In this study, we integrated heterologous EG and CBHI genes into the genome of a BGL-displaying S. cerevisiae, permitting the production of EG and CBHI via cell-surface display, secretion, or a combined strategy. The recombinant strains that produced EG and CBHI in the same space (on the cell surface or in the medium) showed superior performance in the production of cellulosic ethanol. To our knowledge, this is the first report on combining cell-surface display and secretion strategies in a single yeast strain for heterologous cellulase production.

The benefits of attaching BGL to the yeast cell wall have been reported previously [15]. However, suitable strategies for EG and CBH production by cells remained obscure. In the present study, direct ethanol production from β-glucan and PASC was performed using cellulolytic yeast strains. Specifically, the production of ethanol was compared to investigate the suitable strategy for producing various kinds of cellulases. When β-glucan was used as the fermentation substrate, the ethanol production rate of strain EG-D was apparently higher than that of EG-S. Notably, strain EG-D was able to convert β-glucan into ethanol immediately after the start of fermentation, indicating that the displayed BGL and EG were successfully transferred into the fermentation medium with the cell inoculum. In contrast, strain EG-S converted β-glucan into ethanol after a 9-h lag, consistent with the need for this strain to accumulate (via secretion) EG in the medium following inoculation. This phenomenon demonstrated the advantage of an EG-display system to the cellulose fermentation process, since display permitted early onset of the production of ethanol. Compared with β-glucan, PASC contains a higher degree of polymerization and crystallization, rendering this substrate more difficult to degrade by BGL and EG. Indeed, fermentation with PASC yielded apparently lower rates of ethanol production than those seen upon fermentation with β-glucan. The decline of ethanol observed in PASC fermentation is probably due to the consumption via yeast cells when cellulases could not hydrolyze sufficient glucose from PASC. Notably, the rate of ethanol production from PASC by EG-D appeared higher than that by EG-S.

Subsequently, the CBHI gene was successfully expressed along with BGL and EG genes in S. cerevisiae. The T. emersonii CBHI used in this study was fused with the CBM of T. reesei CBH1 [21]; use of the CBM has been shown to enhance the adsorption of enzyme to its substrate and modify substrate surfaces to facilitate enzymatic hydrolysis [26]. We observed that co-production of CBHI with displayed EG enhanced ethanol production by up to twofold, demonstrating that the heterologous CBHI that was present assisted in the degradation of PASC. In the display strategy, the CBHI gene was expressed via the SED1 expression cassette and anchored on the cell surface of S. cerevisiae using the SED1 anchoring domain [22]. It has been reported that the expression level of the SED1 gene was highly induced in the stationary phase by various environmental stresses, such as ethanol [27]. In the present work, expression of both CBHI and EG gene by cell-surface expression cassettes (in strain EG-D-CBHI-D) permitted a doubling of CBHI activity at 96 vs. 0 h of fermentation, confirming the utility of the stress-induced SED1 expression cassette. However, we note that the SED1 anchoring domain was fused to the C-terminus of the CBHI-CBM chimera protein; N- and C-terminal of CBM were fused with CBHI and SED1 anchoring domain, respectively, which may hinder the function of the CBM. Alternatively, an N-terminal anchoring domain, such as the N terminus flocculation functional domain of Flo1p, may be more suitable for displaying the chimeric CBHI containing a C-terminal CBM [7]. Nonetheless, to our knowledge, the present work represents the first report on displaying a CBHI on the cell surface of yeast strain.

In a previous report, a yeast strain displaying BGL, EG, and cellobiohydrolase II (CBHII) on the cell surface yielded higher ethanol production than a strain secreting the corresponding enzymes in free form [28]. Immobilized BGL on the cell wall is considered more appropriate for cellulosic ethanol production compared to free BGL [15]; our use of cell-surface-displayed BGL is presumably one of the reasons for the elevated ethanol yields in the present study. Ethanol production by strain EG-S-CBHI-S was similar to that of strain EG-D-CBHI-D. To understand this interesting result, the mechanism of the enzymatic hydrolysis of cellulose should be taken into account. EG can randomly cleave the amorphous regions of cellulose to produce oligosaccharides and provide free chain ends for CBH activity. Then CBHI can initialize cleavage from the free chain ends, degrading crystalline cellulose into cellobioses in a processive manner [29]. The binding of EG and CBHI onto the cellulose surface, along with the processive movement of CBHI, are considered essential steps in the degradation of cellulose [29]. Besides enzymes, the presence of a living microorganism is also important to the hydrolysis of cellulose. Generally, the radius of a spherical yeast cell is around 2 μm [30], which is nearly 400-fold larger than the hydrodynamic radius of a cellulase protein (≥5 nm) [31]. The immobilization of enzymes onto the cell wall may block the movement of processive enzymes (e.g., CBHI) or cause steric restriction in the collision with cellulose. As shown in Fig. 6a, in the case of strain EG-D-CBHI-D, three different cellulases were displayed on the cell surface in relatively close proximity (e.g., CBHI-BGL distance) compared to the strains constructed by other strategies. Such co-localization is expected to increase the occurrence of synergistic interactions among cellulases [8] and to facilitate the transportation of glucose into cells. However, due to the size of the cell–enzyme complex, the penetration of EG and CBHI into the internal space of cellulose is expected to be limited; such penetration has been reported as an important determinant of hydrolytic rate [14, 32]. Additionally, the processive movement of CBHI may be retarded due to its immobilization on the cell. In previous pre-steady-state analyses of CBHI activity on cellulose, stalling of the processive movement of CBHI was reported to lead to lower specific activity [33, 34]. By contrast, EG and CBHI in strain EG-S-CBHI-S were produced as free forms (Fig. 6d), a strategy that was expected to decrease steric hindrance and to increase the chance of collision with the substrate. Although the cellulases appeared to accumulate in fermentation with EG-S-CBHI-S (rising from 27 mU/mL at 0 h to 48 mU/mL at 96 h), the cellulolytic enzyme activity was still lower than that of EG-D-CBHI-D. Additionally, the diffusion efficiency of enzymes might affect the ability to degrade cellulose, especially in a substrate with higher viscosity (e.g., PASC). Thus, co-locating EG and CBHI in the same space (on the cell surface or in the medium) is favorable for amorphous cellulose-based ethanol fermentation.

