7.2.1 Process integration: pretreatment/hydrolysis interface

An effective pretreatment alters or removes impediments to hydrolysis, both structural and compositional. The result is an improved rate of enzyme hydrolysis and increased yield of fermentable sugars. The key technical challenges in pretreatment optimization are to significantly disrupt cell wall structure to allow for enzyme access, while minimizing degradation and inhibitor formation, without driving up capital and process costs. At the interface of pretreatment and enzyme science, researchers have aimed to understand the fundamental properties of biomass that is ‘well-pretreated.’ A rigorous understanding of the physical and chemical characteristics of optimally pretreated biomass may allow rational design of both improved pretreatment technologies and improved enzymes. Researchers are not in agreement about what factor(s) correlate most with increased enzyme accessibility. Some consider that the increase in surface area and ‘pore’ size of biomass is the property which best correlates with enzyme digestibility (Kumar and Wyman, 2009a, 2009b; Chandra et al., 2007). Others cite reduced cellulose crystallinity as the controlling factor. Since pretreatment serves as a type of fractionation by removing and/or relocating lignin as well as disrupting the hemicellulose (Kim and Holtzapple, 2006; Liu and Wyman, 2005), a number of authors have noted a direct link between enzymatic accessibility and degree of hemicellulose and lignin removal during pretreatment. However, it has also been observed that a near complete removal of hemicellulose and lignin can have a detrimental impact on enzymatic accessibility (Ishizawa et al., 2009). Other frequently addressed parameters include fiber size or degree of polymerization (DP) of the cellulose.

As the goal of the combined pretreatment and hydrolysis steps is to convert as much of the polymeric sugar to fermentable sugar as possible, care must be taken during these steps not to degrade or irreversibly transform the sugars, as they will then be lost, representing a costly reduction in product per ton of starting material. The impact of their loss on process cost depends on the feedstock; for less expensive feedstocks, a process may be optimized wherein some of the sugars may be sacrificed due to inaccessibility following a less severe pretreatment, or at the other extreme, due to degradation during a particularly harsh pretreatment. Much consideration is taken in optimizing the pretreatment conditions. Commonly, a composite design of experiments is used to explore variables such as catalyst concentration, temperature, and duration of process, with the aim of balancing resultant accessibility to enzyme, while minimizing loss of sugar potential due to degradation.

Pretreated slurries have physical and chemical characteristics that can prevent the enzyme proteins from catalyzing the depolymerization of the cellulose and hemicellulose into monomeric sugars for further fermentation. Pretreatment process design also impacts the performance of downstream fermentation processes. A critical area of pretreatment research seeks to minimize inhibitors of enzymatic hydrolysis and fermentation.

Many leading pretreatment technologies take advantage of acid or base catalysts to accelerate biomass deconstruction. The extremes of the pH scale are not well tolerated by the majority of microorganisms and enzymes, and thus pH modification is often required following pretreatment in order to bring biomass slurries to an appropriate pH for optimal enzymatic conversion and sugar fermentation. Commercial fungal cellulases currently on the market have pH optima in the range of pH 4–6. When adding an alkaline chemical to an acid pretreated slurry, or acidic chemical to an alkaline pretreatment, formation of salts is unavoidable. Direct costs for neutralization of chemical inputs, and also for salt disposal that result from the need to manage waste streams at the end of the process flow are key cost drivers in pretreatment. Salts that end up in the soluble fraction after fermentation must be dealt with in wastewater treatment steps, and salts that remain in the insoluble fraction of residual material after fermentation may produce contaminant emissions which are regulated and pose a treatment cost. One example is sulfur oxide, as in the case of a sulfuric acid enriched process.

In addition to these process costs, accumulation of salts in process streams can be inhibitory to enzymes during hydrolysis and to the fermenting organism due to osmotic stress. Salt formation from pH adjustments is a burden to the point that some biorefineries have been designed to avoid using acid as a pretreatment technology as they require greater neutralization chemical inputs.

In addition to the salts generated during pretreatment, there are other compounds that are well known as inhibitors in downstream processes, particularly as biological inhibitors of commonly used fermentation organisms, such as furans, weak acids, and phenolic compounds.

