- Email: [email protected]
Expression, purification and characterization of the receptor-binding domain of botulinum neurotoxin serotype B as a vaccine candidate
Protein Expression and Purification 110 (2015) 122–129
Contents lists available at ScienceDirect
Protein Expression and Purification journal homepage: www.elsevier.com/locate/yprep
Expression, purification and characterization of the receptor-binding domain of botulinum neurotoxin serotype B as a vaccine candidate Alon Ben David, Amram Torgeman, Ada Barnea, Ran Zichel ⇑ Department of Biotechnology, Israel Institute for Biological Research, Ness Ziona, Israel
a r t i c l e
i n f o
Article history: Received 29 January 2015 and in revised form 8 February 2015 Available online 26 February 2015 Keywords: Botulinum vaccine Recombinant protein expression His-tag protein purification
a b s t r a c t The receptor-binding domain of botulinum neurotoxins (the HC fragment) is a promising vaccine candidate. Among the HC fragments of the seven BoNT serotypes, the expression of HC/B in Escherichia coli is considered especially challenging due to its accumulation as a non-soluble protein aggregate. In this study, the effects of different parameters on the expression of soluble HC/B were evaluated using a screening assay that included growing the bacterium at a small scale, a chemical cell lysis step, and a specific ELISA. The highest soluble HC/B expression levels were obtained when the bacterium E. coli BL21(DE3) + pET-9a-HC/B was grown in terrific broth media at 18 °C without induction. Under these conditions, the yield was an order of magnitude higher than previously reported. Standard purification of the protein using a nickel column resulted in a low purity of HC/B. However, the addition of an acidic wash step prior to protein elution released a major protein contaminant and significantly increased the purity level. Mass spectrometry analysis identified the contaminant as ArnA, an E. coli protein that often contaminates recombinant His-tagged protein preparations. The purified HC/B was highly immunogenic, protecting mice from a 106 LD50 challenge after a single vaccination and generating a neutralizing titer of 50 IU/ml after three immunizations. Moreover, the functionality of the protein was preserved, as it inhibited BoNT/B intoxication in vivo, presumably due to blockade of the neurotoxin protein receptor synaptotagmin. Ó 2015 Elsevier Inc. All rights reserved.
Introduction Botulism is a neurologic disease caused by toxins mainly produced by the bacterium Clostridium botulinum. Although botulism cases are rare in humans, botulinum neurotoxins (BoNTs1) are the most poisonous substances known, and they are therefore categorized as a tier 1 select agent by the Center for Disease Control and prevention (CDC) [1]. There are seven serologically distinct serotypes of these neurotoxins (designated A–G), among which serotypes A, B, and E are the most abundant in cases of human intoxication [2]. BoNTs are 150-kDa proteins, consisting of a 100kDa heavy chain (H) joined to a 50-kDa light chain (L) via a disulfide bond. The toxins of all serotypes share a similar architecture, organized into three structural domains that mediate the three steps of the intoxication process. The first step is the attachment of the
⇑ Corresponding author at: Israel Institute for Biological Research, P.B. 19, Ness Ziona 74100, Israel. Tel.: +972 8 9381513; fax: +972 8 9381761. E-mail address: [email protected] (R. Zichel). 1 Abbreviation used: BoNTs, botulinum neurotoxins; CDC, Center for Disease Control and prevention; rBV, recombinant botulinum vaccine; TB, terrific broth; CV, column volumes; TTD, time to death. http://dx.doi.org/10.1016/j.pep.2015.02.008 1046-5928/Ó 2015 Elsevier Inc. All rights reserved.
receptor-binding domain, located at the C-terminus of the heavy chain (the HC fragment), to its receptors and subsequent internalization via endocytosis. The next step is the translocation and release of the light chain into the cytosol, a step considered to be facilitated by the translocation domain found on the N-terminus of the heavy chain (HN). The final step is the cleavage of one of three SNARE (soluble N-ethylmaleimide sensitive factor attachment protein receptor) proteins by the light chain, which possesses endopeptidase activity, thereby inhibiting the release of the neurotransmitter acetylcholine from nerve cells to synapses [2–4]. The prevention and treatment of botulism is based on the presence of toxin-specific neutralizing antibodies in the blood stream. The source of the antibodies can be either self, in the case of prophylactic vaccination, or external (usually of vaccinated equine origin), in cases of post-exposure treatment. Historically, vaccines against botulism have consisted of formalin-inactivated toxins (toxoid) adsorbed to aluminum hydroxide [3]. Although toxoidbased botulinum vaccines are safe and efficient [3,5], their manufacturing process is expensive because it requires large-scale production facilities for this highly hazardous and spore forming agent. Therefore, current efforts to develop botulism vaccines are mainly focused on subunit vaccines consisting of the HC fragment.
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
The HC fragment is not toxic by itself, but it contains many neutralizing epitopes and can be produced recombinantly. In 1995, Clayton et al. reported that vaccination of mice with a recombinant HC fragment of neurotoxin serotype A elicited protective immunity against challenge with its homologous toxin [6]. Since that time, extensive research efforts have been made, and the receptor-binding domains of the seven BoNT serotypes have been produced and shown to induce a protective immune response [5,7]. Moreover, a recombinant botulinum vaccine (rBV) composed of the HC fragments of botulinum neurotoxins A and B, produced by the DynPort Vaccine Company for the United States Department of Defense, is currently under clinical investigation [8]. The first attempts to produce recombinant HC fragments of neurotoxin A were performed using Escherichia coli as a host [6,9]. However, as most of the expressed protein was insoluble in this system, subsequent studies have used the alternative host Pichia pastoris. Using this system, the HC fragments of botulinum A, B, C, D, E, and F have been produced with satisfactory yields [10–15]. Recently, the interest in expressing HC fragments in E. coli has returned, with better success. In the expression of the receptor domains of BoNT/E and BoNT/F, yields of 8–20 mg of protein per liter of culture were obtained [7,16,17]. Greater efforts have been made regarding HC/A expression. Several groups reported yields of 10–70 mg of HC/A per liter of culture [7,18,19], and we recently developed an expression system that yields 350 mg of HC/A per liter [20]. Nevertheless, expression of the receptor-binding domain of BoNT/B has remained challenging, as the reported yields of the expression of this domain are much lower. In the work of Baldwin et al. [7], the HC fragments of seven BoNT serotypes were expressed in E. coli to produce a heptavalent botulinum vaccine. The authors described special difficulty in expressing HC/B, as most of the protein was found in the insoluble cell fraction, with only 2 mg of the protein being obtained from 1 liter of culture. Similar yields were reported by Held et al. (1 mg per liter) [21]. Gao et al. purified HC/B from inclusion bodies with higher yields [22]. However, proper refolding of the protein into its native structure could not be achieved, and the final protein was obtained in 2 M urea. In this study, we developed a simple, efficient, and economic process for expressing HC/B in a soluble form in E. coli, achieving a yield that is an order of magnitude higher than previously reported. The protective properties of the recombinant product are at least comparable to those of HC/Bs produced using other expression systems, and the product is functionally active, as it was able to inhibit BoNT/B intoxication in vivo. During the development of the purification process, we found that a simple acidic wash step is able to remove ArnA, an E. coli protein that frequently contaminates His-tagged protein preparations due to its affinity for nickel ions. This finding has broad implications for the purification of other recombinant proteins. Materials and methods Ethics statement All animal experiments were performed in accordance with the Israeli law and were approved by the Ethics Committee for Animal Experiments at the Israel Institute for Biological Research. Materials All chemicals were purchased from Sigma–Aldrich or Merck, unless otherwise stated. Yeast extract and tryptone were obtained from Becton, Dickinson and Company (Franklin Lakes, NJ). Mouse
123
anti-HC/B monoclonal antibody was prepared as previously described [23]. Rabbit anti-HC/B polyclonal antibodies were purified from the sera of hyperimmune rabbits that were immunized with HC/B, as previously described [24]. A synthetic HC/B gene with optimized codon usage for expression in E. coli was synthesized by GenScript (Piscataway, NJ). Bacteria and toxins E. coli BL21(DE3) and the pET-9a plasmid were purchased from Novagen (Madison, WI). Clostridium botulinum strain B was obtained from the IIBR collection (B592). The neurotoxin gene of this strain complies with that of the Danish strain (Accession Number M81186) [25]. BoNT/B was prepared from concentrated supernatant of a culture grown for 6 days in anaerobic culture tubes. Growth of cultures for optimization experiments During optimization, the cells were grown with shaking (250 rpm) in 250-ml polycarbonate baffled shake flasks (Nalgene; Rochester, NY) containing 40 ml of media and kanamycin (30 lg/ ml). The cultures were inoculated with a 1:100 dilution of a starter culture grown in terrific broth (TB) media. The following media were evaluated: TB (12 g/l tryptone, 24 g/l yeast extract,0.4% (v/v) glycerol, and 89 mM potassium phosphate); TB + sorbitol (660 mM sorbitol, 12 g/l tryptone, 24 g/l yeast extract, 0.4% (v/v) glycerol, 89 mM potassium phosphate, and 2.5 mM betaine); and minimal media [26] (30 g/l glycerol, 13.3 g/l KH2PO4, 4 g/l (NH4)2HPO4,1.7 g/l citric acid, 1.2 g/l MgSO47H2O, 8.4 mg/l EDTA, 2.5 mg/l CoCl26H2O, 15 mg/l MnCl24H2O, 1.5 mg/l CuCl22H2O, 3 mg/l H3BO3, 2.5 mg/l Na2MoO42H2O, 13 mg/l Zn(CH3COO)22H2O, 100 mg/l Fe(III)citrate, and 4.5 mg/l thiamineHCl. Soluble HC/B quantification assay This assay was used to estimate the yield of the soluble HC fragment in cultures grown under various conditions during expression optimization. The assay included two steps. First, samples (0.5 ml) withdrawn from the cultures were chemically disrupted with the CelLytic B Plus Kit (Sigma–Aldrich) according to the manufacturer’s instructions, and the soluble proteins were separated from the insoluble cell fraction by centrifugation. The concentration of the HC fragment in the supernatants was then estimated by sandwich ELISA as follows. Plates (Maxisorp, Nunc; Roskilde, Denmark) were coated with 50 ll of a mouse anti-HC/B monoclonal antibody [23] diluted to a final concentration of 4 lg/ml in coating buffer (50 mM Na2CO3, pH 9.6) and then incubated overnight at 4 °C. The plates were subsequently washed with wash solution (0.9% NaCl, 0.05% Tween 20) and blocked for 1 h at 37 °C with TSTA buffer (50 mM Tris, 0.9% NaCl, 0.05% Tween 20, 2% BSA, 200 ll per well). After washing, the plates were incubated with serial dilutions (50 ll per well, in duplicate) of the tested supernatant and pure HC standard in TSTA for 1 h at 37 °C. The plates were then washed with wash solution and incubated for 1 h with a rabbit anti-HC fragment polyclonal antibody, diluted in TSTA to a final concentration of 0.5 lg/ml. After additional washing, the plates were incubated with 50 ll of alkaline phosphatase-conjugated donkey anti-Rabbit IgG (Jackson ImmunoResearch) diluted 1:1,500 for 1 h at 37 °C. Finally, the plates were washed with wash solution, and the colorimetric reaction was developed using the substrate p-nitrophenyl phosphate (1 mg/ml in 0.2 M Tris buffer). The absorbance at 405 nm was continuously measured for 15 min in 30-s intervals, and the concentration of the HC fragment was determined by interpolation from a standard curve of HC/B.
