Phylodynamics of avian influenza clade 2.2.1 H5N1 viruses in Egypt

The phylogenetic analysis of the H5 gene of Egyptian viruses from 2006 to 2014 indicated the presence of two main subgroups—namely the classic 2.2.1 and the variant 2.2.1.1—according to the updated nomenclature of the WHO/OIE/FAO H5N1 Evolution Working group [9]. The classic group of clade 2.2.1 that was introduced into Egypt in 2006 remained stable through 2009 and represented the original viruses known at that time. The variant clade 2.2.1.1, which emerged in late 2007 from vaccinated commercial poultry, was subdivided into 2 clusters from 2008 to 2011 (2.2.1.1 and 2.2.1.1a). The first cluster emerged in late 2007 (2.2.1.1) and remained until 2009, while the second cluster (2.2.1.1a) emerged in 2008 and remained until 2011. Since then, these variant clusters have not been detected (Figs. 1 and 2). In 2008, the classic viruses evolved into a new clade 2.2.1.2 due to gradual accumulation of genetic mutations in the HA protein, and was the dominant cluster between 2009 and 2014 in both the household and commercial poultry sectors irrespective of their vaccination status (Figs. 1 and 2). Among the 368 H5 sequences analyzed in this study, there were 299 viruses of classic clade (75 belonged to 2.2.1, 224 to 2.2.1.2) and 69 viruses of variant clade (21 of 2.2.1.1 and 48 of 2.2.1.1a) (Fig. 2).

In general, the original viruses of clade 2.2.1 from 2006 to 2009 were distributed along populated areas of the Nile basin. They were mostly detected in the northern Delta region (53/75) with fewer cases (22/75) in Upper Egypt. The results from passive surveillance and notifications through the veterinary authority supported the same findings that Lower Egypt represented the highest record in comparison to Upper Egypt [10]. However, this finding requires further investigation because it may be due to a sampling bias, as most surveillance activity was directed to the Delta governorates at that time.

The endemic 2.2.1.2 cluster was identified first in 2008 and continues to circulate. It was widely distributed in most governorates in Egypt. Unlike 2.2.1, the 2.2.1.2 cluster was equally distributed between Lower and Upper Egypt (each with 112/224). The 2.2.1.2 viruses were mostly detected in Fayoum (37 virus) and Giza (24 virus) in Upper Egypt, as well as in Menofia (32 virus) in Lower Egypt, (Fig. 3).

From 2007 to 2011, most (51/69) viruses from the variant clade 2.2.1.1 were detected in the Delta area, with lower detection rates (18/54) in Upper Egypt (Giza, Beni suef, Menia, Luxor, and Qena governorates). The variant clade was highly prevalent (46/69) in commercial poultry farms, especially in Qalubiya (20/46), then Giza (6/46), Sharkia (5/46), and Dakahlya (5/46). Both 2.2.1.1 and 2.2.1.1a were first detected in Sharkia governorate with further expansion of both variant clusters to Qalubiya, Beheria, and Dakahlya in Lower Egypt and to Giza then to Upper Egypt (Additional file 1: Figure S1).

Analysis of virus population dynamics of the entire data set of the Egyptian H5N1 viruses showed a rise in genetic diversity in the 2.2.1.2 cluster from early 2008, shortly after the first introduction of the H5N1 viruses in the country in 2006. From 2009 to 2014, the 2.2.1.2 cluster exhibited a constant progressive adaptation to poultry and was considered to be an endemic cluster [11]. Genetically and antigenically distinct viruses emerged in Egypt in late 2007 after vaccination began in poultry (referred to as subgroups E and F [12] or subclade B [13] or clade 2.2.1.1 in this study) and estimated to have the highest divergence and rapid evolution rate.

The classic clade (2.2.1) and endemic clade (2.2.1.2) was widely distributed in the household poultry production sector involving chickens (118/134) and ducks (72/75). From 2009–2014, there was an increased detection of the 2.2.1.2 clade in live bird markets (LBM) (25/33), (Fig. 4). In this study, the variant clades (2.2.1.1 and 2.2.1.1a) were mainly reported in commercial chickens (44/93). Although most cases (38/69) were presented with unknown vaccination history, the variant clades were mainly reported in vaccinated flocks (27/69).