The pretreatment is one of most expensive processing steps in the biomass conversion pathway, and thus has a great potential for improvement. Some of the operating expenses have been touched upon in the above discussion, but the capital costs associated with the specialized equipment needed to achieve high temperatures and pressures and resist corrosion in the presence of aggressive catalysts comprise a major part of the combined biorefinery capital expense. Sulfuric acid catalyzed pretreatments in particular have been highlighted as being associated with expensive capital inputs due to the need to build reactors of exotic metals that withstand degradation by strong acid. For other processes, the need to recycle expensive catalysts provides a driver for high pretreatment costs.

Reducing inhibitor generation and reducing the chemical loading in pretreatment can provide a multitude of benefits. Maintaining the effectiveness of a pretreatment in opening up the cell wall to enzymatic attack while decreasing catalyst load such that cheaper metallurgy can be used has been a recent focus of effort. Displacing chemical inputs may require more complex and robust enzymes, but the overall savings in chemical consumption and waste treatment and salt disposal are positive. Perhaps the most substantial benefit is seen in the fermentation process; the impact of lowered salts and decreased inhibitors improves the rates and yields of conversion of sugars to product. For this reason, convergent efforts in the industry have recently pushed to decrease the need for externally added catalysts in pretreatment (Chen et al., 2012).

In addition to reducing the negative impact of inhibitors through improved pretreatments, a range of biological innovations has helped reduce the magnitude of the problem. On the enzyme front, screening for enzymes that are highly active in the presence of biomass-derived inhibitors such as soluble lignin has led to improved cocktails for use of ‘real life’ pretreatment slurries. On the fermentation front, organisms have been developed that are more tolerant to specific inhibitors like furfural and HMF (Geddes et al., 2011). Multiple studies utilizing model substrates have shown that some inhibitors are synergistically detrimental in their action. Due to the synergism between different classes of inhibitors, evolutionary engineering and adaptation to biomass pretreatment slurries has been an attractive path for improving stress tolerance of organisms (Geddes et al., 2011).

Different types of pretreatment lead to different enzyme requirements downstream. While dilute acid pretreatments effectively convert a majority of hemicellulose to monomeric sugars, alkali pretreatments leave a substantial amount of insoluble and/or soluble oligomeric xylan, with or without side chain branching. Many fermentation processes are unable to utilize soluble oligomeric sugars; for this reason, the degree of polymerization of soluble sugars is important, as enzymes may need to convert soluble oligomers to monomers.

In the case of dilute acid pretreated corn stover, insoluble hemicelluloses may be reduced to less than 5% of total insoluble solids in pretreated material. Interestingly, hydrolysis of this material still benefits from the presence of enzymes that can attack this insoluble hemicellulose. Beta-xylosidases, which can convert soluble hemicellulose oligomers to monomers, are required at relatively low levels due to the low content of oligomerized hemicelluloses.

7.2.2 Enzymatic hydrolysis and product fermentation: process design

There are several viable hydrolysis and fermentation options available, each with benefits and drawbacks. The most economically viable process options select for configurations that maximize enzyme performance. Research is dedicated to understanding how feedstock, pretreatment, and fermentation methods holistically change the process economy with respect to hydrolytic conversion of biomass. Ultimately, each feedstock and pretreatment combination should be evaluated on an individual basis to determine the best process configuration to enable the industry. Fermentation of biomass hydrolysates to ethanol has been explored in much more detail than any other biobased product, and thus most of the process design discussion in this chapter focuses on ethanol as the end product.

A central consideration in deriving an optimal process is the impact of total solids (TS) loadings on overall process economy. A constant challenge to biomass-based biorefiners is the need for high ethanol titers at the end of fermentation; ethanol titers less than 5–6% in general can be considered cost prohibitive due to the power input needed for distillation to drive off the additional water. The simplest way to obtain high ethanol concentrations is to ensure high potential sugar content in upstream processes by maintaining the total solids concentration between 20 and 30% TS, or even higher if possible. However, at these high total and insoluble solids concentrations, efficient mixing is difficult. Unlike starch-based fermentation processes where relatively high solids consistencies can be achieved, lignocellulose slurries at high total solids have much higher viscosity due to their intrinsic high water retention properties. The first hours of hydrolysis require comparatively high energy input until the bulk viscosity ‘breaks.’ Sugar concentrations increase as hydrolysis progresses, introducing product inhibition to the existing enzyme inhibitors in the liquid phase; this often results in lower overall hydrolysis yields. To avoid these inhibitor sources, a dilute or lower total solids process could be coupled with an evaporation step or the addition of a concentrated syrup stream (e.g., from a first generation co-located starch biorefinery) to allow for sufficient sugar concentrations to reach ethanol titer targets. However, the capital and operational expenses must be weighed against the distillation cost savings.

