Development of Clade-Specific and Broadly Reactive Live Attenuated Influenza Virus Vaccines against Rapidly Evolving H5 Subtype Viruses

ABSTRACT

We have developed pandemic live attenuated influenza vaccines (pLAIVs) against clade 1 H5N1 viruses on an Ann Arbor cold-adapted (ca) backbone that induced long-term immune memory. In 2015, many human infections caused by a new clade (clade 2.2.1.1) of goose/Guangdong (gs/GD) lineage H5N1 viruses were reported in Egypt, which prompted updating of the H5N1 pLAIV. We explored two strategies to generate suitable pLAIVs. The first approach was to modify the hemagglutinin gene of a highly pathogenic wild-type (wt) clade 2.2.1.1 virus, A/Egypt/N03434/2009 (Egy/09) (H5N1), with its unmodified neuraminidase (NA) gene; this virus was designated Egy/09 ca. The second approach was to select a low-pathogenicity avian influenza H5 virus that elicited antibodies that cross-reacted with a broad range of H5 viruses, including the Egypt H5N1 viruses, and contained a novel NA subtype for humans. We selected the low-pathogenicity A/duck/Hokkaido/69/2000 (H5N3) (dk/Hok/00) virus for this purpose. Both candidate vaccines were attenuated and immunogenic in ferrets, inducing antibodies that neutralized homologous and heterologous H5 viruses with different degrees of cross-reactivity; Egy/09 ca vaccine antisera were more specific for the gs/GD lineage viruses but did not neutralize recent North American isolates (clade 2.3.4.4), whereas antisera from dk/Hok/69 ca-vaccinated ferrets cross-reacted with clade 2.3.4.4 and 2.2.1 viruses but not clade 1 or 2.1 viruses. When vaccinated ferrets were challenged with homologous and heterologous H5 viruses, challenge virus replication was reduced in the respiratory tract. Thus, the two H5 pLAIV candidates are suitable for clinical development to protect humans from infection with different clades of H5 viruses.

IMPORTANCE In response to the continuing evolution of H5N1 avian influenza viruses and human infections, new candidate H5 live attenuated vaccines were developed by using two different approaches: one targeted a specific circulating strain in Egypt, and the other was based on a virus that elicits broadly cross-reactive antibodies against a wide range of H5 viruses. Both candidate vaccines were immunogenic and exhibited protective efficacy in ferrets. Our study permits a comparison of the two approaches, and the data support the further development of both vaccine viruses to optimally prepare for the further spread of clade 2.2.1 or 2.3.4.4 viruses.

INTRODUCTION

Since the first emergence of the H5N1 highly pathogenic avian influenza (HPAI) virus (HPAIV) in poultry in Hong Kong in 1997, viruses of this lineage have continued to evolve in avian species (1). Based on phylogenetic analysis of 1,637 H5 hemagglutinin (HA) sequences, 12 new clades were designated (http://www.who.int/influenza/gisrs_laboratory/h5n1_nomenclature/en/). Viruses that belong to clade 2.2.1 were isolated mostly in Egypt and have been prevalent in poultry in this area since 2006; 346 human cases, including 77 deaths, were reported between 2006 and 2015, and 136 cases, including 39 deaths, were reported in 2015 alone in Egypt (http://www.who.int/influenza/human_animal_interface/2016_07_19_tableH5N1.pdf?ua=1). The Egypt isolates further evolved into two clades, from 2.2.1 to 2.2.1.1 and 2.2.1.2, since 2008 (2–4). Viruses from Egypt also acquired the ability to bind α2,6-linked (human-type) sialic acid receptors, raising concerns about their increased risk of transmission to humans (5). In addition to the evolution of H5 HA, extensive reassortment events have occurred with AIVs of different subtypes in birds in China, resulting in new H5 viruses paired with different neuraminidase (NA) subtypes (H5N2, H5N5, H5N6, and H5N8) and internal protein genes (6–9).

An H5N8 HPAIV was detected in China in 2010 and caused outbreaks in poultry in South Korea and Japan in 2014 (7, 10–12). The HA gene of this virus belongs to clade 2.3.4.4 of HPAI goose/Guangdong (gs/GD) lineage H5N1 viruses. The H5N8 HPAIV spread to Europe, possibly by migratory birds (13, 14), and was detected in North America in wild birds as well as in commercial poultry in the United States in December 2014 (15). In addition, reassortment between the gs/GD lineage H5N8 virus and North American-origin viruses resulted in the emergence of H5N1 and H5N2 viruses in the United States (15, 16).

We previously produced two H5N1 pandemic live attenuated influenza virus vaccines (pLAIVs) bearing HA and NA of early H5N1 isolates (A/Hong Kong/213/2003 and A/Vietnam/1203/2004 [Viet/1203/04]) and the remaining gene segments from the cold-adapted (ca) A/Ann Arbor/6/60 (AA ca) virus vaccine donor strain. These vaccines were tested in animal models as well as in human volunteers (17–19). Although initial phase 1 clinical studies showed that the H5N1 pLAIVs replicated poorly and failed to elicit serum neutralizing antibody (Ab) responses in humans (17), the subsequent administration of an inactivated H5N1 vaccine elicited rapid, robust, and high-quality serum antibody responses, indicating that the pLAIV had induced long-term immune memory (20). This clinical observation was confirmed with H7N7 and H7N9 pLAIVs (21, 22). Based on this knowledge, we sought to develop additional H5 pLAIVs that could be used to protect against the fourth-order clades that have emerged in Egypt since 2015.

