|Home | About | Journals | Submit | Contact Us | Français|
The Bacillus megaterium protein production system based on the inducible promoter of the xyl operon (PxylA) was systematically optimized. Multiple changes in basic promoter elements, such as the −10 and −35 region and the ribosome-binding site, resulted in an 18-fold increase of protein production compared to the production of the previously established system. The production in shaking-flask culture of green fluorescent protein (Gfp) as a model product led to 82.5 mg per g cell dry weight (gCDW) or 124 mg liter−1. In fed-batch cultivation, the volumetric protein yield was increased 10-fold to 1.25 g liter−1, corresponding to 36.8 mg protein per gCDW. Furthermore, novel signal peptides for Sec-dependent protein secretion were predicted in silico using the B. megaterium genome. Subsequently, leader peptides of Vpr, NprM, YngK, YocH, and a computationally designed artificial peptide were analyzed experimentally for their potential to facilitate the secretion of the heterologous model protein Thermobifida fusca hydrolase (Tfh). The best extracellular protein production, 5,000 to 6,200 U liter−1 (5.3 to 6.6 mg liter−1), was observed for strains where the Tfh export was facilitated by a codon-optimized leader peptide of YngK and by the signal peptide of YocH. Further increases in extracellular protein production were achieved when leader peptides were used in combination with the optimized expression system. In this case, the greatest extracellular enzyme amount of 7,200 U liter−1, 7.7 mg liter−1, was achieved by YocH leader peptide-mediated protein export. Nevertheless, the observed principal limitations in protein export might be related to components of the Sec-dependent protein transport system.
For economic production of recombinant proteins, bacterial hosts are usually the systems of choice. Due to the well-documented limitations in the application of the most prominent production host, Escherichia coli, for extracellular and high-molecular-mass protein formation, alternative production hosts, such as Bacillus megaterium, have gained growing interest (26, 33). Similar to E. coli, B. megaterium is able to utilize a wide variety of carbon sources, which allows for its growth on low-cost substances (31); however, the Gram-positive B. megaterium does not produce the endotoxins associated with the outer membrane of Gram-negative E. coli. Furthermore, the bacterium is known for its high protein secretion potential. Secretion of recombinant proteins into the growth medium reduces the efforts and costs of protein purification. In comparison to industrially employed Bacillus subtilis strains, B. megaterium strains have the advantage of highly stable, freely replicating plasmids and the lack of alkaline proteases (30).
In addition to the recently described T7 RNA polymerase-dependent protein production system (13), sucrose- and xylose-inducible promoter systems were successfully applied for heterologous protein production in B. megaterium (4, 22, 23). The most commonly employed gene expression system is based on the xylose-inducible PxylA promoter of the B. megaterium xyl operon, usually localized on a freely replicating plasmid. Repression of promoter activity in the absence of the inducer xylose is mediated by the repressor protein XylR, which is provided by a plasmid-encoded copy of the xylR gene. The first PxylA-based gene expression shuttle vector, pWH1520, already comprised all necessary elements for xylose-inducible gene expression in B. megaterium, as well as for replication and antibiotic-based selection in both B. megaterium and E. coli (22). This plasmid was the starting point for the construction of multiple new vectors representing a genetic toolbox ensuring its broad applicability. In this context, a large multiple cloning site was introduced and obsolete parts of the plasmid were eliminated. DNA sequences encoding C- or N-terminal affinity tags were added (5, 16). Moreover, various plasmids for Sec-dependent secretion of recombinant, heterologous proteins were constructed (3, 17). In general, exoproteins which are secreted via the Sec pathway are synthesized as protein precursors with N-terminal leader sequences. These signal peptides (SPs) facilitate the translocation of the unfolded preproteins through the SecYEG channel in the cytoplasmic membrane. After the passage through the membrane, the SPs are cut off by membrane-anchored signal peptidases and the mature proteins are released into the extracellular space. There, the mature proteins are finally folded and diffuse through the cell wall into the environment (28, 29).
The plasmidless B. megaterium strains MS941 and YYBm1 were generated from the wild-type strain DSM319 by directed gene deletion. Knockout of the major extracellular protease gene nprM resulted in B. megaterium MS941. This strain only shows 1.5% of the wild-type extracellular protease activity and thus is well suited for extracellular protein production (32). Inactivation of the xylA gene for xylose metabolism in B. megaterium MS941 led to the strain YYBm1, which does not metabolize the inducer of gene activation (34).
