Hiav

SJa

b

a

ARR1AA

KIVHHAAIFBpPHH

1

sp

a

a

H

h0

Vaccine 32 (2014) 6421–6432

Contents lists available at ScienceDirect

Vaccine

j our na l ho me page: www.elsev ier .com/ locate /vacc ine

5 N-terminal � sheet promotes oligomerization of H7-HA1 thatnduces better antibody affinity maturation and enhanced protectiongainst H7N7 and H7N9 viruses compared to inactivated influenzaaccine

urender Khuranaa,∗, Elizabeth M. Coylea, Swati Vermaa, Lisa R. Kinga,ody Manischewitza, Corey J. Crevarb, Donald M. Carterb, Ted M. Rossb, Hana Goldinga,∗∗

Division of Viral Products, Center for Biologics Evaluation and Research (CBER), Food and Drug Administration, Silver Spring, MD 20903, USAVaccine and Gene Therapy Institute of Florida, Miami, USA

r t i c l e i n f o

rticle history:eceived 17 June 2014eceived in revised form2 September 2014ccepted 22 September 2014vailable online 3 October 2014

eywords:nfluenzaaccineemagglutininA1ntibodyffinity

mmune Responseerrets

a b s t r a c t

Initiation of mass vaccination is critical in response to influenza pandemic. There is an urgent need of asimple, rapid method for production of influenza vaccine that is more effective than current traditionalinfluenza vaccines. Recent H7N9 transmissions to humans in China with high morbidity/mortality initi-ated extensive vaccine evaluation. We produced the HA1 domains (amino acids 1-320) from H7N9 andH7N7 strains in E. coli. Both were found to contain primarily monomers/trimers with low oligomericcontent. However, when residues from the N-terminal � sheet (first 8 amino acid) of H7 HA1 domainswere swapped with the corresponding amino acids from H5N1, functional oligomeric H7 HA1 were pro-duced (HA1-DS), demonstrating strong receptor binding and hemagglutination. In rabbits, the HA1-DSfrom either H7N9 or H7N7 generated high neutralization titers against both homologous and heterol-ogous H7 strains, superior to the unmodified H7 HA1 proteins. In ferrets, HA1-DS from H7N7 elicitedhigher (and faster) HI titers, better protected ferrets from lethality, weight loss, and reduced viral loadsfollowing challenge with wild-type highly pathogenic H7N7 virus compared with inactivated H7N7 sub-unit vaccine. HA1-DS vaccinated ferrets were also better protected from weight loss after challenge withthe heterologous H7N9 virus compared with inactivated H7N7 subunit vaccine. Importantly, the H7N7

acteriaroteinandemic7N77N9

HA1-DS vaccine induced antibody affinity maturation far superior to the inactivated H7N7 subunit vac-cine, which strongly correlated with control of viral loads in the nasal washes after challenge with eitherH7N7 or H7N9 strains. We conclude that N-terminus � sheet domain-swap can be used to produce stablefunctional oligomeric forms of better recombinant HA1 vaccines in simple, inexpensive bacterial systemfor rapid response to emerging pandemic threat for the global population.

Published by Elsevier Ltd.

. Introduction

Influenza pandemics are likely to occur following transmis-ion of new zoonotic influenza strains to domestic poultry origs with occasional adaptation to the human population. In such

∗ Corresponding author. Division of Viral Products, Center for Biologics Evaluationnd Research (CBER), FDA, Silver Spring, MD, USA 20903. Tel.: +240 402 9632.∗∗ Corresponding author. Division of Viral Products, Center for Biologics Evaluationnd Research (CBER), FDA, Silver Spring, MD, USA 20903. Tel.: +240 402 9561.

E-mail addresses: Surender.Khurana@fda.hhs.gov (S. Khurana),ana.Golding@fda.hhs.gov (H. Golding).

ttp://dx.doi.org/10.1016/j.vaccine.2014.09.049264-410X/Published by Elsevier Ltd.

scenarios, rapid production of vaccines and their release for massvaccination is of primary importance. Alternative approaches ofgenerating influenza vaccines including recombinant technologiesand/or potency reagents for use in formulation and release of vac-cine lots could shorten the time to vaccine availability, which isa high priority for public health authorities around the world torespond to the threat of an influenza pandemic.

The process of constructing a new vaccine strain based on newlycirculating viruses is quite lengthy. It involves in vivo (in chicken

eggs) or in vitro (in cell culture using reverse genetics techniques)reassortment between the internal genes of a donor virus such asA/PR/8/34 with the hemagglutinin (HA) and neuraminidase (NA)of the new influenza strain. The candidate vaccine strains must be

6 cine 3

fhoiipp

gsPpintrttsaorItdom

v2ita

wrotpfnooaNrsiHta

tbvcdbv

2

2

2

422 S. Khurana et al. / Vac

urther selected based on their high growth capability in eggs andigh yield of HA content before they can be used for productionf vaccines. This process is used for the production of seasonalnfluenza vaccines every year, but it may pose a clear imped-ment to initiation of rapid mass vaccination against spreadingandemic influenza, as was evident for the 2009 H1N1 influenzaandemic.

Alternative approaches are needed to make HA immunogens foreneration of recombinant HA based vaccines in the shortest pos-ible time that could save several months of manufacturing time.roduction of recombinant HA proteins in bacterial systems couldrovide a rapid and economical approach for early response to

mpending influenza pandemic. The main challenge to the recombi-ant technology is to ensure that the HA vaccine products resemblehe native virion-associated trimeric spike proteins and can elicitobust immune responses targeting protective conformational epi-opes in the globular domain of HA. Recently, we demonstratedhat bacterially produced influenza HA1 globular domain fromeveral pandemic strains from Group 1 (including H1N1pdm09nd two H5N1 viruses) are properly folded and form functionalligomers that can agglutinate red blood cells (RBC). The oligome-ization signal was mapped to the N-terminal �-sheet of HA1 [1–4].mportantly, the recombinant oligomeric HA1 proteins elicited highiters of neutralizing antibodies in both rabbits and sheep thatemonstrated strong protection of animals against both homol-gous and heterologous H5N1 avian strains in a ferret challengeodel [2–4].Since February 2013, the newly emerged H7N9 avian influenza

irus (AIV) in China has caused human infections, and as of May014, more than 445 infections with case fatality rate of 30%, and

s posing a new pandemic threat [5] [6,7] [8]. Several H7N9 vaccinerials are underway. Both traditional and new vaccine approachesre being evaluated [9] [10].

Therefore, we produced the HA1 globular domain of H7N9 asell as H7N7 that belong to group-2 influenza strains in bacte-

ial expression system and evaluated their functionality in termsf oligomeric structure, receptor binding, and red blood cell agglu-ination (RBC). Unexpectedly, the recombinant H7 HA1 proteinsrimarily formed monomers and trimers with low oligomericractions. Trimer/oligomer formation is an important property ofative HA spike proteins required for cell attachment and forptimal immunogenicity. Comparison of the N-terminus sequencef H7N9 and H7N7 HA with that of H5N1 HA identified threemino acid differences within the first eight amino acids in the-terminus �-sheet that may have led to the reduced oligome-

ization of the H7 HA1 domains. Exchanging the N-terminal �heet (first 8 residues) of H7 HA1 with the corresponding sequencen H5N1 resulted in higher oligomerization of domain-swapped7 HA1 proteins that were functionally superior compared with

he wild type H7 HA1 in terms of receptor binding and RBCgglutination.

The current study demonstrates that the oligomeric H7 HA1 pro-eins promoted generation of higher neutralizing antibodies andetter protection against both homologous and heterologous H7irus strains compared to traditional inactivated H7N7 subunit vac-ine in ferrets. Furthermore, the improved virus-neutralizing titersirectly correlated with significant affinity maturation of the anti-odies elicited by the H7 HA1 protein compared with H7N7 subunitaccine.

