|Home | About | Journals | Submit | Contact Us | Français|
The impending influenza virus pandemic requires global vaccination to prevent large-scale mortality and morbidity, but traditional influenza virus vaccine production is too slow for rapid responses. We have developed bacterial systems for expression and purification of properly folded functional hemagglutinin as a rapid response to emerging pandemic strains. A recombinant H5N1 (A/Vietnam/1203/2004) hemagglutinin globular domain (HA1) was produced in Escherichia coli under controlled redox refolding conditions. Importantly, the properly folded HA1(1-320), i.e., HA1 lacking amino acids 321 to 330, contained ≥75% functional oligomers without addition of foreign oligomerization sequence. Site-directed mutagenesis mapped the oligomerization signal to the HA1 N-terminal Ile-Cys-Ile residues at positions 3 to 5. The purified HA1 oligomers (but not monomers) bound fetuin and agglutinated red blood cells. Upon immunization of rabbits, the oligomeric HA1(1-320) elicited potent neutralizing antibodies against homologous and heterologous H5N1 viruses more rapidly than HA1(28-320) containing only monomers. Ferrets vaccinated with oligomeric HA1 (but not monomeric HA1 with the N terminus deleted) at 15 and 3 μg/dose were fully protected from lethality and weight loss after challenge with homologous H5N1 (A/Vietnam/1203/2004, clade 1) virus, as well as heterologous clade 2.2 H5N1 (A/WooperSwan/Mongolia/244/2005) virus. Protection was associated with a significant reduction in viral loads in the nasal washes of homologous and heterologous virus challenged ferrets. This is the first study that describes the presence of an N-terminal oligomerization sequence in the globular domain of influenza virus hemagglutinin. Our findings suggest that functional oligomeric rHA1-based vaccines can be produced efficiently in bacterial systems and can be easily upscaled in response to a pandemic influenza virus threat.
The recent global spread of swine-origin H1N1 highlighted the need for rapid development of effective vaccines against pandemic influenza viruses. Much of our recent knowledge was derived from studies with the highly pathogenic (HP) H5N1 avian influenza A viruses (AIV) (24). The H5N1 viruses still cause severe human disease with >60% mortality and may undergo adaptation for human-to-human transmission.
Antibodies specific to hemagglutinin (HA) are believed to be the best correlate of protection against influenza virus infection and are the primary endpoint used to evaluate vaccine immunogenicity. The production of HA using recombinant technology could overcome the constraints of traditional influenza virus vaccine manufacturing that (i) require several months for the generation of vaccine viruses using reassortment/reverse genetics and adaptation for high growth in eggs, (ii) suffer from bottlenecks at every step, (iii) are expensive, and (iv) are dependent on the supply of eggs. However, the use of recombinant HA proteins poses several challenges; in addition to proper folding of the HA monomers, trimer formation is an important property of native HA spike proteins required for cell attachment (32) and for optimal immunogenicity (28). On virions, the trimeric HA complex is stabilized by three 76-Å-long helices that form a triple coiled-coil structure and consist of residues primarily from the HA2 region. Stability studies indicated that the HA2 tails contribute 28.4 kcal mol−1 and that the HA1 heads contribute only 5.3 kcal mol−1 to the stability of the trimers (10, 31). The expression of recombinant HA ectodomain in mammalian cells required the addition of multimerization “foldon” at the C terminus in order to produce stable oligomeric structures (28). Therefore, the prediction was that HA1 globular head (without HA2) will not form stable trimers (2).
Expression of recombinant HA proteins in bacterial systems could provide a rapid and economical approach for early response to impending influenza virus pandemic. However, it was not clear whether unglycosylated proteins will present antigenically relevant epitopes. Most of the influenza virus protective antigenic sites are conformation dependent and map primarily to the HA1 globular head (22, 30). Previously, we used H5N1 whole-genome phage display libraries to map the antibody repertoires following human infection with HP H5N1 (A/Vietnam/1203/2004) AIV, as well as in post-H5N1 vaccination sera (11, 12). We have identified large HA1 fragments, encompassing the receptor-binding domain (RBD), that bound broadly neutralizing human monoclonal antibodies and polyclonal sera from H5N1 recovered individuals. Furthermore, in a recent study in our laboratory, bacterially expressed globular HA1(3-130) and HA ectodomain (1-480) derived from novel H1N1 A/California/04/2009 were compared. Both proteins were properly folded. However, only the HA1 globular head (1-330) formed oligomers and agglutinated human red blood cells (RBCs). In contrast, the HA ectodomain (1-480) contained only monomers and did not agglutinate RBCs (13).
