Secretion of heterologous proteins into Escherichia coli cell culture medium offers significant advantages for downstream processing over production as inclusion bodies; including cost and time savings, and reduction of endotoxin. Signal peptides play an important role in targeting proteins for translocation across the cytoplasmic membrane to the periplasmic space and release into culture medium during the secretion process. Alpha toxinH35L (ATH35L) was selected as an antigen for vaccine development against Staphylococcus aureus infections. It was successfully secreted into culture medium of E. coli by using bacterial signal peptides linked to the N-terminus of the protein. In order to improve the level of secreted ATH35L, we designed a series of novel signal peptides by swapping individual domains of modifying dsbA and pelB signal peptides and tested them in a fed-batch fermentation process. The data showed that some of the modified signal peptides improved the secretion efficiency of ATH35L compared with E. coli signal peptides from dsbA, pelB and phoA proteins. Indeed, one of the novel signal peptides improved the yield of secreted ATH35L by 3.5-fold in a fed-batch fermentation process and at the same time maintained processing at the expected site for signal peptide cleavage. Potentially, these new novel signal peptides can be used to improve the secretion efficiency of other heterologous proteins in E. coli. Furthermore, analysis of the synthetic signal peptide amino acid sequences provides some insight into the sequence features within the signal peptide that influence secretion efficiency.
Escherichia coli offers many advantages as a production organism, including growth on inexpensive carbon sources, rapid biomass accumulation, amenability to high cell-density fermentations and simple process scale-up (Mergulhao et al. 2005). Recombinant proteins can be produced in E. coli as intracellular inclusion bodies; however secretion into the extracellular environment is preferred as it simplifies downstream purification processes, protects recombinant proteins from proteolysis by cytoplasmic or periplasmic proteases, reduces endotoxin levels and contamination of the product by host proteins, and enhances biological activity (Gottesman 1996). In E. coli, proteins normally do not secreted into the extracellular circumstance except for a few classes of proteins such as toxin and hemolysin. However, small proteins are frequently released into the culture medium depends on the characteristics of signal sequences and proteins (Choi and Lee 2004; Tong et al. 2000). Many studies have therefore been carried out to improve the secretion efficiency of recombinant proteins in E. coli expression systems (Baneyx and Mujacic 2004; Cornelis 2000; Klatt and Konthur 2012; Mergulhao et al. 2005).
In E. coli, the Sec-dependent (Sec) secretion pathway is the general secretion route and signal peptides linked to the N-terminus of recombinant proteins play a critical role in translocation and secretion. Unfolded precursor proteins are translocated across the cytoplasmic membrane with concomitant cleavage of the signal peptide by signal peptidase and released into the periplasmic space where they fold into their native structure (Green and Mecsas 2016; Mergulhao et al. 2005; Valent et al. 1998). The further export of recombinant proteins into the extracellular environment has been reported for a number of proteins (Kotzsch et al. 2011; Qian et al. 2008). The Sec secretion pathway can also utilize a co-translational mechanism of export that couples translation of proteins by the ribosome with secretion through the SecYEG channel [the signal peptide recognition particle (SRP) pathway]. The SRP pathway relies on the SRP particle, which recognizes an N-terminal signal peptide with highly hydrophobic core during protein secretion and the binding affinity of the SRP particle for signal peptides increases with the hydrophobicity of the h-region of signal peptides (Green and Mecsas 2016; Nilsson et al. 2015).
Signal peptides are short peptides (generally 20–30 amino acid residues in length) and have three distinguishable structural domains with different functions; an amino-terminal region with a net positive charge (the n-region) followed by a hydrophobic region (the h-region) and a protease recognition sequence (the c-region) with a preference for small residues at the −3 (P3) and −1 (P1) positions relative to the cleavage site (Fekkes and Driessen 1999; Paetzel et al. 2000; Paetzel 2014). In order to improve secretion of recombinant proteins in E. coli expression systems, a number of heterologous signal peptides have been evaluated (Velaithan et al. 2014; Jonet et al. 2012; Ismail et al. 2011; Low et al. 2011; Nagano and Masuda 2014). These studies demonstrated that the hydrophobic region in the signal peptide plays an important role for protein translocation across the bacterial cytoplasmic membrane due to the interaction of the h-region with the membrane during protein translocation. Several studies have also reported that translocation efficiency increases with the length and hydrophobicity of the h-region, and a minimum hydrophobicity is required for their secretion function (Duffy et al. 2010; Ryan et al. 1993; Wang et al. 2000).
The pore-forming α-hemolysin protein, also known as α-toxin (AT), is produced by the majority of Staphylococcus aureus (S. aureus) serotypes and secreted as a water soluble monomer (Craven et al. 2009; Kennedy et al. 2010; Ragle and Bubeck 2009; Wardenburg and Schneewind 2008). The AT polypeptide is processed through the secretory machinery to yield a mature ~33 kDa protein of 293 amino acids (Berube and Wardenburg 2013). It is one of the most well-characterized bacterial virulence factors. AT oligomerizes into a heptameric structure on the host cell membrane creating a pore structure resulting in cell lysis. However, substitution of histidine 35 with leucine (ATH35L) leads to the loss of hemolytic activity of AT (Menzies and Kernodle 1994, 1996; O'Reilly et al. 1986; Ragle and Bubeck 2009). Since there are no vaccines available for the prevention of S. aureus infections, the mutant ATH35L has been investigated as a vaccine target (Menzies and Kernodle 1996; Ragle and Bubeck 2009) and, a previous inflammation study has used mutant ATH35L isolated from crude E. coli (Craven et al. 2009; O'Reilly et al. 1986; Saito et al. 2009). In this study, we designed novel signal peptides that significantly improved the secretion of the ATH35L protein into culture medium in E. coli and maintained proper cleavage processing to give the mature ATH35L protein sequence. We also demonstrated that the position of amino acid residues in the h-region is a potentially important factor affecting secretion of recombinant proteins.
