Efficient yeast cell-surface display of exo- and endo-cellulase using the SED1 anchoring region and its original promoter

The search for practical petroleum substitutes from renewable resources has become a global priority as atmospheric carbon dioxide levels continue to rise. Lignocellulosic materials such as sugar cane bagasse, corn stover, rice straw, grasses, wood chips and other agricultural waste contain large amounts of polysaccharides such as cellulose and hemicellulose. These polysaccharides are one of the most abundant renewable resources in nature, and are attracting much attention as feedstocks for second-generation bioethanol production [1]. However, due to the recalcitrance of lignocellulosic materials, the pre-treatment requiring a large amount of energy and the large quantities of multiple enzymes are necessary for the complete hydrolysis of these polysaccharides to fermentable sugars [2–5]. The cost of the bioconversion of these materials to ethanol is much higher than that of conventional sugar sources such as corn starch and cane juice [1]. In particular, the cost of the enzymes is one of the biggest obstacles for commercially viable production of bioethanol from lignocellulosic materials. Therefore, consolidated bioprocessing (CBP), which combines enzyme production, saccharification of polysaccharides, and fermentation into a single process, has a great potential for the cost-effective production of ethanol from lignocellulosic materials [6–8]. To date, however, no natural microorganisms with the capability for efficient enzyme production, lignocellulose saccharification, and ethanol production have been identified.

Cellulose, a polymer of β-(1 to 4) linked glucose residues, is the major component of lignocellulosic materials (40 to 60% by total dry weight) [9]. Saccharomyces cerevisiae (S. cerevisiae) is the most commonly used microorganism for industrial ethanol production. Compared to bacteria, it has more rapid sugar consumption, provides a higher ethanol yield from glucose, and has higher resistance to ethanol and other compounds present in lignocellulosic hydrolysates [10, 11]. However, native S. cerevisiae is not a candidate microorganism for the CBP of cellulose because it lacks the enzymes required for the hydrolyzation of cellulose into glucose. To address this problem, the heterologous expression and secretion of cellulolytic enzymes, including endoglucanase (EG), cellobiohydrolase (CBH) and β-glucosidase (BGL) from yeasts and other microorganisms, have been pursued over the last two decades [12]. On the other hand, efficient whole-cell biocatalysts for simultaneous saccharification and fermentation have also been constructed by displaying cellulolytic enzymes on the cell surface using the glycosylphosphatidylinositol (GPI) anchoring system [13–18]. The cell-surface engineering system has several industrial advantages over the enzyme secretion system. In this system, yeast is transformed by introducing fusion genes coding cellulolytic enzymes and the anchoring domain of GPI protein. The fused proteins self-immobilize into the yeast cell wall, so the activities of the enzymes are retained as long as the yeast continue to grow [19]. As the separation of the biocatalyst from the products is easy, reutilization of the yeast cells also enables re-use of the active enzymes displayed on the cell surface, without the need for the cells to reproduce. This approach would reduce the cost of yeast propagation as well as the cost of enzyme addition [20]. Furthermore, Matano et al. [21] recently reported a cell-surface display system that alleviated the adsorption of free cellulases onto the crystalline cellulose and improved the hydrolysis of cellulose and ethanol production.

In previous studies, the C-terminal domain of α-agglutinin (Sag1), which is a GPI protein in S. cerevisiae, has been mainly used as an anchoring domain in fusion proteins. The genes encoding these fusion proteins have been expressed by constitutive promoters such as TDH3 and PGK1 promoters. However, the cellulolytic activities on the cell surface provided using conventional gene cassettes are insufficient for the efficient CBP of cellulose. There is therefore, urgent need for improvements in yeast cell-surface display cassettes.

In this study, we constructed the novel gene cassettes for the cell-surface display of cellulolytic enzymes that incorporated the promoter and the GPI anchoring region derived from S. cerevisiae SED1, which encodes a major stress-induced structural GPI protein [22]. The gene cassettes for the cell-surface display of Aspergillus aculeatus β-glucosidase (BGL1) were integrated into the HIS3 locus of the S. cerevisiae genome, then the BGL activity of these recombinant strains was evaluated. The results demonstrated that the simultaneous utilization of the SED1 promoter and the anchoring region significantly improved BGL activity on the cell surface. In addition, we also constructed a recombinant yeast strain displaying the Trichoderma reeseii endoglucanase II (EGII) using this novel gene cassette. Hydrolysis activity against the water-insoluble cellulose of this strain also increased significantly. Furthermore, the ethanol productivity from hydrothermally processed rice straw was clearly improved by using this EGII-displaying strain.