Relative to the basic process configuration like the one in Fig. 7.1, process designs with greater and lesser complexity can be explored with respect to the impact of these designs on the enzymes and fermentation organisms. In general, increasing complexity, as illustrated here by considering the impact of splitting process streams, may open possibilities to allow for providing a more optimal environment to hydrolytic enzymes and fermentation organisms.

Solid–liquid separations are often considered at various stages in the process, as this can allow for detoxification, recycling of catalysts, and integration of other water streams. We will consider the impact of solid–liquid separation at two different stages of the process, though this procedure could be employed at several additional points in the process.

As was explained above, pretreatments that hydrolyze hemicellulose to a high degree often produce side products that are inhibitory both to enzymes and to organisms. With this in mind, it may be desirable to introduce a solid–liquid separation step following pretreatment to overcome problems with soluble inhibitors. Detoxification of soluble streams (liquor) generated from pretreatment can be performed, such as over-liming in the case of acid pretreatment, to remove inhibitory components such as furfural (Mohagheghi et al., 2006). Separating the process stream at this stage also offers the potential to further reduce soluble inhibitors in the solid fraction by washing solids to varying levels. The advantage is a dramatic improvement in enzyme performance that is often observed following removal of soluble inhibitors, and an even more pronounced improvement in fermentability of resulting sugars from hydrolysis of the insoluble fraction. Streams can be recombined following hydrolysis of the insoluble fraction, or alternatively, splitting the process streams allows for an option to ferment pentose and hexose streams separately. This translates into shorter fermentation times and faster uptake rates for the individual sugars. The significant drawback to either of these strategies is the addition of water or salts into the process, diluting sugar titers. In all cases, the benefits gleaned from splitting the process stream must outweigh the significant costs of the solid–liquid separation combined with additional costs for downstream evaporation or similar concentration strategy.

Another point where solid–liquid separation is often considered is after the hydrolysis. Separation of lignin-enriched residues also allows for their potential use in high value co-products (Inyang et al., 2010; Kim and Kadla, 2010), and may be necessary for effective recovery of soluble fermentation products. Unfortunately, separation of the very fine particles remaining after near complete enzyme hydrolysis is quite difficult, requiring the use of flocculants to aid separation, which increases operational costs. Loss of sugar potential is also an issue.

While splitting process streams increases the process complexity, consolidation has been an area of intense interest in the past few years. Enzymatic hydrolysis and fermentation are areas which may be consolidated to varying extents.

Hybrid hydrolysis and fermentation (HHF)

A combination of the SHF and SSF process aims to take advantage of the benefits of both these systems through use of a ‘hybrid hydrolysis and fermentation’ or HHF configuration. In this case, hydrolysis is performed under conditions that are optimal for enzyme performance, but the fermentation is initiated before the target hydrolysis conversion level is reached under conditions that are optimal for fermentation; thus, enzymatic conversion and fermentation occur simultaneously in the later part of an HHF. Enzymatic hydrolysis rates will drop over the course of hydrolysis due to product inhibition; reducing the temperature of hydrolysis to accommodate concurrent fermentation after the hydrolysis rate reaches a certain threshold does not compromise hydrolysis yields. A schematic of enzymatic hydrolysis rates versus time is shown as a function of process time in Fig. 7.2. At the optimal enzyme hydrolysis temperature of 50 °C, the hydrolysis rate gradually slows due in part to feedback inhibition since there is no organism to take up glucose and cellobiose. At 32 °C, the optimal yeast fermentation temperature, the cellulase enzymes work at a suboptimal rate due to the lower temperature but no feedback inhibition is experienced due to the simultaneous fermentation of the released sugars. The optimal duration of hydrolysis versus fermentation in the HHF configuration is substrate and pretreatment dependent; examining the rates of product formation in a separate hydrolysis process and in an SSF process will allow for prediction of the optimal timing for switching from hydrolysis to SSF mode in an HHF. When the feedback inhibition at the optimal temperature gets too great, yeast is added and the reaction conditions are altered to match the optimum conditions for the fermentation organism.

