INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The production of fuels and chemicals from cellulosic substrates using microbial biocatalysts offers the potential to reduce the use of fossil fuels and improve the environment (11, 18, 22, 29-31). However, the low activity of cellulase enzymes (15, 30) and the resulting cost of hydrolysis represent major barriers for the use of lignocellulosic feedstocks for fuels, bulk chemicals, and plastics (15, 29, 30, 41). The enzymatic hydrolysis of cellulose has been extensively studied but remains poorly understood (2, 3, 5, 21, 32). Hydrolysis results from the combined action of at least three classes of -1,4-glucanase activities (2, 5, 26, 28): endoglucanases, exoglucanases, and cellobiases that complete the hydrolysis of soluble products (from two to six glucosyl residues) to monomeric glucose. Cellobiose and soluble cellobiosides are potent competitive inhibitors of endo- and exoglucanases that must be removed to prevent autoinhibition. Fungi such as Trichoderma reesei (19, 27, 39) secrete soluble -1,4-glucosidases (cellobiase) to complete the hydrolysis process. In some bacteria, however, hydrolysis of soluble cellobiosides is completed intracellularly. Cellobiose and cellotriose are actively transported by a -glucoside-specific phosphoenolpyruvate-dependent phosphotransferase system (PTS) (18, 20, 40). Intracellular cellobiose-phosphate is subsequently hydrolyzed to glucose and glucose-6-phosphate by a cytoplasmic phospho--glucosidase for immediate entry into glycolysis.

One approach to reduce the costs of cellulase enzymes for bioprocessing is to develop ethanologenic biocatalysts that provide a portion of the cellulolytic activity. This approach is being pursued with several naturally ethanologenic microorganisms such as Saccharomyces cerevisiae (8, 9, 33, 34), Zymomonas mobilis (7, 24, 31), and cellulolytic bacteria such as Clostridium thermocellum and C. thermosaccharolyticum (16, 35). Our laboratory is pursuing an alternative approach: the addition of cellulase activities to enteric bacteria (Escherichia coli B and Klebsiella oxytoca M5A1) that have been previously engineered to produce ethanol by adding the Z. mobilis pdc and adhB genes encoding the pyruvate-to-ethanol pathway (18). K. oxytoca M5A1 contains a native cellobiose-specific PTS and phospho--glucosidase (20, 25, 40), eliminating the need for extracellular -glucosidase. The K. oxytoca operon containing the cellobiose utilization genes (casAB) has been cloned (20) and expressed in ethanologenic E. coli KO11 (18, 25) to construct a second, analogous biocatalyst. Subsequent studies have overexpressed the celZ endoglucanase (CelZ; formerly EGZ) from the plant saprophyte, Erwinia chrysanthemi (6, 10), in both K. oxytoca M5A1 (42) and E. coli B (44). By adding accessory genes (out genes) from E. chrysanthemi, high levels of endoglucanase activity were effectively secreted into the extracellular milieu by both organisms (13, 42, 44).

CelZ represents 95% of the total carboxymethyl cellulase (CMCase) activity in E. chrysanthemi (6). The remaining 5% is attributed to a second endoglucanase, CelY (formerly EGY). The celY gene has also been cloned and sequenced (13). Recent investigations have shown that CelY and CelZ differ in substrate preference (43). When mixed at ratios similar to those produced by cultures of E. chrysanthemi, these two enzymes function synergistically, indicating a potential need for both enzymes for optimal cellulose hydrolysis during fermentation.

In the current study, we have integrated the celY and celZ genes from E. chrysanthemi into the chromosome of K. oxytoca P2, an ethanologenic derivative of M5A1 (40). The secreted cellulase enzymes functioned together with commercial fungal cellulase to increase the production of ethanol during the simultaneous saccharification and fermentation (SSF) of crystalline cellulose.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteria, plasmids, and culture conditions. Strains and plasmids used in this study are listed in Table 1. E. coli DH5 and TOPO10F' were used as hosts during plasmid constructions. The celZ gene was cloned from E. chrysanthemi P86021 (42, 44). The celY gene was cloned by Guiseppi et al. (13) from E. chrysanthemi 3937. The out genes were cloned by He et al. (14) from E. chrysanthemi EC16.