124
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
Purification of HC/B
In vivo inhibition test
For the development of HC/B purification process, E. coli BL21(DE3) carrying the pET-9a-HC/B plasmid was grown without induction in 2-l polycarbonate baffled shake flasks (Nalgene) containing 0.5 l of TB media supplemented with kanamycin (30 lg/ ml). The flasks were incubated at 18 °C with shaking (250 rpm) for 40 h. The cells were harvested by centrifugation, then resuspended in 100 ml of binding buffer (20 mM sodium phosphate, 0.5 M NaCl, 20 mM imidazole, pH 7.4), and disrupted by sonication. The cell extract was clarified by centrifugation (14,000g, 30 min) and loaded onto a HisTrap FF 1 ml column (GE Healthcare) mounted on an AKTA Explorer FPLC system (GE Healthcare). The column was washed with 10 column volumes (CV) of binding buffer, after which different conditions were tested for improving the purity level, as described in the Results section. The modified purification process of HC/B included the following steps: (1) loading of the cell lysate onto a Ni2+ column; (2) washing of unbound proteins with binding buffer (10 CV); (3) acidic washing of contaminants with acidic wash buffer (50 mM sodium phosphate, 50 mM NaCl, pH 4.0, 15 CV); (4) washing with binding buffer containing 40 mM imidazole (10 CV); and (5) elution of HC/B with elution buffer (20 mM sodium phosphate, 0.5 M NaCl, 500 mM imidazole, pH 7.4). Comparison of the purity levels obtained using the modified purification process with a standard process (lacking the acidic wash step) was performed using the Experion Automated Electrophoresis System (Bio-Rad; Hercules, CA) according to the manufacturer’s instructions. Mass spectrometry analysis of ArnA band was done at the Smoler Proteomics Center (Technion – Israel Institute of Technology, Haifa, Israel).
Mice (6 per group) were injected (1 ml, i.p.) with a 5 MsLD50 dose of BoNT/B in the presence (test group) or absence (control group) of different doses of Hc/B. Following injection, survival was monitored for 1 week.
Dialysis buffer optimization experiments These experiments were performed on HC/B eluted from a Ni2+ column in elution buffer (20 mM sodium phosphate, 0.5 M NaCl, 500 mM imidazole, pH 7.4). Samples of the protein (750 ll) were transferred to a Slide-A-Lyzer G2 Dialysis Cassette, 10,000 MWCO (Thermo Scientific). The cassettes were incubated in 750 ml of the tested buffer at 4 °C with stirring, for a total of three buffer changes, each for at least 2 h. Protein determination at the end of the process was performed with a NanoDrop Spectrophotometer (Thermo Scientific), using an extinction coefficient of 105,660 mM 1 cm 1 and molecular weight of 52.9 kDa (calculated using the ExPASy server [27]). Vaccination of mice with HC fragment For all vaccination experiments, the HC/B fragment was diluted with PBS and adsorbed to aluminum hydroxide (0.5% w/v final concentration of Al(OH)3). The vaccines were administered to CD-1 mice (Charles River UK) via the subcutaneous route. For the challenge protection test, mice were immunized once (5 lg of antigen), and after 21 days, they were subjected to a 102, 103, 104, 105, or 106 MsLD50 (mouse LD50) challenge, using three mice per challenge dose. For the potency study mice (5 per group) were vaccinated with nine different HC/B doses, ranging from 3.9 to 1000 ng (2-fold dilution factor between sequential groups). Twenty-one days after vaccination, the mice were challenged with a 103 MsLD50 dose, and their survival was monitored for 7 days. For the generation of neutralizing antibodies, mice (n = 7) were immunized three times with vaccine preparations containing 5 lg of protein per injection, at 21-day intervals. Two weeks after the last injection, mice were bled, and the sera were pooled. The neutralizing antibody concentration was determined according to the European Pharmacopoeia [28].
Statistical analysis A 4-parameter logistic regression model was employed to construct a standard curve for pure HC, using SoftMax Pro 5.4 (Molecular Devices, Sunnyvale, CA), for the estimation of expression levels. Calculation of the ED50 (effective dose protecting 50% of the mice) was carried out with GraphPad Prism 5 software (GraphPad software, San Diego, CA), employing non-linear regression analysis. Comparison of survival curves was conducted with the log-rank (Mantel-Cox) test using GraphPad Prism 5 software. Differences were considered significant at P < 0.05. Results and discussion Construction of the expression system and preliminary protein production The expression of genes originating from C. botulinum in E. coli often suffers from low yields since they exhibit distinct codon usage, which limits translation. To overcome this issue, a synthetic gene with optimized codon usage for the host was prepared. The gene encoded amino acids 859–1290 of BoNT/B (Danish strain). To facilitate purification, a 6His tag was added to the N-terminus of the protein. The gene was cloned into the pET-9a vector, which was found to be optimal for the expression of BoNT/A [20]. The corresponding pET-9a-HC/B plasmid was transformed into E. coli BL21(DE3). Preliminary attempt to produce HC/B included growing the bacterium in TB media without induction, overnight at 37 °C. These conditions were chosen because similar growth conditions resulted in high expression levels of the receptor-binding domain of BoNT/A (350 mg of purified HC/A from 1 l of culture) [20]. However, a poor yield of HC/B was obtained (1–2 mg HC/B per liter), and following purification by means of immobilized metal chelate affinity chromatography (IMAC), the protein contained a considerable amount of contaminants (purity level of 30%, Fig. 1). Analysis of the soluble and non-soluble cell fractions using SDS-PAGE showed that HC/B was the dominant protein in the pellet, while it could not be observed in the supernatant. Hence, the poor yield was due to the tendency of the protein to form insoluble aggregates, as previously reported by others [7,21]. Expression optimization To improve the soluble HC/B yield, a screening assay that allows evaluation of the effects of various expression conditions in a short time was developed. The assay involved small-scale (40 ml) bacterial cell growth under the tested conditions, a chemical cell lysis step, and quantification of HC/B in the soluble cell fraction using a specific ELISA test. A common strategy to improve soluble protein expression is reduction of the growth temperature. Under low temperature, the growth rate of the bacterium and the protein production rate are reduced, and the recombinant protein therefore has more time to undergo proper folding. Therefore, expression was tested at 37, 28, and 18 °C. Another way to reduce the growth rate is by growing the bacterium on minimal medium, in which it must synthesize basic building blocks that are readily available in rich media.
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
125
in the absence of an inducer, as a result of basal expression of T7 RNA polymerase. In other studies [7,21,22], the promoters used for HC/B included the lac operator element, and the expression vectors carried a copy of the lac repressor gene (lacI). Together, these components provide tighter control of gene expression, and HC/B expression requires induction with IPTG. Although a stringent control level can be advantageous for the expression of recombinant proteins that are detrimental to the host, for HC/B expression, the more relaxed control level is preferred. Since the recombinant gene is transcribed at a moderate rate by T7 RNA polymerase when present at a basal level, the protein has a better chance of folding properly than in induced systems in which the gene is transcribed at very high levels upon induction, and accumulation of the unfolded polypeptide occurs. For expression of the receptor-binding domain of BoNT/A, we previously obtained similar results, where the use of the T7 promoter improved soluble HC/A levels by two orders of magnitude in comparison with the T7lac promoter [20]. Fig. 1. Preliminary attempt to produce HC/B. E. coli BL21(DE3) harboring the pET9a-HC/B plasmid were grown in TB media overnight at 37 °C. Following cell disruption the protein was purified using IMAC. Samples from the purification process were subjected to SDS-PAGE (A) and Western blot (B) analyses: 1 – BoNT/B complex; 2 – soluble cell fraction; 3 – non-soluble cell fraction; 4 – the flow through of soluble cell fraction loading step to the IMAC column (unbound proteins); and 5 – the eluted protein. The HC fragment was detected using horse anti BoNT/B complex antibodies.
Accordingly, we tested the use of rich media (terrific broth) and minimal media [26], in which the generation times at 37 °C were 30 and 150 min, respectively. Induction of osmotic stress was also reported to improve soluble expression yields of several recombinant proteins [29–31]. It is assumed that under osmotic stress, the bacterium produces osmoprotectants or induces stress response mechanisms that affect the folding kinetics of the recombinant protein [29,30]. Therefore, the effect of this parameter on the soluble HC/B yield was evaluated by growing the bacterium in TB media containing 660 mM sorbitol. Soluble HC/B levels were most affected by the growth temperature (Table 1). When cells were grown at 37 °C, the volumetric yield was low (4.2 mg HC/B per liter of culture, TB media). Significant improvement in the yield was achieved when the temperature was reduced to 28 °C (51.7 mg/l), and the highest soluble HC/B level was obtained at 18 °C (77.8 mg/l). Induction with IPTG had a negative effect on soluble HC/B levels at all examined temperatures, which could be due to the rapid accumulation of unfolded polypeptides that are produced at very high levels following induction. The use of minimal media did not increase HC/B levels above those obtained with TB media. The specific volumetric yields of cultures grown with defined media were similar to those obtained with TB media. However, because defined media supported lower cell mass accumulation, the overall volumetric yield was also lower. Induction of osmotic shock did not improve soluble HC/B levels as well. The volumetric yield obtained under osmotic shock was similar to that obtained with regular growth in TB media at 18 °C. Because growth of the bacterium under osmotic shock was performed at 18 °C as well, it can be concluded that this treatment did not contribute to obtaining higher soluble protein levels beyond regular expression in TB media at 18 °C. The yields of soluble HC/B that we obtained at 18 °C are an order of magnitude higher than those previously reported for the expression of the protein in E. coli [7,21]. The main difference between our expression system and those described in other reports is the control level of recombinant gene expression. In our system, the transcription of the HC/B gene is controlled by the T7 promoter. This system is leaky, and recombinant gene expression occurs even
Development of HC/B purification process To facilitate the purification of HC/B, the recombinant gene was designed to contain a 6xHis tag at its N-terminus. Recombinant vaccines against malaria that were produced in E. coli and contained a His tag have been found to be safe in phase I clinical trials [32,33]. Although purification of His-tagged proteins on a Ni2+ column usually results in a protein with satisfactory purification level, histidine-rich host cell proteins can often contaminate the preparation [34]. Hence, in some cases, additional chromatographic steps are required to improve the purity level. In our hands, applying a standard IMAC purification procedure resulted in a poor purity of HC/B (Fig. 1). One strategy for removing these contaminants is to perform a wash step prior to the elution step, with an intermediate imidazole concentration that will release the contaminants without eluting the target protein. To determine the optimal purification conditions for HC/B, a cell lysate of E. coli BL(21) + pET-9a-HC/B grown in TB media at 18 °C was loaded onto a 1-ml Ni2+ column, and the column was washed with increasing imidazole concentrations (30–100 mM in 10 mM steps). Finally, the remaining proteins were eluted with 0.5 M imidazole (Fig. 2). Neither of the tested imidazole concentrations was found to be suitable for obtaining satisfactory purity level of HC/B. At imidazole concentrations lower than 60 mM, only a single 90 kDa contaminant was completely removed. Several proteins were washed out at 60–90 mM imidazole, and others were released from the column only at 0.5 M imidazole. Moreover, at all of the tested imidazole concentrations, HC/B was released along with the contaminants, and above a concentration of 50 mM, significant leakage occurred. We next attempted to separate HC/B from the contaminants by washing the column with a low pH buffer. Under pH values lower than the pKa of histidine (6.0), the residue is protonated, and the affinity toward nickel ions is therefore reduced, and proteins can be released from the column. To find a pH value that differentially releases the contaminants and HC/B from the column, a cell lysate was loaded onto a nickel column, and the pH of the wash buffer was decreased from 7.0 to 4.0 by mixing phosphate buffers at pH 7.0 and 4.0 in different ratios. While HC/B was not eluted at any of the tested pH values (7.0–4.0), several contaminants were released from the column at pH 5.5–6.0. Further decreasing the pH with citrate buffer to 3.0 resulted in the formation of a white precipitate in the column, without releasing HC/B. Because several contaminants were released from the column at pH 5.5–6.0 without leakage of HC/B, we wished to examine whether the addition of an acidic wash step to a standard purification process could increase the HC/B purity level. Therefore, we compared the purity obtained through a standard purification
126
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
Table 1 Comparison of soluble HC/B expression levels obtained at different conditions.
a b c d
Growth temperature (°C)
Mediaa
A600nm
37 37 37 37 28 28 18 18 18 18
TB TB TB MM TB MM TB TB TB TB
19.5 20.2 23.3 8 24.3 8.6 19.9 23.7 19.7 25.9
Induction (+/ )b + +
+ +
HC/B yield (mg/l)c
HC/B specific yield (mg/lO.D.)
4.2 2.7 2.3 2.7 52 9.5 78 30 61 43
0.2 0.1 0.1 0.3 2.1 1.1 3.9 1.3 3.1 1.7
Comments Induction at A600nm = 1.0d Induction at A600nm = 5.0d
Induction at A600nm = 0.6 Osmotic shock Osmotic shock + Induction at A600nm = 0.6
TB – terrific broth; MM – minimal media. With IPTG to a final concentration of 0.4 mM (37 °C) or 0.1 mM (18 °C). HC fragment yield values are expressed as mg protein per 1 liter of culture. Following induction samples were withdrawn and analyzed after 2, 3, 4, 5, 6, and 20 h. The results with the highest HC/B yield are shown.
