Epidemiological and genetic characterization of pH1N1 and H3N2 influenza viruses circulated in MENA region during 2009–2017

Recently published data indicates that up to 650,000 deaths are reported annually due to respiratory illness associated with seasonal influenza [1]. Influenza viruses evolve rapidly, escaping natural or vaccine-induced immune response by accumulating mutations especially within surface glycoproteins genes: hemagglutinin (HA) and neuraminidase (NA) [2]. This antigenic drift results in annual influenza epidemics in humans, a major cause of morbidity and mortality and consequently increased burden on health services [3]. In addition to antigenic drift, influenza viruses evolve by genetic reassortment, leading in some occasions to life threatening pandemics. Three influenza A pandemics were reported in last century: 1918 H1N1, 1957 H2N2 and 1968 H3N2 [4, 5]. In June 2009, WHO announced the first pandemic in the twenty-first century, which was caused by swine-origin H1N1 (pH1N1) and resulted in more than 250,000 deaths worldwide [6, 7]. However, the pH1N1 virus replaced the seasonal H1N1 virus, and is currently co-circulating with H3N2 and influenza B viruses.

Antigenic drift mutations are typically localized at the antigenic sites surrounding the receptor binding site (RBS) of the HA protein. Therefore, in addition to contributing to HA antigenicity, they may affect receptor preference and virulence [8]. Four antigenic sites (Ca, Cb, Sa, Sb) have been identified for H1N1 viruses compared to five sites (A-E) for H3N2 viruses [9, 10]. Changes in HA and NA proteins may sometimes lead to acquisition of glycosylation sites, which is believed to efficiently generate antigenic variants and play critical role in virus evolution [11, 12] by masking epitopes, stabilizing HA structures and altering receptor binding specificity [13].

Each season, the World Health Organization (WHO) select virus strains for optimal vaccine performance with the recommendation for the northern hemisphere announced in February and that for the southern hemisphere in September [14]. Following 2009 pandemic, unprecedented surveillance programs have been launched to monitor influenza viruses at the molecular and epidemiological levels [15, 16].

Countries in the Middle East and North Africa (MENA) region are continuously considered as one of the hotspot regions for emerging and reemerging infectious diseases [17]. A major reason is the location of MENA countries in the pathway of three major migratory bird flyways: east Africa–west Asia, Black Sea–Mediterranean, and east Atlantic [18]. Additional risk factors that contribute in the emergence and rapid spread of epidemic diseases in the MENA region include prolonged humanitarian crises which subsequently result in massive population mobility, weak surveillance systems and limited laboratory diagnostic capacity. Further, some countries in this region, such as the Gulf States, are hubs for expatriates that constitute more than half of its population, many of which arrive from East and South-East Asia [19]. Globally shared factors also include climate change and increased human–animal interaction [20].

Until recently, little was known about epidemiology of influenza in MENA region. However, many countries have developed a systemic influenza surveillance programs after 2009 pandemic [21]. Currently, fourteen countries in MENA region are reporting to WHO virological surveillance database ‘FluNet; on weekly basis. However, there is a relatively limited number of genetic sequences available for influenza virus from MENA countries. Moreover, most of publications characterizing the molecular evolution of influenza virus in MENA originated from very few countries, mainly Egypt and Iran [22] . In the current study, we report on the epidemiology of co-circulating influenza A and B viruses in MENA since April 2009 until December 2017. Most of countries in this region are characterized by a subtropical climate [23]. However, climatic parameters such as temperature and humidity are variable among different regions of MENA and hence could affect influenza seasonality patterns. Therefore, we evaluated the epidemiology of influenza viruses among the three areas of MENA region: Gulf States, Levant, and North Africa. Additionally, we analyzed the genetic characteristics and molecular evolution of HA and NA genes of circulating pH1N1 and H3N2influenza A subtypes.