Genetic methods. Standard methods were used for plasmid construction, analyses, and sequencing (1). The ribosome-binding site and promoterless coding region of celY were amplified by the PCR using pMH18 as the template with the following primer pair: N terminus, 5'-CTGTTCCGTTACCAACAC-3', and C terminus, 5'-GTGAATGGGATCACGAGT-3'. The E. chrysanthemi out genes (pCPP2006) were transferred by conjugation using pRK2013 for mobilization (42). Constructions were confirmed by sequencing using the dideoxy method and a LI-COR Model 4000-L DNA sequencer with fluorescent primers. The E. chrysanthemi celY and celZ genes were introduced into K. oxytoca P2 by electroporation using a Bio-Rad Gene Pulser. Recombinants were selected on solid medium containing kanamycin (50 mg/liter) as previously described (23, 42).

Primer extension analysis. Promoter regions were identified by mapping the transcriptional start sites using IRD41-labeled fluorescent primers within the coding regions (44): 5'-ACCATCAGCATCAACGCCCAACAACG-3' for celY and 5'-GACTGGATGGTTATCCGAATAAGAGAGAGG-3' for celZ. Extension products were dissolved in loading buffer and compared to parallel dideoxy sequences (42) using the LI-COR Model 4000-L DNA sequencer (LI-COR, Inc., Lincoln, Nebr.).

Enzyme assay. Endoglucanase activity was determined in vitro using 2% CMC as the substrate. Appropriate dilutions of cell-free culture broth (extracellular activity) or broth containing cells that had been disrupted by ultrasound (total activity) were assayed at 35°C in 50 mM citrate buffer (pH 5.2). Reactions were terminated by heating to 100°C for 10 min. Reducing sugars were measured using 3,5-dinitrosalicylic acid reagent with glucose as a standard (42, 44). Endoglucanase activity (CMCase) is expressed as micromoles of reducing sugar per minute (IU). Results are an average of two or more determinations.

Fermentation. SSF tests were conducted in unbaffled, 500-ml flasks containing 200 ml of broth. Flasks were fitted with a rubber stopper and vented with an 18-gauge needle. Fermentations were conducted at 35°C (120 rpm) in LB medium containing 10% Sigmacell 50 (crystalline cellulose). Inocula were grown for 12 h in LB medium containing 5% glucose. Cells were harvested by centrifugation and resuspended in LB medium. Each flask was inoculated with approximately 64 mg of cells (dry weight).

Materials and chemicals. Tryptone and yeast extract were products of Difco (Detroit, Mich.). Antibiotics, low-viscosity CMC, and Sigmacell 50 were obtained from the Sigma Chemical Company (St. Louis, Mo.). The IRD41-labeled fluorescent primers were purchased from LI-COR, Inc.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of a promoter-probe vector for celY. Although the celY gene from E. chrysanthemi was cloned previously, it was poorly expressed in E. coli from the original plasmid (13). To increase expression, a promoter-probe vector was constructed as follows using celY as the reporter (Fig. 1). The promoterless celY coding region with a native ribosome-binding site (1.2 kbp) was amplified by PCR using pMH18 as the template and inserted into the topoisomerase vector, PCR2.1-TOPO. A clone oriented to express celY from the lac promoter was selected and designated pLOI2311 (5.2 kbp). An EcoRI fragment containing the promoterless celY gene was isolated from pLOI2311. The ends of this fragment were blunted using Klenow polymerase prior to ligation into the HincII site of pUC18. A clone oriented to express celY from the lac promoter was selected (3.9 kbp), and expression was confirmed using endoglucanase indicator plates (pLOI2316). The promoterless celY gene was isolated from pLOI2316 as a 1.2-kbp fragment using EcoRI and HindIII and inserted into the corresponding sites of pLOI2302 (pUC19 derivative) to reverse the direction of the celY gene. As expected, the resulting construct DH5(pLOI2317) was inactive on endoglucanase indicator plates due to the lack of a promoter. To facilitate the insertion of DNA fragments containing promoter regions, plasmid pLOI2317 (3.9 kbp) contains a BamHI site in the polylinker region, immediately upstream from the celY gene (Fig. 1).