Clearly, the choice between an SHF, SSF, or HHF format will depend on the characteristics of the enzyme cocktail employed in a process. For an enzyme preparation that has a very narrow high temperature optimum, and high tolerance to end-product inhibition, a process with extensive conversion at high temperature is favored. A recent trend in enzyme development has been to take advantage of higher rates of reaction accomplished by thermostable cellulases (Viikari et al., 2007), and to engineer beta-glucosidase enzymes that can effectively catalyze cellobiose to glucose conversion even in the presence of high levels of glucose. A result is that the relative performance of enzymes at hydrolysis temperature optima is much higher than at the fermentation temperature, favoring extensive high temperature hydrolysis, even at high solids loadings.

7.4.1 Beta-glucosidase

Beta-glucosidase (bG) (or cellobiase, cellobiose hydrolases) (EC 3.2.1.21) catalyzes conversion of cellobiose and other soluble glucooligomers to glucose. Enzymes with bG activity are found in glycoside hydrolyase families GH1, 3 and 9; they cleave soluble oligosaccharides (Langston et al., 2006; Eyzaguirre et al., 2005). Lower quantities of bG are required for an SSF reaction relative to HHF or SHF reactions, as in SSF, the enzyme cleaves soluble glucooligomers to glucose, which is rapidly taken up by the fermentation organism. Many cellulolytic fungi produce one or more bGs, but often the total bG protein is not commonly abundant, making up less than 1% of total secreted protein in most wild type secretomes (Chundawat et al., 2011). Interestingly, some bacteria do not depend on exogenous beta-glucosidase, instead using cellobiose phosphorylase to drive metabolism from celluolytic degradation, a strategy that was recently engineered into yeast (Sadie et al., 2011). Engineering of fermentation strains with an effective cellobiose transporter (Ha et al., 2011) may allow for improved sugar uptake when enzymes are optimized for SSF.

To support SHF and HHF, biotechnological improvement of bG levels and biochemical properties in fungal enzyme cocktails is required. Beta-glucosidase activity must be maintained in the presence of high concentrations of its product, glucose, to prevent cellobiose from accumulating. Cellobiose is a potent inhibitor of CBH and EG. For industrial biomass conversion targeting high feedstock loads, supplementing bG to non-engineered microbial cellulolytic enzyme preparations can be imperative, because of high cellobiose level during the enzymatic conversion.

With regard to developing enzyme cocktails with improved bG activity, technology for improving bG expression in filamentous fungi is critical. To support optimal performance of bGs in high temperature hydrolysis, at process-relevant pH, improvement of bG stability has been accomplished (Wogulis et al., 2011; Postlethwaite and Clark, 2010; Lamsa et al., 2004; Hill et al., 2012). Characterization of bGs that are resilient to glucose inhibition may also be relevant to their industrial application (Fang et al., 2010).

7.4.2 Cellobiohydrolases

Cellobiohydrolases (CBHs) make up the bulk of many fungal cellulolytic cocktails. The enzymes have tunnel-like active sites, and they processively cleave cellobiose units from the ends of cellulose chains. CBH I enzymes (EC 3.2.1-) cleave sugars from the reducing ends of cellulose, while CBH IIs (EC 3.2.1.91) are specific towards non-reducing ends. In fungi, CBH I and CBH II activities are associated with GH7 and GH6 families, respectively. Many fungal CBHs are associated with type 1 carbohydrate binding modules (CBM1s), and these may assist the catalytic core with processivity (Beckham et al., 2010; Tavagnacco et al., 2011).

In the past few years, insights about molecular limitations of cellobiohydrolases have highlighted potential for their improvement by rational protein design. Processive CBH movement can be obstructed by kinks or other impediments on the cellulose surface (Igarashi et al., 2011) and it has been suggested that k(off) values may be a major factor in CBH efficiency (Kurasin and Valjamae, 2011; Praestgaard et al., 2011). CBH variants with potential for improved activity due to decreased binding of product (product expulsion) have been constructed (Bu et al., 2011). Further, engineering of CBH I to reduce end-product inhibition has been performed (Healey et al., 2012). Stability of CBH I and II is known to be limiting to high temperature hydrolysis performance in a range of cellulase cocktails, and recent developments in stabilization of these enzymes are promising (Heinzelman et al., 2009, 2010; Voutilainen et al., 2009, 2010).

7.4.3 endo-1,4-β-Glucanases

Endoglucanases (EGs, EC 3.2.1.4) hydrolyze internal glycosidic bonds in cellulose. EGs generally have ‘cleft’-like active sites, and they are described as hydrolyzing and then dissociating, though some bacterial EGs are known to act ‘processively’ on crystalline cellulose (Li et al., 2007). As endoglucanases create new ‘ends’ for CBHs to bind to, it is not surprising that significant synergism is observed between EGs and CBHs. EGs are found among numerous GH families, with fungal examples predominantly in GH5, 7, 12, and 45. Fungal EGs are often associated with CBM1s.

