Revealing Nature’s Cellulase Diversity: The Digestion Mechanism of Caldicellulosiruptor bescii CelA

See allHide authors and affiliations

Science  20 Dec 2013:
Vol. 342, Issue 6165, pp. 1513-1516
DOI: 10.1126/science.1244273

An Enzyme Drill

Cellulase enzymes degrade the cell walls of plants by breaking down cellulose into its constituent sugar fragments and thus have attracted interest for biofuels production. Using transmission electron microscopy Brunecky et al. (p. 1513; see the Perspective by Berlin) discovered that an especially active cellulase, CelA, from Caldicellulosiruptor bescii bacteria does not move along the surface of the substrate, but drills into the cellulose to form cavities.


Most fungi and bacteria degrade plant cell walls by secreting free, complementary enzymes that hydrolyze cellulose; however, some bacteria use large enzymatic assemblies called cellulosomes, which recruit complementary enzymes to protein scaffolds. The thermophilic bacterium Caldicellulosiruptor bescii uses an intermediate strategy, secreting many free cellulases that contain multiple catalytic domains. One of these, CelA, comprises a glycoside hydrolase family 9 and a family 48 catalytic domain, as well as three type III cellulose-binding modules. In the saccharification of a common cellulose standard, Avicel, CelA outperforms mixtures of commercially relevant exo‐ and endoglucanases. From transmission electron microscopy studies of cellulose after incubation with CelA, we report morphological features that suggest that CelA not only exploits the common surface ablation mechanism driven by general cellulase processivity, but also excavates extensive cavities into the surface of the substrate. These results suggest that nature’s repertoire of cellulose digestion paradigms remain only partially discovered and understood.

In nature, most cellulolytic enzymes systems are of two general types: those with noncomplexed cellulases and hemicellulases produced by aerobic fungi and most bacteria (1), and those in which polysaccharidases self-assemble onto a common protein scaffold to form large macromolecular assemblies called cellulosomes (2, 3). Only a few anaerobic bacteria and fungi are known to produce cellulosomes. In both cases, the enzymes secreted are primarily equipped with a single catalytic domain. An alternate enzymatic system, in a sense midway between that of the two previous models, is used by the thermophile Caldicellulosiruptor bescii (previously Anaerocellum thermophilum) (46), and some other bacteria in which the individual cellulases secreted are multimodular, containing multiple binding and catalytic domains. The most abundant enzymes secreted by C. bescii are not only multimodular, but possess catalytic domains with different activities (multifunctional). Lochner and co-workers conducted an extensive characterization of the C. bescii secretome and concluded that CelA is the dominant cellulase (5). CelA is a complex, thermally stable enzyme containing an N‐terminal family 9 glycoside hydrolase (GH9) endo‐β‐1,4‐glucanase domain, three family 3 carbohydrate-binding modules (CBM3), and a C‐terminal GH48 exo‐β‐1,4‐glucanase domain. family 9 and 48 catalytic domains have been considered to be highly synergistic (7, 8). CelA was first isolated and partially characterized in terms of its activity by Zverlov and co-workers (9), who reported that CelA demonstrated activity on cellulose, as well as weak activity on xylan. However, they did not report the performance of CelA on biomass and did not propose its mode of action or the complementarity of the two catalytic domains.

C. bescii can operate at temperatures up to 90°C (5, 6), which may explain the advantage of several CBMs in counteracting the loss of binding due to increased temperature. Although C. bescii has been well studied at the microbial level, the structural features of its highly thermally stable enzymes have not been explained owing to the lack of structural information for enzymes from the Caldicellulosiruptor clade. Currently, large‐scale biomass saccharification in bioconversion processes relies exclusively on fungal enzymes operating at 50° to 55°C. C. bescii CelA may have several advantages over other fungal and bacterial cellulases for use in biofuels production, namely, its high specific activity and stability at elevated temperatures.