In this study, we explored two different approaches to develop new H5 pLAIVs on the AA ca platform: the conventional approach of selecting a virus from the clade of interest and an exploratory approach of selecting a virus outside the gs/GD lineage that would induce cross-reactive antibodies and cross-protection against a wider range of H5 viruses.

RESULTS

Selection of viruses for vaccine development.

We selected the human isolate A/Egypt/N03434/2009 (Egy/09) as a donor of the HA and NA genes for a new H5N1 pLAIV that specifically targeted viruses that are a concern in Egypt because they are associated with a large number of human infections and have acquired α2,6-linked sialic acid binding activity (5). As an alternative strategy, we also sought to identify a virus that would induce broadly cross-reactive antibody responses by selecting 9 low-pathogenicity (LP) H5 viruses on the basis of their phylogenetic relationships and geographic locations and generating antisera in mice and ferrets. These LPAIVs bore H5 HAs paired with N1, N2, N3, N5, N8, and N9 NA subtypes that are novel to humans (Table 1). We evaluated the kinetics of replication of the viruses in the respiratory tract of mice in order to establish an animal model for the subsequent evaluation of vaccine efficacy. These data are summarized in Fig. 1. With the exception of A/chicken/Hidalgo/28159-232/1994 (ck/Hida/94), the 9 LP H5 viruses replicated in the upper respiratory tracts of most or all of the mice in each group. Replication of the 9 LP H5 viruses was more consistent in the lungs than in the nasal turbinates (NTs) of mice.

TABLE 1.

Viruses used for this study

Virusa	Subtype	Abbreviation	Source
A/chicken/Pennsylvania/1983*	H5N2	ck/PA/83	South East Poultry Research Laboratory, USDA, Athens, GA
A/chicken/Hidalgo/28159-232/1994*	H5N2	ck/Hida/94
A/chicken/Queretaro/14588-19/1995*	H5N2	ck/Que/95
A/duck/Singapore/F119/3/1997*	H5N3	dk/Sing/97
A/mallard/MN/105/2000*	H5N5	mal/MN/02
A/duck/NY/191255-59/2002*	H5N8	dk/NY/02
A/emu/NY/12716/94*	H5N9	emu/NY/94
A/duck/Potsdam/1402-6/1986*	H5N2	dk/Pots/86	St Jude Children's Research Hospital, Memphis TN
A/Hong Kong/483/97	H5N1	HK/483/97	Centers for Disease Control and Prevention, Atlanta, GA
A/Vietnam/1203/2004	H5N1	Viet/1203/04
A/Indonesia/05/2005	H5N1	Indo/05/05
A/American Green Winged-Teal/WA/195750/2014	H5N1	AGW/WA/14	National Veterinary Services Laboratories, APHIS, USDA, Ames, IA
A/Northern pintail/WA/40964/2014	H5N2	NP/WA/14
A/gyrfalcon/WA/40188-6/2014	H5N8	gyrfal/WA/14
A/chicken/Egypt/1553-1/2010	H5N1	ck/Egy/10	Istituto Zooprofilattico Sperimentale delle Venezie, Italy
A/duck/Hokkaido/69/2000*	H5N3	dk/Hok/00	Hokkaido University, Japan
A/duck/Hokkaido/84/2002*	H5N3	dk/Hok/02
A/mallard/Sweden/52/2002*	H5N9	mal/Swe/02	Erasmus University Medical Center (EMC), Netherlands
A/chicken/Netherlands/EMC/2014	H5N8	ck/NL/14

^aAsterisks indicate viruses used for the selection of vaccine candidates.

FIG 1.

Replication kinetics of H5 influenza viruses in mice following i.n. administration of 10⁶ TCID₅₀ of virus in 50 μl. Virus titers in the nasal turbinates (A) and the lungs (B) of 5 mice per group sacrificed on days 2, 3, 4, and 7 postinfection are expressed as log₁₀ TCID₅₀ per gram of tissue. The dashed line indicates the lower limit of detection.

Generation of postinfection sera in mice and ferrets.

Postinfection sera were generated for the 9 LP H5 viruses in mice and ferrets. Two- to fourfold-higher titers were observed when horse erythrocytes were used instead of turkey erythrocytes in the hemagglutination inhibition (HAI) assay. Most of the H5 viruses induced HAI antibodies by 49 days postinfection (p.i.) in mice and by 27 days p.i. in ferrets (data not shown). Neutralizing antibody titers induced by the 9 viruses in both animal models are presented in Table 2. Although high neutralizing antibody titers were detected in sera of mice, they did not necessarily correlate with the titers of virus in the lungs. In particular, the neutralizing antibody titer was high in mice inoculated with ck/Hida/94, despite the fact that we did not detect viral replication in the NTs and titers in the lungs were low or not detected (Fig. 1 and Table 2).

TABLE 2.

