Production of Recombinant HA1 subunit from H5N1 Avian Influenza Hemagglutinin in Escherichia coli

Influenza A/H5N1 represents a potential risk of a worldwide pandemic event, and as we have witnessed in past influenza outbreaks, the current production chains for vaccination cannot supply the demand for emergencies such as these. Limiting factors include the need for eggs and active virus handling as well as turnaround time and cost. Most of the influenza AH5N1 cases are contained in poultry; however, risk of zoonosis and spread of the virus in humans grows by the day, thus, a mass-produced fast-responding preventive vaccine is required. In this article, we describe the production of a recombinant HA1 subunit of influenza AH5N1 hemagglutinin as a potential key ingredient of a new avian vaccine, using a prokaryotic biotechnological platform. This system is potentially faster, cheaper, and more efficient than current means of vaccine production.

outbreak, local availability of chicken egg will decrease, hampering the production of inactivated virus vaccine, accompanied with an increase of costs per unit [5,6].
Vaccination with live viruses relies heavily on the immune system effective response to hemagglutinin (HA), a homotrimeric integral membrane glycoprotein with each monomer consisting of two domains: a hydrophilic domain (HA1) that binds to sialic acid (a polysaccharide frequently found in receptors on the host cell surface) and a hydrophobic domain (HA2) that anchors the protein in the lipid component of the virus envelope. HA1 is not only responsible for fusion of the viral envelope with the cell membrane, it is the main immunogenic agent within the virus as well [7]. Indeed, serum containing IgG capable of recognizing this protein confers immunity against IAV infection, provided that the subtype has a HA identical to that recognized by the vaccine [8][9][10]. Given the characteristics of the protein, HA is the optimal choice for creating a subunit vaccine, to develop a faster and cheaper technology to cover for the needs of vaccination programs [11].
A number of alternative systems have been proposed to produce influenza vaccines, both for human use and veterinary application [12,13], which include but are not limited to viral propagation in mammalian or insect cells and resorting to recombinant(r) DNA generated proteins [14][15][16]. rHA production stands out among all of these options due to its reported low-cost, ease of implementation, fast turnaround time, and safety [17][18][19].
In this article, we describe the design, development, and evaluation of a reliable, fast, and relatively low-cost system for the production of rHA of IAV in Escherichia coli and discuss its potential as a platform for fast mass-production of influenza A sub-unitary vaccines for veterinary vaccination applications.

Design of expression system and cloning of the HA gene into the expression vector
The gene sequence taken as reference for the design of the rHA expression system was IAV (A/chicken/Egypt/1063/2010(H5N1)) segment 4 HA complete coding sequence (CD) (GenBank accession number HQ198269.2). The sequence was optimized for expression in E. coli, using the GeneOptimizer® assisted Sequence Analysis service (LifeTechnologies, Inc. Carlsbad California, USA).
The 568-amino acid (aa) sequence encoded by the aforementioned gene (accession number ADM85860.1) was analyzed using the Simple Modular Architecture Research Tool (SMART) [20] to identify the domains present in the protein.
Subsequently, its signal peptide (aa 1 to 16) and its transmembranal region (aa 532 to 568) were removed from the sequence to avoid interference with the three-dimensional structure of the protein, leaving a 515-aa chain. A methionine-coding codon was added at the beginning and two stop codons at the end of the corresponding synthetic nucleotide sequence to generate the first CD to be expressed, which was called HAq. A second sequence, named HAp, coding only the globular subunit of the HA1 was generated, comprising the aa residues 17 to 338 from the reference sequence, plus a start methionine-coding codon and two stop codons as well (see Table 1). The putative three-dimensional structures of both constructs were predicted using the SWISS-MODEL tool [21]. The theoretical molecular weight and isoelectric point of the proteins to be encoded by HAq and HAp sequences were inferred informatically by using the tool ProtParam, included in the ExPASy protein informatics package [22].
The two sequences were synthesized by GeneArt (Invitrogen, Carlsbad California, USA), and cloned by PX' Therapeutics (Grenoble, France) into the expression vector pET-30b(+), carrying kanamycin resistance and the T7 promoter under the regulation of an element of the lac operon, rendering the plasmids named pVIT_HAp and pVIT_HAq. Figure 1 shows a simplified map of these plasmids. The recombined plasmids were then propagated in Escherichia coli BLR (DE3) (Novagen, Merck, Darmstadt, Germany), with genotype F-ompT hsdSB(rB -mB -) gal lac ile dcm Δ (srl -recA) 306::Tn10 (tetR) (DE3), for which Ca 2+ competent cells were prepared and transformed by heat shock. Transformed cells were cultured in Luria-Bertani agar medium containing 50 µg/ml of kanamycin seeds. Then, they were grown overnight and the colonies obtained were analysed by polymerase chain reaction (PCR) for verification of successful transformation. Oligonucleotides required to verify by PCR the presence of our synthetic constructs in the plasmids containing bacteria were designed and tested in silico using OligoAnalyzer 3.1 (Integrated DNA Technologies®, Coralville Iowa, USA) and PrimerBlast3.1 [23]. The primers and their physicochemical properties are described in Table 2. Amplicons were analyzed by agarose gel electrophoresis. for batch culture supplemented with 30µg/mL kanamycin. At an OD600 of 1.5, each of the inoculums were added to new tuves containing 90 mL of MRM, in 500 mL baffled beakers, which, once reaching an OD600 of 1.5, were used each to inoculate 400 mL of MRM in a one-liter BioFlo® Bioreactor (Eppendorf, Hamburg, Germany). The cultures were kept at 37°C with stirring conditions required to keep dissolved oxygen constant until an OD600 of 1.5 was reached. Once this density was reached, temperature was kept at 25°C 1M IPTG was added to a final concentration of 0.5mM, to attain the induction stage, constantly monitoring the process throughout 20 h. Cells harvesting was performed centrifuging at 13,000g for 25 min at 4°C. The recovered biomasses were resuspended in 50 mL of depleted MRM (10x more concentrated that the original harvest) and then preserved at -20°C.