. Materials and Methods

.1. Expression vector and cloning of H7N9-HA1 derivatives

The complete HA0 gene segment of H7N7-A/Netherlands/19/03 and H7N9-A/Shanghai/1/2013 was generated by chemical

2 (2014) 6421–6432

synthesis, and was used for cloning. pSK is a T7 promoterbased expression vector where the desired polypeptide can beexpressed as a fusion protein with His6 tag at the C-terminus.DNA encoding HA1 (1-320) of the H7N7-A/Netherlands/219/03 orH7N9-A/Shanghai/1/2013 and its various derivatives were clonedas NotI-PacI inserts in the pSK expression vector. A shorter HA1from H7N9 (amino acids 28-320) was also expressed and purifiedfrom E. coli. (Fig. 1A, B)

2.2. Protein expression, refolding and purification

E. coli BL21 cells (Novagen) were used for expression ofH7N7-A/Netherlands/219/03 and H7N9-A/Shanghai/1/2013 HA1derivatives. Following expression, inclusion bodies were isolatedby cell lysis and multiple washing steps with 1% Triton X-100. FinalInclusion Bodies (IBs) pellets were resuspended in denaturationbuffer containing 6 M Guanidine Hydrochloride and dithioerythre-itol (DTE) at final protein concentration of 10 mg/ml and werecentrifuged to remove residual debris. For refolding, supernatantswere slowly diluted 100-folds in redox folding buffer. The refoldedproteins were dialyzed against 20 mM Tris HCl pH 8.0 to remove thedenaturing agents. The dialysates were filtered through a 0.45 �Mfilter and were subjected to purification by HisTrap Fast flow chro-matography.

2.3. Gel filtration Chromatography

Proteins at a concentration of 5 mg/ml were analyzed onSuperdex S200 XK 16/60 column (GE-Healthcare) pre-equilibratedwith PBS, and the protein elutions were monitored at 280 nm.Protein molecular weight marker standards (GE healthcare)were used for column calibration and generation of standardcurves, to identify the molecular weights of the test proteinsamples.

2.4. Hemagglutination Assay

Human erythrocytes were separated from whole blood (Lam-pire Biologicals). After isolation and washing, 30 �l of 1% humanRBC suspension (vol/vol in 1% BSA-PBS) were added to 30 �l serialdilutions of purified recombinant HA1 proteins (rHA1) or influenzavirus in 1% BSA-PBS in a U-bottom 96-well plate (total volume,60 �l). Agglutination was read after incubation for 30 min at roomtemperature

2.5. Receptor binding assay using surface plasmon resonance

Biotinylated �-2,3 sialic acid glycan were provided bythe Consortium for Functional Glycomics (http://www.functionalglycomics.org). The biotinylated receptor homologwas coupled to a NLC sensor chip at 400 resonance units (RU)in the test flow cells. Binding of different rHA1 derivatives wasanalyzed at 25 ◦C using a ProteOn surface plasmon resonancebiosensor (BioRad Labs). Samples of 60 �l freshly prepared H7N7-HA1 and H7N9-HA1 proteins at 10 �g/ml were injected at a flowrate of 30 �l/min (120-sec contact time). The Flow was directedover a mock surface to which no glycan was bound, followed by the�-2,3 sialic acid glycan- bound surface. Responses from the glycansurface were corrected for the response from the mock surfaceand for responses from a separate, buffer only, injection. Bindingkinetics and data analysis were performed with BioRad ProteON

manager software (version 2.0.1). Binding to both �-2,6 and �-2,3sialic acid glycans was evaluated for H7 derived from H7N9 andH7N7. Similar to H5N1, H7 HA1 proteins bound primarily to �-2,3sialic acid glycans.

S. Khurana et al. / Vaccine 32 (2014) 6421–6432 6423

Figure 1. Characterization of purified H7N9, H7N7 and H5N1 HA1 proteins from E. coli by reducing gel electrophoresis and gel filtration chromatography.(A) Hemagglutinin globular head from H7N9 A/Shanghai/01/2013 (H7N9) and A/Netherlands/219/03 (H7N7) HA1 domain (aa 1-320 wt) or truncated HA1 (aa 28-320) wereexpressed in E. coli as fusion proteins with His6 tag at the C-termini. (B) The purified proteins ran as single bands at the expected molecular weights in reducing SDS-PAGE.(C-F) Superdex S-200 gel filtration chromatography of bacterially produced H7N9, H7N7 and H5N1 HA1 proteins. Purified H7N9 HA1 protein with N-terminus deletion(28-320) (C), HA1 with intact N-terminus (1-320) (D), H7N7 A/Netherlands/219/03 protein with intact N-terminus (HA1 1-320) (E), and H5N1 A/Vietnam/1203/2004 HA1protein with intact N-terminus (1-320) from the A/Vietnam/1203/2004 strain (F) were subjected to gel filtration. The panels present superimposed elution profiles of purifiedH olum

2

i

A proteins (blue line) overlaid with calibration standards (grey line). The elution v

.6. Microneutralization assay

Viral-neutralizing activity was analyzed in a microneutral-zation assay based on the methods of the pandemic influenza

es of protein species are shown.

reference laboratories of the Center for Disease Control and Pre-vention (CDC). Low pathogenicity H7N9 (A/Shanghai/1/2013) andH7N7 (HA derived from H7N7-A/Netherlands/219/03) viruses, gen-erated by reverse genetics, were obtained from CDC and the lab

6 cine 3

ocf

2

tHMtm

2

2

saFm2w7nsaigwlglcaTspGWIcA

2

pmuwfw(Aoc0cw(tbarspn

424 S. Khurana et al. / Vac

f Maryna Eichelberger, CBER, respectively. The experiments wereonducted with three replicates for each serum sample and per-ormed at least twice.

.7. Rabbit immunization

Female New Zealand rabbits (3 experiments) were immunizedwice intra-muscularly at 21-days interval with 50 �g of purifiedA1 proteins and its derivatives with Titermax adjuvant (Titer-ax Inc, Norcross, Georgia, US). Sera were collected 8 days after

he first and second vaccination, respectively, and analyzed in aicroneutralization assay against H7N9 and H7N7 virus strains.

.8. Ferret immunization and Challenge studies

.8.1. Vaccination of ferrets and blood collectionAll ferrets (Triple F Farms) used in the study were confirmed as

eronegative for circulating seasonal influenza A (H1N1 and H3N2)nd influenza B viruses by hemagglutination inhibition assay (HI).emale Fitch ferrets (n = 6 in each group) were vaccinated intra-uscularly in the quadriceps muscle on day 0, boosted on day

1, and challenged with virus on day 35. Control animals (n = 6)ere mock vaccinated with phosphate buffered saline (PBS; pH

.2). Each animal was vaccinated twice with 15 �g of recombi-ant H7 HA1-DS in sterile 0.9% saline or with H7N7 inactivatedubunit vaccine (Sanofi Pasteur) containing 15 �g of HA antigens indicated on the vaccine vial (and confirmed using SRID). Bothmmunogens were mixed with the adjuvant formulation, Emulsi-en at a 1:1 ratio. In addition, H7 HA1-DS protein was administeredithout adjuvant as a comparator. The volume for all intramuscu-

ar vaccinations was 0.5 ml. The first and second vaccinations wereiven in the left and right hind legs, respectively. Blood was col-ected from anesthetized ferrets via the anterior vena cava. Theollected blood was transferred to a tube containing a serum sep-rator and clot activator and allowed to clot at room temperature.ubes were centrifuged at 6000 rpm for 10 minutes; serum waseparated, aliquoted, and stored at −80 ± 5 ◦C. All procedures wereerformed in accordance with the National Research Council (NRC)uidelines for the Care and Use of Laboratory Animals, the Animalelfare Act, and the Centers for Disease Control (CDC)/National

nstitutes of Health (NIH) Bio-Safety Guidelines in Microbiologi-al and Biomedical Laboratories and approved by the Institutionalnimal Care and Use Committee (IACUC).