Fig. 2. Elution of HC/B and host cell proteins from Ni2+ column at different imidazole concentrations. A cell lysate of E. coli BL21(DE3) carrying the pET-9a-HC/B plasmid and grown at 18 °C in TB media was loaded onto Ni2+ column and then it was washed with increasing imidazole concentrations (30–100 mM in 10 mM steps). Elution of the remaining proteins was done with 0.5 M imidazole. (A) The chromatogram of the purification process (A280nm – grey; A260nm – black; and imidazole concentration – dashed) (B) SDS-PAGE analysis of protein sample eluted at different imidazole concentrations: 1 – 30 mM; 2 – 40 mM; 3 – 50 mM; 4 – 60 mM; 5 – 70 mM; 6 – 80 mM; 7 – 90 mM; 8 – 100 mM; 9 – 500 mM. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
process with a modified process that included a wash step with phosphate buffer at pH 4.0 (Fig. 3). Both processes were carried out using equal volumes of the same cell lysate. The purity level of HC/B obtained via the standard process was 63% (estimated using automated electrophoresis analysis). This preparation included a main contaminant of 74 kDa that accounted for 12.5% of the total proteins in the preparation and two minor contaminants of 22 and 25 kDa. The acidic wash significantly increased HC/B purity (85%). This improvement was achieved mainly due to complete removal of the main contaminant at 74 kDa during the acidic wash step. In addition, the presence of other contaminants was reduced as well during the acidic wash step. Overlaying the electropherograms of the two preparations revealed that the peak of HC/B purified via the modified process was slightly higher than that of HC/B purified by the standard process. Therefore, no loss of HC/B occurred during the acidic wash. ArnA is the main contaminant removed by the acidic wash The addition of the acidic wash step to the standard purification process significantly improved the purity level of HC/B. This washing step is simple, economical, does not hamper the yield, and avoids the inclusion of an additional chromatographic step. The main contribution of the acidic wash was the removal of the main contaminant of 74 kDa. To identify this contaminant, the corresponding band was excised from a gel and analyzed via mass spectrometry. The protein was identified as ArnA, an E. coli enzyme involved in the biosynthesis of 4-amino-4-deoxy-L-arabinose
[35]. ArnA is considered to be one of the two most abundant E. coli proteins (along with SlyD) that are co-purified with recombinant His-tagged proteins [36]. ArnA contains histidine-rich patches on its surface, which are believed to interact with Ni2+ ions. Commonly, elution of ArnA from a Ni2+ column occurs at imidazole concentrations of 55–80 mM [34]. Because HC/B began to be eluted from the column at 50 mM imidazole, it was surprising that under an acidic wash, ArnA was completely removed from the column, while HC/B remained bound. The cause of this result could be explained by differences between ArnA and Hc/B in the nature of their interactions with nickel ions. The tertiary structure of ArnA is a hexamer, consisting of a dimer of two trimers [37] (Fig. 4). The histidine-rich patches found in ArnA are located at the center of each trimer, and are formed by spatial proximity of histidine residues from each monomer. It is possible that by reducing the pH, the hexameric structure of ArnA dissociated into monomers, and the interaction with nickel ions was therefore lost. On the other hand, in HC/B, the histidine residues are in linear proximity, and their interaction with nickel ions is therefore not dependent on the spatial structure of the protein. Recently, two research groups developed E. coli strains that allow purification of His-tagged protein with reduced host cell protein contamination. Robichon et al., from New England BioLabs, developed the commercial strain NiCo21 [38]. In this strain, ArnA is fused to a chitin binding domain. Following the standard IMAC purification process, ArnA is separated from the recombinant protein on a chitin column, a step that can hamper the overall yield and increase process costs. Anderson et al. developed the strain
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
127
Fig. 3. Acidic wash of column-bound HC/B increases its purity. (A) Electropherograms overlay of HC/B purified using standard His-tagged protein purification process (grey) and using a modified process that included an acidic wash step (black). (B) SDS-PAGE analysis of HC/B purified using a standard and modified purification process. 1 – supernatant of broken cells; 2 – pellet of broken cells; 3 – HC/B purified using standard process; 4 – HC/B purified using a modified process that included an acidic wash step. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 4. Distribution of histidine residues on the surface of the contaminant ArnA. ArnA is a hexamer that is formed by a dimer of two trimers. The protein contains a histidinerich patch on its surface at the trimer interface. This patch is likely to accounts for the ArnA binding to Ni2+ ions. At low pH the protein may dissociate to monomers, thereby disrupting the interaction with nickel. The Fig. was prepared with PyMOL molecular graphics system using the ArnA structure published by Gatzeva-Topalova et al. [37] (pdb code 1Z7E). The ArnA monomers are colored differently and histidine residues are colored blue.
LOBSTR (low background strain) in which the affinity of ArnA toward nickel ions was reduced through mutagenesis, and ArnA is released from the column at 30 mM imidazole [36]. Both genetically engineered strains are derivatives of E. coli BL21(DE3), which is the most prevalent strain used for protein expression. However, there are additional E. coli strains that are employed for recombinant protein expression, which bear special characteristics. For example, the Rosetta (Novagen) and Codonplus (Stratagene) strains enable the expression of genes with rare codons for E. coli, and the Origami (Novagen) strain is used for the expression of proteins that contain disulfide bonds. In cases where the recombinant protein must be expressed in a host other than BL21(DE3), the use of LOBSTR or NiCo21 is not applicable. Nevertheless, removal of contaminants through an acidic wash is generic to all E. coli strains, as they contain the same contaminants. In addition, an acidic wash may be beneficial for the purification of His-tagged proteins expressed in other systems, such as P. pastoris or Chinese hamster ovary. Optimization of the HC/B formulation Following the purification of HC/B using a Ni2+ column, it was found in the elution buffer. This buffer contained high salt and imidazole concentrations (0.5 M each), and the protein therefore had to be dialyzed to a buffer that is suitable for injection. Eluted HC/B was initially dialyzed against 50 mM sodium phosphate,
50 mM sodium chloride, pH 6.5 (buffer 1, Table 2). At the end of the dialysis process, a considerable amount of protein precipitate was obtained. Precipitate formation could be observed as early as 2 h after the beginning of the dialysis process. The estimated loss of HC/B during dialysis was 50%. Therefore, we wished to determine the basis for HC/B precipitation and reduce its loss by optimizing the composition of the dialysis buffer. Following the purification of HC/B using IMAC, the protein was divided into separate dialysis cassettes, each of which was dialyzed against a different buffer (Table 2). At the end of the dialysis process, the protein solutions were centrifuged to remove the precipitated protein, and their protein contents were determined. The solubility of HC/B was most strongly affected by the salt concentration. At a high salt concentration (0.5 M NaCl, buffer 4), precipitation of HC/B was not observed, whereas in a buffer with the same composition except for the salt content (50 mM, buffer 1), 49% of the protein precipitated. The mechanism underlying this stabilization by a high salt concentration is presumably shielding of charges on the protein surface, thereby reducing intermolecular interactions (also known as ‘‘salting in’’). Although a 100% yield was obtained with buffer 4, its high salt content makes it less optimal for injection. The solubility of HC/B was also affected by the pH of the buffer. Major precipitation occurred when using Tris buffer (buffer 5, pH 8.0), while the protein was more stable when dialyzed against acidic buffers (yields of 87% for buffer 2, pH 5.5, and 68% for buffer 3, pH 5.0). It is possible that the precipitation observed for
128
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
Table 2 Yields of HC/B following dialysis against different buffers compositions. Buffer composition
Protein concentration (mg/ml)
Yield (%)
(1) 50 mM sodium phosphate, 50 mM NaCl, pH 6.5 (2) 50 mM sodium citrate, pH 5.5 (3) 50 mM sodium acetate, pH 5.0 (4) 50 mM sodium phosphate, 0.5 M NaCl, pH 6.5 (5) 20 mM TrisHCl, 50 mM NaCl, pH 8.0
2.7
51
4.6 3.6 5.3
87 68 100
1.9
35
Fig. 5. Dose-response relationships in mice vaccinated with HC/B. Mice (5 per group) were vaccinated with nine different HC/B doses, ranging from 3.9 to 1000 ng (2-fold dilution factor between sequential groups). Twenty one days after the vaccination the mice were challenged with 103 MsLD50, and survival was monitored for 7 days. The ED50 of HC/B was 58 ng, with a 95% confident limits of 23.5–128 ng.
a dose of 103 LD50, and in the work of Zichel et al., mice were protected against challenge with a dose of 106 LD50 after vaccination with 5 lg HC/B. However, in these works, mice received three HC/ B injections before toxin challenge, whereas in our work, mice were protected after a single vaccination. To study the potency of HC/B, mice were vaccinated once with the protein at nine different doses ranging from 3.9 to 1000 ng and then challenged with 103 MsLD50 of BoNT/B (Fig. 5). At doses below 16 ng, no survival was observed, while 100% survival was obtained at doses higher than 250 ng. The calculated ED50 (effective dose protecting 50% of the mice) of HC/B was 58 ng, with 95% confident limits of 23.5 and 128 ng. These results are in accordance with other reports concerning HC/B potency. Potter et al. [39] obtained full protection against challenge with 104 MsLD50 at HC/B doses higher than 625 ng. Gao et al. [22] reported full protection of mice against challenge with 103 MsLD50 at HC/B doses above 250 ng and no survival below 2 ng. However, in their study, the vaccination regime included three injections with HC/B, rather than a single injection, as in our study. To test the ability of HC/B to generate neutralizing antibodies, a group of seven mice was vaccinated three times with 5 lg of aluminum hydroxide-adsorbed HC/B. The neutralizing antibody concentration in the pooled sera of the mice was 50 IU/ml. This value is consistent with previous reports. Smith et al. reported that three vaccinations of mice with 1 lg of HC/B produced using P. pastoris generated a neutralizing antibody concentration of 28.7 IU/ml [5]. Gao et al. [22] vaccinated mice with HC/B expressed in E. coli and produced from inclusion bodies. Following three vaccinations with a dose of 1.25 lg, the neutralizing antibody concentration was 17.1 IU/ml, and when the HC/B dose was 6.25 lg, the concentration was 19.2 IU/ml. In vivo inhibition of botulinum neurotoxin B by the HC fragment
Fig. 6. HC/B inhibits mice intoxication with BoNT/B. Mice (n = 6) were exposed to 5 MsLD50 of BoNT/B in the presence (broken line) or absence (solid line) of HC/B, and survival was monitored. Time to death was significantly delayed in the HC/B-treated group (P = 0.0057 by log-rank test).