The above-described xylose-inducible protein production system for B. megaterium has proven to be a useful tool for several applications (31). Nevertheless, compared to the production levels of the most widely used bacterial host, E. coli, the amount of intracellularly produced heterologous protein in B. megaterium was lower. For example, Biedendieck et al. (5) produced 5.2 mg per g cell dry weight (gCDW) in a fed-batch cultivation employing the xylose-inducible protein production system in B. megaterium. In comparison, Dürrschmid et al. (10) produced 73.7 mg green fluorescent protein (Gfp) per gCDW by using E. coli with a T7 RNA polymerase-dependent gene expression system in a comparable cultivation approach. Consequently, in this work, a directed systematic optimization of the xylose-inducible protein production and export system was carried out in order to reach the protein production efficiency of E. coli. First, different plasmid-encoded signals directing the protein production process at different stages, including transcription and translation, were systematically optimized. The impacts of the genetic optimizations on heterologous intracellular protein production were quantified using an enhanced Gfp variant as the model protein. Based on this optimized system, the Sec pathway for protein export was also targeted for further improvements. Until now, only two different signal peptides (SPLipA and SPPac) were employed to facilitate the export of heterologous target proteins in B. megaterium (3). In this work, new B. megaterium SPs were screened and experimentally evaluated for their capability to promote protein secretion of the heterologous model hydrolase from Thermobifida fusca (Tfh).
The molecular biology methods used were outlined previously (24). The synthetic oligonucleotides used in this work can be found in Table S1 in the supplemental material. The plasmids constructed or employed are listed in Table Table11 . E. coli strain DH10B (Invitrogen, San Diego, CA) was used for all cloning purposes.
The basic expression plasmid of this investigation, p3STOP1622, containing one stop codon in each possible reading frame downstream from its multiple cloning site (mcs), was constructed by site-directed mutagenesis (QuikChange II; Agilent Technologies, Santa Clara, CA) of pSTOP1622 (5). The oligonucleotides QC-p3STOP1622-for and QC-p3STOP1622-rev were used as primers for the PCR to introduce two additional stop codons downstream from the mcs into pSTOP1622. Starting from the resulting p3STOP1622, two unique restriction sites were introduced to allow for simple genetic modification of the DNA region between PxylA and the mcs. For this purpose, a PacI site was inserted between the −10 and −35 regions of PxylA by site-directed mutagenesis using the primer pair QC-p3STOP1623-for and QC-p3STOP1623-rev, resulting in p3STOP1623. Afterwards, an NheI site was introduced by replacing the DNA fragment between the PacI and BsrGI restriction sites with the synthetic oligonucleotide pair p3STOP1624-for and p3STOP1624-rev, resulting in p3STOP1624. The gfp gene was amplified by PCR from pRBBm34 using the primers gfp-for and gfp-rev (5). The gene named gfp used in this study differed from the egfp gene employed by Biedendieck et al. (5) by a single base exchange which leads to an amino acid exchange at the C terminus of the protein, without obvious influence on the protein function. Insertion of gfp into the mcs of p3STOP1622, p3STOP1623, and p3STOP1624 via the corresponding BglII and EagI restriction sites led to p3STOP1622-gfp, p3STOP1623-gfp, and p3STOP1624-gfp, respectively.
The oligonucleotide pairs −10+-for and −10+-rev, utr+-for and utr+-rev, and rbs+-for and rbs+-rev carrying optimized genetic elements for transcription and translation were introduced into p3STOP1624 after PacI-NheI or NheI-BsrGI digestion. The resulting plasmids were named pKMBm1 (optimized −10 region [−10+]), pSSBm40 (optimized utr [utr+]), and pSSBm44 (optimized rbs [rbs+]). The gfp gene was introduced into these plasmids as described above, resulting in pKMBm9, pSSBm46, and pSSBm50. The plasmids pSSBm84 and pSSBm78, with an optimized −35 region (−35+) for the PxylA promoter, were generated by site-directed mutagenesis. For the necessary PCRs, the primers QC-35+-for and QC-35+-rev were used in combination with pSSBm39 and pSSBm50 as the template. Fusion of the XhoI-NheI fragment from pKMBm1 harboring an optimized −10 region for the PxylA promoter with the vector pSSBm44 containing an optimized ribosome binding site resulted in plasmid pSSBm74. Ligation of the oligonucleotides utr+-for and utr+-rev with the NheI- and BsrGI-cut pSSBm44 vector resulted in pKMBm4. The gfp gene was inserted into mcs of both plasmid pSSBm74 and pKMBm4 as described above, resulting in pSSBm76 and pKMBm10. Fusion of the AatII-PacI fragments derived from pSSBm78 and from pSSBm76 resulted in the plasmid pSSBm81, which contains both optimized promoter elements (−10+ and −35+) in combination with the optimized rbs and the gfp gene. Site-directed mutagenesis of pSSBm78 using the primer pair QC-ΔNheI-for and QC-ΔNheI-rev led to the NheI-lacking plasmid pSSBm85.
The mcs- and affinity tag-encoding regions were subcloned from pC-HIS1622, pN-HIS-TEV1622 (5), and p3STOP1624 into pSSBm85 using the restriction sites BsrGI and PstI, thereby generating the high-performance (hp) plasmids pC-HIS1623hp, pN-HIS-TEV1623hp, and p3STOP1623hp.
