Influenza viruses are members of the Orthomyxoviridae family. This family was recently updated by the International Committee on Taxonomy of Viruses (ICTV, 2017) and at the moment accommodates seven genera (and nine species): Alpha influenza viruses (IAV), Beta influenza viruses (influenza B virus), Delta influenza viruses (influenza D virus), Gamma influenza viruses (influenza C virus), Isavirus (Salmon isavirus), Quaranjavirus (Johnston Atoll quaranjavirus, Quaranfil quaranjavirus) and Thogotovirus (Dhori thogotovirus and Thogoto thogotovirus) (ICTV, 2017). Influenza C, D, and Thogoto viruses may very rarely cause infections in humans (Lambert et al., 2015; White et al., 2016; Nesmith et al., 2017). Contrarily, incursions of influenza A and B viruses in humans are frequent, as these virus cause yearly epidemics, being relevant pathogens for human health. According to the World Health Organization, these viruses are important contributors of lower respiratory infections and in 2016 occupied the 4th place of cause of death worldwide. Influenza virus A and B genera differ in host range and pathogenicity, and IAVs are the only viruses that pose a significant risk of zoonotic infection, host switch, and the generation of pandemics (Kuiken et al., 2006; Horman et al., 2018; Mostafa et al., 2018), being the focus of this review. Influenza A is a zoonotic virus, widespread throughout highly diverse ecological niches as it infects various animals including humans, pigs, bats, horses, and wild aquatic birds, the latter being the primary reservoir for most IAV (Yoon et al., 2014). The diversity among IAV has led to the establishment of categories based on the viral antigens, i.e., the two main proteins at the surface of the virus: hemagglutinin (H or HA) and neuraminidase (N or NA). There are 18 different hemagglutinin and 11 neuraminidase subtypes (Tong et al., 2013), with 16 HA and 9 NA being found in aquatic birds (Webster et al., 1992) and the remainder in bats. Influenza A in its reservoir infects the intestinal tract, causing mild to asymptomatic disease although with some cost to the animal, including reduced migratory capacity (van Gils et al., 2007). Interspecies transmission is thought to occur mainly from wild to domestic birds (Shortridge, 1992), from where it disseminates to other species (Lam et al., 2013). Many adaptations are required to permit establishment in different hosts which have not been fully understood (Schrauwen and Fouchier, 2014). In hosts other than the reservoir, the disease is more confined to the respiratory tract, although, in cases of highly pathogenic and virulent strains a systemic infection could be established. The high frequency of circulating influenza viruses in diverse animal populations increases the potential for emergence of pandemic virus. These strains have, however, thus far been limited to interspecies transmission from avian and swine viruses to humans (Taubenberger and Kash, 2010; Vijaykrishna et al., 2011).

This year marks the 100th anniversary of one of the deadliest pandemic outbreaks, the 1918 Spanish Flu that was caused by an avian IAV. This virus infected one third of the world population (500 millions), provoking 20–50 million deaths (Johnson and Mueller, 2002). Unusual circumstances underpinned the deadly proportions of the outbreak, including the first world war that devastated populations, causing poverty, malnutrition, poor sanitation, overcrowding and massive troop movements, as well as excessive contact with animals (Wever and van Bergen, 2014). In addition, lack of suitable therapeutic and health care means, and absence of pre-existing immunity to the zoonotic virus, combined with an unusual pathogenicity of the viral strain contributed to excess mortality and spread of the disease. Since then, three additional influenza pandemics of severe consequences have been reported: the 1957 H2N2 ‘Asian Flu’ pandemic (0.7 million–1.5 million deaths) (Viboud et al., 2016), the 1968 H3N2 “Hong Kong Flu” (1 million deaths) (Mathews et al., 2009), and most recently, the 2009 novel H1N1 virus causative of “Swine flu” (151,700–575,500 deaths) (Dawood et al., 2012). Of note, pandemic outbreaks have been and are still impossible to predict. Key aspects favoring the incursions of influenza in humans include its, already mentioned, widespread distribution, and viral evolution. Establishment of a virus in a specific host requires adaptation to that host, rendering viral sub-types restricted to particular animal species (Taubenberger and Kash, 2010; Schrauwen and Fouchier, 2014). IAV evolution is fast with different contributor factors (Lowen, 2018). The viral polymerase has low replicative fidelity that results in an average of 2–3 mutations per replicated genome, and therefore the viral progeny consists of genetically diverse population (termed quasispecies) (Pauly et al., 2017), containing many defective virions. The increased risk of producing defective or semi-infectious progeny viruses associated with the viral polymerase is largely compensated by the emergence of escape mutants that bypass neutralizing antibodies, host immunity and antiviral drugs. In addition, the number of quasispecies is increased by the segmented genome. The segmentation of the viral genome permits mixing of different parental strains in co-infections, a process known as reassortment. Reassortment originates chimeric viral genomes, where one (or more) entire segments are swapped when two different viruses co-infect the same cell. The genetic mixing in infected cells can lead to the emergence of viruses with pandemic potential, especially when genome recombination occurs between viruses adapted to different species. The advantages of having an eight-partite genome are evident for viral evolution, although the segmented genome increases the complexity of the assembly of fully infectious virions. The way this genome assembles is very relevant to human health and, in part, the topic of this review. The subject is controversial and has many unresolved questions, despite decades of dedicated research. Here, I will explore the knowns, unknowns and currently under debate views, and for this I will briefly introduce some aspects of the structure of the virion, lifecycle and mechanisms of viral genome assembly.