Titers of cross-neutralizing antibodies in postinfection ferret and mouse antisera induced by infection with H5 viruses^a

Virus	GMT of cross-neutralizing antibodies in ferret sera (n = 2) induced by:
	dk/NY/02		mal/MN/02		emu/NY/94		ck/Hida/94		dk/Hok/02		dk/Hok/00		dk/Pots/86		mal/Swe/02		dk/Sing/97
	F	M	F	M	F	M	F	M	F	M	F	M	F	M	F	M	F	M
dk/NY/02	508	425	27	320	127	320	201	452	320	508	806	906	320	806	226	320	201	320
mal/MN/02	508	201	201	254	201	40	31	508	160	320	160	452	100	201	127	160	63	25
emu/NY/94	31	≤10	≤10	≤10	160	226	≤10	28	≤10	28	113	31	28	20	31	40	≤10	≤10
ck/Hida/94	226	254	50	452	40	56	254	806	127	508	403	1,280	201	640	201	452	100	403
dk/Hok/02	452	508	127	254	113	31	113	452	160	452	226	640	226	452	226	226	127	226
dk/Hok/00	403	508	113	403	160	40	127	806	160	906	640	1,015	320	640	508	320	160	320
dk/Pots/86	640	508	160	320	201	113	226	905	226	806	452	1,280	254	1,612	640	508	201	905
mal/Swe/02	403	905	113	640	226	113	160	1,280	254	806	905	1,810	640	1,015	508	508	113	806
dk/Sing/97	452	254	80	226	63	56	160	320	201	508	640	1,015	403	806	452	320	226	508

^aNumbers in boldface type indicate homologous titers. Underlining indicates titers that are ≥4-fold lower than the homologous titer. F, ferret; M, mouse.

All of the animals survived infection when 10⁶ 50% tissue culture infective doses (TCID₅₀) and 10⁷ TCID₅₀ of each virus were administered to mice and ferrets, respectively.

Cross-reactivity of the neutralizing antibody response in mouse and ferret sera.

To identify a virus that elicited a broadly cross-reactive neutralizing antibody response, postinfection sera induced in ferrets and mice were tested against the 9 viruses in a checkerboard fashion (Table 2). A neutralizing antibody titer that was within 4-fold of the homologous antibody titer was considered to be indicative of cross-reactivity. In mice, A/duck/Hokkaido/69/2000 (H5N3) (dk/Hok/00) elicited a high homologous antibody titer, and the antiserum neutralized all of the viruses except A/emu/NY/12716/94 (emu/NY/94) (Table 2). In ferrets, dk/Hok/00 elicited antibody that neutralized all of the viruses except emu/NY/94 and A/mallard/MN/105/2000 (mal/MN/02) (Table 2). Interestingly, antibodies induced by all 8 heterologous viruses in both mice and ferrets were poorly cross-reactive with the emu/NY/94 virus.

Cross-reactive neutralizing antibodies against highly pathogenic avian influenza viruses.

To determine whether the antisera elicited by the LPAI H5 viruses cross-reacted with HPAI viruses, we tested them in vitro against 5 HPAI viruses, Viet/1203/04 (H5N1), A/Indonesia/05/2005 (Indo/05/05) (H5N1), A/Hong Kong/483/97 (H5N1) (HK/483/97), A/chicken/Pennsylvania/1983 (H5N2) (ck/PA/83), and A/ck/Queretaro/14588-19/95 (H5N2) (ck/Que/95) (Table 3). Cross-reactive neutralizing activity against ck/Que/95 (H5N2) was detected in all of the postinfection ferret antisera, with the highest titers being found in sera from ferrets infected with dk/Hok/00. However, none of the sera cross-reacted with the HPAIV H5 human isolates from Indonesia or Vietnam.

TABLE 3.

Titers of cross-neutralizing antibodies tested against H5 HPAIVs induced by infection with H5 LPAIVs in ferrets

Virus	GMT of cross-neutralizing antibodies in ferret sera (n = 2) induced bya:
	dk/NY/02	malMN/02	emu/NY/94	ck/Hida/94	dk/Hok/02	dk/Hok/00	dk/Pots/86	mal/Swe/02	dk/Sing/97
Viet/1203/04 (H5N1)	≤10	≤10	≤10	≤10	≤10	≤10	≤10	≤10	≤10
Indo/05/05 (H5N1)	≤10	≤10	≤10	≤10	≤10	≤10	≤10	≤10	≤10
HK/483/97 (H5N1)	40	≤10	≤10	≤10	64	40	20	50	57
ck/PA/83 (H5N2)	57	≤10	≤10	28	113	57	40	80	80
ck/Que/95 (H5N2)	254	113	50	160	320	640	100	127	80
Homologous virus	425	254	160	254	452	905	640	80	403

^aNumbers in boldface type indicate homologous titers.

Generation of reassortant H5 ca candidate vaccine viruses.