Four signal peptide (SP) coding regions for SPYngK, SPVpr, SPNprM, and SPYocH were amplified by PCR from genomic B. megaterium DSM319 DNA using the primer pairs spyngK-for and spyngK-rev, spvpr-for and spvpr-rev, spnprM-for and spnprM-rev, and spyocH-for and spyocH-rev, respectively. The resulting PCR fragments and the annealed oligonucleotides sp+yngK-for and sp+yngK-rev, which encode a codon-adapted SPYngK, were inserted into pMM1525 via the EagI and SpeI sites, resulting in the plasmids pSSBm22, pSSBm23, pSSBm24, pSSBm25, and pSSBm27. Using vector pYYBm9 as the template, the codon-optimized gene of a T. fusca hydrolase (Tfh) fused to a His6 tag was amplified via PCR using the primers tfh-for and tfh-rev. Plasmids pSSBm22, pSSBm23, pSSBm24, pSSBm25, and pSSBm27 were cut with EagI and BglII and subsequently ligated with the amplified tfh-his6. The resulting vectors were named pSSBm28, pSSBm29, pSSBm30, pSSBm31, and pSSBm33. DNA fragments containing coding sequences for either an artificial signal peptide (SPAsp) or the signal peptide from B. megaterium penicillin amidase (SPPac) were amplified by PCR using the templates pADBm20 and pRBBm26 and the primer pairs spasp/pac-for and spasp-rev and spasp/pac-for and sppac-rev, respectively. Both PCR products were inserted into pSSBm28 via the restriction sites SpeI and XhoI, creating the plasmids pSSBm34 and pSSBm35. The DNA regions encoding the different SP-Tfh-His6 fusions were subcloned from the various basic expression plasmids into the high-performance plasmid p3STOP1623hp utilizing the restriction sites BsrGI and PstI, thereby generating the plasmid series pSSBm94 to pSSBm101.
The insertion of the mcs- and His6 tag-coding DNA region derived from pC-HIS-1623hp into the vectors pSSBm94 to pSSBm100 via the SpeI-PstI restriction sites finally resulted in new high-performance plasmids for extracellular protein production. According to the encoded signal peptides, these plasmids were named pSPYngK-hp, pSPVpr-hp, pSPNprM-hp, pSPYocH-hp, pSP+YngK-hp, pSPAsp-hp, and pSPPac-hp. Plasmid pSPLipA-hp was constructed by the insertion of splipA-mcs of pHIS1525 (16) into the high-performance plasmid derivate pSSBm101. The desired structures of all constructed plasmids were verified by DNA sequence analysis.
The B. megaterium strain MS941, an nprM deletion mutant of DSM319, was used as protein production host for all shaking-flask experiments (32). Protoplasted B. megaterium cells were transformed with the appropriate expression plasmids using a polyethylene glycol-mediated procedure described before (1). All B. megaterium plasmid strains were grown in baffled shake flasks at 37°C in Luria-Bertani (LB) medium (24). The medium was supplemented with tetracycline to a final concentration of 10 mg liter−1 to sustain the selective pressure on the stable replication of the corresponding plasmids. Recombinant expression of genes under transcriptional control of the xylose-inducible promoter was induced by the addition of 0.5% (wt/vol) xylose at an optical density measured at 578 nm (OD578) of 0.4. Samples were taken at time points indicated in the corresponding figure legends after the induction of heterologous gene expression. Cells were separated from the growth medium by centrifugation (14,000 × g). If necessary, the precipitated cells were stored at −20°C and the cell-free supernatant at 4°C prior to analysis.
A RALF Plus 3.7-liter bioreactor (Bioengineering, Wald, Switzerland) was used to operate a high-cell-density cultivation process in fed-batch mode. For the initial batch phase, the bioreactor was prepared with 1 liter of a minimal medium [3.52 g liter−1 KH2PO4, 6.62 g liter−1 Na2HPO4·2 H2O, 0.3 g liter−1 MgSO4·7 H2O, 25 g liter−1 (NH4)2SO4, 15 g liter−1 fructose, 80 mg liter−1 MnCl2·4 H2O, 106 mg liter−1 CaCl2·2 H2O, 5 mg liter−1 FeSO4·7 H2O, 4 mg liter−1 (NH4)6Mo7O24·4 H2O, 2.2 mg liter−1 CoCl2]. Subsequently, the medium was supplemented with 10 mg liter−1 tetracycline. The bioreactor was inoculated with cells of a starter culture of B. megaterium YYBm1 carrying pSSBm85 [PxylA-(−35+ rbs+)-gfp] to a final OD578 of 0.1. The cultivation was controlled to pH 7.0 and 37°C. At the end of the batch phase, Gfp production was induced by the addition of 7 g liter−1 xylose and a dissolved-oxygen (DO) controlled feeding profile was applied. The feed contained 150 g liter−1 fructose in a buffer solution containing 9.9 g liter−1 KH2PO4, 14.98 g liter−1 Na2HPO4, 0.3 g liter−1 MgSO4·7 H2O, 25 g liter−1 (NH4)2SO4, 40 mg liter−1 MnCl2·4 H2O, 53 mg liter−1 CaCl2·2 H2O, 2.5 mg liter−1 FeSO4·7 H2O, 2 mg liter−1 (NH4)6Mo7O24·4 H2O, and 1.1 mg liter−1 CoCl2. Additionally, it was supplemented with 5 g liter−1 xylose and 10 mg liter−1 tetracycline.
