Genomics Review of Holocellulose Deconstruction by Aspergilli

Abstract

SUMMARY

Biomass is constructed of dense recalcitrant polymeric materials: proteins, lignin, and holocellulose, a fraction constituting fibrous cellulose wrapped in hemicellulose-pectin. Bacteria and fungi are abundant in soil and forest floors, actively recycling biomass mainly by extracting sugars from holocellulose degradation. Here we review the genome-wide contents of seven Aspergillus species and unravel hundreds of gene models encoding holocellulose-degrading enzymes. Numerous apparent gene duplications followed functional evolution, grouping similar genes into smaller coherent functional families according to specialized structural features, domain organization, biochemical activity, and genus genome distribution. Aspergilli contain about 37 cellulase gene models, clustered in two mechanistic categories: 27 hydrolyze and 10 oxidize glycosidic bonds. Within the oxidative enzymes, we found two cellobiose dehydrogenases that produce oxygen radicals utilized by eight lytic polysaccharide monooxygenases that oxidize glycosidic linkages, breaking crystalline cellulose chains and making them accessible to hydrolytic enzymes. Among the hydrolases, six cellobiohydrolases with a tunnel-like structural fold embrace single crystalline cellulose chains and cooperate at nonreducing or reducing end termini, splitting off cellobiose. Five endoglucanases group into four structural families and interact randomly and internally with cellulose through an open cleft catalytic domain, and finally, seven extracellular β-glucosidases cleave cellobiose and related oligomers into glucose. Aspergilli contain, on average, 30 hemicellulase and 7 accessory gene models, distributed among 9 distinct functional categories: the backbone-attacking enzymes xylanase, mannosidase, arabinase, and xyloglucanase, the short-side-chain-removing enzymes xylan α-1,2-glucuronidase, arabinofuranosidase, and xylosidase, and the accessory enzymes acetyl xylan and feruloyl esterases.

INTRODUCTION

Plants capture photons from sunlight to fix carbon dioxide into sugars via the Calvin cycle, and they deposit most of the carbohydrates onto plant cell walls (1). Plant cell walls are major constituents of what is commonly defined as biomass, a huge component of all terrestrial habitats that is mostly represented by the cell walls of dead plants, trees, grasses, and leftovers of actively managed croplands. Biomass is also a natural waste product that accumulates along sugar mills and alcohol refineries worldwide (2, 3).

However, plants, with their robust and refractory cell walls, die and are completely recycled in nature. Polysaccharides are deconstructed and resulting sugars assimilated by all microorganisms that live in soil and forest floors (4). Furthermore, natural decomposition of woody (most of the available biomass) materials remains a recalcitrant process that takes place in soil microbiomes and perhaps employs numerous enzymes derived from not one organism but a whole community of microbes (5).

Bacteria are major components of soil environments and are equipped with many of the enzymes needed to degrade plant cell walls (6, 7). Fungi are also present in soil and in the rumina of herbivores and are rich in polysaccharide-degrading enzymes and active in recycling plant cell walls (8, 9). For example, Aspergillus nidulans grown on sorghum stover under solid-state culture conditions secreted a total of 294 proteins, predominantly hemicellulases, cellulases, polygalacturonases, chitinases, esterases, and lipases, over a 2-week period, while the fungus used only 30% of the total available hemicellulose and cellulose accessible from sorghum stover (10).

However, not all fungi are alike; some fungi, such as Trichoderma reesei and Aspergillus niger, employ predominantly a classical acid catalysis hydrolytic model of degrading plant cell wall polymers, while others, such as Myceliophthora thermophila (11, 12) and Phanerochaete chrysosporium (13, 14), appear to take advantage of an oxidative route of breaking down glycoside bonds. Even though some fungi predominantly hydrolyze biomass while others employ an oxidative mechanism, most fungi contain a mixed set of hydrolases and oxidases. Neurospora crassa (15, 16) and Aspergillus species appear to harbor a mixed acid hydrolysis and oxidative system.

In the last decade, biomass degradation by fungal enzymes has received renewed attention because of the need to utilize biomass as a food source to drive fermentation processes that manufacture biofuels and other feedstocks for chemicals and pharmaceuticals. Seamless integration of a biomass-to-sugar conversion process with traditional sugar-to-ethanol fermentation is the bottleneck of the biofuel industry (17–23). Biomass carbohydrate heterogeneity and structural fiber complexity (recalcitrance) result in water-deprived structures that hinder the access of enzymes.

The genomic DNA sequences of numerous Aspergillus species have been determined, including Aspergillus oryzae (24), Aspergillus fumigatus (25), and Aspergillus niger (26). Genome-wide reports comparing Aspergillus flavus with A. oryzae (27) and A. nidulans with A. fumigatus and A. oryzae (28) and a curated comparative genomics resource for aspergilli (29) are available as well. Several articles describing plant cell wall-degrading enzymes have been published, mainly focusing on glycoside hydrolases (30, 31) or fungal sets of plant polysaccharide-degrading enzymes (32, 33), as well as comparisons of A. nidulans with A. niger and A. oryzae (34–36). The last comprehensive review of cellulose- and hemicellulose-degrading enzymes produced by Aspergillus was published in 2001 (37).

Here we propose a comprehensive genomic review of seven sequenced Aspergillus genomes and compare them with the model biomass degrader T. reesei and the genetic model system N. crassa. In examining gene models encoding cellulases and hemicellulases in fungi, one encounters gene duplications and protein products with overlapping biochemical functions, illustrating an evolutionary effort to acquire functions capable of degrading the recalcitrant plant cell wall. As a result, the combination of genetic multiplicity (gene duplications or multiple acquisitions of the same gene, followed by evolution into specialized functions) and functional redundancy (genes with diversified biochemical functions, such as exo- and endohydrolysis of the same substrate) creates a complex and overlapping repertoire of enzymes, for which we provide a comprehensive overview.

HOLOCELLULOSE ENZYME BREAKDOWN SYSTEMS

Holocellulose is the carbohydrate fraction of biomass (lignocellulose) that includes the total polysaccharide fraction obtained after extractives and lignin have been removed from natural materials. (A rough schematic is shown in Fig. 1.) Holocellulose (∼65 to 85%) contains polymers, such as cellulose (∼30%), hemicellulose (∼20 to 30%), and pectin (∼5 to 30%), that incorporate sugars as their basic repeating units (38). These sugars represent a massive pool of C—H-bonded energy molecules locked into recalcitrant polymers.

FIG 1.

Canonical holocellulose structure and deconstructive hydrolytic enzyme interactions. The main polymers integrating biomass are lignin (boxes) and holocellulose, which includes hemicellulose (light-colored, loosely branched chains) and cellulose (black linear bundled chains). Sugars: X, xylose; A, arabinose; Gc, glucuronic acid; M, mannose. Open hexagons, ferulic acid; closed circles, acetyl groups. Biomass is the principal carbon sink on earth and recruits numerous enzymes to deconstruct cellulose and hemicellulose to sway the carbon cycle via the central energy metabolism. Enzymes needed to deconstruct holocellulose include the following: cellulases, i.e., cellobiohydrolases and endoglucanases, with and without CBMs, β-glucosidases, copper-dependent lytic polysaccharide monooxygenases (LPMOs), and cellobiose dehydrogenases; and hemicellulases, i.e., xylanases, mannosidases, xyloglucanases, xylan acetyl esterases, feruloyl esterases, arabinanases, glucuronidases, and arabinofuranosidase xylosidases.

Cellulose, hemicellulose, and lignin quantities are variable in plants (Table 1). For example, in sugar cane bagasse, the main hemicellulose is arabinoxylan, and in soft- and hardwoods, hemicellulose contains glucomannan and arabinan (39, 40). In addition, pectin also contains a carbohydrate-formulated polymeric complex. However, it appears at various rates in different plant tissues. While abundant in fruits, very little pectin is present in traditional biomass, such as wood, sorghum, and corn stover or sugar cane bagasse. Thus, the main pectin-degrading activities lie outside the scope of this review.

TABLE 1.