Lab adapted influenza A viral strains are typically small and spherical, with diameters of ∼100 nm (Figures 1A,B). Clinical isolates, however, when initially analyzed in the lab display a range of morphologies with the prevalent form being spherical particles, but presenting many other morphologies described as filaments [for comprehensive review (Dadonaite et al., 2016)] and bacilliform (Vijayakrishnan et al., 2013) (Figure 1C). Spherical and bacilliform virions contain, in 80% of the cases, complete genomes (Nakatsu et al., 2016). Filaments can be very long (over 10 μm) and homogeneous in width (100 nm), or heterogeneous with a bulbous head with a diameter greater than 200 nm known as Archetti body (Archetti, 1955), and variable lengths. Interestingly, many of these irregular shaped and long budding virions lack viral genomes (Vijayakrishnan et al., 2013).

The virion illustrated in Figure 2A contains a host derived plasma membrane that harbors three viral transmembrane proteins unevenly distributed and present in non-stoichiometric amounts. HA is involved in viral entry, being the most abundant of the three and is a glycoprotein that forms trimers (Copeland et al., 1986). NA cleaves sialic acids being involved in viral exit. It is also a glycoprotein, the second in abundancy, and forms tetramers that cluster (Harris et al., 2006; Calder et al., 2010; Wasilewski et al., 2012; Hutchinson et al., 2014; Chlanda et al., 2015). The proton channel M2 is the least abundant viral transmembrane protein (Hutchinson et al., 2014), playing a crucial role in the release of viral material from endosomes (Kemler et al., 1994). Beneath the lipid membrane is a layer of the viral protein M1 that plays a determinant role in virion morphology (Roberts et al., 1998). M1 oligomerizes to form a helical matrix with a range of curvatures (Calder et al., 2010). The virion core accommodates a segmented viral genome composed of eight segments, and will be described in detail below. The genome displays an arrangement with seven external segments surrounding a central RNA piece (7+1 configuration) in Figure 2B (top view) (Oxford and Hockley, 1987; Noda et al., 2006). In most cases, the eight segments are found hanging from the tip of the virion and it is still controversial if they can adopt anti-parallel orientations (Sugita et al., 2013). IA virions incorporate selected host proteins, some shared by all influenza strains, and others strain and host specific (Hutchinson et al., 2014).

