Conceived and designed the experiments: LS TP LT. Performed the experiments: VN UK AP JH JT AJ GL YH PD VT SU LR RB HL YZ GM. Analyzed the data: LS VN UK AP JH GL SU WW TP LT. Wrote the paper: LS LT.
Current address: Sanofi Pasteur, Swiftwater, Pennsylvania, United States of America
All the authors are employees of the VaxInnate Corporation.
It is known that physical linkage of TLR ligands and vaccine antigens significantly enhances the immunopotency of the linked antigens. We have used this approach to generate novel influenza vaccines that fuse the globular head domain of the protective hemagglutinin (HA) antigen with the potent TLR5 ligand, flagellin. These fusion proteins are efficiently expressed in standard
Influenza is one of the major infectious disease threats to the human population. It affects individuals of all ages, causes repeated infections throughout life, and is responsible for recurrent seasonal epidemics as well as periodic global pandemics of varying severity. Vaccines are central both to the effective control of seasonal outbreaks and to pandemic preparedness. Hemagglutinin (HA) has been the key protective antigen in seasonal influenza vaccines for the last forty years. While its structure and the basis of its efficacy are well understood, the genetic variability of HA coupled with current methods of vaccine production make it exceedingly difficult to simultaneously meet seasonal and pandemic needs on a global basis. HA changes antigenically to evade the immune response and on average, the prevalent influenza strains in circulation will acquire three to four amino acid changes per year in HA, mostly in regions of HA that are recognized by protective antibodies. Mutations accumulate over time and approximately every three to five years the virus evolves into an antigenically distinct strain
The current inter-related nature of seasonal and pandemic vaccine production has led to intense interest in the development of innovative technologies which could support both seasonal and pandemic influenza vaccine production. Improvements in influenza vaccine production by the industry have recently focused on cell culture. This approach alleviates the significant manufacturing issues associated with egg based manufacturing, but does not improve production efficiency. The intense focus on cell culture production stems from the historical view that protective forms of HA antigens must be manufactured using eukaryotic cells, like those of humans and chickens. The reason for this is that HA undergoes host cell dependent post-translational modification and even though the location and number of different glycosylation sites are not conserved among HAs, it is thought that glycosylation aids in correct folding of the molecule
In addition to improvements in vaccine production efficiency, enhancement of the immunopotency of influenza vaccines will be required in order to meet seasonal and pandemic needs on a global scale. It is now well established that physical linkage of Toll-like receptor (TLR) ligands and vaccine antigens enhances the immunopotency of the linked antigen. TLRs are expressed on various cell types, including professional antigen presenting cells (APC), where they act as primary sensors of microbial infection and then activate signaling pathways that lead to the induction of immune and inflammatory genes. TLR agonists are molecules such as lipoproteins, lipids, sugars or nucleic acids that are specifically associated with pathogenic organisms. Engagement of TLRs by their cognate agonists and the subsequent signaling within APC leads to enhanced processing and presentation of antigens that are co-delivered to those APC
We here present an approach that addresses many of the production and immunopotency barriers currently associated with seasonal and pandemic influenza vaccines. We have identified a single domain based on the globular head domain of HA which is a self-sufficient protective subunit that can be produced using prokaryotic expression systems. This globular head domain spans the majority of the neutralizing epitopes in HA and stably refolds to faithfully form these conformationally sensitive epitopes. We have genetically fused the globular head subunit to the TLR5 ligand flagellin to create an immunologically potent, highly protective vaccine that is very efficiently manufactured. The increased production efficiency associated with these vaccines means that they can be produced to meet national and even global needs in a period of several months with minimal investments in manufacturing infrastructure.
Structural studies have shown that two polypeptides, HA1 and HA2, form the monomeric subunit of the HA trimer. The HA1 polypeptide extends up from a membrane proximal stalk, spans the globular head domain and then returns to the stalk. Based on the architecture of HA1, we designed a subunit vaccine which encompassed the neutralizing epitopes of the globular head and also contained the structural elements necessary for spontaneous and efficient folding to correctly display these epitopes after recombinant protein expression in
A) Ribbon diagram of the trimeric PR8 HA0 ectodomain with a monomeric subunit of the HA trimer circled. B) Ribbon diagram of a monomer with the globular head circled. C) Ribbon diagram of the globular head with the boundaries of the HA1-1, HA1-2 and HA1-3 constructs indicated by crosses. Each construct is presented in detail to the right. The beginning and ending residue numbers, in PR8, for the three constructs are labeled. The important secondary structure elements, such as β-sandwich in HA1-1 and the closing β-sheet in HA1-2 are also marked.