Polysaccharide composition of energy crops and wood

Energy crop	% Sugars in juice	% Biomass in bagasse			Reference
		Cellulose	Hemicellulose	Lignin
Sugar cane	9.8	43	24	22	123
Sweet sorghum	11.8	45	27	21	123
Hardwood		38–51	17–38	27–32	171
Softwood		33–42	22–40	21–31	171

Cellulose, a linear polymer, consists of d-glucose units connected to each other by glycoside β-1,4 linkages. Each glucose unit is rotated 180° from its neighbor molecule, enabling long linear chains with lengths of 2,000 to 25,000 glucose residues. Typical cellulose microfibrils are composed of 36 parallel cellulose chains in a paracrystalline, linear, hexagonal arrangement with a 10-nm diameter, known as type I cellulose (41). The hydrogen bonds between adjacent cellulose polymers form crystalline structures that give plants structural strength. Cellulose occurs in helically wound reinforcing crystalline microfibrils as well as in an amorphous form glued with lignin and hemicellulose to serve as a support matrix (42, 43). Cellulosic microfibrils are approximately 70% crystalline, have an unusually high tensile strength, are impermeable to water and resistant to chemical and biological attack, and are so stable that they are very difficult to break.

To further enhance cellulose recalcitrance, the glucose repeating unit has six hydrogen bond donors and nine hydrogen acceptors, offering numerous ways of establishing inter- and intramolecular hydrogen bonds. Moreover, due to idiosyncratic arrangements of the pyranose ring and conformational changes of hydroxymethyl groups, cellulose chains form microfibrils which exhibit distinct crystal structures (44–48).

Four distinctive crystalline allomorphs of cellulose have been recognized based on X-ray diffraction patterns and solid-state ¹³C nuclear magnetic resonance (NMR) spectra (49–51).

Cellulose I, the most abundant form found in nature, is important for bioenergy production and is a mixture of two distinct crystalline forms: cellulose Iα, a triclinic one-chain unit with parallel chain stacks with progressive shear parallel to the chain axis, the predominant form isolated from bacteria and fresh water algae (52); and cellulose Iβ, a monoclinic two-chain unit with parallel cellulose chains stacked with alternating shear, the major form in higher plants, such as cotton, wood, sugar cane, sorghum, switch grass, ramie, and animal cellulose (53). Cellulose II is the most crystalline, thermodynamically stable form and is a monoclinic unit with antiparallel chains (parallel in cellulose Iβ) forming a highly rigid macromolecule due to the presence of a three-dimensional (3-D) hydrogen bond network in addition to the C—O—C bonds between the glucopyranose rings. In the absence of such hydrogen bond networks, the chains become flexible (45). Cellulose III_I, III_II, and IV are structural variations which appear when cellulose II or I is treated with chemicals such as ammonia and/or subjected to high-temperature treatments (54–56).

The second major polysaccharide fraction is hemicellulose (∼30%), a carbohydrate component loosely defined as polymers extractable by alkaline solutions. Typical alkali-soluble polymers are xylans, mannans, arabinoxylan, xyloglucans, and pectins that contain, in addition to d-glucose, other hexoses (d-mannose, d-galactose, d-fucose, l-rhamnose, and d-galacturonic acid) and pentoses (d-xylose and l-arabinose) (57, 58). Examples of hemicellulosic polysaccharides are xylans, with a linear xylose β-1,4-linked backbone with few side chain substitutions, and xyloglucan, a β-1,4-linked glucan chain to which xylose residues are bound via α-1,6-glycosidic linkages.

Four main types of backbones are common in hemicelluloses (58, 59). Xylans possess a β-1,4-d-xylopyranose backbone, with glucoronoxylan and arabinoxylan being variants found in plant cell walls. Xyloglucans contain a β-1,4-d-glucopyranan backbone decorated with α-d-xylopyranose residues at position 6 and form hydrogen bonds with cellulose microfibrils. Mannans are polymers possessing a β-1,4-d-mannopyranose backbone, whereas galactomannans and glucomannans are variants that are part of the secondary cell wall of softwoods and, finally, β-glucans are β-(1,3-1,4)-d-glucans found in cereal grains.

Pectins are also extracted with alkali and contain galacturonic acid, rhamnose in backbone structures, and apiose, galactose, and arabinose as side chains, but they are abundant in fruits and not in biomass, such as wood, sorghum, sugar cane bagasse, or corn stover (60).

The third most abundant polymer present in biomass is lignin (∼30%), a polymer that surrounds secondary cell walls, resulting in lignified tissues creating a structure that provides mechanical support (e.g., branches and twigs of trees or stems of herbaceous plants). Lignin is formed by polymerization of phenyl propane derivatives, i.e., coumaryl, coniferyl, and sinapyl alcohols, resulting in a solid polymeric structure primarily linked by ether bonds and chemically (covalent bonds) interlinked with hemicellulose (61, 62).

Hemicellulose molecules are cross-linked by phenols, the most abundant of which are para-coumaryl and feruloyl acids. Complexes of feruloylated xyloglucan and a p-coumaroylated arabinoxylan have been isolated (63, 64), and feruloylated α-1,5-linked arabinan and β-1,4-linked galactan have also been observed (64). The pulp of spruce and pine wood yielded lignin carbohydrate β-1,4-d-galactan complexes (61). Interestingly, a small amount of arabinose was also found in lignin-carbohydrate complexes not associated with arabinoxylan (65), thus suggesting cross-linking via ferulic and/or p-coumaric esters to arabinogalactan, α-1,5-linked arabinan, and β-1,4-linked galactan (65).

In summary, cellulose is a linear polymer that is crystalline, strong, and resistant, while hemicellulose is branched and amorphous, with little strength, and pectin along with lignin amalgamates the plant cell wall matrix.

Nearly all fungi and other microorganisms, such as bacteria and archaea, produce enzymes that at least partially degrade plant cell wall polysaccharides (66, 67). They deconstruct cellulose, hemicellulose, lignin, and pectin into simple sugars and phenols, which upon assimilation feed the central energy metabolism to produce chemical energy (ATP) through oxidation, releasing carbon dioxide and water and thus completing the carbon cycle, whereupon carbon dioxide can again be fixed through photosynthesis to produce sugars and plant cell walls.

Figure 1 is a schematic that depicts all significant components of plant cell walls, along with known fungal enzymes adequate for whole conversion of holocellulose (cellulose and hemicellulose) into simple sugars, which are easily metabolized further by fungi and other microorganisms into bioproducts.

The carbohydrate-active enzymes (CAZy) database collectively compiles and assigns into families glycoside hydrolase (GH)-, glycosyl transferase (GT)-, polysaccharide lyase (PL)-, carbohydrate esterase (CE)-, and auxiliary activity (AA)-encoding genes, according to a classification system developed by Henrissat and coworkers (68, 69), based on amino acid sequence similarity, secondary and tertiary fold conservation, and stereochemical architecture of catalytic mechanisms, i.e., inversion or retention of the anomeric configuration (68, 70). According to Jovanovic and coworkers (71), only 22 of 114 GH families are critical for biomass decomposition, and 20 are populated with genes from filamentous fungi, including those encoding endo- and exo-acting cellulases, hemicellulases, backbone-degrading and debranching enzymes, and glucosidases (71).

In the historical view of cellulose degradation, only three types of cellulases are required (Fig. 1): cellobiohydrolases that attack the nonreducing (CAZy family GH6) or reducing (GH7) end of a cellulose chain, endoglucanases (GH5 to GH9, GH12, GH44, GH45, GH48, GH51, GH61, and GH124) that cleave internal linkages of cellulose molecules, producing cellobiose, and β-glucosidases (GH1 and GH3) that cleave cellobiose into two glucose molecules. However, some of the “weak” endoglucanases formerly classified as GH61 enzymes were recently recognized to function as copper-dependent lytic polysaccharide monooxygenases (LPMOs) along with cellobiose dehydrogenase, which breaks down cellulose via an oxidative reaction, and therefore were regrouped into a new family: the auxiliary activity family 9 (AA9) for LPMOs (15, 72–75).

For hemicellulose backbone degradation, at least nine different enzymes are needed, depending on the plant's specific cell wall composition (Fig. 1): xylanases (GH10 and GH11) cleave xylose glycoside linkages (76) from xylan homopolymers (77); arabinofuranosidases (GH3, GH51, GH54, and GH62) promote hydrolysis of α-1,2-, α-1,3-, and α-1,5-l-arabinofuranosidic bonds in arabinoxylan, arabinan, and other arabinose-containing hemicelluloses (78); alpha-glucuronidases (GH67 and GH115) hydrolyze the ester linkage between the 4-O-methyl-d-glucuronic acid of glucoronoxylan and lignin alcohols (79); xyloglucanases hydrolyze glycosidic bonds of branched (GH74) or unbranched (GH12) glucose residues of xyloglucan (80); arabinanases (GH43 and GH93) hydrolyze α-l-arabinofuranoside linkages of α-1,5-l-arabinan and release α-1,5-l-arabinobiose from the nonreducing end (81); mannosidases or mannanases (GH2, GH5, and GH26) catalyze random hydrolysis of β-1,4-mannosidic linkages of mannans (82); acetyl xylan esterases (CE1, CE3 to CE5, CE7, and CE16) hydrolyze the O-acetyl substituent present in xylan backbones (83); feruloyl esterases (CE1) catalyze the cleavage of ester bonds between plant cell wall polysaccharides and phenolic acid, mainly ferulic acid (84); and xylosidases (GH3 and GH43) hydrolyze successive xylose residues from the nonreducing end of xylo-oligomers (85).