Reassortant 6:2 H5 ca viruses were generated by reverse genetics with the six internal protein gene segments from AA ca, the master donor virus that confers the characteristic temperature-sensitive (ts), ca, and attenuation (att) phenotypes, and the HA and NA gene segments from the respective H5 wild-type (wt) viruses. The HA gene from the HPAI Egy/09 virus was modified to remove the multiple basic amino acid cleavage motif. The 6:2 ca Egy/09 virus bearing HA and NA genes synthesized based on reported sequences grew poorly in embryonated hen eggs (<10⁷ TCID₅₀/ml). In order to identify mutations that would be associated with improved yields, the virus was passaged several times in Madin-Darby canine kidney (MDCK) cells, and large plaques were isolated for sequence analysis. A number of mutations were identified in the HA (D94G, P194Q, R208K, and A238T) and NA (E145G) genes. These mutations were introduced into the plasmids in different combinations, and new 6:2 variant viruses on the AA ca backbone were generated and evaluated for their titers in eggs and reactivity with antisera raised against the parental Egy/09 ca virus. Finally, the variant with a minimal sequence change in HA (P194Q) that reached a titer of approximately 10^8.5 TCID₅₀/ml was selected for further characterization as a vaccine candidate. This amino acid change did not affect viral antigenicity, as confirmed by microneutralization (MN) assays using ferret antisera, although Egy/09 ca P194Q immunized ferret serum had lower reactivity to the Egy/09 ca parent virus (Table 4). Ferrets immunized with Egy/09 ca with the P194Q change had similar neutralizing antibody titers against the Egy/09 ca parent and the P194Q variant. The second dose boosted antibody titers. The 6:2 dk/Hok/00 ca virus with the HA and NA sequences derived from the dk/Hok/00 wt virus generated by reverse transcription-PCR (RT-PCR) from viral RNA (vRNA) reached a titer of 10^7.9 TCID₅₀/ml; this virus may require further adaptation to reach a higher titer.

TABLE 4.

The P194Q ca virus is antigenically similar to the Egy/09 ca parent virus^a

Expt	Method	Virus	Antibody titer in postinfection antiserum raised against:
			Egy/09 ca parent			Egy/09 ca P194Q
			Ferret 1	Ferret 2	Ferret 3	Ferret 1	Ferret 2	Ferret 3
1	HAI	Egy/09 ca parent	8	16	32	32	32	16
		Egy/09 ca P194Q	32	32	64	128	256	128
2	MN	Egy/09 ca parent	Not done	Not done	Not done	320	640	1,280
		Egy/09 ca P194Q	Not done	Not done	Not done	320	640	1,280

^aSera from ferrets vaccinated with two doses of the indicated viruses at day 42 or day 56 postinfection were tested in separate experiments by HAI or microneutralization, respectively. Homologous titers are indicated in boldface type. The starting dilution for both assays was 1:10.

In vitro characterization of vaccine viruses.

Since the human Egy/09 wt virus isolate was not available, we used the ck/Egy/00 wt virus for comparison with the Egy/09 ca virus. The amino acid identity between the Egy/09 and A/chicken/Egypt/1553-1/2010 (ck/Egy/10) HA genes is 96.9%. The dk/Hok/00 and ck/Egy/10 wt viruses replicated equally well at 39°C and 33°C. Both the dk/Hok/00 ca and Egy/09 ca viruses were restricted in replication at 39°C, with a 1,000-fold reduction in titers compared to the titer at 33°C, confirming the ts phenotype. Both ca viruses replicated more efficiently than the H5 wt counterparts in incubators set at 25°C; the titers of the ca reassortant viruses at the lower temperature were within 100-fold of the titer at 33°C, while the H5 wt viruses were restricted in replication at the lower temperature. Thus, both H5 ca viruses displayed the expected ts and ca phenotypes (Table 5).

TABLE 5.

H5 ca viruses exhibit ts and ca phenotypes^a

Virus	Mean virus titer (log10 TCID50/ml) ± SD at temp			Presence of phenotype
	39°C	33°C	25°C	tsb	cac
dk/Hok/00 wt	7.3 ± 0.15	7.3 ± 0.17	3.1 ± 0.09	−	−
dk/Hok/00 ca	4.3 ± 0.15	8.0 ± 0.17	7.5 ± 0.13	+	+
ck/Egy/10	8.3 ± 0.16	7.2 ± 0.13	5.0 ± 0.2	−	−
Egy/09 ca	5.3 ± 0.21	8.7 ± 0.2	7.5 ± 0.12	+	+

^aCEK cells were inoculated with H5 wt and ca viruses and incubated in incubators at 39°C, 33°C, and 25°C. After incubation for 6 days at 39°C and 33°C or 10 days at 25°C to reach peak titers, cells were examined for cytopathic effects, and virus titers at each temperature were determined.

^bts, difference between the mean titer at 33°C and that at 39°C of ≥100-fold.

^cca, difference between the mean titer at 33°C and that at 25°C of ≤100-fold.

Attenuation of the Egy/09 ca and dk/Hok/00 ca viruses in ferrets.

The ck/Egy/10 wt virus replicated efficiently in the NTs and modestly in the lungs of ferrets (Fig. 2). Replication of the Egy/09 ca vaccine virus was detected in the NTs of all animals on days 1 and 3 when titers of the samples were determined in eggs, although replication was not detected in MDCK cells. In contrast, viral replication was not detected in the lungs of any of the ferrets infected with the Egypt/09 ca virus at any time point. Similarly, the dk/Hok/00 wt virus replicated in both NTs and lungs, while replication of the dk/Hok/00 ca vaccine virus was detected in the NTs but not in the lungs. Histopathological examination revealed significant inflammation, including necrosis, in the lungs of ferrets infected with the wt viruses, but this was absent in ferrets infected with either ca vaccine virus (Fig. 3). Thus, both the Egypt/09 ca and dk/Hok/00 ca vaccine viruses demonstrated the desired attenuation phenotype in ferrets.

