Expression of recombinant Rhipicephalus (Boophilus) microplus, R. annulatus and R. decoloratus Bm86 orthologs as secreted proteins in Pichia pastoris

Background Rhipicephalus (Boophilus) spp. ticks economically impact on cattle production in Africa and other tropical and subtropical regions of the world. Tick vaccines constitute a cost-effective and environmentally friendly alternative to tick control. The R. microplus Bm86 protective antigen has been produced by recombinant DNA technology and shown to protect cattle against tick infestations. Results In this study, the genes for Bm86 (R. microplus), Ba86 (R. annulatus) and Bd86 (R. decoloratus) were cloned and characterized from African or Asian tick strains and the recombinant proteins were secreted and purified from P. pastoris. The secretion of recombinant Bm86 ortholog proteins in P. pastoris allowed for a simple purification process rendering a final product with high recovery (35–42%) and purity (80–85%) and likely to result in a more reproducible conformation closely resembling the native protein. Rabbit immunization experiments with recombinant proteins showed immune cross-reactivity between Bm86 ortholog proteins. Conclusion These experiments support the development and testing of vaccines containing recombinant Bm86, Ba86 and Bd86 secreted in P. pastoris for the control of tick infestations in Africa.


Background
Rhipicephalus (Boophilus) spp. ticks are distributed in tropical and subtropical regions of the world with range expansion for some species due to changes in climatic conditions [1][2][3]. Infestations with the cattle tick, Rhipicephalus (Boophilus) microplus, economically impact cattle production by reducing weight gain and milk production, and by transmitting pathogens that cause babesiosis (Babesia bovis and B. bigemina) and anaplasmosis (Anaplasma marginale) [4]. R. annulatus and R. decoloratus also affect cattle production and vector pathogens in regions of Latin America, Africa or Asia [2].
Control of tick infestations has been difficult because ticks have few natural enemies. Integrated tick management strategies include the adaptation of different control methods to a geographic area. A major component of integrated tick control methods is the application of acaricides. However, use of acaricides has had limited efficacy in reducing tick infestations and is often accompanied by serious drawbacks, including the selection of acaricideresistant ticks, environmental contamination and contamination of milk and meat products with drug residues [5]. Furthermore, development of new acaricides is a long and expensive process. All of these issues reinforce the need for alternative approaches to control tick infestations [5]. Other approaches proposed for tick control have included the use of hosts with natural resistance to ticks, pheromone-impregnated decoys for attracting and killing ticks, biological control agents and vaccines [6][7][8].
In the early 1990s, vaccines were developed that induced immunological protection of vertebrate hosts against tick infestations. These vaccines contained the recombinant R. microplus Bm86 gut antigen [8][9][10][11][12]. Two vaccines using recombinant Bm86 were subsequently registered in Latin American countries (Gavac) and Australia (TickGARD) during 1993-1997 [13]. These vaccines reduce the number of engorging female ticks, their weight and reproductive capacity. Thus the greatest vaccine effect was the reduction of larval infestations in subsequent generations. Vaccine controlled field trials in combination with acaricide treatments demonstrated that an integrated approach resulted in control of tick infestations while reducing the use of acaricides [12][13][14]. These trials demonstrated that control of ticks by vaccination has the advantages of being cost-effective, reducing environmental contamination and preventing the selection of drug resistant ticks that result from repeated acaricide application. In addition, these vaccines may also prevent or reduce transmission of pathogens by reducing tick populations and/or affecting tick vectorial capacity [13][14][15].
Controlled immunization trials have shown that R. microplus Bm86-containing vaccines also protect against related tick species, R. annulatus and R. decoloratus [16][17][18]. However, R. microplus strain-to-strain variations in the susceptibility to Bm86 vaccination have been reported, which suggests that Bm86 sequence and/or tick physiological differences may influence the efficacy of the vaccine [8,[19][20][21][22]. Therefore, the cloning, expression and vaccine formulation with recombinant Bm86 from local tick strains may be required for vaccine efficacy in some geographic regions [20].
The recombinant Bm86 has been expressed in Escherichia coli [10], Aspergillus nidulans and A. niger [23] and Pichia pastoris [11,24,25]. Of these expression systems, P. pastoris has been shown to be the more efficient for protein secretion [26,27]. Furthermore, production of Bm86 in P. pastoris may increase the antigenicity and immunogenicity of the recombinant antigen [28,29]. However, the process previously reported for the production of recombinant Bm86 in P. pastoris is not based on protein secretion but on the expression of the antigen anchored to the yeast membrane, making necessary the purification under denaturing conditions followed by refolding of an antigen with high number of disulfide bonds [24,25,30]. Recently, R. decoloratus Bm86 orthologs were cloned, expressed in E. coli and partially characterized [31]. However, the cloning and expression of recombinant R. annulatus and R. decoloratus Bm86 orthologs in P. pastoris have not been reported.
The objectives of this study were (i) to clone and express in P. pastoris the recombinant R. microplus, R. decoloratus and R. annulatus Bm86 orthologs from African or Asian tick strains and (ii) to simplify the Bm86 production process by secreting recombinant proteins encoded by Bm86 orthologs in P. pastoris.