ENZYME STRUCTURE-FUNCTION AND SUBSTRATE RELATIONSHIPS

Microbial gene models encoding plant cell wall-degrading enzymes exist abundantly in bacteria and fungi. Genes appear as multiple copies (duplications or multiple sequential acquisitions) in the genomes of sequenced microorganisms, and enzymes (proteins) appear to be functionally redundant (with different three-dimensional structural folds harboring similar biochemical functions) or have overlapping biochemical functions (illustrated by GH10 and GH11 xylanases) (60, 86). Furthermore, cellulose-degrading enzymes (Table 2), such as cellobiohydrolases and endoglucanases, comprise similar catalytic functions: endoglucanases hydrolyze glucose-glucose glycoside bonds randomly, while cellobiohydrolases cleave the same glycosyde bond from terminal ends of cellulose molecules, producing cellobiose. Furthermore, two types of cellobiohydrolases are routinely found: CAZy families GH6 and GH7, which cleave cellobiose from the nonreducing and reducing ends, respectively. Endoglucanases also appear as functional variants, based on the structure relationships of the CAZy classification system (GH5, GH7, GH12, and GH74).

TABLE 2.

Complete cellulose degradation gene complement^a

^aFor complete gene and protein information, refer to Tables S1 to S5 in the supplemental material. Gray-shaded entries are totals.

Hemicelluloses (Fig. 1) are structurally complex polymers containing a variety of monomers (xylose, arabinose, glucose, and glucuronic acid), backbones, and side chains with substituted sugars and attached phenols and therefore require a larger number of distinctive enzymes for complete degradation (see Table 8). Thus, the more recalcitrant a polymer is, even if it is bonded homogeneously, the more redundant functional enzymes are employed, and the more diversified a structural design is, the more functional types with less redundancy are demanded. This apparent gene redundancy and functional diversity related to the substrate are reflected in the genome content of the group of fungi studied here (see Tables 2, 8, and 10).

TABLE 8.

Fungal hemicellulases^a

^a±, with or without.

TABLE 10.

Genome-wide acetyl xylan and feruloyl esterase content in aspergilli

In order to determine the functional importance of this observed redundancy (of genes) and functional multiplicity (of proteins), we examined the full gene complements of seven Aspergillus genomes to precisely define which genes encode functionally similar proteins and to spot genes that appear to be redundant. For example, we show that aspergilli encode exactly two cellobiohydrolases belonging to family GH7, which are structurally and functionally identical at the catalytic domain; however, one contains a carbohydrate-binding module (CBM), while the other does not. On the other hand, we also show that extracellular β-glucosidases are encoded by multiple loci with acquired functional specialization, such as pH and temperature sensitivity as well as susceptibility to inhibitors and activators. As we proceed with the descriptive analysis of the complete holocellulose gene complement, the initial notions of genetic multiplicity (multiple gene copies) and functional redundancy (proteins with similar functions) rapidly disappear once enzymes are organized into their functional and interactive components, and only a few cases of gene duplication remain.

The presented evidence provides strong indications that, over time, fungal genomes acquire multiple gene copies, perhaps through horizontal gene transfer, as shown for other systems (87) that are modified through recombination in order to adapt to the notoriously recalcitrant substrate accessibility of plant cell walls.

Cellulases

Cellulose is a structurally uncomplicated polysaccharide that is recalcitrant toward degradation due to its crystallinity and interconnection with other cell wall polymers, such as hemicellulose and lignin. Thus, even though hydrolysis of cellulose is a simple glucose-glucose glycoside bond rupture, several enzymes are needed to perform this function.

The available three-dimensional structures of cellobiohydrolases (Fig. 2) suggest a tunnel-like conformation around the active site, fitting a single cellulosic chain at the reducing (GH7) or nonreducing (GH6) terminus (88–90). In contrast, endoglucanases (GH5 to -9, GH12, GH44, GH45, GH48, GH51, GH61, and GH124) are shaped by an open groove or cleft, into which a linear amorphous cellulose chain can fit randomly (Fig. 3).

FIG 2.

GH6 (A to C) and GH7 (D to F) cellobiohydrolases. (B and E) Typical tunnel-shaped catalytic cleft found in cellobiohydrolases. (C and F) Cleft depth. Cellobiohydrolases fold into an enclosed catalytic core shaped by a β-sandwich with two large, antiparallel β-sheets packed onto each other, forming a long cellulose-binding tunnel (226). The cellulosic substrate chain has to travel through the tunnel, where β-1,4-glycosyl bonds of cellobiose molecules (dimers) are hydrolyzed off the ends (GH6 or GH7 enzymes). The three-dimensional structures are for Trichoderma reesei GH6 (CBHII; PDB entry 1QK2) (108) and GH7 (CBHI or Cel7A; PDB entry 4C4C) enzymes (227).

FIG 3.

Three-dimensional structures of GH5 (A and B), GH7 (C and D), and GH12 (E and F) endoglucanases. (B, D, and F) Well-defined open clefts in these endoglucanase families. Endoglucanases from the GH5 family show a catalytic module with a typical compact 8-fold β/α barrel architecture, forming an open cleft similar to those of GH7 and GH12 endoglucanases, which share the β-jelly-roll topology with an extended, open substrate-binding groove. Endoglucanases with the open cleft configuration bind randomly to internal portions of a cellulose chain and cleave β-1,4-glycosidic bonds, resulting in shortened fragments. The three-dimensional structures are for the Thermoascus aurantiacus GH5 endoglucanase Cel5A (PDB entry 1GZJ), the Trichoderma reesei GH7 enzyme EGI (PDB entry 1EG1), and the Trichoderma reesei GH12 enzyme EGIII (Egl3 or Cel12A; PDB entry 1H8V) (116, 119, 122).

The model T. reesei cellulase complement has been studied extensively and comprises four endoglucanases (EGI/Cel7B, EGII/Cel5A, EGIII/Cel12A, and EGV/Cel45A), a lytic polysaccharide monooxygenase (AA9; originally incorrectly identified as a hydrolase and therefore called EGIV/Cel61A), and two cellobiohydrolases (CBHI/Cel7A and CBHII/Cel6A) that act synergistically to break down cellulose into cellobiose (91–95). In addition, two β-glucosidases (BGLI/Cel3A and BGLII/Cel1A) hydrolyze cellobiose into glucose (91, 96–98).

Table 2 describes aspergillus hydrolytic cellulases, cellobiohydrolases, endoglucanases, and β-glucosidases as well as oxidative enzymes, lytic polysaccharide monooxygenases, and cellobiose dehydrogenase.

Cellobiohydrolases.

Two cellobiohydrolases, GH6 and GH7, hydrolyze cellobiose from the cellulosic nonreducing and reducing ends, respectively. Enzymes that sequentially remove cellobiose molecules from a cellulose chain are termed processive enzymes (99). Two forms of cellobiohydrolases are present in aspergilli: cellobiohydrolases with and without a CBM (Table 3). Enzymes with CBMs bind tightly to cellulose molecules and aid in the removal of cellobiose molecules from terminal ends, and cellobiohydrolases lacking CBMs are also present in all surveyed fungal genomes, showing clear processive activity (100, 101).

TABLE 3.