The genome is divided into eight segments of single-stranded, negative-sense RNA that vary in length from ∼0.9 to ∼2.3 kb having a total size of ∼13.5 kb. Each segment (illustrated in Figure 2A) is encapsidated as individual rod-like vRNPs (Pons et al., 1969) of 10–15 nm in diameter and 30–120 nm in length (Compans et al., 1972; Noda et al., 2006). All vRNPs are composed of one copy of the heterotrimeric RNA-dependent-RNA-polymerase (RdRp) formed by PB1 (catalytic unit), PB2 and PA (priming transcription units) (Horisberger, 1980; Honda et al., 1990; Guilligay et al., 2008; Yuan et al., 2009), viral nucleoprotein (NP) and viral RNA. The 5′ and 3′ ends of the RNA are composed of conserved RNA sequences of 13 and 12 nucleotides that display partial inverted complementarity and base-pair in an antiparallel manner to form an approximately 15-base pair-long panhandle (Detjen et al., 1987), where the RdRp assembles (Murti et al., 1988). As a result, a double-helical hairpin structure is formed, with a loop at one end folding back on itself (Pons et al., 1969; Hsu et al., 1987). The double-helix contains major and minor grooves caused by the oligomerization of NP that forms a trail of positive residues enabling binding of RNA to the outside of the complex (Ye et al., 2006, 2012; Arranz et al., 2012; Moeller et al., 2012), but rendering vRNA susceptible to RNAse digestion (Yamanaka et al., 1990). Studies showed that 2 molecules of NP interact with PB1 and PB2 (Biswas et al., 1998; Martin-Benito et al., 2001; Arranz et al., 2012; Moeller et al., 2012). The periodicity of NP on RNA is ∼21 nucleotides (Williams et al., 2018). Recent studies show that the ssRNA is coiled around the NP monomers in an unevenly fashion, with some regions tightly bound to NP and others with NP dynamically turning over and therefore exposed (Moeller et al., 2012; Gallagher et al., 2017; Lee et al., 2017; Williams et al., 2018). High-NP content is prevalent on G-rich and U-poor RNA regions, indicating some sort of sequence bias for NO binding as analyzed by HITS-CLIP (Lee et al., 2017) and low-NP content is found in predicted RNA secondary structures by PAR-CLIP (Williams et al., 2018). These regions were proposed to be functionally linked to efficient vRNP packaging in virions (Lee et al., 2017; Williams et al., 2018) with synonymous mutations attenuating influenza replication (Williams et al., 2018).

vRNPs are individual replication units of a specific segment, being able to transcribe and replicate the encoded material under appropriate conditions (Pleschka et al., 1996; Mullin et al., 2004). The way vRNPs are incorporated or packaged in a virion, assembled into a genome, and transmitted/received by neighboring cells/individuals is a key parameter in viral perpetuation. In the case of influenza, as mentioned, this is particularly challenging as the viral genome is segmented, which imposes additional layers of complexity to the assembly of the viral genome. Two models were proposed for the inclusion or packaging of individual vRNPs into budding virions: the random and the selective models. The first model implies that a complete genome is formed by chance without any mechanism to distinguish between different segments. The selective model accommodates a mechanism whereby each virion packages one copy of each unique vRNP (Kingsbury, 1970; Compans et al., 1972). To understand the current situation where the data overwhelmingly favors the selective packaging model, readers are remitted to this comprehensive review (Hutchinson et al., 2010). Briefly, studies demonstrating competition between full-length segments and defective interference particles (segments that have internal deletions) (Davis et al., 1980; Jennings et al., 1983; Duhaut and McCauley, 1996; Duhaut and Dimmock, 1998), and genetic and biochemical analyses revealing that the influenza A genome is haploid, started paving the way favoring the selective model (Hatada et al., 1989; Fujii et al., 2003). Other important contributing studying used reverse genetics, mapped regions responsible for the packaging of specific segments (Duhaut and Dimmock, 2000; Muramoto et al., 2006; Gog et al., 2007; Hutchinson et al., 2008, 2009; Essere et al., 2013; Gavazzi et al., 2013b). Final evidence in favor of the selective model was the visual inspection of virions by electron microscopy/tomography revealing eight segments inside the virion (Noda et al., 2006; Fournier et al., 2012b; Noda et al., 2012; Sugita et al., 2013) with no indication of incorporation of excess number of segments (Harris et al., 2006; Yamaguchi et al., 2008). Seminal studies have captured a 7+1 configuration of vRNPs inside budding virions at the plasma membrane (see Figure 2B). The differences in vRNP length (as a consequence of different number of nucleotides) permitted ranking vRNPs by length into three categories by 3D-tomography: Segments 1 – 3 as long; segments 4 – 6 as medium and segments 7 and 8 as small. Assessment of segment length indicated that the 80% of virions incorporated one complete genome set (Nakatsu et al., 2016). Despite the strong evidence for a stringent, selective model, these data do not constitute proof of either a specific layout of vRNPs in virions or of the establishment of inter-segment RNA interactions. Recently, however, interactions between RNPs have been observed suggesting that the genome actually forms a supramolecular complex (Fournier et al., 2012a,b; Noda and Kawaoka, 2012), held together by base pairing of packaging signals on RNA (Fournier et al., 2012b). Until the present date, it is accepted that RNA–RNA (Essere et al., 2013; Gavazzi et al., 2013a,b), but not protein-RNA interactions, guide genome packaging. Mechanistically, it was proposed that secondary structures on vRNPs (Ferhadian et al., 2018), and specially in conserved regions of low-NP content could allow the establishment of non-covalent inter-segment interactions likely by hybridization (Lee et al., 2017; Williams et al., 2018). After released from the cells, vRNPs in isolated virions studied by cryo-tomography were shown to maintain the double-helical structure and remain attached to one end of filamentous particles (Harris et al., 2006; Calder et al., 2010; Wasilewski et al., 2012; Vijayakrishnan et al., 2013). In terms of morphology, RNP filaments varied in length, curvature and handedness of the helix, not being uniformly rigid structures (Arranz et al., 2012; Moeller et al., 2012; Gallagher et al., 2017) on account of NP-NP interactions (Gallagher et al., 2017).

