Because the correct disulfide arrangements of disulfide-containing peptides are difficult to obtain, the yeast P. pastris is widely used to express peptides containing intramolecular disulfide bridges. The main advantage of P. pastris as a host is that it is expected to secrete peptides with correct disulfide bridges directly into culture medium (Daly and Hearn 2005). Several CSαβ-type antimicrobial peptides have been produced by using P. pastris, and these were well characterized (Wang et al. 2008; Zhang et al. 2011; Wiens et al. 2011). However, a previous study showed that P. pastris is not a suitable host for large-scale production of ABF-2. Although the antimicrobial spectrum of ABF-2 was investigated, structural and mutational studies of ABF-2 have not yet been conducted because of the low yield of ABF-2 in P. pastris (Kato et al. 2002). Therefore, it is very important to develop an alternative expression method that enhances the expression level of ABF-2.
In this study, we selected E. coli as an expression host and at first tried to express ABF-2 directly in E. coli as an inclusion body. Unfortunately, the direct expression of ABF-2 in E. coli was insufficient due to the instability of the expressed ABF-2 itself in the cells (Figure 2). We speculated that this low-level expression of ABF-2 was caused by the degradation of expressed ABF-2 by proteases of E. coli. Therefore, we considered that the prevention of ABF-2 degradation by proteases is essential to enhance the expression level of ABF-2 as an inclusion body.
To prevent the degradation of target peptides, fusion protein systems have been used in many studies. The attachment of soluble proteins to target peptides has been observed to prevent degradation and promote proper folding (Xu et al. 2006). However, when an antimicrobial peptide such as ABF-2 is expressed in a soluble form, it may damage host cells by disrupting their cell membranes. Moreover, in fusion protein systems, enzymatic or chemical cleavage is inevitable to remove the fusion protein tag. Enzymatic cleavage often degrades recombinant peptides, because widely used proteases, such as enterokinase and factor Xa, often show nonspecific cleavages at unexpected sites. Furthermore, many peptides contain potential cleavage sites cleaved by chemicals. For instance, CNBr is commonly used to cleave peptide bonds C-terminal to methionine residues in proteins and peptides. However, because ABF-2 contains methionine in its amino acid sequence, CNBr cannot be used to separate ABF-2 from the fusion protein.
Expression of inclusion bodies may also be useful to avoid proteolytic degradation of the target peptide. Walsh et al. succeeded in expressing a large amount of amyloid β-peptide (Aβ) directly as an inclusion body (Walsh et al. 2009). Because the Aβ peptide has a strong intrinsic propensity to aggregate, we considered that overexpression of the Aβ peptide would readily lead to the formation of an inclusion body in E. coli. This method is a very simple and inexpensive way to produce a large amount of Aβ peptides. However, it is generally difficult to control the inclusion body formation in E. coli cells, especially in the case of small and common peptides. As another way to produce fusion proteins, the utilization of insoluble protein tags has also been reported (Park et al. 2009). This method is expected to prevent both degradation and toxicity. However, when an insoluble protein tag is used for the fusion expression, it is difficult to employ enzymatic cleavage due to the insolubility of the fusion protein, and chemical cleavage is usually necessary. This cleavage step is thus one of the drawbacks of this method.
In some studies, the coexpression of an insoluble partner protein has been reported to enhance the inclusion body formation of the target peptide and protect it from proteolytic degradation by protease. Saito et al. reported that the expression level of somatomedin C was enhanced in the case of coexpression of insulin-like growth factor І (Saito et al. 1987). Jang et al. succeeded in producing a potent antimicrobial peptide, buforin ІІb, by coexpression of human gamma interferon (Jang et al. 2009). To obtain a large amount of ABF-2, we decided to apply this method to ABF-2 and examine the effects of aggregation-prone partner proteins in detail.
In previous studies, translationally coupled two-cistron plasmids were used to coexpress target peptides and partner proteins (Saito et al. 1987; Jang et al. 2009). However, genetic manipulation was needed to construct the translationally coupled two-cistron expression systems. In the present study, we simply used the commercially available pCOLADuet-1 vector to coexpress ABF-2 and partner protein genes (Figure 1). Because this vector is designed for the cloning and coexpression of two genes, the construction of a coexpression plasmid is very easy. Therefore, our method can be easily applied to many proteins and peptides.
In this study, we chose four kinds of proteins as aggregation-prone partners. Lysozyme and α-lactalbumin appear to have evolved from a common ancestral protein, as evidenced by the similarity of their amino acid sequences and three-dimensional structures as well as by the high conservation of disulfide bridges (McKenzie and White 1991). Some studies have clearly shown that these four proteins each form a large amount of inclusion bodies when overexpressed in E. coli (Svensson et al. 2000; Li and Su 2002). Interestingly, the isoelectric points of these proteins are different despite their sequential similarity. Thus, we selected these proteins as good models of aggregation-prone partners in order to evaluate the effect of the charge of the partner protein on the expression level of cationic ABF-2 (pI 9.1).