There is an apparent lack of disease notification and reporting in the commercial poultry sector in Egypt [14]. Thus HA gene sequences of H5N1 viruses since 2012 from this sector is lacking in most governorates. Phylogeography can highlight the drivers of H5N1 emergence and spread. Qalubiya appears to represent a popular location for virus transmission as also has been explored in previous study [15]. In addition, Sharkia and Dakahlia in Delta and Giza were of the same character as they have all virus clusters recorded in different time periods (Fig. 3). However, there remains uncertainty about virus spread to and from those locations and thus more research needs to be conducted in order to investigate this phenomenon.

The Nonsynonymous/Synonymous nucleotide substitution ratio (dN/dS) per site was greater than one for nine individual sites in the HA1 domain of the HA gene (Table 1), indicating the presence of positive selection driving the evolution of Egyptian H5N1 viruses in these sites. In particular, the 2.2.1.2 clade showed five distinct positively selected sites (120, 129, 154, 155 and 162), 2.2.1.1 has three prominent sites (140, 141 and 162), while 2.2.1.1a has five characteristic sites (140, 141, 154, 162 and 185) (Table 1). The results indicated that the 2.2.1.2 clade is under positive selection pressure that leads to more adaptation of the viruses to the environment and the maintenance of its endemic state.

The population dynamics analysis revealed a rapid increase in the genetic diversity of A/goose/Guangdong/1/96 lineage viruses from mid-1999 to early 2000 [7]. In this study, it was shown that the Egyptian H5N1 viruses exhibited high evolution dynamics in almost all governorates of the country. The viruses from clades 2.2.1.2 and 2.2.1.1a had the highest record of positive selection sites (Table 1), which may be attributed to vaccination pressure due to long-standing application of vaccines with high virus load in the endemic environment. This reflects the continuous adaptation of Egyptian viruses to the poultry and to their environment with persistent changes every season [16]. The genetic variation among the Egyptian viruses was previously reported and the presence of positive selection was recorded. In this regard, Cattoli et al. [13] indicated that evolutionary dynamics and positive selection significantly increased in virus populations in countries applying the avian influenza vaccination for H5N1, compared to viruses in countries that had never used vaccination. They also indicated that the rapid evolution of H5N1 viruses in Egypt was possibly linked to vaccination pressure due to sub-optimal use of vaccines.

The Egyptian viruses showed a high rate of evolution since 2006, as the original clade 2.2.1 was 4 × 10⁻³ substitution/site/year and lasted for 4 years. The variant clade, conversely, had 6.1 × 10⁻³ substitution/site/year, distributed as 3.8 × 10⁻³ for 2.2.1.1 over 2 years and 7.2 × 10⁻³ for 2.2.1.1a over 4 years. The 2.2.1.2 clade showed higher and slower evolution rate in comparison to variant viruses, it was 6.9 × 10⁻³ over a period of 6 years (Table 2). In addition, the Bayesian skyride analysis of the 2.2.1.2 viruses from 2009 to 2014 showed that the genetic diversity is directly proportional to the annual prevalence peaks. The genetic diversity of the variant clusters from 2007 to 2010 showed a higher pattern of increase followed by a sharp decline in 2011(Fig. 5a, b).

The evolutionary analysis of Egyptian viruses revealed that these viruses have progressive rates of evolution. The factors related to this increase were mainly attributed to the sub-optimal use of vaccines and long-lasting virus persistence in the environment leading to the endemic prevalence of 2.2.1.2 viruses over six successive years. Cattoli et al. [13] revealed that the two main Egyptian clades (designated as A and B) have co-circulated in domestic poultry since late 2007 and exhibited different profiles of positively selected codons and rates of nucleotide substitution. The mean evolutionary rate of clade 2.2.1 H5N1 viruses was estimated in their study as 4.07 × 10⁻³ nucleotide substitutions per site, per year whereas clade 2.2.1.1 viruses possessed a markedly higher substitution rate (8.87 × 10⁻³) and that reflected the high genetic diversity among Egyptian viruses.

The analysis of the 368 HA genes enabled us to examine changes in the receptor binding site (RBS), antigenic sites (AS) and the cleavage site.