Lysis and preparation of inclusion bodies
From the concentrated harvest, aliquots (10% from the total volume) were taken and centrifuged for 20 min at 12,000g at 4°C. Pellets were resuspended in 30 mL of 50 mM potassium phosphate (monobasic) solution to wash every pellet, obtaining an OD600 of approximately 10. Samples were centrifuged again for 20 min at 12,000g, at 4°C. The supernatants were discarded, and the pellet were resuspended in 3 mL of lysis buffer (50 mM Tris, 5 mM EDTA, 1 mg/mL lysozyme, 0.5 M, NaCl, 1 µg/mL pepstatin A, and 1 µg/mL leupeptin, pH 7.5) and distributed in 1 mL aliquots (to facilitate handling and speed up freezing/thawing). The aliquots were subjected to 6sixfreeze-thaw cycles by placing the samples in a tube rack, inside an ultra-freezer (REVCO, Thermo Electron Corporation Waltham, Massachusetts, USA) at -70°C for 5 min. Thawing immediately at 42°C with agitation (300 rpm) in thermomixer (vortex vigorously to mix well). The lysate were centrifuged 5 min at 12,000g at 4°C and the supernatant were discarded to recover the pellet containing the inclusion bodies.

Solubilization and refolding of recombinant rHA
Solubilization of inclusion bodies found in the lysates were carried out as follows: the pellets obtained in the previous stage were resuspended and pooled in 6 mL solubilization buffer (1 mM glycine, 1 µg/mg pepstatin A, 1 µg/mL leupeptin, 50 mM sodium phosphate (monobasic), 5 mM β-mercaptoethanol, and 8 M urea) and incubated overnight at 6-8°C, and then centrifuging at 12,000g for 20 min at 6-8°C, collecting the supernatant and preserving it at -20°C until further processing. In order to allow the rHA to properly fold, 900 µL refolding buffer (400 mM arginine, 2 mM EDTA, and 4 mM CHAPS in PBS at pH 7.3) were gradually added to 100 µL of solubilized protein suspensions under a light vortexing speed and after this, speed was gradually increased for 10 sec to ensure homogeneity. The refolded protein suspensions were then centrifuged at 15,000g for 30 min at 6°C, separated from the scarce pellet formed, and incubated overnight at 6-8°C. These supernatants were stored at -20°C. Recovery percentage were estimated using native SDS-PAGE as described elsewhere [25] and measured as described previously [26].

Immunological assay
An Enzyme Linked Immunosorbent Assay (ELISA) was carried out according to manufacturer instructions using the H5N1 (Avian Flu) Hemagglutinin Elisa Pair Set SEK002 (Sino Biological Inc., Beijing, China). Standard of HA provided by the kit was used to read a curve of serial dilutions to calculate the concentration of rHA in the unknown samples. The samples included in the assay were direct and diluted aliquots from the solubilized rHA, the protein diluted in refolding buffer, a rHA control provided by PX'Therapeutics, and an inactivated IAV H5N1.

Rational design of HAp and HAq.
Prior to decide the CDs to be synthesized to render the expression cassettes for rHA form AIV, structural analysis of the  Figure 2). In order to maximize three-dimensional similarity between native hemagglutinin and the recombinant protein, two sequences were designed taking these features into account: HAp, which contained only the viral protein domain, and HAq, which comprised both the viral protein domain and the HA2 domain. Signal peptide was removed to ensure epitope structure, because this sequence is not found in functional hemagglutinin. The transmembrane region was also removed to improve stability and decrease hydrophobicity.
Physicochemical properties of HAq and HAp are shown in Table 3.
Once the theoretical properties of both proteins were inferred bioinformatically, both sequences were modeled threedimensionally to characterize their behavior in three-dimensional space considering primary structure. These models are shown in Figure 3. Results of this analysis were matched with previously reported three-dimensional models and crystallographic analysis of isolated native hemagglutinins [SMTL ID: 4kth.2 (Structure of A/Hubei/1/2010 H5 HA); 2wr1.1 (Structure of influenza H2 hemagglutinin with human receptor)]. This theoretical data supported the hypothesis that these synthetic constructs could have the same structure as their native counterparts.

