INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The cost of cellulase enzymes is one of the principal economic bottlenecks for the commercialization of lignocellulosic biomass-to-ethanol technology (13, 16, 18, 29, 32). Three different approaches have been proposed to reduce the cost of cellulase for the bioethanol industry: (i) improved on-site production of cellulase (16, 32), (ii) optimized reconstitution of cellulase components from different sources into a more effective artificial cellulase system (12, 43, 44), and (iii) development of improved ethanologenic biocatalysts which supply a portion of the cellulase needed for the direct microbial conversion of cellulose into ethanol (6, 17). Our lab has focused on the latter approach.

Recombinant strains of Escherichia coli B have been engineered which produce ethanol efficiently from all sugar constituents of lignocellulose by using the Zymomonas mobilis genes encoding the pyruvate-to-ethanol pathway (pdc and adhB) (20, 30). Genes for cellulose hydrolysis are now being added (27, 47). The solubilization of cellulose involves three classes of enzymes: endoglucanase, exoglucanase, and -glucosidase (32). The requirement for the addition of -glucosidase has been eliminated by the functional insertion of genes for cellobiose utilization from a related organism, Klebsiella oxytoca M5A1. These K. oxytoca genes encode a phosphoenolpyruvate-dependent phosphotransferase system and phosphocellobiase (22).

Further improvement of cellulose utilization will require extracellular secretion or release of endoglucanase and exoglucanase activities. E. coli is very limited in its ability to secrete proteins into the extracellular environment (34). Recombinant proteins such as endoglucanases, which are secreted by their source organisms, are typically accumulated in the periplasmic space in E. coli (11, 26, 47). In related gram-negative bacteria, however, at least three different types of protein secretion systems have been described (type I, type II, and type III) (25, 34). Considerable progress has been made in our understanding of the type II secretion system (5, 14, 15, 23-25, 28, 34, 37), the most widely used system for protein secretion. Type II secretion involves a two-step process in which a pro-protein containing an N-terminal signal peptide is exported to the periplasm and processed into a mature enzyme by using a highly conserved Sec pathway. The mature enzyme is then secreted into the extracellular environment through the peptidoglycan and outer membrane. Seven type II secretion systems have been cloned from gram-negative bacteria (2, 15, 24, 28, 34, 35, 38, 41), and all contain 12 to 15 gene products which share homology. Although heterologous proteins are often exported to the periplasm by the Sec system, the genes which complete extracellular secretion are highly species specific. This is true even for closely related enteric bacteria such as Erwinia chrysanthemi and E. carotovora (15, 24, 35). Two type II secretion systems, encoded by pul genes from K. oxytoca and the out genes from E. chrysanthemi, have been reconstituted in K-12 strains of E. coli (15, 34). The E. chrysanthemi EC16 out gene products have been shown to secrete EC16 pectate lyase in recombinant E. coli (15). In EC16, this system also appears to facilitate the secretion of the endoglucanase EGZ (3).

Our laboratory has previously cloned celZ from E. chrysanthemi P86021 (4) and demonstrated that the encoded enzyme (EGZ) can function with commercial fungal cellulase to improve ethanol production (47). In these studies, EGZ was accumulated in the E. coli B periplasm and required cell disruption (solvents, detergent, or heat) for release. Total EGZ production was approximately 3 IU per ml of broth.

In this study, we have identified a strong promoter to enhance the expression of celZ endoglucanase in E. coli B and demonstrated that this enzyme can be effectively secreted (or released into the environment) by using the out genes from E. chrysanthemi EC16, thus eliminating the need for cell disruption.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Organisms and culture conditions. Bacterial strains and plasmids used in this study are listed in Table 1. E. coli DH5 was used as the host for plasmid construction. The celZ gene was previously cloned in our laboratory from E. chrysanthemi P86021 (4, 47). The out genes were cloned by He et al. (15) from E. chrysanthemi EC16.

Genetic methods. E. chrysanthemi out genes (pCPP2006) were conjugated into E. coli by using pRK2013 for mobilization (10, 28). Standard methods were used for all plasmid constructions (1, 40). Small-scale plasmid isolation was performed by the TELT procedure (1). Large-scale plasmid isolation was performed with a Promega Wizard Kit. DNA fragments were isolated from gels with a Qiaquick Gel Extraction Kit from Qiagen. E. coli and Z. mobilis chromosomal DNAs were isolated as described by Cutting and Horn (8) and Yomano et al. (49), respectively. Purified chromosomal DNA from E. coli DH5 served as a template for the amplification of two glycolytic gene promoters by using the following primer pairs: gap promoter, 5'-CGAATTCCTGCCGAAGTTTATTAGCCA-3' and 5'-AAGGATCCTTCCACCAGCTATTTGTTAGTGA-3'; and eno promoter, 5'-AGAATTCTGCCAGTTGGTTGACGATAG-3' and 5'-CAGGATCCCCTCAAGTCACTAGTTAAACTG-3'.