Several recent studies report that the optimal growth temperature for C. bescii is near 80°C (5, 6). We examined the cellulolytic performance of purified CelA [supplementary material (SM) text S1 and S2] acting on Avicel, a model cellulose substrate that is generally used to evaluate cellulase action, at 60°, 75°, and 85°C. We compared CelA to a binary mixture containing Trichoderma reesei Cel7A exoglucanase, currently the most common exoglucanase found in commercial cellulase preparations (10, 11), and Acidothermus cellulolyticus Cel5A endoglucanase (11), which mimics the two cellulolytic (endo‐ and exoglucanase) activities found in CelA. The percentage of glucan converted over a 7-day digestion study is shown in Fig. 1A. CelA retained high activity compared to Cel7A at all temperatures tested, converting 60% of glucan at 85°C compared to 28% glucan conversion by Cel7A at its optimal temperature of 50°C (Fig. 1A). The extent of conversion obtained in this study for Cel7A is consistent with those reported by several other research groups (1214). Furthermore, the activity of this enzyme acting on Avicel, on a molar basis, is seven times as high as that of the common exo- and endocellulase standard mixture, Cel7A and Cel5A (SM text S3). This high activity may be due to the proximity of a chain‐end–forming endoglucanase and an efficient cellobiohydrolase in the same molecule, thereby increasing the intramolecular synergy. These results illustrate that when acting on Avicel, CelA is a far more active single enzyme than the dominant enzyme in today’s commercial cellulase formulations, T. reesei Cel7A. End-product inhibition of cellulases is a well-known characteristic (15), and CelA in particular is known to produce both cellobiose and glucose. To assess this issue, we tested the sensitivity of CelA to cellobiose by adding low loadings of a thermally stable β‐d‐glucosidase from Thermotoga maritima (16) to the CelA mix. The addition of β‐d‐glucosidase markedly improved the performance of CelA acting on Avicel, producing 100% conversion in 7 days (Fig. 1A).

Fig. 1 Activity of CelA.

(A) Avicel conversion by CelA at different temperatures (15 mg of CelA/g glucan or 14 mg of CelA + 1 mg β‐glucosidase/g glucan) compared to Cel7A/E1 (12 mg/g + 3 mg/g glucan, respectively). (B) CelA conversion of switchgrass at 15 mg/g glucan. (C) Corn stover conversion by CelA at 20 mg/g (AP) or 15mg/g (DA) glucan and improvement of CelA by addition of β‐d‐glucosidase on DA corn stover (14 mg CelA + 1 mg β‐glucosidase/g glucan). (D) CelA conversion of xylan from native switchgrass at 15 mg/g glucan. Error bars represent standard deviations from three different experiments.

The performance of CelA acting on biomass is predictably lower than when acting on Avicel; CelA performs best on alkaline peroxide (AP)– (17, 18) pretreated corn stover (~50% conversion) (Fig. 1C). When tested on other substrates, such as dilute acid–pretreated switchgrass (DA) (19), native switchgrass (SG), and ammonia fiber expansion–treated switchgrass (AFEX) (20, 21) (Fig. 1B), CelA performed better on the untreated substrate even though the entire rationale of pretreatment is to improve enzymatic hydrolysis of biomass substrates. CelA activity was lowest when acting on AFEX-pretreated switchgrass, where the glucan conversion was 20%. Also, the activity of purified CelA on biomass was similar to that of the C. bescii purified secretome (ExtP) fraction (fig. S6). The importance of removing end-product inhibition when CelA acts on DA-pretreated corn stover is shown by the results in Fig. 1C, where CelA alone converts only ~30% of the glucan content to sugars and the addition of β‐d‐glucosidase increases this conversion by almost 75%: a notable improvement, given the relatively low overall enzyme loading.

While performing these digestion experiments, we also observed that CelA could achieve 60% conversion of xylan in native switchgrass, which showcases its potential for use in an industrial process using mild or no pretreatment (Fig. 1D). We tested both catalytic modules from CelA and found that the GH48 is largely responsible for this xylan-degrading activity (SM text S4). To further examine the xylan-degrading ability of CelA, we crystallized the two catalytic domains (GH9 and GH48) of CelA (SM text S5). It is probable that the additional ability to degrade xylans as well as glucans is defined by small changes in the conformational properties of the active sites of glycoside hydrolases. However, closer examination of the CelA GH9 and GH48 structures shows no obvious features that would prevent or favor xylan binding over cellodextrins.