.8.2. Infection and monitoring of ferretsAnimal experiments with H7N7 and H7N9 influenza virus were

erformed in the AALAC-accredited ABSL-3 enhanced facility. Ani-als were infected and monitored as previously described, except

sing 5% isofluorane anesthesia. Briefly, ferrets were anesthetizedith isofluorane and infected intranasally with 1 × 107 50% plaque

orming units (pfu), the highest dose of H7N9 (A/Shanghai/1/2013),hich is equivalent to 1000 LD50 (107 pfu) of wild type H7N7

A/Netherlands/219/03) viruses in a volume of one milliliter.nimals were daily monitored for temperature, weight loss, lossf activity, nasal discharge, sneezing and diarrhea following viralhallenge. To determine viral load in nasal washes, 1.5 ml of.9% saline was administered to each nare and the washes wereollected each day post-challenge from each ferret. Temperaturesere measured using an implantable temperature transponder

BMDS, Sayre, PA) and were recorded at approximately the sameime each day. Pre-infection values were averaged to obtain aaseline temperature for each ferret. Clinical signs of sneezingnd nasal discharge, inappetence, dyspnea, neurological signs,

espiratory distress, and level of activity were assessed daily. Acoring system was used to assess activity level where 0 = alert andlayful; 1 = alert but playful only when stimulated; 2 = alert butot playful when stimulated; 3 = neither alert nor playful when

2 (2014) 6421–6432

stimulated. Based on the daily scores for each animal in a group, arelative inactivity index was calculated.

2.8.3. Determination of viral loadsViral loads in nasal washes were determined by the plaque

assay. Briefly, MDCK cells plated in 6-well tissue culture plateswere inoculated with 0.1 ml of virus-containing sample, seriallydiluted in Dulbecco’s modified Eagle’s medium (DMEM). Virus wasadsorbed to cells for 1 h, with shaking every 15 min. Wells wereoverlaid with 1.6% w/v Bacto agar (DIFCO, BD Diagnostic Systems,Palo Alto, CA, USA) mixed 1:1 with L-15 media (Cambrex, EastRutherford, NJ, USA) containing antibiotics and 0.6 mg/ml trypsin(Sigma, St. Louis, MO, USA). Plates were incubated for 5 days. Cellswere fixed for 10 minutes using 70% v/v Ethanol and then overlaidwith 1% w/v crystal violet. Cells were then washed with deionizedwater to visualize plaques. Plaques were counted and compared touninfected cells.

2.8.4. Hemagglutination Inhibition (HI) assayRDE-treated ferret sera were serially diluted in v-bottom

96-well microtiter plates followed by the addition of 4 hemagglu-tination units (HAU) of influenza virus. Following incubation forapproximately 30 minutes, 1% suspension of human RBC in PBS(pH 7.2) was added and mixed by agitation. The human RBCs wereallowed to settle for 30 minutes at room temperature and HI titerswere determined by the reciprocal value of the last dilution ofsera that completely inhibited hemagglutination of human RBCs.A negative titer was defined as 1:10.

2.8.5. Ethics StatementThe rabbit study was carried out in strict accordance with the

recommendations in the Guide for the Care and Use of Labora-tory Animals of the National Institutes of Health. The protocol wasapproved by the Institutional Animal Care and Use Committee ofthe CBER-FDA.

The ferret study was carried out in strict accordance with therecommendations in the Guide for the Care and Use of Labora-tory Animals of the National Institutes of Health. The protocol wasapproved by the Institutional Animal Care and Use Committee ofthe Vaccine and Gene Therapy Institute of Florida (Protocol #TR1.1).All surgery was performed under sodium pentobarbital anesthesia,and all efforts were made to minimize suffering.

2.8.6. Affinity measurements by surface plasmon resonanceSteady-state equilibrium binding of post-H7 vaccination animal

sera was monitored at 25 ◦C using a ProteOn surface plasmon res-onance biosensor (BioRad Labs) as previously described for H5N1and pH1N1 post-vaccination sera [4,11,12]. The H7 rHA1 proteinwas coupled to a GLC sensor chip (BioRad Labs) with amine cou-pling with 500 resonance units (RU) in the test flow cells. Proteinconcentrations on the chip were optimized for detection of mono-valent binding per antibody molecule. Ten-fold dilution of animalsera (60 �l) was injected at a flow rate of 30 �l/min (120-sec con-tact time). Flow was directed over a mock surface to which noprotein was bound, followed by the rHA1 protein coupled sur-face. Responses from the protein surface were corrected for theresponse from the mock surface and for responses from a sepa-rate, buffer only, injection. MAb 2D7 (anti-CCR5) and naïve rabbitsera were used as a negative control antibody in the experiments.Binding kinetics for the animal sera and the data analyses were per-formed with BioRad ProteON manager software. Antibody off-rateconstants which describe the stability of the complex, i.e. the frac-

tion of complexes that decays per second, were determined directlyfrom the serum/plasma sample interaction with HA1 protein usingSPR (as described above) and calculated using the BioRad ProteOnmanager software for the heterogeneous sample model. To improve

cine 3

tt

2

ttatrt

3

3H

ec[

cceapwfT(

ipra3(tfsr((a

3a

ips3NHitcnpctswl

S. Khurana et al. / Vac

he measurements, the off-rate constants were determined fromwo independent SPR runs.

.8.7. Statistical analysisThe area under the curve (AUC) of the body weight and the viral

iters was compared between all vaccine groups using ANOVA. Inhe analysis of the affinity measurements, comparisons of the aver-ge off-rates between groups were performed using the Student’s-test. Spearman correlations are reported for the calculation of cor-elations between off-rate and MN titers. All p values reported arewo-sided.

. Results

.1. Expression and purification of HA1 domain from7N9-A/Shanghai/1/2013.

We previously demonstrated that HA1 proteins (1-320) fromither pH1N1 or H5N1 strains produced under controlled refoldingonditions contained significant fractions of oligomers and trimers1–3] and Fig. 1F.

Soon after the sequence of the H7N9 became available, theomplete HA0 was generated by chemical synthesis and used forloning. pSK plasmid (a T7 based expression vector) was used toxpress the H7 HA1 globular domain (either the complete 1-320mino acids or N-terminus- deleted 28-320 amino acids) as fusionroteins with His6 tag at the C-termini (Fig. 1A). Both proteinsere expressed in E. coli BL21 cells. The proteins were purified

rom inclusion bodies under controlled redox refolding conditions.hey ran as single bands in SDS-PAGE under reducing conditionsFig. 1B).

To determine the distribution of different molecular speciesn the recombinant HA1 proteins derived from H7N9 strain, bothurified rHA1 proteins were subjected to size exclusion chromatog-aphy (SEC). As expected, the H7 HA1 (28-320) ran exclusivelys a monomer (Fig. 1 C). However, unexpectedly, the H7 HA1 (1-20) contained predominantly trimers with no oligomeric formsFig. 1D). In order to determine if this finding was unique tohe HA1 derived from H7N9 we also expressed the HA1 (1-320)rom the highly pathogenic H7N7 (A/Netherlands/219/03) virustrain [13,14]. Again it was found that the H7N7 HA1 (1-320)an primarily as monomer/trimers with low oligomeric fractionFig. 1E) compared to the HA1 (1-320) produced from the H5N1A/Vietnam/1203/2004) that exist forms predominantly oligomerss previously described (Fig. 1F).