Samples were taken for biomass, xylose, fructose, and Gfp concentration measurements. Xylose and fructose concentrations were analyzed by high-pressure liquid chromatography (HPLC) analysis (Hitachi, Tokyo, Japan) on a Metacarb 87C column (Bio-Rad, München, Germany). Ultra-pure H2O at a flow rate of 0.6 ml min−1 (85°C) was used as the mobile phase. The amount of Gfp was determined by its fluorescence as described below.
To analyze the composition of the intracellular proteins, precipitated bacterial cells were enzymatically disrupted with lysozyme (17). DNA was degraded by the addition of Benzonase (Merck, Darmstadt, Germany) to the lysis buffer. After incubation, soluble proteins were separated from insoluble ones and cell debris by centrifugation (14,000 × g, 4°C). Soluble proteins of 5 × 108 cells were separated via 12% SDS-PAGE gels and visualized by Coomassie brilliant blue staining. Proteins in the cell-free growth medium were precipitated using 70% (wt/vol) ammonium sulfate (12). The solution was mixed gently for 2 h at 4°C, and the precipitated proteins were collected by centrifugation (14,000 × g, 30 min, 4°C). After removal of the supernatant, the proteins were suspended in 50 mM Tris-HCl buffer (pH 7.5) containing 8 M urea. Proteins equivalent to 1.5 ml of cell-free culture medium were subsequently separated via 12% SDS-PAGE.
Recombinant Gfp was quantified via fluorescence spectroscopy measurements using an LS50B luminescence spectrometer (PerkinElmer, Boston, MA) and a fluorescence cuvette (type 104.002F-QS; Helma, Müllheim, Germany). For this purpose, B. megaterium cells were harvested by centrifugation (14,000 × g). The resulting cell pellet was suspended in 100 mM sodium phosphate buffer (pH 7.0) to a final concentration of 1 × 109 cells per ml. One-hundred-microliter amounts of these samples were added to 900 μl buffer and mixed well. These samples, containing 1 × 108 cells ml−1, were excited using a wavelength of 475 nm, while fluorescence emission was recorded at 512 nm. The relative levels of fluorescence of different amounts of purified Gfp were determined and quantified using the following linear correlation: Gfp (mg ml−1) = relative emission maxima × (3.42 × 10−6) × dilution factor. Since Biedendieck et al. (5) established that 0.334 g liter−1 cell dry weight (CDW) equals an OD578 of 1 for B. megaterium MS941, the Gfp amount could also be specified in mg per gCDW. In this study, this equation, originally obtained for purified Gfp, was applied to estimate Gfp amounts inside bacterial cells. Due to additional light absorption and scattering by the cells, a slight underestimation of the amount of Gfp produced has to be taken into account.
The hydrolase activity of the enzyme Tfh was measured by detection of the hydrolysis product of p-nitrophenylpalmitate (pNPP) as described by Dresler et al. (9). For this purpose, 40 μl of cell-free culture supernatant was added to 960 μl of freshly prepared pNPP solution. The enzyme activity assay was carried out at 30°C. The enzymatic release of p-nitrophenol was photometrically detected at 400 nm for 90 s. One enzyme unit was defined as the amount that caused the release of 1 μmol p-nitrophenol per minute under the given assay conditions. The extinction coefficient of p-nitrophenol is 9.62 cm2 μmol−1. For the calculation of hydrolase amounts (mg liter−1), the specific activity value for purified enzyme determined by Yang et al. (35) and a temperature correlation factor were employed.
The major goal of this investigation was the significant enhancement of recombinant intra- and extracellular protein production by B. megaterium. Based on a previously established protein production system, various directed optimization strategies were followed.
The quantity of an intracellular protein formed by a microbial cell is mainly determined by several closely interconnected steps: transcription, mRNA stability, and translation. Consequently, promoter sequences, mRNA signatures, and ribosome-binding sites of the plasmid-borne, xylose-inducible protein production system of B. megaterium were the targets of directed genetic optimization approaches. In order to make the basic plasmid, p3STOP1622, susceptible to multiple genetic optimizations, two new unique restriction sites were introduced between the promoter PxylR and the translational start site located upstream of the multiple cloning site (mcs) to allow for a cassette exchange strategy. Since DNA changes in this region might affect the production of a target protein, both sites were chosen to cause only minimal changes in the native DNA sequence. By the introduction of a PacI restriction site between the −35 and −10 region of the promoter PxylA, the plasmid p3STOP1623 was created. Subsequently, an additional NheI recognition site was inserted downstream of the PxylA transcriptional start site into the plasmid p3STOP1623, resulting in p3STOP1624. The effects of these DNA modifications on protein production in B. megaterium were evaluated. For this purpose, these plasmids and the starting plasmid p3STOP1622 were equipped with gfp as a model gene. Gfp is an excellent reporter protein since it is not harmful for the cells and is independent of substrates and cofactors, readily measurable, and provides high sensitivity (20, 27). B. megaterium MS941 cells were transformed with each of these plasmids individually and Gfp production experiments were carried out.