Cloning and sequence analysis of Bm86, Bd86 and Ba86
The Bm86 orthologs were cloned by RT-PCR from Mozambique R. microplus (Bm86), Israeli R. annulatus (Ba86) and South African R. decoloratus (Bd86) tick strains. Partial sequences were obtained and used to search the NCBI nr database for sequence identity. The first four BLAST hits (E-value = 0.0) showed that cloned Bm86, Bd86 and Ba86 sequences were identical (90-97% identity) to previously reported Bm86 (Australian Yeerongpilly reference strain; GenBank accession number M29321), Bm95 (Argentinean A strain; AF150891) and Bd86-1 and Bd86-2 (Kenyan strain; DQ630523 and DQ630524) sequences. The only fragment of 1,107 nucleotides previously reported for Ba86 (Mexican strain; AF150897) had 99.9% identity to the Ba86 sequence reported here with a single A × G substitution at position 1,674 (position 1 corresponds to the adenine in the initiation codon of the M29321 reference sequence). The Bm86 sequence of the Mozambique R. microplus strain reported here had a deletion of 66 nucleotides between positions 554 and 619 not found in other Bm86 sequences, which suggested that this region encoding for 22 amino acids may not be important for protein function. The Bd86 sequence of the South African R. decoloratus strain had an 18 nucleotides insertion between positions 1,690 and 1,691, similar to Bd86-2 and three nucleotides longer than in Bd86-1 [31].
Pairwise nucleotide and amino acid sequence alignments were conducted between cloned Bm86, Ba86 and Bd86 sequences and those identified above to have identity to these sequences ( Table 1). The results showed that sequence identity was higher between Bm86 and Ba86 than with Bd86 sequences.

Production and characterization of P. pastoris strains for the expression of recombinant Bm86, Bd86 and Ba86
The plasmids pPAMoz9, pPADec8 and pBaI were transformed into P. pastoris strains GS115, KM71H and X33 for expression of recombinant Bm86, Bd86 and Ba86 proteins. Single colonies of P. pastoris transformants for each gene were grown in an orbital shaker under induction conditions. Culture supernatants were spotted on a nitrocellulose membrane for dot-blot analysis of recombinant proteins. Expression of Bm86 and Bd86 was obtained in GS115 and KM71H strains while Ba86 was expressed in strain X33 only ( Table 2). Expression levels varied between 1.0 and 6.0 mg·L -1 , representing 1.5% to 13.2% of total proteins in the supernatant ( Table 2). For recombinant Bm86 and Bd86, differences in expression levels were not observed between GS115 and KM71H strains. The highest expression levels were obtained for Ba86 in strain X33 ( Table 2). The recombinant strains GS115Moz9-2, KM71HDec8-1 and X33pBaI-3 with highest expression levels of Bm86, Bd86 and Ba86, respectively, were selected for fermentation scale up in a 5-L bioreactor.
The GS115Moz9-2, KM71HDec8-1 and X33pBaI-3 high expression strains had a Mut S phenotype (Table 3). It has been demonstrated that transformation of P. pastoris with plasmids using the AOX1 expression system may lead to three mutant phenotypes with regard to methanol utilization [32]. The Mut + phenotype grows on methanol at the wild-type rate and requires high feeding rates of metha-nol, the Mut S phenotype has a disruption in the AOX1 gene and has a slower specific growth rate in methanol and the Mutis unable to grow in methanol. Although transformation of X-33 and GS115 strains with linearized constructs favor single crossover recombination at the AOX1 locus and generates a Mut + phenotype, double crossover recombination that results in the disruption of the wild-type AOX1 gene and the generation of a Mut S phenotype is possible. The P. pastoris strains with a Mut S phenotype grow slower in methanol but may be better hosts for the secretion of recombinant proteins [33].