.2. H5N1 N-terminal ̌ sheet promotes oligomerization of H7N9nd H7N7 rHA1 proteins

In a previous study with H5N1 group-1 influenza strain, wedentified an oligomerization sequence in the first 8 aa of the HA1rotein that constitutes the first beta sheet in the influenza HAtructure. The core trimerization signal was mapped to amino acids

to 5 (ICI) [3]. Therefore, we compared the HA sequence at the-termini eight amino acid residues of group-2 influenza strains,7N7 and H7N9 vs. the H5N1 (Fig. 2A). Amino acid changes were

dentified at amino acid position 2, 5, and 7. Of these, the muta-ion in amino acid 5 from Isoleucine to Leucine is a conservativehange, not expected to significantly reduce the trimerization sig-al. On the other hand, the changes from Glutamine to Lysine inosition 2 and Tyrosine to Histidine in position 7 introduce positiveharged side chain amino acid changes. As can be seen in the struc-

ure of the trimeric HA1 structure (Fig. 2B) the I-C-I trimerizationignal faces the interface between the N-termini of the monomers,hile amino acids Q-2 and Y-7 are facing outwards and are

ikely to be involved in the trimer-trimer interactions required for

2 (2014) 6421–6432 6425

oligomerization. Therefore, charge changes in these two aminoacids could possibly destabilize the oligomerization of trimers. Toaddress this possibility, the three variable residues in the first 8 aaof the two H7 globular heads (HA1 1-320) were swapped with thethree corresponding residues from H5N1- A/Vietnam/1203/2004strain to generate � sheet domain-swapped H7N9 and H7N7 HA1-DS proteins. In both cases, the purified domain-swapped HA1-DSproteins contained significantly higher oligomeric fractions (Fig. 2 Cand 2D) compared to the wild type H7N9 and H7N7 HA1 counter-parts (Fig. 1D and 1E, respectively).

To evaluate whether the enhanced oligomerization of thedomain-swapped H7N9 rHA1 proteins also promote their func-tional activity, HA1 proteins were tested in the SPR based receptorbinding assay using � 2- 3 sialic acid-glycan immobilized on sen-sor chip and in the hemagglutination assay using human redblood cells (Fig. 2E and 2F respectively). In the SPR receptorbinding assay, the recombinant H7 HA1-DS (Fig. 2E red curve)demonstrated binding kinetics similar to the H7N9 vaccine virus(Fig. 2E, black curve). The wild type H7N9 HA1 (1-320) (blue curve)showed very weak binding, in agreement with the low content oftrimeric/oligomeric forms, while the monomeric H7N9 HA1 (28-320) (green curve) did not bind to the glycan, consistent with theabsence of trimeric/oligomeric forms. Similar binding profiles werefound for the H7N7 HA1-DS vs. wild type HA1 (not shown). Inthe hemagglutination assay, the recombinant wt H7N9 HA1 (1-320) or (28-320), which lack oligomeric forms, did not agglutinateRBC (Fig. 2F top two rows). In contrast, the N-terminus domain-swapped H7 HA1-DS from both H7N9 and H7N7 (3rd and 4th rows)gave significant RBC agglutination, similar to the H7N9 virus (Fig. 2Fbottom row). These findings are in full agreement with our previ-ous studies demonstrating that receptor binding requires at leasttrimers, while the hemagglutination assay requires oligomeric HA(rosettes) for cross-linking RBC and lattice formation [3].

3.3. Rabbit vaccination: generation of better cross-reactiveneutralizing antibodies by domain-swapped H7 HA1 proteins.

To determine the impact of the N-terminus domain swap andhigher oligomer fractions on the immunogenicity of the H7 HA1proteins, rabbits were primed and boosted with H7N9 derivedHA1 (28-320), HA1 (1-320 wt), or HA1-DS mixed with a com-mercial adjuvant (Titermax). We also immunized rabbits withH7N7 derived HA1-wt and HA1-DS proteins (Table 1). All threeH7N9-derived rHA1 proteins generated neutralizing titers afterthe second (but not first) vaccination against the homologousH7N9 virus in a microneutralization assay. These results confirmedthat both monomers and oligomers were properly folded andpresented “relevant” antigenic targets for virus neutralization tothe immune system. In terms of hierarchy, the domain-swappedH7N9 HA1-DS generated the highest H7N9 neutralization titersfollowed by the wild type H7N9 HA1-wt- and the N-terminustruncated HA1 (28-320). The same rabbit sera were tested forcross-neutralization against virus strain containing the H7 fromthe highly pathogenic H7N7-A/Netherlands/219/03 avian influenzavirus. The cross-neutralization titers against H7N7 were only 4-fold lower than against H7N9 for the domain-swapped HA1-DSimmune sera. In contrast, the H7N7 neutralization titers after vacci-nation with the wild type H7N9 HA1-wt and HA1 (28-320) immunesera were 8-fold and 16-fold lower, respectively (Table 1, toppanel). Rabbits vaccinated with the H7N7 derived rHA1 proteinsalso showed higher neutralization titers (homologous and heterol-ogous) with post-second boost sera from H7N7 HA1-DS immunized

animals compared to those immunized with the H7N7 HA1-wt(Table 1 bottom panel). It was of interest to find higher overallneutralization titers after vaccination with the H7N9 derived rHA1proteins compared with the H7N7 rHA1 proteins, which may reflect

6426 S. Khurana et al. / Vaccine 32 (2014) 6421–6432

Figure 2. Biochemical and functional characterization of bacterially produced HA1 and N-terminal domain swap proteins from H7N7 and H7N9 virus strains.(A) Alignment of N-terminal eight amino acids of A/Netherlands/219/03 (H7N7), A/Shanghai/01/2013 (H7N9) and A/Vietnam/1203/2004 (H5N1) virus hemagglutinin. (B) Aribbon structure of the HA1 trimer is shown. The N-terminus trimerization signal (I-C-I) is shown in the enlarged window. The residue changes in position 2 and 7 between H5and H7 are shown in blue (H5). They both face outwards and can contribute to inter-trimer associations. (C-D) Size exclusion chromatography of properly folded bacteriallyproduced H7N7 HA1-DS and H7N9 HA1-DS with N-terminal eight amino acids domain swapped with H5N1-A/Vietnam/1203/2004. The panel present superimposed elutionprofiles of purified HA protein (blue line) overlaid with calibration molecular weight standards (grey line). The elution volumes of protein species are shown. (E) Bindingkinetics of purified H7N9 HA1 proteins and its derivatives in a SPR based receptor binding assay. Steady-state equilibrium analysis of different H7N9-HA1 proteins to �2,3 Sialic acid glycan was analyzed at 25 ◦C using a ProteOn surface plasmon resonance biosensor (BioRad Labs). Samples of purified bacterial H7N9 HA1 proteins (10 �g/ml)H7N9 HA1-DS, red; H7N9 HA1-wt, blue; H7N9 (28-320) green and H7N9 virus, black. (F) Agglutination of human RBC by properly folded bacterially produced H7N7 andH7N9 HA1 proteins and their domain swapped (DS) derivatives, along with H7N9 (A/Shanghai/01/2013) virus. Serial dilutions of purified HA1 proteins were mixed withh

ion

F

uman RBC and hemagglutination was read after 30 min at RT.

nherent differences in immunogenicity of the two strains. On the

ther hand, the antibodies elicited by the H7N7 derived HA1-DSeutralize H7N7 and H7N9 equally (Table 1 bottom panel).