buffer 5 was due to the proximity of the pH of the buffer (8.0) to the isoelectric point of the protein (calculated pI of 7.21, using the ExPASy server [27]). Hence, we chose to use citrate buffer, pH 5.5, for HC/B formulation. SDS-PAGE analysis of HC/B dialyzed against the different buffers showed that the contaminant content in the preparations was inversely related to the HC/B yield. Thus, the precipitate obtained during dialysis is of HC/B, while the contaminants remain stable. Therefore, reducing precipitation improves the yield of the process as well the purity level of the protein. Immunogenicity of HC/B Single vaccination of mice with 5 lg HC/B resulted in full protection against challenge with 102–106 MsLD50 BoNT/B. This protection is at least equivalent to that of HC/B produced using other expression systems. Baldwin et al. reported that vaccination of mice with 1 lg HC/B conferred protection against challenge with
We have recently demonstrated that HC/A can inhibit intoxication with BoNT/A and BoNT/E in vivo [20]. The presumed mechanism of this inhibition is that HC/A competes with the toxins for receptor binding (SV2). HC/A was not able to inhibit intoxication with BoNT/B, as this toxin serotype utilizes a different receptor (synaptotagmin) to enter the cells [40]. To determine whether BoNT/B intoxication can be inhibited by HC/B, mice (n = 6) were exposed to 5 MsLD50 of the toxin in the presence or absence of HC/B (1 mg), and their survival was monitored (Fig. 6). HC/B significantly delayed intoxication by BoNT/B. In the absence of HC/B, the time to death (TTD) ranged from 23.5 to 40 h, and the median TTD was 26 h, while in the presence of HC/ B, the TTD ranged from 36 to 60 h, and the median TTD was 43.4 h (P = 0.0057). One animal in the HC/B-treated group survived the challenge. The ability of HC/B to compete with the toxin for receptor binding indicates that the recombinant protein is expressed and folded into its native three dimensional structure. This characteristic is of great importance because, as an antigen, HC/B displays spatial epitopes that may contribute to the generation of neutralizing antibodies. Conclusions In this work, an efficient expression system for the receptorbinding domain of BoNT/B was developed. The system provides an order of magnitude improvement in the soluble protein yield compared with previously reported works, and it is simple and economical. The protein exhibits good immunogenicity, and its protective properties are at least comparable to those of HC/Bs produced using other expression systems. Furthermore, the native function of the protein is preserved, as it is able to inhibit BoNT/
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
B intoxication in vivo. In addition, we found that the contaminant ArnA can be removed through a simple wash step with an acidic buffer, thereby avoiding the need for an additional chromatographic step. As ArnA is a common contaminant of His-tagged proteins produced in E. coli, we believe that this finding may aid in the purification of other recombinant proteins. Acknowledgment This work was funded by Israel Institute for Biological Research grant SB/5122-18. References [1] R.G. Darling, C.L. Catlett, K.D. Huebner, D.G. Jarrett, Threats in bioterrorism I: CDC category a agents, Emerg. Med. Clin. North Am. 20 (2002) 273–309. [2] L.A. Smith, J.M. Rusnak, Botulinum neurotoxin vaccines: past, present, and future, Crit. Rev. Immunol. 27 (2007) 303–318. [3] J.M. Rusnak, L.A. Smith, Botulinum neurotoxin vaccines: past history and recent developments, Hum. Vaccin. 5 (2009) 794–805. [4] M. Montal, Botulinum neurotoxin a marvel of protein design, Annu. Rev. Biochem. 79 (2010) 591–617. [5] L.A. Smith, Botulism and vaccines for its prevention, Vaccine 27 (2009) D33– D39. [6] M.A. Clayton, J.M. Clayton, D.R. Brown, J.L. Middlebrook, Protective vaccination with a recombinant fragment of Clostridium botulinum neurotoxin serotype A expressed from a synthetic gene in Escherichia coli, Infect. Immun. 63 (1995) 2738–2742. [7] M.R. Baldwin, W.H. Tepp, A. Przedpelski, C.L. Pier, M. Bradshaw, E.A. Johnson, J.T. Barbieri, Subunit vaccine against the seven serotypes of botulism, Infect. Immun. 76 (2008) 1314–1318. [8] J.D. Shearer, M.L. Vassar, W. Swiderski, K. Metcalfe, N. Niemuth, I. Henderson, Botulinum neurotoxin neutralizing activity of immune globulin (IG) purified from clinical volunteers vaccinated with recombinant botulinum vaccine (rBV A/B), Vaccine 28 (2010) 7313–7318. [9] H.F. LaPenotiere, M.A. Clayton, J.L. Middlebrook, Expression of a large, nontoxic fragment of botulinum neurotoxin serotype A and its use as an immunogen, Toxicon 33 (1995) 1383–1386. [10] J. Boles, M. West, V. Montgomery, R. Tammariello, M.L.M. Pitt, P. Gibbs, L. Smith, R.D. LeClaire, Recombinant C fragment of botulinum neurotoxin B serotype (rBoNTB (HC)) immune response and protection in the rhesus monkey, Toxicon 47 (2006) 877–884. [11] M.P. Byrne, T.J. Smith, V.A. Montgomery, L.A. Smith, Purification, potency, and efficacy of the botulinum neurotoxin type A binding domain from Pichia pastoris as a recombinant vaccine candidate, Infect. Immun. 66 (1998) 4817– 4822. [12] M.P. Byrne, R.W. Titball, J. Holley, L.A. Smith, Fermentation, purification, and efficacy of a recombinant vaccine candidate against botulinum neurotoxin type F from Pichia pastoris, Protein Expr. Purif. 18 (2000) 327–337. [13] M.P. Dux, R. Barent, J. Sinha, M. Gouthro, T. Swanson, A. Barthuli, M. Inan, J.T. Ross, L.A. Smith, T.J. Smith, R. Webb, B. Loveless, I. Henderson, M.M. Meagher, Purification and scale-up of a recombinant heavy chain fragment C of botulinum neurotoxin serotype E in Pichia pastoris GS115, Protein Expr. Purif. 45 (2006) 359–367. [14] M.P. Dux, J. Huang, R. Barent, M. Inan, S.T. Swanson, J. Sinha, J.T. Ross, L.A. Smith, T.J. Smith, I. Henderson, M.M. Meagher, Purification of a recombinant heavy chain fragment C vaccine candidate against botulinum serotype C neurotoxin [rBoNTC(Hc)] expressed in Pichia pastoris, Protein Expr. Purif. 75 (2011) 177–185. [15] K.J. Potter, W. Zhang, L.A. Smith, M.M. Meagher, Production and purification of the heavy chain fragment C of botulinum neurotoxin, serotype A, expressed in the methylotrophic yeast Pichia pastoris, Protein Expr. Purif. 19 (2000) 393– 402. [16] M.R. Baldwin, W.H. Tepp, C.L. Pier, M. Bradshaw, M. Ho, B.A. Wilson, R.B. Fritz, E.A. Johnson, J.T. Barbieri, Characterization of the antibody response to the receptor binding domain of botulinum neurotoxin serotypes A and E, Infect. Immun. 73 (2005) 6998–7005. [17] Y.Z. Yu, S.M. Zhang, Y. Ma, H.Q. Zhu, W.B. Wang, Y. Du, X.W. Zhou, R.L. Wang, S. Wang, W.Y. Yu, P.T. Huang, Z.W. Sun, Development and evaluation of candidate vaccine and antitoxin against botulinum neurotoxin serotype F, Clin. Immunol. 137 (2010) 271–280.
129
[18] R. Chen, J. Shi, K. Cai, W. Tu, X. Hou, H. Liu, L. Xiao, Q. Wang, Y. Tang, H. Wang, Improved soluble expression and characterization of the Hc domain of Clostridium botulinum neurotoxin serotype A in Escherichia coli by using a PCR-synthesized gene and a Trx co-expression strain, Protein Expr. Purif. 71 (2010) 79–84. [19] Y.Z. Yu, Z.W. Sun, S. Wang, W.Y. Yu, High-level expression of the Hc domain of clostridium botulinum neurotoxin serotype A in Escherichia coli and its immunogenicity as an antigen, Chin. J. Biotechnol. 23 (2007) 812–817. [20] A. Ben David, E. Diamant, A. Barnea, O. Rosen, A. Torgeman, R. Zichel, The receptor binding domain of botulinum neurotoxin serotype A (BoNT/A) inhibits BoNT/A and BoNT/E intoxications in vivo, Clin. Vacc. Immun. 20 (2013) 1266–1273. [21] D.M. Held, A.C. Shurtleff, S. Fields, C. Green, J. Fong, R.G.A. Jones, D. Sesardic, R. Buelow, R.L. Burke, Vaccination of rabbits with an alkylated toxoid rapidly elicits potent neutralizing antibodies against botulinum neurotoxin serotype B, Clin. Vacc. Immun. 17 (2010) 930–936. [22] Y.L. Gao, S. Gao, L. Kang, C. Nie, J.L. Wang, Expression of Hc fragment from Clostridium botulinum neurotoxin serotype B in Escherichia coli and its use as a good immunogen, Hum. Vaccin. 6 (2010) 462–466. [23] E. Diamant, B.E. Lachmi, A. Keren, A. Barnea, H. Marcus, S. Cohen, A. Ben David, R. Zichel, Evaluating the synergistic neutralizing effect of anti-botulinum oligoclonal antibody preparations, PLoS One 9 (2014) e87089. [24] R. Zichel, A. Mimran, A. Keren, A. Barnea, I. Steinberger-Levy, D. Marcus, A. Turgeman, S. Reuveny, Efficacy of a potential trivalent vaccine based on Hc fragments of botulinum toxins A, B, and E produced in a cell-free expression system, Clin. Vacc. Immun. 17 (2010) 784–792. [25] S.M. Whelan, M.J. Elmore, N.J. Bodsworth, J.K. Brehm, T. Atkinson, N.P. Minton, Molecular cloning of the Clostridium botulinum structural gene encoding the type B neurotoxin and determination of its entire nucleotide sequence, Appl. Environ. Microbiol. 58 (1992) 2345–2354. [26] D.J. Korz, U. Rinas, K. Hellmuth, E.A. Sanders, W.D. Deckwer, Simple fed-batch technique for high cell density cultivation of Escherichia coli, J. Biotechnol. 39 (1995) 59–65. [27] E. Gasteiger, A. Gattiker, C. Hoogland, L. Ivani, R.D. Appel, A. Bairoch, ExPASy: the proteomics server for in-depth protein knowledge and analysis, Nucleic Acids Res. 31 (2003) 3781–3788. [28] European Pharmacopoeia, Botulinum antitoxin, 2011. [29] Y. Fundoiano-Hershcovitz, L. Rabinovitch, Y. Langut, V. Reiland, G. Shoham, Y. Shoham, Identification of the catalytic residues in the double-zinc aminopeptidase from Streptomyces griseus, FEBS Lett. 571 (2004) 192–196. [30] S. Picaud, M.E. Olsson, P.E. Brodelius, Improved conditions for production of recombinant plant sequitepene synthase in Escherichia coli, Protein Expr. Purif. 51 (2007) 71–79. [31] N. Oganeysyan, I. Ankoudinova, S.H. Kim, R. Kim, Effect of osmotic stress and heat shock in recombinant protein overexpression and crystallization, Protein Expr. Purif. 52 (2007) 280–285. [32] C.F. Ockenhouse, E. Angov, K.E. Kester, C. Diggs, L. Siosson, J.F. Cummings, Phase I safety and immunogenecity trial of FMP1/AS02A, a Plasmodium falciparum MSP-1 asexual blood stage vaccine, Vaccine 24 (2006) 3009–3017. [33] M.A. Thera, O.K. Doumbo, D. Coulibaly, D.A. Diallo, A.K. Kone, Safety and immunogenecity of an AMA-1 malaria vaccine in malian adults: results of a phase 1 randomized control trial, PLoS One 3 (2008) e1465. [34] V.M. Bolanos-Garcia, O.R. Davies, Structural analysis and classification of native proteins from E. coli commonly co-purified by immobilised metal affinity chromatography, Biochim. Biophys. Acta 1760 (2006) 1304–1313. [35] G.J. Williams, S.D. Breazeale, C.R.H. Raetz, J.H. Naismith, Structure and function of both domains of ArnA, a dual function decarboxylase and a formyltransferase, involved in 4-amino-4-deoxy-L-arabinose biosynthesis, J. Biol. Chem. 280 (2005) 23000–23008. [36] K.R. Andersen, N.C. Leska, T.U. Schwartz, Optimized E. coli expression strain LOBSTR eliminates common contaminants from his-tag purification, Proteins 81 (2013) 1857–1861. [37] P.Z. Gatzeva-Topalova, A.P. May, M.C. Sousa, Structure and mechanism of ArnA: conformational change implies ordered dehydrogenase mechanism for polymyxin resistance, Structure 13 (2005) 929–942. [38] C. Robichon, J. Luo, T.B. Causey, J.S. Benner, J.C. Samuelson, Engineering Escherichia coli BL21(DE3) derivatives strains to minimize E. coli protein contamination after purification by immobilized metal affinity chromatography, Appl. Environ. Microbiol. 77 (2011) 4634–4646. [39] K.J. Potter, M.A. Bevins, E.V. Vassilieva, V.R. Chiruvolu, T. Smith, L.A. Smith, M.M. Meagher, Production and purification of the heavy-chain fragment C of botulinum neurotoxin, serotype B, expressed in the methylotrophic yeast Pichia pastoris, Protein Expr. Purif. 13 (1998) 357–365. [40] M. Dong, D.A. Richard, M.C. Goodnough, W.H. Tepp, E.A. Johnson, E.R. Chapman, Synaptotagmins I and II mediate entry of botulinum neurotoxins B into cells, J. Cell Biol. 162 (2003) 1293–1303.
