Baculovirus-derived hemagglutinin vaccines protect against lethal influenza infections by avian H5 and H7 subtypes John Crawford a, 1 , Bethanie Wilkinson b , Andrei Vosnesensky b , Gale Smith b , Maricarmen Garcia a, 2 , Henry Stone a , Michael L. Perdue a, * a Southeast Poultry Research Laboratory, ARS, USDA, 934 College Station Road, Athens, GA 30605, USA b Protein Sciences Corporation, 1000 Research Parkway, Meriden, CT 06450, USA Received 2 July 1998; received in revised form 3 December 1998; accepted 4 December 1998 Abstract Baculoviruses were engineered to express hemagglutinin (HA) genes of recent avian influenza (AI) isolates of the H5 and H7 subtypes. The proteins were expressed as either intact (H7) or slightly truncated versions (H5). In both cases purified HA proteins from insect cell cultures retained hemagglutination activity and formed rosettes in solution, indicating proper folding. Although immunogenic in this form, these proteins were more eective when administered subcutaneously in a water-in-oil emulsion. One or two-day-old specific pathogen free (SPF) White Rock chickens, free of maternal AI antibodies, responded with variable serum HI titers, but in some cases the titers were comparable to those achieved using whole virus preparations. Vaccination of three-week-old chickens with 1.0 mg of protein per bird generated a more consistent serum antibody response with an average geometric mean titer (GMT) of 121 (H5) and 293 (H7) at 21 days postvaccination. When challenged with highly pathogenic strains of the corresponding AI subtypes, the vaccinated birds were completely protected against lethal infection and in some cases exhibited reduced or no cloacal shedding at 3 days postinfection. Vaccine protocols employing these recombinant HA proteins will not elicit an immune response against internal AI proteins and thus will not interfere with epidemiological surveys of natural influenza infections in the field. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Avian influenza; Hemagglutinin; Baculovirus-origin subunit vaccine 1. Introduction Avian influenza (AI) viruses, type A members of the Orthomyxoviridae family, have eight negative sense RNA segments encoding 10 proteins [1]. Four of these proteins are important targets of the host immune re- sponse. The nucleoprotein (NP) and matrix 1 (M1) protein elicit both humoral and cellular responses fol- lowing infection [2–5], which help to clear the virus from the host but do not neutralize virus due to the in- ternal location of these proteins. The hemagglutinin (HA) and neuraminidase (NA) genes encode virulence- associated surface glycoproteins [6], and antibody to either, inhibits infection [7] or prevents disease [8]. The HA protein is the most abundant surface glycoprotein [9], is responsible for attachment of virus to terminal sialic acid residues on host cell receptors [10], and mediates fusion between viral and cellular membranes [11]. Additionally, antibody re- sponse to the HA protein is the basis for dividing influenza into its 15 distinct antigenic subtypes [12–15]. Avian influenza virus subtypes 5 and 7 (H5 and H7) are unique in that they have undergone rapid phenoty- pic shifts to highly pathogenic (HP) variants both in field and laboratory settings [16, 17]. Avian influenza remains an economic threat to commercial poultry Vaccine 17 (1999) 2265–2274 0264-410X/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0264-410X(98)00494-0 1 Present address. Department of Surgery, Center for AIDS research, Duke University Medical Center, Durham, NC 27710, USA. 2 Present address. Department of Avian Medicine, University of Georgia, 935 College Station Road, Athens, GA 30605, USA. * Corresponding author. Tel.: +1-706-546-3435; fax: +1-706-546- 3161; e-mail: [email protected].
Transcript
Baculovirus-derived hemagglutinin vaccines protect against lethalin¯uenza infections by avian H5 and H7 subtypes
John Crawforda, 1, Bethanie Wilkinsonb, Andrei Vosnesenskyb, Gale Smithb,Maricarmen Garcia a, 2, Henry Stonea, Michael L. Perduea, *
aSoutheast Poultry Research Laboratory, ARS, USDA, 934 College Station Road, Athens, GA 30605, USAbProtein Sciences Corporation, 1000 Research Parkway, Meriden, CT 06450, USA
Received 2 July 1998; received in revised form 3 December 1998; accepted 4 December 1998
Abstract
Baculoviruses were engineered to express hemagglutinin (HA) genes of recent avian in¯uenza (AI) isolates of the H5 and H7subtypes. The proteins were expressed as either intact (H7) or slightly truncated versions (H5). In both cases puri®ed HAproteins from insect cell cultures retained hemagglutination activity and formed rosettes in solution, indicating proper folding.