We also recognized that inclusion of the small β-sandwich could be required to further stabilize the globular head domain subunit. Therefore, the second prototypic construct, designated as HA1-1, includes this additional secondary structure, and is similar to the thermolysin released fragment previously described by Bizebard
In order to evaluate the importance of secondary structural elements such as the second β-sheet and the small β-sandwich in the stabilization of the independent head domain and the consequent display of conformational epitopes, we designed a third PR8 construct, HA1-3, as a control. The boundary for HA1-3 (PR8) was placed at residues N101 to G276 to form a construct which is similar in design and size to an HA subunit previously reported by Jeon and Arnon
Each of the globular head constructs were recombinantly linked to the C terminus of
The STF2.HA1-1 (PR8), STF2.HA1-2 (PR8) and STF2.HA1-3 (PR8) fusion proteins were expressed using standard
STF2.HA1-1, STF2.HA1-2 and STF2.HA1-3 proteins were expressed and purified. Refolded proteins were analyzed by SDS-PAGE and Western blot analyses. A) Coomassie stained gel showing the proteins run in the presence (R) or absence (N) of reductant. Bands of the appropriate molecular weight were observed for each construct. B) Western blot analyses using the anti-flagellin monoclonal antibody, 6H11. C) Western blot using PR8-specific immune sera raised following a sub-clinical infection of mice with the PR8 virus.
To evaluate the conformational integrity of the HA subunit of the fusion proteins, Western blots of the SDS-PAGE gels were probed with either the flagellin-specific monoclonal antibody, 6H11, which recognizes a linear epitope (
To further evaluate the conformational integrity of the HA moiety of the STF2.HA1-1 (PR8), STF2.HA1-2 (PR8) and STF2.HA1-3 (PR8) fusion proteins, ELISA plates were coated with serial dilutions of the full length HA (ecto-domain, HA0, produced in Hi5 cells), PR8 virus and the fusion proteins. The coated plates were probed with either naïve or convalescent mouse sera (
ELISA plates were coated with the 0.2 µg/well of indicated STF2 fusion proteins, PR8 virus or the full length PR8 HA0 ectodomain expressed in Hi5 cells. Plates were probed with either naïve or PR8 convalescent sera at indicated dilution. Following incubation with HRP-conjugated goat anti-mouse IgG, plates were developed with UltraTMB substrate. Results reflect the delta value of OD450 (Convalescence-Naïve) of samples performed in duplicate. Naïve values (data not shown) were below 0.02.
An
Taken together, the results confirm the importance of secondary structure in the appropriate refolding of the HA globular head domain. While both the STF2.HA1-1 and STF2.HA1-2 protein fold properly, the STF2.HA1-2 recombinant protein folds more efficiently under the conditions tested. The globular head component of STF2.HA1-3 fails to fold efficiently.
The conformational integrity of defined antigenic regions of the PR8 globular head was tested using a panel of monoclonal antibodies known to be specific for neutralizing epitopes located in the globular head
A) Ribbon diagram depicting of the known antigenic regions of the HA globular head. B–E) ELISA plates were coated with either PR8 virus or the STF2 fusion proteins and probed with a mAb specific for the Sb, Sa, Ca1, Ca2 and Cb region. All STF2 fusion proteins are produced from E coli. F–I) ELISA plates were coated with STF2.HA1-2 and competed against soluble form of HA1-1, HA1-2, STF2.HA1-2, STF2.HA1-3 in binding of the panel of monoclonal antibodies (H37-64, 18 ng/ml; H37-77, 8 ng/ml; H36-11, 188 ng/ml and H163, 500 ng/ml). Bound antibodies were detected by 450 nm absorption. HA1-1 and STF2.HA1-1 were produced in insect cell culture. Both proteins have C-terminal 6His tag. HA1-2, STF2.HA1-2 and STF2.HA1-3 were produced in E. coli. HA1-2 has 6His tag at C-terminus.