FIG 2.

Replication of candidate ca vaccine viruses and the corresponding wt viruses in the respiratory tract of ferrets. Ferrets were inoculated intranasally with 10⁷ TCID₅₀ of each virus in a volume of 1 ml, and tissues were harvested on days 1, 3, and 5 postinfection. Virus titers are expressed as log₁₀ TCID₅₀ per gram. For Egypt viruses, titers were also determined in eggs and are expressed as EID₅₀ per gram. The dashed horizontal line represents the lower limit of detection (<10^1.5 TCID₅₀/g or EID₅₀/g). Shown are virus titers in nasal turbinates (circles) or lungs (squares) from ferrets infected with ck/Egy/10 wt (A), Egy/09 ca (B), dk/Hok/00 wt (C), or dk/Hok/00 ca (D).

FIG 3.

Images of hematoxylin-and-eosin-stained lung sections collected following infection with the dk/Hok/00 (A) and Egy/10 (B) wt viruses, compared to the dk/Hok/00 (C) and Egy/09 (D) ca viruses. Stars highlight areas of intense airway-associated (intrabronchiolar) inflammation and necrosis. All images were taken on day 5 postinfection at a ×20 magnification. The black bar is equivalent to 50 μm on images (original magnification, ×20).

Immunogenicity in ferrets.

One dose of the Egy/09 ca vaccine failed to elicit a detectable serum neutralizing antibody response 27 days after immunization in this study. However, 2 weeks after the second dose of the vaccine, a neutralizing antibody response was detected (geometric mean titer [GMT] of 1:259) against the ck/Egy/10 wt virus, which belongs to the same clade (2.2.1) as that of the source virus for the sequences of HA and NA of the Egy/09 ca virus (Table 6). The sera also cross-reacted well with Indo/05/05 (clade 2.1) but did not react well with Viet/1203/04 (clade 1). Cross-reactivity with North American lineage H5 viruses, ck/PA/83 (H5N2) or Eurasian lineage clade 2.3.4.4 H5 viruses recently isolated in North America, was absent or barely detectable.

TABLE 6.

Serum neutralizing antibodies against homologous and heterologous H5 viruses in ferrets immunized with the Egy/09 ca and dk/Hok/00 ca vaccines

Vaccine virus	Days after 1st dose (days after 2nd dose)	GMT (95% CI)a
		ck/Egy/10 (H5N1)	Indo/05/05 (H5N1)	Viet/1203/04 (H5N1)	ck/PA/83 (H5N2)	AGW/WA/14 (H5N1)	NP/WA/14 (H5N2)	gyrfal/WA/14 (H5N8)	dk/Hok/00 (H5N3)
Egy/09 ca	14	≤10	≤10	ND	≤10	ND	ND	ND	≤10
	27b	≤10	≤10	ND	≤10	ND	ND	ND	≤10
	42 (14)	259 (167–398)	160 (79–324)	23 (16–32)	58 (18–185)	20 (NA)	≤10	≤10	95 (37–247)
	52 (24)	206 (112–379)	58 (28–119)	32 (NA)	28 (13–62)	≤10	≤10	≤10	44 (19–101)
dk/Hok/00 ca	14	≤10	≤10	ND	≤10	ND	ND	ND	<10
	27b	≤10	≤10	ND	18 (14–24)	ND	ND	ND	92 (55–153)
	42 (14)	93 (55–159)	≤10	≤10	115 (63–210)	29 (16–53)	72 (37–142)	65 (33–128)	581 (334–1,013)
	52 (24)	90 (57–142)	≤10	≤10	70 (27–184)	25 (21–30)	64 (27–148)	51 (25–106)	411 (181–933)

^aShown are geometric mean titers (95% confidence intervals [CI]) of data from 6 ferrets. ND, not done; NA, not applicable (only one animal had a measurable titer of ≥10). Prevaccination sera had no detectable antibody.

^bOne day before the second dose was administered.

In contrast, one dose of the dk/Hok/00 ca vaccine elicited neutralizing antibody against the homologous wt virus, and the titer increased significantly with the second dose. In addition, following the boost, cross-reactivity against the North American lineage H5 virus ck/PA/83 (H5N2) and some reactivity with different clades of gs/GD lineage H5 viruses, including 2014 North American isolates, but not Viet/1203/04 (clade 1) or Indo/05/05 (clade 2.1) viruses, were noted.

Protective efficacy of the dk/Hok/00 ca and Egy/09 ca viruses in ferrets.

Ferrets immunized with 2 doses of the Egy/09 ca or dk/Hok/00 ca vaccine were challenged with homologous and heterologous wt H5 viruses. Since several wt H5 viruses do not cause serious signs of disease and are not lethal for ferrets, protective efficacy was evaluated by the reduction in challenge virus titers in the respiratory tract at 3 days postchallenge compared to the titers in mock-vaccinated animals. Statistically significant reductions in virus titers were seen for ck/Egy/10, Indo/05/05, and ck/PA/83 in the NTs of ferrets immunized with Egy/09 ca (Fig. 4A). Because the dk/Hok/00 wt virus was not detected in the lungs of mock-vaccinated ferrets on day 3 postchallenge, we were unable to assess protection from dk/Hok/00 wt virus challenge in the lower respiratory tract.

FIG 4.