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Construction of promoter-probe vector for celY. Sau3A1 fragments of Z. mobilis chromosomal DNA were ligated into the BamHI site of pLOI2317 to provide a strong, surrogate promoter for the celY coding region (solid segments). The Z. mobilis DNA fragment (promoter 1) is shown as an open segment. Replicons and antibiotic resistance genes are stippled; other vector DNA is shown as thin connecting lines. The arrows indicate the direction of transcription.

Construction of plasmids with increased expression of celY in E. coli DH5. Sau3A1 fragments of Z. mobilis chromosomal DNA were used to provide a heterologous promoter that would not be subject to native regulatory mechanisms in K. oxytoca or interfere with subsequent integration into the K. oxytoca chromosome (42). Fragments of 0.5 to 1.5 kbp were isolated and randomly ligated into the BamHI site of pLOI2317 to generate a library of surrogate promoters (Fig. 1). Approximately 7,500 colonies were screened on endoglucanase indicator plates. One-third of the clones actively produced celY. The most active 100 colonies were identified by zone size, purified, and retested. The 30 clones with the largest zones of activity were grown overnight in LB medium and assayed for CMCase activity. The five most active clones are listed in Table 2 and exhibited approximately sevenfold-higher activity than the original clone, pMH18. Plasmid pLOI2323 was selected for further investigation.

View this table: [in this window] [in a new window]   TABLE 2.   Expression of celY in E. coli DH5 using Sau3A1 fragments of Z. mobilis chromosomal DNA as surrogate promotersa

View this table: [in this window] [in a new window]   TABLE 3.   Transcriptional initiation sites and putative promoter regions for the celY promoter in DH5(pLOI2323)a

Construction of a vector for the integration celY and celZ into the chromosome of K. oxytoca P2. Plasmid pLOI2307 (7.2 kbp) was previously constructed and expressed celZ from a surrogate Z. mobilis promoter at high levels in recombinant E. coli DH5 (44) and K. oxytoca M5A1 (42). To facilitate subcloning of this hybrid celZ gene and promoter (4.5 kbp), an EcoRI linker was inserted into the T4 DNA polymerase-treated SphI site of pLOI2307 to provide flanking EcoRI sites for convenient excision (pLOI2349). Prior to constructing a plasmid containing celY and celZ, a random 3-kbp fragment of EcoRI-digested K. oxytoca M5A1 chromosomal DNA was inserted into pLOI2323 containing celY (and surrogate promoter) to serve as a guide for homologous recombination (pLOI2348; 8 kbp). This 3-kbp M5A1 fragment was partially sequenced and appears to encode the complete M5A1 glgP gene. In pLOI2348 (8 kbp), flanking AscI sites allowed the excision of a single 5.5-kbp fragment containing the M5A1 glgP gene, the Z. mobilis surrogate promoter, and E. chrysanthemi celY.

View larger version (29K): [in this window] [in a new window]   FIG. 2.   Construction of pLOI2352 for the functional integration of celY and celZ into the chromosome of ethanologenic K. oxytoca P2. Coding regions for celY and celZ are shown as solid segments. Fragments of Z. mobilis DNA that serve as promoters (prom1 and prom2) are shown as open segments. The K. oxytoca DNA fragment which serves as a guide for chromosomal integration (guide) is stippled. Replicons and antibiotic resistance genes are stippled; other vector DNA is shown as a thin connecting line. The arrows on segments indicate the direction of transcription. The small open arrows represent the FRT sites which are recognized by the flp recombinase. FRT sequences are asymmetrical and are arranged to allow the deletion of plasmid DNA (replicon and selectable marker) after chromosomal integration.