The results of these experiments showed that both restriction sites have some impact on target protein formation by B. megaterium. The introduction of the PacI recognition site caused an increase in Gfp production of approximately 50%, from 4.6 mg per gCDW to 6.8 mg per gCDW, while the additional insertion of the NheI site reduced the Gfp production by 40%, back to 4.2 mg g−1. In total, only a minor response was observed and p3STOP1624 harboring both new restriction sites was used as the basis for further genetic optimizations.
Based on p3STOP1624, different stages of the protein formation process were the targets of genetic optimizations (Fig. (Fig.1).1). The −35 (TTGAAA) and the −10 region (TATGAT) of PxylA were exchanged into their predicted optimal counterparts, termed −35+ (TTGACA) and −10+ (TATAAT), respectively. This usually leads to increased RNA polymerase binding and higher transcription initiation frequency and, thereby, to an acceleration of the whole transcriptional process (7, 19). The DNA region surrounding the palindromic XylR-binding motif termed utr (AGTTAGTTTATTGGATAAACAAACTAACT) was modified to utr+ (GGAATTGTAGTTAGTTTACAATTCCAACAAACTAACT) (the DNA sequence in italics remained unchanged, while the underlined regions were inserted into or replaced the native DNA sequence) for an increased mRNA half-life time via the formation of a hairpin loop in the 5′ untranslated RNA region (UTR) of the target protein's mRNA. An optimized ribosome-binding site (Shine-Dalgarno sequence) for B. megaterium (AAGGAGGTGA) was already developed and successfully employed for recombinant production of DsrS (17). This adapted ribosome-binding site (rbs+) is highly complementary to the 3′ end of 16S rRNA of the B. megaterium ribosome. It increases the ribosome's affinity to the mRNA of interest and thus enhances the translation process. After exchanging each of the above-mentioned genetic elements of p3STOP1624 into their improved counterparts individually, the plasmids were equipped with gfp as a reporter gene. B. megaterium cells were transformed with the different plasmids, and protein production experiments were carried out. Cells carrying the basic plasmid p3STOP1624-gfp were used as the reference.
When tested separately, all improved elements showed the expected enhancing effect on Gfp production in B. megaterium (Fig. (Fig.2).2). The Gfp amounts visualized by SDS-PAGE correlated well with the measured Gfp fluorescence values (compare Fig. 2a and b). Obviously, the Gfp was properly folded and active. Consequently, the overall Gfp production can reliably be quantified by fluorescence.
B. megaterium cells carrying pKMBm9 (−10+) contained a nearly 3-fold-greater amount of Gfp (20.4 mg per gCDW) than the reference cells (7.0 mg per gCDW). The optimized −35 promoter sequence (−35+) revealed the greatest positive impact on protein formation. Compared to the amount in the reference, the Gfp quantity was increased 11-fold, to 76.5 mg per gCDW. The modified element causing a hairpin loop in the 5′ UTR of the target mRNA (utr+) enhanced Gfp production 3.8-fold (26.6 mg per gCDW). The positive impact of an optimized ribosome-binding site on protein formation originally described by Malten et al. (17) was clearly confirmed for the production of Gfp. B. megaterium cells harboring pSSBm50 encoding this optimized Shine-Dalgarno sequence (rbs+) contained double the amount of Gfp (14.8 mg per gCDW) in the reference cells.
A combination of the genetic elements enhancing target protein production was expected to cause even greater protein yields. Therefore, new production plasmids were constructed combining two or more of the optimized DNA elements. First of all, the translation-enhancing Shine-Dalgarno sequence rbs+ was combined with each of the other optimized sequences. Protein production experiments with Gfp as the model protein were performed in B. megaterium. Cells harboring the plasmids pSSBm76 (−10+ rbs+) and pKMBm10 (utr+ rbs+) showed dramatically increased intracellular Gfp accumulations of 62.9 mg per gCDW and 75.0 mg per gCDW, respectively (Fig. (Fig.3).3). This increase is significantly larger than the sum of the protein amounts produced by the strains carrying the plasmids with only a single modification (−10+, 20.4 mg per gCDW; utr+, 26.6 mg per gCDW; rbs+, 14.8 mg per gCDW). In contrast, the combination of the optimized −35 promoter element (−35+) and the adapted ribosome-binding site (rbs+) did not show this strong effect. Cells harboring the plasmid pSSBm78 (−35+ rbs+) showed only a slight increase in Gfp content (80.6 mg per gCDW) over that of cells employing pSSBm84 (−35+) for protein production (76.5 mg per gCDW). Analogously, the combination of both consensus promoter elements (−35+ −10+) with the optimized Shine-Dalgarno sequence (rbs+) did not yield the desired improvement in protein formation (16.7 mg per gCDW). In fact, a strong decrease in Gfp production was detected in comparison with that in cells employing either pSSBm76 (−10+ rbs+) or pSSBm78 (−35+ rbs+), which encode a less conserved promoter.
