There are many extant bioenergy grass feedstock varieties (genotypes), which are sufficient for select conversion processes. For example, specific maize and sugarcane genotypes have been successful bioenergy grass feedstocks since high-yielding genotypes (grain and juice, respectively) have been grown at large scale for decades, and the conversion process (yeast fermentation) is well understood at the industrial level. Recent attention has been given to the more difficult problem of 2nd generation lignocellulose biomass conversion into profitable bioenergy products, which has the potential for accessing the photosynthate locked into the plant cell wall for conversion into useful products. Clearly, 2nd generation genotypes that produce high dry weight yields are of paramount importance, which is the opposite direction of the Green Revolution which led to small plants with high grain yield . However, the identification and improvement of bioenergy grass genotypes with high biomass that efficiently respond to a given conversion process is ideal.
While there is much potential for bioenergy grasses as feedstock into thermal conversion processes (e.g. combustion, torrefaction, pyrolysis, and gasification), in this section we explore traits relevant to lignocellulose biochemical conversion processes which convert biomass into fermentable products through enzymatic hydrolysis (saccharification) . The bioenergy grass feedstock traits that underlie conversion efficiency are being elucidated opening the door to genetic enhancement from existing feedstock.
Cellulase enzyme cost is estimated to be ~50% of the total cost of the commercial hydrolysis process . In addition, the enzymatic hydrolysis of lignocellulosic material experiences a reduction in activity over time. This reduction in activity has been attributed to hydrolysis inhibition (end product and other [15–18]), reduction in easily accessible cellulose (e.g. crystalline vs. amorphous cellulose ), and reduction in efficient enzyme adsorption. Increasing enzyme accessibility to cellulose has been shown to play a crucial role in improving enzymatic hydrolysis [20–24]. Finding efficient means to increase enzymatic hydrolysis is vital to the success of lignocellulosic bioenergy production.
Chemical inhibition of cellulase reduces the total amount of reducing sugar produced for fermentation. High concentrations of end-products have been known to cause a reduction in cellulase activity. For example, while cellobiose is often a product of cellulases, it has also been shown to be a significant inhibitor of the activity of some cellulase . This inhibition has been shown to be reduced by supplementing β-glucosidase to cellulase solutions lacking sufficient β-glucosidase activity . End-product inhibition by glucose has been shown to inhibit late stage hydrolysis rates [27–29]. In addition to cellobiose, glucose has been shown to inhibit cellulase activity in cellulases derived from Trichoderma species [30, 31]. However, inhibitory effects of glucose do not appear to affect Aspergillus species to the same degree [32–35]. This often leads to Trichoderma cellulases being supplemented with Aspergillus β-glucosidase to increase saccharification efficiency on an industrial level [36, 37]. Additionally, xylose and arabinose, which are produced during the hydrolysis of hemicellulose, have been shown to inhibit cellulase activity [18, 38]. Substrate inhibition of cellulases has led to simultaneous saccharification and fermentation (SSF) systems becoming popular, alleviating end-product inhibition.
In addition to end-product inhibition, metal ions have been shown to be inhibitory to cellulase hydrolysis reactions. It is suggested that the Fe(II) and Cu(II) oxidize the reducing ends of cellulose, inhibiting the exo-cellulolytic activity of cellulase [39–43]. However, not all metal ions cause an inhibitory effect on hydrolysis. Kim et al. found that while Hg++, Cu++ and Pb++ caused decrease in the production of total reducing sugars, other metal ions (Mn++, Ba++, and Ca++) caused an increase in the total production of reducing sugars, indicating a stimulating effect on hydrolysis . Two of these ions (Hg++ and Mn++) were shown to play a direct role in enzyme adsorption. Additionally, Mg++ was shown to stimulate the activity of glucanase from Bacillus cellulyticus. The activity of cellulase produced from Chaetomium thermophilum was shown to be increased by Na+, K+ and Ca++, but inhibited by Hg++, Zn++, Ag+, Mn++, Ba++, Fe++, Cu++, and Mg++. This indicates that metal ions play an important role in enzyme efficacy during hydrolysis, and that knowledge of the correct ratio of metal ions is essential to increasing hydrolysis activity.