7.4.4 GH61s

GH61 proteins are abundant in diverse fungal systems. The presence of this type of protein in fungal secretomes and genomes was described 15 years ago (Saloheimo et al., 1997), but their industrial relevance was first appreciated by scientists who uncovered GH61s as potent enhancers of cellulolytic cocktails. Addition of recombinant GH61s to Trichoderma cellulase cocktails was shown to greatly enhance the activity of these cellulases in lignocellulose degradation, lowering the required enzyme concentration for substrate breakdown by a factor of two (Harris et al., 2010). Originally, T. reesei and Aspergillus kawachii GH61s were classified as weak endoglucanases (Saloheimo et al., 1997; Karlsson et al., 2001; Koseki et al., 2008). The first published crystal structures of a GH61 indicated that these are novel enzymes and not EGs (Karkehabadi et al., 2008; Harris et al., 2010). No obvious cleft or hydrolytic active site could be identified in the structure; instead, a divalent metal-binding site was observed on the surface of the protein. Structural similarities between GH61 and a chitin binding protein 21 (CBP21, belonging to CBM33) were reported. Further clues to the mechanism of GH61 were obtained when it was shown that the protein could not stimulate hydrolysis in the presence of cellulases on relatively pure cellulose substrates, which suggested that a chemical present in lignocellulose but absent in cellulose might be a requirement for GH61 activity (Harris et al., 2010).

Elucidation of the function of GH61 (and the related chitin cleaving CBM33 enzymes from bacteria) has recently been accomplished. Cleavage of cellulose (or chitin, for CBM33) has been demonstrated, and a clear requirement for a redox-active factor is shown (Langston et al., 2011; Phillips et al., 2011; Beeson et al., 2012; Forsberg et al., 2011; Quinlan et al., 2011; Westereng et al., 2011). Quinlan et al. (2011) showed that pretreated biomass contains soluble redox cofactors that potentiate GH61 activity, which explains the previous results where no activity was found on pure cellulose in the absence of these soluble cofactors. GH61s use reducing equivalents (provided by small molecule reductants or by cellobiose dehydrogenase) and oxidatively cleave cellulose in a copper-dependent fashion. The mechanistic details of this novel reaction mechanism are still being explored. In some studies, cleavage products are oxidized on the reducing end (Langston et al., 2011; Phillips et al., 2011; Westereng et al., 2011), while others show non-reducing end oxidation (Langston et al., 2011; Phillips et al., 2011) or oxidation at both positions (Quinlan et al., 2011). Recently, the structures of two more members of the GH61 family were solved, and the observation of trapped oxygen species in the crystal supports a recently framed mechanism of action wherein oxygen is inserted into cellulose at C–H bonds adjacent to the glycosidic bond, rendering the cellulose chain unstable, which ultimately causes the bond to break (Beeson et al., 2012; Li et al., 2012).

7.4.5 endo-β-Xylanases and β-xylosidases

As was mentioned in Section 7.2.1, the nature of the biomass pretreatment greatly impacts the degree of hemicellulose conversion, and thus has a direct effect on the relative requirement for xylanases and beta-xylosidases. The total xylanase activity secreted by wild type fungi is often insufficient for effective conversion of pretreated biomass. Even when near-complete conversion of xylan to monomeric sugars is accomplished during a pretreatment, as is the case with dilute acid catalyzed pretreatments, a xylanase may benefit the conversion of cellulose by removing the small amounts of remaining insoluble xylan material, thus allowing greater access to cellulose. For biomass feedstocks that have high insoluble xylan content, xylanase and beta-xylosidase enzymes are critical to effective hydrolysis.

While the structure of hemicellulose is diverse and depends on the source of the feedstock, endo-xylanases (EC 3.2.1.8, xylanases) are known to be active on materials with different O-substitutions by acetyl, glucuronoyl, arabinosyl, and other modifications, although the ability to cleave within the xylan backbone close to these substitutions is variable. Xylanases are found widely among different GH families, including GH8, 10, 11, 30, and 43, but the ones from fungi that are most commonly used in biomass applications belong to GH10 and GH11, and these differ in specificity. GH10 xylanases are more active on substituted xylan relative to GH11s and produce shorter oligosaccharides (Ustinov et al., 2008). Xylanases often contain carbohydrate binding modules (CBMs).