Although immunogenic in this form, these proteins were more e�ective when administered subcutaneously in a water-in-oilemulsion. One or two-day-old speci®c pathogen free (SPF) White Rock chickens, free of maternal AI antibodies, responded withvariable serum HI titers, but in some cases the titers were comparable to those achieved using whole virus preparations.
Vaccination of three-week-old chickens with 1.0 mg of protein per bird generated a more consistent serum antibody responsewith an average geometric mean titer (GMT) of 121 (H5) and 293 (H7) at 21 days postvaccination. When challenged with highlypathogenic strains of the corresponding AI subtypes, the vaccinated birds were completely protected against lethal infection and
in some cases exhibited reduced or no cloacal shedding at 3 days postinfection. Vaccine protocols employing these recombinantHA proteins will not elicit an immune response against internal AI proteins and thus will not interfere with epidemiologicalsurveys of natural in¯uenza infections in the ®eld. # 1999 Elsevier Science Ltd. All rights reserved.
Avian in¯uenza (AI) viruses, type A members of the
Orthomyxoviridae family, have eight negative sense
RNA segments encoding 10 proteins [1]. Four of these
proteins are important targets of the host immune re-
sponse. The nucleoprotein (NP) and matrix 1 (M1)
protein elicit both humoral and cellular responses fol-
lowing infection [2±5], which help to clear the virus
from the host but do not neutralize virus due to the in-ternal location of these proteins. The hemagglutinin(HA) and neuraminidase (NA) genes encode virulence-associated surface glycoproteins [6], and antibody toeither, inhibits infection [7] or prevents disease [8]. TheHA protein is the most abundant surfaceglycoprotein [9], is responsible for attachment of virusto terminal sialic acid residues on host cellreceptors [10], and mediates fusion between viral andcellular membranes [11]. Additionally, antibody re-sponse to the HA protein is the basis for dividingin¯uenza into its 15 distinct antigenic subtypes [12±15].
Avian in¯uenza virus subtypes 5 and 7 (H5 and H7)are unique in that they have undergone rapid phenoty-pic shifts to highly pathogenic (HP) variants both in®eld and laboratory settings [16, 17]. Avian in¯uenzaremains an economic threat to commercial poultry
Vaccine 17 (1999) 2265±2274
0264-410X/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0264-410X(98 )00494 -0
1 Present address. Department of Surgery, Center for AIDS
research, Duke University Medical Center, Durham, NC 27710,
USA.2 Present address. Department of Avian Medicine, University of
Georgia, 935 College Station Road, Athens, GA 30605, USA.
worldwide. Frequent isolation of H5 and H7 subtypeAI viruses from live bird markets [18] and migratorywaterfowl [19, 20] suggest these reservoirs are likelysources of introduction into the larger commercial set-tings, sometimes resulting in devastating consequences.Examples include the North American highly patho-genic outbreaks in Pennsylvania in 1983 [21] and incentral Mexico during 1994±1995 [22], and recurrentH7 HP outbreaks in Australia [23±25]. These out-breaks almost always result in signi®cant economiclosses either in quarantine and indemni®cation/clean-up costs or import restrictions on the country orregion of origin [26,27]. The recent direct transmissionof lethal AI viruses of the H5 subtype to humans inHong Kong [28], as well as the documented cases oftransmission of H7 subtype to humans and othermammals [29±32], has elevated the need to control AIbeyond economic considerations.