Genome-wide distribution of fungal cellobiohydrolases

Fungus	Cellobiohydrolase(s)
	GH6 (nonreducing end)		GH7 (reducing end)
	Without CBM	With CBM	Without CBM	With CBM
Strains with a complete cellobiohydrolase set
A. clavatus NRRL-1	ACLA_025560	ACLA_062560	ACLA_088870	ACLA_085260
A. nidulans FGSC A4	AN1273	AN5282	AN5176	AN0494
A. niger CBS 513.88	ANI_1_1704074	ANI_1_300104	ANI_1_2134064	ANI_1_1574014
A. terreus NIH 2624	ATEG_00193	ATEG_07493	ATEG_03727	ATEG_05002
Strains with an incomplete cellobiohydrolase set
A. fumigatus Af293		AFUA_3G01910	AFUA_6G07070	AFUA_6G11610
A. flavus NRRL-3357	AFLA_069820		AFLA_067550, AFLA_021870
A. oryzae RIB 40	AOR_1_734074		AOR_1_608164, AOR_1_1654194
Strains used for reference
H. jecorina taxid 51453		GUX2_HYPJE		GUX1_TRIRE
N. crassa OR 74A	NCU03996, NCU07190	NCU09680	NCU05104	NCU07340

Cellobiohydrolases (Fig. 2) belonging to family GH6 (EC 3.2.1.91), also known as cellulase family B (Cel6A and CBHII), are thought to hydrolyze β-1,4-d-glycosidic linkages in cellulose and cellotetraose, releasing cellobiose from the nonreducing ends of cellulose molecules (102–104). GH6 cellobiohydrolases function through an inversion of anomeric stereochemistry, as indicated by NMR (105) with cellobiohydrolase II from T. reesei (Hypocrea jecorina). The first 3-D structure of cellobiohydrolase II (Trichoderma reesei Cel6A) provided evidence recognizing the catalytic general acid in the inverting mechanism (106–108).

Cellobiohydrolases belonging to family GH7 (EC 3.2.1.176), also known as cellulase family C, are thought to hydrolyze β-1,4-d-glycosidic linkages in cellulose, releasing cellobiose from the reducing ends of cellulose molecules. GH7 enzymes retain two catalytic amino acid residues near the consensus, Glu-X-Asp-X-X-Glu, where the first Glu acts as the catalytic nucleophile and the second Glu as a general acid/base (109, 110). Processivity is the key property of cellobiohydrolases, which employ reducing-end exo- and endo-mode initiation in parallel (88, 90, 111, 112). Intrinsic processivity (absence of obstacles or blocking) is greater than observed processivity, indicating that cellobiohydrolase activity is limited by something blocking the enzyme rather than by the enzyme's intrinsic tendency to fall off a cellulose chain (112).

Two forms of cellobiohydrolase family GH7 (Cbh1 and CelD) are common to all aspergilli and have structurally similar catalytic domains (Fig. 2). However, only Cbh1 contains a CBM that binds to cellulose. Both enzymes are catalytically functional, but CelD does not have a CBM and has a four-amino-acid deletion on the tunnel-obstructing loop, providing a continuous opening in the absence of a CBM (100). The fact that only Cbh1 binds to the substrate and in combination with CelD exhibits strong synergy only when Cbh1 is present in excess suggests that Cbh1 unties enough chains from cellulose fibers to enable processive access of CelD (100). Interestingly, a direct correlation of water content of the substrate and the absence or presence of CBMs on cellobiohydrolases was established, suggesting that, in nature, mixing and matching of catalytic domains with cellulose-binding domains result in functional interactions that optimize the catalytic output (101).

Endoglucanases.

Endoglucanases hydrolyze β-1,4-glycoside bonds intramolecularly and randomly along the noncrystalline portion of cellulose molecules. Four CAZy families are present in aspergilli: GH5, GH7, GH12, and GH45. Most CAZy family endoglucanases show one gene per family, sometimes one with and another without a CBM. Some multiplicity is also observed in GH12 and GH5 (Table 2).

GH5 endoglucanases (Fig. 3), formerly known as cellulase A, are the main cellulases in fungi and other organisms as well (113, 114). They hydrolyze β-1,4-d-glycosidic bonds randomly and internally of the amorphous region of cellulose. The catalytic core domain of Cel5A from T. reesei determined at 2.05 Å shows a substrate-binding pocket consisting of a deep catalytic cleft within a shallow groove, consistent with other structural studies of GH5 endoglucanases (115). The Thermoascus aurantiacus GH5 endoglucanase, which consists of a catalytic module with compact 8-fold β/α-barrel architecture (116), has a long, tryptophan-rich substrate-binding groove suggesting substrate-binding subsites at positions −4 to +3, in addition to the two conserved catalytic glutamates (116).

GH7 endoglucanases (Fig. 3B) are similar to GH16 enzymes found in plants and to bacterial agarases, cleaving β-1,3- and/or β-1,4-glycosidic linkages. Members are related by amino acid sequence similarity, the retaining hydrolytic mechanism, and catalytic residue identity (117). These endoglucanases share the β-jelly-roll topology and the retaining catalytic mechanism (118–120). The endoglucanase catalytic core domain from T. reesei, determined at 3.6-Å resolution, reveals an extended, open substrate-binding cleft rather than a tunnel like the one found in cellobiohydrolases (119), showing that the tunnel-forming loops have been deleted, which results in an open active site enabling random and internal binding (119).

GH12 endoglucanases (Fig. 3C) form a β-jelly-roll structure with two β-sheets, of six and nine strands, packed on top of one another, and one α-helix. The concave surface forms a substrate-binding groove in which the active site residues, a carboxylic acid trio similar to those of GH7 and GH16 glycoside hydrolases, are located (121, 122). The GH12 endoglucanase from A. niger has also been crystallized and follows a similar three-dimensional architecture, even though the amino acid sequence similarity is somewhat less conserved (123).

Phylogenetic, functional, and substrate specificity analyses of 30 endoglucanases belonging to six GH families (GH5, GH6, GH7, GH9, GH12, and GH45) suggest a structure-function relationship based on active site conformation and the catalytic mechanism (124). Moreover, GH5 endoglucanases are part of a large family that can be grouped into subfamilies (over 31), while all aspergillus endoglucanases fall into two GH5 subfamilies (subfamilies 5 and 7) showing little structural variation (125).

Table 4 shows that compared to cellobiohydrolases, endoglucanases associate least often with CBMs, although ∼30% of all observed endoglucanases are linked to a CBM. On average, the aspergillus genome encodes one or two GH5 endoglucanases with a CBM and one or two without and at least one GH12 endoglucanase with no CBM. GH7 endoglucanases appear to be intermediate, as they are sometimes partially represented or absent, and GH45 endoglucanases are rare. There seems to be some redundancy (same gene model) of endoglucanases with similar structural folds and catalytic activities, especially among families GH5 and GH12. However, whether these apparently redundant gene copies have diversified over time and the encoded proteins acquired novel specialized functions, such as differentiated pH and temperature sensitivity or inhibition and/or activation sites, remains unknown.

TABLE 4.

Genome-wide endoglucanase content in aspergilli^a

^aAmong the aspergilli, there were 18, 9, 9, and 6 cellulases in the GH5, GH7, GH12, and GH45 families, respectively, for a total of 37 cellulases. ±, with or without.

Copper-dependent lytic polysaccharide monooxygenases.

The recently created auxiliary activity family 9 (AA9), formerly included in glycoside hydrolase family 61, is in fact comprised of copper-dependent lytic polysaccharide monooxygenases (LPMOs).

The entire family was originally classified as glycoside hydrolase family 61 because the original protein expressed in yeast had some endoglucanase activity and was regulated similarly to other T. reesei cellulase genes (94). However, GH61 (AA9) enzymes significantly enhance cellulolytic activity on lignocellulosic substrates in combination with other cellulases, due to their action on the cleavage of cellulose chains through an oxidative process of C-1, C-4, and C-6 carbons (126). LPMOs use copper as the catalytic metal, oxygen, and reducing agents for their activity (126). The secretome of Thielavia terrestris contains six AA9 proteins, comprising 10% of the total soluble protein secreted into the medium of cultures grown in the presence of cellulose (72). Thus, the classical hydrolytic mechanism for the degradation of plant polysaccharides was recently challenged by the landmark discovery (15, 75, 127) that oxidoreductase systems, such as fungal LPMOs, directly oxidize cellulose, breaking glycoside bonds and generating aldones and lactones (15, 75, 127–129).

The first AA9 LPMO high-resolution structural model was derived from T. reesei (129, 130). The protein core is a twisted β-sandwich built up of nine β-strands forming a compact single-domain β-sandwich with a large buried ionic network (Fig. 4). A functional metal-ion-binding site is coordinated by three conserved histidines located at the surface near the N terminus, and the requirement for a divalent metal ion for catalytic activity was determined experimentally (74, 75). There is a noteworthy structural similarity between AA9 LPMOs and the chitin-binding protein CBP21 (from Serratia marcescens), a protein that stimulates the chitin-degrading activity of chitinases, with no hydrolytic activity itself (130). Notably, alignment of GH61 endoglucanases with other glucanases did not support the positioning of the two conserved catalytic acidic residues that are present in the catalytic site of all glycoside hydrolases, supplying the acid component that is indispensable for hydrolysis of glycoside bonds (72).