Replication of challenge viruses in the respiratory tracts of ferrets. Groups of ferrets were mock vaccinated (L-15) or immunized intranasally with two doses of 10⁷ TCID₅₀ of the Egy/09 ca or dk/Hok/00 ca virus and challenged with 4 different H5 viruses on day 56 postvaccination. Tissues were harvested 3 days later. (A) Virus titers in tissues are expressed as log₁₀ TCID₅₀ per gram. The dashed horizontal line represents the lower limit of detection (<10^1.5 TCID₅₀/g). (B) Virus titers in nasal turbinates and lungs of ferrets immunized with L-15 or with the dk/Hok/00 ca vaccine virus. Asterisks indicate the statistical significance (P < 0.05) of virus titers in tissues of vaccinated versus mock-vaccinated animals (data were analyzed by Student's t test using Prism 6 software [GraphPad Software Inc.]).

A similar pattern was seen in ferrets immunized with the dk/Hok/00 ca vaccine; viral titers in the NTs of vaccinated ferrets challenged with the ck/Egy/10, Indo/05/05, ck/PA/83, and dk/Hok/00 wt viruses and titers in the lungs of ferrets challenged with ck/PA/83 were significantly lower than those in mock-immunized animals (Fig. 4B). However, there was no statistically significant correlation between serum neutralizing antibody titers and challenge virus titers in the tissues, with the exception that the titer of the dk/Hok/00 virus in the NTs was significantly lower in animals with high serum neutralizing Ab titers (r² = −0.87; P = 0.002) (data not shown).

DISCUSSION

Since its emergence in 1997, the gs/GD lineage of H5 HPAIV has posed a threat to global public health, with more than 800 human infections being reported over nearly 20 years, although the areas affected and the numbers of infected people vary by year (http://www.who.int/influenza/human_animal_interface/Influenza_Summary_IRA_HA_interface_07_19_2016.pdf?ua=1). Moreover, the gs/GD H5 viruses have evolved genetically and antigenically and diversified into many different clades and subclades (1). There is limited antigenic cross-reactivity among viruses that belong to different clades or subclades; thus, the WHO has recommended that vaccine candidates be generated to keep pace with the rapid evolution of these viruses in nature (http://www.who.int/influenza/vaccines/virus/candidates_reagents/summary_a_h5n1_cvv_20160229.pdf?ua=1).

The large number of human infections with H5N1 viruses in Egypt in recent years is a great concern (http://www.emro.who.int/egy/egypt-news/upsurge-h5n1-human-poultry-cases-may-2015.html). A genetically distinct cluster within clade 2.2.1.2 emerged in 2014 and was associated with an increase in the number of human cases (2). However, no marked difference in antigenicity or receptor preference was noted in this group of viruses compared to viruses isolated prior to 2014 (4). The Egy/09 ca vaccine that we developed contains the HA and NA gene sequences of a human isolate from 2009. The vaccine was immunogenic and protective against a 2010 wt isolate from the same clade (2.2.1), and the lack of antigenic change in recently circulating viruses in Egypt suggests that the vaccine virus should be effective against them as well. The serum antibody response induced by the Egy/09 ca vaccine cross-neutralized Indo/05/05 (clade 2.1.3.2) but showed low cross-reactivity with other clades, such as Viet/1203/04 (clade 1), recent North American isolates (clade 2.3.4.4), and a North American lineage H5N2 virus.

An influenza vaccine is likely to provide the best protection if it is generated from the circulating strain. However, the processes of production and distribution are time-consuming, so a strain-specific vaccine may not be ready for timely distribution, as was the case for the 2009 H1N1 pandemic (23). Therefore, we explored a parallel approach to develop an H5 vaccine that would elicit broad reactivity across a range of H5 viruses. We have used a similar approach to select candidate vaccine viruses among animal influenza A viruses of different subtypes, including H2, H3, H6, and H7 (24–27). dk/Hok/00 (H5N3), an LPAI Eurasian lineage virus, was selected based on its ability to elicit a robust homologous HAI titer in both mice and ferrets and the observation that the antisera cross-reacted with the largest number of other H5 viruses (Table 2). This virus was reported to react with a panel of monoclonal antibodies raised against a North American lineage H5 virus in an enzyme-linked immunosorbent assay (ELISA) and reacted well with polyclonal antibodies elicited by a Eurasian lineage H5 virus in a neutralization assay (28). Since the virus is of low pathogenicity for chickens, genetic modification to remove the sequence motif associated with the virulence of highly pathogenic H5 viruses was not necessary. The presence of an N3 NA gene, which has not circulated in humans, may reduce the potential inhibition of replication of the pLAIV by preexisting antibody against N1 or N2 NAs. The replication of one North American and two gs/GD lineage challenge viruses, Egy/09 and Indo/05, was greatly reduced in the respiratory tracts of ferrets immunized with the dk/Hok/00 ca vaccine. Sera from ferrets immunized with the dk/Hok/00 ca vaccine cross-neutralized North American lineage H5 viruses and clade 2.3.4.4 gs/GD lineage H5 viruses that emerged in North America in 2014 but not earlier clades of gs/GD lineage H5 viruses (Table 6). However, it is not known whether and how this in vitro cross-neutralizing activity would correlate with protection from viral infection The antigenic sites of clade 2 H5 viruses correspond to the antigenic sites of H1N1 viruses (29). We aligned the amino acid sequences of the antigenic sites of the Egy/09 and dk/Hok/00 HAs with those of the viruses used in challenge and serological assays to understand why the dk/Hok/00 ca virus provided greater cross-reactivity and cross-protection than the Egy/09 ca vaccine against the 2014 North American H5 isolates. Unfortunately, the amino acid alignment did not provide any insights into the basis for the differences in cross-reactivity, consistent with the conclusion of Shore et al. that sequence identity across the surface of the HA protein is an insufficient measure for determining antigenic relatedness (29). It should be noted that although LAIVs are expected to elicit serum and mucosal antibodies, as well as cellular immune responses, a reliable correlate of protection of LAIV has not been established. In mice and ferrets, protection against heterologous challenge viruses is often seen with LAIVs (30, 31).