Contents lists available at ScienceDirect
Protein Expression and Purification journal homepage: www.elsevier.com/locate/yprep
Expression, purification and characterization of the receptor-binding domain of botulinum neurotoxin serotype B as a vaccine candidate Alon Ben David, Amram Torgeman, Ada Barnea, Ran Zichel ⇑ Department of Biotechnology, Israel Institute for Biological Research, Ness Ziona, Israel
a r t i c l e
i n f o
Article history: Received 29 January 2015 and in revised form 8 February 2015 Available online 26 February 2015 Keywords: Botulinum vaccine Recombinant protein expression His-tag protein purification
a b s t r a c t The receptor-binding domain of botulinum neurotoxins (the HC fragment) is a promising vaccine candidate. Among the HC fragments of the seven BoNT serotypes, the expression of HC/B in Escherichia coli is considered especially challenging due to its accumulation as a non-soluble protein aggregate. In this study, the effects of different parameters on the expression of soluble HC/B were evaluated using a screening assay that included growing the bacterium at a small scale, a chemical cell lysis step, and a specific ELISA. The highest soluble HC/B expression levels were obtained when the bacterium E. coli BL21(DE3) + pET-9a-HC/B was grown in terrific broth media at 18 °C without induction. Under these conditions, the yield was an order of magnitude higher than previously reported. Standard purification of the protein using a nickel column resulted in a low purity of HC/B. However, the addition of an acidic wash step prior to protein elution released a major protein contaminant and significantly increased the purity level. Mass spectrometry analysis identified the contaminant as ArnA, an E. coli protein that often contaminates recombinant His-tagged protein preparations. The purified HC/B was highly immunogenic, protecting mice from a 106 LD50 challenge after a single vaccination and generating a neutralizing titer of 50 IU/ml after three immunizations. Moreover, the functionality of the protein was preserved, as it inhibited BoNT/B intoxication in vivo, presumably due to blockade of the neurotoxin protein receptor synaptotagmin. Ó 2015 Elsevier Inc. All rights reserved.
Introduction Botulism is a neurologic disease caused by toxins mainly produced by the bacterium Clostridium botulinum. Although botulism cases are rare in humans, botulinum neurotoxins (BoNTs1) are the most poisonous substances known, and they are therefore categorized as a tier 1 select agent by the Center for Disease Control and prevention (CDC) [1]. There are seven serologically distinct serotypes of these neurotoxins (designated A–G), among which serotypes A, B, and E are the most abundant in cases of human intoxication [2]. BoNTs are 150-kDa proteins, consisting of a 100kDa heavy chain (H) joined to a 50-kDa light chain (L) via a disulfide bond. The toxins of all serotypes share a similar architecture, organized into three structural domains that mediate the three steps of the intoxication process. The first step is the attachment of the
⇑ Corresponding author at: Israel Institute for Biological Research, P.B. 19, Ness Ziona 74100, Israel. Tel.: +972 8 9381513; fax: +972 8 9381761. E-mail address: [email protected] (R. Zichel). 1 Abbreviation used: BoNTs, botulinum neurotoxins; CDC, Center for Disease Control and prevention; rBV, recombinant botulinum vaccine; TB, terrific broth; CV, column volumes; TTD, time to death. http://dx.doi.org/10.1016/j.pep.2015.02.008 1046-5928/Ó 2015 Elsevier Inc. All rights reserved.
receptor-binding domain, located at the C-terminus of the heavy chain (the HC fragment), to its receptors and subsequent internalization via endocytosis. The next step is the translocation and release of the light chain into the cytosol, a step considered to be facilitated by the translocation domain found on the N-terminus of the heavy chain (HN). The final step is the cleavage of one of three SNARE (soluble N-ethylmaleimide sensitive factor attachment protein receptor) proteins by the light chain, which possesses endopeptidase activity, thereby inhibiting the release of the neurotransmitter acetylcholine from nerve cells to synapses [2–4]. The prevention and treatment of botulism is based on the presence of toxin-specific neutralizing antibodies in the blood stream. The source of the antibodies can be either self, in the case of prophylactic vaccination, or external (usually of vaccinated equine origin), in cases of post-exposure treatment. Historically, vaccines against botulism have consisted of formalin-inactivated toxins (toxoid) adsorbed to aluminum hydroxide [3]. Although toxoidbased botulinum vaccines are safe and efficient [3,5], their manufacturing process is expensive because it requires large-scale production facilities for this highly hazardous and spore forming agent. Therefore, current efforts to develop botulism vaccines are mainly focused on subunit vaccines consisting of the HC fragment.
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
The HC fragment is not toxic by itself, but it contains many neutralizing epitopes and can be produced recombinantly. In 1995, Clayton et al. reported that vaccination of mice with a recombinant HC fragment of neurotoxin serotype A elicited protective immunity against challenge with its homologous toxin [6]. Since that time, extensive research efforts have been made, and the receptor-binding domains of the seven BoNT serotypes have been produced and shown to induce a protective immune response [5,7]. Moreover, a recombinant botulinum vaccine (rBV) composed of the HC fragments of botulinum neurotoxins A and B, produced by the DynPort Vaccine Company for the United States Department of Defense, is currently under clinical investigation [8]. The first attempts to produce recombinant HC fragments of neurotoxin A were performed using Escherichia coli as a host [6,9]. However, as most of the expressed protein was insoluble in this system, subsequent studies have used the alternative host Pichia pastoris. Using this system, the HC fragments of botulinum A, B, C, D, E, and F have been produced with satisfactory yields [10–15]. Recently, the interest in expressing HC fragments in E. coli has returned, with better success. In the expression of the receptor domains of BoNT/E and BoNT/F, yields of 8–20 mg of protein per liter of culture were obtained [7,16,17]. Greater efforts have been made regarding HC/A expression. Several groups reported yields of 10–70 mg of HC/A per liter of culture [7,18,19], and we recently developed an expression system that yields 350 mg of HC/A per liter [20]. Nevertheless, expression of the receptor-binding domain of BoNT/B has remained challenging, as the reported yields of the expression of this domain are much lower. In the work of Baldwin et al. [7], the HC fragments of seven BoNT serotypes were expressed in E. coli to produce a heptavalent botulinum vaccine. The authors described special difficulty in expressing HC/B, as most of the protein was found in the insoluble cell fraction, with only 2 mg of the protein being obtained from 1 liter of culture. Similar yields were reported by Held et al. (1 mg per liter) [21]. Gao et al. purified HC/B from inclusion bodies with higher yields [22]. However, proper refolding of the protein into its native structure could not be achieved, and the final protein was obtained in 2 M urea. In this study, we developed a simple, efficient, and economic process for expressing HC/B in a soluble form in E. coli, achieving a yield that is an order of magnitude higher than previously reported. The protective properties of the recombinant product are at least comparable to those of HC/Bs produced using other expression systems, and the product is functionally active, as it was able to inhibit BoNT/B intoxication in vivo. During the development of the purification process, we found that a simple acidic wash step is able to remove ArnA, an E. coli protein that frequently contaminates His-tagged protein preparations due to its affinity for nickel ions. This finding has broad implications for the purification of other recombinant proteins. Materials and methods Ethics statement All animal experiments were performed in accordance with the Israeli law and were approved by the Ethics Committee for Animal Experiments at the Israel Institute for Biological Research. Materials All chemicals were purchased from Sigma–Aldrich or Merck, unless otherwise stated. Yeast extract and tryptone were obtained from Becton, Dickinson and Company (Franklin Lakes, NJ). Mouse
123
anti-HC/B monoclonal antibody was prepared as previously described [23]. Rabbit anti-HC/B polyclonal antibodies were purified from the sera of hyperimmune rabbits that were immunized with HC/B, as previously described [24]. A synthetic HC/B gene with optimized codon usage for expression in E. coli was synthesized by GenScript (Piscataway, NJ). Bacteria and toxins E. coli BL21(DE3) and the pET-9a plasmid were purchased from Novagen (Madison, WI). Clostridium botulinum strain B was obtained from the IIBR collection (B592). The neurotoxin gene of this strain complies with that of the Danish strain (Accession Number M81186) [25]. BoNT/B was prepared from concentrated supernatant of a culture grown for 6 days in anaerobic culture tubes. Growth of cultures for optimization experiments During optimization, the cells were grown with shaking (250 rpm) in 250-ml polycarbonate baffled shake flasks (Nalgene; Rochester, NY) containing 40 ml of media and kanamycin (30 lg/ ml). The cultures were inoculated with a 1:100 dilution of a starter culture grown in terrific broth (TB) media. The following media were evaluated: TB (12 g/l tryptone, 24 g/l yeast extract,0.4% (v/v) glycerol, and 89 mM potassium phosphate); TB + sorbitol (660 mM sorbitol, 12 g/l tryptone, 24 g/l yeast extract, 0.4% (v/v) glycerol, 89 mM potassium phosphate, and 2.5 mM betaine); and minimal media [26] (30 g/l glycerol, 13.3 g/l KH2PO4, 4 g/l (NH4)2HPO4,1.7 g/l citric acid, 1.2 g/l MgSO47H2O, 8.4 mg/l EDTA, 2.5 mg/l CoCl26H2O, 15 mg/l MnCl24H2O, 1.5 mg/l CuCl22H2O, 3 mg/l H3BO3, 2.5 mg/l Na2MoO42H2O, 13 mg/l Zn(CH3COO)22H2O, 100 mg/l Fe(III)citrate, and 4.5 mg/l thiamineHCl. Soluble HC/B quantification assay This assay was used to estimate the yield of the soluble HC fragment in cultures grown under various conditions during expression optimization. The assay included two steps. First, samples (0.5 ml) withdrawn from the cultures were chemically disrupted with the CelLytic B Plus Kit (Sigma–Aldrich) according to the manufacturer’s instructions, and the soluble proteins were separated from the insoluble cell fraction by centrifugation. The concentration of the HC fragment in the supernatants was then estimated by sandwich ELISA as follows. Plates (Maxisorp, Nunc; Roskilde, Denmark) were coated with 50 ll of a mouse anti-HC/B monoclonal antibody [23] diluted to a final concentration of 4 lg/ml in coating buffer (50 mM Na2CO3, pH 9.6) and then incubated overnight at 4 °C. The plates were subsequently washed with wash solution (0.9% NaCl, 0.05% Tween 20) and blocked for 1 h at 37 °C with TSTA buffer (50 mM Tris, 0.9% NaCl, 0.05% Tween 20, 2% BSA, 200 ll per well). After washing, the plates were incubated with serial dilutions (50 ll per well, in duplicate) of the tested supernatant and pure HC standard in TSTA for 1 h at 37 °C. The plates were then washed with wash solution and incubated for 1 h with a rabbit anti-HC fragment polyclonal antibody, diluted in TSTA to a final concentration of 0.5 lg/ml. After additional washing, the plates were incubated with 50 ll of alkaline phosphatase-conjugated donkey anti-Rabbit IgG (Jackson ImmunoResearch) diluted 1:1,500 for 1 h at 37 °C. Finally, the plates were washed with wash solution, and the colorimetric reaction was developed using the substrate p-nitrophenyl phosphate (1 mg/ml in 0.2 M Tris buffer). The absorbance at 405 nm was continuously measured for 15 min in 30-s intervals, and the concentration of the HC fragment was determined by interpolation from a standard curve of HC/B.
124
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
Purification of HC/B
In vivo inhibition test
For the development of HC/B purification process, E. coli BL21(DE3) carrying the pET-9a-HC/B plasmid was grown without induction in 2-l polycarbonate baffled shake flasks (Nalgene) containing 0.5 l of TB media supplemented with kanamycin (30 lg/ ml). The flasks were incubated at 18 °C with shaking (250 rpm) for 40 h. The cells were harvested by centrifugation, then resuspended in 100 ml of binding buffer (20 mM sodium phosphate, 0.5 M NaCl, 20 mM imidazole, pH 7.4), and disrupted by sonication. The cell extract was clarified by centrifugation (14,000g, 30 min) and loaded onto a HisTrap FF 1 ml column (GE Healthcare) mounted on an AKTA Explorer FPLC system (GE Healthcare). The column was washed with 10 column volumes (CV) of binding buffer, after which different conditions were tested for improving the purity level, as described in the Results section. The modified purification process of HC/B included the following steps: (1) loading of the cell lysate onto a Ni2+ column; (2) washing of unbound proteins with binding buffer (10 CV); (3) acidic washing of contaminants with acidic wash buffer (50 mM sodium phosphate, 50 mM NaCl, pH 4.0, 15 CV); (4) washing with binding buffer containing 40 mM imidazole (10 CV); and (5) elution of HC/B with elution buffer (20 mM sodium phosphate, 0.5 M NaCl, 500 mM imidazole, pH 7.4). Comparison of the purity levels obtained using the modified purification process with a standard process (lacking the acidic wash step) was performed using the Experion Automated Electrophoresis System (Bio-Rad; Hercules, CA) according to the manufacturer’s instructions. Mass spectrometry analysis of ArnA band was done at the Smoler Proteomics Center (Technion – Israel Institute of Technology, Haifa, Israel).