Past vaccination initiatives to counter AI in UScommercial poultry were banned, impractical, or posedan unacceptable risk. Conventional parenteral vaccinesrequire a prolonged postvaccination waiting periodwhen oil emulsion adjuvants are employed [33]. Hostresponse to vaccine-origin NP prevents distinctionbetween vaccine and ®eld strains [34, 35], and with livevirus vaccines, the possibility of reassortment among®eld and vaccine RNA segments exists [36]. Crawfordet al. demonstrated that oral immunization against AIwas e�cacious and did not induce anti-NP antibodiesin vaccinated chickens; however, multiple doses wererequired [37]. A more practical approach may be asubunit vaccine that avoids all of the above pitfalls. Inthis report, we describe the development and successfuluse of recombinant, baculovirus-origin H5 or H7 HAprotein as an e�ective vaccine in chickens.
2. Materials and methods
2.1. Virus isolates, histories, and replication
Eight viruses were used for this study: A/chicken/Pennsylvania/1370/1983 (H5N2) (CP83); A/chicken/Queretaro/19/1995 (H5N2) (CQ95), NationalVeterinary Services Laboratories (NVSL), Ames, IA;A/FPV/Rostock (Germany)/1934 (H7N2) (CG34); A/chicken/Jalisco/1994 (H5N2) (CJ94); A/chicken/NewYork/1995 (H7N2) (CNY95); A/pekin robin/California/1994 (H7N1) (PR94); A/chicken/Victoria/1976 (H7N7) (CV76); and A/turkey/Oregon/1971(H7N3). All isolates were ampli®ed in speci®c-patho-gen-free (SPF) 10-day-old embryonating eggs from¯ocks at, Southeast Poultry Research Laboratory
(SEPRL). Allantoic ¯uid was harvested for use as viralstocks.
Ampli®ed virus harvested from inoculated embryo-nating eggs was extracted using acid±guanidinium±phenol [38]. A single tube, reverse transcription-poly-merase chain reaction protocol (RT-PCR) was used togenerate and amplify HA-speci®c cDNA from viralRNA [39]. This product was sequenced directly usingdye-deoxy2 labeled terminators and a 373 automatedDNA sequencer [40], as speci®ed by AppliedBiosystems protocols (Foster City, CA). Speci®c pri-mers used for sequencing are available on request.Nucleotide sequences were compiled and analyzedusing GeneWorks 2.452 software (Intelligenetics, Mt.View, CA). The ®nal plasmid constructs were alsosequenced using the same primers. Sequences of theHA coding regions in the plasmids are deposited inGenbank, accession numbers: AF071775 andAF071776.
2.3. Cloning HA genes from H5 and H7 RNA
A/chick/Jalisco/94 (H5N2) and A/chick/New York/95 (H7N2) strains were selected to produce recombi-nant hemagglutinin. Puri®ed RNA from infected allan-toic ¯uid was used as a template to make cDNAutilizing Moloney Murine Leukemia Virus (M-MuLV)reverse transcriptase (Gibco-BRL, Gaithersburg, MD).The primer used for cDNA synthesis was homologousto the 3 0 end of all in¯uenza gene virion segments (5 0-AGCAAAAGCAGG-3 0).
Ampli®cation of the HA genes were accomplishedby the polymerase chain reaction (PCR) using GeneAmp kits obtained from Perkin Elmer (Foster City,CA). The 5 0 and 3 0 primers contained restrictionenzyme sites not found within the HA genes (Sma1and Kpn1). The separate PCR reaction mixtures (100ml) contained 20 pmol each of primers speci®c for 5 0
and 3 0 ends for each of the CJ94 and CNY95 HAgenes. PCR-ampli®cation was carried out for 25±30cycles each consisting of 1 min of denaturation at948C, 2 min at 358C for reannealing and 2 min at728C for extension. The resulting HA PCR productswere digested with Sma1-Kpn1, ligated into transferplasmids with compatible ends (TA cloning system,Invitrogen), ampli®ed in bacteria, and veri®ed byDNA sequence analysis. The HA gene transfer plas-mids were subcloned to create H5 and H7 HABaculovirus expression vectors [41].
J. Crawford et al. / Vaccine 17 (1999) 2265±22742266
2.4. Protein puri®cation and assay
Proteins were puri®ed from infected insect cells byion-exchange chromatography to greater than 95%purity. Final protein concentration in PBS wasmeasured by the BCA protein assay (PierceBiochemicals, Rockport, IL).