Influenza A virus initiates infection upon binding of hemagglutinin to sialic acid on transmembrane proteins on the host cell surface, entering via receptor-mediated endocytosis. As pH drops during endosome maturation, the membranes of the virus and of the endosome fuse and the proton pump activity of the viral protein M2 allows the release of vRNPs to the cytosol. vRNPs are subsequently transported to the nucleus where transcription and replication of influenza RNA occurs. As already mentioned, each segment constitutes an independent transcription-replication unit equipped with its own polymerase (Wise et al., 2009; Dou et al., 2018), with the same vRNA template being transcribed into capped and polyadenylated viral mRNA or copied as a complementary RNA (cRNA). cRNA will then serve as template to produce multiple copies of vRNA as reviewed in (Fodor, 2013; Hutchinson and Fodor, 2013; Te Velthuis and Fodor, 2016; Pflug et al., 2017). vRNP nuclear export requires a daisy chain reaction using the viral proteins M1 and NS2 and the cellular protein CRM1 (Martin and Helenius, 1991; O’Neill et al., 1998; Neumann et al., 2000; Elton et al., 2001; Watanabe et al., 2008, 2014). Upon exiting the nucleus, vRNPs accumulate around the microtubule organizing centre (MTOC) (Momose et al., 2007; Bruce et al., 2010; Jo et al., 2010; Amorim et al., 2011; Eisfeld et al., 2011a; Momose et al., 2011; Avilov et al., 2012b). Subsequently, vRNPs are transported to the plasma membrane by unclear mechanisms. Together with host and the other viral major structural proteins (M1, M2, HA, and NA), vRNPs are packaged into progeny virions, which bud and are released from the cell [reviewed recently in (Pohl et al., 2016; Dou et al., 2018)].

Transport of the vRNPs to the cell surface, assembly of influenza A genome into a supramolecular complex, virion budding and release from the cell have been the subject of many recent seminal manuscripts, many of which involving the host GTPase Rab11 [as reviewed in (Hutchinson and Fodor, 2013; Gerber et al., 2014; Eisfeld et al., 2015; Giese et al., 2016; Lakdawala et al., 2016; Pohl et al., 2016; Dou et al., 2018)]. Interestingly, Rab11 is emerging as a key factor subverted by evolutionary unrelated viral families (Paramyxo, Bunya, and Orthomyxoviruses, among many others) and bacteria (Salmonella and Shigella) of relevance to human health. The interactions and role of Rab11 in bacterial and viral infections are explored in these reviews (Guichard et al., 2014; Vale-Costa and Amorim, 2016). Importantly, Rab11 function in several infections might hold common biological principles that could be explored therapeutically. It is therefore important to obtain a roadmap of Rab11 behavior and interactions for each infection. This review deals with the current knowledge, challenges and controversies of the interaction between Rab11 with IAV.

Rab11 is a member of a large family of GTPases called Ras-related in brain (Rab) with 44 subfamilies encoded in the human genome (Diekmann et al., 2011). The activity of every Rab depends on GDP/GTP association: the GTP-bound active form and the GDP-bound inactive form. A guanine nucleotide exchange factor (GEF) catalyzes GTP-binding and causes a conformational change. The GTP-bound conformation allows recruitment of molecular motors, tethers and SNAREs to drive, dock and fuse, respectively, vesicles to their cognate membranes (Stenmark, 2009). It is the association of lipids and proteins in the vesicles with the ones in organelles that ensures the high fidelity of the vesicular transport, and delivery of material where and when is needed [reviewed in (Vale-Costa and Amorim, 2016)]. During transport, tethering and fusion, Rabs are converted back to the GDP-bound form through hydrolysis of GTP, stimulated by a GTPase-activating protein (GAP) [with one inorganic phosphate (P_i) being released]. In the GDP-bound form, Rabs can be recycled back to specific membranes via the action of many different enzymes (Diekmann et al., 2011; Hutagalung and Novick, 2011).