The structural integrity of these neutralizing epitopes was further examined using competition assays. Since the panel of monoclonal antibodies reacted equally well with plate-bound STF2.HA1-2 (PR8) and virus particles, the ELISA plates were coated with the STF2.HA1-2 (PR8) fusion protein. The panel of monoclonal antibodies was incubated with the HA1-2 globular head protein alone, STF2.HA1-2 or STF2.HA1-3 for 2 hours. HA1-1 globular head alone or the STF2.HA1-1 protein produced in baculovirus (HA1-1 bv) were included in the evaluation. The mixture was transferred to the ELISA plates and the specific monoclonal antibody reactivity was measured following washing and blocking of the plates. The results (
Groups of 10 BALB/c mice were immunized on days 0 and 14 with 3, 0.3 and 0.03 µg of STF2.HA1-2. A group of naïve mice was included as a negative control. On day 10, animals were bled and the sera of individual animals examined for HA-specific IgG by probing ELISA plates coated with the PR8 virus (
BALB/c mice (10/group) were immunized on day 0 and 14 with 3, 0.3 or 0.03 µg of STF2.HA1-2. A group receiving formulation buffer alone was included as a negative control. A) Sera were harvested on day 21 and evaluated for HA-specific antibody responses by ELISA. B–C) On day 28, mice were challenged
The same principles of design were applied to the HA of the currently circulating seasonal strain A/Solomon Islands/3/2006 (SI). Protein was expressed and purified using the methodologies as for the prototypic PR8 construct. To evaluate the immunogenicity and efficacy of this construct, groups of 10 BALB/c mice were immunized on day 0 and day 14 with 3 or 0.3 µg of STF2.HA1-2 (SI), and bled on day 21. Sera were analyzed for hemagglutination inhibition of SI virus using chicken red blood cells as the target. Geometric mean titers (GMT) were 1∶320 (range 1∶160–1∶640) at the 3 µg dose level and 1∶226 (range 1∶80–1∶640) at the 0.3 µg dose level. Ferret immune sera, raised on natural infection and obtained from CDC, exhibited an HAI titer of 1∶320. To further characterize the potency of the vaccine, 15 BALB/c mice were immunized twice with 10, 1 or 0.1 µg of STF2.HA1-2 (SI). Sera were harvested 1 week post the boosting immunization and evaluated for microneutralization titers. In instances where a mouse adapted strain of the virus is not available, as is the case for this seasonal strain, neutralization titers of ≥1∶40 are generally accepted as a correlate of efficacy. The results are reported as GMT in
Groups [N] | Dose | GMT | 95% CI |
STF2.HA1-2 (SI) | 10 | 845 | 472–1511 |
STF2.HA1-2 (SI) | 1 | 297 | 106–833 |
STF2.HA1-2 (SI) | 0.1 | 8 | 4–20 |
Naive | 5 | 5–5 | |
STF2.HA1-2 (SI) | 15 | 640 | 404–1014 |
STF2.HA1-2 (SI) | 5 | 453 | 211–971 |
Naïve | 40 | 40–40 |
µg/animal;
geometric mean titers;
95% confidence intervals;
,
To further evaluate the immunopotency of STF2.HA1-2 (SI), groups of 6 New Zealand White rabbits were immunized twice,
We have identified the structural elements necessary for efficient refolding of a protective HA subunit after recombinant protein expression in
We find that of the three prototypic constructs evaluated, STF2.HA1-2 provides the level of expression and ease of conformationally correct refolding required to support a truly efficient, scalable manufacturing process. This molecule expressed well in our prokaryotic system and refolded easily using rapid dilution. STF2.HA1-1 also refolded albeit with somewhat lower efficiency than STF2.HA1-2, presumably as a consequence of the additional domain and two additional disulfide bonds. Western blots of STF2.HA1-1 run under reducing and non-reducing conditions and probed with PR8 convalescent sera reveal a band in the non-reduced sample that co-migrates with the main band in the reduced sample suggesting that a significant proportion of the protein remained misfolded. STF2.HA1-3 refolded the least efficiently, most likely due to the absence of secondary structures required for stable refolding. Reactivity with a panel of defined neutralizing monoclonal antibodies further supports this conclusion.
STF2.HA1-2 (PR8) was found to be highly immunogenic and efficacious against a lethal challenge in the mouse model. Mice receiving doses of STF2.HA1-2 (PR8) as low as 0.3 µg were protected against a lethal challenge of virus. These data demonstrate that a subunit of HA based on the globular head domain can be fully protective in a standard mouse lethal challenge model. When the same principles of design were applied to the currently circulating seasonal strain, A/Solomon Islands/3/2006, we found that STF2.HA1-2 (SI) was highly immunogenic in both mice and rabbits. In mice doses of 1 µg elicited geometric mean neutralizing titers of 1∶297 and in rabbits doses of 5 µg elicited titers of 1∶453. Thus, the principles of design for these protective subunit vaccines can be applied to different HA molecules.