Coexpression of BLZ, whose isoelectric point is 6.5, modestly enhanced the expression level of ABF-2. In contrast, ABF-2 was produced effectively as an inclusion body in the case of coexpression with an anionic partner protein (HLA or BLA). Coexpression of HLZ, whose isoelectric point is high, yielded no change in the expression level of ABF-2. Although it has been conjectured that the charge of both target and partner proteins influences inclusion body formation, the effect of the charge of the partner protein on the expression level of the target peptide has not been elucidated in detail. In this study, we experimentally showed that the charge of the partner protein is an important factor for enhancing the inclusion body formation of the target peptide. Interestingly, although HLA and BLA have almost the same isoelectric points and molecular weights, the expression level of ABF-2 was enhanced more by coexpression of BLA than by that of HLA. Because not only electrostatic but also hydrophobic interactions are thought to be critical factors for inclusion body formation (Murby et al. 1995), we compared their GRAVY scores, which express the total hydrophobicity of a protein. However, we could not find a correlation between the expression level of ABF-2 and the GRAVY scores of the partner proteins. There may be unknown factors that affect inclusion body formation.
It has been reported that the presence of impurities in an inclusion body, such as nucleic acids and non-plasmid-encoded proteins, affects the final refolding yield of the target (Maachupalli-Reddy et al. 1997). These impurities can be removed by washing the body using a low concentration of detergent. Therefore, inclusion body washing is very important to enhance the refolding yield of the target. However, in this study we observed that the inclusion body composed of ABF-2 without partner proteins was gradually solubilized during the washing process (data not shown). Because of this unfavorable solubilization of the body, ABF-2 was lost during the washing process and we could not obtain even crude ABF-2 in the experiment using expression without partner proteins. On the other hand, the inclusion body composed of ABF-2 and the partner protein showed almost no solubilization during the washing process, suggesting that a robust inclusion body was formed via coexpression of the peptide and a partner protein. From these results, it can be said that coexpression of an aggregation-prone protein is effective not only to enhance the expression level of a target peptide as an inclusion body but also to prevent the unfavorable loss of an inclusion body during washing.
Because the charge of ABF-2 is opposite that of BLA, we succeeded in separating ABF-2 from BLA efficiently by a simple one-step cation-exchange chromatography without enzymatic or chemical cleavage (Figure 3). After cation-exchange chromatography, we succeeded in obtaining 54 mg of crude ABF-2 without disulfide bridges from 1 L of culture. Although crude ABF-2 was refolded by a very simple standard dialysis refolding protocol, we obtained as much as 7.8 mg of correctly folded ABF-2. Because refolding additives, such as arginine, have been used to suppress the aggregation of proteins during refolding in many studies (Arakawa et al. 2007), optimization of the refolding protocol may enhance the refolding yield of ABF-2.
To confirm that refolded ABF-2 was correctly folded and active, purified ABF-2 was evaluated by a colony-forming unit assay as well as by CD and NMR spectroscopy (Table 3, Figures 4, 5). We confirmed that purified ABF-2 was active against both S. aureus and E. coli. Moreover, S. aureus is more sensitive than E. coli. These results were identical to the findings reported previously (Kato et al. 2002). Quantitative analysis of the CD spectra of purified ABF-2 indicated a secondary structural content that included 28% α-helix and 17% β-sheet. This result was in agreement with the categorization of ABF-2 as a CSαβ-type antimicrobial peptide. Moreover, the HSQC spectrum of 15 N-labeled ABF-2 is sharp and well dispersed. From these data, we concluded that purified ABF-2 was correctly folded. This sample will enable us to analyze the structure and molecular mechanism of ABF-2 by using NMR in future studies.
It is known that the long C-terminal regions of ABFs are divergent and vary in length (Froy 2005). Therefore, the difference in the C-terminal region of ABFs is thought to affect their antimicrobial spectrum. We believe that the method described in this work could be applied not only to ABF-2 but also to other ABFs. In future studies, we plan to use this technique to elucidate the effects of the differences in the C-terminal regions of ABFs on their antimicrobial spectra.
Here, we demonstrated that ABF-2 could be expressed as an inclusion body in large quantities by coexpression of an aggregation-prone protein. Moreover, the expression level of ABF-2 was greatly enhanced by coexpression of anionic partner proteins. After the refolding and purification processes, we obtained milligram quantities of correctly folded ABF-2. Finally, this expression method allowed stable isotopic labeling of recombinant ABF-2, which is required for structural studies using multidimensional heteronuclear NMR spectroscopy.