Expression of recombinant Ba86, Bd86 and Bm86 proteins in P. pastoris
The GS115Moz9-2, KM71HDec8-1 and X33pBaI-3 strains were used for bench-top fermentation exploiting the methanol utilization ability of P. pastoris strains in PM medium. This medium was previously used for P. pastoris fermentations to express high levels of recombinant Bm86 [24,34].
The initial phase of the fermentation process (biomass production phase) ended after 20-24 hrs and induction of recombinant protein expression started at the onset of methanol-adoption and utilization phases. As expected, all strains behaved similarly when growing on glycerol as the sole carbon source (Table 3). Cell densities before induction and maximum growth rates on glycerol were very similar and similar to those previously reported in P. pastoris [33,35].
The selected fed-batch strategy to feed methanol was identical for all strains. Once glycerol used as carbon source in the initial batch and fed-batch phases was consumed, recombinant protein expression was induced by the addition of methanol to the culture medium. An exponential growth phase was then observed during the next 20-24 hrs with maximum growth rates of 0.005, 0.002 and 0.003 h -1 for the strains GS115Moz9-2, KM71HDec8-1 and X33pBaI-3, respectively. However, after 24 hrs growth in methanol, cells stop growing and a steady increase in pO 2 levels revealed that a stationary growth phase was achieved. Nevertheless, total protein production continued to increase gradually to 274, 194 and 170 mg·L -1 for the strains GS115Moz9-2, KM71HDec8-1 and X33pBaI-3, respectively (Table 3 and Figs. 1 and 2).
In this first approach to obtain recombinant Bm86, Bd86 and Ba86 secreted to the culture medium, methanol was supplied at 1 ml·h -1 ·L of the initial fermentation volume for the first two hrs and then methanol supply was increased in 10% increments every 30 min to a rate of 3 ml·h -1 ·L. This strategy probably did not allow maintain-ing a steady concentration of methanol throughout the whole fermentation process and either starvation or accumulation of methanol could have occurred. This fact may explain lower growth rates and expression levels of recombinant Bm86, Bd86 and Ba86 when compared to the 65 g·L -1 dry weight and 1.5 g·L -1 of recombinant protein previously reported for membrane-bound Bm86 in P. pastoris [11,24,34]. These results suggest that recombinant Bm86, Bd86 and Ba86 protein expression levels may be increased by the optimization of the fermentation and methanol induction processes.  The presence of recombinant proteins in the culture supernatant was demonstrated at the end of the fermentation process by SDS-PAGE and Western blot (Fig. 3). Recombinant Bm86, Bd86 and Ba86 secreted in P. pastoris appeared in SDS-PAGE and Western blots as a major wide band with a size range of 100 to 110 kDa and smaller degradation fragments (Fig. 3). The recombinant Bm86 previously expressed in P. pastoris also had degradation products and a major wide band, but with a size ranging from 90 to 100 kDa [11]. These differences in estimated molecular weight of the proteins may be due to strain differences in glycosylation, which is responsible for the wide appearance of the protein band in the SDS-PAGE and Western blot [11].