Alignment of the HA sequences from H7N7 and H7N9 (Suppl.ig. 1) identified 15 amino acid differences in the HA1 domain and

only 4 amino acid differences in the HA2 domain. When presented

on the structure of the HA trimer (Fig. 3), it is apparent that severalof the amino acid changes in the HA1 globular domains of H7N7and H7N9 strains are located around the receptor binding domain(RBD) (Fig. 3 green residues).

S. Khurana et al. / Vaccine 32 (2014) 6421–6432 6427

Table 1Geometric Mean and Standard deviations for the end-point microneutralization titers of H7N7-HA1 or H7N9-HA1 immunized rabbits against H7N9 (A/Shanghai/1/2013)and H7N7 (A/Netherlands/219/03) virus strains.

Rabbit Sera Immunizationa Microneutralization titersb (Mean + SD)

H7N9 (A/Shanghai/1/2013) H7N7 (A/Netherlands/219/03)

H7N9 HA1 IMMUNIZEDH7N9 HA 28-320 1 <20 <20

2 853.3 ± 369.5 66.7 ± 23.1H7N9 HA1-wt 1 <20 <20

2 1706.7 ± 739 266.7± 92.4H7N9 HA1-DS 1 <20 <20

2 4266.7 ± 1478 1066.7 ± 369.5H7N7 HA1 IMMUNIZEDH7N7 HA1-wt 1 <20 <20

2 33.3 ± 11.5 33.3 ± 11.5H7N7 HA1-DS 1 <20 <20

2 853 ± 369 640

a Animals were immunized with 50 �g proteins mixed with Titermax adjuvant every three weeks. H7N

( e of t

3fc

fiiH

F(Av(o[d

b MN data (Mean with Standard Deviation) are shown for post-H7N7 andA/Shanghai/1/2013) and H7N7 (A/Netherlands/219/03) virus strains. Representativ

.4. HA1-DS vaccination provides better protection in ferretsrom homologous H7N7 and heterologous challenge H7N9 strainsompared with inactivated subunit H7N7 vaccine

The rabbit immunization data suggested that the oligomeric

orms of HA1 are more efficient in eliciting neutralizing antibod-es. To further evaluate the efficacy of such HA1-based vaccinen influenza virus challenge studies, ferrets were immunized with7N7 HA1-DS, either with oil-in-water adjuvant commonly used

igure 3. Structural map of differences in the HA1 domain between H7N9A/Shanghai/01/2013) and H7N7 (A/Netherlands/219/03) virus hemagglutinin.mino acid differences in the HA1 domain of H7N9 (A/Shanghai/01/2013)irus hemagglutinin (Supplementary figure 1) compared to the H7N7A/Netherlands/219/03) are shown as surface-exposed colored residues (green) onne HA monomer within the H7N7 (A/Netherlands/219/03) HA trimer structureProtein Data Bank (PDB) identifier 4DJ6]. The HA1 globular domain and HA2 stalkomain are indicated. This structure is shown in the side view.

9 following each immunization in a microneutralization assay against H7N9hree experiments.

in veterinary vaccines (Emulsigen) or unadjuvanted, and comparedwith the inactivated subunit H7N7 vaccine. Since the unadjuvantedinactivated H7N7 vaccine was found to be poorly immunogenic inhumans [15], we included a group of ferrets that were vaccinatedwith the inactivated subunit H7N7 vaccine (Sanofi Pasteur) adju-vanted with Emulsigen as well. At the time we have initiated theseferret challenge experiments, the inactivated H7N9 subunit vaccinewas not available for our studies.

Hemagglutination inhibition (HI) titers after the first vacci-nation with the Emulsigen adjuvanted inactivated H7N7 subunitvaccine were all <1:40 against H7N9 and only one seroconvertedagainst the homologous H7N7. After the second vaccination, theimmune sera reached mean titer of 1:80 against homologous H7N7and 1:60 against heterologous H7N9 in vaccinated ferrets (Fig. 4Aand B, blue circles). The unadjuvanted H7N7 HA1-DS also elicitedlow HI titers after the first vaccination but all animals seroconvertedafter a second vaccination, with average HI titers of 1:160 and 1:93against H7N7 and H7N9, respectively (Fig. 4A and B, green cir-cles). Importantly, all ferrets (but one) vaccinated with adjuvantedH7N7 HA1-DS seroconverted after single vaccination against thehomologous H7N7 virus (average HI titer of 1:93), and all fer-rets seroconverted against the heterologous H7N9 virus (averageHI titer of 1:83). The second vaccination with adjuvanted H7N7HA1-DS generated very robust HI titers against both homologousH7N7 and heterologous H7N9 strains (average HI titers of 1:826and 1:680, respectively (Fig. 4A and B, red circles). The HI titersgenerated following adjuvanted H7N7 HA1-DS vaccination weresignificantly higher than the other two vaccinated groups follow-ing both first and second vaccination. As was observed in the rabbitexperiments, the HI titers against H7N7 and H7N9 were very sim-ilar in the ferrets immunized with the H7N7 HA1 proteins (Fig. 4 Avs. B) suggesting targeting of common (conserved) antigenic sitesin the two globular heads.

Influenza virus challenge of the vaccinated ferrets withhighly pathogenic avian influenza (HPAI) H7N7 (A/Netherlands/219/2003) strain resulted in progressive weight loss in the unvac-cinated control group such that all ferrets had to be euthanized byday 14 with body weight loss of ≥ 20% (Fig. 5 A, black curve). Theadjuvanted inactivated H7N7 subunit vaccine group demonstratedlower weight loss than unvaccinated group (ranging from 5-14%),and stabilized after day 8 (Fig. 5A, blue curve). The unadjuvantedHA1-DS vaccinated ferrets showed an initial weight loss similar to

the subunit vaccine group until day 5, when the animals startedto gain weight and recovered to their original body weight by day12-14 (Fig. 5A, green curve). The group that was vaccinated twicewith the adjuvanted H7N7 HA1-DS showed no significant body

6428 S. Khurana et al. / Vaccine 32 (2014) 6421–6432

Figure 4. Domain swapped H7N7 HA1 elicits higher neutralizing antibody titers antibodies against both homologous H7N7 (A/Netherlands/219/03) and heterologous H7N9(A/Anhui/01/2013) virus compared with inactivated H7N7 subunit vaccine in ferrets.Ferrets were immunized twice (3 weeks apart) with recombinant H7N7 HA1-DS (rHA1) proteins with or without Emulsigen adjuvant or with subunit H7N7 vaccine mixedw analyH to be ss tion, w

wp

ivtcvHiuHwocvpit

3ecH

tteewmc(navlvonfl

ith Emulsigen. (Sera were collected 2 weeks after first or second vaccination and7N9 (B) virus strains. Horizontal line represent an HI titer of 40 that is considered

ubunit inactivated H7N7 vaccine groups were highly significant after each vaccina

eight loss after HPAI H7N7 challenge and demonstrated completerotection (Fig. 5A, red curve).

The wild type H7N9 (A/Anhui/1/2013) virus is less pathogenicn ferrets than the wild type HPAI H7N7 (A/Netherlands/213/03)irus. The mock control group lost on average 15% body weight buthen started to re-gain weight after day 8-9 following H7N9 virushallenge (Fig. 5B, black curve). Ferrets vaccinated with the adju-anted inactivated H7N7 subunit vaccine and unadjuvanted H7N7A1-DS demonstrated similar patterns of body weight loss follow-

ng heterologous H7N9 virus challenge that was less severe than thenvaccinated group (Fig. 5B blue and green curves, respectively).owever, as observed in the HPAI H7N7 challenge, the animals thatere vaccinated twice with the adjuvanted H7N7 HA1-DS showed

nly marginal body weight loss after H7N9 challenge (Fig. 5B, redurve). Area under the curve measurements showed that the adju-anted H7 HA1 vaccination provided statistically significant betterrotection from body weight loss compared to the adjuvanted

nactivated H7N7 subunit vaccine against both homologous wildype H7N7 (p= 0.0026) and heterologous H7N9 virus p= 0.013).