Functional integration of celY and celZ into the K. oxytoca P2 chromosome. Plasmid pLOI2352 was introduced into P2 by electroporation, followed by selection for kanamycin resistance. Approximately 150 colonies were recovered, and all were positive on endoglucanase indicator plates. Ten clones with the largest zones of activity were purified, grown in broth, and assayed for endoglucanase activity. These produced 5 to 6 IU of endoglucanase activity per ml. One clone was selected for further study and designated SZ12.

Construction of a celZ knockout mutation. To confirm the presence of a functional celY in SZ21, a knockout mutation of the chromosomally integrated celZ was constructed by double, homologous recombination using plasmid pLOI2357 (Fig. 3). Plasmid pUC19 was digested with SmaI and HindIII, treated with Klenow polymerase, and self-ligated to eliminate many of the polylinker sites (pLOI2354). The remaining unique EcoRI site was used to insert a 4.5-kbp EcoRI fragment containing the promoter and celZ gene from pLOI2349 to make pLOI2355 (7.2 kbp). The 1.8-kbp SmaI fragment from pHP45aac containing the apramycin resistance gene (aac) was then ligated into the central region of celZ, replacing a small internal PstI fragment (after treatment with T4 DNA polymerase) to produce pLOI2356 (9 kbp). The 6.3-kbp EcoRI fragment from this plasmid was isolated and inserted into the core integration vector, pLOI2224, to produce pLOI2357 (8.2 kbp). This plasmid contains a conditional R6K replicon and kanamycin resistance gene in addition to a celZ gene that is interrupted by an apramycin resistance gene.

View larger version (32K): [in this window] [in a new window]   FIG. 3.   Construction of pLOI2357 for the inactivation of celZ by double homologous recombination. Coding regions for celY are shown as solid segments. After disruption, the N-terminal and C-terminal segments are indicated by celZ' and celZ", respectively. The fragment of Z. mobilis DNA that serves as a promoter (prom1) is shown as an open segment. Replicons and antibiotic resistance genes are stippled; other vector DNA is shown as a thin connecting line. The arrows on segments indicate the direction of transcription. The small open arrows represent the FRT sites which are recognized by the flp recombinase.

Effect of the E. chrysanthemi out genes (pCPP2006) on the extracellular secretion of CelY and CelZ in derivatives of K. oxytoca P2. Table 4 summarizes the endoglucanase activities exhibited by cellulolytic derivatives of ethanologenic K. oxytoca P2. Strain SZ6 (42) contains a chromosomally integrated hybrid celZ gene with the same promoter fragment used to construct SZ21. Despite the presence of two endoglucanase genes in SZ21, extracellular and total endoglucanase activities were 13% lower in this strain than in SZ6. Most of the endoglucanase activity produced by SZ21 can be attributed to celZ. SZ22, a celZ mutant of SZ21, expressed only 11% of the endoglucanase produced by the parent containing functional celY and celZ genes. In strains SZ6 and SZ21 containing a functional celZ, most of the endoglucanase activity (primarily CelZ) was cell associated. In strain SZ22, containing a functional celY alone, half of the endoglucanase activity was extracellular.