We compared the relative contribution of influenza types (A and B) and subtypes (pH1N1 and H3N2) to influenza epidemics and investigated seasonality patterns in Arabian Gulf, Levant and North Africa regions during 2009–2017. Influenza surveillance data were retrieved from the global web-based tool ‘FluNet database’ which was launched first in 1997 by WHO for influenza surveillance [24]. A preliminary check of data available revealed that reported influenza cases were unevenly distributed across countries, in addition to lack of data from three countries (Saudi Arabia, Libya and Yemen), thus a total number of 17 countries were involved in current study. A study comparing influenza seasonality patterns among MENA countries has been published earlier in 2018 by Caini and his colleagues [46]. In the current study, we compared influenza seasonality based on WHO classification of influenza transmission zones and divided MENA countries in three main regions: Arabian Gulf (5 countries) and Levant (6 countries) which belong to Western Asia transmission zone and North Africa (5 countries) which is classified as North Africa transmission zone [33].

Influenza A was dominant in all years, with the pH1N1 accounting for the majority of cases except in 2012, during which H3N2 dominated. Influenza A and B accounted for 79.5 and 20.5% of all influenza cases during the study period, respectively. Generally, seasonality of influenza epidemics in all regions under study was in line with the northern hemisphere with a typical peak placed between December and March. Although Levant and Arabian Gulf lies in West Asia transmission zone (based on WHO classification), seasonality trends in Gulf were similar to that of North Africa rather than Levant. In all years, influenza activity started during October at high levels in Gulf and North Africa compared to a weak, late commence of the season in Levant, albeit the earlier winter seasons in the later. Furthermore, influenza positive cases were gradually increasing/decreasing in Gulf and North Africa, whereas a sharp increase/decrease of activity was reported in Levant (Additional file 2: Figure S1). The less defined seasonality pattern in Gulf countries could be partially explained by the tropical-like climate of these countries which is generally characterized by a wide primary peak in winter in addition to secondary smaller peaks seen throughout the year [34]. In contrast, the location of Levant countries (Turkey-Lebanon-Syria-Jordan-Cyprus) in temperate regions was demonstrated by a clear seasonal variations of influenza epidemics, with a marked peak in cold winter months [46‐48].

In addition to monitoring activity of circulating influenza viruses, WHO utilizes surveillance data in studying virus evolution and assessing the antigenic characteristics of circulating viruses before selecting the most appropriate influenza vaccine strains for both hemispheres. Therefore, the molecular evolution of surface glycoproteins (HA and NA) of influenza A subtypes were also investigated in current study. All HA and NA sequences of pH1N1 and H3N2 viruses depicted in IRD since 2009 (total of 1260 sequences) were analyzed. Due to the limited number of sequences available, sequence analysis was done per MENA as whole rather than independent three regions. Phylogenetic analysis of both genes revealed that influenza viruses clustering in the phylogenetic tree is mainly chronological for both subtypes. Additionally, the phylogenetic analysis of the HA sequences revealed similar evolution of mutations circulating worldwide and hence similar clustering with globally circulating strains. In accordance to published data [25], majority of pH1N1 viruses after 2014 clustered within the genetic clade 6B and related sub-clade, 6B.1, while in H3N2 viruses’ sub-clades 3C.2 and related sub-clade 3C.2a were the predominating. In contrast to our findings, though, pH1N1 viruses belonging to group 6 related sub-clades (6A, 6B and 6C) were detected globally during 2013–2014 compared to an earlier appearance (2011–2012) in MENA countries. Moreover, group 6C pH1N1 viruses were not found after 2013 in MENA countries, however, were found in viruses isolated in other countries during 2015. Likewise, phylogenetic analysis of H3N2 viruses has revealed some differences compared to globally reported data [25]. Although global reports indicate that the majority of H3N2 viruses circulated during 2011–2012 fell into sub-clade 3C, viruses were mainly clustering within subclades 3A and 3B in MENA countries during that period.

