In February 2013, two patients living in Shanghai were admitted to the Shanghai Fifth Hospital with fever, cough and respiratory tract infection, followed by severe pneumonia, respiratory distress and multiorgan dysfunction . While the first patient, an 87-year-old man, did not present a history of exposure to live birds during the preceding 2 weeks, the second patient, a 27-year-old man, was a butcher at a market selling live birds. A 35-year-old female from the Anhui Province of China, the third patient who became infected, visited a chicken market a week before her symptoms started [2, 3]. All three patients died, and their infections did not appear to be epidemiologically linked .
On March 29, 2013, the Chinese Center for Disease Control and Prevention confirmed that the illness was caused by a new reassortant H7N9 strain, and on March 31, 2013, these infections were reported to the World Health Organization [2-5]. By April 18, 2013, the H7N9 virus was identified in six provinces from China and it caused 82 infections and 18 deaths [6, 7]. There were 126 laboratory-confirmed cases reported by April 30, 2013, and 132 cases confirmed by May 13, 2013, including 33 deaths [8, 9]. Most infections occurred in older urban men, who reported exposure to chickens or pigeons grown in captivity, professionally or through visits to poultry markets, 3–8 days before the onset of symptoms [1, 8]. The first infection outside Mainland China was in a 53-year-old male from Taiwan who returned on April 9, 2013 from Suchow, the Jiangsu Province, and did not report a history of contact with sick patients or with animals during his travel .
Antigenic drift and antigenic shift
The lack of proofreading, a general characteristic of RNA viruses, accounts for the high-error rates, between 10−3 and 10−5 mutations per nucleotide per replication cycle [11, 12]. The replication error rate of influenza viruses, approximately 7.2 × 10−5 bp−1, corresponds to about one mutation per genome replication, and provides a source of sequence diversity that enhances their ability to adapt to various environments [13-15]. In sharp contrast, much lower error rates are observed in DNA viruses – for example, 1.8 × 10−8 mutations per nucleotide are thought to occur during one genomic replication of the herpes simplex virus type 1 [11, 16, 17]. The error-prone replication results in the gradual incorporation of mutations into the influenza virus genome, in a process known as antigenic drift, allowing the virus to escape immune detection [18-21]. In addition to this characteristic that is shared by all RNA viruses, the influenza virus harbours a second feature that is instrumental for pathogenesis. The organisation of the genome into several segments (eight for type A and B viruses, seven for type C viruses) facilitates the exchange of genetic information when two or more strains co-infect the same cell, leading to the formation of new, reassortant viruses. This process is responsible for major changes in the antigenicity of the virus, a process that is referred to as antigenic shift and was only described for type A influenza viruses [22, 23]. Two parental viruses that each has eight segments can generate 256 distinct combinations by this mechanism, providing a powerful mechanism for genetic diversity [24, 25].
The H7N9 virus
Human infections with other H7 viruses, including H7N1, H7N2, H7N3 and H7N7, were previously reported in several countries worldwide [5, 7, 26-30]. However, human infection with an N9 type of influenza virus has not been reported to date . Based on phylogenetic analyses, it was proposed that the H7N9 virus emerged through multiple reassortment events among viruses from at least four different avian origins, and it entered the human population on at least two different occasions [3, 7, 31]. The H7 gene is thought to have originated from H7N3 viruses infecting ducks in eastern Asia, the N9 gene is closest to genes from viruses isolated from Korean wild ducks and migratory birds, and the six internal genes originated from at least two groups of H9N2 avian viruses infecting chickens [7, 8]. An interesting particularity of the H7N9 virus is that, unlike H5N1, which previously caused death in birds, this new virus causes mild or no symptoms. This makes detection more challenging, and renders culling a less effective intervention to prevent or limit transmission [6, 8]. While no evidence for sustained human-to-human transmission of H7N9 was found, only approximately 40% of those who have died or were severely ill were reported to have had contacts with poultry, and live poultry markets explained some, but not all the reported human infections .