.5. H7N7 HA1 reduced viral loads in the nasal washes moreffectively than inactivated subunit H7N7 vaccine in ferretshallenged with wild type homologous H7N7 and heterologous7N9 virus

In addition to survival and weight retention, the capacityo reduce viral replication in the upper and lower respiratoryract is an important attribute of successful influenza vaccines,specially against pandemic strains H7N7 and H7N9. To thatnd, viral loads in the nasal washes from ferrets challengedith H7N7-A/Netherlands/219/03 or H7N9-A/Anhui/1/2013 wereeasured every alternate day following virus challenge. As

an be seen in Fig. 5C, following homologous H7N7 challengeA/Netherlands/219/03), high viral loads were measured in theasal washes of unvaccinated animals. Ferrets vaccinated withdjuvanted inactivated H7N7 subunit vaccine did not decrease theiral loads in nasal washes at any day following HPAI H7N7 chal-enge compared to the unvaccinated mock control (Fig. 5 C, blues. black curves). In comparison, the viral loads in nasal washes

f animals vaccinated with adjuvanted H7N7 HA1-DS were sig-ificantly lower (3-5 log10 reduction of average viral titers) in all

errets at all days following homologous HPAI H7N7 virus chal-enge compared to the unvaccinated or the adjuvanted inactivated

zed in HI assay using human RBC against homologous H7N7 (A) and heterologouseroprotective. The HI titers of the post-vaccination ferret sera between the HA1 vs.ith p< 0.05 (t-Test).

H7N7 subunit vaccine (red curve). Even in the unadjuvanted H7N7HA1-DS vaccinated ferrets, the nasal viral loads were significantlylower (>2 log10 reduction of average viral titers) in all ferrets atall days following homologous HPAI H7N7 virus challenge (greencurve). Interestingly, after heterologous H7N9 virus challenge(A/Anhui/1/2013), the lowest viral loads were observed in nasalwashes from ferrets vaccinated with the adjuvanted H7N7 HA1-DS (>2 log10 reduction of average viral titers) (Fig. 5D red curve),followed by unadjuvanted H7N7 HA1-DS (green curve), relativeto the mock-vaccinated and inactivated H7N7- subunit vaccinatedferrets (Fig. 5D black and blue curves, respectively)). Area under thecurve measurements showed that the adjuvanted H7 HA1 vaccina-tion provided statistically significant better protection from virusreplication compared to the adjuvanted inactivated H7N7 subunitvaccine against both homologous wild type H7N7 (p = 0.000001)and heterologous H7N9 virus (p = 0.000001). This was an unex-pected finding and may indicate differences in either the specificityor quality of the antibodies generated by the oligomeric H7 HA1-DSversus the inactivated H7N7 subunit vaccine.

3.6. Adjuvanted H7 HA1-DS promoted better antibody affinitymaturation against H7 hemagglutinin compared with inactivatedH7N7 subunit vaccine that correlated strongly with HI titers andinversely correlated with viral loads

To gain a better understanding of the differences in quality of theantibody response generated following vaccination with the differ-ent H7 immunogens, all post-vaccination ferret sera were subjectedto real time SPR based kinetic assay using H7N7 rHA1 coated chips.To that end, we measured the antibody affinity (antigen-antibodycomplex dissociation rates) of post-first and post-second vaccina-tion ferret sera from all vaccinated ferrets as previously described[4,11]. As can be seen in Fig. 6A, the adjuvanted inactivated H7N7vaccine elicited very low affinity antibodies (faster dissociationrates) after one vaccination, which were improved after the secondvaccination (i.e. slower dissociation rates) (Fig. 6A, blue circles).The unadjuvanted H7-HA1-DS gave intermediate antibody affini-ties with increase in affinity observed between the first and secondvaccination (Fig. 6A, green circles). Importantly, the adjuvanted

H7-HA1-DS immune polyclonal sera showed very high bindingaffinity antibodies already after the first vaccination (average offrates: 4 × 10−4/sec) and a further increase in antibody affinity afterthe second vaccination (Fig. 6A, red circles), that were significantly

S. Khurana et al. / Vaccine 32 (2014) 6421–6432 6429

Figure 5. H7N7 HA1 provides better control of weight loss and viral load compared with subunit H5N1 vaccine following challenge of ferrets with wild-type homologous(H7N7) or heterologous (H7N9) viruses.Six ferrets in each group were immunized with either the unadjuvanted H7N7 HA1-DS (H7-HA1; green circles), or adjuvanted H7N7 HA1-DS (H7-HA1 + Emulsigen; redcircles), or with adjuvanted inactivated H7N7 subunit vaccine (Inactivated + Emulsigen; blue circles) or PBS (Mock; black circles). Following two immunizations, ferrets (sixper group) were infected intranasally with 1 × 107 plaque forming units (pfu) of wild type homologous H7N7-A/Netherlands/219/03 (A and C) and heterologous H7N9-A/Anhui/01/2013 (B and D). A-B Animals were scored for percent original body weight for two weeks. (C-D) Viral loads in nasal washes following challenge of vaccinatedferrets with wild type homologous H7N7-A/Netherlands/219/03, or heterologous H7N9-A/Anhui/01/2013 on days 1, 3, 5 post-virus challenge. Horizontal lines representa uvantt type H

hbHi(cC

wcvopt

verage pfu of viral loads in the nasal washes of each group (6 ferrets per group). Adjo the adjuvanted inactivated H7N7 subunit vaccine following homologous (C) wild

igher affinity compared to adjuvanted H7N7 subunit vaccine afteroth first and second vaccination. Importantly, the polyclonal anti-A1 antibody off rates for all the ferrets demonstrated very strong

nverse correlation with the HI titers against both homologousH7N7) and heterologous (H7N9) virus strains following first (openircles) and second (filled circles) vaccination (Fig. 6, panels B and, respectively).

Importantly, as can be seen in Fig. 7, viral loads in the nasalashes of all ferrets challenged with either H7N7 or H7N9 inversely

orrelated with the anti-HA1 antibody off rates after the second

accination either on day 1 or day 3 post-challenge with homol-gous wild type H7N7 and heterologous H7N9 viruses (Fig. 7,anels A-B and C-D, respectively). Similar to the weight loss results,he viral loads were most efficiently controlled in the ferrets that

ed H7 HA1 vaccinated ferrets had statistically significant lower viral load compared7N7 (p = 0.000001) and (D) heterologous H7N9 virus (p = 0.000001) virus challenge.

were vaccinated with the adjuvanted H7N7 HA1-DS (Fig. 7, redcircles) and least controlled in the animals vaccinated with theadjuvanted inactivated H7N7 subunit vaccine (Fig. 7 blue circles),which inversely correlated with polyclonal sera antibody affinityprior to challenge.