In this study, we developed novel gene cassettes for the efficient cell-surface display of cellulolytic enzymes using the S. cerevisiae SED1 promoter and anchoring region (Figure 1). BGL-displaying strains using the novel gene cassettes (BY-BG-TS, SA and SS) achieved higher BGL activity on the cell surface than a conventional strain (BY-BG-TA) harboring a cassette carrying the TDH3 promoter and the SAG1 anchoring region (Figure 2B). The BY-BG-SS strain, which had a gene cassette with the S. cerevisiae SED1 promoter and its anchoring region (SS cassette), achieved the highest BGL activity (Figure 2B). Comparison of GPI anchoring domains for effective cell-surface display using the inducible GAL7 promoter has been reported [25]. However, there have been no reports on the effect of the combinations of a promoter and an anchoring domain for enzyme cell-surface display. To our knowledge, this is the first report on the synergetic effect of a GPI anchoring domain and its original promoter for the cell-surface display.

Previously, the activity of heterologous enzymes displayed on a yeast cell surface was improved by increasing the copy number of the gene cassettes, for example by using the multi-copy genome integration method (δ-integration) [26]. However, the enzyme activity of recombinant strains constructed by the δ-integration method is relatively unstable, due to the random genome integration of gene cassettes. Yamada et al. [26] constructed a recombinant S. cerevisiae strain, MT8-1/δBEC, by δ-integration. MT8-1/δBEC had six copies of a gene cassette for BGL display using the PGK1 promoter and the Sag1 anchoring domain. MT8-1/δBEC achieved a BGL activity of 15 U/g wet cells, which is comparable to approximately 100 U/g dry cells after 72 h cultivation. On the other hand, the BGL activity of BY-BG-SS reached 235 ± 28 U/g dry cells after only 48 h cultivation, even though it carried only a single gene cassette for BGL display. Although these strains cannot be compared directly, because MT8-1 and BY4741 strains have different genetic backgrounds, this novel gene cassette will thus allow the more stable construction of a heterologous enzyme-displaying yeast strain with high activity. The BGL activity of BY-BG-SS (235 ± 28 U/g dry cells after 48 h) is close to the specific activity of purified A. aculeatus BGL1 (280.7 U/g protein) at pH 5.0 and 30°C [27]. The K _m value of the BGL on the BY-BG-SS strain for p NPG (0.198 mM) is lower than that of purified A. aculeatus BGL1 (0.23 mM) [27]. Furthermore, SS cassettes also have a significant effect on the cell-surface display of T. reeseii EGII, which is an endo-type cellulolytic enzyme. Hydrolysis activity of BY-EG-SS against the water-insoluble substrate, AZCL-HE-Cellulose, was 106-fold higher than that of the conventional strain, BY-EG-TA (Table 3). These results suggest that recombinant yeast cells displaying cellulolytic enzymes using SS cassettes are highly promising biocatalysts for efficient ethanol production from cellulosic materials.

It has been reported that the SED1 expression level increases during the stationary phase [22]. As shown in Figure 2A, the expression level of the BGL1 controlled by SED1 promoter rose dramatically and rapidly after 48 h cultivation. The BGL1 activity of BY-BG-SA and SS was 3.5- and 7.4-fold higher than that of BY-BG-TA and TS after 48 h, respectively (Figure 2B). The high expression level of BGL1 controlled by the SED1 promoter is likely one of the causes for the high BGL activity of the BY-BG-SA and SS strains. On the other hand, it was previously reported that the mRNA level of SED1 varies periodically during the cell cycle and that the peak of SED1 expression is M phase, whereas no apparent peak was observed in TDH3 expression during the cell cycle [28]. In the cell cycle of S. cerevisiae, the synthesis of the new cell wall is activated in the M phase. The expression of many genes involved in the synthesis of cell-wall components and their delivery also increases in this phase [28], suggesting that the M phase is the most suitable for efficient cell wall integration of GPI proteins. Although further investigation using other constitutive promoters such as PGK1 promoter is needed, the cell cycle-regulated expression controlled by the SED1 promoter might be also one of the causes of high BGL activity of the BY-BG-SA and SS strains (Figure 2B).