2.5. Animal hosts and housing
Speci®c pathogen free (SPF) White Rock chickens(SEPRL) were used for vaccine studies. For challengestudies, chickens were transferred to a bio-safety level3-agriculture facility, housed in stainless steel isolationcabinets (ventilated through high e�ciency particulateair (HEPA) ®lters to maintain negative-pressure), andgiven feed and water ad libitum. Chicken groupsreceiving di�erent challenge virus strains were housedin separate, negative-pressure HEPA ®ltered cabinets.
2.6. Experimental design
Four vaccination trials were performed to evaluatechicken serum antibody response to vaccine, e�cacyagainst morbidity and mortality, and reduction of viralshed following a lethal avian in¯uenza challenge.Results from four trials are presented in this report;designs for the four trials are as follows:
2.6.1. Trial 1Seven groups of birds were employed in trial 1. One
control group (24 chickens), received no vaccination;six groups (10 chickens each) received various vaccinetreatments as outlined in Table 1. Each group was sub-sequently split into halves and challenged with eitherA/chicken/Queretaro/1995 (H5N2) or A/chicken/Pennsylvania/1983 (H5N2). Each chicken in the sixtreatment groups received a single 0.2-ml subcutaneousinjection containing the listed number of hemaggluti-nating units at one day of age. Groups 1±3 were deliv-ered as partially puri®ed protein in solution; groups 4and 5 vaccines contained immunostimulating lipo-some/HA complexes (ISCOM); group 6 birds receivedthe HA protein emulsi®ed in a water-in-oil adjuvant(ADJV) as previously described [42]. Hemagglutinationinhibition (HI) titers of sera were measured at 28 dayspostvaccination. Sera were tested using either A/tur-key/Wisconsin/1968 (H5N9) or A/chicken/Jalisco/1994(H5N2) as antigen. Based on results from these trials,all remaining trials were performed with proteins for-mulated in the water-in-oil adjuvant.
2.6.2. Trial 2Trial two was designed to evaluate the response of
three-week-old birds to increasing amounts of puri®edHA protein in adjuvant. Five groups of ®ve birds eachwere inoculated with four concentrations of HA asoutlined in Table 2. Serum HI titers were measured at
Table 1
Summary of host response and virus shedding in 1-day-old SPF White Plymouth Rock chicks using baculovirus-origin H5 HA protein
Vaccine formulationa Challenge virus HI respondersb Morbidity Mortality Virus recovery
oral cloacal
Control CQ95 0/12 12/12 12/12 12/12 11/12
CP83 0/12 12/12 12/12 12/12 12/12
250 HAU CQ95 1/5 5/5 3/5 5/5 5/5
CP83 0/5 5/5 4/5 5/5 5/5
25 HAU CQ95 1/4 4/4 4/4 4/4 4/4
CP83 0/5 5/5 5/5 5/5 5/5
2.5 HAU CQ95 2/5 5/5 5/5 5/5 5/5
CP83 0/5 5/5 5/5 5/5 5/5
250 HA+ ISCOM CQ95 1/5 5/5 4/5 5/5 5/5
CP83 0/5 5/5 5/5 5/5 5/5
25 HA+ ISCOM CQ95 2/5 5/5 3/5 5/5 5/5
CP83 0/5 5/5 4/5 5/5 5/5
250 HAU+ADJV CQ95 5/5 0/5 0/5 5/5 0/5
CP83 3/3c 0/5 0/5 5/5 0/5
a HAU means hemagglutinating units, ISCOM immune stimulating complex and ADJV water-in-oil emulsion adjuvant.b Birds responding with a serum hemagglutination inhibition titer (HI) of greater than 4 at 28 days post vaccination, with number of chickens
responding over total tested.c Two sera were lost.
J. Crawford et al. / Vaccine 17 (1999) 2265±2274 2267
14 and 21 days postvaccination at 10 days postchal-lenge. Virus shedding was measured as outlined below.
2.6.3. Trial 3This trial was designed to evaluate the H7 hemag-
glutinin preparation at two di�erent doses and chal-lenge with two lethal AI viruses that had been isolated60 years apart. Results are presented in Table 3 andFig. 3.