A key benefit with this approach is that the STF2.HA1-2 recombinant fusion protein can be made quickly, inexpensively and in quantities sufficient to meet global needs. The efficiency of this technology translates approximately into a 1,000 fold gain in production. As a point of reference, the average yield for cell culture is 3 mg/L; for egg based production, 7 mg/L; for baculovirus recombinant synthetic protein, 13 mg/L and for the standard prokaryotic system described here, 3,700 mg/L. This increase in production capacity, along with the fact that it is carried out in a prokaryotic system, provides an opportunity to address several shortcomings of the current egg-based system. One advantage deriving from increased capacity is the ability to increase the dose of antigen. Studies have shown that persons greater than 65 years of age respond less well to the standard vaccine, and that increasing the dose of HA four to five-fold substantially improves the immune response in this segment of the population. Formulation of a “high-dose” vaccine for the elderly becomes a practical possibility with an unconstrained supply of antigen. A second set of advantages comes from eliminating the growth of virus from the manufacturing process. Currently, vaccine production strains are created by crossing the HA and NA genes from candidate circulating strains onto an egg-adapted virus, generally the PR/8/34 strain. Manufacturers then further adapt these production strains to create high-yield viruses. The adaptation process results in selection of mutations in the upper part of the HA globular head near the receptor binding site
In conclusion, the manufacturing approach described herein has major advantages over existing technologies in that it allows faster molecular development, rapid manufacturing and very high levels of productivity at small manufacturing scales. These advantages are critical to the successful production of seasonal and pandemic influenza vaccines.
The codon optimized synthetic genes of the hemagglutinin (HA) globular head domain of PR8 were fused to the C-terminus of the full-length sequence of
The synthetic genes encoding HA1-1 (PR8) or in fusion with STF2 were codon-optimized for Baculovirus expression (DNA2.0 Inc., Menlo Park, CA) and cloned into pFastBac™. The honey bee mellitin sequence (MKFLVNVALVFMVVYISYIYAD PS) was fused to the amino terminus of recombinant proteins to provide a secretion signal and hexahistidine was tagged to the carboxyl terminus to facilitate purification. The synthetic genes were cloned to the pFastbac1 vector. The recombinant Baculovirus generation followed standard Bac-to-Bac® Baculovirus Expression protocol (Invitrogen, Carlsbard, CA).
High expresser clones were cultured overnight and used to inoculate fresh LB medium supplemented with 25 µg/ml kanamycin, 12.5 µg/ml tetracycline and 0.5% glucose. At an OD600 = 0.6 protein expression was induced with 1 mM IPTG for 3 h at 37°C. Cells were harvested by centrifugation (8,000g for 7 minutes) and disrupted by microfluidizer (18,000 psi). The inclusion body was washed with 1% Triton X100 and dissolved in 8 M urea. The filtered protein solution in 25 mM NaCl and 50 mM Acetate, pH 4.0 was applied to a SP Sepharose Fast Flow column (GE/Amersham). The fraction peak was eluted by salt gradient and buffer exchanged to 50 mM Tris, 25 mM NaCl and 8 M urea, pH 8.0. Protein refolding was achieved by rapid dilution (1∶10) into 100 mM Tris-HCl buffer (pH 8.0), and further purified by anion exchange (Source Q, GE/Amersham). For final polishing and endotoxin removal, a Superdex 200 gel filtration column (10/300 GL, GE/Amersham) was used. The protein peak was eluted using 100 mM Tris, 150 mM NaCl, 1% glycerol and 1% Na-deoxycholate elution buffer. Peak fractions were pooled, dialyzed against 1×PBS and stored at −80°C. For all 6xHis tagged proteins, the metal chelating column was employed. Protein was loaded to a Ni-NTA column equilibrated in 20 mM Tris, pH 8, 0.5 M NaCl and eluted in a gradient of 0–0.5 M imidazole. The target protein was further purified by size exclusion column (10/300 GL, GE/Amersham). The peak fractions were pooled, concentrated and dialyzed against 1×PBS. Aliqoted protein solution was stored at −80°C. Endotoxin contamination was assayed by using standard Chromogenic Limulus Amebocyte Lysate assay (Cambrex, Walkersville, MD) as directed by the manufacturer.
Many aspects of the ELISA methods were held in common. ELISA plates were coated with the indicated proteins in PBS overnight at 4°C or one hour at room temperature. All washes between reagent addition steps were performed 3 times with 1X PBS/0.05% Tween-20. Plates were blocked with 200–300 µl/well of Assay Diluent Buffer (ADB; BD Pharmingen) for 1–3 hour at 23–27°C. After incubation with the indicated detection antibodies, HRP-labeled goat anti-mouse antibody (Jackson Immunochemical) diluted in ADB was added and the plates were incubated at 23–27°C for 1 hour. After adding TMB Ultra substrate (Pierce) and monitoring color development, the reaction was stopped with 1 M H2SO4 and OD450 was measured on a microplate spectrophotometer.