Protein recovery and purification
To obtain a clarified supernatant for recombinant protein purification, a primary centrifugation step was performed at 3,900 × g. Due to the fact that P. pastoris culture centrifugation at g-forces between 3,000-5,000 results in a significant product entrainment [36], a washing step of cell pellets was made for the full recovery of secreted proteins.
P. pastoris secretes few autologous proteins [37]. Therefore, heterologous protein secretion serves as the major first step in recombinant protein purification. However, unclear supernatants and recombinant protein purities ranging between 55% and 66% suggested the presence of contaminants in the supernatant after cell separation Secretion of recombinant Bm86, Bd86 and Ba86 by P. pastoris Characterization of the growth of P. pastoris strains during the fermentation process Canales et al., Figure 1 Characterization of protein secretion in P. pastoris strains during the fermentation process Canales et al., Figure 2 ( Table 4). This observation suggested that probably cell lysis occurred during the stationary phase of the fermentation process due to suboptimal growth conditions. Cell lysis during the fermentation may have contributed to protein degradation, thus affecting recombinant protein yield and reinforcing the need for optimization of the fermentation process to reduce protein degradation and increase expression levels.
It has been demonstrated in previous cell fractionation experiments of P. pastoris that a wide range of particles densities and sizes are present in a disrupted suspension of the yeast [38,39]. Therefore, to separate particles in suspension from secreted recombinant proteins, supernatants were filtered throughout 5, 0.45 and 0.22 μm filtration systems, which resulted in 20-25% increase in recombinant protein purity (Table 4). Finally, size exclusion and diafiltration through a 50 kDa cut-off membrane resulted in 80-85% pure recombinant proteins (Table 4 and Fig. 4).
The purity of recombinant proteins reported herein after protein secretion and a simple centrifugation-filtration purification process was higher than that obtained for membrane-bound Bm86 [24,34]. The purification of the membrane-bound Bm86 required cell disruption, washing of cell pellet, denaturation, renaturation and protein precipitation procedures [24,34]. In spite of the high level expression obtained during fermentation [11,34] and the optimization of the purification process [24,40-43] for the membrane-bound Bm86, the secretion of recombinant Bm86 in P. pastoris reported herein allowed for higher recovery and purity of recombinant protein after purification.
Additionally, an important advantage of secreting recombinant proteins in P. pastoris, particularly for proteins with complex structures and a high number of disulfide bonds such as Bm86 [44], is that the isolation of a membranebound form under denaturing conditions followed by refolding is very unlikely to reform all disulfide bonds correctly and reproducibly. By contrast, if disulfide bond formation occurs through the natural cell processing and secretion machinery as reported herein, the product is more likely to have a reproducible conformation closely resembling the native protein.
Other expression systems using arthropod cell lines have been considered. However, despite recent advances in the application of insect cell culture technology for the production of recombinant proteins, the process is still more expensive and difficult to scale-up when compared to proteins expressed in E. coli and P. pastoris [45]. The secretion of recombinant Bm86 ortholog proteins reported here in P. pastoris is easy to scale-up, simple, reproducible and likely to result in a product with high antigenicity and immunogenicity [28,29].

Characterization of recombinant Bm86, Bd86 and Ba86
Although differences may exist in antigen recognition between cattle and rabbits [46], rabbits have been proven to recognize some Bm86 protective epitopes [11,47] and were therefore considered a suitable host to evaluate immune cross-reactivity between recombinant Bm86 ortholog proteins.
The purified recombinant Bm86, Bd86 and Ba86 were adjuvated and used to immunize rabbits. The sera from immune rabbits were used to evaluate by Western blot the immune cross-reactivity between Bm86 ortholog proteins. The results showed that recombinant Bm86, Bd86 and Ba86 contained cross-reactive epitopes (Fig. 5)

Conclusion
We have cloned and secreted in P. pastoris the recombinant R. microplus, R. decoloratus and R. annulatus Bm86 orthologs from African or Asian tick strains. To our knowledge, this is the first study of Bm86, Bd86 and Ba86 secretion in P. pastoris. The results reported herein have shown that in P. pastoris, Bm86 ortholog recombinant proteins are secreted and purified from the culture supernatant with high yield and purity. The purification process for secreted proteins was simpler than that described for membrane-bound Bm86, which suggests the possibility of simplifying the purification process for recombinant Bm86 when secreted in P. pastoris. Additionally, secretion of recombinant Bm86 ortholog proteins in P. pastoris is likely to result in a more reproducible conformation closely resembling the native protein. Finally, the preliminary immunological characterization of recombinant Bm86, Bd86 and Ba86 evidenced the presence of crossreactive epitopes among these proteins. These results suggest that these recombinant antigens can be used for the development of vaccines for the control of tick infestations in Africa. The control of livestock Rhipicephalus spp. infestations in Africa would contribute to improve animal health and production in this region.  Immune cross-reactivity between Bm86 ortholog proteins . RNA was reverse transcribed for 45 min at 45°C prior to PCR consisting of an initial step of 2 min at 94°C followed by 35 cycles of a denaturing step of 30 sec at 94°C and an annealing-extension step of 2 min at 68°C. Control reactions were done using the same procedures, but without RNA added to control contamination of the PCR. PCR products were electrophoresed on 1% agarose gels to check the size of amplified fragments by comparison to a DNA molecular weight marker (1 Kb Plus DNA Ladder, Promega). The amplicon was resin purified (Wizard, Promega) and cloned into pGEM-T vector (Promega