Despite these compelling properties exhibited by CelA, several limitations may also be attributed to its multimodular architecture. Although the levels of glucan conversion achieved by CelA acting on highly crystalline cellulose (Avicel) are much higher than that of T. reesei Cel7A, the overall performance of CelA acting on commercially relevant feedstocks, such as switchgrass and corn stover, are lower, even with the use of a β‐d-glucosidase (Fig. 1C). There are several important differences between biomass feedstocks and Avicel. Biomass is structurally and chemically more complex than Avicel. Smaller enzymes, such as Cel7A, may be better able to penetrate into the plant cell walls, even after pretreatment, whereas the larger CelA, with its multiple CBMs, may be too large and be more prone to nonproductive binding. Some of these characteristics may be enhanced by pretreatments that markedly change the structural integrity of the biomass, perhaps reflected by our observation that CelA works best with the AP-pretreated biomass, a pretreatment that substantially removes lignin (22, 23) (Fig. 1C).

Chemical species other than glucan present in biomass, such as lignin and hemicelluloses, may trap CelA in nonproductive binding states. Naturally, all cellulases will experience some level of nonproductive binding on biomass. However, owing to the larger molecular weight of CelA and because enzyme loadings are done on an enzyme weight basis, the molar enzyme loading of CelA is much lower than it is for Cel7A. Therefore, the activity of a CelA would be affected to a greater extent by nonproductive adsorption than would an equivalent mixture of cellulases with lower molecular weights. This might explain the results on AFEX-pretreated biomass, where lignin‐carbohydrate complexes (LCCs) are redeposited on plant cell walls after such pretreatment (22, 23). LCCs represent a barrier for all enzymes, but the problem seems amplified for a complex enzyme such as CelA.

Further comparison of CelA to its fungal counterparts reveals another fascinating aspect of this cellulase. Transmission electron microscopy (TEM) imaging results indicate that CelA has a mode of action distinct from that ascribed to fungal enzymes, such as those used in the commercial formulation CTec2 (Novozymes).

We examined changes in the surface and internal structure of Avicel PH101 particles treated with CelA. These particles were recovered from digestions carried out to ~65% cellulose conversion and then compared to Avicel particles digested to ~60% conversion using CTec2 (CTec2 is composed primarily of Cel7A). Our analysis focused on the most electron-translucent particles within each sample, where individual cellulose microfibrils could often be identified within the particles (Fig. 2). Particles digested by the Cel7A-containing formulation displayed morphology previously reported where one end of the particle was finely tapered to a narrow point (Fig. 2, A and B) and an opposite end displayed a blunt edge with a slight angle from the long axis (Fig. 2, A’ and B’) (24). CelA-digested particles, by contrast, displayed narrowed, irregular, but not finely tapered morphology on one end (Fig. 2, C and D) and an irregular, scalloped, angled morphology on the opposite end (Fig. 2, C’ and D’). In addition to a surface ablation activity typical of fungal enzymes that seem to work only on the surface of the substrate (Fig. 2, A and B), CelA appears to excavate down into the layers of the substrate, generating cavities (Fig. 2, E to H, and fig. S8). TEM evidence for this excavation mechanism was notable: CelA-digested Avicel particles displayed surfaces marked by irregularly spaced cavities distributed along the length of the particles (Fig. 2, E and F). The cavities vary widely in cross-sectional area at the surface between 25 and ~1000 nm2. Most cavities are less than 500 nm2 in cross-sectional area and have widths of 15 to 20 nm and lengths of 15 to 30 nm (Fig. 3). No cavities were observed on the surface of Avicel particles digested by CTec2. Additionally, on the basis of the predicted conformations that CelA can attain, as inferred by molecular dynamics simulations, it is clear that CelA fits into these cavities (Fig. 3). From these simulations, the effective size of CelA is estimated to be between 10 and 35 nm. These spatial dimensions for CelA correlate very well with the smaller cavity dimensions (diameter) in the range of 15 to 30 nm (Fig. 3). The schematic in Fig. 4 summarizes the differences in the digestion mechanism of CelA and Cel7A.