To better understand the phenomenon of oligomerization of HA1 globular domain in the absence of HA2 sequence, we expressed a series of H5N1-derived HA1 proteins with N- and C-terminal deletions and point mutations and correlated their ability to form oligomers with functional HA properties, including receptor binding and agglutination of RBCs. Furthermore, to assess the importance of oligomerization for immunogenicity and cross-protection, these HA1 proteins were used in rabbit vaccinations and in the ferret influenza virus HP H5N1 virus challenge model. Our findings suggest that functional oligomeric rHA1 proteins can be produced efficiently in bacterial systems and applied for rapid development of effective vaccines against emerging influenza virus strains.
cDNA corresponding to the HA gene segment of H5N1-A/Vietnam/1203/2004 was generated from RNA isolated from egg-grown virus strain and was used for cloning. pSK is a T7 promoter-based expression vector where the desired polypeptide can be expressed as a fusion protein with a His6 tag at the C terminus (12). DNA encoding HA1(1-330) of the A/Vietnam/1203/2004 and its various amino- and carboxy-terminal deletions were cloned as NotI-PacI inserts in the pSK expression vector (see Fig. S1 in the supplemental material).
Escherichia coli Rosetta Gami cells (Novagen) were used for the expression of various H5N1-A/Vietnam/1203/2004 HA1 strains and its various deletions. After expression, inclusion bodies were isolated by cell lysis and multiple washing steps with 1% Triton X-100. A final inclusion body pellet was resuspended in denaturation buffer containing 6 M guanidine hydrochloride and dithioerythreitol at a final protein concentration of 10 mg/ml and was centrifuged to remove residual debris. For refolding, supernatant was slowly diluted 100-fold in redox folding buffer. The renaturation protein solution was dialyzed against 20 mM Tris-HCl (pH 8.0) to remove the denaturing agents. The dialysate was filtered through a 0.45-μm-pore-size filter and subjected to purification by HisTrap fast-flow chromatography.
To demonstrate that the bacterially expressed HA fragments are properly folded, they were analyzed by circular dichroism (CD) melt spectroscopy. For CD spectroscopy in solution, H1N1-HA proteins were dissolved in 20 mM phosphate-buffered saline (PBS; pH 7.4) at 0.5 mg/ml. The change in ellipticity at 222 nm (to follow the unfolding of α-helices) during unfolding was monitored by using a J-715 CD system (Jasco). The unfolding reaction was initiated by subjecting the protein in PBS to 1°C/min increments. The experiments were carried out in triplicate.
Proteins at a concentration of 5 mg/ml were analyzed on Superdex S200 XK 16/60 column (GE Healthcare) pre-equilibrated with PBS, and the protein elution was monitored at 280 nm. Protein molecular weight marker standards (GE Healthcare) were used for column calibration and the generation of standard curves to identify the molecular weights of the test protein samples.
Human erythrocytes were separated from whole blood (Lampire Biologicals). After isolation and washing, 30 μl of a 1% human RBC suspension (vol/vol in 1% bovine serum albumin [BSA]-PBS) were added to 30-μl serial dilutions of purified HA1 proteins or influenza virus in 1% BSA-PBS in a U-bottom 96-well plate (total volume, 60 μl). Agglutination was read after incubation for 30 min at room temperature.
Hemagglutination inhibition experiments were performed by using anti-H5N1 human monoclonal antibody (MAb) FLA5.10 (12). Experiments were performed as described earlier, except that prior to addition to the RBCs, the HA proteins were incubated for 15 min at room temperature with the human MAb.
Binding of different HA1 derivatives to fetuin (natural homolog of sialic acid cell surface receptor proteins) and its asialylated counterpart (asialo-fetuin) was analyzed at 25°C using a ProteOn surface plasmon resonance (SPR) biosensor (Bio-Rad Labs). Fetuin or asialo-fetuin (Sigma) were coupled to a GLC sensor chip with amine coupling at 1,000 resonance units in the test flow cells. Samples of 60 μl of freshly prepared H5N1-HA1 proteins at 10 μg/ml were injected at a flow rate of 30 μl/min (120-s contact time). The flow was directed over a mock surface to which no protein was bound, followed by the fetuin-coupled or asialo-fetuin-coupled surface. Responses from the protein surface were corrected for the response from the mock surface and for responses from a separate, buffer-only, injection. Binding kinetics and data analysis were performed using Bio-Rad ProteOn manager software (version 2.0.1).
Virus-neutralizing activity was analyzed in a microneutralization assay based on the methods of the pandemic influenza virus reference laboratories of the U.S. Centers for Disease Control and Prevention (CDC). The following low-pathogenicity H5N1 viruses, generated by reverse genetics, were obtained from the CDC: A/Vietnam/1203/2004 (SJCRH, clade 1), A/Indonesia/5/2005 (PR8-IBCDC-RG2, clade 2.1), A/Turkey/1/05 (NIBRG-23, clade 2.2), and A/Anhui/1/05 (IBCDC-RG5, clade 2.3.4). The experiments were conducted with three replicates for each serum sample and performed at least twice.