Phenolic compounds are also known to inhibit cellulolytic enzymes. These phenolics are often found in lignin, and are released (as well as their derivatives) during pretreatment processes. The types of phenolics present depends largely on the composition of biomass in combination with the type of pretreatment method employed [47–49]. A variety of released phenolic compounds have been identified during chemical pretreatment of lignocellulosic biomass [50–52], which have been shown to inhibit conversion of carbohydrates into ethanol as well as to inhibit cellulase activity [38, 53–56]. Cellulases, hemicellulases, and β-glucosidase enzymes have all been shown to be inhibited by these phenolic compounds [54, 56–59]. The magnitude of inhibition may specific to enzyme source as Aspergillus niger β-glucosidase was shown to be more resilient to phenolic inhibition when compared to Trichoderma reesei β-glucosidase, requiring a 4x higher concentration for inhibition . Introduction of tannic acid degrading enzymes (Tannases) has been shown to increase enzymatic hydrolysis, likely by reducing tannic acid’s propensity to interact and inhibit cellulase . Additionally, polyethylene glycol has been shown to reduce inhibition of cellulase by tannins  by breaking up tannin-protein complexes. Tween 80 and PEG-4000 have been shown to prevent inhibition of β-glucosidase by reducing the tannins ability to bind the cellulase protein [61, 62]. Finding additional methods to reduce the role of inhibitors in enzymatic hydrolysis is an important factor in increasing hydrolysis efficiency and profitability. Reducing the process-specific release of cellulase inhibitors through tailored feedstock genotypes is an attractive approach to enhancing enzymatic hydrolysis.
Lignocellulosic material is a complex matrix of cellulose, hemicellulose and lignin [63, 64]. In un-pretreated lignocellulosic samples, only a fraction of the cellulose is accessible to enzymatic hydrolysis, while the rest of the exposed biomass is lignin and hemicellulose. In order to increase access to cellulose, pretreatment methods are employed that aim to remove the lignin and hemicellulose fraction and leave cellulose available for hydrolysis. In addition, phenolic compounds such as ferulate play an important role in crosslinking lignin within the cell wall (see reviews [65–70]) and have the potential to be genetically modified to aid in the removal of specific cell wall components. There are many grass-specific features of the cell wall which have the potential to be exploited for increased bioenergy production . For example, the composition of grass lignin is composed of syringyl (S), guaiacyl (G) and p- hydroxyphenyl (H) subunits that when present in varying ratios may lead to increased digestibility . However, debate remains involving the role of lignin subunits in conversion efficiency [72–75].
Removal of structural components such as hemicellulose via dilute sulfuric acid pretreatment has been shown to increase accessibility to cellulose for enzymatic hydrolysis . Removal of hemicellulose has been reported to increase pore volume and surface area further increasing the accessibility of cellulase . Drying lignocellulosic substrates after chemical pretreatment results in the collapse of the newly formed pores, resulting in a decrease in enzymatic hydrolysis rate through reduction in available cellulose for hydrolysis [24, 77]. Another pretreatment strategy which uses ionic liquids on switchgrass was shown to increase the porosity by over 30 fold, greatly increasing the accessibility of cellulose to enzymatic digestion . This indicates that pore size and volume may play a significant role in increasing the rate of enzymatic hydrolysis. The identification of bioenergy grass feedstock genotypes that respond favorably to chemical pretreatment can increase end-product yield.