The first H5 pLAIV on the Russian A/Leningrad/134/17/1957 ca virus backbone also used HA from a low-pathogenicity AIV, A/duck/Potsdam/1402-6/86 (H5N2), for the generation of vaccines against H5N1 viruses. This vaccine provided cross-protection against antigenically distinct, highly pathogenic H5N1 viruses in mice (32, 33), and cross-reactive antibody against highly pathogenic H5N1 virus was induced in humans (34). More recently, a computationally optimized broadly reactive antigen (COBRA) technology was applied to overcome the diversity of different clades of H5N1 viruses. Virus-like particles (VLPs) containing COBRA HA based on clade 2 H5N1 viruses induced protective levels of HAI antibodies against representatives of each subclade of clade 2 viruses, and animals immunized with clade 2 COBRA VLPs were completely protected from challenge infection with a clade 2.2 H5N1 virus (35, 36). Recently, it was also shown that a cocktail of H5N1 COBRA VLP vaccines based on clade 2 and 3 viruses elicited HAI antibody titers against viruses from multiple H5N1 virus clades (37).

In summary, we developed two candidate H5 pLAIVs using two different approaches: directly targeting the currently circulating enzootic strain in Egypt and selecting a virus that elicits broadly cross-reactive antibodies against a wide range of H5 viruses. Both vaccines were immunogenic and exhibited protective efficacy in ferrets. Although neither vaccine provided complete protection across all clades or lineages of H5 viruses, our study permits a comparison of the two approaches, and it is noteworthy that the dk/Hok/00 ca vaccine elicited antibodies that cross-reacted with the 2014 North American H5 viruses derived from the gs/GD lineage, while the Egy/09 ca virus did not. The data from ferrets support the further development of both vaccine viruses to optimally prepare for the further spread of clade 2.2.1 or 2.3.4.4 viruses. In addition to the selection of an appropriate vaccine virus, there may be ways to optimize and improve the outcome of vaccination by combining vaccine platforms in a “prime-boost” strategy, with priming with a live attenuated, DNA, or adenovirus vaccine followed by a boost with an inactivated vaccine (22, 38–40).

MATERIALS AND METHODS

Viruses and cells.

The H5 viruses used in this study are listed in Table 1. Virus stocks were propagated in the allantoic cavities of 10- to 11-day-old embryonated specific-pathogen-free hen eggs (Charles River Laboratories, North Franklin, CT) at 35°C. The allantoic fluid was harvested at 72 h postinoculation and tested for hemagglutinating activity using 1.0% horse red blood cells or 0.5% turkey red blood cells (Lampire Biological Laboratories, Pipersville, PA). Allantoic fluid was pooled, aliquoted, and stored at −80°C until use. Virus titers were determined in MDCK cells or chicken embryo kidney (CEK) cells (ATCC, Manassas, VA) as described previously (19).

MDCK, CEK, and 293T cells were maintained in modified Eagle's medium with 10% fetal calf serum and l-glutamine. All HPAI H5 viruses were handled by personnel approved by the NIH Biosurety Program and Select Agent Program in enhanced biosafety level 3 (BSL-3) containment laboratories approved for use by the U.S. Department of Agriculture (USDA) and the Centers for Disease Control and Prevention (CDC).

Animals.

BALB/c mice (4 to 6 week olds and female) were purchased from Taconic Farms (Germantown, NY), and 4- to 6-month-old female ferrets were purchased from Triple F Farm (Sayre, PA) or obtained from Simonsen Laboratory (Gilroy, CA). All animal experiments with HPAI H5 viruses were conducted in animal BSL-3 (ABSL-3) containment laboratories approved for use by the USDA and the CDC. The generation of mouse and ferret antisera and evaluation of the kinetics of replication of low-pathogenicity H5 viruses were done in animal BSL-2 laboratories at the NIH in compliance with the guidelines of the Institutional Animal Care and Use Committee.

Viral titration.

Harvested tissues were homogenized in Leibovitz-15 (L-15) medium at 10% (wt/vol) for lung or 5% (wt/vol) for nasal turbinate samples and clarified by centrifugation at 2,500 rpm for 10 min. Clarified homogenates were titrated on MDCK cell monolayers. The virus titer in the lungs of ferrets was also determined in 10- to 11-day-old embryonated hen's eggs since they are a more sensitive substrate for the titration of virus from lung tissue homogenates.

Generation of mouse antisera against H5 viruses.