The most characteristic change in the receptor binding site of the Egyptian viruses included in this study was the observation of one amino acid deletion at site 129 (129∆) that was not recorded in the ancestral strain (A/goose/Guangdong/1/96). This change was linked to the emergence of 2.2.1.2 clade in 2008. The viruses with 129∆ were found in the majority of human infections in Egypt in 2009 and have been found in all H5N1 human infections afterwards [17]. The presence of the 129∆ mutation may affect the binding of the virus to human receptors. Another important change in the receptor binding site was linked to the variant clades (2.2.1.1 and 2.2.1.1a), which showed S129L substitution. The substitution was more pronounced in the 2.2.1.1a cluster and was linked with another substitution (P74S) in most cases (28/35) (Table 3).

The most recently isolated viruses (n = 56) that were collected between 2012 and 2014 were examined for HA gene mutations and belonged to the 2.2.1.2 clade in which three mutations (129∆, I151T, and S120(D,N)) were shown to be constant (Table 3).

The loss of HA154–156 glycosylation site was shown to enhance H5N1 virus binding to terminally α-2,6 sialic acid receptors and so increased the transmissibility to mammals [17‐19]. The majority of the Egyptian 2.2.1 viruses lacked this site. There were few (6/75) 2.2.1 viruses that had the HA154–156 glycosylation site. Conversely, the majority of the Egyptian 2.2.1.2 viruses (154/224) had this site (Table 3).

The H5N1 viruses from Egypt displayed four characteristic mutations (D43N, S120(D,N), (S,L)129∆, and I151T). The results showed that 57 % of the HA sequenced genes showed a triple mutation (129∆, S120(D,N), and I151T) (Table 3). These triple mutations are characteristic in 2.2.1.2 clusters from different bird species such as chicken, duck, turkey, geese, ostrich, and quail; however, few of the 2.2.1.2 viruses did not carry them. The percentage of those viruses with the triple mutation reached 100 % from 2012 to 2014. Two mutations of those (129∆, I151T) had increased attachment and infectivity to the human lower respiratory tract, but not in the larynx [17]; that indicates an increasing possibility of human infections in Egypt.

The majority of variant viruses of clade 2.2.1.1 had no changes in the receptor binding site at position 129 (129∆) (Table 3), and they were not responsible for any human infection in Egypt, except for one case in early 2008 during the beginning of widespread prevalence of this cluster [20]. In addition, Perovic et al. [21] indicated the extensive evolution of Egyptian H5N1 HPAI virus towards human. They reported that all G2 viruses (referred to as 2.2.1.2 clade in our study) displayed four characteristic mutations (D43N, S120(D,N), (S,L)129∆ and I151T). The other mutations that are linked to increased affinity to human receptors like “S223N, D183G, E186G, Q192R, Q222L and G224S” were not found in the Egyptian H5N1 viruses.

Amino acid substitutions at the antigenic sites can cause antigenic drift, possibly leading to vaccine escape as observed in the field. In comparison to the virus introduced in 2006, the characteristic changes that occurred in the antigenic sites of HA gene are very specific to each cluster (Table 3). There were 35/75 of 2.2.1 viruses that showed no antigenic changes in comparison to the parent Egyptian virus of 2006 and the precursor virus A/Bar-headed Goose/Qinghai/5/05. In addition, 26/75 of the viruses showed one antigenic change at either P74S or D154(E,N) or R162(I,K) or S141P sites. The remaining (14/75) of 2.2.1 viruses showed two to three changes. The most predominant antigenic change for those viruses was R162I which has been observed in 17 viruses. Approximately 60 % (132/224) of the viruses that belong to 2.2.1.2 cluster have four mutations that occurred at the same sites, i.e., S120(D,N), I151(T,L), D154(A,N) and R162K (Table 3). The Egyptian variant viruses (2.2.1.1 and 2.2.1.1a) carry characteristic mutations in the antigenic sites (at positions 74, 129, 140, 141, 154 and 162), while 2.2.1.2 cluster carries characteristic mutations at positions 120, 129, 151, 154, and 162; all these mutations can differentiate each cluster of Egyptian viruses from the others. Four characteristic mutations P74S, R162K, 140G and 141P were frequently detected together in the variant cluster 2.2.1.1 (17/31), whereas the variant cluster 2.2.1.1a had another four characteristic mutation that occurred together in sites P74S, R162(K,E), 141P and D154N (Table 3).