New Zealand rabbits were immunized thrice intramuscularly at 21-days interval with 100 μg of purified HA1 proteins and its derivatives with TiterMax adjuvant (TiterMax, Inc., Norcross, GA).
All ferrets (Marshall Farms) used in the study were confirmed as seronegative for circulating seasonal influenza A (H1N1 and H3N2) and influenza B viruses by hemagglutination inhibition. Female Fitch ferrets (n = 5 per group) were vaccinated intramuscularly in the quadriceps muscle on day 0, boosted on day 21, and challenged with virus on day 35. Control animals (n = 5) were mock vaccinated with PBS (pH 7.2). Each animal was vaccinated with one of two doses (15 μg or 3 μg) of recombinant HA in sterile 0.9% saline. Each vaccine was mixed with the adjuvant formulation, TiterMax, at a 1:1 ratio. The volume for all intramuscular vaccinations was 0.5 ml. The first and second vaccinations were given in the left and right hind legs, respectively. Blood was collected from anesthetized ferrets via the anterior vena cava. The collected blood was transferred to a tube containing a serum separator and clot activator and allowed to clot at room temperature. Tubes were centrifuged at 6,000 rpm for 10 min; the serum was separated, divided into aliquots, and stored at −80 ± 5°C. All procedures were in accordance with the National Research Council Guidelines for the Care and Use of Laboratory Animals, the Animal Welfare Act, and CDC/National Institutes of Health (NIH) Bio-Safety Guidelines in Microbiological and Biomedical Laboratories and were approved by the Institutional Animal Care and Use Committee (IACUC).
Animal experiments with H5N1 influenza virus were performed in the AALAC-accredited ABSL-3 enhanced facility. Animals were infected and monitored as previously described (35), except using 5% isoflurane anesthesia. Briefly, ferrets were anesthetized with isoflurane and infected intranasally with 106 50% egg infectious doses (EID50; i.e., ~105.75 50% tissue culture infective doses [TCID50]/ml) of A/Vietnam/1203/2004 (clade 1) or A/Whooperswan/Mongolia/244/2005 (clade 2.2) in a volume of 1 ml. Animals were monitored for temperature, weight loss, loss of activity, nasal discharge, sneezing, and diarrhea daily after viral challenge. To determine the viral load from nasal washes, 1.5 ml of 0.9% saline was administered to each nare, and the wash was collected each day postchallenge from each ferret. Temperatures were measured through use of an implantable temperature transponder (BMDS, Sayre, PA) and were recorded at approximately the same time each day. Preinfection values were averaged to obtain a baseline temperature for each ferret. Clinical signs of sneezing and nasal discharge, inappetence, dyspnea, neurological signs, respiratory distress, and level of activity were assessed daily. A scoring system was used to assess activity level as follows: 0, alert and playful; 1, alert but playful only when stimulated; 2, alert but not playful when stimulated; and 3, neither alert nor playful when stimulated. Based on the daily scores for each animal in a group, a relative inactivity index was calculated (35).
Viral loads in nasal washes were determined by the plaque assay. Briefly, MDCK cells plated in six-well tissue culture plates were inoculated with 0.1 ml of virus-containing sample, serially diluted in Dulbecco modified Eagle medium. Virus was adsorbed to cells for 1 h, with shaking every 15 min. Wells were overlaid with 1.6% (wt/vol) Bacto agar (Difco/BD Diagnostic Systems, Palo Alto, CA) mixed 1:1 with L-15 medium (Cambrex, East Rutherford, NJ) containing antibiotics and 0.6 mg of trypsin (Sigma, St. Louis, MO)/ml. The plates were incubated for 5 days. Cells were fixed for 10 min with 70% (vol/vol) ethanol and then overlaid with 1% (wt/vol) crystal violet. The cells were then washed with deionized water to visualize the plaques. Plaques were counted and compared to uninfected cells.
RDE (receptor destroying enzyme)-treated ferret sera were serially diluted in V-bottom 96-well microtiter plates, followed by the addition of 8 hemagglutination units of influenza virus. After an incubation of ~20 min, a 0.5% suspension of horse RBCs (HRBCs) in PBS (pH 7.2) were added and mixed using agitation. The HRBCs were allowed to settle for 30 min at room temperature, and hemagglutination inhibition titers were determined from the reciprocal value of the last dilution of sera that completely inhibited the hemagglutination of HRBCs. A negative titer was defined as 1:10.
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals (NIH). The protocol was approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh (permit A-3187-01). All surgery was performed under sodium pentobarbital anesthesia, and all efforts were made to minimize suffering.