Based on these observations, an optimized expression plasmid was designed. Starting with the plasmid containing the combination of genetic elements (−35+ and rbs+) facilitating the highest Gfp production (80.6 mg per gCDW), the NheI site was changed back into its native DNA sequence, since the introduction of this unique restriction site revealed a negative impact on protein production in B. megaterium. B. megaterium cells transformed with pSSBm85 displayed slightly increased Gfp production (82.5 mg per gCDW).
Next, the combination of the optimized elements −35+ and rbs+ and the deletion of the NheI site were used as the basis for the creation of new high-performance plasmids. For this purpose, combinations with coding sequences for N- or C-terminally localized His6 tags for protein purification by affinity chromatography were generated. An encoded tobacco etch virus (TEV) protease recognition site enables the removal of the N-terminally located His6 tag. The novel high-performance vectors p3STOP1623hp, pC-HIS1623hp, and pN-HIS-TEV1623hp were constructed based on the established parental plasmids pSTOP1622, pC-HIS1622, and pN-HIS-TEV1622 (5). Due to intensive investigation and documentation of the performance of these parental plasmids (5), we refrained from somewhat redundant testing of the novel, fusion tag-encoding high-performance plasmids. The only difference in the high-performance vector equivalent for intracellular Gfp production tested here lies in the N- and C-terminally fused His6 tags.
B. megaterium YYBm1 transformed with pSSBm85 [PxylA-(−35+ rbs+)-gfp] was used for fed-batch cultivations aiming for upscaled protein production (Fig. (Fig.4).4). Strain YYBm1 is a defined xylA and nprM deletion mutant of the wild-type B. megaterium DSM319 (34). The xylA mutation prevents consumption of the gene expression inducer xylose, while NprM represents the major extracellular protease. In high-cell-density cultivations, a minimal medium containing 15 g liter−1 fructose was used for the initial batch phase. The complete consumption of fructose indicated the end of the batch phase and led to a sudden rise of dissolved oxygen (DO). At this time point, approximately 11.5 h after inoculation, heterologous gfp expression was induced by the addition of xylose. At the same time, a DO-controlled feeding profile was started. Gfp production was determined via fluorescence measurements. A volumetric maximum in Gfp concentration of 1.25 g liter−1 was reached at the end of the cultivation (20.75 h). The Gfp yield per cell dry weight rose continuously until the end of the cultivation process, up to 36.8 mg per gCDW.
The signal peptides (SPs) for protein export via the Sec pathway of B. megaterium were computationally predicted using PrediSi software (http://www.predisi.de/) (15). During this process, all open reading frames (ORFs) of the B. megaterium genome database MegaBac (version 2) were analyzed for signal peptide-coding sequences. The ORFs had been previously determined using Glimmer 2 software (http://www.cbcb.umd.edu/software/glimmer/) (8). Most of the predicted B. megaterium signal peptides showed high similarity with the consensus sequence for type I signal peptides described for B. subtilis by Tjalsma et al. (28, 29). They had lengths close to the predicted average of 28 amino acids and showed the typical three domains. The three leader sequences originated from the proteins Vpr, NprM, and YngK showed a high signal peptide probability and were therefore chosen to be examined. These proteins were previously found to be efficiently secreted by B. megaterium (31). However, the original DNA sequence of the signal peptide from YngK (spyngK) showed a poor codon adaptation index (CAI) of 0.33 for B. megaterium when analyzed by JCat software (http://www.jcat.de/) (14). This low CAI indicated the use of codons which are rare in B. megaterium. For production of T. fusca hydrolase (Tfh) in B. megaterium, it was shown that the use of rare codons in a coding sequence may lead to a slowdown of the translational process or to complete abortion of translation (35). Thus, in addition to the native spyngK, a codon-optimized version, sp+yngK (CAI = 0.99), was synthesized de novo and tested. The signal peptide of YocH was chosen from the list of predicted B. megaterium SPs due to its very high similarity to the consensus signal peptide structure for Sec-dependent protein secretion described by Tjalsma et al. (28, 29). So far, the protein YocH has not been detected in any secretome analysis of B. megaterium. In addition to these native B. megaterium SPs, an artificial signal peptide (SPAsp) was constructed. This in silico-designed “optimal” signal peptide was deduced from the average amino acid sequence of all signal peptides currently known for Gram-positive bacteria. A hidden Markov model approach implemented in the program PrediSi (15) was used for the development of this peptide. Based on a systematic database search, the amino acid residue which occurs most frequently at each position was chosen for that position. Two SPs already used in recombinant protein production and secretion in B. megaterium served as positive controls in this work. The signal peptides of penicillin amidase (SPPac) (18) and of the esterase LipA (SPLipA) (21) were already tested to facilitate the secretion of a heterologous levansucrase, LevΔ773, in B. megaterium (3). SPLipA was also successfully employed in B. megaterium to mediate the efficient export of a hydrolase from Tfh (35).