Primer extension analysis of transcriptional initiation. Promoter regions were identified by mapping the transcriptional start sites with a primer within the celZ gene. RNA was isolated from cells in late exponential phase by using a Qiagen RNeasy kit. Cells were treated with 400 µg of lysozyme per ml in TE containing 0.2 M sucrose (25°C, 5 min) prior to lysis. After ethanol precipitation, RNA was dissolved in 20 µl of Promega avian myeloblastosis virus (AMV) reverse transcriptase buffer (50 mM Tris-HCl, pH 8.3; 50 mM KCl; 10 mM MgCl₂; 0.5 mM spermidine; 10 mM dithiothreitol). IRD41-labeled primer (5'-GACTGGATGGTTATCCGAATAAGAGAGAGG-3'; LI-COR, Inc.) was added. The sample was denatured at 80°C for 5 min, annealed at 55°C for 1 h, and purified by alcohol precipitation. Annealed samples were dissolved in 19 µl of AMV reverse transcriptase buffer containing 500 µM deoxynucleoside triphosphates and 10 U of AMV reverse transcriptase and then incubated for extension (1 h at 42°C). Products were treated with 0.5 µg of DNase-free RNase A per ml, precipitated, dissolved in loading buffer, and compared to parallel dideoxy sequences by using the LI-COR model 4000-L DNA sequencer.

Endoglucanase activity. Expression of celZ was initially evaluated by using the Congo red procedure of Wood and Bhat (48). Recombinant clones were transferred to gridded CMC plates and incubated for 18 h at 30°C prior to staining. Endoglucanase-positive clones formed yellow zones on a red background. The diameters of these zones were recorded as a relative measure of celZ expression.

Ultrastructural analysis. For ultrastructural analysis, fresh colonies from Luria agar plates were fixed in 2% glutaraldehyde in 0.2 M sodium cacodylate buffer (pH 7) and then in 1% osmium tetroxide, followed by treatment with 1% uranyl acetate in distilled water. Samples were dehydrated in ethanol and embedded in Spurr's plastic (42). Ultrathin sections were examined with a Zeiss EM-10CA electron microscope.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of a low-copy-number promoter probe vector with celZ as the reporter. To facilitate the isolation of strong promoters, a low-copy-number vector was constructed with a pSC101 replicon and a BamHI site immediately preceding a promoterless celZ gene (pLOI2171). Plasmid pLOI1620 was used as a source of celZ and is a pUC18 derivative with expression from consecutive lac and celZ promoters. The BamHI site in this plasmid was eliminated by digestion and Klenow treatment (pLOI2164). The celZ gene was isolated as a promoterless NdeI fragment after Klenow treatment. The resulting blunt fragment was digested with HindIII to remove downstream DNA and then ligated into pUC19 (HindIII to HincII) to produce pLOI2170. In this plasmid, celZ is oriented opposite to the direction of lacZ transcription and was weakly expressed. The BamHI (amino terminus)-SphI (carboxyl terminus) fragment from pLOI2170 containing celZ was then cloned into the corresponding sites of pST76-K, a low-copy-number vector with a temperature-sensitive replicon, to produce pLOI2171 (Fig. 1). Expression of celZ in this vector was extremely low, facilitating its use as a probe for strong promoters.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Diagram illustrating plasmid pLOI2171, a low-copy-number promoter probe vector Abbreviations: Kan, kanamycin resistance gene; Rep (ts), temperature-sensitive pSC101 replicon.

Expression of celZ from E. coli gap and eno promoters. Chromosomal DNA from DH5 was used as the template to amplify the gap and eno promoter regions by PCR. The resulting fragments of approximately 400 bp each were digested with EcoRI and BamHI and cloned into the corresponding sites in front of a promoterless celZ in pLOI2171 to produce pLOI2174 (gap promoter) and pLOI2175 (eno promoter). As a control, the EcoRI-SphI fragment from pLOI2164 containing the complete celZ gene and native E. chrysanthemi promoter was cloned into the corresponding sites of pST76-K to produce pLOI2173.