Construction of expression plasmids
Bm86, Ba86 and Bd86 coding regions were excised from pGEM-T by Xho I and Xba I digestion (restriction sites introduced during PCR by CZABM5 and CZABM3 primers, respectively) and cloned into P. pastoris expression vector pPICZαA (Invitrogen) digested with Xba I and Xho I. In this way, Bm86 orthologs were cloned under the control of the alcohol oxidase (AOX1) promoter, in frame with the yeast alfa-factor secretion signal but without the C-terminal c-myc/His tag due to a translation termination site introduced by CZABM3 primer during PCR. The expression constructs were sequenced at both ends and selected constructs with correct sequences were named pPAMoz9 (Bm86), pPADec8 (Bd86) and pBaI (Ba86) and used for transformation of P. pastoris.

Pichia pastoris transformation and screening for recombinant protein expression
Expression plasmids were linearized by restriction with Sac I and transformed into P. pastoris strains GS115, KM71H and X33 (Invitrogen) by electroporation as described [49]. Transformants were selected on YPDS plates containing 100 μg·ml -1 Zeocin and incubated at 30°C. A functional assay to directly screen for high expression recombinant clones was made by culturing the transformants in an orbital shaker at 250 rpm and 30°C. Single colonies were inoculated in 1 ml YPDS containing 100 μg·ml -1 Zeocin and grown overnight. Cultures were divided into two parts of 500 μl each. Five hundred μl were transferred to 5 ml fresh YP medium with 20 g·L -1 glycerol, grown for 24 hrs and inoculated into 250 ml fresh YP medium supplemented with 20 g·L -1 glycerol. Growth in glycerol was resumed after 24 hrs and then methanol was added daily to 1% (v/v) during the course of induction. After 5 days growing on methanol, supernatants were collected by centrifugation for 15 min at 15,000 × g in a Beckman Allegra™ X-22R centrifuge, rotor F2402H (Beckman-Coulter, Palo Alto, CA, USA) and dot blots were made to screen for expression of recombinant proteins. The other 500 μl were also transferred to 5 ml fresh YP medium with 20 g·L -1 glycerol, grown for 24 hrs and mixed with glycerol to 250 g·L -1 . Long term stocks were prepared as 100 μl aliquots and stored frozen at -80°C. Plates were incubated at 30°C for 3 days and cell growth was observed and compared to controls.

Fermentation process
Pre-inoculums and inoculums for bioreactor cultures were grown in a shaker at 30°C and 250 rpm. Two 100 μl long term stock vials were seeded in 1 ml YP medium, grown for 12 hrs and transferred into 4 × 50 ml tubes containing 5 ml of YP medium with 20 g·L -1 glycerol. After 24 hrs, cultures were mixed again and 5 ml were used to inoculate 2 L Erlenmeyer flasks containing 250 ml of YP medium with 20 g·L -1 glycerol. Cells were grown to an O.D. 600 nm between 15 and 20 and then cultures were inoculated into a 5-L working volume Biostat B bioreactor (B. Braun Biotech, Melsungen, Germany) containing 3.5 L of PM with 40 g·L -1 glycerol.
During the fermentation process, temperature was kept at 30°C and dissolved oxygen was maintained at 30% saturation by regulating agitation and aeration rates. A threephase cultivation protocol was used in the fermentation. The glycerol growth phase included a 12 to 14 hrs batch stage from the starting point followed by a 10 to 12 hrs glycerol fed-batch stage. A glycerol solution of 50% (v/v) was added to the fermentor for 4 hrs to reach an equivalent total quantity of 60 g·L -1 in the culture medium. Upon exhaustion of glycerol, indicated by a sharp increase in dissolved oxygen, methanol induction was made by adding 1% (v/v) methanol to the culture medium and 3 hrs later the fed-batch phase was started by feeding methanol according to the P. pastoris Fermentation Process Guideline [49]. The pH was allowed to drop to 3.5 during the whole glycerol phase and it was maintained in this value by the addition of NH 4 OH. Prior to methanol induction, pH was adjusted and maintained at 5.5 by adding NH 4 OH or H 3 PO 3 . Throughout the fermentation processes, supplements of 20 ml TES and VT solutions were added to the culture medium every 24 hrs. Additionally, GS115 strain cultures were supplemented with 0.04 g·L -1 L-Histidine every 24 hrs.