2.6.4. Trial 4The ®nal trial was designed to measure responses to
a bivalent vaccine containing both H7 and H5 hemag-glutinin proteins. It was designed similar to the otherswith control birds compared to two groups receivingeither 0.1 or 1.0 mg of each protein formulated in thewater in oil emulsion. Results are presented in Table 4.
2.7. Serologic evaluation, challenge with lethal AIstrains and virus reisolation from birds
For serology, blood (0.3 ml) was drawn from eachchicken at time points indicated, incubated at 378Cwith 20% sodium citrate (0.75% ®nal concentration)for 1 h, when used to evaluate serum antibody re-sponses to H5 and/or H7 vaccination using the hemag-glutination inhibition (HI) assay [43]. Challenge virusstrain information and administration time points areas indicated in Tables 1±4. Virus challenges were deliv-ered intranasally (IN) using 10,000 times the meanchick lethal dose (CLD50). Ten days postchallenge,remaining live birds were bled for serum HI titerevaluation. Cloacal and oral swabs were taken at 3days postchallenge, freeze±thawed one time, mixed vig-orously, and held at room temperature for 1 h.Samples were clari®ed by centrifugation and inoculatedinto three 10-day-old embryonating eggs per sample.Embryo viability was recorded for one week, afterwhich allantoic ¯uid was collected from remaining liveeggs and assayed for a hemagglutination (HA) titer [43].
Table 2
Response to increasing doses of H5 vaccine
Vaccine dose Number of
HI positive
birds
Average HIa titer postvaccination Number of
survivors
Number of birds
shedding virus at
3 days
postchallenge
14 days PVb 21 days PV postchallenge oral cloacal
None 0/5 0 0 0 0/5 N/A
0.05 mg/bird 1/5 2 0 0 0/5 2/2 2/2
0.20 mg/bird 3/5 2 10.6 96 3/5 3/3 2/3
1.0 mg/bird 5/5 9.6 121.6 484.4 5/5 4/5 1/5
5.0 mg/bird 5/5 224 146 361.2 5/5 1/5 1/5
a Hemagglutination inhibition: average geometric mean titer.b Postvaccination.
Table 3
Protection in 3-week-old White Rock chickens against lethal challenge with Eurasian-origin avian in¯uenza strains using recombinant H7 vaccine
Group Challengea Morbidityb Mortalityc Oral+d Cloacal+e
A, 10 mg Ck/Ger/34 0/5 0/5 4/6 1/6
P/Rob/94 1/4 0/4 3/4 2/4
B, 1 mg Ck/Ger/34 0/6 0/6 2/5 1/5
P/Rob/94 0/3 0/3 2/3 3/3
Control P/Rob/94 4/4 4/4 3/3f 2/3f
a Challenge at 38 days postvaccination (40 days old), using A/chicken/Germany/1934 (H7N2) or A/Peking robin/California/1994 (H7N1).b Number of chickens a�ected over total number of chickens, observed over a 6-day period postchallenge.c Number of chickens a�ected over total number of chickens, observed over a 6-day period postchallenge.d Number of chickens positive for oral virus isolation over total number of chickens. Samples were collected 3 days postchallenge.e Number of chickens positive for cloacal virus isolation over total number of chickens. Samples were collected 3 days postchallenge.f One chicken died prior to the 3 day virus collection point.
J. Crawford et al. / Vaccine 17 (1999) 2265±22742268
3. Results
3.1. Characterization of the HA genes of viral strainsfor cloning in baculovirus
Two isolates representative of currently circulationstrains of H5 and H7 subtypes of avian in¯uenzaviruses were selected based upon sequence analysis ofthe HA gene from extracted viral RNA and phyloge-netic comparison with other isolates [44±46]. The com-plete HA coding sequences of these two strains arepresented in Fig. 1. The HA sequences were comparedwith each other and with the human H3 HA originallyused for HA crystallization. Analysis of thesesequences indicates conservation of all cysteine resi-dues, no unusual sequences or artifacts associated withcloning, and illustrate the exact start and stop pointsof the coding sequences relative to each other. Otherfeatures include conservation of both the fusion pep-tide sequences and previously characterized conservedcarbohydrate addition sites [44]. The sequence of theA/NY/95 (H7) strain is the only complete HAsequence shown. The sequence of the Aichi strain istruncated because of cleavage by bromelain in theoriginal crystallization study. The sequence of the A/Jalisco/94 H5 strain is shorter by design.