To better understand the role of HA1 structure-function and its effect on generating protective immunity after immunization, we expressed a series of H5N1-derived HA1 proteins with N- and C-terminal deletions and evaluated their ability to form oligomers and to agglutinate RBCs. The intact H5N1 HA1 and a series of truncated proteins were expressed in E. coli and isolated from inclusion bodies by denaturation and slow renaturation under controlled redox refolding conditions as previously described (11, 12). The His6-tagged fusion proteins were purified using Ni-NTA chromatography to >95% purity (Fig. (Fig.1A1A and see Fig. S2 in the supplemental material). Proper folding was confirmed by binding to a panel of H5N1-neutralizing human MAbs that recognize conformational epitopes in the HA-RBD (12) and do not bind to unfolded HA proteins (11). As shown in Fig. Fig.1B1B and Fig. S3 in the supplemental material, all bacterially expressed HA1 proteins containing RBD bound human MAb FLA5.10 (as well as human MAbs [huMAbs] FLD21.140 and FLD.3.14) with similar kinetics, as determined by SPR. HA (1-104), which does not contain the RBD, did not bind to the three huMAbs.
Agglutination of red cells is a surrogate for the binding of influenza virus to its sialic acid receptors. The interaction between individual HA RBD and the sialyloligosaccharides moieties is rather weak (Kd > 10−4 M) (16), and increased avidity is accomplished by binding of multimeric HA spikes to multiple cell receptors. To determine whether the recombinant HA1 proteins contain functionally active forms, they were evaluated in a human RBC hemagglutination assay. As positive controls, we used the H5N1 vaccine strain rgA/Vietnam/1203/2004 and the licensed H5N1 inactivated vaccine (Fig. (Fig.1C).1C). Both HA1(1-330) and HA1(1-320) proteins agglutinated RBC, with endpoints of 97 and 4 ng/ml, respectively. CD melt spectroscopy (see Fig. S4 in the supplemental material) demonstrated that the HA1(1-320) protein was more stable than the HA1(1-330) protein (melting temperatures of 54.3 and 51.8°C, respectively). Therefore, the deletion of the 10-amino-acid sequence at the carboxy terminus of HA1 had a stabilizing effect on the HA1 protein and improved hemagglutination (Fig. (Fig.1C).1C). In contrast, all of the N-terminal deletions (i.e., amino acids 5 to 320, 9 to 330, 17 to 330, 28 to 330, and 28 to 320) did not agglutinate RBCs (Fig. (Fig.1C1C).
The hemagglutination mediated by HA1(1-330) and HA1(1-320) was specific since it was blocked in a concentration-dependent manner by preincubation with the H5N1-neutralizing huMAb FLA5.10 (but not by irrelevant MAb 2D7; data not shown) (Fig. (Fig.1D1D).
The hemagglutination results suggested that the intact HA1(1-330) and HA1(1-320), but not HA1 proteins with the N terminus deleted, contain higher-order quaternary forms required for RBC lattice formation. To address this possibility, the HA1 derivatives were subjected to gel filtration (Fig. (Fig.22 and Table Table1).1). It was found that HA1(1-320) contained ~80% high-molecular-weight (high-MW) oligomeric forms (Fig. (Fig.2A).2A). In comparison, all of the mutants with the N terminus deleted appeared as monomers only (Fig. 2B and C and Table Table1).1). The H5N1 inactivated subunit vaccine (Sanofi Pasteur) contained only oligomers (Fig. (Fig.2E).2E). Interestingly, HA1(1-104) segment, devoid of the RBD, also formed oligomers (Fig. (Fig.2D).2D). In addition to size chromatography, the monomeric and oligomeric peaks of HA1(1-320) were isolated from the gel filtration and were analyzed by native SDS-PAGE (Fig. (Fig.2F),2F), as well as by reduced SDS-PAGE (Fig. (Fig.2G).2G). The H5N1 vaccine was included as a positive control. In the native gel, monomeric fraction of HA1(1-320) ran at the expected MW (Fig. (Fig.2F,2F, lane 1), while the oligomeric fraction contained multiple high-MW species, similar to the H5N1 inactivated vaccine (Fig. (Fig.2F,2F, lanes 2 and 3). In SDS-PAGE under reducing conditions, the bacterially expressed HA1 monomeric and oligomeric fractions all ran as monomers (Fig. (Fig.2G,2G, lanes 1 and 2). As expected, the vaccine H5N1 HA was dissociated into HA1 and HA2 (Fig. (Fig.2G,2G, lane 3). Sedimentation velocity data collected by analytical centrifugation suggested that the oligomeric fraction of the rHA1(1-320) contained multiples of trimers with majority of oligomers consisting of four to six trimers (data not shown).