Many pretreatments can solubilize hemicellulose, but do not completely convert oligomeric xylan to xylose. Inclusion of beta-xylosidases (BX, EC 3.2.1.37) in enzyme cocktails is beneficial in many cases to complete conversion of soluble xylooligosaccharides. BXs have catalytic cores belonging to the GH3, 30, 39, 43, 52, and 54 families.

As for cellulases, the hemicellulases show synergy between different classes of activity, so it is often of significant benefit to include a complete array of xylanase, beta-xylosidase, and debranching activities. When working with hardwood or feedstocks (such as many agricultural residues and energy crops) that are classified as grasses, the enrichment in arabinoxylan may require careful optimization of hemicellulase components.

7.4.6 Hemicellulase ‘debranching’ enzymes

Among ‘debranching’ enzymes which can catalyze the removal of arabinosyl substitutions on xylan, α-L-arabinofuranosidases (EC 3.2.1.55) may be beneficial for industrial enzyme applications where the hemicellulose structure has not been greatly disrupted during pretreatment. These enzymes are found in GH3, 43, 51, 54, and 62, and often in addition contain CBMs (Saha, 2000). Depending on GH origin, they prefer singly substituted xylose sites (O2, 3, or 5), or rather disubstituted xylose sugars (Ara esterifying O2 and 3). As for arabinose substitutions, it is sometimes also beneficial to include enzymes that remove glucuronoyl or glucuronoyl methyl esters from the xylan backbone, through use of alpha-glucuronidases (EC 3.2.1.139), which can be found in GH67 and GH115. Likewise, removal of galactose substitutions, which are found in galactomannan, pectin, and other hemicelluloses, can sometimes improve performance, and this is accomplished by alpha-galactosidases (EC 3.2.1.22), a group of enzymes whose catalytic cores belong to GH4, 27, 36, 57, and 110 families. Softwoods and their abundant arabinoglucuronoxylan, and grasses which contain arabinoxylans are substrates which may require these debranching activities, depending on the pretreatment type and severity.

Several types of carbohydrate esterases are relevant to biomass hydrolytic enzyme cocktails. Esterases that can cleave acetyl, feruloyl, and glucoronyl moieties within hemicellulose are advantageous in order to allow for complete conversion of some highly decorated hemicellulose backbones. Acetyl xylan esterases (EC 3.1.1.72), feruloyl esterases (EC 3.1.1.73), and/or glucuronoyl esterases (EC 3.1.1.-) may enhance the activity of other hemicellulase components, especially when the substrate is acetylated hardwood xylan or ferulated grass arabinoxylan, and when a lower severity pretreatment is used. Different feruloyl esterases have different specificity towards different hydroxycinnamoyl ester bonds, which are involved in linking hemicellulose to lignin (Benoit et al., 2008). Glucuronoyl esterases can assist α-glucuronidases by hydrolyzing glucuronyl ester linkages.

7.4.7 Other activities

A range of other substrates may also be beneficial to biomass hydrolytic cocktails, depending on the feedstock that is used. These include mannanases, which may help with softwood degradation. For substrates that contain pectin, such as sugar beet pulps and orange peels, a wide range of pectinolytic enzymes are required. Xyloglucanases and beta-glucanases may be of use, and are often present in complex secretomes of naturally occurring cellulose-degrading microbes.

In addition to components where the biochemical activity is known, various other proteins have been reported to confer positive benefit to hydrolysis when included in combination with cellulases and hemicellulases. It should be noted that even non-catalytic proteins such as bovine serum albumin (BSA) have an effect if the protein to biomass ratio is high enough. Acting as a ‘blocking agent,’ non-catalytic protein can non-specifically bind to components such as lignin and thereby prevent inhibitory adsorption by cellulases and hemicellulases. Including BSA or other ‘negative control’ proteins may help to discriminate proteins with biomass specific, and potentially catalytically, active candidates.