All signal peptides were introduced into both the plasmid containing the native xylose-inducible expression system and the plasmids harboring the improved high-performance expression system (Fig. (Fig.1).1). All plasmids were equipped with a codon-adapted variant of the tfh gene fused to the coding region of a C-terminally located His6 tag (35). For Tfh, a simple and reliably reproducible enzyme activity assay which makes the exoenzyme a viable model for heterologous protein production and secretion in B. megaterium is available (35).
In production experiments, it was shown that all SPs employed facilitate the secretion of active heterologous Tfh into the growth medium (Fig. (Fig.5).5). Culture supernatant of B. megaterium cells transformed with a plasmid without the tfh gene (pMM1525) was used as negative control and did not show any hydrolase activity at all. The Tfh protein amounts quantified via SDS-PAGE gels correlated well with the volumetric hydrolase activities detected, indicating that all secreted enzyme was in its active conformation (Fig. (Fig.55).
Generally, B. megaterium cells employing the high-performance production system (pSSBm94 to pSSBm101) showed increased amounts of Tfh secreted into the growth medium compared to the amounts secreted by cells employing the corresponding basic expression systems (pSSBm28 to pSSBm31, pSSBm33 to pSSBm35, and pYYBm9). Used in combination with the basic expression plasmids, the four signal peptides SPVpr, SPNprM, SPPac, and SPAsp facilitated the secretion of low levels of Tfh (SPVpr, 430 U liter−1; SPNprM, 470 U liter−1; SPPac, 910 U liter−1; and SPAsp, 1,100 U liter−1). The signal peptide SPYngK mediated export of medium quantities of the model enzyme (SPYngK, 2,700 U liter−1), and the three leader sequences SP+YngK, SPYocH, and SPLipA facilitated the transport of high levels of Tfh into the growth medium (SP+YngK, 5,000 U liter−1; SPYocH, 6,200 U liter−1; and SPLipA, 6,200 U liter−1) (Fig. (Fig.5).5). When employing either SPAsp or SPPac in combination with the high-performance expression system in B. megaterium, the amounts of Tfh secreted into the culture supernatants were dramatically increased compared to the amount secreted by the basal system. For SPAsp, the increase was >5-fold (5,800 U liter−1), while for SPPac, even a 6-fold increase (5,500 U liter−1) was detected. Still, a significant increase in exported Tfh was detected in the culture broth of cells which carried the high-performance expression system in combination with the signal peptides SPVpr (1,300 U liter−1), SPNprM (1,500 U liter−1), and SPYngK (5,600 U liter−1). The three leader peptides, which performed well in combination with the basic expression system, facilitated the secretion of comparable Tfh amounts when used in the high-performance system (SP+YngK, 4,900 U liter−1; SPYocH, 7,200 U liter−1; and SPLipA, 5,600 U liter−1).
The established combination of mcs and a coding sequence for a C-terminally located His6 tag of pHIS1525 (16) was inserted into the high-performance production plasmids comprising one of each of the SPs presented here (pSSBm94 to pSSBm101). Thus, a desired target protein can be cloned into these plasmids, generating His6-tagged proteins which can be purified from B. megaterium culture supernatant by metal-affinity chromatography as demonstrated by Malten et al. (16).
In this work, the protein production potential of the xylose-inducible expression system for B. megaterium was systematically improved. Enhanced transcription was achieved by the insertion of consensus promoter elements for the −10 or −35 region. mRNA was improved by modifications which led to a predicted hairpin loop in the 5′ untranslated region of the target gene's mRNA. Furthermore, a ribosome binding site (rbs) adapted to B. megaterium 16S rRNA was applied to accelerate translation. Employed individually, all of these genetic elements yielded positive effects on recombinant protein production in B. megaterium. The combinations of −10+ and rbs+ or utr+ and rbs+ facilitated strong improvements in Gfp formation (62.9 mg per gCDW and 75.0 mg per gCDW). Surprisingly, these levels of production were much greater than the sums of the positive effects of these elements when tested individually. The resulting increases in protein production were nearly multiplications of the fold changes observed for the single optimizations. Obviously, a larger amount or a more stable mRNA represents an improved target for the ribosome, which is attracted by the perfect Shine-Dalgarno sequence. Taken together, this yields a highly synergistic protein synthesis process.