Materials and methods
Escherichia coli strains and growth conditions
Escherichia coli strains BL21 (DE3) [fhuA2 [lon] ompT gal (λ sBamHIo ∆EcoRI-B int: lacI: PlacUV5::T7 gene1) i21 ∆nin5)[dcm]∆hsdS] and BL21 Star™ (DE3) [F−ompT hsdSB (r−B, m−B) galdcmrne131 (DE3)] were chosen as hosts for recombinant protein expression. Recombinant cells were cultured in seed medium (20 g/L yeast extract) or rich growth medium (20.3 g/L yeast extract (BioSpringer, Milwaukee, WI, USA) 10.1 g/L sodium sulfate anhydrous (JT baker, Center Valley, PA, USA) and 7 g/L K2HPO4 (JT baker, Center Valley, PA, USA) both supplemented with 50 μg/mL kanamycin (Sigma, St. Louis, MO, USA) for expression of recombinant ATH35L proteins at 30 °C.
Construction of expression plasmids
Expression vector, pJ411, provided by DNA 2.0 Inc. (Menlo Park, CA, USA) was used for T7 promoter driven-expression of signal peptide variants linked to the ATH35L gene. The kanamycin resistance gene was used as a selection marker. To generate ATH35L gene (Gene bank accession no. KY474302), wild-type of AT gene (Gene bank accession no. CP006838.1) was mutated by substitution of histidine 35 with leucine and the mutant was codon-optimized for expression in E. coli. Nucleic acid and amino acid sequences of the codon-optimized ATH35L gene used in this study are shown in Table 1. Codon-optimization of nucleic acid sequences for expression in E. coli, gene synthesis and DNA sequencing analysis of ATH35L with different signal peptides were performed by DNA 2.0 Inc (Menlo Park, CA, USA). A summary of all signal peptides and recombinant plasmids used in this study is shown in Table 2 and nucleic acid sequences of all signal peptides used in this study are summarized in Table 3.
Micro 24 fed-batch culture processes
Small volume fed-batch cultures (3 mL working volume) were performed in a Micro 24 bioreactor system (Applikon Biotechnology, Forster City, CA, USA). Feed solutions and other supplements were sterilized by autoclaving or filtration through a 0.22-μm pore size filter, while the rich growth culture medium [20.3 g/L yeast extract (BioSpringer, Milwaukee, WI, USA), 10.1 g/L sodium sulfate anhydrous (JT baker, Center Valley, PA, USA) and 7 g/L K2HPO4 (JT baker, Center Valley, PA, USA)] was separately autoclaved and added later under aseptic conditions. The culture was initiated by inoculation of 2.8% culture volume into the rich growth medium containing 50 μg/mL of kanamycin (Sigma, St. Louis, MO, USA), 7.6 g/L of trace metal cocktail solution (55 g/L sodium citrate dehydrate (Sigma, St. Louis, MO, USA), 27 g/L FeCl3·6H2O (Sigma, St. Louis, MO, USA), 0.5 g/L CoCl2·6H2O (Sigma, St. Louis, MO, USA), 0.5 g/L Na2MoO4·2H2O (Sigma, St. Louis, MO), 0.95 g/L CuSO4·5H2O (Sigma, St. Louis, MO, USA), 1.6 g/L MnCl2·4H2O (Sigma, St. Louis, MO, USA), 1.3 g/L ZnCl2 (Sigma, St. Louis, MO, USA) and 2 g/L CaCl2 (Sigma, St. Louis, MO, USA) and 0.8% culture volume of the glycerol and Epsom salt solution [315 g/L of glycerol (Sigma, St. Louis, MO, USA) and 31.4 g/L of MgSO4 (Sigma, St. Louis, MO, USA)]. During the fed-batch cultivation, the impeller speed was initially set to 800 rpm and later controlled to keep the dissolved oxygen level (DO) at 60% saturation. In fed-batch mode, 55% (v/v) glycerol for the carbon source and 33% (w/v) yeast extract were used as feed solutions. Recombinant ATH35L gene expression was induced by addition of 0.5 mM isopropyl-β-D1-thiogalactopyranoside (IPTG) (Biovectra, Charlottetown, PE, USA) when the cell reached at an optical cell density (OD600) of 10. Cell culture was continued at 30 °C for an additional 14 h after induction. Cultured cells were collected at 14 h post-induction, and cell culture medium was collected from the harvest samples by centrifugation at 13,300g for 5 min for analysis and quantification of extracellular ATH35L. Periplasmic and cytoplasmic fractions were prepared from cell lysate from the harvest samples using PeriPreps™Periplasting kit (Epicentre, Madison, WI, USA).
