramsey, MN). Protein samples were desalted using a HiPrep 26/10 desalting column (GE Healthcare Bio-Sciences, Piscataway, NJ). Desalted proteins were loaded onto an anion exchange, HiTrap Q XL column (GE Healthcare Bio-Sciences). Before the start of 1 m NaCl gradient, the column was run with 20 mm Tris-HCl buffer pH 7.0 alone for 10 min, and then, the 1 m NaCl gradient was started and run for 20 min. Protein fractions were collected from 0 to 30 min with a flow rate of 5 mL/min (2 mLs per fraction). Eluted protein fractions were used for enzyme analysis and fractions from 41 to 60 and 39 to 60, which showed CBH I and E1 activities, respectively, were pooled separately. Pooled protein solutions were concentrated using a Spin-X UF concentrator (Corning, Corning, NY), 2 mL samples were loaded onto a Sephacryl S-200 Hi Prep, 1.6 × 60 column (GE Healthcare Bio-Sciences), and protein fractions were collected at flow rate of 0.5 mL/min using 20 mm Tris-HCl buffer pH 7.0. Eluted protein fractions were used for enzyme analysis, and fractions from 45 to 55 and 75 to 87, which showed CBH I and E1 activity, respectively, were pooled separately. Pooled protein fractions were concentrated using Spin-X UF concentrators and buffer exchanged with 50 mm sodium acetate pH 5.0. Purified proteins were assessed for relative purity by SDS–PAGE and remaining volumes lyophilized.
Validation of extraction and quantification of enzyme
Enzyme extraction was carried out as described above. Each resultant pellet was used for an additional four extractions. The supernatant from each extraction was collected separately and used for enzyme assay and Western blot analysis.
The source of cellulose was Sigmacell (Sigma Chemical Co.). 4-Methylumbelliferyl β-D-cellobioside (MUC) was obtained from Sigma Chemical Co., St. Louis, MO (M2018).
Microbial growth was monitored using Sigmacell as the sole carbon source to indicate the release of fermentable sugars and to ensure that no major interfering by-products were in the reaction (Jimenez-Flores et al., 2010). Control enzymes were from T. reesei (Sigma # C8546, Celluclast) and β-glucosidase (Sigma # C6105). A starter culture of Saccharomyces cerevisiae was allowed to grow in 5 mL of YPD broth on a rotary shaker at 37 °C and 225 r.p.m for 4–6 h; 0.1 mL of the culture (OD550 =0.5 of a 1 : 60 dilution) was used as inoculum for the cellulose fermentation assays. The assay was carried out under sterile conditions in BacT (BioMerieux, St. Louis, MO) bottles containing 6.25 mg/mL cellulose (Sigma # S3504, Sigmacell) suspended in 40 mL of 140 mm citrate/90 mm bicarbonate buffer pH 5.0. Treatments were as follows: no enzyme added, 4 μL Celluclast, 4 μL β-glucosidase, or 4 μL β-glucosidase with 3.75 mg purified plant-derived CBH I and 0.1 mg of purified plant-derived E1. All treatments contained 250 mg Sigmacell, 100 μL of yeast (grown to log phase in YPD (a standard medium for growing yeast comprising 1% yeast extract, 2% peptone and 2% glucose). Glucose is one of the media components used to grow the yeast. After growing the yeast to log phase, cells were harvested by centrifugation and washed with 140 mm citrate/90 mm bicarbonate buffer pH 5.0, to remove the sugar and other components of YPD medium. After washing, cells were resuspended in the citrate buffer, aliquoted into glycerol stocks and stored at −80 °C.
Bottles were incubated at 37 °C and consisted of the following treatments: No enzyme, 1 μL Celluclast, 1 μL β-glucosidase, or 1 μL β-glucosidase with 939 μg purified plant-derived CBH I and 27.5 μg of purified plant-derived E1. All treatments contained 62.5 mg Sigmacell and citrate buffer to a total volume of 10 mL.
Glucose oxidase assay
The GLOX assay was conducted as described by the Worthington Biochemical website (http://www.worthington-biochem.com/gop/assay.html) with the following modifications. Peroxidase and glucose oxidase were resuspended to 1mg/mL, nonoxygenated o-dianisidine (Sigma Chemical Co.) was resuspended in DMSO (Sigma Chemical Co.) to a stock concentration of 2% (v/v) and used in the assay at a concentration of 0.016%, and a 10% D-glucose stock solution was left to mutarotate for a minimum of 1 h prior to use. The total assay volume was 200 μL: 150 μL o-dianisidine solution in 0.1 m potassium phosphate buffer pH 6.0, 10 μL peroxidase, 10 μL glucose oxidase and 30 μL of glucose standard or cellulase reaction sample. Three independent GLOX reactions were performed for each treatment, and mean glucose concentrations are reported. GLOX reactions were conducted at room temperature, and readings were taken at 460 nm every 30 s for 5 min. SoftmaxPro5.4 (Molecular Devices, Sunnyvale, CA) software was used to analyse reaction rates. Samples for both the GLOX assay and HPLC analysis were removed from the same reaction bottles.
Carbohydrate concentrations were obtained using protocols established by the National Renewable Energy Laboratory Technical Report (NREL/TP-510-42623, available online at: http://www.nrel.gov/biomass/pdfs/42623.pdf). Analysis was performed on a Shimadzu Prominence Series HPLC with a Bio-Rad Aminex (HPX-87P) column, a Bio-Rad de-ashing pre-column and an Agilent 1200 Series Refractive Index Detector. Results shown are the median of three replicate samples.
The analysis was performed in SAS version 9.1 (SAS® business analytics software and services, Cary, NC) using PROC MIXED. Four parallel analyses were performed for BCH0101/SP122, BCH0101/SP114, BCC0206/SP122 and BCC0206/SP114. Each analysis was a one-way ANOVA with E1 (or CBH I) per cent dry weight as the response and generation as the only factor. The LSMEANS command was used to compare the mean response at each generation to every other generation and to the initial T1 value. These comparisons were adjusted using Tukey’s method. Each analysis used a 5% significance level.
This work was supported by a grant from the US Department of Energy (DE FG36 GO88025) with cost share from the Wal-Mart Foundation, the Walton Family Foundation, and Arkansas State University. We also thank Dr John Walker, Cal Poly State University, for his help in statistical analysis of the data.