Mice (6 per group) were injected (1 ml, i.p.) with a 5 MsLD50 dose of BoNT/B in the presence (test group) or absence (control group) of different doses of Hc/B. Following injection, survival was monitored for 1 week.
Dialysis buffer optimization experiments These experiments were performed on HC/B eluted from a Ni2+ column in elution buffer (20 mM sodium phosphate, 0.5 M NaCl, 500 mM imidazole, pH 7.4). Samples of the protein (750 ll) were transferred to a Slide-A-Lyzer G2 Dialysis Cassette, 10,000 MWCO (Thermo Scientific). The cassettes were incubated in 750 ml of the tested buffer at 4 °C with stirring, for a total of three buffer changes, each for at least 2 h. Protein determination at the end of the process was performed with a NanoDrop Spectrophotometer (Thermo Scientific), using an extinction coefficient of 105,660 mM 1 cm 1 and molecular weight of 52.9 kDa (calculated using the ExPASy server [27]). Vaccination of mice with HC fragment For all vaccination experiments, the HC/B fragment was diluted with PBS and adsorbed to aluminum hydroxide (0.5% w/v final concentration of Al(OH)3). The vaccines were administered to CD-1 mice (Charles River UK) via the subcutaneous route. For the challenge protection test, mice were immunized once (5 lg of antigen), and after 21 days, they were subjected to a 102, 103, 104, 105, or 106 MsLD50 (mouse LD50) challenge, using three mice per challenge dose. For the potency study mice (5 per group) were vaccinated with nine different HC/B doses, ranging from 3.9 to 1000 ng (2-fold dilution factor between sequential groups). Twenty-one days after vaccination, the mice were challenged with a 103 MsLD50 dose, and their survival was monitored for 7 days. For the generation of neutralizing antibodies, mice (n = 7) were immunized three times with vaccine preparations containing 5 lg of protein per injection, at 21-day intervals. Two weeks after the last injection, mice were bled, and the sera were pooled. The neutralizing antibody concentration was determined according to the European Pharmacopoeia [28].
Statistical analysis A 4-parameter logistic regression model was employed to construct a standard curve for pure HC, using SoftMax Pro 5.4 (Molecular Devices, Sunnyvale, CA), for the estimation of expression levels. Calculation of the ED50 (effective dose protecting 50% of the mice) was carried out with GraphPad Prism 5 software (GraphPad software, San Diego, CA), employing non-linear regression analysis. Comparison of survival curves was conducted with the log-rank (Mantel-Cox) test using GraphPad Prism 5 software. Differences were considered significant at P < 0.05. Results and discussion Construction of the expression system and preliminary protein production The expression of genes originating from C. botulinum in E. coli often suffers from low yields since they exhibit distinct codon usage, which limits translation. To overcome this issue, a synthetic gene with optimized codon usage for the host was prepared. The gene encoded amino acids 859–1290 of BoNT/B (Danish strain). To facilitate purification, a 6His tag was added to the N-terminus of the protein. The gene was cloned into the pET-9a vector, which was found to be optimal for the expression of BoNT/A [20]. The corresponding pET-9a-HC/B plasmid was transformed into E. coli BL21(DE3). Preliminary attempt to produce HC/B included growing the bacterium in TB media without induction, overnight at 37 °C. These conditions were chosen because similar growth conditions resulted in high expression levels of the receptor-binding domain of BoNT/A (350 mg of purified HC/A from 1 l of culture) [20]. However, a poor yield of HC/B was obtained (1–2 mg HC/B per liter), and following purification by means of immobilized metal chelate affinity chromatography (IMAC), the protein contained a considerable amount of contaminants (purity level of 30%, Fig. 1). Analysis of the soluble and non-soluble cell fractions using SDS-PAGE showed that HC/B was the dominant protein in the pellet, while it could not be observed in the supernatant. Hence, the poor yield was due to the tendency of the protein to form insoluble aggregates, as previously reported by others [7,21]. Expression optimization To improve the soluble HC/B yield, a screening assay that allows evaluation of the effects of various expression conditions in a short time was developed. The assay involved small-scale (40 ml) bacterial cell growth under the tested conditions, a chemical cell lysis step, and quantification of HC/B in the soluble cell fraction using a specific ELISA test. A common strategy to improve soluble protein expression is reduction of the growth temperature. Under low temperature, the growth rate of the bacterium and the protein production rate are reduced, and the recombinant protein therefore has more time to undergo proper folding. Therefore, expression was tested at 37, 28, and 18 °C. Another way to reduce the growth rate is by growing the bacterium on minimal medium, in which it must synthesize basic building blocks that are readily available in rich media.
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
125
in the absence of an inducer, as a result of basal expression of T7 RNA polymerase. In other studies [7,21,22], the promoters used for HC/B included the lac operator element, and the expression vectors carried a copy of the lac repressor gene (lacI). Together, these components provide tighter control of gene expression, and HC/B expression requires induction with IPTG. Although a stringent control level can be advantageous for the expression of recombinant proteins that are detrimental to the host, for HC/B expression, the more relaxed control level is preferred. Since the recombinant gene is transcribed at a moderate rate by T7 RNA polymerase when present at a basal level, the protein has a better chance of folding properly than in induced systems in which the gene is transcribed at very high levels upon induction, and accumulation of the unfolded polypeptide occurs. For expression of the receptor-binding domain of BoNT/A, we previously obtained similar results, where the use of the T7 promoter improved soluble HC/A levels by two orders of magnitude in comparison with the T7lac promoter [20]. Fig. 1. Preliminary attempt to produce HC/B. E. coli BL21(DE3) harboring the pET9a-HC/B plasmid were grown in TB media overnight at 37 °C. Following cell disruption the protein was purified using IMAC. Samples from the purification process were subjected to SDS-PAGE (A) and Western blot (B) analyses: 1 – BoNT/B complex; 2 – soluble cell fraction; 3 – non-soluble cell fraction; 4 – the flow through of soluble cell fraction loading step to the IMAC column (unbound proteins); and 5 – the eluted protein. The HC fragment was detected using horse anti BoNT/B complex antibodies.
Accordingly, we tested the use of rich media (terrific broth) and minimal media [26], in which the generation times at 37 °C were 30 and 150 min, respectively. Induction of osmotic stress was also reported to improve soluble expression yields of several recombinant proteins [29–31]. It is assumed that under osmotic stress, the bacterium produces osmoprotectants or induces stress response mechanisms that affect the folding kinetics of the recombinant protein [29,30]. Therefore, the effect of this parameter on the soluble HC/B yield was evaluated by growing the bacterium in TB media containing 660 mM sorbitol. Soluble HC/B levels were most affected by the growth temperature (Table 1). When cells were grown at 37 °C, the volumetric yield was low (4.2 mg HC/B per liter of culture, TB media). Significant improvement in the yield was achieved when the temperature was reduced to 28 °C (51.7 mg/l), and the highest soluble HC/B level was obtained at 18 °C (77.8 mg/l). Induction with IPTG had a negative effect on soluble HC/B levels at all examined temperatures, which could be due to the rapid accumulation of unfolded polypeptides that are produced at very high levels following induction. The use of minimal media did not increase HC/B levels above those obtained with TB media. The specific volumetric yields of cultures grown with defined media were similar to those obtained with TB media. However, because defined media supported lower cell mass accumulation, the overall volumetric yield was also lower. Induction of osmotic shock did not improve soluble HC/B levels as well. The volumetric yield obtained under osmotic shock was similar to that obtained with regular growth in TB media at 18 °C. Because growth of the bacterium under osmotic shock was performed at 18 °C as well, it can be concluded that this treatment did not contribute to obtaining higher soluble protein levels beyond regular expression in TB media at 18 °C. The yields of soluble HC/B that we obtained at 18 °C are an order of magnitude higher than those previously reported for the expression of the protein in E. coli [7,21]. The main difference between our expression system and those described in other reports is the control level of recombinant gene expression. In our system, the transcription of the HC/B gene is controlled by the T7 promoter. This system is leaky, and recombinant gene expression occurs even
Development of HC/B purification process To facilitate the purification of HC/B, the recombinant gene was designed to contain a 6xHis tag at its N-terminus. Recombinant vaccines against malaria that were produced in E. coli and contained a His tag have been found to be safe in phase I clinical trials [32,33]. Although purification of His-tagged proteins on a Ni2+ column usually results in a protein with satisfactory purification level, histidine-rich host cell proteins can often contaminate the preparation [34]. Hence, in some cases, additional chromatographic steps are required to improve the purity level. In our hands, applying a standard IMAC purification procedure resulted in a poor purity of HC/B (Fig. 1). One strategy for removing these contaminants is to perform a wash step prior to the elution step, with an intermediate imidazole concentration that will release the contaminants without eluting the target protein. To determine the optimal purification conditions for HC/B, a cell lysate of E. coli BL(21) + pET-9a-HC/B grown in TB media at 18 °C was loaded onto a 1-ml Ni2+ column, and the column was washed with increasing imidazole concentrations (30–100 mM in 10 mM steps). Finally, the remaining proteins were eluted with 0.5 M imidazole (Fig. 2). Neither of the tested imidazole concentrations was found to be suitable for obtaining satisfactory purity level of HC/B. At imidazole concentrations lower than 60 mM, only a single 90 kDa contaminant was completely removed. Several proteins were washed out at 60–90 mM imidazole, and others were released from the column only at 0.5 M imidazole. Moreover, at all of the tested imidazole concentrations, HC/B was released along with the contaminants, and above a concentration of 50 mM, significant leakage occurred. We next attempted to separate HC/B from the contaminants by washing the column with a low pH buffer. Under pH values lower than the pKa of histidine (6.0), the residue is protonated, and the affinity toward nickel ions is therefore reduced, and proteins can be released from the column. To find a pH value that differentially releases the contaminants and HC/B from the column, a cell lysate was loaded onto a nickel column, and the pH of the wash buffer was decreased from 7.0 to 4.0 by mixing phosphate buffers at pH 7.0 and 4.0 in different ratios. While HC/B was not eluted at any of the tested pH values (7.0–4.0), several contaminants were released from the column at pH 5.5–6.0. Further decreasing the pH with citrate buffer to 3.0 resulted in the formation of a white precipitate in the column, without releasing HC/B. Because several contaminants were released from the column at pH 5.5–6.0 without leakage of HC/B, we wished to examine whether the addition of an acidic wash step to a standard purification process could increase the HC/B purity level. Therefore, we compared the purity obtained through a standard purification
126
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
Table 1 Comparison of soluble HC/B expression levels obtained at different conditions.
a b c d
Growth temperature (°C)
Mediaa
A600nm
37 37 37 37 28 28 18 18 18 18
TB TB TB MM TB MM TB TB TB TB
19.5 20.2 23.3 8 24.3 8.6 19.9 23.7 19.7 25.9
Induction (+/ )b + +
+ +
HC/B yield (mg/l)c
HC/B specific yield (mg/lO.D.)
4.2 2.7 2.3 2.7 52 9.5 78 30 61 43
0.2 0.1 0.1 0.3 2.1 1.1 3.9 1.3 3.1 1.7
Comments Induction at A600nm = 1.0d Induction at A600nm = 5.0d
Induction at A600nm = 0.6 Osmotic shock Osmotic shock + Induction at A600nm = 0.6
TB – terrific broth; MM – minimal media. With IPTG to a final concentration of 0.4 mM (37 °C) or 0.1 mM (18 °C). HC fragment yield values are expressed as mg protein per 1 liter of culture. Following induction samples were withdrawn and analyzed after 2, 3, 4, 5, 6, and 20 h. The results with the highest HC/B yield are shown.