The result shown in Figure 3 suggests that the amount of BGL1 immobilized in the cell wall varies depending on the C-terminal anchoring domain, and accounts for the difference in BGL activity between the BY-BG-SA and SS strains. It has been estimated that S. cerevisiae cell contains the 60 GPI proteins, and most of them can be integrated into the cell wall by a covalent attachment to the β-(1 to 6) glucan chains in the glucan layer [29]. In recombinant yeasts, heterologous GPI-anchoring fusion proteins must compete with the native GPI proteins for the limited protein incorporation capacity of the glucan layer [25]. Therefore, it is possible that the integration efficiency of GPI-anchoring enzymes strongly affects the population and activity of the enzymes in the cell wall [25]. It was recently reported that different GPI proteins exhibit different transportation and integration efficiencies into the cell wall [30–34]. Analysis of a large panel of biochemically verified GPI proteins suggested that specific residues in the upstream region of the ω-site, which is the cleavage/attachment site of the C-terminal GPI-anchoring signal recognized by a GPI transamidase, seem to favor cell wall localization [32–34]. Hamada et al. [33] reported that the amino acid sequence in the upstream region of ω-site differed substantially between Sed1 and Sag1, and that reporter protein fused with the Sed1 anchoring domain had higher cell-wall integration efficiency than protein fused with the Sag1 anchoring domain. The results shown in Figures 2 and 3 are in good agreement with these reports.

Another possible cause for the high BGL activity of the BY-BG-SS strain is the specific synergy of the SED1 promoter with the Sed1 anchoring domain in SS cassette. We constructed novel gene cassettes for BGL1 display with the S. cerevisiae CWP2 promoter and/or its anchoring region based on the SS cassette (Additional file 1). Cwp2 is a major GPI protein in S. cerevisiae, as is Sed1 [35]. The mRNA peak level of CWP2 during the cell cycle occurs during the G2 phase [28]. BGL-displaying strains harboring these cassettes (BY-BG-SC, CS and CC) were constructed and their BGL activity was evaluated (Additional file 2). BY-BG-CC achieved the highest BGL activity among these strains (148 ± 1 U/g dry cells after 48 h), whereas BY-BG-SC exhibited less than half the BGL activity of BY-BG-CC (71 ± 2 U/g dry cells after 48 h) (Additional file 2). These results suggest that each gene encoding GPI anchoring protein is expressed most efficiently at different phases during the cell cycle, as regulated by its original promoter in S. cerevisiae cells. After the attachment of the GPI anchor in the endoplasmic reticulum (ER), GPI proteins leave the ER in COPII-coated vesicles and travel via the Golgi to the plasma membrane [36]. During this intracellular transportation, each GPI protein and its GPI anchors are subject to a variety of modifications [29]. Although many aspects of the intracellular transportation and modification systems for each GPI protein remain unclear [29], cell cycle-regulated expression of a gene encoding GPI protein under the control of its original promoter might favor efficient intracellular transportation and modification. Further validation for other GPI proteins and their promoters is needed to prove this synergetic effect. Regardless, the simultaneous utilization of an anchoring domain of GPI protein and its original promoter is a good option for constructing gene cassettes providing efficient cell-surface display.

The cell wall of yeast is made of a thick microfibrillar array of β-(1 to 3) glucan and β-(1 to 6) glucan chains [37]. It is possible that a part of GPI anchoring enzymes expose their catalytic domain to external surface of the cell wall while the remainder is completely embedded in the glucan layer [25]. Small substrates such as cellobiose are accessible to all integrated enzymes because these substrates penetrate the cell wall. In contrast, large substrates such as cellulose can only access enzymes exposed on the external surface. Therefore, the hydrolysis of highly polymerized substrates by enzymes integrated into the yeast cell wall is more difficult than hydrolysis of small substrates [25, 38]. In this study, we constructed several T. reeseii EGII-displaying strains (BY-EG-TA, TS, SA and SS) and evaluated the enzyme activity of these cells using AZCL-HE-Cellulose as the water-insoluble substrate. Whereas BY-EG-TA and SA had very little activity against AZCL-HE-Cellulose, BY-EG-TS and SS showed significant activity (Table 3). This result suggests that the Sed1 anchoring domain provides significant advantages over the Sag1 anchoring domain not only in the efficiency of cell wall integration, but also in localizing the enzyme at the external surface of the cell wall. We constructed the novel gene cassette for EGII display with the S. cerevisiae CWP2 promoter and its anchoring region (Additional file 1). The EG-displaying strain harboring this cassette (BY-EG-CC) was constructed and the hydrolysis activity against AZCL-HE-Cellulose was evaluated. The hydrolysis activity of BY-EG-CC (A₅₉₀ = 0.021 ± 0.001) was lower than BY-EG-TA (0.030 ± 0.004), even though BGL-displaying strain using the CC cassette (BY-BG-CC) achieved high BGL activity (Additional file 2). This result also suggests that the anchoring domain has significant effect on the localization of enzymes in the cell wall. On the other hand, the possibility of the effect of the steric hindrance by different anchoring domains fused with EGII cannot be excluded. Although the full length of Sed1 was used as the Sed1 anchoring domain in this study, it has been reported that C-terminal-229 amino acids of Sed1 can be used as an anchoring domain [25]. It cannot be denied that the extra N-terminal sequence of Sed1 could function as a linker, causing the positive effect of the substrate accessibility of EGII. Investigation of the surface accessibility of the displayed enzymes by other experiments such as immunofluorescence labeling [39] would be necessary for more understanding about the localization of Sed1 in the cell wall.