For further evolution analysis of the pH1N1 and H3N2 HAs, we determined the positions of identified substitutions with respect to their localization in HA protein structure. Generally, more substitutions were detected in H3N2 viruses (n = 32) compared to pH1N1 viruses (n = 24) during study period. In accordance with our findings, analysis of antigenic dynamics of both subtypes’ HAs revealed that H3N2 viruses evolve faster than human H1 subtype [49]. The average amino acid mutation rate of the HA protein for H3N2 virus was 3.6 per year compared to 2.45 per year for H1N1 [50]. One reason could be the continuous circulation of H3N2 virus in human population since 1968, resulting in higher immune pressure as compared to the relatively new pH1N1 virus [51]. Substitutions were most frequently found in the head domain in pH1N1 viruses, including eight changes in the antigenic sites and five in the RBS, while interestingly, changes were mainly localized in stem domain in H3N2 viruses. Yet, most of substitutions identified in head domain of H3N2 viruses were localized in or near major antigenic sites (n = 8). Of note, changes of the antigenic HA composition of H3N2 viruses were accompanied by a six-time exchange of the H3N2 vaccine strains which exert more pressure on H3 antigenic sites compared to only two changes in pH1N1 vaccine strains during study period (Additional file 1: Table S1). Overall, the H3N2 strains displayed more diversity and faster accumulation of mutations in antigenic epitopes compared to pH1N1 strains [49]. HA sequences from the H3N2 strains possessed two amino acid changes in one epitope (B; L157S and N189K) during the 2009. In 2011, this number increased to seven changes in three epitopes: two in epitope A, three in epitope B, and two in epitope C (Fig. 4b). In following years, more antigenic drift occurred due to changes in H3 antigenic site, however, most of these changes were found in recommended vaccine strain of subsequent seasons. The continuous change in H3N2 vaccine strains could also explain H3N2 lower activity despite the larger number of accumulated substitutions (Fig. 7b). Reports measuring influenza vaccine effectiveness have shown that the overall vaccine effectiveness (VE) was highest during 2010–2011 season (60%), while lowest during the 2014–2015 season (19%) [52]. This was associated with change in vaccine strain of H3N2 from A/Victoria/361/2011 to A/Texas/50/2012. Thus, vaccine composition was reformulated again for 2015–2016 season to contain A/Switzerland/2013 and was shown to be more effective (48%) [52]. In 2015, H3N2 viruses were carrying four substitutions in antigenic sites that were not found in A/Texas/2012 strain including: N144S and N145S in A3 site, and Y159F and K160 T in B1 site. Moreover, 87% of pH1N1 viruses had acquired S179 N mutation in 2015, which was associated with acquisition of potential glycosylation site. Together, these changes in H3N2 and pH1N1 viruses isolated in 2015 might -at least- partially explain the lower vaccine effectiveness during earlier months of 2015.