3.2. Properties of recombinant A/chicken/Jalisco/94 andA/chicken/New York/95 hemagglutinins
Recombinant A/chicken/Jalisco/94 (H5) and A/chicken/New York/95 (H7) hemagglutinins were
expressed as noncleaved HA (HA0) glycoproteins.Puri®ed recombinant HA0 (rHA0) from AI H5 and H7strains migrate as single major polypeptides of ap-proximately 69,000 molecular weight on SDS PAGE.Western blot analysis of puri®ed H5 and H7 rHA0 isshown in Fig. 2a. Peptide-N-glycosidase F treatmentof both H5 and H7 recombinant hemagglutininsdemonstrated that these rHA0s were glycosylated whenexpressed in the baculovirus system (data not shown).
The HA0 precursor of in¯uenza is a homotrimer,and treatment with trypsin releases a single argininefrom each unit to generate the HA1 and HA2 domains.HA0 monomers, in contrast, are degraded by trypsin,and trypsin sensitivity is used as an assay for trimerformation. Trypsin treatment of puri®ed, recombinantH5 and H7 HAs converts them from HA0 into HA1
and HA2 sized proteins (Fig. 2b), suggesting a trimerformation. Recombinant H5 and H7 proteins aggre-gate into micelle formations, as observed by electronmicroscopy (not shown). In this form the rHA0s canagglutinate chicken red blood cells, demonstratingtheir ability to recognize cellular receptors (data notshown).
3.3. Evaluation of immunogenicity of proteinpreparations
Initially vaccination dosages were based upon HAunits. Increasing HA units of the H5 protein injectedsubcutaneously, without adjuvant, in 1-day-old chick-ens yielded variable HI responses against H5 subtypeviruses (Table 1). Formulations with H5 protein into
Table 4
Response to a combined H5/H7 vaccine
Vaccine dosea Number of HIb
positive birds
Average prechallenge
HI titers:
Challenge virus Average
postchallenge
titers in
survivors
Number
of survivors
H5 H7 14 days
PVc21 days
PV
H5 H7
H5 H7 H5 H7
None 0/10 0/10 0 0 0 0 A/Ck/Que/95 (H5) NA NA 0/5
A/Ck/Ger/34 (H7) NA NA 0/5
0.1 mg/bird each Ð H5 and H7 7/10 9/10 1 12 2 236 A/Ck/Que/95 (H5) 128 1176 4/5
A/Ck/Ger/34 (H7) 16 2304 2/5
1.0 mg/bird each Ð H5 and H7 10/10 10/10 15 195 55 384 A/Ck/Que/95 (H5) 112 672 4/4d
A/Ck/Ger/34 (H7) 48 >4096 4/4
a Delivered in water-in-oil emulsion.b Hemagglutination inhibition.c Postvaccination.d Due to miscommunication only eight of the 10 vaccinates in this group were transferred to high containment for challenge.
J. Crawford et al. / Vaccine 17 (1999) 2265±2274 2269
an ISCOM (immune stimulating complex) had noapparent e�ect on the immunogenicity. The highestconcentration of H5 protein using water-in-oil emul-sion adjuvant protected chickens completely fromlethal challenge. Birds in this group had prechallengeHI titers, with an average GMT of 127. The vacci-nation with 250 HA units also reduced cloacal viruslevels below detectable limits following challenge.After encountering di�culty in reproducibly relatingHA units to quantity of HA (data not shown), it wasdetermined subsequent trials would be based on micro-grams of HA protein per dose.
In Table 2, the e�ective concentration of the recom-binant H5 protein was measured in 3-week-old birdsby increasing dose and measuring survival and shed-ding of challenged vaccinates. Sixty percent protectionfrom death was observed using a 0.2 mg vaccine and100% at 1.0 mg per chicken. At a dose of 1.0 mg perbird, viral shed decreased such that only one chickenof ®ve was positive for AI virus in the cloaca. Titersfrom vaccinated chickens indicated that an optimalimmune response followed doses between 1.0 and 5.0mg of protein per bird.