FIG 4.

Lytic polysaccharide monooxygenases (LPMOs). LPMOs, which are classified in the AA9 family (formerly GH61), are bivalent ion-dependent lytic polysaccharide monooxygenases. These proteins cleave cellulose chains with oxidation of various carbons (C-1, C-4, and C-6). The LPMO three-dimensional structures are for Neurospora crassa (PDB entry 4EIR) (129) (A), Hypocrea jecorina (PDB entry 2VTC) (130) (B), Thielavia terrestris (PDB entry 3EJA) (72) (C), and Phanerochaete chrysosporium (PDB entry 4B5Q) (126) (D).

Because of the large number of potential LPMOs (formerly GH61 endoglucanases), we performed a neighbor-joining phylogenetic tree analysis in order to group similar proteins into five distinctive clans. Table 5 shows all AA9 LPMOs of aspergilli grouped into five clans (c1 to c5), based on phylogenetic neighbor-joining relationships (see Table S3 in the supplemental material).

TABLE 5.

Fungal lytic polysaccharide monooxygenases of the AA9 family

Aspergilli, in general, have an average of 8 AA9 LPMOs, whereas A. clavatus has 6, A. niger 7 and A. terreus 12. Aspergilli possess one AA9c1, one or two AA9c2, two or three AA9c3, one AA9c4, and several AA9c5 LPMOs. Most AA9 LPMOs are not linked to CBMs, except for AA9 clan 1, in which every enzyme is linked to a CBM. Thus, AA9 LPMOs are a diverse group of enzymes whose activity remains to be elucidated fully. The interesting aspect of this group of enzymes is the fact that they apparently do not work alone; they interact with other enzymes, such as cellobiose dehydrogenase (CDH) (15, 127), enhancing overall cellulolytic activity.

Cellobiose dehydrogenase.

Cellobiose dehydrogenase (EC 1.1.99.18) is an enzyme that oxidizes cellobiose to cellobiolactone in the presence of an electron acceptor, such as cytochrome c, dichlorophenol-indophenol, or ferricyanide, producing cellobiono-1,5-lactone and a reduced acceptor (131, 132). Cellobiose dehydrogenases are well studied in white and brown rot and plant-pathogenic as well as composting fungi from the dikaryotic phyla Basidiomycota and Ascomycota under cellulolytic culture conditions (133).

CDH is the major oxidoreductase secreted by some fungi (but not aspergilli) growing on biomass that contains cellulose, and it catalyzes the oxidation of cellobiose and longer cellodextrins to 1-5-δ-lactones (134). CDH enhances crystalline cellulose degradation by coupling the oxidation of cellobiose to reductive activation of copper-dependent polysaccharide monooxygenases that catalyze the insertion of oxygen into C—H bonds adjacent to the glycoside linkage (15). Deletion of cdh-1, the gene that encodes the major N. crassa CDH, resulted in reduced cellulase activity, and addition of purified CDH from M. thermophila to the Δcdh-1 strain resulted in a 1.6- to 2.0-fold stimulation of cellulase activity (15). These results suggest that CDH acts as a cofactor for LPMO enzymes. This may (or may not) be true in nature, because redox agents, such as gallate or ascorbate (135), can substitute for CDH.

Lactones hydrolyze spontaneously in solution or are hydrolyzed enzymatically by lactonases to generate aldonic acids (136). Oxidation of cellobiose takes place in the flavin domain following electron transfer to the heme domain, and the reduced heme is able to reduce a wide variety of substrates, including quinones, metal ions, and organic dyes. Reduced cellobiose dehydrogenase can also react with molecular oxygen and interact with the newly discovered copper-dependent lytic polysaccharide monooxygenases that directly oxidize crystalline cellulose (15, 137). The current hypothesis for the function of cellobiose dehydrogenase involves the generation of hydroxyl radicals, formed via reduction of an extracellular ferric complex (138, 139) that takes part in Fenton chemistry, with hydrogen peroxide produced by CDH transferred to an array of acceptor oxidases, such as LPMOs and others that remain unknown (128, 140).

Phylogenetic analysis of all known cellobiose dehydrogenase-encoding genes showed a separation into three classes: class I, found only in Basidiomycota; class II, found in Ascomycota, frequently fused to a CBM (133, 139); and class III, found only in Ascomycota, where they lack a CBM (141).

Cellobiose dehydrogenases are typically monomeric proteins consisting of two domains joined by a protease-sensitive linker region (142). Domains found in cellobiose dehydrogenase, such as heme-binding cytochrome, GMC oxidoreductase, Rossmann-fold NAD(P)(⁺) binding, choline dehydrogenase, and flavoprotein (BetA) domains, are highly conserved among all aspergilli (Table 6).

TABLE 6.

Fungal cellobiose dehydrogenases

Fungus	Cellobiose dehydrogenase(s)	No. of CDH genes per genome
A. clavatus NRRL1	ACLA_076510, ACLA_094490	2
A. flavus NRRL3357	AFLA_001890, AFLA_023820	2
A. fumigatus Af293	AFUA_2G17620, AFUA_2G01180	2
A. nidulans A4	AN7230.2, AN3962	2
A. niger CBS513.88	ANI_1_168174	1
A. oryzae RIB40	AOR_1_98134, AOR_1_712114, AOR_1_2566154	3
A. terreus	ATEG_09993, ATEG_08150	2
H. jecorina		0
N. crassa OR74A	NCU05923, NCU00206	2

In general, aspergilli (Table 6) contain two CDHs, whereas A. niger has one and A. oryzae three. CDHs are not linked to a CBM. To date, all investigated CDHs have been reported to bind specifically to cellulose. For some CDHs, the cellulose-binding ability is attributed to a divergent carbohydrate-binding module (143), whereas other types of CDHs, with no cellulose-binding domain at all, bind to the cellulose surface through the FAD-binding domain (144).

β-Glucosidases.

β-Glucosidases are key enzymes in lignocellulosic hydrolysis because they convert cellobiose and other cellooligosaccharides into glucose (145). β-Glucosidases are classified by CAZy into two families: GH1 and GH3 (69, 146, 147). The structure of a GH1 β-glucosidase from T. reesei (TrBgl2; PDB entry 3AHY) has a classical (α/β)₈-TIM barrel fold. The active site of TrBgl2 consists of a 15- to 20-Å-deep slot-like cleft located on connecting loops at the C-terminal end of the β-sheets of the TIM barrel (148).

The crystal structure for a GH3 β-glucosidase from A. aculeatus has been published (PDB entry 4IIB-H) (149). AaBGL1 is a dimer in solution, and the monomer consists of three domains: a catalytic TIM barrel-like domain, an α/β sandwich domain, and an FnIII (fibronectin type III) domain. Linkers connect these domains.

The genomes of seven aspergilli encode extracellular β-glucosidases belonging to two CAZy families: GH1 and GH3. Table 7 shows that genes for extracellular GH1 β-glucosidases are scarce, averaging 1 gene per genome, while genes for GH3 glucosidases are widely abundant, averaging 7 gene models per genome, with A. flavus harboring 11 and A. fumigatus harboring 5. On average, aspergilli encode 18 β-glucosidases per genome, with 9 being extracellular and 9 intracellular (see Table S5 in the supplemental material). Intracellular β-glucosidases are distributed between the GH1 and GH3 families similarly to extracellular ones, with aspergilli carrying, on average, 2 and 7 gene models, respectively (see Table S5). The β-glucosidase gene model content per Aspergillus genome varies from 10 to 22 (see Table S5), suggesting duplications or sequential acquisition events for similar gene models within this group of fungi.

TABLE 7.

Fungal β-glucosidases

Aspergillus cellulose degradation summary.

Aspergilli contain, on average, 36 cellulase gene models (Table 2), while H. jecorina and N. crassa contain an average of 27, that can be divided into two main functional categories: hydrolytic (27 gene models per average genome) and oxidative (10 gene models per average genome).

Hydrolytic enzymes are abundant and can be grouped further by the way they interact with the substrate, the three-dimensional shape of the catalytic domain, and the type of hydrolysis they promote (retaining or inverting).