Lignin has been shown to play a large role in enzymatic conversion efficiency . In Miscanthus sinesens, Yoshida et al. showed that removal of lignin via sodium chlorite resulted in an increase in enzymatic hydrolysis rate . Yoshida et al. further demonstrated that the addition of hemicellulases resulted in an increase in overall hydrolysis rate, indicating that hemicellulose is an additional inhibitor of cellulose hydrolysis rates . Zhao et al. also reported an increase in the enzymatic hydrolysis rate of sugarcane bagasse after the removal of lignin with paracetic acid . Dissolution of lignocellulosic material with ionic liquid has been shown to increase enzymatic hydrolysis rates in wheat straw , corn stover  and switchgrass . Kimon et al. showed that disolving lignocellulosic material in ionic liquid at temperatures >150°C has a large effect on saccharification of sugarcane bagasse . Additionally ionic liquid pretreatment of switchgrass was shown to increase hydrolysis kinetics by over 39 fold over untreated switchgrass . Ionic liquid pretreatment has also been shown to break inter and intra-molecular hydrogen bonding between cellulose strands causing an increase in the removal of amorphous components (lignin, hemicellulose) as well as an increase in surface area for cellulase adsorption . These methods were both shown to superiorly increase hydrolysis rates when compared to traditional methods (dilute acid and ammonium hydroxide, respectivley). Singh et al. reported that ionic liquid caused disruption of the inter and intra-molecular hydrogen bonding between lignin and cellulose which initially causes swelling of the plant cell wall followed by complete dissolution . Organosolv pretreatment of switchgrass was shown to preferentially remove both lignin and hemicelluloses, leaving a larger cellulose fraction which resulted in an increase in the enzymatic hydrolysis rate . Rollin et al. showed that treating switchgrass with organozolv resulted in a similar increase in the surface area causing increased cellulase adsorption . It is important to note that the promising field of ionic liquid pretreatment it still in its infancy. The current high costs of ionic liquid pretreatment limits its application to industrial scale-up, and like enzyme costs, must be reduced in order to be economically feasible on a large scale.
In addition to chemical pretreatment, naturally occurring mutations found in grasses have been shown to increase the rate of enzymatic hydrolysis via reductions in lignin. Brown midrib (bmr) is a phenotype found in grasses (maize , sorghum  and pearl millet ) that is associated with a mutation in genes involved in monolignol biosynthesis. These mutations have been shown to lead to a reduction in the total lignin content of the plant [92, 93]. The brown colored midrib of the leaf has been shown to associate with a mutation in cinnamyl-alcohol dehydrogenase (CAD), which causes incorporation of cinnamyl-aldehydes in place of cinnamyl-alcohol during lignin biosynthesis [72, 94, 95]. Additional bmr varieties have been shown to have mutation in caffeic acid O-methyltransferase (COMT) [96–98]. However, both CAD and COMT mutants only exhibit reduced monolignol biosynthesis as opposed to total cessation of monolignol biosynthesis, indicating that other CAD and COMT genes may individually override complete cessation of monolignol biosynthesis. Theerarattananoon et al. found that a bmr mutant sorghum variety had less total lignin than forage, grain, sweet and photoperiod sensitive sorghum varieties . In addition to lower lignin contents, bmr varieties have been shown to have increased susceptibility to chemical pretreatments. In sorghum, it was found that bmr mutants were more susceptible to alkaline pretreatment than non-bmr varieties . Corredor et al. demonstrated that bmr sorghum varieties had a 79% hexose yield after enzymatic hydrolysis, which was higher than two non-bmr varieties which yielded 43% and 48% . Additionally, sorghum varieties that contain both the mutations in COMT and CAD have been shown to have lower lignin contents than either mutant individually . It is possible that there are additional genes and alleles leading to lowered lignin or other traits associated with higher hydrolysis rates. The identification of new as well as known lignification genes could lead to novel breeding programs where stacking of genes could result in intrinsic increases in lignocellulosic digestibility.
It is important to note that some maize bmr varieties have been characterized as being susceptible to lodging . However, these susceptibilities were not seen in other maize studies which may be attributed to differences in genetic background [104, 105]. This suggests that selecting an optimal genotype for the bmr mutation may be important in creating a superior feedstock. In addition to lodging, bmr mutants have been labeled as more susceptible to disease and pathogen attack due to reduction in the lignin barrier. However, accumulation of lignin precursors has been shown to prevent the production of virulence factors as well as limit fungal pathogens [106–108]. It has also been widely reported that bmr varieties experience a decrease in yield associated with reduced lignin content. This has been seen in maize [104, 109, 110] and sorghum [111, 112]bmr varieties. However, sorghum bmr hybrid varieties have been created that experience yields similar to wild type , suggesting that the genetic background of the mutant variety is important in overcoming yield reduction.