Cloning Z. mobilis DNA fragments as promoters for celZ expression. Previously, we have used random fragments of Z. mobilis DNA as an effective source of surrogate promoters for the high-level expression of heterologous genes in E. coli (7, 19). A similar approach was used to find promoters for Erwinia celZ expression. Z. mobilis chromosomal DNA was extensively digested with Sau3A1. The resulting fragments were ligated into pLOI2171 at the BamHI site and transformed into E. coli DH5 to generate a library of potential promoters. Resulting colonies were transferred to gridded CMC plates and stained for endoglucanase activity after incubation (Table 2). Approximately 20% of the 18,000 clones tested were CMC positive. The 75 clones which produced larger zones than the control, pLOI2173, were examined further by using E. coli B. In general, recombinants of DH5 produced larger zones than recombinants of strain B. However, the relative promoter strength in each host was similar for most clones. Based on zone size with CMC plates, four clones appeared to express celZ at approximately 6-fold-higher levels than the construct with the original E. chrysanthemi celZ gene (pLOI2173) and at 10-fold-higher levels than the E. coli glycolytic promoters.

Production and apparent secretion of endoglucanase. Eight plasmid derivatives of pST76-K (pLOI2177 to pLOI2184) were selected from group I and group II (Table 2) and assayed for total endoglucanase activity in strain B (Table 3). The four plasmids with the largest zones on CMC plates also had the highest endoglucanase activities (pLOI2177, pLOI2180, pLOI2182, and pLOI2183). The activities were approximately sixfold higher than that of the unmodified celZ (pLOI2173), which is in excellent agreement with our estimate by using the square of the radius of the cleared zone on CMC plates. Figure 2 shows a comparison of activity estimates from plates and in vitro enzyme assays for strain B containing a variety of different promoters, with and without the out genes. Although there is some scatter, a direct relationship is clearly evident, which validates the plate method for estimating relative activity. The original construct in pUC18, a high-copy-number plasmid, was also included for comparison (pLOI2164). This construct with consecutive lac and celZ promoters produced less EGZ activity than three of the low-copy-number plasmids with surrogate promoters (pLOI2177, pLOI2182, and pLOI2183). The DNA fragment containing celZ and the most effective surrogate promoter was isolated from pLOI2183 (EcoRI-SphI) and inserted into pUC19 with transcription oriented opposite to that of the lac promoter (pLOI2307). With this high-copy-number plasmid, endoglucanase activity was further increased twofold.

View larger version (21K): [in this window] [in a new window]   FIG. 2.   Relationship between cleared zones on CMC indicator plates and endoglucanase activity in disrupted cells.

Sequence of Z. mobilis fragment with strong promoter activity. The sequences of the four strongest surrogate promoters (pLOI2177, pLOI2180, pLOI2182, and pLOI2183) were determined. To facilitate this process, each was fused with pUC19 at the PstI site. The resulting plasmids, pLOI2196, pLOI2197, pLOI2198, and pLOI2199, were produced at high copy numbers (ColE1 replicon) and could be sequenced in both directions by using M13 and T7 sequencing primers. All four plasmids contained identical pieces of Z. mobilis DNA and were siblings. Each was 1,417 bp in length and contained four internal Sau3A1 sites. DNA and translated protein sequences (six reading frames) of each piece were compared to the current database. Only one fragment exhibited a strong match in BLAST searches (281-bp internal fragment); it was 99% identical in DNA sequence to part of the Z. mobilis hpnB gene which is proposed to function in cell envelope biosynthesis (39). Primer extension analysis revealed both a single major start site, 67 bp upstream from the Sau3A1/BamHI junction site with celZ, and a second minor start site farther upstream (Fig. 3). Sequences in the 10 and 35 regions were compared to the conserved sequences for E. coli sigma factors (45, 46). The dominant promoter region (ca. 85% of the total start site) appears similar to a ⁷⁰ promoter, whereas the secondary promoter site resembles a ³⁸ promoter.

View larger version (34K): [in this window] [in a new window]   FIG. 3.   Partial nucleotide sequence of the Z. mobilis DNA fragment in pLOI2183 which served as a surrogate promoter. The full sequence has been assigned GenBank accession number AF109242. Sequence numbering began with the base adjacent to the upstream Sau3A1 site (GATC). Two transcriptional start sites were identified by primer extension analysis and are marked with a "#" symbol. Putative 35 and 10 regions are marked by double and single underlines, respectively. Vector and celZ DNA are shown as lowercase letters. The Shine-Dalgarno region and the celZ translational start are marked in boldface.

Periplasmic inclusion bodies. Little difference in cell morphology was observed between recombinants and the parental organism by light microscopy. Under the electron microscope, however, small polar inclusion bodies were clearly evident in the periplasm of strain B(pLOI2164) and are presumed to contain EGZ (Fig. 4). With strain B(pLOI2307), which produced twofold-higher activity, these polar bodies were very large and occupied up to 20% of the total cell volume. The large size of these polar bodies suggests that activity measurements may underestimate total EGZ production. Typically, polar inclusion bodies were smaller in constructs which also contained the out genes (not shown). No periplasmic inclusion bodies were evident in strain B(pUC19), which served as a control.