Fermentor fed-batch culture processes
Large volume fed-batch cultures were performed in a DasGip fermentor (SaniSure Inc, Moorepark, CA, USA) with 1 L working volume. Feed solutions, culture medium and other supplements were prepared as previously described for the small volume fed-batch culture processes. The culture was initiated by inoculation of 2.8% of culture volume into the prepared culture medium containing 50 μg/mL of kanamycin (Sigma, St. Louis, MO, USA), 7.6 g/L of trace metal cocktail solution, 15.8 g/L of glycerol (Sigma, St. Louis, MO, USA) and 1% (v/v) P2000 antifoam (Alfa Aesar, Reston, VA, USA) solution. Air space velocity was 1 vvm and the temperature was maintained at 30 °C. Ammonium hydroxide (23.5% v/v) (Sigma, St. Louis, MO, USA) and glacial acetic acid (50% v/v) solutions (Sigma, St. Louis, MO, USA) were used to maintain cultures at pH 7. During batch experiments, the impeller speed was initially set to 1200 rpm and later controlled to keep the DO at 60% saturation. In fed-batch mode, 55% (v/v) of glycerol for the carbon source and 33% (w/v) of yeast extract solutions were used as feed solutions. During feeding, the impeller speed was maintained constant at 1200 rpm, while the DO saturation was automatically kept at 60%. Recombinant ATH35L gene expression was induced by addition of 0.5 mM IPTG (Biovectra, Charlottetown, PE, USA) at an optical cell density of 80 (OD600). After induction, cell culture was continued in fed-batch mode at 30 °C for an additional 12 h. Cultured cells were collected at different time points post-induction to determine the profiles of secreted ATH35L protein, osmolality and the concentration of glycerol and acetate in culture medium.
For SDS-PAGE analysis, supernatants from the harvest samples were treated with 4× Bolt® LDS sample buffer (Life Technology, Frederick, MD, USA) which contains both lithium dodecyl sulfate as a denaturing agent and dithiothreitol (DTT) (Life Technology, Frederick, MD, USA) as a reducing agent. All samples were heated at 90 °C for 3 min before loading on SDS-PAGE gels. SDS-PAGE was performed using 10% pre-cast Bis–Tris NuPAGE SDS gels (Life Technology, Frederick, MD, USA). Electrophoresis was performed at a constant 200 V for 45 min in MOPS running buffer under denaturing conditions (Life Technology, Frederick, MD, USA). The separated protein bands were visualized by staining with the Simply Blue Safe Stain solution (Life Technology, Frederick, MD, USA) or transferred onto nitrocellulose membranes for Western blot analysis.
Detection of ATH35L protein by Western blot
Harvested samples were separated by SDS-PAGE under denaturing conditions and transferred onto nitrocellulose membranes using an iblot transfer kit (Life Technology, Frederick, MD, USA). Membranes were incubated in blocking buffer (5% skim milk in 0.1% TTBS buffer) at room temperature for 1–1.5 h, then incubated with 1 μg/mL of anti-Staphylococcal alpha hemolysin toxin mAb (LC10 mAb; MedImmune Inc, Gaithersburg, MD, USA) in 0.1% TTBS buffer at room temperature for 1 h or at 4 °C overnight. After washing membranes with 0.1% TTBS at room temperature, the membrane was incubated with 1 μg/mL of HRP conjugated Goat anti Mouse Ab (Bethyl Laboratories, Montgomery, TX, USA) in 0.1% TTBS at room temperature for 1 h. All membranes were then visualized with the SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL, USA) and scanned on a ChemiDoc XRS™ system (Biorad, Hercules, CA, USA).
Quantification of extracellular ATH35L protein
The concentration of ATH35L in cell culture medium was determined by a customized Octet assay. The assay was performed on the Octet QKe (Fortebio, MenloPark, CA, USA) using anti-Staphylococcal alpha hemolysin toxin mAb (LC10 mAb; MedImmune Inc, Gaithersburg, MD, USA) to capture the ATH35L protein. All samples were diluted at a 1:10 and 1:20 ratio with the kinetics buffer (Fortebio, MenloPark, CA, USA) in a 96-well plate (Corning, Tewksbury, MA, USA). For the Octet assay process, LC10 mAb was bound to the Protein A biosensor (Fortebio, MenloPark, CA, USA) at 300 rpm for 300 s, followed by the base line step in the kinetics buffer at 300 rpm for 60 s, the sample association step at 300 rpm for 150 s, and the dissociation step in the kinetics buffer at 300 rpm for 60 s in the basic kinetic mode. Experimental curves were recorded for the individual samples, and data was processed and analyzed using the Octet data analysis software 7.0 (Fortebio, MenloPark, CA, USA). Finally, samples were quantified by comparison with a standard curve generated from serial dilutions of affinity-purified ATH35L protein. Standard deviation values between the 1:10 and 1:20 diluted samples were less than 10%.
Purification of ATH35L from cultivation medium
For N-terminal peptide sequence analysis, extracellular ATH35L in cultivation medium was purified using a combination of ammonium sulfate precipitation and Poros XS cation exchange chromatography (Life Technologies, Grand Island, NY, USA). Briefly, the ATH35L culture was harvested by centrifugation at 850g for 30 min to remove the cells. The resulting culture medium was adjusted to pH 5.2 with 1 M acetic acid (Sigma, St. Louis, MO, USA), and centrifuged at 9500g for 15 min to remove precipitant. The supernatant was then purified using ammonium sulfate precipitation, followed by using Poros XS cation exchange resin (Life Technologies, Grand Island, NY, USA).
N-terminal polypeptide sequencing
Purified ATH35L was fractionated by SDS-PAGE and then transferred onto a nitrocellulose membrane using an iblot transfer kit (Life Technology, Frederick, MD, USA). The N-terminal amino acid sequencing analyses of the isolated ATH35L samples were performed by Covance (Greenfield, IN, USA) using an automated protein/peptide sequencing system.