The H7 HA recombinant protein was evaluated in2-day-old chickens, and the response was e�ective at adose as low as 1.0 mg of HA protein per bird (Fig. 3;Table 3). Individual chicken responses indicated areasonably uniform response at 21 days postvaccina-tion with little change through day 50 (10 days post-challenge). Vaccinated chickens were challenged withtwo di�erent H7 strains, Ck/Ger/34 and P/Rob/94,and both 1.0 mg and 10 mg dosage groups were 100%protected against both strains. With the 1934 Germanstrain, signi®cant reduction in cloacal shedding wasalso noted.
In the ®nal experiment a bivalent H5 and H7 com-bined vaccine using 4-week-old chickens was inocu-lated into two dosage groups, 0.1 or 1.0 mg each HAper bird. The groups were subdivided such that onehalf was challenged with HP H5 AI virus and theother with HP H7 at 21 days postvaccination (Table 4).The chickens responded well to the combined vaccineswith 100% protection against either Ck/Ger/34 or P/Rob/94 at 1.0 mg/bird doses. Titers against the H7both pre- and postchallenge, appeared signi®cantlyhigher than did those against the H5. Additionally,
Fig. 1. Features of the coding sequences of H5 (top line) and H7 (second line) HA baculovirus inserts. The coding sequences of the two inserted
HA genes obtained from A/Ck/Jalisco/94 H5N2 (genBank No. AF071775) and A/Ck/New York/95 H7N1 (genBank No. AF071776) are aligned
and compared to the sequence for the human H3 A/Aichi/68 H3N2 (third line), which is in the Brookhaven protein data base ®les of solved crys-
talline structures. The sequences common to all three are noted on the bottom line and the receptor binding site for the H3 structure is under-
lined. Potential glycosylation sites are boxed.
J. Crawford et al. / Vaccine 17 (1999) 2265±22742270
Fig. 2. (a) SDS PAGE analysis of puri®ed recombinant A/chicken/Jalisco/94 (H5) and A/chicken/New York/95 (H7) hemagglutinins. Puri®ed H5
and H7 rHA0 proteins were separated on an 11.5% SDS PAGE and either stained with Coomassie blue (left panels) or analyzed by the Western
blot with anti-A/Jalisco/94 or anti-A/New York/95 antiserum (right panels). Puri®ed rHA0 protein migrates as a single polypeptide of approxi-
mately 69,000 molecular weight. In insect cells, hemagglutinins of low pathogenic strains are not cleaved into HA1 and HA2, since insect cells
lack the protease needed for this cleavage. (b) Trypsin treatment of puri®ed recombinant A/chicken/Jalisco/94 (H5) and A/chicken/New York/95
(H7) hemagglutinins. The samples were treated with 0.25 to 50 mg/ml TCPK-trypsin or ice for 30 min, separated by 11.5% SDS PAGE and ana-
lyzed by the Western blot with anti-A/Jalisco/94 or anti-A/New York/95 antiserum. The arrows indicate HA0, HA1 and HA2.
J. Crawford et al. / Vaccine 17 (1999) 2265±2274 2271
titers against the H7 and H5 HAs appeared to rise fol-lowing challenge with the heterologous strain. Thismay have been due simply to a natural increase in titerbetween 3 weeks postvaccination and the 10 days fol-lowing challenge.
4. Discussion
Vaccination of poultry is a fairly widespread andfruitful process. Standard virus vaccines routinely inuse in commercial poultry today include those againstNewcastle disease, Fowlpox, Marek's disease,Infectious bronchitis, Infectious bursal disease andInfectious laryngotracheitis. Vaccination against avianin¯uenza has been prohibited in the US because AIposes only regional problems and vaccination inhibitsthe ability to follow natural ®eld infections. The natu-ral infection induces circulating antibodies againstboth the surface glycoproteins and the internal viralNP and M1 proteins. Since these natural ®eld virusescan sometimes become highly virulent, regulatoryagencies understandably wish to keep the backgroundlevels for detection of AI as low as possible. The useof recombinant HA would remove the antibody re-sponse to internal proteins and allow distinctionbetween vaccinates and infected birds. Other recombi-
nant approaches to expression of the AI HA have alsobeen successful [48]. The recent outbreak of highlypathogenic avian in¯uenza in Hong Kong that waslethal in 6 of 18 cases in humans has raised new con-cerns regarding interaction between birds and humans.One mechanism to protect against such a zoonosis isto generate immunity in the animal vector of the dis-ease, such that shedding of the virus into the environ-ment is e�ectively eliminated. In this study, suitablevaccines, satisfying this criterion were readily devel-oped from baculovirus-vectored HA genes expressingfunctional protein in insect cells.