Plates were coated with serial dilutions of proteins. After block, plates were probed with monoclonal antibody specific for flagellin (6H11; Inotek) or convalescent sera against PR8 virus overnight at 4°C.
Plates were coated with 100 µl/well HA1-1 produced in insect cells in PBS (5 µg/ml). Dilutions of the sera in ADB were added (100 µl/well) and the plates were incubated overnight at 4°C.
Sucrose density gradient purified PR8 virus (Advanced Biotechnologies Inc.,) or STF2 tagged recombinant HA proteins were diluted to 4 µg/ml in 1X PBS and 100 µl coated in triplicates. After block, plates were incubated with 100 µl of HA-specific antibodies diluted in ADB at 25°C for 2.5 hours.
Plates were coated overnight with 100 µl/well STF2.HA1-2 (PR8) at 2 µg/ml. Antibodies were pre-incubated with serially diluted recombinant proteins for 2 hours at 25°C and added to the washed and blocked STF2.HA1-2 (PR8) coated plates for a further 2 hour incubation followed by detection antibody. The amount of antibody used for each epitope was pre-determined to be in the linear range of a saturation curve in ELISA.
MDCK cells were obtained from ATCC and maintained in DMEM supplemented with 5% FBS, 2 mM L-glutamine, 100 units/ml Penicillin and 100 µg/ml Streptomycin. Influenza viruses, mouse adapted A/PR/8/34 and A/Solomon Islands/3/2006, were obtained from Dr. Y. Kawaoka (University of Wisconsin) and CDC, respectively, and propagated in either MDCK cells or 11-day old SPF embryonated hen's eggs (Charles River Laboratories, North Franklin, CT).
The bioactivity of purified recombinant proteins was tested as previously described
BALB/c mice 6–8 weeks old were purchased from the Jackson Laboratory (Bar Harbor, ME) and housed in either the Yale University vivarium (New Haven, CT) or the Princeton University vivarium (Princeton, NJ). All studies were performed in accordance with the University Institutional Animal Care and Use Committees (IACUC). Recombinant proteins were prepared in one of two vehicles: PBS (phosphate-buffered saline) or formula F147 (10 mM L-histidine, 150 mM NaCl, 5% trehalose, 0.02% polysorbate 80, 0.1 mM EDTA, 0.5% ethanol, 10 mM Tris, pH 7.2). Vehicles were used interchangeably without detectable impact on the results. Mice were immunized subcutaneously (
Studies with female and male New Zealand White rabbits were performed at Covance Research Products (Denver, PA). Rabbits (6/group) were immunized intramuscularly (
Serum samples were treated with receptor destroying enzyme II (RDE, Denka Seiken Co., Ltd., Tokyo, Japan) and co-cultivated with 100 TCID50 of influenza virus A/PR/8/34 or A/Solomon Islands/3/2006 for 1.5 hr in series dilution (duplicate). MDCK cells (4×104 /well) in DMEM supplemented with 1% BSA, 20 mM HEPES, and 100 IU/ml Penicillin and 100 µg/ml Streptomycin were then added and incubated for 20 hours at 37°C. Cells were washed, fixed, air-dried and incubated with a monoclonal anti-influenza A nucleoprotein antibody (1∶2,000, clones A1 and A3, ATCC/BEI resources). Signals were detected by OD450. Virus back titration, positive serum control, virus controls (VC), and cell controls (CC) were included in the assay. The end point of virus neutralizing antibody for each serum was determined using 50% of specific signal = [(Average OD of VC wells)–(Average OD of CC wells)]/2+Average OD of CC wells. Values below this value are considered positive for neutralizing activity.
The titers of neutralizing antibodies were transformed into natural logarithm, and subjected to ANOVA/Tukey tests. Survival curves between different groups were compared with Log-rank test. Data analysis used GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego California USA,
To assess efficacy, mice immunized on days 0 and 14 as described above were challenged on day 35 by intranasal administration of 1×LD90 (dose lethal to 90% of mice; 1×103 TCID50) of influenza A isolate, PR8. Animals were monitored daily for 21 days following the challenge for survival and weight loss.
We thank Dr. Walter Gerhard (The Wistar Institute) for assisting with the selection of PR8 specific monoclonal antibodies; Dr. Yoshihiro Kowaoka (University of Wisconsin) and Dr. Lynda Cauley (University of Connecticut) for providing PR8 virus. We thank Bruce Weaver, Dallas Jock and Lorine Lantz for technical assistance and Dr. Alan Shaw, Dr. Robert Becker, Dr. Richard Flavell and Dr. Ruslan Medzhitov for helpful discussion and reading this manuscript.