Screening homologous and heterologous signal peptides for enhanced ATH35L secretion in E. coli
Since the selection of the signal peptide has a major impact on recombinant protein secretion in E. coli systems (Sjostrom et al. 1987), three E. coli signal peptides, two from the Sec pathway (pelBss and phoAss) and one from the SRP pathway (dsbAss), were screened to identify a signal peptide for efficient ATH35L secretion. It has been previously shown that some proteins are successfully secreted using their native signal peptides in heterologous expression systems (Rigi et al. 2014; Shahhoseini et al. 2003). Therefore, we also assessed the native signal peptide of AT (NTss) even though the structural elements of signal peptides from extracellular proteins are different between S. aureus and E. coli. A summary of the signal peptides screened is shown in Table 2. Recombinant cell strains containing different signal peptide-ATH35L fusion constructs were cultured in fed-batch conditions using a Micro24 system. The relative secretion levels of ATH35L protein into the cell culture medium were compared by Western blot analysis using purified mature ATH35L as a reference. The molecular weight of ATH35L precursor protein is approximately 3 kDa higher than mature ATH35L due to the presence of the signal peptide. Precursor and mature AT could therefore be distinguished by their size difference on SDS-PAGE gels.
Analysis of cell culture medium, cytoplasmic and periplasmic samples revealed that mature ATH35L protein was secreted into the culture medium as well as into the periplasmic space after export from the cytoplasm. Although some of soluble ATH35L still remained in the cytoplasm, more than 50% of expressed ATH35L protein for the E. coli signal peptides was secreted into the periplasmic space and the culture medium (Additional file 1: Figure S3).
Figure 1 shows the relative levels of ATH35L secreted into the culture medium from the different signal peptides. DsbAss showed the highest level of secreted ATH35L product, whereas NTss produced the lowest level. The n-region of NTss has a similar net positive charge (+1.1) to the selected E. coli signal peptides pelBss (+1), phoAss (+1) and dsbAss (+2). Furthermore, the length of all four signal peptides was similar. However, NTss has the longer h-region (Table 2) with a low similarity of amino acid sequences with the E. coli signal peptides (data not shown). Also, hydrophobicity in the h-domain of NTss is lower (50%) than the E. coli signal peptides (>80%). This observation implies that hydrophobicity strongly influences the secretion efficiency of recombinant proteins in E. coli. For the E. coli signal peptides screened, pelBss and phoAss showed a lower level of secreted ATH35L compared with dsbAss (Fig. 1). The amino acid sequences of these three signal peptides show a high level of similarity using the PRALINE multiple sequence alignment tool (Simossis et al. 2005). In particular, the identity of amino acid sequence between dsbAss and pelBss was as high as 45% (data not shown). Therefore, to investigate the impact on the ATH35L secretion of individual signal peptide domains and to improve the secretion efficiency of ATH35L, novel signal peptides were designed by modifying the native signal peptide sequences.
Design and screening of novel signal peptides for improving ATH35L secretion
With regard to Set I novel signal peptides, dsbAss and pelBss were selected to initiate the design of new signal peptides. In spite of the high level of amino acid sequence identity between the dsbAss and pelBss, these signal peptides showed different ATH35L secretion efficiencies. Six novel signal peptides, NSP1-6, were created by rearranging individual domains of dsbAss and pelBss and cloned in-frame into the ATH35L expression vector (Table 2; Fig. 2b). After transformation of recombinant plasmids into BL21 Star™ (DE3), recombinant cells were cultured in 1L scale fed-batch conditions to evaluate the secretion of ATH35L.
The data in Fig. 3 show that NSP2 and NSP4 increased the secretion of ATH35L into cell culture medium by 2.5-fold (0.4 g/L) and fivefold (0.8 g/L) respectively compared with dsbAss (0.15 g/L) at 10 h post-induction. Intriguingly, these two novel signal peptides share the same n- and h-domains in their structures (D–N and P–H), but not the c-domains (Table 2; Fig. 2b). The observed improvement in secretion suggests that combination of the n-domain of dsbAss (D–N) and the h-domain of pelBss (P–H) results in a favorable structure for efficient translocation of ATH35L across the cytoplasmic membrane. Although both signal peptides improved the ATH35L secretion into cell culture medium, the impact on secretion efficiency of ATH35L was different depending on the c-domain in their structure. The ATH35L secretion was more efficient when the c-domain of pelBss (P–C) was combined with D–N and P–H domains in NSP4 (Figs. 2b, 3).
In contrast to NSP2 and NSP4, the level of secreted ATH35L in cell culture medium was reduced for NSP6 (0.02 g/L) compared with dsbAss (0.15 g/L), even though NSP6 also contains the h-domain of pelBss like NSP2 and NSP4 (P–H; Fig. 2b) and the same c-domain as NSP2 (D–C; Fig. 2b). Interestingly, NSP1 and NSP5 contain the n-domain of pelBss like NSP6 (P–N; Fig. 2b) and the secreted ATH35L in cell culture medium was also reduced for these two signal peptides.
Unlike NSP1, NSP5 and NSP6, ATH35L secretion was not reduced for NSP2, NSP3 and NSP4 in comparison to dsbAss. NSP2, NSP3 and NSP4 all contain the n-domain of dsbAss in their structure (D–N in Fig. 2b). NSP3 shares the same n-and c-domains (D–N and P–C) with NSP4, but contains a different h-domain than NSP2 and NSP4 (D–H for NSP3 and P–H for NSP2 and NSP4). Figure 3 demonstrates that NSP3 did not improve the yield of secretory ATH35L (0.15 g/L), whereas NSP4 had the highest yield of secretory ATH35L (0.8 g/L). The present data denote that the n-domain of dsbAss (D–N) is a favorable for translocating ATH35L across the cytoplasmic membrane. Moreover, ATH35L secretion is more efficient when D–N and P–H domains are combined (NSP2 and NSP4 in Fig. 3) compared with the combination of D–N and D–H (NSP3 in Fig. 3). In addition, we detected ATH35L precursor protein by Western blot only from NSP4, although NSP3, NSP4 and NSP5 all contain the same c-domain (P–C) which determines signal peptide cleavage. One explanation is that that the detected ATH35L precursor protein was released from dead cells.