One class of proteins which may have potential in a biorefinery setting are the expansins; these proteins play a role in cell wall ‘loosening’ in plants (Sampedro and Cosgrove, 2005). Fungi and bacteria have been noted to secrete proteins with sequence homology to plant expansins; fungal candidates have been dubbed ‘swollenins’ or ‘loosenins.’ Swollenin from T. reesei was noted to disrupt the structure of cotton and filter paper cellulose without increasing reducing ends (Saloheimo et al., 2002). Reports in the literature describing cellulose enhancement by fungal swollenins and loosenin (Wang et al., 2010; Wang et al., 2011; Jager et al., 2011; Chen et al., 2010; Quiroz-Castaneda et al., 2011) and also by bacterial expansin-like proteins (Lee et al., 2010; Kim et al., 2009) have emerged in the past few years. Further study will be required in order to determine the mechanism by which these molecules increase lignocellulose conversion and allow industrial exploitation of this class of protein.

7.4.8 Thermostabilization and development of ‘thermally active’ enzyme cocktails

Including appropriate biochemical activities and designing appropriate mixtures to optimize synergy is a fundamental activity in enzyme development. In addition it is helpful to obtain highly active representatives from each enzyme type. Enzyme kinetics are very much temperature driven; in general, the higher the temperature, the faster the reaction. However, as temperature increases, so does protein denaturation/inactivation. Discovery of wild type enzymes from thermotolerant fungi is a popular strategy for obtaining highly efficient enzymes (Viikari et al., 2007). In addition, engineering of proteins so that they retain activity during high temperature is effective, through use of rational design and/or protein evolution campaigns. The multicomponent nature of effective cocktails often requires that a number of protein components be stabilized. By iteratively introducing more and more stable components, the temperature profiles of a complete mixture are often shifted to a higher temperature. The goal is to achieve higher performance in conversion per unit protein by increasing the rates of reaction of the biological catalysts through improvement of ‘thermal activity.’

7.5.1 Benchmarking the state of enzyme technology

Iterative improvements to enzyme cocktails by corrections of individual enzymes, addition of new components, and optimizing of mixtures for improved synergy and high temperature performance all contribute positively to biochemical conversion economy. Arguably, monitoring enzyme performance improvements is best done by comparing the enzyme dose required for a given degree of carbohydrate to monomer conversion of a given, pretreated feedstock. Comparative observations of enzyme initial rates of reaction, while important toward understanding the fundamental properties and mechanisms of individual enzymes, may not predict performance in an industrially relevant setting, where high extents of conversion of biomass to sugars are required. Looking at enzyme dose to achieve an economically relevant conversion target allows for monitoring of dose requirement, and the reduction in enzyme dose that is achieved through engineering of a complex cocktail is a measure of technological improvement. In benchmarking performance, the ideal conditions should be ones that mimic industrially relevant conversion process flows. As was described in Section 7.2.2, this may necessitate that hydrolysis is performed in the presence of soluble inhibitors, as the costs of solid–liquid separations may be prohibitive. Also, the benchmarking of enzyme performance should be done at high total solids loadings, reflecting the importance of maintaining high sugar concentrations for production of high product titers.

As an example of published fold enzyme reductions, Novozymes in 2005 reported progress on a Department of Energy (DOE) funded subcontract to the National Renewable Energy Laboratory (NREL) (Teter and Cherry, 2005a, 2005b). During the course of a four-year project, Novozymes and NREL reduced the enzyme requirement for converting 80% of insoluble cellulose to glucose in dilute acid pretreated corn stover (PCS) by six-fold (Fig. 7.4). The lab-scale assay (50 g) was run under conditions that are relevant to an industrial process (pH 5.0 and 50 °C for 7 days), with the caveat that the substrate used was washed NREL dilute acid PCS.

A second enzyme improvement project has been concluded at Novozymes (Project DECREASE, September 2008–October 2011). Researchers have further reduced the enzyme dose required for 80% conversion of total glucan and xylan-based carbohydrate in NREL dilute acid PCS by 1.9-fold (Fig. 7.5). In the most recent benchmarking, a larger lab-scale reaction was used (500 g), with 21.5% TS whole slurry (unwashed) PCS as the substrate for 7 days of hydrolysis. For 80% conversion, a 1.9-fold dose reduction is observed for CTec3 relative to the reference enzyme cocktail.

7.5.2 Techno-economic modeling and optimizing biochemical conversion processes

While measuring the relative improvements in enzyme performance through benchmarking dose reductions is valuable, the industrially relevant question is whether their use enables a process which allows for competitive production costs. Techno-economic modeling is performed to quantify costs that the current technology would allow in a mature industry (so-called ‘n^th generation’ plant assumptions). As the process is composed of discrete unit steps, the modeling seeks to quantify the contribution of individual steps to capital and operating costs. In this context, various models have been used to quantify the enzyme cost component.