Fig. 2 TEM micrographs of partially digested small Avicel PH101 particles.

Those digested to ~60% conversion by a CTec2 cellulase mixture (primarily comprising T. reesei Cel7A) display tip sharpening (A and B) on one end and blunted morphology on the other (A’ and B’) Particles digested to ~65% conversion with CelA display slightly narrowed, tapered, blunt ends (C and D) and irregular blunt or angled ends (C’ and D’), as well as cavities of various sizes on the surface [arrows, (E) and (F)]. Occasionally, some of the material being removed to form the cavities appears to remain attached to the cavity edge (G and H). All scale bars are 500 nm.

Fig. 3 Evidence that CelA fits in the cavities.

(A and B) The histograms show dimensions of the cavities created by CelA (black), extracted from the TEM micrographs in Fig. 2, E and F, and the calculated dimensions of CelA itself (gray) over the course of a 40-ns molecular dynamics simulation, with the smallest dimension in (A) and largest dimension in (B).

Fig. 4 Schematic representation of digested cellulose microfibril bundles.

The diagrams contrast the surface ablation and reducing-end oriented mechanism of Cel7A (left) with the surface ablation and cavity-forming mechanism of CelA (right). This representation suggests how these distinct deconstruction mechanisms could be synergistic by specializing in different aspects of the nanoscale architecture of biomass materials and exposing new surfaces.

To better understand the digestion mechanism of CelA, we constructed a kinetic model of enzymatic digestion of cellulose to test the hypothesis that CelA has a specific mode of action that leads to cavity formation, a mode not found in simpler cellulase enzymes (SM text S6). In the model, CBMs are designed to bind and unbind from cellulose; catalytic domains have binding, engaging, and processive digestion functionality, with enhanced binding when a linker-attached module is bound. The model reproduces the surface ablation mode of digestion found in the commercial formulations when applied to enzyme with only a single CBM linked to a single catalytic domain (fig. S13C). The same digestion, dominated by single-layer ablation, is observed if identical functionality is used with the complex CelA-type model (fig. S10A). The model shows that the cavity formation occurs (fig. S10B) when the additional constraint is put into the complex model so that, once fully bound, the processive digestion is slowed owing to competing digestion directions and increased binding affinity, which inhibits repeated unbinding and rebinding in other locations.

On the basis of these results, it appears that CelA and multifunctional cellulases represent a distinct paradigm for cellulose digestion and high activity created by combining complementary modules separated by long linker peptides on the same gene product. We propose that this hydrolysis mechanism could synergize with other cellulases representing the previously known paradigms—free enzymes and cellulosomes—because they offer distinctive mechanisms. For example, in nature, enzymes from mesophilic saprophytic fungi are unable to mix with enzymes secreted into high-temperature environments. We thus propose that considerable synergism can be afforded if it is possible to bring together enzymes from these natural cellulolytic paradigms.

Supplementary Materials

Materials and Methods

Figs. S1 to S17

Supplementary Text S1 to S6

Tables S1 to S8

References (2535)

References and Notes

  1. Acknowledgments: This work was supported by the BioEnergy Science Center (BESC). BESC is a U.S. Department of Energy (DOE) Bioenergy Research Center supported by the Office of Biological and Environmental Research in the U.S. DOE Office of Science. We acknowledge colleagues at the Biomass Conversion Research Laboratory at Michigan State University for providing the AFEX-pretreated materials. Structures have been deposited in the Protein Data Bank with PDB codes 4DOD (GH9), 4DOE (GH9-CB), and 4EL8 (GH48).
View Abstract

Navigate This Article