In summary, these studies demonstrated that in the case ofgroup 2 avian strains (H7N7 and H7N9) the ability of the globulardomain to form functional oligomers can be significantly improvedafter swapping the N-terminal 8 amino acids that form �-sheetwith the corresponding 8 amino acids from the group 1 avian strain

H5N1 virus. This �-sheet domain-swap significantly increased thefraction of functional oligomeric forms and resulted in higher virusneutralization titers (rabbits) and HI titers (ferrets) against bothhomologous and heterologous H7 strains. Most importantly, the

6430 S. Khurana et al. / Vaccine 3

Figure 6. Oligomeric H7N7 HA1 elicits higher affinity HA1 binding antibodiescompared with inactivated subunit H7N7 vaccine that significantly correlates withhomologous H7N7 and heterologous H7N9 neutralizing antibody titers in ferrets.Real time antibody kinetics of ferret sera immunized with H7N7 HA1-DS orinactivated H7N7 subunit vaccine were measured by SPR. Post-vaccinationferret sera were collected 8 days following first and second immunizations withadjuvanted subunit H7N7 vaccine (Inact + Em; blue circles), adjuvanted oligomericH7N7 HA1-DS (rHA1 + Em; red circles), or unadjuvanted oligomeric H7N7 HA1-DS(rHA1; green circles). Sera (1:10 dilution) were injected simultaneously onto H7N7(A/Netherlands/219/03) HA1 protein immobilized on a sensor chip through thefree amine group, and onto a blank flow cell, free of peptide. (A) Post-vaccinationserum antibody affinity to HA1 domain as measured by antigen-antibody complexoff-rate constants in SPR analysis of immunized ferret sera. Serum antibody off-rateconstants were determined as described in material and methods. (B-C) Correlationbetween serum HI titers for 18 ferrets (six per group) against homologous H7N7(A/Netherlands/219/03) virus after first and second immunization (B), or against

2 (2014) 6421–6432

recombinant HA1 -DS generated high affinity antibodies in ferretsthat gave superior protection from weight loss and better controlof virus replication in the upper respiratory track (URT) comparedwith the traditional inactivated H7N7 subunit vaccine, which isextremely important to reduce virus transmission. The study alsosuggested that in the face of real pandemic threat, a rHA1-basedvaccine can be rapidly produced and the immunogen (adjuvanted)may possibly generate protective antibody titers after a single vac-cination, not previously observed in naïve animals, more efficientlythan traditional inactivated subunit vaccines.

4. Discussion

The recently circulating H7N9 strains in China and highlypathogenic H7N7 avian flu in Europe with proven binding to bothavian and human receptors pose a risk of further adapting to repli-cation in humans upper respiratory tract. Transmission betweenferrets and pigs was already demonstrated and thus, only a fewadditional mutations may be required for efficient human-to-human transmission [6] [16] [8]. The lack of pre-existing immunityagainst H7N7 and H7N9 virus may result in a severe influenzapandemic.

The structure of HA suggest that most of the inter subunit saltbridges and hydrophobic interactions are between the HA2 chainsdue to coiled-coil structure which forms the stem of the HA trimer.These earlier HA-structural studies did not describe the oligomeri-zation signal in the HA1 globular domain identified in the currentstudy, suggesting that in the presence of HA2 the N-terminus�-sheet structure is engaged in HA1-HA2 bridge and not in HA1 oli-gomerization. However, in the absence of HA2 sequence, the HA1globular head formed oligomers and agglutinate human RBC, aswas observed with multiple group-1 influenza subtype strains pro-duced by our group [1–4]. It is likely that in the native spikes theN-terminal �-sheets of the three HA1 globular domains are not insufficient proximity to form oligomers, but in the absence of HA2they are free and close enough to provide the needed oligomeriza-tion signal.

Shortly after the isolation of H7N7 and H7N9 viruses fromhumans, we initiated cloning and expression of HA1 and HA2domains from H7N7 and H7N9 hemagglutinin in our bacterialsystem that was successfully used previously in the productionof HA1 and HA2 proteins from H5N1 and pH1N1 [2–4]. How-ever, the bacterially produced wild type HA1 (1-320) from H7N7(A/Netherlands/219/03) and H7N9 (A/Shanghai/01/2013) did notcontain significant fraction of oligomers. The oligomerizationsequence in HA1 (of H5N1) was previously mapped to the first8 amino acids that form a beta sheet, and specifically to the I-C-Icore (aa 3-5) [3]. Alignment of the N-termini of H7N9, H7N7and H5N1 hemagglutinin sequences identified three amino aciddifferences; K2Q, L5I, and H7Y. Therefore, we swapped the threevariable residues in the N-terminal 8 amino acids of H7 with thecorresponding amino acid residues from H5N1 since this part of theHA1 is not involved in any of the main antigenic sites in the globu-lar head. The domain swapped H7 HA1-DS proteins were found tobe properly folded and to contain primarily oligomeric forms. Theoligomeric forms were not simply aggregates of monomers, but

contained multiples of functional trimers. In subsequent testing,we found that H7 HA1-DS from H7N7 and H7N9 strains are supe-rior to both HA1-wt, in the following assays: a) binding to the sialicacid receptor �-2,3-SA glycan in SPR; b) agglutination of human

heterologous H7N9-A/Anhui/01/2013 (C), and the individual serum antibody dis-sociation rates against H7-HA1 derived from H7N7 (A/Netherlands/219/03). Colorsare the same as in Panel A. Open circles, post first vaccination; close circles, postsecond vaccination. Correlation statistics of the off-rate constants for post-vaccineferret sera with HI titers were highly significant, with p< 0.05.

S. Khurana et al. / Vaccine 32 (2014) 6421–6432 6431

Figure 7. H7N7 HA1-DS generates higher affinity antibodies compared with inactivated H7N7 vaccine in immunized ferrets that correlate with lower viral loads in nasalwashes following challenge with homologous wild type H7N7-A/Netherlands/219/03 or heterologous H7N9-A/Anhui/01/2013 virus. Post-vaccination ferret serum antibodyaffinity to HA1 derived from H7N7 (A/Netherlands/219/03) was measured by antigen-antibody complex off-rate constants in SPR (Fig. 6A). Dissociation off-rates for individualferret polyclonal sera were correlated with the viral loads (pfu/ml) in nasal washes on days 1 (A-B) or day 3 (C-D) post challenge with (A-C), or virus (B-D). Colored circlesrepresent the different groups: Emulsigen adjuvanted H7 HA1-DS (red circles), unadjuvanted H7 HA1-DS (green circles), and adjuvanted subunit inactivated H7N7 vaccine( signifia 0.5).

l

RreHsHcaH

bastsacetsN

i

blue circles). The polyclonal serum anti-HA1 antibody off-rates show statistically

nd H7N9-A/Anhui/01/2013 (B and D) on day 1 and 3 following virus challenge (R2 >oads in nasal washes were highly significant, with p< 0.05.

BC; c) generation of neutralizing antibodies in rabbits; d) in fer-ets the adjuvanted HA1-DS from H7N7 (A/Netherlands/213/03)licited higher HI titers compared with adjuvanted inactivated7N7 subunit vaccine, with most ferrets seroconverting after a

ingle vaccination; e) most importantly, the adjuvanted H7A1-DS elicited higher affinity antibodies that were effective in

ontrolling virus replication and completely prevented weight lossfter challenge with the homologous (H7N7) or the heterologous7N9 strains.

We have previously demonstrated the production of recom-inant HA1 proteins from both pH1N1 and H5N1 (A/Vietnamnd A/Indonesia) (1, 2, 3, 4). All of the rHA1 proteins containedignificant oligomeric fractions, which are required for RBC agglu-ination and high immunogenicity. We also shown the N-terminusequences from 16 different influenza strains from both Group 1nd Group 2 (Ref. 3, Suppl. Figure 5). All 16 strains contain theore trimerization signal (amino acids 2-5) (I-C-I or I-C-L). How-ver in positions 2 and 7 that were identified in the current studyo be important for stable oligomerization, several other group 2

trains have different residues that could change the charge on the-terminal �-sheet and result in lower oligomeric fractions.