Synergism between recombinant E. chrysanthemi endoglucanase (CelZ and CelY) and fungal cellulase (Spezyme CE) during the fermentation of cellulose to ethanol. SSF experiments were performed with flasks without pH control to evaluate the combined effects of fungal cellulase (Spezyme) and cellulase enzymes produced by the biocatalysts on ethanol production from Sigmacell 50, a highly crystalline substrate (Table 5). Although very low levels of ethanol were produced by all strains in the absence of Spezyme, strains SZ6(pCPP2006) and SZ21(pCPP2006) containing functional celZ genes produced higher levels of ethanol (P  0.05) than strain SZ22(pCPP2006) containing only a functional celY gene and strain P2(pCPP2006) lacking both endoglucanase genes. In the absence of both Spezyme and Sigmacell 50, all strains produced 0.22 g of ethanol per liter. The additional increment of ethanol produced by SZ6(pCPP2006) and SZ21(pCPP2006) during incubation with Sigmacell 50 is attributed to hydrolysis of the small fraction of amorphous cellulose in the substrate by CelZ (42-44). Digestion of amorphous cellulose by CelY alone produces saccharides that are too large to be transported and metabolized without further hydrolysis (43).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Although the development of a recombinant biocatalyst that can supply all of the enzymes needed for the hydrolysis of cellulose and ethanol production has yet to be achieved, incremental progress has been made with both bacterial and yeast systems. Plasmid-based strains of Z. mobilis have been developed that express low levels (estimated to be <100 U/liter) of both an endoglucanase and a cellobiase from Xanthomonas albilineans (31). Higher levels of activity (600 to 1,000 U/liter) were produced by plasmid-based strains of Z. mobilis expressing an endoglucanase gene from Cellulomonas uda (24) and the celZ gene (CelZ) from E. chrysanthemi (7). Although the C. uda endoglucanase remained cell associated in recombinant Z. mobilis, up to 40% of the CelZ was released from the periplasm in late stationary phase without apparent lysis (7). No data were presented concerning the utilization of cellobiose or -1-4-linked glucosides for growth and ethanol production by this strain.

Cellobiase and three different cellulase genes have been coexpressed from a single plasmid in recombinant Saccharomyces cerevisiae (33): cellobiase (bgl1) from Endomyces fibuliger, endoglucanase (end1) from Butyrivibrio fibrisolvens, cellobiohydrolase (cbh1) from Phanerochaete chrysosporium, and cellodextrinase (cel1) from Ruminococcus flavefaciens. Secretion of the gene products was facilitated by adding various leader sequences. URA3-based autoselection in an auxotrophic host was used to ensure plasmid retention. The resulting strain expressed higher levels of cellobiase activity than cellulase activity, sufficient to allow slow growth of the recombinant on cellobiose. An analogous construct expressing two cellulase genes (34) produced higher levels of endoglucanase (approximately 1,200 U/liter). No data were presented regarding ethanol production by these strains or growth on polymeric -1-4-linked glucosides.

A recent study has shown that substantial levels of cellulases can be produced and secreted in recombinant S. cerevisiae by integrating up to 50 copies of the -glucosidase gene from Bacillus circulans and an endo- or exocellulase gene from Bacillus sp. strain DO4 into multiple chromosomal  sites of Ty1 retrotransposons (8, 9). Both genes contained added leader sequences to facilitate efficient secretion. In batch fermentations, 300 to 500 U of both activities were produced per liter. Recombinant strains achieved approximately 30% more cell mass and ethanol than the parent strain during growth on a mixture of glucose and soluble cellobiosides (9). A novel two-step process was developed using the recombinant in which high levels of cell mass and cellulase were initially produced aerobically from glucose (9). After supplementing this broth with additional commercial cellulase, crystalline cellulose substrate was added and fermented to ethanol. Although broth from the first step contained up to 700 filter paper units (FPU) of cellulase activity per liter (Whatman filter paper hydrolysis), no ethanol was produced without the further addition of commercial cellulase. Production of these enzymes by recombinant S. cerevisiae allowed a 40% reduction in the amount of commercial enzyme required, from 30 to 18 FPU/g of cellulose.

Our approach to reduce the requirement for fungal cellulase has been to functionally express genes encoding the Z. mobilis pathway in K. oxytoca, an abundant organism in pulp mill waste that has the natural ability to transport and metabolize cellobiose (20, 40). This recombinant can rapidly and efficiently produce 40 to 50 g/liter of ethanol from cellobiose (40). We have now added two endoglucanase genes from a closely related organism that macerates plant cell walls and tissues in nature, i.e., E. chrysanthemi (6, 10). High levels of endoglucanase activity were produced (over 20,000 U/liter), equivalent to 1% of the endoglucanase present in concentrated commercial cellulase products (Table 4 and reference 27). These activities are more than an order of magnitude higher than was previously reported for engineered strains of S. cerevisiae or Z. mobilis. To minimize some of the problems associated with the expression of heterologous genes in industrial strains, unregulated promoters were isolated from random fragments of Z. mobilis DNA using functional assays, hybrid genes were integrated into the chromosome to ensure stability, and the antibiotic resistance markers used in construction were deleted using the FLP recombinase system (23) to facilitate future genetic modifications. Approximately 95% of the CMCase activity produced by this strain was attributed to the celZ gene product, with the balance attributed to the celY product.