To further investigate which HA1 forms are required for receptor binding and RBC agglutination, we established a fetuin-based SPR assay that mimics the simultaneous interactions between the virion HA spikes with sialic acid moieties (23). All H5N1-HA1 mutants and truncated proteins were tested for binding to fetuin coated on biosensor chips. As shown in Fig. Fig.3A,3A, HA1(1-320) showed higher binding to fetuin-coated surface than did HA1(1-330). The H5N1 vaccine bound fetuin at rate similar to that of HA1(1-320) but dissociated more slowly (Fig. (Fig.3A).3A). The levels of fetuin binding by HA1(1-320) versus HA1(1-330) correlated well with the RBC agglutinations demonstrated in Fig. Fig.1C1C (top two rows) and confirmed that the 10-amino-acid C-terminus deletion stabilized the functional oligomeric HA1. No binding to asialo-fetuin was observed, confirming the binding specificity of these proteins to sialylated glycoproteins (Fig. (Fig.3B).3B). To better understand the role of monomers, trimers, and oligomers in receptor binding and hemagglutination, a preparative gel filtration column was used to isolate monomers, trimers, and oligomers of HA1(1-320). Only fractions containing trimers and oligomers, but not monomers, bound to fetuin in the SPR assay (Fig. (Fig.3C,3C, red, blue, and green curves). The monomeric HA1 was properly folded, as determined by binding to the three conformation-dependent human MAbs (data not shown), and yet it did not bind fetuin. All of the HA1 proteins with N-terminal deletions consisted only of monomers and did not bind fetuin (Fig. (Fig.3A3A and Table Table11).
In the hemagglutination assay, only the oligomeric fraction of HA1(1-320) agglutinated RBC (Fig. (Fig.3D).3D). The trimeric HA1 that bound to sialic acid receptor (Fig. (Fig.3C)3C) also bound to RBCs, but for hemagglutination the cross-linking of multiple RBCs is required to generate the lattice. This cross-linking was achieved by the addition of anti-His polyclonal IgG that bound to the C-terminal His6 tag (Fig. (Fig.3C).3C). On the other hand, monomeric HA1 did not agglutinate RBCs even in the presence of anti-His antibodies (Fig. (Fig.3C,3C, second row), suggesting that interaction between monomeric HA1(1-320) and the sialyloligosaccharides moieties is rather weak (16). The N-terminal HA fragment (amino acids 1 to 104) formed oligomers (Fig. (Fig.2D)2D) but did not bind fetuin and did not agglutinate RBC since it does not contain the RBD (Table (Table11).
These data revealed a hierarchy of requirements for biological functions of bacterial HA1, based not only on proper folding but also on the presence of stable higher-MW quaternary forms reminiscent of what has been documented for full-length HA molecule purified from influenza virus. These data confirmed that receptor binding requires a functional trimeric HA1, but oligomeric forms are required for RBC lattice formation resulting in hemagglutination. Moreover, the bacterially purified HA1 globular head with intact N terminus contained the trimers and oligomers required for these functions.
Alignment of the N-terminal amino acids of the HA protein from representative strains of 16 different influenza virus A hemagglutinin subtypes (see Fig. S5 in the supplemental material) identified amino acids I3C4I5G6 as highly conserved. Since the deletion of only four residues in the N terminus of HA1 (HA 5-320) was sufficient to prevent RBC agglutination (Fig. (Fig.1),1), we constructed two mutants of HA1 (I3C4I5 > A3A4A5, termed HA 1-330-AAA) and (I3C4I5 > G3A4G5, termed HA 1-330-GAG). These mutations did not affect protein folding, as determined by binding to huMAb FLA5.10 (see Fig. S6 in the supplemental material and CD melt; data not shown). However, both mutated proteins contained only monomers (Table (Table1).1). Most importantly, they did not bind to receptor in the fetuin-based SPR assay and did not agglutinate RBCs (Fig. (Fig.4A,4A, red and green curves, Fig. Fig.4B,4B, top two rows).
Furthermore, to investigate the role of cysteine at position 4, we constructed the HA1-Cys4Ala mutant and evaluated its structure and functional properties. In the context of the intact HA, Cys4 in HA1 normally forms a disulfide bond with Cys462 in HA2. In the absence of HA2, free Cys4 could potentially form disulfide bonds between HA1 monomers during the renaturation process. Gel filtration analysis of the purified HA1-C4A mutant protein (HA 1-330-C4A) showed that this protein consists of ~60% oligomeric HA1 forms and 40% monomers (Table (Table1).1). In the fetuin SPR binding assay, both HA1(1-330) and HA1(1-330;C4A) bound receptors (Fig. (Fig.4A,4A, blue and black curves), which correlated with the presence of oligomers (Table (Table1).1). No binding to asialo-fetuin was observed, confirming the binding specificity of these proteins to sialylated glycoproteins (data not shown). Importantly, the HA 1-330-C4A mutant protein agglutinated RBCs with comparable activity to that of the unmutated HA1(1-330) (Fig. (Fig.4B,4B, bottom row). These results suggested that the Cys4 in the N terminus of HA1 is not absolutely required for oligomerization and functional activity of the HA1 globular domain.