Expression of recombinant hemagglutinin
After expression screening of transformed bacteria, the clones containing pVIT-HAp demonstrated a much higher expression level than those containing pVIT-HAq (data not shown) and were thus chosen for subsequent production of the rHA. The rational for this situation, implies that is easier to express small versions of the rHA for bacterial cells. In this way it is possible to increase the yields for rHA expression using the same culture media and expression strains. Additional to this, (See Figure 6) Riesenberg media demonstrate the higher growth rate (LB 0.248h-1, Riesenberg 0.356h-1 and Terrific broth 0.332h-1) to culture the recombinant E. coli cells being a good option for scale up the rHA expression. During the fermentation, a series of culture samples were taken for monitoring the process; particularly, optical density (absorbance at 600 nm), as a biomass yield indicator. Besides, the rHAp yield was estimated by SDS-PAGE and bands densitometry.  Figure 7 shows the protein profiles of the samples taken throughout the semipurification process, from the harvest to the solubilization of the rHAp.

Immunological assay
All of the samples tested through ELISA showed detectable immunoreactivity. The highest titer was given by the solubilized rHAp, followed by the refolded protein, the PX'Therapeutics' batch, and finally, the inactivated IAV used as positive control came from a Mexican pharmaceutical company (Laboratorio Avi-Mex S.A. de CV.). The ELISA showed that the hemagglutinin produced in this work had a similar immunoreactivity as a commercial veterinary-grade hemagglutinin. The results can be seen in Figure 8.

Discussion
Using the biotechnological platform previously described, a recombinant modified Influenza A/H5N1 HA1 domain, suitable for subunitary vaccine production was prepared safely, without need of eukaryotic components, in a short time, and at high yield. An overall performance of this approach can be seen on Table 1. Despite the theoretical differences between the rHA and the hemagglutinin as part of viral particles, the ELISA assays demonstrated that the former was recognized by anti-HA antibodies. Hemagglutination assays (not shown) carried out using HAp did not display hemagglutination activity. Similar results were obtained by other groups [27] with some hemagglutinins tested. This could be due to the lack of a structure able to keep several HA1 domain particles together, as it does the viral particle in vivo. However, HAp showed in vitro immunoreactivity, consistent with guidelines for vaccines in poultry.
One of the main advantages of the platform is that the gene encoding HAp can be modified or constructed de novo using genetic editing or chemical synthesis with slight adjustments to the fermentation and purification processes, to produce abundant quantities of relatively pure rHA in independence from live virus and cell culture or pathogen-free chicken eggs.
This represent a considerable advantage as compared to current methods, which require as much as two eggs per dose [28] and no less than six months after the strain is identified and isolated [29]. The reliability and flexibility of the system was tested by adapting the platform for the production of rHA1 from the A/Egypt/1063/2010(H5N1) chicken isolate but can be adapted for production of other IAV strains, which are currently undergoing evaluation. Given that bacterial expression system such as the one described in this article are commonplace within biotechnological production chains, it can also be adopted in case of a pandemic event, since most academic institutions and biotechnological, and pharmaceutical companies possess the equipment needed for the implementation of this platform.
It is important to realize that these subunit antigens for vaccines, although promising in its initial results, has yet to be tested in live animals. Several subunitarian influenza virus vaccines have been developed based on rHA, using mammalian, insect, or bacterial cells as expression systems, and testing shows very encouraging results both in humans and animals [30,31].
However, previous testing with baculovirus-dependent systems have shown that protection against IAV H5 HA induced a neutralizing antibody response in only 23% of individuals with a single dose of 90 µg and a maximum of 52% after two doses of 90 µg [2]; results that do not comply with guidelines for vaccines in case of pandemics or epidemics due to the high cost of production, high dosage required, and deficient capability to immunize the subject. At the same time, several adjuvants are in development, so reformulation of recombinant hemagglutinin-dependent vaccines with the inclusion of one of these adjuvants could increase their immunogenic capabilities and surpass the need for large doses [32,33]. Other reports evaluated the immunogenicity of partial rHA1 sequences produced in bacteria against immunogenicity of complete rHA produced in insect cells. The results of these studies show that non-glycosylated partial region from recombinant hemagglutinin elicited neutralizing titers only four times lower than the insect cells-derived HA, although both inhibited viral entry into host animal cells [30].
Even considering the remaining challenge research on alternative production platforms of recombinant vaccines against IAV, the subunitarian approach looks promising. The available platforms for chicken egg-dependent vaccine production do not have the wherewithal to either rapidly or broadly respond in case of a pandemic or epidemic outbreak [34,35]. Further, studies with our platform will include modification of the protein, testing with adjuvants, assessment in live poultry, and improvement of the purification methodology. Still, our methodology provides for a scalable cost-effective platform for the production of AIV vaccines, which can potentially be adapted for different variants as required, that we anticipate will be ready to meet the required standards to cover the needs in the case of a health crisis.

Competing interests
The authors declare that they have no potential competing interests.