Fig. 2. Elution of HC/B and host cell proteins from Ni2+ column at different imidazole concentrations. A cell lysate of E. coli BL21(DE3) carrying the pET-9a-HC/B plasmid and grown at 18 °C in TB media was loaded onto Ni2+ column and then it was washed with increasing imidazole concentrations (30–100 mM in 10 mM steps). Elution of the remaining proteins was done with 0.5 M imidazole. (A) The chromatogram of the purification process (A280nm – grey; A260nm – black; and imidazole concentration – dashed) (B) SDS-PAGE analysis of protein sample eluted at different imidazole concentrations: 1 – 30 mM; 2 – 40 mM; 3 – 50 mM; 4 – 60 mM; 5 – 70 mM; 6 – 80 mM; 7 – 90 mM; 8 – 100 mM; 9 – 500 mM. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
process with a modified process that included a wash step with phosphate buffer at pH 4.0 (Fig. 3). Both processes were carried out using equal volumes of the same cell lysate. The purity level of HC/B obtained via the standard process was 63% (estimated using automated electrophoresis analysis). This preparation included a main contaminant of 74 kDa that accounted for 12.5% of the total proteins in the preparation and two minor contaminants of 22 and 25 kDa. The acidic wash significantly increased HC/B purity (85%). This improvement was achieved mainly due to complete removal of the main contaminant at 74 kDa during the acidic wash step. In addition, the presence of other contaminants was reduced as well during the acidic wash step. Overlaying the electropherograms of the two preparations revealed that the peak of HC/B purified via the modified process was slightly higher than that of HC/B purified by the standard process. Therefore, no loss of HC/B occurred during the acidic wash. ArnA is the main contaminant removed by the acidic wash The addition of the acidic wash step to the standard purification process significantly improved the purity level of HC/B. This washing step is simple, economical, does not hamper the yield, and avoids the inclusion of an additional chromatographic step. The main contribution of the acidic wash was the removal of the main contaminant of 74 kDa. To identify this contaminant, the corresponding band was excised from a gel and analyzed via mass spectrometry. The protein was identified as ArnA, an E. coli enzyme involved in the biosynthesis of 4-amino-4-deoxy-L-arabinose
[35]. ArnA is considered to be one of the two most abundant E. coli proteins (along with SlyD) that are co-purified with recombinant His-tagged proteins [36]. ArnA contains histidine-rich patches on its surface, which are believed to interact with Ni2+ ions. Commonly, elution of ArnA from a Ni2+ column occurs at imidazole concentrations of 55–80 mM [34]. Because HC/B began to be eluted from the column at 50 mM imidazole, it was surprising that under an acidic wash, ArnA was completely removed from the column, while HC/B remained bound. The cause of this result could be explained by differences between ArnA and Hc/B in the nature of their interactions with nickel ions. The tertiary structure of ArnA is a hexamer, consisting of a dimer of two trimers [37] (Fig. 4). The histidine-rich patches found in ArnA are located at the center of each trimer, and are formed by spatial proximity of histidine residues from each monomer. It is possible that by reducing the pH, the hexameric structure of ArnA dissociated into monomers, and the interaction with nickel ions was therefore lost. On the other hand, in HC/B, the histidine residues are in linear proximity, and their interaction with nickel ions is therefore not dependent on the spatial structure of the protein. Recently, two research groups developed E. coli strains that allow purification of His-tagged protein with reduced host cell protein contamination. Robichon et al., from New England BioLabs, developed the commercial strain NiCo21 [38]. In this strain, ArnA is fused to a chitin binding domain. Following the standard IMAC purification process, ArnA is separated from the recombinant protein on a chitin column, a step that can hamper the overall yield and increase process costs. Anderson et al. developed the strain
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
127
Fig. 3. Acidic wash of column-bound HC/B increases its purity. (A) Electropherograms overlay of HC/B purified using standard His-tagged protein purification process (grey) and using a modified process that included an acidic wash step (black). (B) SDS-PAGE analysis of HC/B purified using a standard and modified purification process. 1 – supernatant of broken cells; 2 – pellet of broken cells; 3 – HC/B purified using standard process; 4 – HC/B purified using a modified process that included an acidic wash step. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 4. Distribution of histidine residues on the surface of the contaminant ArnA. ArnA is a hexamer that is formed by a dimer of two trimers. The protein contains a histidinerich patch on its surface at the trimer interface. This patch is likely to accounts for the ArnA binding to Ni2+ ions. At low pH the protein may dissociate to monomers, thereby disrupting the interaction with nickel. The Fig. was prepared with PyMOL molecular graphics system using the ArnA structure published by Gatzeva-Topalova et al. [37] (pdb code 1Z7E). The ArnA monomers are colored differently and histidine residues are colored blue.
LOBSTR (low background strain) in which the affinity of ArnA toward nickel ions was reduced through mutagenesis, and ArnA is released from the column at 30 mM imidazole [36]. Both genetically engineered strains are derivatives of E. coli BL21(DE3), which is the most prevalent strain used for protein expression. However, there are additional E. coli strains that are employed for recombinant protein expression, which bear special characteristics. For example, the Rosetta (Novagen) and Codonplus (Stratagene) strains enable the expression of genes with rare codons for E. coli, and the Origami (Novagen) strain is used for the expression of proteins that contain disulfide bonds. In cases where the recombinant protein must be expressed in a host other than BL21(DE3), the use of LOBSTR or NiCo21 is not applicable. Nevertheless, removal of contaminants through an acidic wash is generic to all E. coli strains, as they contain the same contaminants. In addition, an acidic wash may be beneficial for the purification of His-tagged proteins expressed in other systems, such as P. pastoris or Chinese hamster ovary. Optimization of the HC/B formulation Following the purification of HC/B using a Ni2+ column, it was found in the elution buffer. This buffer contained high salt and imidazole concentrations (0.5 M each), and the protein therefore had to be dialyzed to a buffer that is suitable for injection. Eluted HC/B was initially dialyzed against 50 mM sodium phosphate,
50 mM sodium chloride, pH 6.5 (buffer 1, Table 2). At the end of the dialysis process, a considerable amount of protein precipitate was obtained. Precipitate formation could be observed as early as 2 h after the beginning of the dialysis process. The estimated loss of HC/B during dialysis was 50%. Therefore, we wished to determine the basis for HC/B precipitation and reduce its loss by optimizing the composition of the dialysis buffer. Following the purification of HC/B using IMAC, the protein was divided into separate dialysis cassettes, each of which was dialyzed against a different buffer (Table 2). At the end of the dialysis process, the protein solutions were centrifuged to remove the precipitated protein, and their protein contents were determined. The solubility of HC/B was most strongly affected by the salt concentration. At a high salt concentration (0.5 M NaCl, buffer 4), precipitation of HC/B was not observed, whereas in a buffer with the same composition except for the salt content (50 mM, buffer 1), 49% of the protein precipitated. The mechanism underlying this stabilization by a high salt concentration is presumably shielding of charges on the protein surface, thereby reducing intermolecular interactions (also known as ‘‘salting in’’). Although a 100% yield was obtained with buffer 4, its high salt content makes it less optimal for injection. The solubility of HC/B was also affected by the pH of the buffer. Major precipitation occurred when using Tris buffer (buffer 5, pH 8.0), while the protein was more stable when dialyzed against acidic buffers (yields of 87% for buffer 2, pH 5.5, and 68% for buffer 3, pH 5.0). It is possible that the precipitation observed for
128
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
Table 2 Yields of HC/B following dialysis against different buffers compositions. Buffer composition
Protein concentration (mg/ml)
Yield (%)
(1) 50 mM sodium phosphate, 50 mM NaCl, pH 6.5 (2) 50 mM sodium citrate, pH 5.5 (3) 50 mM sodium acetate, pH 5.0 (4) 50 mM sodium phosphate, 0.5 M NaCl, pH 6.5 (5) 20 mM TrisHCl, 50 mM NaCl, pH 8.0
2.7
51
4.6 3.6 5.3
87 68 100
1.9
35
Fig. 5. Dose-response relationships in mice vaccinated with HC/B. Mice (5 per group) were vaccinated with nine different HC/B doses, ranging from 3.9 to 1000 ng (2-fold dilution factor between sequential groups). Twenty one days after the vaccination the mice were challenged with 103 MsLD50, and survival was monitored for 7 days. The ED50 of HC/B was 58 ng, with a 95% confident limits of 23.5–128 ng.
a dose of 103 LD50, and in the work of Zichel et al., mice were protected against challenge with a dose of 106 LD50 after vaccination with 5 lg HC/B. However, in these works, mice received three HC/ B injections before toxin challenge, whereas in our work, mice were protected after a single vaccination. To study the potency of HC/B, mice were vaccinated once with the protein at nine different doses ranging from 3.9 to 1000 ng and then challenged with 103 MsLD50 of BoNT/B (Fig. 5). At doses below 16 ng, no survival was observed, while 100% survival was obtained at doses higher than 250 ng. The calculated ED50 (effective dose protecting 50% of the mice) of HC/B was 58 ng, with 95% confident limits of 23.5 and 128 ng. These results are in accordance with other reports concerning HC/B potency. Potter et al. [39] obtained full protection against challenge with 104 MsLD50 at HC/B doses higher than 625 ng. Gao et al. [22] reported full protection of mice against challenge with 103 MsLD50 at HC/B doses above 250 ng and no survival below 2 ng. However, in their study, the vaccination regime included three injections with HC/B, rather than a single injection, as in our study. To test the ability of HC/B to generate neutralizing antibodies, a group of seven mice was vaccinated three times with 5 lg of aluminum hydroxide-adsorbed HC/B. The neutralizing antibody concentration in the pooled sera of the mice was 50 IU/ml. This value is consistent with previous reports. Smith et al. reported that three vaccinations of mice with 1 lg of HC/B produced using P. pastoris generated a neutralizing antibody concentration of 28.7 IU/ml [5]. Gao et al. [22] vaccinated mice with HC/B expressed in E. coli and produced from inclusion bodies. Following three vaccinations with a dose of 1.25 lg, the neutralizing antibody concentration was 17.1 IU/ml, and when the HC/B dose was 6.25 lg, the concentration was 19.2 IU/ml. In vivo inhibition of botulinum neurotoxin B by the HC fragment
Fig. 6. HC/B inhibits mice intoxication with BoNT/B. Mice (n = 6) were exposed to 5 MsLD50 of BoNT/B in the presence (broken line) or absence (solid line) of HC/B, and survival was monitored. Time to death was significantly delayed in the HC/B-treated group (P = 0.0057 by log-rank test).