Vaccine formulation and analysis
Prior to adjuvation of the vaccine, protein solutions were adjusted to a concentration of 120 μg·ml -1 and filtered through 0.45 and 0.22 μm cartridges (Sartorius AG) under sterile conditions in a laminar flow to obtain a sterile antigen solution. Adjuvation was made by mixing a solution of anhydromannitoletheroctodecenoate (Montanide ISA 50 V; Seppic, Paris, France) with the recombinant protein solution in batch-by-batch processes using a high-speed mixer Heidolph DIAX 900 (Heidolph Elektro, Kelheim, Germany) at 8,000 rpm and the vaccine was filled manually under sterile conditions in glass bottles of 20 ml (Wheaton, Millville, NJ, USA). Quality controls were made by testing mechanical and thermal stability of vaccine emulsions as described by Canales et al. [24].

Rabbit immunization with recombinant proteins
Two New Zealand White rabbits per group was each immunized with 3 doses (weeks 0, 4 and 8) containing 50 μg/dose of purified recombinant proteins formulated as described above or Gavac (Revetmex, Mexico City, Mexico) as control. Rabbits were injected subcutaneously with 1 ml/dose using a 1 ml tuberculin syringe and a 27 1/2G needle. Two weeks after the last immunization, blood samples were collected from each rabbit into sterile tubes and maintained at 4°C until arrival at the laboratory. Serum was then separated after centrifugation and stored at -20°C. Rabbits were cared for in accordance with standards specified in the Guide for Care and Use of Laboratory Animals.

SDS-PAGE, dot blot and Western blot analyses
Protein samples were analyzed by denaturing SDS-PAGE with a 12% PAGEgel-SDS cassette gel (PAGE-gel Inc, San Diego, CA, USA) under reducing conditions. Protein bands were visualized by either Coomassie Brilliant Blue R250 or silver staining. Samples were treated with dithiothreitol (DTT) reducer (PAGE-gel Inc.), heated in boiling water for 5 min before loading onto the gel and electrophoresed for 80 min at 90 mA constant current.
Electrophoretic transfer of proteins from gels to nitrocellulose membranes (PROTRAN BA85; Schleicher and Schuell, Dassel, Germany) for Western blot analysis was carried out in a Minie-Genie Electroblotter semi-dry transfer unit (Idea Scientific, Corvallis, OR, USA) according to manufacture's protocol. Protein samples of 2 μl were absorbed onto nitrocellulose membrane by gravity flow to perform the dot blot analysis. A standard curve was constructed with known amounts of recombinant Bm86 extracted from Gavac (Revetmex) and was used for semiquantitative analysis in dot-blots. The supernatant of the GS115/Albumin strain (Invitrogen) grown under the same conditions was used as a negative control in both dot-and Western-blots. Membranes for dot or Western blots were blocked with 5% skim milk for 1 hr at room temperature, washed three times in TBS (25 mmol/L Tris·HCl, 150 mmol/L NaCl, pH 7.6) and probed with sera from rabbits immunized with Gavac (Revetmex) (1:1000 dilution) or recombinant proteins (1:5000 dilution) as described above. The antisera were diluted in 3% BSA in TBS. Membranes were then washed three times with TBS and incubated with an anti-rabbit IgG horseradish peroxidase (HRP) conjugate (Sigma-Aldrich) diluted 1:1000 in TBS. After washing the membranes again, color was developed using TMB stabilized substrate for HRP (Promega).