The recombinant in¯uenza HA proteins had hemag-glutinating activity, formed homotrimers thatassembled into rosettes in solution and were immuno-genic. When formulated with water-in-oil adjuvant andinjected at concentrations as low as 0.2 mg/bird, theproteins a�orded 100% protection against challengewith lethal H5 and H7 strains. In the case of the H5challenge virus, a strain sharing greater than 98%identity with the HA sequence of the baculovirus con-struct was used. In the case of the H7, however, nosuitable recent lethal strains were available from NorthAmerica, consequently two Eurasian-origin strainswere used instead. The original `fowl plague' isolate(A/Chick/Germany/34 H7N2) has been repeatedlyshown to be 100% lethal in White Rock chickens with
Fig. 3. Chicken serum antibody responses (average geometric mean titers) to baculovirus-vectored HA protein. Nine 2-day-old White Rock
chickens were each inoculated with 10 or 1 mg of baculovirus origin, puri®ed H7 hemagglutinin protein (A/chicken/New York/1995). At 40 days
postvaccination the birds were challenged with either A/Chick/Germany/34 or A/Pekin Robin/94 (arrow). Control chickens did not survive chal-
lenge; therefore, there was not a 50 day time point for this group. Error bars indicate standard error.
J. Crawford et al. / Vaccine 17 (1999) 2265±22742272
a mean time to death of 24±48 h when inoculatedintranasally. The HA of this virus shares only 84%identity with the A/NY/95 isolate. A second highlypathogenic laboratory Asian isolate (A/pekin robin/CA/94 H7N1) was derived from an isolate obtainedfrom quarantined birds in California [47]. The vacci-nated birds were fully protected from lethal infectionwith both strains. It is thus clear that protectionagainst lethal disease does not necessarily require strin-gent identity between the HA of the vaccine and chal-lenge strains.
In some of the trials reported here, virus sheddingwas e�ectively reduced at three days postchallenge athigher HA concentrations. This time point has beenpreviously shown to be a useful one in determininge�ciency of AI vaccines [37, 48] An ideal AI vaccinewould be one which prevented shedding completelyand these studies suggest that a recombinant HA con-struct sharing signi®cant sequence identity with thechallenge virus would be e�ective.
The two subtypes chosen for the study have beenshown to undergo rapid and unpredictable shifts tohighly pathogenic phenotypes both in nature and inthe laboratory. Routine, widespread vaccination ofpoultry against them would reduce and likely eliminatethe possibility of such a shift. The relative ease ofpreparation of these gene constructs and rapid puri®-cation of the protein suggests that such a strategywould be physically feasible. Although in thesestudies, there were di�erent levels of response to eachsubtype, the results obtained from combining the twoHA subtypes suggested that multivalent vaccines couldbe prepared to protect against multiple subtypes ormultiple antigenic variants within subtypes.Widespread use of these HA vaccines in areas withpersistently circulating avian in¯uenza viruses such asin Minnesota or in central Mexico might ultimatelylead to antigenic drift and emergence of new variants.However, since the vast majority of poultry in NorthAmerica remain free of AI, vaccination with a subunitvaccine such as these, would more likely prevent estab-lishment of endemic infections rather than exacerbatethem.
Acknowledgements
The authors express gratitude for the expert techni-cal assistance of Ms. Joan Beck and Ms. Patsy Decker.We also thank Dr. David Swayne for valuable inputto the research and manuscript.
References
[1] Lamb R.A., Krug R.M. Orthomyxoviridae: the viruses and
their replication. In: Fields B.N., Knipe D.M., Howley P.M.,