Transgenic approaches have already shown potential to increase saccharification efficiency in grasses. Overexpression of miR156, which suppresses SQUAMOSA PROMOTER BINDING PROTEIN LIKE (SPL) genes, in switchgrass caused an increase in overall biomass accumulation coupled with an increase in conversion efficiency of 24.2% – 155.5% in non-pretreated lignocellulosic material and between 40.7%–72.3% increase in acid pretreated samples . In addition, moderate overexpression of miR156 caused switchgrass plants not to flower, reducing the possibility of transgenic gene escape. However, it should be noted that overexpression of miR156 caused dwarfism in both rice  and maize , which greatly reduces the plants value as a bioenergy feedstock. In addition, overexpression of R3R3-MYB4 transcription factors has been shown to repress lignin biosynthesis in several species [117–120]. In switchgrass, overexpression of PvMYB4 resulted in a three-fold increase in hydrolysis efficiency . However, like the overexpression of miR156, these plants experienced a smaller stature than control varieties, limiting the gains made from increased hydrolysis efficiency. Clearly, the identification of active small RNA regulatory genes that do not affect biomass yield using genomic approaches is an exciting avenue towards bioenergy grass improvement.
Crystallinity index (CI) is a parameter that is used to determine the relative amount of crystalline cellulose in lignocellulosic material. Increased crystallinity of cellulose causes reduction in cellulase binding to cellulose due to reduced surface area. Conversely, increased amorphous cellulose causes an increase in the surface area, causing an increase in hydrolysis rates. CI has been measured using x-ray diffraction , solid-state 13C NMR , infrared spectroscopy (IR) [124–126] and Raman spectroscopy . CI has been shown to be correlated with enzymatic hydrolysis of lignocellulosic material. In Sorghum bicolor, CI has been shown to be negatively correlated with hydrolysis rate in whole plant tissue . It has also been shown in sorghum as well as maize that stem has a higher crystalline content than leaf tissue . Furthermore, sorghum bmr mutants as well as wild type varieties experience an increase in CI after pretreatment with 1M NaOH. This observation is attributed to the removal of the amorphous component of the lignocellulosic biomass, leaving a larger fraction of crystalline material. However, it was also observed that an increase in the concentration of NaOH to 5M showed a decrease in CI, which was attributed to the crystal structure change and cellulose amorphization . A similar trend was seen in dilute acid pretreatment of five sorghum varieties. Dilute acid pretreatment of sorghum at 140°C resulted in an increase in CI, however increasing the temperature during pretreatment to 165°C resulted in a decrease in the CI of 4 of 5 sorghum varieties . This change in cellulose composition after pretreatment has been previously demonstrated in various industrial cellulose samples pretreated with NaOH [130, 131]. Sugarcane bagasse was also shown to experience an increase in crystallinity after pretreatment with peracetic acid, which was attributed to a decrease in the amorphous component of the plant biomass . Corredor et al. demonstrated dilute acid pretreatment of bmr and non-bmr sorghum varieties were shown to increase CI after pretreatment . In addition, hydrolysis of the same samples resulted in a reduction in CI. Liu et al. found that like sorghum, acid pretreatment of maize biomass causes an increase in CI. However, the harshest pretreatment conditions cause a decrease in crystallinity, likely due to disruption of the cellulose crystalline structure . This trend was confirmed by Mittal et al., who also demonstrated that crystallinity of corn stover depends on specific conditions of alkali pretreatment. Additionally, Barl et al. demonstrated that maize husks experienced an increase in CI after both acid (H2SO4) and alkali (NaOH) pretreatment processes . It should be noted that previous studies have demonstrated that the cellulose binding domain of cellulases disrupt cellulose crystalline structure and causes a decrease in CI [134, 135]. This suggests that cellulose binding plays a role in conjunction with a decrease in cellulose content in the reduction in crystallinity index during enzymatic hydrolysis. Therefore, finding favorable genetic variation in endogenous and pretreated CI is a logical approach to improve hydrolysis yield .