View larger version (172K): [in this window] [in a new window]   FIG. 4.   Electron micrographs of E. coli B cells harboring different plasmids. (A) pUC19 containing no visible periplasmic inclusion body. (B) pLOI2164 containing small periplasmic inclusion bodies. (C) pLOI2307 containing large periplasmic inclusion bodies. Abbreviations: im, inner membrane; pib, periplasmic inclusion body. Bars, 0.1 µm.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Despite extensive investigation of technology for the bioconversion of lignocellulose to fuel ethanol, the cost of cellulase is still one of the major barriers for commercialization (13, 16, 18, 29, 32). The development of inexpensive enzymatic methods for cellulose hydrolysis has great potential for improving the efficiency of substrate utilization and the economics of the process. Prior to fermentation, insoluble substrates such as cellulose must be converted into a soluble form which can be transported and metabolized. Fungal cellulase is commercially produced on a large scale and is used in food processing, detergents, and pulping (32). Although industrial experience with these enzymes makes it likely that they will also be used in cellulose-to-ethanol processes, it should be possible to develop ethanologenic biocatalysts that provide at least part of the required cellulase activities as a recombinant coproduct with ethanol. Ethanologenic derivatives of E. coli B have been previously engineered to metabolize cellobiose and cellotriose by inserting genes encoding the phosphoenolpyruvate-dependent phosphotransferase system and phospho--glucosidase from K. oxytoca M5A1 (22), eliminating a need for the least-stable component of fungal cellulase, -glucosidase (27). However, enzymes which solubilize polymeric cellulose must be present in the extracellular environment to be effective.

Previous studies have identified E. chrysanthemi EGZ as an effective endoglucanase which can be used in combination with commercial fungal cellulase to improve ethanol production from cellulose (47). The gene encoding this enzyme was expressed well (ca. 3,000 IU per liter), but higher levels of expression would be desirable. The native E. chrysanthemi promoter was more effective in E. coli than either of two E. coli glycolytic promoters (gap and eno). However, Z. mobilis DNA served as an excellent source of surrogate promoters. Sau3A1 fragments with strong promoter activity were identified which provided a sixfold increase in EGZ production in a low-copy-number vector and a twofold increase in a high-copy-number vector. In both cases, most of the activity was cell associated (periplasmic, intracellular, or bound). Relative activity correlated with the size of periplasmic inclusion bodies. Based on activity, EGZ is calculated to represent 4 to 6% of the total cellular protein in the most active construct. However, this may be an underestimate of total EGZ protein since inclusion bodies represented approximately 20% of the total cell volume. It is possible that some of the proteins retained in the inclusion bodies are misfolded or partially degraded.

Native EGZ was secreted or released from the periplasm to the extracellular environment by the out gene products in E. chrysanthemi (21), and the out gene products have been previously demonstrated to secrete recombinant E. chrysanthemi pectate lyases in E. coli (15). Addition of the out genes increased the EGZ activity produced in all E. coli recombinants up to twofold and facilitated the apparent secretion of approximately one-half of the activity into the extracellular environment. The increase in EGZ activity could result from improved folding or reduced degradation. At this time, the basis for incomplete secretion or release remains unknown but may be growth associated. Preliminary experiments revealed that over 90% of the EGZ activity was extracellular in early exponential phase. The fraction of extracellular activity slowly declined and continued to decline during stationary phase, although total activity continued to increase. It is possible that the stability or synthesis of the Out proteins or Sec proteins limited EGZ export during the late-exponential and stationary phases, while the unregulated promoters driving EGZ production continued to function.

The ability of these E. coli strains to secrete or release high levels of EGZ demonstrates an approach which can be used to produce EGZ as a coproduct with ethanol in ethanologenic strains of E. coli B and eliminates the need for additional process steps for the release of endoglucanase activity (47). Although this approach may not be readily used with genes from other hosts, it does demonstrate the potential for E. coli to be further developed as a host for recombinant protein secretion. Considerable progress has been made in defining the basis for protein recognition and secretion in the analogous systems from seven different organisms (2, 5, 14, 15, 23-25, 28, 34, 35, 37, 38, 41). Structural information for extracellular secretion by type II systems does not appear to be localized within a particular region and may involve the entire structure of the secreted protein (14). Recognition of target proteins for secretion appears to reside in the outC and outD gene products (24). With continued progress in defining this relationship it should be possible to secrete large amounts of recombinant products for biotechnology applications by using E. coli as a host.