In Fig. 3, NSP1, NSP5 and NSP6 have the same n-domain (P–N), but not the h-domain. The secretion of ATH35L for NSP1 is less reduced than NSP5 and NSP6 although NSP1 has the same n- and h-domains with NSP5 (P–N and D–H). In Fig. 2b, NSP1 is composed of homologous h- and c-domains (D–H and D–C) while NSP5 and NSP6 are composed of heterologous h- and c-domains (D–H and P–C or P–H and D–C). We hypothesized that the signal peptide structure containing homologous h-and c-domains is more efficient for secretion of recombinant proteins. To verify this hypothesis, we also compared the ATH35L secretion between NSP2 and NSP4, and the result shows that these two signal peptide also have similar structural features for the secretion of ATH35L (P–H and D–C for NSP2 and P–H and P–C for NSP4). Further studies are still required to determine if the amino acid residues in the h- and c-domains influence each other to affect ATH35L secretion. However, in comparison among signal peptides containing the same n-domain, the data clearly showed that the composition of homologous h- and c-domains was more efficient for ATH35L secretion than the composition of heterologous h- and c-domains (NSP1>NSP5 or NSP6 and NSP4>NSP2 or NSP3 in Fig. 3).
Effects of altering the position of amino acids in the h-domain on the secretion of ATH35L
Two further series of novel signal peptides were created by modifying the position of amino acids in the h-regions of NSP3 and NSP4 in order to investigate whether modification of the hydrophobic amino acid position in the h-region affected the secretion efficiency of the ATH35L protein. NSP3 and NSP4 mutants were generated by shuffling the position of amino acids in the h-region or substituting all the h-region amino acids in the case of NSP4b and NSP4c (Fig. 2a, c). Cell culture medium samples from NSP3 (Set III) and NSP4 (Set II) mutant constructs were collected at 12 h post-induction for evaluating the productivity of secreted ATH35L.
For the Set II signal peptides (Fig. 2c; Table 2), NSP4a was created by rearranging the position of hydrophobic residues in the h-domain in order to investigate whether the position of hydrophobic residues affects secretion of ATH35L. Since total hydrophobicity of the h-domain is an important factor in determining the secretion efficiency of a recombinant protein (Hikita and Mizushima 1992a, b; Chou and Kendall 1990), we additionally generated a couple of mutants, NSP4b and NSP4c. Total hydrophobicity of these mutants were changed by replacing residues of the h-domain with polyleucine or polyalanine (Fig. 2c). The result shown in Fig. 4 indicates that NSP4 mutant signal polypeptides reduced overall ATH35L secretion levels. NSP4 produced 0.9 g/L of secretory ATH35L in cell culture medium, but NSP4a reduced the yield of secretory ATH35L to 0.28 g/L. Also, the yield of secretory ATH35L for both NSP4b and NSP4c was significantly reduced to 0.04 g/L regardless of increasing or diminishing total hydrophobicity of the h-domain (Fig. 4b). In Fig. 4a, mainly unprocessed ATH35L precursor was detected by Western blot for NSP4c. The total hydrophobicity of the h-domain in NSP4c was significantly decreased and it seems that NSP4c_ATH35L was not translocated successfully across the inner membrane from the cytoplasmic space. Most likely, the accumulated ATH35L precursors in the cytoplasmic space leaked out from the intracellular space after cell death.
For NSP3, there was no change of the ATH35L secretion efficiency compared to dsbAss despite having the same n- and c-domains of NSP4 in its structure (Figs. 2b, 3). In the study of the Set II NSP4 mutants, the secretion efficiency of ATH35L was significantly affected by the position of hydrophobic residues in the h-region (NSP4a in Figs. 2c, 4) and NSP4 did not affect the level of translation of this protein (Additional file 1: Figure S3). Thus, four mutants of NSP3, Set III, were generated by rearranging the location of amino acid residues in the h-domain in order to investigate the relationship between the amino acid position and the secretion efficiency of ATH35L. Amino acid residues in the h-domain was grouped and each group contained three amino acid residues. The groups were shuffled to relocate their position (Fig. 2a, d). The original position order of each amino acid in the h-region is numbered next to amino acids in the Table 2 and Fig. 2d. The secretion of ATH35L by the Set III signal peptides is shown in Fig. 5. NSP3b and NSP3c improved the yield of secretory ATH35L by 2.7-fold (0.48 g/L) and 1.4-fold (0.25 g/L) respectively compared with NSP3 (0.18 g/L). In contrast, NSP3a and NSP3d reduced the yield of secretory ATH35L to 0.1 g/L and 0.14 g/L, respectively (Fig. 5b). In Western blot analysis of Set III signal peptides, ATH35L precursor was not detected from any of NSP3 mutants (Fig. 5a).
Determining the optimal induction period for NSP4_ATH35L secretion
Figure 3 shows that the secretion of ATH35L into cell culture medium by NSP4 is approximately fivefold higher than dsbAss at 10 h post-induction. At the same time, ATH35L precursor protein was additionally detected from NSP4_ATH35L by Western blot, whereas there was no detection of ATH35L precursor from the other signal peptides. Thus, we performed a series of fed-batch cell culture experiments to determine the optimal induction time for maximizing the secretion of ATH35L while minimizing the contamination with ATH35L precursor.