The pandemic strain, pH1N1, originated from swine reservoir and circulated in human population for much shorter time (~ 9 years) compared to H3N2 viruses that has been circulating since 1968. Therefore, pH1N1 viruses are still considered to be in adaptive phase in terms of recognition of new receptor as well as avoiding immune response of the new host. Evidence of such adaptation process was revealed while analyzing mutations acquired by different pH1N1 viruses since 2009. Positive selection test of HA gene in pH1N1 viruses revealed that two amino acids in Ca antigenic site (D239G/E/N and Q240R) showed evidence of adaptive evolution. Owing to its presence in HA antigenic and RBS, variants of 239 position, in particular, has been found to be a key determinant of virus capability to bind epithelial cells in upper and/or lower respiratory tract [53]. D239G variant has demonstrated the ability to bind α2–6 receptors in addition to increased binding affinity for α2–3 receptors expressed by pneumocytes in lower respiratory tract, and hence resulting in enhanced illness [37]. 225G and 225 N variants were only detected in specimens obtained from patients with severe clinical symptoms, whereas D225E variants were detected in both severe and mild cases with similar frequencies [53, 54]. In this study, D239 variants occurred in 10.3% of pH1N1 viruses. In contrast to D239E (8%) which appeared during 2009–2010, D239G and D239N substitutions formed no genetic cluster and reported sporadically suggesting that there was no spreading of viruses carrying this variant. Of note, variants of D239 position were also reported in seasonal strains circulated before 2009 [39]. Receptor binding site of HA protein represented by 130-loop (148–152), 190-helix (201–208) and 220-loop (235–242) (H1 numbering) is one of the major determinants of virus transmission ability [55]. Therefore, amino acid residues in and around the receptor binding site were predicted to change during adaptation process to human receptors. Those amino acids include: S200, S202, D204, A214, I233 and E241 [56]. Four of these sites were found to be mutated in pH1N1 viruses included in this study. Of these, S200P, S202 and A214T were detected sporadically while I233T was introduced successfully in more than 80% of viruses after 2014. In 2013, de Vries and his colleagues has reported that S200P and S202 T substitutions would increase the receptor-binding avidity whereas the presence of A214T substitution was linked to decrease binding avidity [57]. Interestingly, sequences expressing A214T variant were always associated with S200P and S202 T, which could be considered as compensatory process. With respect to I233T mutation, the presence of non-polar Isoleucine in 233 position has been found to disrupt the positioning of residues in the RBS from making a stable network of interactions [58]. The substitution the Isoleucine with the polar amino acid Lysine, has been shown to significantly increase HA affinity to human receptor [58]. Similarly, the displacement of the non-polar Isoleucine residue at 233 position with a polar amino acid, Threonine, could generate a stable ionic interaction between polar T233 residue in RBS and the acidic E241 residue in 220 loop, consequently increasing binding affinity to human receptor in comparison to earlier pH1N1 isolates [58]. In addition to mutations in RBS, limited number of mutations in HA stem domain were also acquired in an early stage of pH1N1 evolution and thereafter introduced successfully in virus population. Among these mutations are E391K and E516K, which are found in the vicinity of the fusion peptide and transmembrane domain respectively. Such mutations are usually considered to be under negative selection because they might alter HA stability or functions [59]. However, for a newly emerged virus such as pH1N1, such mutations may be prone to positive selection because it has to adapt to the new host. It is noteworthy, both subtype viruses acquired fixed mutations generally rapidly, however, more fixed mutations were accumulated in H3N2 compared to more non-fixed substitution in pH1N1, another evidence that the latter is still in adaptation process.

In addition to antigenic drift, glycans on the globular head of HA can shield antigenic sites from neutralizing antibodies, but at the same time can affect HA binding to host cell receptors [38]. Compared to vaccine strains, two amino acid mutations involving S179 N (Sa epitope) and K160 T (B epitope) in globular head of H1 and H3, respectively, have resulted in acquisition of putative glycosylation sequons. Glycosylation at the antigenic sites of HA head could be associated with increased viral pathogenicity by hiding antigenic sites from immune recognition. Interestingly, S179 N glycosylation was also reported in seasonal H1N1 viruses circulated during 1940–1948 before being replaced by N177 glycosylation in subsequent seasons until 2009 [60]. Additional glycosylation sequons in positions 42 and 367 have been also acquired by N1 and N2 respectively. Unexpectedly, mutation rate of N2 (3.09 × 10^− 3) was found to be higher than that of H3 (1.47 × 10^− 3), however, this could be related to unusual number of mutations in N2 sequences which were not reported elsewhere. That in addition to low number of N2 sequences which were mainly deposited from single country (Iran).