These data suggested that in the absence of HA2, the HA1 globular domain can use an oligomerization signal in the N terminus that encompasses the highly conserved amino acid residues at positions 3 to 5 of influenza virus HA.
We next compared the immunogenicity of bacterially expressed monomeric HA1(28-320) with that of the HA1(1-320) protein (~80% oligomers) in rabbits. Microneutralization assay was used to evaluate both the homologous and the heterologous neutralizing capacity of postvaccination rabbit sera after three to four consecutive immunizations (100 μg of protein per dose) (Fig. (Fig.5A).5A). After two immunizations, the monomeric HA1(28-320) elicited modest neutralizing antibody titers (1:80) against homologous virus (A/Vietnam/1203/2004, clade 1), which increased 4-fold by the fourth immunization. Cross-neutralization of A/Turkey/1/2005 (clade 2.2) and A/Anhui/1/2005 (clade 2.3.4), but not of A/Indonesia/5/2005 (clade 2.1), was also observed (Fig. (Fig.5A,5A, top panel). In contrast, rabbits immunized with oligomeric HA1(1-320) showed a faster kinetics of immune response and broader cross-clade neutralization. A neutralizing antibody titer of 1:160 against A/Vietnam/1203/2004 was measured after the second immunization and increased dramatically to 1:5,120 after the third vaccination (Fig. (Fig.5A).5A). Importantly, cross-clade neutralizing antibody titers were also very robust against heterologous HP H5N1 AIV, including A/Indonesia/5/2005 (clade 2.1), which is more difficult to cross-neutralize (Fig. (Fig.5A,5A, bottom panel).
In order to determine whether vaccination with oligomeric HA1 elicit antibodies that are oligomer specific, postvaccination sera from rabbit K1 [vaccinated with HA1(1-320)] and rabbit K3 [vaccinated with HA1(28-320)] were absorbed with the monomeric (28-320) or oligomeric (1-320) proteins, followed by binding to SPR sensor chips coated with the oligomeric fraction of HA1 (Fig. (Fig.5B)5B) or the monomeric fraction of HA1(1-320) (Fig. (Fig.5C).5C). Adsorption of either sera with HA1(1-320) removed SPR binding to the two proteins (Fig. 5B and C, gray and purple curves). On the other hand, K1 serum that was adsorbed with the monomeric HA1(28-320) still bound at a low level to the chip coated with the oligomeric HA1(1-320) protein (Fig. (Fig.5B,5B, green curve) but not to the chip coated with the monomeric protein (Fig. (Fig.5C).5C). These findings suggested the presence of oligomeric-specific antibodies in the sera of K1 rabbit, which were not adsorbed by the monomeric HA1(28-320) protein. The presence of trimer-specific anti-HA antibodies (seasonal) has been previously suggested (6).
To further evaluate the ability to generate protective immunity with bacterially expressed HA1 proteins, we used the ferret influenza virus challenge animal model, which is extremely susceptible to highly pathogenic H5N1 influenza virus infections. Since the pattern of influenza virus attachment to the lower respiratory tract resulting in influenza virus-associated pneumonia in ferrets resembles influenza virus infections in humans, this model has been widely used to evaluate influenza virus pathogenesis and vaccines (15, 26). Ferrets were vaccinated twice with 3 or 15 μg of either oligomeric HA1 globular protein (HA 1-320) or monomeric HA 28-320 with the N terminus deleted on days 0 and 21. The antigen doses were selected based on seasonal influenza virus vaccines and the need for dose sparing. At 14 days after the second immunization, unvaccinated and vaccinated animals were challenged intranasally with highly pathogenic H5N1 A/Vietnam/1203/2004 (clade 1, homologous to the vaccine stain) or with the H5N1 A/Whooperswan/Mongolia/244/2005 (clade 2.2) AIV at a predetermined lethal dose (106 EID50). The animals were monitored for 10 days for lethality, weight loss, and sickness scores.
The hemagglutination inhibition titers following two vaccinations with rHA1(1-320) ranged between 1:40 and 1:640 (average, 1:204) and between 1:10 and 1:320 (average, 1:141) for the 15- and 3-μg doses, respectively. rHA1(28-320) did not generate hemagglutination inhibition titers in any of the vaccinated ferrets. After intranasal challenge with HP avian viruses, A/Vietnam/1203/2004 and A/Whooperswan/244/2005, all unvaccinated ferrets developed severe symptoms, lost weight progressively, and died within 3 to 7 days of challenge (Fig. 6A to D, black open circles). HA1(28-320) with the N terminus deleted, which contains only monomers, did not protect animals from weight loss and lethality at the 3-μg dose (Fig. 6A and B, green squares), and only one animal survived homologous H5N1 A/Vietnam/1203/2004 challenge at the 15-μg vaccine dose (Fig. 6A and B, blue squares). In contrast, ferrets vaccinated with HA1(1-320) with either a 3-μg (red circles) or a 15-μg (black filled circles) dose were fully protected from lethality (Fig. (Fig.6B).6B). These animals showed only a minor transient weight loss on day 3 (≤10%), followed by a full recovery without any signs or symptoms by day 4 after homologous (A/Vietnam/1203/2004) virus challenge (Fig. (Fig.6A).6A). Importantly, HA1(1-320) immunization also protected ferrets against heterologous challenge with highly pathogenic clade 2.2 virus (H5N1 A/Whooperswan/244/2005), resulting in an 80% survival rate and <10% weight loss in both the high- and low-dose-vaccinated groups (Fig. 6C and D, filled red and black circles).