Not all pretreatment strategies lead to an increase in CI. Pretreatment strategies that are particularly harsh initially increase CI through removal of amorphous components, followed by subsequent dissolution of crystalline cellulose. For example, Kimon et al. demonstrated that dissolving sugarcane lignocellulosic material with ionic liquids at temperatures >150°C causes a reduction in the cellulose CI and a large increase in glucan saccharification, while temperatures <150°C has a small effect on crystallinity, which was associated with a slower initial rate of glucan saccharification . Therefore, a screen for bioenergy grass genotypes that respond to harsh pretreatments in a favorable way could identify better feedstocks.
CI has been shown to differ between plant species, as well as different varieties within a species. When compared to different sorghum varieties, maize has been shown to have a higher CI . Vandenbrink et al. demonstrated that CI differed between 18 different varieties of Sorghum bicolor, and these differences in CI were associated with hydrolysis rate . Harris et al. found that crystallinity index differed among a large variety of plants which included sweet sorghum, switchgrass, giant Miscanthus, sweet Miscanthus, flame Miscanthus, gamagrass, big bluestem and Arabidopsis. However, it must be pointed out that many of these species were only tested on a small number of varieties, which may not give an accurate depiction of CI in a diverse population where one genotype is one data point. These studies provide evidence that due to differences in CI between species and variety, there may be a significant genetic component that is associated with the trait.
There is much debate about the changes in crystallinity experienced during enzymatic hydrolysis of lignocellulosic materials. Various studies have demonstrated that amorphous cellulose components are hydrolyzed preferentially to crystalline components, resulting in an increase in crystallinity as enzymatic hydrolysis occurs [80, 137, 138]. However, various other studies have demonstrated that hydrolysis results in little change to crystallinity over the course of enzymatic hydrolysis [139, 140], which was attributed to the synergistic action of endo and exo-glucanase activities [87, 141]. However, it should be noted that studies have shown that the cellulose binding domain of multiple cellulases disrupt the supermolecular structure of cellulose, resulting in a decrease in CI [134, 135]. This creates a difficult task in measuring changes in CI during enzymatic hydrolysis.
Non-specific cellulase adsorption to biomass plays a crucial role in determining the effectiveness of enzymatic hydrolysis. Due to the high cost of enzymes for commercial scale hydrolysis, adsorption and desorption rates in specific genotypes should be pre-determined. After hydrolysis, enzymes can either remain adsorbed to the substrate or unbound in the hydrolysate . Cellulase adsorption depends largely on the concentration of the protein, as well as cellulase concentration and available surface area . Initial protein adsorption has been shown to correlate with the initial rate of cellulose hydrolysis [19, 144]. Multiple studies have shown that total enzyme adsorption is directly related to hydrolysis rate and yield [145–148]. Strong correlations between available surface area and rate of hydrolysis have also been observed [23, 149, 150]. This increase in hydrolysis rate can be attributed to increased adsorption. Nutor et al. found that initial protein adsorption occurs quickly, reaching a maximum in 30 minutes, followed by 55-75% desorption . Increasing the amount of enzyme adsorbed onto cellulose substrate is a potential avenue to increase hydrolysis rates, and it remains untested if specific cellulases are better adsorbed in specific bioenergy grass feedstock varieties.
Cellulase adsorption to lignin reduces cellulase activity by sequestering the enzyme away from its substrate. After the completion of hydrolysis, non-specific binding to lignin that has been freed during hydrolysis has been shown to occur, where 30-60% remains bound to the lignin fraction [152, 153]. This non-specific binding has been shown to be only partly reversible . Adsorption of cellulases to isolated lignin has been reported, supporting claims that non-specific binding occurs to the lignin fraction during hydrolysis [155, 156]. Any cellulase bound to lignin is not available to hydrolyze cellulose, limiting overall efficiency. Hydrolysis rates of cellulose has been shown to be correlated with the tightness and affinity of adsorption . Removal of lignin does not only reduce the steric hindrance to the enzyme, but also reduces the lignin available for non-specific binding [158, 159].