By contrast, ethanol production levels were apparently lower in strains EG-D-CBHI-S and EG-S-CBHI-D. We infer that yeast cells attach to the surface of cellulose to facilitate the collision between immobilized cellulases and cellulose, but that this proximity may block the access of secreted EG or CBHI to the surface of the substrate (Fig. 6b, c). For instance, the cellulase activities of strain EG-D-CBHI-S at 96 h were higher than those of EG-S-CBHI-S, while PASCase activity was lower than that of strain EG-S-CBHI-S, effects that may be due to the steric hindrance mentioned above. Additionally, the hydrolysis efficiency of strain EG-S-CBHI-D appeared to be further decreased, presumably by the retarded movement of immobilized CBHI on the cell surface, as evidenced by the lower PASCase activity and ethanol production obtained in strain EG-S-CBHI-D. Interestingly, the individual enzyme activities in strain EG-D-CBHI-D were raised while PASCase activity kept constant, assuming that only the enzymes located in the contact region between yeast cell and cellulose could participate in cellulose degradation. Although the individual enzyme activities in whole cell were raised, the improvement of cellulolytic activity in “attaching region” is limiting. In contrast, most of the free cellulases can bind to cellulose and then participate in hydrolysis; the increase of individual enzyme activities could be reflected on improvement of cellulolytic activity. This may also explain why the overall enzyme activity of strain EG-S-CBHI-S (secretion system) rose observably (about 77 %) along with the increase of individual enzyme activities. Nonetheless, our results do indicate that the involvement of the microorganism in the synergism of cellulases may affect their hydrolysis efficiency towards cellulose. Moreover, the interactions between microbial cells and cellulose may also affect the cellulose degradation process. For instance, Francisco et al. reported that the engineered E. coli with displayed Cex CBH on the surface exhibited specific adhesion capacity toward cellulose [35], suggesting that the cell-to-cellulose adhesion pattern may lead to a distinct mechanism from that of free cellulases to break down cellulose.

From an industrial point of view, the reuse of yeast cells through various rounds of fermentation may be important in the ethanol production process [36]. Cell-recycle batch fermentation (CRBF) is a semi-continuous operation strategy using high densities of recycled cells to produce ethanol continuously. In previous reports, the CRBF of lignocelluloses was performed by the addition of large amount of commercial cellulases, a step that represents one of the main bottlenecks for commercialization [37, 38]. The application of a cellulolytic yeast strains to CRBF is expected to decrease the need for the addition of costly commercial enzymes [4]. Thus, in this study, the direct ethanol production from PASC using recycled cellulolytic cells was conducted without the addition of commercial cellulases. After a 3-batch recycle fermentation, our strain EG-D-CBHI-D retained 60 % of the PASC degradation ability that the strain had in the 1st batch. This retention of activity was 1.7-fold higher than that seen in strain EG-S-CBHI-S. The apparently lower degradation ability in strain EG-S-CBHI-S likely can be attributed to the loss of cellulases in each cycle, as most secreted enzymes were separated from cells during the cell collection step at the beginning of each batch. Although the amount of enzyme produced was sufficient for saccharification in the first batch, the ability to generate cellulases in each new cycle apparently declined in subsequent batches. In contrast, the enzymes immobilized on strain EG-D-CBHI-D showed more consistent activity than the activities in strain EG-S-CBHI-S during the recycling. Khaw et al. reported that the ethanol production rate of a yeast strain producing displayed α-amylase was maintained during a number of repetitions [39]. Consequently, the construction of cellulolytic yeast by cell-surface display strategy is more applicable to CRBF compared with the secretion strategy. Matano et al. even reported that with the addition of extraneous cellulases (10 FPU/g-biomass), the fermentation ability of a cellulase-displaying yeast strain remained constant after 5-cycle repeated-batch fermentation [40]. However, the required amount of additional cellulase was still too high, precluding economic feasibility for an industrial process [40]. In future work, we propose to further improve the cellulolytic activity of yeast strains and apply the cell-surface display strategy to ethanol production from lignocellulosic biomass (e.g., rice straw) with addition of trace amounts of exogenous cellulases.