Cellobiohydrolases interact with the reducing (average of 3 per genome) or nonreducing (average of 3 per genome) terminus of crystalline cellulose fibers, splitting off cellobiose, occasionally linked to a CBM (average of 1 per genome) or without a CBM (average of 2 per genome).

Endoglucanases interact randomly with amorphous cellulose through an open-cleft catalytic domain that appears in three structural folds: the (α/β)₈ (GH5; averages 2 gene models per genome), jelly roll (GH7 and GH12; average 1 gene model [each] per genome), and 7-fold propeller (GH74; averages 1 gene model per genome) structural configurations. Endoglucanases bind internally to cellulose molecules and hydrolyze glycoside bonds through a retaining (GH5, GH7, and GH12; average 1 gene model [each] per genome) or inverting (GH74; averages 1 gene model per genome) mechanism, thus exponentially multiplying cellulose termini accessible to cellobiohydrolases and β-glucosidases.

Oxidative enzymes include the major cellobiose dehydrogenase (average of 2 gene models per genome), which produces oxygen radicals utilized by lytic polysaccharide monooxygenases (average of 8 gene models per genome) to oxidize glycoside linkages, breaking long crystalline cellulose chains to make them accessible to other hydrolytic enzymes.

Hemicellulases

While cellulose is a simple linear polymer that is recalcitrant to enzymatic digestion, hemicellulose is a heterogeneous mix of polymers, with linear and branched sections composed of various types of sugars and decorated by side groups. Thus, hemicellulases comprise a diverse group of enzymes which catalyze hydrolysis of sugar bonds and side chains, such as ferulic acid and acetyl groups (Fig. 1). While cellulose is invariable in structure among plants, hemicellulose composition varies significantly between plant species. Thus, the makeup of hemicellulose in biomass, e.g., corn stover versus sugar cane bagasse, varies considerably.

Hemicellulases can be grouped into the following functional types: enzymes that hydrolyze backbones, enzymes that remove side chains, and accessory enzymes that remove decorations such as acetyl groups.

Hemicellulose backbone-hydrolyzing enzymes.

Backbone-hydrolyzing hemicellulases comprise all the enzymes needed to degrade the main polymer of xylan, mannan, and arabinan, as well as mixed structures, such as arabinoxylan, glucoronoxylan, xyloglucan, galactomannans, and glucomannans, composed largely of xylose, mannose, and arabinose monomers.

(i) Endo-1,4-β-xylanases.

Xylanases are defined as enzymes that promote hydrolysis of β-1,4-d-xylosidic linkages. Xylanases belonging to the GH10 and GH11 families are retaining enzymes with stereoselective hydrolysis of xylan or β-xylobiosides through a double-displacement mechanism involving a covalent xylobiosyl-enzyme intermediate (150–152). The double-displacement mechanism involves a glycosyl-enzyme intermediate, which is formed and hydrolyzed with general acid/base catalytic assistance, usually involving a Glu (153). Some xylanases promote incomplete hydrolysis of xylan substrates, resulting in xylobiose, xylotriose, and xylotetraose (154, 155).

GH10 xylanases exhibit 8-fold TIM barrel [(β/α)₈] structures containing a deep active site groove, consistent with the endo mode of action (156). GH11 xylanases exhibit β-jelly-roll folding structures with two β-sheets and an α-helix, resembling a partially closed right hand (157).

Table 8 shows that A. niger has a single GH10 but four GH11 xylanases. All other aspergilli have more than one GH10 or GH11 xylanase gene, averaging three genes per genome. Some of the GH10 xylanases are linked to a CBM, i.e., A. clavatus, A. flavus, A. fumigatus, and A. terreus have one gene each, and none of the GH11 xylanases are linked to a CBM. Thus, it seems that xylanases have been acquired multiple times by most fungal genomes, and both structural types (GH10 and GH11) are represented in all analyzed fungi. Table 8 shows that the presence of a CBM is only occasionally found in GH10 xylanases, while the vast majority of GH10 and GH11 xylanases are not linked to a CBM.

Most known xylanases are grouped into the GH10 and GH11 families, although a few bacterial xylanases were recently characterized in detail and ascribed to the GH5 family (158). We did not find GH5 xylanases in aspergilli, although the CAZy database moved some GH5 members into GH30, and we found one characterized GH5 xylanase from T. reesei (159), which had only two gene models that may be homologous in aspergilli. However, the amino acid sequence homology was not clear. Since none of the other aspergilli showed GH5 xylanases, we did not include them in our tables.

(ii) β-Xylosidases (4-β-d-xylan xylohydrolases).

β-Xylosidases hydrolyze β-1,4-d-xylans, oligomers, and xylobiose to remove successive single d-xylose residues from the nonreducing terminus. GH3 xylosidases follow xylanases, with a retaining mechanism of hydrolysis, while GH43 xylosidases are inverting enzymes (160). Talaromyces emersonii (β-XTE) and T. reesei (β-XTR) hydrolyze xylobiose, producing β-d-xylose (160). In Aspergillus awamori, the enzymatic hydrolysis of p-nitrophenyl β-d-xylopyranoside occurs with overall retention of the substrate anomeric configuration, suggesting cleavage of xylosidic bonds through a double-displacement mechanism (161, 162). The catalytic amino acid residues of the extracellular β-d-xylosidase, which hydrolyzes p-nitrophenyl β-d-xyloside as a substrate, involve a carboxylate and a protonated group part for binding of the substrate; however, only a carboxylate group is needed for cleavage (163).

Only GH43 xylosidases have been crystallized, and they display a five-blade β-propeller fold (164, 165). This three-dimensional structural fold also applies to other GH43 members, such as arabinoxylan arabinofuranohydrolases and arabinanases.

In aspergilli (Table 8), most GH3 xylosidases appear multiple times, averaging three copies per genome, while GH43 xylosidases are represented only once or are absent (A. niger).

(iii) Mannanases and β-mannosidase.

GH2 and GH5 β-mannosidases (EC 3.1.25) promote hydrolysis of terminal, nonreducing β-d-mannose residues in β-d-mannosides. GH26 mannan endo-β-1,4-mannosidase (or endo-β-mannanase [EC 3.2.1.78]) promotes random hydrolysis of β-1,4-mannosidic linkages in the main chain of mannans, glucomannans, and galactomannans (166).

The catalytic general acid/base residue is a glutamate, which is separated in sequence by ∼100 residues from the other catalytic nucleophile, another glutamate (167). Immediately preceding the general acid/base residue in sequence is an asparagine that makes interactions with the 2-hydroxyl group of the substrate (167).

GH5 and GH26 β-mannanases have been crystallized, and a common three-dimensional structure has been observed. Like other members of these families and xylanases from the GH10 family, they exhibit the typical 8-fold TIM barrel [(β/α)₈] structure, with the two key active site glutamic acids located at the C-terminal ends of β-strands 4 (acid/base) and 7 (nucleophile) (168–170).

Endo-β-mannanases (GH26) are not common among aspergilli (Table 8): only four species (A. flavus, A. nidulans, A. niger, and A. oryzae) present this type of endo-acting mannanase, while other mannosidases (end-cutting or exo types) are present more frequently and belong to the GH2 and GH5 families (33, 37, 82).

All three types of mannanases have similar three-dimensional structures, suggesting that they function in similar fashions, and the fact that one cuts a mannan chain randomly (endo) and another splits off dimers from a polymeric substrate is more related to the spatial location of the catalytic amino acids within the substrate-binding domain than to the three-dimensional structure itself (168–170).

Side chain-hydrolyzing enzymes.

Carbohydrate side chain-hydrolyzing enzymes are enzymes that hydrolyze sugars linked to the main chain of hemicellulose. Arabinose is the second most abundant sugar in hemicellulose and pectin (171), being found in arabinoxylan and arabinan. Arabinoxylan is constituted by a β-1,4-linked xylopyranose backbone with heterogeneous side chains, such as l-arabinose, O-acetyl, ferulic acid, p-coumaric acid, and 4-O-methylglucuronic acid (172). Arabinan is formed by an α-1,5-linked arabinose backbone with branching by α-1,2- or α-1,3-linked arabinofuranose side chains (173). Enzymes known to hydrolyze side chains are arabinofuranosidases, arabinases, and glucuronidases.

(i) Arabinofuranosidases.

Arabinofuranosidases (EC 3.2.1.55) hydrolyze α-1,2-, α-1,3-, and α-1,5-l-arabinofuranosidic bonds in l-arabinose-containing hemicelluloses, such as arabinoxylan and l-arabinan (174, 175).