Examples of techno-economic modeling include NREL’s design reports, formulated to model the economics of producing ethanol from cellulose. Published over the last two decades, these ASPEN PLUS-based models show the assumptions and crude data used to build the models in exhaustive detail (Aden et al., 2002; Wooley et al., 1999; Humbird et al., 2011). These models aimed to quantify the impact of technology that the researchers deemed reasonable to achieve in 2012. The key parameter calculated was the minimum ethanol selling price (MESP, $/gal), defined as the lowest price required to generate a 10% internal rate of return (IRR). Important design considerations can be found in the reports themselves, and summarized in Foust et al. (2009). Underscoring interest in producing a range of fuels and commodity chemicals from biomass sugars, NREL’s 2011 design report included a section that ‘backed-out’ the cost of producing sugar at a cellulosic ethanol plant, toward capturing reference costs for producing mixed pentose and hexose sugars from biomass in this context. The model postulated a 2012 minimum ethanol selling price of $2.15/gal (2007 dollar basis) for a 2,205 ton/day plant.

Numerous other examples of techno-economic models for production of cellulosic ethanol exist, with published methodology presented in less detail than in the NREL design cases. An example includes a model for costs associated with conversion of AFEX-pretreated corn stover where various co-products in addition to combined heat and power (CHP) were assumed, and where enzymes were produced on site (Lau et al., 2012). A model simulating a mature CBP process with ‘aggressive’ performance parameters using NREL-based/ASPEN-based models was published by Lynd et al. (2005).

In the studies cited, the enzyme costs are quite variable, as is also noted by Olson et al. (2012). While this suggests that there is disagreement regarding economy of enzyme use, in fact the differences are not surprising given that the processes themselves are quite variable. As an illustrative example, the quantity of requirements for conversion of well-washed pretreated corn stover is lower than for unwashed slurry, which is responsible in part for the differences in estimated enzyme use costs. Optimizing a process for total cost reduction is not the same as optimizing for lowest enzyme cost. As a new generation of enzymes is developed, producers may elect to keep total enzyme quantity per gallon at the same level as for a previous less effective cocktail, but alter the process to allow for more dramatic cost reductions. Improved enzymes may allow for a higher total solids loading, or for a shorter process time, or both (Simms-Borre, 2012).

Toward understanding how Cellic® enzymes impact total ethanol production costs, Novozymes has developed a techno-economic model. The cost model is an Excel-based model derived from industrial partners’ input and the NREL 2002 Lignocellulosic Biomass to Ethanol Process Design and Economics report describing a dilute sulfuric acid pretreatment followed by hydrolysis and fermentation. The model accounts for the mass flows through the plant and scales equipment using vendor quotes and scalability factors from the NREL report or from Novozymes’ partner input. Table 7.1 describes some of the key assumptions used in the model. Lab-scale data reflecting mass balance flows of the pretreatment, hydrolysis, and fermentation processes are input into the model; thus, the model reflects the current state of technology.

While enzyme hydrolysis in a dedicated hydrolysis was shown in Fig. 7.5 as an example of benchmarking, for total process cost optimization, the best format for enzyme use is a hybrid hydrolysis and fermentation (HHF), and performance of Cellic®CTec3 with an advanced pentose/hexose co-fermenting strain is shown in Fig. 7.6. Cellic CTec3 was added at a commercially relevant dose at day 0. The proprietary C5/C6 co-fermenting yeast strain was added at a 1 g/L initial pitch at days 5, 0, and 3, respectively. As shown, the SHF and HHF have very similar and superior performance to the SSF configuration, both reaching high ethanol titers by day 7 or 8. Final ethanol titers show a clear benefit for several days of dedicated hydrolysis prior to fermentation, but there is significant flexibility in process time distribution. Using the base case model, sensitivity analysis reveals an ethanol selling price of $2.50/gal, before subsidies. The assumption is made that first generation plants that will be constructed will likely not achieve the economies predicted, but it is expected that a mature plant using the modeled technology should achieve the predicted economic parameters (n^th generation plant design).

The model is particularly useful for providing a basis for understanding where the total of process and capital costs can be reduced. Predicting a process design whereby lowered costs can be achieved is performed on a case-by-case basis, using input from partners, and data with variously pretreated feedstocks. ‘Sweet spots’ in the process indicate how a globally optimized process can best be designed. In the hydrolysis step, the most sensitive inputs are the solids and enzyme loading and the glucan conversion to glucose.