SED1 is a major stress-induced gene in S. cerevisiae, and the expression level increases during stressful industrial fermentation such as sake brewing [22, 40]. Hasunuma and Kondo [1] suggested that the environment of an economically feasible CBP using cellulase-displaying yeast strains must also be stressful (highly concentrated yeast cells, high osmotic pressure, high temperature, low nutrition capacity and fluctuating processes). Therefore, the stress-induced property of the SED1 gene should be favorable for the cell-surface display of cellulolytic enzymes in CBP. On the other hand, the transcriptional levels of BGL1 controlled by the SED1 promoter decreased dramatically after 48 h cultivation (Figure 2A). This result suggested that any factors other than stress regulate the expression of SED1 promoter. Identification of the regulatory factors in the expression of SED1 promoter would be necessary for further enhancement of the cellulolytic activities on the cell surface provided using the SS cassette.

In the direct ethanol production from hydrothermally processed rice straw, the BY-EG-SS strain enhanced the saccharification and fermentation of the pretreated rice straw in the presence of a commercial cellulase cocktail (Figure 4). EGII was displayed in the cell surface of the BY-EG-SS strain and successfully compensated for cellulolytic activity of the commercial enzymes under stressful conditions. As stated in the Background Section, the cost of commercial enzymes is one of the biggest obstacles to cost-effective bioethanol production from lignocellulosic materials. The result shown in Figure 4 indicates that cellulase-displaying yeast strains using the SS cassette have significant potential for reducing the cost of commercial enzymes and for the cost-effective CBP of lignocellulosic biomass.

It has been reported that the synergetic effects of EG, CBH and BGL are necessary for the effective hydrolyzation of cellulose using cell-surface-displaying yeasts, and that optimization of the ratios of the activities of these enzymes is also important for the hydrolyzation efficiency [26, 41, 42]. The BY-EG-SS strain clearly improved the ethanol productivity from pretreated rice straw, but this strain expressed only EGII (Figure 4). The cellulolytic activity of this strain could increase with the additional display of other enzymes in the future. In addition, we revealed that the combination of the promoter and anchoring domains in the gene cassette has significant effects on the efficiency of enzyme display. If the activity of various enzymes on the cell surface can be controlled artificially by the combinations of their promoters and anchoring domains, we will be able to construct the recombinant yeast strains optimized for the CBP of lignocellulosic materials.

We have developed novel gene cassettes for the efficient cell-surface display of exo- and endo-type cellulolytic enzymes. We demonstrated that the combinations of a promoter and an anchoring domain in the display cassette had significant effects on the efficiency with which the enzyme integrated into the cell wall. Simultaneous utilization of the SED1 promoter and Sed1 anchoring domain in a gene cassette enabled highly efficient enzyme integration into the cell wall. The BGL-displaying yeast strain carrying this cassette achieved high BGL activity within a relatively short cultivation time. Furthermore, this novel display cassette is also suitable for the hydrolyzation of insoluble substrates including lignocellulosic biomass. Ethanol productivity from hydrothermally processed rice straw was clearly improved by using the novel gene cassettes for EGII display. If cell-surface cellulolytic enzyme activities can be controlled and optimized artificially by the combinations of the promoter and anchoring domain, then cellulase-displaying yeasts will be an attractive option for cost-effective CBP of lignocellulosic materials.