Figure 6 shows the secretion of ATH35L from NSP4 and dsbAss into cell culture medium at various induction time points. Secretion of NSP4_ATH35L increased from 0.69 g/L at 8 h post-induction up to 1 g/L at 12 h post-induction. In comparison with NSP4, the productivity of secreted ATH35L for dsbAss increased from 0.12 g/L at 8 h post-induction up to 0.2 g/L at 12 h post-induction. The yield of secreted ATH35L for NSP4 was 5.7-fold higher than with dsbAss at 8 h post-induction and was consistently increased by approximate fivefold at 12 h post-induction compared with dsbAss. For NSP4_ATH35L, the ATH35L precursor was not detected at 8 h post-induction, but started to appear after 10 h post-induction. In the cell growth profile, the cell growth rate of NSP4_ATH35L was gradually reduced from 8 h post-induction (Additional file 1: Figure S2). In contrast to NSP4, the cell growth rate and the productivity of secreted ATH35L for dsbAss_ATH35L were consistently increased up to 12 h post-induction (Figs. 6b and Additional file 1: Figure S2). After 12 h post-induction, the productivity of secreted ATH35L for dsbAss did not increase (data not shown). Additionally, we did not detect ATH35L precursor in the cell culture medium of dsbAss_ATH35L up to 12 h post-induction (Fig. 6a). Thus, the optimal induction times for secretory ATH35L from NSP4_ATH35L and dsbAss_ATH35L were at 8 and 12 h post-induction respectively when the yield of secretory ATH35L for NSP4 was ≥3.5-fold higher than for dsbAss. Therefore, the present data show that NSP4 substantially improved the levels of ATH35L secretion in a shorter induction time compared with dsbAss.
In the secretion of recombinant proteins, the site and consistency of cleavage of the signal peptide by the signal peptidase is an important product quality attribute. To verify the cleavage of ATH35L from NSP4, secreted ATH35L in culture medium was purified and analyzed by N-terminal peptide sequencing (Additional file 1: Figure S1). The data confirmed that the novel signal peptide, NSP4, was correctly cleaved from ATH35L precursor protein and mature ATH35L protein in cell culture medium was released from the periplasmic space (Additional file 1: Table S1).
The goals of our study were to evaluate the impact of different signal sequences on the secretion of ATH35L into cell culture medium and to increase the secretion of properly processed ATH35L into cell culture medium. In our study, a comparative analysis by Western blot showed that dsbAss was the most effective in directing the secretion of ATH35L into cell culture medium among the initial four signal peptides tested. In contrast, pelBss showed the lowest secretion efficiency of ATH35L among E. coli signal peptides (Fig. 1), although amino acid sequences of these two signal peptides are 45% identical using Needleman–Wunsch algorithm (Needleman and Wunsch 1970). This result suggests that different domain compositions of signal peptides can lead to different secretion efficiencies of recombinant proteins despite the high degree of similarity between different signal peptide sequences.
Since the individual domains of the signal peptide have different roles in protein targeting and translocation (Choi and Lee 2004; Lehnhardt et al. 2012), we hypothesized that the secretion efficiency of a recombinant protein can be influenced by modifying the signal peptide domains. In screening the Set I novel signal peptides to investigate this hypothesis, NSP2, NSP3 and NSP4 improved or maintained the secretion of ATH35L into cell culture medium (Fig. 3). In contrast, NSP1, NSP5 and NSP6 reduced the ATH35L secretion (Fig. 3). Several studies have shown that the net positive charge of basic residues in the n-domain is an important feature the highly basic n-domain promotes the interactions of the n-domain with SRP, influencing the translocation of a recombinant protein and removing basic residues from signal peptides reduced the rate of recombinant protein export (Hikita and Mizushima 1992a, b; Low et al. 2013; Nesmeyanova et al. 1997; Nilsson et al. 2015; Peterson et al. 2003; Tian and Bernstein 2009). Consistent with these observations, in this study, the n-domain of dsbAss has higher positive net charge (+2) than the n-domain of pelBss (+1), and signal peptides containing the n-domain of dsbAss (NSP2, NSP3 and NSP4) showed more favorable translocation of ATH35L.
NSP2, NSP3 and NSP4 contain a favorable n-domain (D–N) for ATH35L secretion (Figs. 2b and 3). Nevertheless, these signal peptides varied in the secretion of ATH35L and only NSP2 and NSP4 positively impacted the ATH35L secretion over dsbAss (Fig. 3). One of their structural differences is that NSP2 and NSP4 have the h-domain of pelBss (P–H) and NSP3 has the h-domain of dsbAss (D–H). Previous studies have shown that the h-domain also plays a critical role in its secretion activity and total hydrophobicity of the h-region is a key determinant of secretion efficiency (Choi and Lee 2004; Duffy et al. 2010; Hikita and Mizushima 1992a; Sjostrom et al. 1987; Zanen et al. 2005). Additionally, increased total hydrophobicity in the h-domain of signal peptides has been demonstrated to improve recombinant protein secretion (Chou and Kendall 1990; Jonet et al. 2012; Klatt and Konthur 2012). Thus, we analyzed total hydrophobicity values of the h-domain for the selected E. coli signal peptides including NSP4b and NSP4c (Table 4). Interestingly, the h-domain of NSP2 and NSP4 (P–H) has a lower total hydrophobicity than the h-domain of NSP3 (D–H). Since NSP2, NSP3 and NSP4 share the same n-domain, variation in the basic amino acid residues in the n-domain cannot explain the differences in ATH35L secretion observed with these signal peptides. Similarly, the h-domain of phoAss has the highest hydrophobicity among the selected E. coli signal peptides (phoAss > dsbAss > pelBss; Table 4), but dsbAss showed better ATH35L secretion efficiency than phoAss (Fig. 1). Moreover, the ATH35L secretion efficiency was reduced for NSP4b regardless of the increased total hydrophobicity by using substitution of amino acid residues in the h-domain with polyleucine (Fig. 2c; Table 4). This observed lack of correlation of overall h-domain hydrophobicity with secretion may be associated with the composition of the hydrophobic amino acid residues. Several studies have noted that the composition of hydrophobic amino acids in the h-domain are not random and the organization of hydrophobic residues in the h-domain of signal peptide is important for leading the translocation of a recombinant protein in secretion process (Duffy et al. 2010; Lehnhardt et al. 2012). Thus, this would explain why NSP4b with its homogeneous hydrophobic h-domain, significantly decreased the secretion of ATH35L (Fig. 4; Table 4). Not surprisingly, hydrophobic amino acids in the h-domain are required for secretion because when the NSP4 h-domain hydrophobicity was demolished by replacing all residues in the h-domain with polyalanine in NSP4c, there was failure of ATH35L secretion. This suggests that placing strong hydrophobic amino acid residues in the signal peptide are required for the secretion of a recombinant protein, but the composition of amino acids in the h-domain is more critical than merely increasing total hydrophobicity to improve secretion of recombinant proteins.