There have been 21 potential antigenic sites identified in the HA of H5N1 HPAI. It was shown that a single amino acid change in the HA of H5N1 HPAI virus can affect immune response and protection [22]. The recent Egyptian viruses from 2013 and 2014 carry four prominent mutations in the antigenic sites (at positions 120, 151, 154 and 162), all of which belong to the endemic clade 2.2.1.2 viruses; two of these mutations (154 and 162) were distinguishing the new viruses from the earlier 2008–09 viruses, which indicate limited changes in these sites in comparison to the earlier viruses of this cluster. However the recent viruses circulating during the last few years in Egypt from 2011 showed a clear separation from the ancestral viruses indicating a gradual evolution of those viruses.

The antigenic analysis of the earlier H5N1 variant strains in Egypt demonstrated antigenic variation [12, 23], which was driven by multiple mutations primarily occurring in the major antigenic sites at the globular head of HA [24]. Other studies showed that the classic clade 2.2.1 strains are antigenically related and cross-reactive to the ancestral Asian H5N1 strains, but demonstrated weak cross-reactivity with the Egyptian variant 2.2.1.1 strains [23, 25]. The majority of these mutations, alongside the other 19 amino acid mutations, were located within or adjacent to the receptor binding domain (RBD) in the HA1 that may affect the virus replication and transmission. There were six conserved mutations in previously reported antigenic sites (D43N, S120(N,D), S129∆, I151T, D154N and R162K) between the early 2006 strain and the endemic 2.2.1.2 cluster strains. It was shown that the D43N mutation resulted in antigenic drift between classic 2.2.1 and 2.2.1.2 clusters [26].

The results of the present study showed clear differences among the virus clusters in terms of the absence or presence of certain changes in the antigenic sites. For instance, the majority of 2.2.1.2 cluster had four changes occurring at the same time in the antigenic sites (S120(D,N), R162K, I151(T,L) and D145(A,N)), while the latter two sites were lacking in the 2.2.1.1 cluster. Ibrahim et al., 2013 [26] observed an antigenic variation between different H5N1 clusters, especially between the variant 2.2.1.1a and the endemic 2.2.1.2 cluster strains showing a significant antigenic drift. Some residues were located in different antigenic sites like 133S, 154D and 156A, 190L and 192Q and 71L. They also confirmed that the 2.2.1.1 cluster showed a broader reactivity to all strains that represent different H5N1 clusters circulating in Egypt, as it shared residues with all the strains in the major antigenic sites.

There were 13 groups of viruses identified based on the amino acid sequences at the cleavage site. Most of the H5N1 HPAI virus isolates belonging to 2.2.1 (42/75), 2.2.1.2 (105/224) and 2.2.1.1a (7/38) as well as all the viruses that belong to 2.2.1.1 cluster (31/31) have a common cleavage site of “ERRRKKR”, described as the consensus cleavage site for clade 2.2 viruses [27]. However, the pattern “EKRRKKR” became dominant from 2013 and replaced the previous pattern which disappeared after 2012 (Table 4). The pattern “EGRRKKR” of amino acid sequence was exclusively present in 2.2.1.1a cluster and represented the highest proportion (23/38) among the other six patterns in the same cluster, while the pattern “ERRRKR” was observed only in 30.6 % (23/75) of the viruses that belong to the 2.2.1 cluster (Table 4).

The currently dominant amino acid cleavage site pattern “PQGEKRRKKR/GLF” is closely associated with the mutation 129∆ at the receptor binding site, and mutations at the antigenic sites (S120D, I151T, D145N and R162K) which are known to increase the binding ability to human receptors. All these mutations characterize the dominant cluster (2.2.1.2) from 2011 onwards [26]. The substitution R325G was found at the cleavage site in Egyptian 2.2.1.1a viruses, while R325K characterized recent 2.2.1.2 viruses from 2011–2014. The R325G substitution was shown to significantly reduce pathogenicity without altering the transmission efficiency of H5N1 HPAI virus [28] and shows that non-adaptive mutations can play a role in virus evolution as the 2.2.1.1a cluster disappeared in Egypt since 2011.

The high diversity of the HA gene in relation to some governorates indicates active virus circulation in different locations. In this study, the hypervariability of the HA gene was noticed in relation to the geographical location (data not shown). Viruses from Qalubiya, Giza, Menufia and Dakahlya governorates had the highest number of heterogeneous amino acid sequences of the HA gene. In addition, unequal virus distribution was noticed among governorates and that favour virus persistence in an endemic area (Fig. 3). All the above mentioned results support the existence of genetic diversity of HA gene in Egypt with progressive virus evolution in a model of intermittent re-emerging H5N1 viruses to a clean areas located inside an endemic environment.