The H7N7 that caused human infections in the Netherlandsn 2003 and in Italy in 2013 contains the polybasic cleavage site

cant inverse correlation with virus titers of H7N7-A/Netherlands/219/03 (A and C)Correlation statistics of the off-rate constants for post-vaccine ferret sera with viral

in HA, a property of highly pathogenic influenza viruses, andwas associated with health complaints, especially conjunctivi-tis [17] [14]. There are several amino acid differences betweenH7N7 and H7N9 hemagglutinins, indicating different virus evo-lution. Some of the different amino acids within the H7N7 andH7N9 hemagglutinin are in exposed regions of the HA1 globularhead including a site close to the RBD [18] (Fig. 3). The highestH7N7 virus neutralizing titers were found in sera from rabbitsvaccinated with the domain swapped H7N9 HA1 protein, thatwas only four fold lower than the titer against homologous H7N9virus (vaccine strain) (Table 1). These titers were even higherthan observed after vaccination of rabbits with the H7N7 derivedHA1-DS. This finding is of interest and may reflect overall higherimmunogenicity of the H7N9 HA. Ongoing trials in humans (withadjuvanted and unadjuvanted H7N9 inactivated vaccines) willaddress this question in light of the very poor immunogenicityreported in human trials with the H7N7 (A/Netherlands) subunitvaccine [15]. It will be of interest to determine if vaccination ofhumans with adjuvanted subunit or VLP H7N9 vaccines elicits

cross-reactive H7N7 neutralizing antibodies. Our findings with H7rHA1 proteins are in full agreement with previous studies fromour lab and others demonstrating that oligomeric forms of HA areimportant for generation of robust protective immune responses

6 cine 3

as

abHHdovp

C

A

rgMF

A

f2

R

[

[

[

[

[

[

[

[

[

[

432 S. Khurana et al. / Vac

gainst homologous and also heterologous influenza virustrains [3,4,19,20].

The findings in the current study show for the first time that domain-swap of the N-terminal 8 amino acid residues coulde used for stabilizing trimeric/oligomeric functional forms ofA1 from emerging influenza strains with pandemic potential.ighly protective domain swapped HA1 vaccine could be pro-uced rapidly and efficiently in bacterial system after sequencingf novel influenza strains before the derivation of high growthaccine virus strain, as part of a rapid response to influenzaandemic.

onflict of Interest

The authors have no conflict of interest to report

cknowledgements

We are thankful to Carol Weiss and Zhiping Ye for thorougheview of the manuscript. The H7N3 containing the H7 hemag-lutinin from H7N7-A/Netherlands/02/2003 was kind gift fromaryna Eichelberger, CBER, FDA. The research was funded by CBER,

DA.

ppendix A. Supplementary data

Supplementary data associated with this article can beound, in the online version, at http://dx.doi.org/10.1016/j.vaccine.014.09.049.

eferences

[1] Khurana S, Larkin C, Verma S, Joshi MB, Fontana J, Steven AC, et al. Recom-binant HA1 produced in E. coli forms functional oligomers and generatesstrain-specific SRID potency antibodies for pandemic influenza vaccines. Vac-cine 2011;29:5657–65.

[2] Khurana S, Verma S, Verma N, Crevar CJ, Carter DM, Manischewitz J, et al.Properly folded bacterially expressed H1N1 hemagglutinin globular head andectodomain vaccines protect ferrets against H1N1 pandemic influenza virus.

PLoS ONE 2010;5:e11548.

[3] Khurana S, Verma S, Verma N, Crevar CJ, Carter DM, Manischewitz J, et al.Bacterial HA1 vaccine against pandemic H5N1 influenza virus: evidence of oli-gomerization, hemagglutination, and cross-protective immunity in ferrets. JVirol 2010;85:1246–56.

[

2 (2014) 6421–6432

[4] Verma S, Dimitrova M, Munjal A, Fontana J, Crevar CJ, Carter DM, et al.Oligomeric recombinant H5 HA1 vaccine produced in bacteria protects ferretsfrom homologous and heterologous wild-type H5N1 influenza challenge andcontrols viral loads better than subunit H5N1 vaccine by eliciting high-affinityantibodies. J Virol 2012;86:12283–93.

[5] Gao R, Cao B, Hu Y, Feng Z, Wang D, Hu W, et al. Human infection with a novelavian-origin influenza A (H7N9) virus. N Engl J Med 2013;368:1888–97.

[6] Zhu H, Wang D, Kelvin DJ, Li L, Zheng Z, Yoon SW, et al. Infectivity, transmis-sion, and pathology of human-isolated H7N9 influenza virus in ferrets and pigs.Science 2013;341:183–6.

[7] Watanabe T, Kiso M, Fukuyama S, Nakajima N, Imai M, Yamada S, et al.Characterization of H7N9 influenza A viruses isolated from humans. Nature2013;501:551–5.

[8] Li Q, Zhou L, Zhou M, Chen Z, Li F, Wu H, et al. Epidemiology of human infec-tions with avian influenza A(H7N9) virus in China. N Engl J Med 2014;370:520–32.

[9] Smith GE, Flyer DC, Raghunandan R, Liu Y, Wei Z, Wu Y, et al. Development ofinfluenza H7N9 virus like particle (VLP) vaccine: homologous A/Anhui/1/2013(H7N9) protection and heterologous A/chicken/Jalisco/CPA1/2012 (H7N3)cross-protection in vaccinated mice challenged with H7N9 virus. Vaccine2013;31:4305–13.

10] Dormitzer PR, Suphaphiphat P, Gibson DG, Wentworth DE, Stockwell TB, AlgireMA, et al. Synthetic generation of influenza vaccine viruses for rapid responseto pandemics. Sci Transl Med 2013;5:185ra68.

11] Khurana S, Verma N, Yewdell JW, Hilbert AK, Castellino F, Lattanzi M, et al.MF59 adjuvant enhances diversity and affinity of antibody-mediated immuneresponse to pandemic influenza vaccines. Sci Transl Med 2011;3:85ra48.

12] Khurana S, Verma N, Talaat KR, Karron RA, Golding H. Immune response follow-ing H1N1pdm09 vaccination: differences in antibody repertoire and avidity inyoung adults and elderly populations stratified by age and gender. J Infect Dis2011;205:610–20.

13] Capua I. Avian influenza: We have the chance to make a difference. Vet J2007;174:213–4.

14] Koopmans M, Wilbrink B, Conyn M, Natrop G, van der Nat H, Vennema H,et al. Transmission of H7N7 avian influenza A virus to human beings dur-ing a large outbreak in commercial poultry farms in the Netherlands. Lancet2004;363:587–93.

15] Couch RB, Patel SM, Wade-Bowers CL, Nino D. A randomized clinical trial of aninactivated avian influenza A (H7N7) vaccine. PLoS One 2012;7:e49704.

16] Zhang Q, Shi J, Deng G, Guo J, Zeng X, He X, et al. H7N9 influenza viruses aretransmissible in ferrets by respiratory droplet. Science 2013;341:410–4.

17] Bos ME, Te Beest DE, van Boven M, van Beest Holle MR, Meijer A, BosmanA, et al. High probability of avian influenza virus (H7N7) transmission frompoultry to humans active in disease control on infected farms. J Infect Dis2010;201:1390–6.

18] Yang H, Carney PJ, Donis RO, Stevens J. Structure and receptor complexesof the hemagglutinin from a highly pathogenic H7N7 influenza virus. J Virol2012;86:8645–52.

19] Wei CJ, Xu L, Kong WP, Shi W, Canis K, Stevens J, et al. Comparative efficacy of

neutralizing antibodies elicited by recombinant hemagglutinin proteins fromavian H5N1 influenza virus. J Virol 2008;82:6200–8.

20] Weldon WC, Wang BZ, Martin MP, Koutsonanos DG, Skountzou I, Compans RW.Enhanced immunogenicity of stabilized trimeric soluble influenza hemagglu-tinin. PLoS ONE 2009:2009.