In addition to protection from mortality and morbidity, viral loads in the nasal washes of HA1(1-320)-vaccinated animals were reduced by 2 to 5 logs on days 3 and 5 postchallenge compared to unvaccinated animals or animals vaccinated with monomeric HA1 (Fig. 6E and F). Reduction in viral loads following heterologous challenge was more modest (1 to 2 logs) (Fig. 6G and H). Such reduction in viral loads in the nasal cavities is predicted to also reduce virus transmission.
Together, our data demonstrated that bacterially expressed HA1 proteins that are properly folded and contain functional oligomers can elicit protective immunity against both highly pathogenic vaccine-matched and heterologous avian influenza viruses.
Expression of recombinant HA proteins in bacteria could provide a rapid and economical approach for early response to impending influenza virus pandemic. Early studies demonstrated that protective influenza virus antigenic sites are conformation dependent and map primarily to the HA1 globular domain. Therefore, producing HA1 proteins in a properly folded state is imperative to eliciting protective antibody responses.
In the present study we dissected the structure-function requirements of bacterially expressed HA1 proteins and evaluated their potential use as prophylactic vaccines against HP H5N1 AIV. The main findings are as follows. (i) A panel of H5N1-A/Vietnam/1203/2004 HA1 proteins with N- and C-terminal deletions purified from E. coli under careful redox conditions were shown to be properly folded by binding to conformation-dependent huMAb. (ii) HA1 with an intact N terminus contained oligomers in addition to monomers, whereas HA1 with N-terminal deletions contained only monomers. (iii) A fetuin receptor-binding assay demonstrated that only HA1 proteins with intact N termini, containing oligomers, bound receptors. (iv) Hemagglutination required oligomeric HA1. (v) Site-directed mutagenesis of Ile-Cys-Ile residues at positions 3 to 5 disrupted oligomer formation, fetuin binding, and RBC agglutination, with no effect on HA1 folding. (vi) In rabbits, properly folded HA1 containing oligomers generated more rapid potent neutralizing antibodies than monomeric HA1 and cross-neutralized several H5N1 clades, including A/Indoensia/5/2005. (vii) Vaccination of ferrets with HA1(1-320) at either 3 or 15 μg of protein per dose protected animals from lethality and morbidity after challenge with homologous (A/Vietnam/1203/2004) or heterologous (A/Whooperswan/Mongolia/244/2005) HP AIV challenge. In contrast, monomeric HA1(28-320) was not immunogenic in ferrets at the same doses and did not protect animals from H5N1 challenge.
The structure of HA from highly pathogenic H5N1 A/Vietnam/1203/2004 resembles the 1918 and other human H1 HAs (21, 33, 34). Most of the intersubunit salt bridges and hydrophobic interactions are between the HA2 chains due to a coiled-coil structure which forms the stem of the HA trimer (4, 6-9, 30). These earlier HA-structural studies did not describe the oligomerization signal in the HA1 globular domain identified in the present study, suggesting that in the presence of HA2 the N-terminus β-sheet structure is engaged in an HA1-HA2 bridge and not in HA1 oligomerization. This may explain why most recombinant HA ectodomain proteins exist as monomers and require the addition of multimerization sequences such as “foldon” at the C terminus in order to produce stable oligomeric structures (3, 28, 29). This was further confirmed in a recent study in our laboratory with bacterially expressed HA proteins from the novel H1N1 A/California/04/2009 comparing the composition and immunogenicity of globular HA1(1-330) to that of the HA ectodomain (1-480). Both proteins were properly folded. However, only the HA1 globular head (1-330) formed oligomers and agglutinated human RBCs, whereas the HA ectodomain (1-480) contained only monomers and did not agglutinate RBC (13). It is likely that in the native spikes the N-terminal β-sheets of the three HA1 globular domains are not in sufficient proximity to form oligomers, but in the absence of HA2 they are free and close enough to provide the needed oligomerization signal. This was confirmed by our finding that an N-terminal fragment HA1(1-104) without the RBD appeared primarily as oligomers in gel filtration chromatography (Fig. (Fig.2D2D).