H7N9 sequencing revealed an interesting combination of mutations. Two of the first three sequenced strains, and several of the ones that were isolated subsequently, harboured a Q226L mutation (position 226 in H3 numbering and position 217 in H7 numbering) in the hemagglutinin (HA) gene, a change that decreases binding to the alpha 2,3 receptors found in the avian enteric tract and increases binding to the alpha 2,6 receptors found in the human upper respiratory tract [30, 31]. This mutation was key to the ability of the 1957 and 1968 pandemic viruses to acquire human transmissibility [33-35]. In all three viruses that were initially sequenced, and several isolates that were sequenced later, the HA gene harboured the T160A mutation (H3 numbering) [3, 36]. By leading to the loss of N-glycosylation at a nearby amino acid, this mutation increases the binding of the virus to human receptors [36, 37]. In addition, all three viruses harboured a 5-amino-acid deletion in the neuraminidase (NA) stalk [3, 6]. Stalk length is important for both pathogenicity and host range and, in mice, viruses with shorter NA stalks appear to be more virulent than those with longer stalks [30, 38, 39]. A key feature of the H7N9 virus is the replacement of glutamic acid at position 627 of the PB2 protein with a lysine (E627K), a hallmark of virulence and adaptation to mammals [30, 40]. This mutation, which was shown to improve the enzymatic activity of the viral polymerase at lower temperatures, is thought to confer temperature-dependency, allowing the virus to better replicate at the lower, 32–33 °C temperatures found in the human upper respiratory tract than at the higher, 41–42 °C temperatures from the avian enteric tract [40-42]. It is interesting to remark that, while the E627K mutation was present in viruses infecting three patients from the Hangzhou Province, an isolate originating in poultry faeces from a market visited by one of the patients had glutamic acid at this site (627E), indicating potential avian to human adaptation pressure . An additional feature of several H7N9 strains is the presence of the S31N mutation in the M2 homotetrameric ion channel, which confers resistance to the M2 channel blockers amantadine and rimantadine [30, 36]. On the other hand, the HA cleavage site, which harbours several basic amino acids in virulent influenza viruses, such as certain highly pathogenic H5N1, H7N7 or H7N1 strains (RRRKKR*G, KRRRR*G, or KKRKKR*G, respectively, where the asterisk indicates the cleavage site), but a single arginine in low-pathogenicity viruses, contained a single arginine residue (EIPKGR*G) in several H7N9 strains that were sequenced, representing a low-pathogenicity feature [30, 43-47]. HA molecules that have several basic amino acids at this location are highly cleavable, and the acquisition of enhanced cleavability, which is essential for pathogenesis, often accompanies the conversion of low-pathogenicity strains to high-pathogenicity ones [45, 48].
Thus, it appears that while genetic reassortment helped the formation of this novel H7N9 strain, antigenic drift facilitated the emergence of several adaptive mutations responsible for its virulence, drug resistance and zoonotic potential [30, 32]. Importantly, examining the HA, NA and PB2 sequences from several H7N9 isolates from human, poultry and environmental origins that were sequenced by early May 2013 revealed that there were 36 sites with at least one mutated amino acid, pointing towards the diversity of mutations that can be seen in these viruses .
A study of the first 109 patients with H7N9 influenza reveals additional interesting features. Even though individuals aged 2–91 years old developed H7N9 influenza, two-thirds of the infections occurred in people over 50 and two-thirds occurred in males. Infections were twice more frequent in women aged 50–64 years and 65–79 years than in women from the 20- to 34-year age group. Similarly, infections were four to five times more frequent in men aged 50–64 years and 65–79 years than in men from the 20- to 34-year age group. While several explanations exist for this trend, including the possibility of greater exposure to poultry among elderly people, antibody-dependent enhancement was advanced as a potential explanation . Antibody-dependent enhancement, a phenomenon reported for several viruses, occurs when low levels of pre-existing non-neutralising antibodies that are cross-reactive, but not cross-protective, enhance viral entry and infectivity [49-55]. In the case of dengue hemorrhagic fever, it is known that individuals having a secondary infection are at higher risk to develop severe disease as compared with individuals at their first infection. Antibody-dependent enhancement was proposed, and is a widely accepted mechanism to explain the increased susceptibility to develop severe disease upon infection with a new serotype in people with pre-existing immunity to the virus [56, 57].
Integrating past lessons and developing new frameworks
Particularly during the initial stages of an outbreak, when the behaviour of new viruses is insufficiently known, it is essential to appreciate the importance and the impact of non-pharmaceutical interventions, such as social distancing or school closures, which were shown to mitigate transmission . In addition, a fundamental teaching that emerged from the SARS outbreak is the role of nosocomial transmission in shaping the pandemic, as patients admitted for unrelated conditions, or transferred within or between hospitals, significantly contributed to infecting others [59-62].
Not all individuals within a population have the same likelihood to become infected or to infect others. A small number of individuals, known as ‘super-spreaders’, appear to be responsible for most transmission events within a population, and this heterogeneity, known as the 20/80 rule, is increasingly being recognised as a fundamental characteristic of the host-pathogen interaction [63, 64]. For example, when a 9-year-old child from North Dakota was diagnosed in July 1998 with tuberculosis, 56 (20%) of his schoolmate contacts had a positive tuberculin skin test and were presumed to have become infected, yet his twin brother had a mild case and was not infectious . During the SARS outbreak, most patients appeared to have had very low, if any, infectivity, but a minority of the infected people transmitted the virus to a larger number of secondary contacts than most others in the population [66-69]. Relatively little attention has been devoted, historically, to capturing this variability in infectiousness and transmissibility, and this concept, while still insufficiently understood, is increasingly becoming an important facet of public health initiatives [70, 71].
In the wake of the current H7N9 outbreak, and to ensure that we are better equipped to understand transmission in occupational and non-occupational settings, it is imperative to analyse influenza virus biology in context of the triad that incorporates and integrates knowledge about the pathogen, the host and the environment. Teachings from previous epidemics and pandemics, along with those from the current outbreak, will complete and define a framework that should actively fuel public health initiatives and shape epidemic and pandemic preparedness plans. More than ever before, these are becoming increasingly important, challenging and rewarding tasks that define the current era of the global village .