CelY and CelZ appear to be secreted by different mechanisms. Approximately 70% of the CelZ was secreted as an extracellular product when the E. chrysanthemi out genes were added on a plasmid (pCPP2006), consistent with a type II secretion system (16). Half of the CelY activity was secreted in the presence or absence of the out genes, consistent with a type IV secretion system (17). The integrated genes have been maintained without selection for over 6 months. However, pCPP2006 containing the out genes is less stable and requires antibiotic selection to ensure maintenance. In SSF experiments with added commercial cellulase, two of the new K. oxytoca recombinants, SZ21 and SZ22, produced more ethanol than the parent strain lacking E. chrysanthemi cellulases. Both of these strains also produced ethanol levels equivalent to the best yeast SSF experiments (9) using approximately one-third of the amount of added commercial cellulase (5 PFU/g of cellulose versus 18 FPU/g of cellulose for recombinant yeast [9]).

Previous studies have shown that endoglucanases CelY and CelZ function synergistically to affect the hydrolysis of CMC and amorphous cellulose (43). This synergy is based on differences in substrate range and mechanism of action. CelZ hydrolyzes cellotriose, cellotetraose, cellopentaose, amorphous cellulose, and CMC. CelY hydrolyzes polymeric substrates to products of approximately 10 glucosyl residues. Optimum synergy was observed with a high ratio of CelZ to CelY, similar to that produced in nature by E. chrysanthemi and by SZ21. Strains SZ6(pCPP2006) and SZ21(pCPP2006) expressing CelZ and CelY+CelZ, respectively, produced small amounts of ethanol from Sigmacell 50 even without the addition of fungal enzymes (Table 5). Both ethanol values were significantly higher than strains grown in broth without added cellulose and strain P2(pCPP2006) with cellulose (no endoglucanase genes). This small amount of ethanol is attributed to the hydrolysis of a minor amorphous component of Sigmacell 50.

SSF experiments (Table 5) demonstrated that bacterial cellulases produced by ethanologenic K. oxytoca functioned synergistically with added commercial cellulase to increase ethanol production (7 to 22%) from crystalline cellulose (Sigmacell 50). Surprisingly, the beneficial effect was attributed almost exclusively to CelY, despite the fact that CelY activities were quite low in comparison to CelZ. Strain SZ22 expressing CelY was nearly equivalent to strain SZ21 expressing both activities. Strain SZ6 expressing only CelZ showed little increase in ethanol despite the production of over 20,000 U of endoglucanase activity per liter. Differences in the effectiveness of CelY and CelZ in combination with Spezyme may result from differences in substrate specificities and modes of action. Fungi such as T. reesei produce multiple endoglucanase activities which are presumed to function together with exoglucanases during the hydrolysis of crystalline cellulose (26, 32, 39). It is possible that the fungal activity resembling CelZ is not limiting hydrolysis in dilutions of Spezyme. In contrast to CelZ, CelY does not hydrolyze soluble cellobiosides but preferentially acts on longer-chain substrates, producing ends which can function as new sites for exoglucanase activity. Thus, endoglucanases with an activity profile resembling that for CelY may be limiting in fungal preparations supplied commercially.

In the absence of fungal cellulase additions, previous studies have shown that CelY and CelZ function synergistically to degrade amorphous cellulose (43). In nature, lignocellulosic substrates are depolymerized by mixtures of extracellular enzymes produced by consortia of fungi and bacteria. Thus, it is not surprising that a mixture E. chrysanthemi enzymes and T. reesei enzymes can improve the digestion of lignocellulosic substrates during bioconversion to ethanol.