Rab11 subfamily is composed of Rab11 a, b, and c, the latter also called Rab25. Rab11a and b, that share 89% homology, have been implicated in IAV lifecycle, although Rab11a has been studied in more detail and for this reason all subsequent mentions of Rab11 refer to the Rab11a isoform. In uninfected cells, Rab11 is the master regulator of the endocytic recycling compartment (ERC), one of the systems the cell uses for delivering endocytosed material, as well as, specific cargo from the TGN, to the cell surface (Grant and Donaldson, 2009; Vale-Costa and Amorim, 2016). It has been shown that in healthy cells, the ERC derives from early endosomes by the action of KIF13A (Delevoye et al., 2014) and actin (Ripoll et al., 2016) and whether there is a pool of ERC-vesicles derived from the TGN formed in a distinct manner is not known.

In healthy conditions, Rab11-GTP recruits a series of molecular motors using molecular adaptors called Rab11-family interacting proteins (FIPs) (Horgan and McCaffrey, 2009) to drag vesicles on cytoskeletal tracks, but also to attach vesicles to transition zones between microtubules and actin. Rab11 containing vesicles tether to the plasma membrane using SEC15, a member of the exocyst, or Munc13-4. Finally, the fusion step at the plasma membrane involves SNAP25 (a SNARE), and/or SYN4 and/or VAMP8 (Wallace et al., 2002; Aikawa et al., 2006; Sato et al., 2008; Collins et al., 2012; Schafer et al., 2014; Marshall et al., 2015; Johnson et al., 2016).

Many Rabs, including Rab5, Rab7, and Rab9 were shown to be important for IAV infection. Assessment of their role was straightforward, as their depletion would affect viral entrance (Lakadamyali et al., 2003). Conversely, the function of Rab11 in IAV infection has been difficult to demonstrate and is still under debate. Rab11 was shown to affect viral production, especially when Rab11a and Rab11b were knocked down simultaneously, being stipulated that these factors are required for efficient viral replication (Bruce et al., 2009). When assessing the step in which Rab11 was required, studies have consistently found defects only in the late stages of infection. It was proposed that Rab11 transported vRNPs and promoted assembly of the genome and that the processes were intimately related. However, the molecular mechanisms bridging the two steps in the viral lifecycle are controversial. For simplicity, I will separately introduce involvement of Rab11 in vRNP transport (see section “Rab11 and vRNP Transport”) and IAV genome assembly (see section “Briding vRNP Transport With Genome Assembly”) and will finalize exploring the bridging mechanisms that are currently under evaluation (see “Briding vRNP Transport With Genome Assembly”).

In recent years, exciting reports have contributed to advances in understanding IAV genome assembly. It has become accepted that influenza A genome forms a supramolecular complex and basic principles are expected to apply to engineer the 7+1 arrangement vRNPs adopt inside a virion. Being able to predict these principles is important to determine next years’ epidemics strain, as well as foresee viruses with pandemic potential. These studies have traditionally been done by studying vRNP structure, vRNP inter-segment biochemical interactions and assessing geographical locations of specific vRNP types in budding or isolated virions of many different IAV strains and in distinct hosts. However, knowledge about IAV genome assembly also requires fully tackling the molecular mechanisms underlying its formation inside the host cell. Several reports have suggested a model coupling IAV genome assembly with vRNP transport involving the host protein Rab11, whereby “kissing” between vesicles resulted in the establishment of inter-segment interactions. However, this model conflicts with several experimental results and requires more investigation. In particular, a bulk of evidence suggests that Rab11 is modified by infection but the outcome is ill-defined. Recent advances on cellular compartmentalization are revolutionizing the way we understand how spatial organization promotes biological reactions in time and in a densely populated cytosolic environment. This emerging topic might permit explaining sophisticated phenomena such as assembly of segmented viral genomes. More insights will arise from developing approaches to overcome some of the technical limitations we current have including tracking individual vRNPs, or several organelles, or resolving assembled genomes inside the living cell. Finally, innovative models to study viral infection, in a way that takes into account cell polarization without compromising the resolving power inside individual living cells, will also be crucial to understand how proteins that promote trafficking and establish cell polarization are involved in viral assembly.

MJA wrote the manuscript.

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.