We changed the pattern of amino acid residues in the h-domain whilst maintaining the overall composition and hydrophobicity of the h-domain to investigate whether ATH35L secretion could be improved. In NSP4a, altering the position of individual amino acids significantly impacted the secretion efficiency of ATH35L (Fig. 4). Similarly, NSP3a,b,c and d showed differing secretion efficiencies after rearranging the position of amino acid residues in the h-domain of NSP3 (Fig. 5). In the NSP3b and c mutants, the location of strong hydrophobic residues, such as leucine and valine, at the center of the h-domain or close to the c-domain increased the ATH35L secretion. In contrast, when these hydrophobic residues were located close to the n-domain as in NSP3a and NSP3d, ATH35L secretion was decreased. Consistent with our hypothesis, NSP4a also showed reduced ATH35L secretion when strongly hydrophobic amino acid residues in the h-domain were located close to the n-polar region. This also suggests that hydrophobic residues in the h-domain interact with amino acids in the n- and c-domains.
In conclusion, in designing and testing novel signal peptides we have both improved the secretion of recombinant ATH35L from E. coli (Additional file 1: Figure S3) and also increased the understanding of the influence of the composition and interaction of the signal peptide domains on secretion. The data presented here demonstrated that (1) the combination of the n-domain of dsbAss and the h-domain of pelBss is favorable for ATH35L secretion, (2) hydrophobicity in the h-region is a critical component for translocation of recombinant proteins across the cytoplasmic membrane, (3) the composition and arrangement of hydrophobic amino acids in the h-domain influence the secretion efficiency of recombinant proteins. From these studies, we have identified a novel signal peptide (NSP4) that significantly improved ATH35L secretion and decreased the induction time, which are both very beneficial for commercial production although the exact mechanism of the secretion pathway for NSP4_ATH35L is not yet clear. Furthermore, we believe that the novel signal peptides designed in the present study could be used to improve the secretion efficiency of other recombinant proteins in the E. coli expression platform.
alpha hemolysin toxin
alpha hemolysin toxinH35L mutant
a bacterial disulfide bond formation protein A signal sequence
native signal sequence
novel signal peptide
pectate lyase B signal sequence of Erwinia carotovora CE
alkaline phosphatase signal sequence of E.coli
sodium dodecyl sulfate polyacrylamide gel electrophoresis
a protease recognition sequence
a hydrophobic region
an amino-terminal region with a net positive charge
the Sec-dependent secretion pathway
the signal peptide recognition particle
the twin-arginine translocation secretion pathway
tris-buffered saline (TBS) and Polysorbate 20
Baneyx F, Mujacic M (2004) Recombinant protein folding and misfolding in Escherichia coli. Nat Biotechnol 22:1399–1408
Craven RR, Gao X, Allen IC, Gris D, Bubeck Wardenburg J, McElvania-Tekippe E, Ting JP, Duncan JA (2009) Staphylococcus aureus alpha-hemolysin activates the NLRP3-inflammasome in human and mouse monocytic cells. PLoS ONE 4(10):7446. doi:10.1371/journal.pone.0007446
Hikita C, Mizushima S (1992a) The requirement of a positive charge at the amino terminus can be compensated for by a longer central hydrophobic stretch in the functioning of signal peptides. J Biol Chem 267(17):12375–12379
Ismail NF, Hamdan S, Mahadi NM, Murad AM, Rabu A, Bakar FD, Klappa P, Illias RM (2011) A mutant L-asparaginase II signal peptide improves the secretion of recombinant cyclo dextrin glucano transferase and the viability of Escherichia coli. Biotechnol Lett 33:999–1005
Jonet MA, Mahadi NM, Murad A, Rabu A, Bakar F, Rahim R, Low K, Illias R (2012) Optimization of a heterologous signal peptide by site-directed mutagenesis for improved secretion of recombinant proteins in Escherichia coli. J Mol Microbiol Biotechnol 22:48–58
Kennedy AD, Bubeck Wardenburg J, Gardner DJ, Long D, Whitney AR, Braughton KR, Schneewind O, DeLeo FR (2010) Targeting of alpha-hemolysin by active or passive immunization decreases severity of USA300 skin infection in a mouse model. J Infect Dis 202:1050–1058
Lehnhardt S, Inouye S, Inouye M (2012) Functional analysis of signal peptide for protein secretion with use of oligonucleotide-directed site specific mutagenesis. In: Inouye M, Sarma R (eds) Protein engineering: applications in science, medicine and industry. Academic Press Inc., London, pp 157–172
Menzies BE, Kernodle DS (1994) Site-directed mutagenesis of the alpha-toxin gene of Staphylococcus aureus: role of histidines in toxin activity in vitro and in a murine model. Infect Immun 62(5):1843–1847
Nagano R, Masuda K (2014) Establishment of a signal peptide with cross-species compatibility for functional antibody expression in both Escherichia coli and Chinese hamster ovary cells. Biochem Biophys Res Commun 447:655–659