buffer 5 was due to the proximity of the pH of the buffer (8.0) to the isoelectric point of the protein (calculated pI of 7.21, using the ExPASy server [27]). Hence, we chose to use citrate buffer, pH 5.5, for HC/B formulation. SDS-PAGE analysis of HC/B dialyzed against the different buffers showed that the contaminant content in the preparations was inversely related to the HC/B yield. Thus, the precipitate obtained during dialysis is of HC/B, while the contaminants remain stable. Therefore, reducing precipitation improves the yield of the process as well the purity level of the protein. Immunogenicity of HC/B Single vaccination of mice with 5 lg HC/B resulted in full protection against challenge with 102–106 MsLD50 BoNT/B. This protection is at least equivalent to that of HC/B produced using other expression systems. Baldwin et al. reported that vaccination of mice with 1 lg HC/B conferred protection against challenge with
We have recently demonstrated that HC/A can inhibit intoxication with BoNT/A and BoNT/E in vivo [20]. The presumed mechanism of this inhibition is that HC/A competes with the toxins for receptor binding (SV2). HC/A was not able to inhibit intoxication with BoNT/B, as this toxin serotype utilizes a different receptor (synaptotagmin) to enter the cells [40]. To determine whether BoNT/B intoxication can be inhibited by HC/B, mice (n = 6) were exposed to 5 MsLD50 of the toxin in the presence or absence of HC/B (1 mg), and their survival was monitored (Fig. 6). HC/B significantly delayed intoxication by BoNT/B. In the absence of HC/B, the time to death (TTD) ranged from 23.5 to 40 h, and the median TTD was 26 h, while in the presence of HC/ B, the TTD ranged from 36 to 60 h, and the median TTD was 43.4 h (P = 0.0057). One animal in the HC/B-treated group survived the challenge. The ability of HC/B to compete with the toxin for receptor binding indicates that the recombinant protein is expressed and folded into its native three dimensional structure. This characteristic is of great importance because, as an antigen, HC/B displays spatial epitopes that may contribute to the generation of neutralizing antibodies. Conclusions In this work, an efficient expression system for the receptorbinding domain of BoNT/B was developed. The system provides an order of magnitude improvement in the soluble protein yield compared with previously reported works, and it is simple and economical. The protein exhibits good immunogenicity, and its protective properties are at least comparable to those of HC/Bs produced using other expression systems. Furthermore, the native function of the protein is preserved, as it is able to inhibit BoNT/
A. Ben David et al. / Protein Expression and Purification 110 (2015) 122–129
B intoxication in vivo. In addition, we found that the contaminant ArnA can be removed through a simple wash step with an acidic buffer, thereby avoiding the need for an additional chromatographic step. As ArnA is a common contaminant of His-tagged proteins produced in E. coli, we believe that this finding may aid in the purification of other recombinant proteins. Acknowledgment This work was funded by Israel Institute for Biological Research grant SB/5122-18. References [1] R.G. Darling, C.L. Catlett, K.D. Huebner, D.G. Jarrett, Threats in bioterrorism I: CDC category a agents, Emerg. Med. Clin. North Am. 20 (2002) 273–309. [2] L.A. Smith, J.M. Rusnak, Botulinum neurotoxin vaccines: past, present, and future, Crit. Rev. Immunol. 27 (2007) 303–318. [3] J.M. Rusnak, L.A. Smith, Botulinum neurotoxin vaccines: past history and recent developments, Hum. Vaccin. 5 (2009) 794–805. [4] M. Montal, Botulinum neurotoxin a marvel of protein design, Annu. Rev. Biochem. 79 (2010) 591–617. [5] L.A. Smith, Botulism and vaccines for its prevention, Vaccine 27 (2009) D33– D39. [6] M.A. Clayton, J.M. Clayton, D.R. Brown, J.L. Middlebrook, Protective vaccination with a recombinant fragment of Clostridium botulinum neurotoxin serotype A expressed from a synthetic gene in Escherichia coli, Infect. Immun. 63 (1995) 2738–2742. [7] M.R. Baldwin, W.H. Tepp, A. Przedpelski, C.L. Pier, M. Bradshaw, E.A. Johnson, J.T. Barbieri, Subunit vaccine against the seven serotypes of botulism, Infect. Immun. 76 (2008) 1314–1318. [8] J.D. Shearer, M.L. Vassar, W. Swiderski, K. Metcalfe, N. Niemuth, I. Henderson, Botulinum neurotoxin neutralizing activity of immune globulin (IG) purified from clinical volunteers vaccinated with recombinant botulinum vaccine (rBV A/B), Vaccine 28 (2010) 7313–7318. [9] H.F. LaPenotiere, M.A. Clayton, J.L. Middlebrook, Expression of a large, nontoxic fragment of botulinum neurotoxin serotype A and its use as an immunogen, Toxicon 33 (1995) 1383–1386. [10] J. Boles, M. West, V. Montgomery, R. Tammariello, M.L.M. Pitt, P. Gibbs, L. Smith, R.D. LeClaire, Recombinant C fragment of botulinum neurotoxin B serotype (rBoNTB (HC)) immune response and protection in the rhesus monkey, Toxicon 47 (2006) 877–884. [11] M.P. Byrne, T.J. Smith, V.A. Montgomery, L.A. Smith, Purification, potency, and efficacy of the botulinum neurotoxin type A binding domain from Pichia pastoris as a recombinant vaccine candidate, Infect. Immun. 66 (1998) 4817– 4822. [12] M.P. Byrne, R.W. Titball, J. Holley, L.A. Smith, Fermentation, purification, and efficacy of a recombinant vaccine candidate against botulinum neurotoxin type F from Pichia pastoris, Protein Expr. Purif. 18 (2000) 327–337. [13] M.P. Dux, R. Barent, J. Sinha, M. Gouthro, T. Swanson, A. Barthuli, M. Inan, J.T. Ross, L.A. Smith, T.J. Smith, R. Webb, B. Loveless, I. Henderson, M.M. Meagher, Purification and scale-up of a recombinant heavy chain fragment C of botulinum neurotoxin serotype E in Pichia pastoris GS115, Protein Expr. Purif. 45 (2006) 359–367. [14] M.P. Dux, J. Huang, R. Barent, M. Inan, S.T. Swanson, J. Sinha, J.T. Ross, L.A. Smith, T.J. Smith, I. Henderson, M.M. Meagher, Purification of a recombinant heavy chain fragment C vaccine candidate against botulinum serotype C neurotoxin [rBoNTC(Hc)] expressed in Pichia pastoris, Protein Expr. Purif. 75 (2011) 177–185. [15] K.J. Potter, W. Zhang, L.A. Smith, M.M. Meagher, Production and purification of the heavy chain fragment C of botulinum neurotoxin, serotype A, expressed in the methylotrophic yeast Pichia pastoris, Protein Expr. Purif. 19 (2000) 393– 402. [16] M.R. Baldwin, W.H. Tepp, C.L. Pier, M. Bradshaw, M. Ho, B.A. Wilson, R.B. Fritz, E.A. Johnson, J.T. Barbieri, Characterization of the antibody response to the receptor binding domain of botulinum neurotoxin serotypes A and E, Infect. Immun. 73 (2005) 6998–7005. [17] Y.Z. Yu, S.M. Zhang, Y. Ma, H.Q. Zhu, W.B. Wang, Y. Du, X.W. Zhou, R.L. Wang, S. Wang, W.Y. Yu, P.T. Huang, Z.W. Sun, Development and evaluation of candidate vaccine and antitoxin against botulinum neurotoxin serotype F, Clin. Immunol. 137 (2010) 271–280.
129
[18] R. Chen, J. Shi, K. Cai, W. Tu, X. Hou, H. Liu, L. Xiao, Q. Wang, Y. Tang, H. Wang, Improved soluble expression and characterization of the Hc domain of Clostridium botulinum neurotoxin serotype A in Escherichia coli by using a PCR-synthesized gene and a Trx co-expression strain, Protein Expr. Purif. 71 (2010) 79–84. [19] Y.Z. Yu, Z.W. Sun, S. Wang, W.Y. Yu, High-level expression of the Hc domain of clostridium botulinum neurotoxin serotype A in Escherichia coli and its immunogenicity as an antigen, Chin. J. Biotechnol. 23 (2007) 812–817. [20] A. Ben David, E. Diamant, A. Barnea, O. Rosen, A. Torgeman, R. Zichel, The receptor binding domain of botulinum neurotoxin serotype A (BoNT/A) inhibits BoNT/A and BoNT/E intoxications in vivo, Clin. Vacc. Immun. 20 (2013) 1266–1273. [21] D.M. Held, A.C. Shurtleff, S. Fields, C. Green, J. Fong, R.G.A. Jones, D. Sesardic, R. Buelow, R.L. Burke, Vaccination of rabbits with an alkylated toxoid rapidly elicits potent neutralizing antibodies against botulinum neurotoxin serotype B, Clin. Vacc. Immun. 17 (2010) 930–936. [22] Y.L. Gao, S. Gao, L. Kang, C. Nie, J.L. Wang, Expression of Hc fragment from Clostridium botulinum neurotoxin serotype B in Escherichia coli and its use as a good immunogen, Hum. Vaccin. 6 (2010) 462–466. [23] E. Diamant, B.E. Lachmi, A. Keren, A. Barnea, H. Marcus, S. Cohen, A. Ben David, R. Zichel, Evaluating the synergistic neutralizing effect of anti-botulinum oligoclonal antibody preparations, PLoS One 9 (2014) e87089. [24] R. Zichel, A. Mimran, A. Keren, A. Barnea, I. Steinberger-Levy, D. Marcus, A. Turgeman, S. Reuveny, Efficacy of a potential trivalent vaccine based on Hc fragments of botulinum toxins A, B, and E produced in a cell-free expression system, Clin. Vacc. Immun. 17 (2010) 784–792. [25] S.M. Whelan, M.J. Elmore, N.J. Bodsworth, J.K. Brehm, T. Atkinson, N.P. Minton, Molecular cloning of the Clostridium botulinum structural gene encoding the type B neurotoxin and determination of its entire nucleotide sequence, Appl. Environ. Microbiol. 58 (1992) 2345–2354. [26] D.J. Korz, U. Rinas, K. Hellmuth, E.A. Sanders, W.D. Deckwer, Simple fed-batch technique for high cell density cultivation of Escherichia coli, J. Biotechnol. 39 (1995) 59–65. [27] E. Gasteiger, A. Gattiker, C. Hoogland, L. Ivani, R.D. Appel, A. Bairoch, ExPASy: the proteomics server for in-depth protein knowledge and analysis, Nucleic Acids Res. 31 (2003) 3781–3788. [28] European Pharmacopoeia, Botulinum antitoxin, 2011. [29] Y. Fundoiano-Hershcovitz, L. Rabinovitch, Y. Langut, V. Reiland, G. Shoham, Y. Shoham, Identification of the catalytic residues in the double-zinc aminopeptidase from Streptomyces griseus, FEBS Lett. 571 (2004) 192–196. [30] S. Picaud, M.E. Olsson, P.E. Brodelius, Improved conditions for production of recombinant plant sequitepene synthase in Escherichia coli, Protein Expr. Purif. 51 (2007) 71–79. [31] N. Oganeysyan, I. Ankoudinova, S.H. Kim, R. Kim, Effect of osmotic stress and heat shock in recombinant protein overexpression and crystallization, Protein Expr. Purif. 52 (2007) 280–285. [32] C.F. Ockenhouse, E. Angov, K.E. Kester, C. Diggs, L. Siosson, J.F. Cummings, Phase I safety and immunogenecity trial of FMP1/AS02A, a Plasmodium falciparum MSP-1 asexual blood stage vaccine, Vaccine 24 (2006) 3009–3017. [33] M.A. Thera, O.K. Doumbo, D. Coulibaly, D.A. Diallo, A.K. Kone, Safety and immunogenecity of an AMA-1 malaria vaccine in malian adults: results of a phase 1 randomized control trial, PLoS One 3 (2008) e1465. [34] V.M. Bolanos-Garcia, O.R. Davies, Structural analysis and classification of native proteins from E. coli commonly co-purified by immobilised metal affinity chromatography, Biochim. Biophys. Acta 1760 (2006) 1304–1313. [35] G.J. Williams, S.D. Breazeale, C.R.H. Raetz, J.H. Naismith, Structure and function of both domains of ArnA, a dual function decarboxylase and a formyltransferase, involved in 4-amino-4-deoxy-L-arabinose biosynthesis, J. Biol. Chem. 280 (2005) 23000–23008. [36] K.R. Andersen, N.C. Leska, T.U. Schwartz, Optimized E. coli expression strain LOBSTR eliminates common contaminants from his-tag purification, Proteins 81 (2013) 1857–1861. [37] P.Z. Gatzeva-Topalova, A.P. May, M.C. Sousa, Structure and mechanism of ArnA: conformational change implies ordered dehydrogenase mechanism for polymyxin resistance, Structure 13 (2005) 929–942. [38] C. Robichon, J. Luo, T.B. Causey, J.S. Benner, J.C. Samuelson, Engineering Escherichia coli BL21(DE3) derivatives strains to minimize E. coli protein contamination after purification by immobilized metal affinity chromatography, Appl. Environ. Microbiol. 77 (2011) 4634–4646. [39] K.J. Potter, M.A. Bevins, E.V. Vassilieva, V.R. Chiruvolu, T. Smith, L.A. Smith, M.M. Meagher, Production and purification of the heavy-chain fragment C of botulinum neurotoxin, serotype B, expressed in the methylotrophic yeast Pichia pastoris, Protein Expr. Purif. 13 (1998) 357–365. [40] M. Dong, D.A. Richard, M.C. Goodnough, W.H. Tepp, E.A. Johnson, E.R. Chapman, Synaptotagmins I and II mediate entry of botulinum neurotoxins B into cells, J. Cell Biol. 162 (2003) 1293–1303.