Systematic monitoring for influenza viruses is also important for evaluating the efficacy of the influenza NA inhibitors. Currently, neuraminidase inhibitors (NAIs) such as oseltamivir and zanamivir are the main drugs for treatment of influenza infections [61]. The effectiveness of oseltamivir in treating influenza infections was threatened by the predominance of oseltamivir resistance among seasonal H1N1 viruses 2009–2010, even in countries where oseltamivir had not been used [62]. However, seasonal virus was subsequently displaced by the pH1N1 virus, which is largely sensitive to oseltamivir [63]. It is noteworthy, oseltamivir-resistant strains were rarely reported during 2009–2010 [64]. In following years, though, a notable increase in the proportion of oseltamivir-resistant pH1N1 viruses were reported worldwide amongst patients with or without NAI treatment [65, 66]. In accordance with our findings, most of reported oseltamivir-resistant cases exhibited geographical and temporal clustering. Such cases were reported during 2011 in New England, Australia [67] and during 2013 in Sapporo, Japan [68]. Here, all oseltamivir-resistant strains were reported were clustering in Oman and Iran during 2013 and 2015, respectively, suggesting efficient transmission of oseltamivir-resistant viruses among humans [69]. This raises the concern that prevalence of oseltamivir-resistant pH1N1 viruses may increase in the future as the case of the previously circulating seasonal H1N1 viruses. Interestingly, two of NA fixed mutations reported in current study, V241I and N369K, have been found to offer robust fitness on pH1N1 viruses carrying H275Y mutation by enhancing NA surface expression and its enzymatic activity [44]. The presence of such permissive mutations in all pH1N1 viruses suggesting that they are more permissive to the acquisition of H275Y and hence increase the risk of oseltamivir-resistant strains to spread globally. Here, we found oseltamivir-resistance mutation, H275Y, in only 4% of pH1N1 virus while no resistance mutations were reported in H3N2 viruses besides both subtypes have retained susceptibility to zanamivir. Unfortunately, we were unable to find published resources or reports regarding the prevalence of NAIs use in MENA countries.

There are limitations in this study. Surveillance data was not available from Saudi Arabia, Yemen or Libya. The lack of influenza surveillance data from Saudi Arabia especially during Umra and Haj seasons could underestimate the actual prevalence of influenza infection in MENA region. Nevertheless, several studies investigating the burden of respiratory infection during Hajj season were published [70]. Moreover, there is a relatively small number of HA and NA genes depicted in Influenza Research Database form MENA countries, especially for H3N2 viruses. This mandates further molecular characterization of circulating viruses in the region. Analysis of small number of sequences could preclude exploring the actual genetic diversity of influenza viruses.

In the context of the national influenza surveillance, we estimated the epidemiology of influenza types A and B in Arabian Gulf, Levant and North Africa regions since 2009. The pandemic strain pH1N1 was the dominant subtype in all years except in 2012 which was dominated by H3N2. Although WHO includes both Gulf and Levant in Western Asia, we found that seasonality patterns of Gulf and North Africa regions are similar compared to Levant region. We have also characterized the genetic evolution of influenza A subtypes: pH1N1 and H3N2 viruses during same period. Comparison of identified mutations of pH1N1 and H3N2 viruses encountered over 9 years revealed significant differences with regard to position and function of identified substitutions. pH1N1 viruses accumulated more substitutions in the head domain whereas H3N2 viruses had most of the reported substitutions in the stem domain. Moreover, H3N2 viruses acquired more fixed substitutions that were successfully introduced in viral population compared to larger number of non-fixed substitutions in pH1N1. More importantly, amino acids in the antigenic sites of H3N2 viruses were showing more variability than those of pH1N1 viruses, probably indicating greater immunological pressure. Interestingly, the increase of substitutions in HA globular head of pH1N1 virus was seen in strong influenza activity while that was not the case in H3N2 viruses despite the higher number of accumulated mutations. A possible explanation could be that the H3N2 viruses have been circulating in human population since 1968 resulting in circulation of different variants of virus compared to the relatively novel pH1N1. On the other hand, molecular analysis of the neuraminidase genes revealed the clustering of Y275H oseltamivir resistant mutation in pH1N1 viruses isolated from Oman and Iran, however, neither NAs carried any of substitutions associated with reduced susceptibility to zanamivir. Major challenges are yet to be faced by MENA countries in order to contribute to global surveillance of influenza virus by improving their surveillance systems for detection and response to all public health events.