Nesmeyanova MA, Karamyshev AL, Karamysheva ZN, Kalinin AE, Ksenzenko VN, Kajava AV (1997) Positively charged lysine at the N-terminus of the signal peptide of the Escherichia coli alkaline phosphatase provides the secretion efficiency and is involved in the interaction with anionic phospholipids. FEBS Lett 403:203–207
Nilsson I, Lara P, Hessa T, Johnson AE, von Heijne G, Karamyshev AL (2015) The code for directing proteins for translocation across ER membrane: SRP cotranslationally recognizes specific features of a signal sequence. J Mol Biol 427:1191–1201
O’Reilly M, de Azavedo JC, Kennedy S, Foster TJ (1986) Inactivation of the alpha-haemolysin gene of Staphylococcus aureus 8325-4 by site-directed mutagenesis and studies on the expression of its haemolysins. Microb Pathog 2:125–138
Qian ZG, Xia XX, Choi JH, Lee SY (2008) Proteome-based identification of fusion partner for high-level extracellular production of recombinant proteins in Escherichia coli. Biotechnol Bioeng 101:587–601
Saito H, Inoue M, Tomiki M, Nemoto H, Komoriya T, Kimata J, Watanabe K, Kohno H (2009) Identification and sequence determination of recombinant Clostridium perfringens alpha-toxin by use of electrospray ionization mass spectrometry. Rinsho Biseibutshu Jinsoku Shindan Kenkyukai Shi 20(1–2):9–20
Shahhoseini M, Ziaee AA, Ghaemi N (2003) Expression and secretion of an α-amylase gene from a native strain of Bacillus licheniformis in Escherichia coli by T7 promoter and putative signal peptide of the gene. J Appl Microbiol 95:1250–1254
Tian P, Bernstein HD (2009) Identification of a post-targeting step required for efficient cotranslational translocation of proteins across the Escherichia coli inner membrane. J Biol Chem 284(17):11396–11404
Tong L, Lin Q, Wong WKR, Ali A, Lim D, Sung WL, Hew CL, Yang DSC (2000) Extracellular expression, purification, and characterization of a winter flounder antifreeze polypeptide for Escherichia coli. Protein Expr Purif 18:175–181
Valent QA, Scotti PA, High S, de Gier JWL, Heijne GV, Lentzen G, Wintermeyer W, Oudega B, Luirink J (1998) The Escherichia coli SRP and SecB targeting pathways converge at the translocon. EMBO J 17:2504–2512
Zanen G, Houben EN, Meima R, Tjalsma H, Jongbloed JD, Westers H, Oudega B, Luirink J, van Dijl JM, Quax WJ (2005) Signal peptide hydrophobicity is critical for early stages in protein export by Bacillus subtilis. FEBS J 272(18):4617–4630
SH designed and conducted the bulk of the experiments, performed analysis and interpretation of data, and drafted the manuscript. SM and GX participated in some of the experiments. RS coordinated the study and helped in the drafting of the manuscript. MB and TL gave advice in the study and edited the manuscript. All authors read and approved the final manuscript.
We would like to thank our co-workers in providing the ATH35L DNA sequence information and bioreactor processes for this study: the Vaccine Platform Research Group and George Chacko (Department of Cell Culture and Fermentation Sciences).
The authors declare that they have no competing interests.
Availability of data and materials
The datasets supporting the conclusions of this article are included within the article. The ATH35L nucleotide sequences used in this study is published in the GenBank (National Center for Biotechnology information) with the accession numbers KY474302.
Ethics approval and consent to participate
This article does not contain any studies with human participants and animals performed by any of the authors.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Additional file 1: Table S1. Identified peptide sequences of secreted ATH35L in culture medium. Figure S1. SDS-PAGE analysis of purified ATH35L protein. Secreted ATH35L by NSP4 in culture medium was purified and used for N-terminal peptide sequencing analysis. Figure S2. The cell growth profiles of NSP4_ATH35L and dsbAss_ATH35L. Recombinant ATH35L gene expression was induced by addition of 0.5 mM IPTG at an optical cell density of 80 (OD600). After induction, cell culture was continued in fed-batch mode at 30 °C for an additional 12 h. Cultured cells were collected at different time points postinduction to determine the profiles of secreted ATH35L protein in culture medium. Figure S3. Western blot analysis of cell lysates and cell culture medium from dsbA_ATH35L and NSP4_ATH35L in a Micro24 fed-batch process. (a) Arrow indicates ATH35L protein in cell crude lysates. (b) Arrow indicates released mature ATH35L protein in cell culture medium. Purified ATH35L protein was used as a reference.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Han, S., Machhi, S., Berge, M. et al. Novel signal peptides improve the secretion of recombinant Staphylococcus aureus Alpha toxinH35L in Escherichia coli
AMB Expr7, 93 (2017). https://doi.org/10.1186/s13568-017-0394-1