Protein adsorption interactions are usually non-covalent (hydrogen bonding, electrostatic or hydrophobic interactions ). Surface characteristics of lignocellulosic material are thought to play a major role in cellulase adsorption where the high surface area hydrophobicity results in increased adsorption. Cellulases have been shown to have hydrophobic amino acids exposed on the outside of the protein, which interact with the hydrophobic surface of cellulase . The affinity of cellulase for hydrophobic substrates may explain non-specific binding to lignin which is highly hydrophobic. In addition to this, metal ions have been shown to increase (in the case of Mn++) and decrease (in the case of Hg++) the adsorption affinity and tightness of binding to the hydrophobic surface of cellulose .
In order to drive down the cost of enzymatic hydrolysis, strategies to recycle cellulases are being developed [141, 162–165]. Enzymes can be recovered from either bound substrate or from the liquid hydrolysate that remains after the first round of hydrolysis. Recovery of the enzyme from bound substrate can be achieved through washing with surfactant (such as Tween 20 ) or through recovery of the solid substrate in which the cellulase remains bound . Use of cellulase recovered from lignocellulosic residue for subsequent rounds of hydrolysis have been shown to experience reduced activity, which has been attributed to accumulation of bound lignin after each successive round of hydrolysis [154, 163]. Recovery of enzyme from the liquid hydrolysate has been traditionally been done through ultracentrifugation techniques [142, 167, 168]. While this method has been proven effective, it would be costly to scale up to industrial magnitudes. A more effective method may be to exploit cellulase affinity for cellulose, in which the addition of cellulose to cellulase-containing hydrolysate results in re-adsorption onto the fresh cellulose substrate [163, 169, 170]. Tu et al. found that addition of fresh substrate to hydrolysate recovered ~50% of cellulases . Additionally, bound enzyme was shown to be able to be recovered by contacting the bound substrate with fresh substrate . However, sequential hydrolysis with recovered enzyme results in decreasing hydrolysis rates due to non-specific binding. Additionally it must be noted that β-glucosidase does not bind to cellulose substrate, and must be added at the beginning of each round of hydrolysis in order to prevent the buildup of cellobiose and the resulting substrate inhibition . It is therefore necessary to develop techniques that are able to efficiently desorb cellulase from bound substrate. Deshpande et al. found that 90% of cellulase was recoverable from steam-exploded wheat straw . Jackson et al. found that using a surfactant such as Tween 80 resulted in a recover of 6 – 77%, depending on concentration of Tween 80 and pH of the solution . Additionally, Jackson et al. revealed that the highest protein recovery does not necessarily dictate the highest activity recovery, and that alkali conditions may be responsible for deactivation of the enzyme. Otter et al. demonstrated that Tween 80 and Triton X were able to desorb 65-68% of bound cellulase under alkaline conditions . Qi et al. demonstrated that enzyme recycling of alkali and dilute-acid wheat straw was comparable when using ultracentrifugation and additional substrate techniques . However, the additional substrate technique requires addition of β-glucosidase after each round of hydrolysis, whereas ultracentrifugation does not. Finally, there was a noticeable difference in enzyme recovery between dilute-acid and alkali pretreated samples, where alkali pretreated samples were able to desorb a larger amount of cellulase. While this discussion is focused on the putative industrial processes, it may be that specific feedstock varieties naturally exhibit lower adsorption rates that would further enhance the engineering endeavors.
In order for bioenergy to become a sustainable alternative to traditional fossil-fuel based transportation fuels, significant improvements to current enzymatic hydrolysis methods must be made. Reduced enzyme activity has been shown to be related to end-product inhibition, production of phenolic compounds from lignin, as well as metal ion inhibition. Additionally the reduction in easily accessible cellulose through steric hindrance and high crystalline to amorphous cellulose levels cause a reduction in cellulose available for enzymatic hydrolysis. Non-specific binding of cellulases to solubilized lignin has also been associated with reduced hydrolysis rates. Finally, adsorption has been shown to be correlated with the initial rate of hydrolysis, while enzyme desorption is essential for enzyme recycling and reducing the cost of enzymes in bioenergy production. While these process components are being examined at the engineering level, a simple screen of existing bioenergy grass varieties could identify genotypes with a favorable trait baseline making the process engineering task less difficult.