To generate mouse antisera, groups of five 4- to 6-week-old female BALB/c mice were lightly anesthetized and infected intranasally (i.n.) with 10⁵ TCID₅₀ of each H5 virus in 50 μl. Mice were monitored for the development of disease and mortality. All of the mice were bled before virus administration, and postinfection sera were collected at one or more of the following time points: 27, 34, and 43 days p.i.

Generation of ferret antisera against H5 viruses.

Two 4- to 5-month-old female ferrets that were seronegative for antibodies to currently circulating H3N2 and H1N1 human influenza viruses were used to generate antiserum for each virus. Ferrets were anesthetized and infected i.n. with 500 μl of each H5 virus containing 10⁷ TCID₅₀.

Ferrets were monitored daily for clinical signs of infection, and body temperatures were recorded twice daily. Sera were collected at 21 and 27 days p.i.

Generation of H5N1 and H5N3 ca reassortant viruses by reverse genetics.

The HA and NA genes of Egy/09 were synthesized based on previously reported sequences (DNA accession numbers 184824 and 184825) and cloned into the plasmid vector pAD3000. This plasmid is a derivative of plasmid pHW2000 that contains polymerase I (Pol I) and Pol II promoters for the expression of vRNA and mRNA, respectively, from the full-length viral gene segment insert (41). One amino acid substitution (P194Q) was introduced into the HA gene to enhance the growth of Egy/09 ca in embryonated hen's eggs. A 6:2 reassortant vaccine virus was generated by cotransfecting the plasmids encoding HA and NA of Egy/09 and the 6 internal protein genes of AA ca into cocultured 293T and MDCK cells. At 3 to 5 days posttransfection, the transfected cell supernatant was inoculated into the allantoic cavities of 10- to 11-day-old embryonated hen's eggs (Charles River Laboratories, Franklin, CT) and incubated at 33°C for 3 days. The HA and NA sequences of the rescued virus were verified by sequencing of cDNA amplified from viral RNA by RT-PCR.

The HA and NA gene segments of dk/Hok/00 were amplified from viral RNA by RT-PCR using primers that are universal to the influenza A virus HA and NA genes and cloned into the plasmid vector pAD3000 without further modification. The generation of the 6:2 reassortant dk/Hok/00 ca virus was carried out as described above.

Serological assays.

For measurement of serum antibody levels, HAI using 0.5% turkey or 1% horse red blood cells and MN assays were performed as previously described (19). Individual serum samples were used for all serological analyses.

Replication of H5 viruses in mice.

Groups of 20 female BALB/c mice were lightly anesthetized with 4% isoflurane and inoculated i.n. with 10⁵ TCID₅₀ of each of 9 H5 viruses diluted in L-15 medium (Invitrogen) to a final volume of 50 μl/mouse. Mock-infected control groups of mice were inoculated with L-15 medium alone. Five mice per virus were sacrificed on days 2, 3, 4, and 7 p.i., and lungs and NTs were harvested and stored at −80°C. Organs were weighed and homogenized in L-15 medium containing 2× antibiotic-antimycotic (catalog number 15240062; Invitrogen) to make 10% (wt/vol) and 5% (wt/vol) tissue homogenates of lungs and NTs, respectively. Tissue homogenates were clarified by centrifugation at 1,500 rpm for 5 min, and virus titers were determined in MDCK cells and calculated by the Reed-Muench method (42).

Replication of candidate ca vaccine viruses in ferrets.

To assess the replication of the H5 ca vaccine candidates Egy/09 ca and dk/Hok/00 ca, groups of 3 individually housed 8- to 12-week-old female ferrets from Simonsen Laboratory (Gilroy, CA) were inoculated i.n. with 10⁷ TCID₅₀ of the Egy/09 ca or dk/Hok/00 ca virus in 0.5 ml. One group of ferrets received 0.5 ml of L-15 medium as a mock-infected control. Three ferrets from each group were sacrificed at days 1, 3, and 5 p.i., and virus titers in the NTs and lungs were determined in MDCK cells or embryonated hen eggs and expressed as the TCID₅₀ or 50% egg infective dose (EID₅₀) per gram of tissue, respectively.

To evaluate the immunogenicity and protective efficacy of the H5 vaccine candidates, groups of 6-month-old female ferrets (three ferrets/group) that were seronegative for antibodies to circulating H3N2 and H1N1 human influenza viruses were immunized i.n. with 2 doses (day 0 and day 28) of 10⁷ TCID₅₀ of the Egy/09 ca or dk/Hok/00 ca virus or L-15 (mock immunized) in 0.5 ml, and serum samples were collected on days 0, 14, 28, 42, and 52 p.i. The animals were transferred to an ABSL-3 facility on day 56 p.i. Sera were collected, and the ferrets were challenged i.n. with 10⁷ TCID₅₀ of various H5 wt viruses, including ck/Egy/10 (H5N1), Indo/05/05 (H5N1), ck/PA/83 (H5N2), and dk/Hok/00 (H5N3). The ck/Egy/10 strain was used for challenge infection since we did not have access to the Egy/09 wt virus that would be the parent virus of the Egy/09 ca vaccine. The animals were euthanized 3 days later, and NTs and lungs (right middle lobe and the caudal portion of the left cranial lobe) were harvested. Tissue homogenates of NTs and lungs were titrated in MDCK cells, and the virus titers were expressed as TCID₅₀ per gram of tissue.