We found three families of arabinofuranosidases in fungi: GH51, GH54, and GH62 (Table 9). Some operate with wide substrate specificity, acting on arabinofuranosides at O-5, O-2, and/or O-3 as a single substituent (171). GH51 arabinofuranosidases hydrolyze small substrates only (176), GH54 enzymes hydrolyze polymeric substrates, such as arabinoxylans, in addition to the small substrates, and GH62 enzymes act only on arabinoxylans (151, 176).

TABLE 9.

Fungal accessory enzymes

^a±, with or without.

The three-dimensional structures of the catalytic domains of fungal GH51 and GH62 arabinofuranosidases are not yet known. However, the bacterial enzymes show a classical (β/α)₈ barrel structure, and there is also evidence that arabinofuranosidases act as hexamers (177, 178). The three-dimensional structure of GH54 arabinofuranosidase establishes a β-sandwich (176).

Table 9 shows that genes encoding arabinofuranosidases are abundant in aspergilli, averaging three GH51, one GH54, and two GH62 gene models per genome. None of the examined arabinofuranosidases are linked to a CBM.

(ii) Arabinanases.

Most monosaccharides are present in their d-form. l-Arabinose is the exception and is found in its furanose form as a constituent of hemicellulose and pectin (171). Arabinanases and arabinofuranosidases catalyze the hydrolysis of α-1,5-arabinofuranosidic bonds in arabinose-containing polysaccharides (151, 179). GH43 and GH93 arabinanases share extensive amino acid sequence similarities with β-fructosidases (GH32 and GH68), with multiple homologous domains (180).

Arabinanases act by depolymerizing arabinopolysaccharides, producing arabinose or arabino oligomers, depending on the preference of the enzyme for substrate termini (exo type) (181) or random cleavage (endo type) (182).

Three-dimensional structures for fungal GH43 arabinanases are not yet available, and the closest crystallized arabinanase is from Cellvibrio japonicus (E value = 2E−35) and displays the classical five-blade β-propeller fold (181, 183). A V-shaped groove that is partially enclosed at one end forms a single extended substrate-binding surface across the face of the propeller (183), and three carboxylates deep in the active site provide the general acid and base components for glycosidic bond hydrolysis, with inversion of the anomeric configuration (183).

GH93 arabinanases recognize linear α-1,5-l-arabinan as the substrate and release α-1,5-l-arabinobiose from the nonreducing end of the polysaccharide (81). Two fungal GH93 arabinanases (81, 184) show a six-blade β-propeller fold with a typical “Velcro ring” closure. The substrate-binding groove is enclosed at one end by two residues, Glu and Tyr, which contribute to the recognition of the nonreducing chain end of the polysaccharide (184).

Table 9 shows that genes encoding arabinanases are also abundant, with the genomes averaging five GH43 and one GH93 arabinanase gene. None of the arabinanases are linked to CBMs.

(iii) β-Glucuronidases and xylan α-1,2-glucuronosidase.

Glucuronidases cleave the glucuronic acid of glucuronoglycans found in plant cell wall structures. Two types of glucuronoglycans are common: glucuronans (alginic acid), constituted exclusively of glucuronic acid residues, and glucuronoglycans (pectins, gums, and mucilages), whose main chain is composed of glucuronic acid and other sugars as side chains (185). Glucuronans are often acetylated, which interferes with glucuronan-hydrolyzing enzymes (186).

Aspergillus glucuronidases are found in two CAZy families, GH2 and GH79, and none of them have been characterized. Table 9 shows that β-glucuronidases are much less abundant than arabinofuranosidases and arabinanases, averaging one or no GH2 or GH79 gene model. Xylan α-1,2-glucuronidases are found in two CAZy families, GH67 and GH115, and none have been characterized. Table 9 shows that xylan α-1,2-glucuronosidases are represented at averages of 1 and 2 genes for GH67 and GH115 enzymes, respectively.

(iv) Xyloglucanases.

Xyloglucan consists of a cellulose-like backbone chain of β-1,4-glucan with xylosyl side chains attached to the O-6 position of glycosyl residues and associates with cellulose microfibrils through hydrogen bonds, forming cellulose-xyloglucan (187). Enzymes responsible for the hydrolysis of the xyloglucan backbone are xyloglucan endo-β-1,4-glucanases or “xyloglucanases” (EC 3.2.1.151). EC 3.2.1.151 reflects many different enzyme sequences, structures, and hydrolytic mechanisms, with either inversion or retention of the anomeric carbon. In the sequence-based CAZy classification, enzymes defined as xyloglucanases are found in retaining families GH5, GH12, and GH16 and inverting families GH44 and GH74 (188). To date, aspergillus GH12 and GH74 xyloglucanases are the best described (189).

While some endoglucanases are promiscuous and can hydrolyze both unbranched and branched β-1,4-glucan chains (189), xyloglucan-specific endoglucanases (XEGs) constitute a relatively new class of enzymes (190) and were described as EC 3.2.1.151 and EC 3.2.1.150 enzymes (191).

In EC 3.2.1.151 XEGs, the reaction involves endo hydrolysis of β-1,4-d-glycosidic linkages in xyloglucan, with retention of the beta-configuration of the glycosyl residues (192–194). In EC 3.2.1.150 XEGs (oligoxyloglucan reducing-end-specific cellobiohydrolases [OREX]), the reaction involves the hydrolysis of cellobiose from the reducing end of xyloglucans consisting of a β-1,4-linked glucan carrying alpha-d-xylosyl groups on O-6 of the glucose residues (195).

OREX belong to the GH74 family. Two GH74 enzymes have been shown to be reducing-end-specific cellobiohydrolases (EC 3.2.1.150) (195, 196) releasing XG, LG, or FG from xyloglucan [X, G, L, and F designate α-d-Xylp-(1→6)-β-d-Glcp, d-Glcp, β-d-Galp-(1→2)-α-d-Xylp-(1→6)-β-d-Glcp, and α-l-Fucp-(1→2)-β-d-Galp-(1→2)-α-d-Xylp-(1→6)-β-d-Glcp, respectively] (197). The exo-acting activity is believed to be due to a loop insertion closing off the positive subsites (198).

The OREX X-ray crystal structure shows two seven-blade β-propeller domains forming a large cleft and a loop where the substrate binds (Fig. 5). The substrate-cleaving region is located near the loop region, believed to be the region of hydrolysis. Deletion of the loop region resulted in endo random cleavages of the substrate, suggesting that the loop region directs the exo activity (198–200). Moreover, an endo-processive xyloglucanase (XEG74) that contains four unique tryptophan residues, in the negative subsites (W61 and W64) and the positive subsites (W318 and W319), was isolated from Paenibacillus. The positive subsites (W318 and W319) were essential for processive degradation and were responsible for maintaining binding interactions with xyloglucan (201).

FIG 5.

GH74 xyloglucanobiohydrolases. (A) The three-dimensional structure consists of two tandem repeats of a seven-blade β-propeller domain which forms a large cleft and a loop where the substrate is bound. (B and C) Two views of the open cleft. The three-dimensional structure of Geotrichum sp. GH74 endoglucanase is from PDB entry 1SQJ (199, 200).

Xyloglucanases are not abundant in fungi. According to CAZy and mycoCLAP (202), to date, there are no GH5, GH16, or GH44 XEGs characterized from fungi. Thus, we considered only GH12 and GH74 XEGs from aspergilli, due to the lack of characterized gene models for the GH5, GH16, and GH44 families. Furthermore, XEGs from these families were not identified in secretomes of aspergilli growing on biomass (10, 14, 203, 204). Table 9 shows that aspergilli contain one or no GH74 xyloglucanases but several GH12 enzymes. A. flavus and A. oryzae have no genes encoding GH74 xyloglucanases, and A. nidulans has two. Other aspergilli have one GH74 xyloglucanase. Unlike A. flavus, all other aspergilli contain one GH12 xyloglucanase. GH12 xyloglucanases are not linked to a CBM (see Table S7 in the supplemental material).

Accessory enzymes.

(i) Acetyl xylan esterases.

Acetyl xylan esterases promote hydrolysis of acetyl groups from xylan β-1,4-linked d-xylopyranoside backbones with heterologous side chains (O-acetyl, ferulic acid, p-coumaric acid, arabinose, and 4-O-methylglucuronic acid groups), mainly present in plant cell walls of hardwoods.