The oligomeric forms were not simply aggregates of monomers but contained multiples of functional trimers, as determined by equilibrium ultracentrifugation. In the native structure, the Cys4 in of HA1 is engaged in a disulfide bond with Cys462 in the HA2. Therefore, it was possible that during refolding this unpaired Cys4 form bonds with other Cys4 residues in adjacent HA1 molecules. To that end, we mutated Cys4 and found that the HA1 (Cys4Ala) mutant protein still contained a prominent oligomeric structure (60 to 75% of the total protein) that ran similar to the unmutated HA1 oligomers on gel filtration chromatography. More importantly, the Cys4Ala mutant bound well to receptor in the fetuin SPR and agglutinated RBCs. In contrast, triple mutants with Ile3-Cys4-Ile5 converted to either Ala-Ala-Ala or Gly-Gly-Gly contained only monomers and lost receptor binding and hemagglutination.
In mammalian and eukaryotic cells, posttranslational glycosylation of HA was shown to play an important role in proper folding, trimer stabilization, and transport to the cell outer membrane (5, 6, 18). On the other hand, we have demonstrated here and in previous studies that bacterially expressed unglycosylated HA1 (and HA0) can be purified as properly folded proteins, as determined by CD spectra analysis and binding to conformation-dependent neutralizing antibodies (12, 13), in agreement with the findings in another recent study (1).
Importantly, our present study demonstrated that, in addition to proper folding, HA1 oligomers were required for high-avidity receptor (fetuin) binding and for cross-linking of RBCs, resulting in hemagglutination. Importantly, the traditional inactivated subunit vaccine generated from egg-grown virus contains primarily oligomeric forms (Fig. (Fig.1).1). Previous reports on the production of recombinant HA in mammalian cells, insect cells, or bacterial systems did not provide information on the presence and function of oligomers versus monomeric forms of HA (1, 14, 17, 19, 20, 25, 27). More recent publications emphasized the importance of high-MW oligomers for optimal immunogenicity of influenza virus recombinant HA proteins. Trimerization domains such as GCN4-pII and foldon were added to the C termini of HA ectodomains produced in mammalian or insect cells in order to produce stable oligomeric structures and to elicit optimal neutralizing antibody titers (3, 28, 29). On the other hand, our study demonstrates oligomer formation with bacterially expressed HA1 with no requirement for the addition of any foreign trimerization sequence. Therefore, our bacterially expressed and properly folded HA1 proteins with intact N termini behave similarly to inactivated H5N1 subunit vaccine in terms of in vitro functions, including receptor binding and RBC agglutination.
The ferret protection data with highly pathogenic avian H5N1 studies provide strong evidence that bacterially expressed HA1 proteins, which are properly folded and contain functional oligomers, are potent inducers of protective immunity against pathogenic influenza viruses. Although all H5N1 viruses are between 95 and 98% identical regardless of clade, there is poor cross-reactivity between antibodies elicited to clade 1 HP H5N1 viruses, such as A/Vietnam/04, and clade 2 H5N1 viruses that predominate among recently transmitted strains, resulting in high human lethality. The cross-protection against heterologous strains is important since it is not certain which of the avian H5N1 influenza virus strains will adapt to human-to-human transmission. The level of protection achieved with the oligomeric forms of H5N1 HA1 bacterial proteins was equivalent to that observed after vaccination with inactivated H5N1 vaccine.
The combination of recombinant technology and improved purification approaches, combined with analytical assays to confirm proper folding and higher-order quaternary structures, will facilitate the large-scale production of HA in bacterial systems. Within 2 weeks of pandemic strain isolation, large quantities of HA1 proteins can be produced (currently 40 to 50 mg/liter in a batch culture; with 8- to 10-fold-higher yields in small-scale continuous fermentation cultures). Thus far, we have generated bacterially expressed properly folded HA1 from two H5N1 strains (A/Vietnam/1203/2004 [clade 1] and A/Indonesia/5/2005 [clade 2.1]), novel H1N1 (A/California/04/2009), H3N2 (A/Wisconsin/15/2009 and A/Victoria/210/2009), and H7N7 (A/Netherlands/219/03), and all were shown to form functional oligomers (≥70%), with lot-to-lot consistency.
Therefore, the production of HA1(1-320) proteins in bacterial systems is a viable and scalable approach for rapid vaccine production in response to emerging influenza virus strains with little or no preexisting immunity (such as H5N1 influenza virus), especially for individuals with known egg allergies.
This study was funded in part by an American Recovery and Reinvestment Act supplement to NIH/NIAID grant UO1-AI077771 to T.M.R. This study was also partly supported by IAA 224-10-1006 from DMID, NIH.
We thank Zhiping Ye and Vladimir Lugovtsev for their thorough review of the manuscript.
Published ahead of print on 17 November 2010.
†Supplemental material for this article may be found at http://jvi.asm.org/.