The optimization of both the −35 and −10 region, resulting in a complete consensus promoter for gene expression, was realized in pSSBm81 (−35+ −10+ rbs+). Surprisingly, this led to lower protein yields (16.7 mg per gCDW) and the loss of external promoter control. Ellinger et al. (11) observed RNA polymerase stalling at promoters with very high similarity to the consensus during in vivo experiments in E. coli. Obviously hampered by strong promoter affinity, the rate-limiting step was found to be the release of the sigma factor and, thus, the transition to an elongation complex (11). Here, this stalling effect might have led to reduced transcription rates which consequently resulted in the formation of smaller amounts of Gfp. Under noninducing conditions, detectable amounts of Gfp were found, indicating a loss of promoter controllability (data not shown). Due to the highly increased affinity of the RNA polymerase complex to the consensus promoter (−35+ −10+), the XylR repressor binding might have lost its competitiveness. Since the binding sequence for the XylR repressor is located just a few base pairs downstream of PxylA, it seems probable that the major repressing effect of XylR is caused by competitive binding to the promoter region, as was shown for the lac repressor of E. coli by Schlax et al. (25). B. megaterium cells producing Gfp mediated by a plasmid containing a combination of −35+, utr+, and rbs+ showed behavior similar to that of cells harboring pSSBm81 (−35+ −10+ rbs+) (data not shown). This might indicate that DNA sequence changes introduced to utr+ do not influence the mRNA's half-life exclusively but also significantly increase the RNA polymerase-binding affinity to its promoter.
B. megaterium cells employing pSSBm78 (−35+ rbs+) produced only a little more Gfp (80.6 mg per gCDW) than cells carrying pSSBm84 (−35+) (76.3 mg per gCDW). Here, the expected multiplier effect was not detected. Although the introduction of the NheI restriction site into the nonoptimized basic vector p3STOP1623 caused a significant reduction in Gfp production, almost no differences in Gfp production were measured when comparing cells carrying the optimized plasmids with or without the NheI restriction site (pSSBm78, 80.6 mg per gCDW; pSSBm85, 82.5 mg per gCDW). These observations led to speculation about general limitations of the protein formation process. On one site, this might be due to specific characteristics of the target gene. For example, the augmented use of certain codons that are rare in B. megaterium might lead to limitation of the corresponding tRNAs and would therefore hamper protein production. Another reason might be the limited supply of specific amino acids. This theory is supported by observations made during the fed-batch cultivations. Here, on the one hand, high volumetric Gfp yields (1.25 g liter−1) were achieved, but on the other hand, only smaller amounts of Gfp per unit of cell dry weight (36.8 mg per gCDW) than in shaking-flask cultures (82.5 mg per gCDW) were reached. An explanation might be the use of defined synthetic medium in the fed-batch cultivations and of complex medium in the shaking flasks.
Nevertheless, the combined optimized features employed in the new high-performance expression plasmids facilitated the production of 82.5 mg per gCDW Gfp in B. megaterium shaking-flask cultivations, which is even more than described for fed-batch cultivations of E. coli (73.7 mg per gCDW) (10). Thus, a competitive protein production system for the alternative bacterial host B. megaterium was developed.
In experiments in the production and secretion of heterologous Tfh in B. megaterium, the secretion facilitated by the mediocre SPPac was enhanced up to 6-fold when employing the optimized high-performance plasmids. However, the amount of Tfh secreted was not strongly enhanced when using well-suited SPs like SPLipA, SPYocH, or SP+YngK. Neither the new signal peptides that were tested nor their use in combination with the high-performance plasmids facilitated more than 16% greater secretion of Tfh (SPYocH in high-performance plasmid, 7,200 U liter−1, which equals 7.7 mg liter−1) than was mediated by SPLipA used in combination with the nonoptimized expression system (6,200 U liter−1 or 6.6 mg liter−1). These results clearly show limitations of the Sec-dependant secretion process, since the intracellular protein production was drastically increased, up to 18-fold (4.6 versus 82.5 mg per gCDW), as shown for Gfp production employing the high-performance production system.
Furthermore, it was shown that a computational approach to create an artificial signal peptide is in part possible. The application of an SPAsp-containing expression plasmid in protein secretion experiments proved the sufficient biological functionality of the artificial leader peptide. Since no prediction method for the right combination of SP and target protein is currently applicable with success (6), screening for the correct SP-protein combination is still necessary. Nevertheless, in combination with the newly developed high-performance expression system, six (SPYocH, SPYngK, SP+YngK, SPAsp, SPPac, and SPLipA) out of eight SPs tested (75%) facilitated high-yield secretion of Tfh, while only three SPs (SPYocH, SP+YngK, and SPLipA) (38%) achieved an equivalent performance when employed in the nonoptimized production system. The described toolbox comprising high-performance expression plasmids for both intracellular and extracellular protein production is commercially available from MoBiTec GmbH, Göttingen, Germany.
This work was financially supported by Deutsche Forschungsgemeinschaft (SFB578) and Codexis, Inc., Redwood City, CA.
Thanks to Karsten Hiller for supporting the computational part of this work and to Florian David for providing media and strategies for the fed-batch cultivations. Also, thanks to the undergraduate students Heinrich Schlums, Johannes Schwerk, Melanie Busch, and Jan Hellert for scientific assistance during the laboratory work.
Published ahead of print on 30 April 2010.
†Supplemental material for this article may be found at http://aem.asm.org/.