Few acetyl xylan esterases have been studied so far, with one from A. niger belonging to carbohydrate esterase family 1 (CE1) and two CE5 enzymes, from T. reesei and Penicillium purpurogenum, being described (205–207). A. awamori CE1 acetyl xylan esterase (accession no. XP_001395572) releases acetic acid from birchwood and is active toward α-naphthylacetate (C₂) and α-naphthylpropionate (C₃) but not toward acyl substrates containing four or more carbons (206), and site-directed mutagenesis that abolished deacetylation indicated that Ser₁₁₉ and Asp₂₀₂ are key. T. reesei CE5 acetyl xylan esterase (AXE1_TRIRE) catalyzes acetylated xylo-oligomers and releases acetic acid from birchwood xylan (207). A novel acetyl xylan esterase (Aes1) from T. reesei, with no CAZy CE classification, was studied using 2-, 3-, and 4-O-acetyl 4-nitrophenyl β-d-xylopyranosides as substrates, and Aes1 hydrolyzed the three substrates with an initial rate ratio of 1:19:17.7, suggesting that Aes1 prefers positions 3 and 4, while the other acetyl xylan esterases prefer position 2 acetyl groups (208–210).

Carbohydrate esterases are classified into 16 CAZy CE families, whereas fungal acetyl xylan esterases appear in 8 of them (211). Most carbohydrate esterases are serine-type esterases also acting on low-molecular-weight substrates, such as 4-nitrophenyl or 4-methylumbelliferyl acetate. An exception is the acetyl xylan esterases from family CE4, which do not operate on aryl acetates (212, 213).

The 3-D structure of acetyl xylan esterase consists of a three-layer α/β/α sandwich fold (214). The central β-sheet consists of six parallel β-strands delimited by four α-helices (two on each side). In addition, acetyl xylan esterase docks an additional two α-helices and four 3₁₀-helices (214). In acetyl xylan esterases, cysteine forms disulfide bridges, and the active site is located in a cleft near the C-terminal end of the third β-strand, where there is a Ser-His-Asp triad (214).

Table 10 shows that the better-understood acetyl xylan esterases belong to the CE1 family; however, aspergilli also have acetyl xylan esterases belonging to the CE3, CE4, CE5, and CE16 families. Most aspergilli harbor genes belonging to at least three or four families in their genomes, with the exception of A. oryzae, which has only two CE families represented. Acetyl xylan esterases belonging to families CE1 and CE5 are sometimes linked to a CBM (Table 10).

(ii) Ferulic acid esterases.

Ferulic acid (4-hydroxy-3-methoxycinnamic acid), a hydroxycinnamic acid, is the main aromatic acid building block of lignocellulosic materials (215, 216). Ferulic acid is generally not found free but instead is esterified to arabinose in various polysaccharides, e.g., arabinoxylans and pectins (217).

Feruloyl esterase, also known as cinnamoyl esterase, is an enzyme that hydrolyzes ferulate ester groups from hemicellulose molecules, which represent cross-links among hemicellulose chains and lignin polymers (218). Feruloyl esterases belong to the CE1 family and hydrolyze the ester bond between hydroxycinnamic acid and sugars present in plant cell walls (219).

The three-dimensional structure of A. niger feruloyl esterase (PDB entries 1UWC and 1USW) displays an α/β hydrolase fold consisting of a nine-stranded β-sheet core surrounded by α-helices and two additional β-strands (220). The catalytic triad (Ser133-His247-Asp194) forms the active site of this enzyme. The active site cavity is confined by a lid, analogous to the case in lipases, and by a loop that confers plasticity to the substrate-binding site (221). Feruloyl esterases were initially classified by aromatic functional categories (222, 223), and later the classification was extended to subfamilies A, B, C, and D, based on primary amino acid sequence similarity and substrate specificity against four model substrates, i.e., methyl 3-methoxy-4-hydroxycinnamate (MFA), methyl 3,4-dihydroxycinnamate (MCA), methyl 4-hydroxycinnamate (MpCA), and methyl 3,5-dimethoxy-4-hydroxycinnamate (MSA) (224, 225).

Table 10 shows that feruloyl esterases are represented in variable copy numbers in all aspergilli, in some cases showing much more than 2 copies (3 copies in A. clavatus, 8 in A. flavus and A. terreus, 10 in A. oryzae, and 11 copies in A. niger). None of the feruloyl esterases are linked to CBMs.

Aspergillus hemicellulase degradation summary.

Aspergilli contain an average of 30 hemicellulase gene models (Tables 8 to 10), while H. jecorina and N. crassa contain an average of 17 each, distributed among the following seven distinct functional categories: the backbone-attacking enzymes xylanase, mannosidase, arabinase, and xyloglucanase and the short-side-chain-removing enzymes xylan α-1,2-glucuronidase, arabinofuranosidase, and xylosidase. In addition, aspergilli contain, on average, seven accessory gene models involved in the liberation of acetyl groups from acetylated polysaccharides and in cleavage of the ester links between monomeric or dimeric ferulic acid and the polysaccharide main chain of xylans.

CONCLUSIONS

Industrial enzymology is a billion-dollar market that capitalizes on the inherent catalytic specificity, speed, and robustness of natural enzymes to carry out complex reactions in a clean and environmentally friendly manner. Holocellulose-degrading enzyme applications are well established within the paper, food, and feedstock industries. Cellulase and hemicellulase cocktails are essential components in any kind of biorefinery core that relies on biomass materials as the input source.

In the present study, based on the genome-wide content of seven aspergilli (a natural biomass biorefinery), we unraveled hundreds of gene models encoding holocellulose-degrading enzymes, suggesting the occurrence of dozens of apparent duplications, but after a systematic organization, these fell into smaller coherent functional groups according to domain organization, biochemical activity, and genome distribution.

Current commercial enzyme cocktails are catalytically incomplete, demanding high enzyme loads and long residence times, and are subject to contamination. Most commercially available cocktails lack one or more enzyme activities that we found in the genome-wide survey presented here.

The present study provides a valuable reference on carbohydrate degradation by aspergilli, giving a significant extension of previous studies (33, 34, 37). The mapping of holocellulose enzyme breakdown systems in filamentous fungi and their relationships with substrates provides the foundation for genome-wide expression analysis studies, as well as a powerful framework for further functional and structural works on fungal carbohydrate-active enzymes.

Biographies

Fernando Segato is an Assistant Professor in the Department of Biotechnology at the Engineering School of Lorena at the University of São Paulo, Brazil. He received his Ph.D. in Molecular Biology and Genetics of Microorganisms from the Ribeirão Preto School of Medicine at the University of São Paulo. His research focuses on prospection of hydrolytic and oxidative enzymes applied in lignocellulose depolymerization, exploration of thermophilic fungal genomes, and improvement of fungal cell factories.

André R. L. Damásio is a Young Research Scientist and works at the Brazilian Bioethanol Science and Technology Laboratory (CTBE) at the Brazilian Center for Research in Energy and Materials (CNPEM) in Campinas, Brazil. He received his Ph.D. in Biochemistry from the Ribeirão Preto School of Medicine at the University of São Paulo. His research focuses mainly on the improvement of Aspergillus strains for secretion of client proteins and the function of glycosylation in the folding and secretion of plant cell wall-degrading enzymes.

Rosymar C. de Lucas is a Postdoctoral Fellow at Novozymes Brazil. She received her Ph.D. in Biochemistry from the Ribeirão Preto School of Medicine at the University of São Paulo. Currently, she is studying degradation of biomass cell walls by enzymes from microorganisms to improve second-generation (2G) ethanol.

Fabio M. Squina is a Senior Researcher at the Brazilian Bioethanol Science and Technology Laboratory (CTBE) at the Brazilian Center for Research in Energy and Materials (CNPEM) in Campinas, Brazil. He received his Ph.D. in Molecular Cell Biology from Ribeirão Preto School of Medicine at the University of São Paulo. His research interests combine molecular biology, “-omics” sciences, and high-throughput screening approaches, aiming to engineer enzymes and biotechnological routes for plant biomass conversion into bioproducts.

Rolf A. Prade is a Microbiology Professor in the Department of Microbiology and Molecular Genetics at Oklahoma State University in Stillwater, OK. He received his Ph.D. in Biochemistry from the Ribeirão Preto School of Medicine at the University of São Paulo. His research focuses on bioinformatics and molecular genetics of fungi, including carbon regulation and enzymatic pathways that metabolize plant cell wall polymers.