The influenza RNA synthesis machine

Abstract

The influenza A viruses are the causative agents of respiratory disease that occurs as yearly epidemics and occasional pandemics. These viruses are endemic in wild avian species and can sometimes break the species barrier to infect and generate new virus lineages in humans. The influenza A virus genome consists of eight single-stranded, negative-polarity RNAs that form ribonucleoprotein complexes by association to the RNA polymerase and the nucleoprotein. In this review we focus on the structure of this RNA-synthesis machines and the included RNA polymerase, and on the mechanisms by which they express their genetic information as mRNAs and generate progeny ribonucleoproteins that will become incorporated into new infectious virions. New structural, biochemical and genetic data are rapidly accumulating in this very active area of research. We discuss these results and attempt to integrate the information into structural and functional models that may help the design of new experiments and further our knowledge on virus RNA replication and gene expression. This interplay between structural and functional data will eventually provide new targets for controlled attenuation or antiviral therapy.

Introduction

The influenza disease appears as yearly epidemics and occasionally more severe pandemics. The most relevant pandemic occurred in 1918 and was responsible for 20–40 million deaths. Later pandemics appeared in 1957 and 1968 and the most recent one started in 2009.^1,2 Influenza type A viruses, members of the Orthomyxoviridae family, cause most of these epidemics and all pandemics. They constitute an extremely heterogeneous virus population comprising many subtypes, which are defined as any combination of their surface glycoproteins, haemaglutinin (HA) and neuraminidase (NA).³ Sixteen HA and nine NA classes have been recognised. Viruses containing many of their combinations can be isolated from aquatic and terrestrial avian species, which constitute their natural reservoir. Transfer of some of these viruses to humans, or transfer of genes from avian to human influenza viruses, originated the influenza pandemics and caused the establishment of new lineages of human viruses.^1,4 Indeed, sporadic cases of highly pathogenic H5N1 avian influenza in humans continue to occur since 2003 and pose a risk for the origin of a devastating new pandemic (http://www.who.int/csr/disease/avian_influenza/en/).

The genome of the influenza A viruses comprise 8 single-stranded RNA molecules of negative polarity that form ribonucleoprotein complexes (vRNPs) by association with the virus polymerase and many nucleoprotein (NP) monomers.^5,6 These vRNPs are independent functional units during transcription and replication. Within the RNPs, the polymerase complex is responsible for both transcription and replication.^7,8 The virus enters the cell by receptor-mediated endocytosis and releases the genomic RNPs into the cytoplasm by acid pH-dependent membrane fusion at the late endosome.³ Contrary to most RNA-containing viruses, the influenza viruses transcribe and replicate their genome in the nucleus of the infected cells and hence depend on host nucleo-cytoplasmic trafficking and nuclear functions to carry out these processes. Upon entry into the nucleus, parental RNPs are first transcribed (primary transcription) and the synthesis of new virus proteins is necessary to proceed to RNA replication.⁹ The transcription process is initiated by cap-snatching, i.e., the cleavage of 5′-capped RNA fragments from host premRNAs. 10 Subsequently, these are used as primers to copy the template and the mRNAs are finally polyadenylated. Virus RNA replication occurs in two steps: complementary RNAs (cRNAs) of positive polarity are synthesized first and serve as replication intermediates to generate large amounts of progeny virus vRNPs. These can then act as templates for late (secondary) transcription and eventually are exported from the nucleus and become incorporated into virions at the plasma membrane.^5,6

Although the basics on influenza virus replication and transcription have been known for decades, a detailed understanding on the structure of the viral transcription and replication apparatus and the mechanisms underlying these processes is still lacking. In addition, genetic analysis of the adaptation of avian viruses to the mammalian hosts has revealed that host-dependent specific interactions of the virus polymerase and RNP with cellular factors are at the basis of an efficient replication in humans of influenza viruses transferred from the avian reservoir.^11–15 Here we summarise new structural, biochemical and genetic evidences that have set the stage for a more profound understanding of influenza virus transcription and replication, and discuss alternative models to describe the mechanisms of these processes (reviewed in ref. 16–18). This new knowledge will help in the identification of novel targets for antiviral treatment of seasonal and pandemic influenza and the development of new attenuated virus strains.

The Virus RNA Synthesis Machine

In contrast to the viruses containing positive-polarity RNAs as genome, the genome of influenza viruses, as those of other negative-polarity RNA viruses, are ribonucleoprotein complexes and not naked RNAs. The viral polymerase cannot replicate a normal virus RNA in the absence of the NP (but see below). Each of the eight viral RNPs behaves as an independent replication and transcription unit. This is at the basis of the high frequency of gene reassortment in doubly-infected cells.¹⁹ Indeed, minireplicon systems have been available since long ago, in which a single virus-like RNA serves as a model template that can replicate and transcribe in vivo with the sole trans-action of the polymerase and NP.^20–23

When isolated from purified virions, the genomic RNPs appear as ribbon-like, closed superhelical structures.^24–27 The structural features of these RNPs are in part due to the NP, as in vitro RNA-NP complexes resemble native RNPs²⁸ and even RNA-devoid NP can form similar structures.²⁹ The maintenance of the closed structure depends on the presence of the polymerase.³⁰

The heterogeneity and flexibility of the purified virion RNPs precluded their detailed structural characterisation, but immunoelectron microscopy documented the presence of the polymerase at one of the ends of the helical structure.³¹ The generation and purification of a panel of recombinant RNPs containing deleted versions of viral RNA segment 8 allowed for the first time to carry out two-dimensional structural analysis of the RNP.³² These studies indicated that the RNPs could adopt circular, elliptical or supercoiled structures depending on the length of the template RNA. They also allowed for the first time the visualisation of the RNP-associated polymerase and determined that the average length of the RNA associated to each NP monomer was 24 nt, in agreement with previous biochemical estimates.³³ The three-dimensional structure of one such recombinant RNP (NS clone 23) was determined by electron microscopy and image processing from negative-stained and frozen samples.^34,35 These recombinant RNPs, containing a deleted version of viral NS RNA segment with a length of 248 nt, were generated by amplification in vivo and were able to carry out transcription in vitro.³⁴ Furthermore, similarly prepared recombinant RNPs could be rescued into infectious virus.³⁶ Thus, the RNPs analysed structurally contain negative-stranded virus RNA³² and are probably equivalent to the progeny vRNPs that accumulate in infected cells and are eventually incorporated into virions.

The RNP structure reported shows a ring-like configuration including 9 NP monomers, two of which are connected to the polymerase complex by non-equivalent interactions (Fig. 1A). A resolution of 12 Å was estimated for the NP ring, while the polymerase and the adjacent NP monomers showed 18 Å resolution. At this stage, it is not possible to establish the precise location of the template RNA in the structure, but the association of the polymerase was RNAse-sensitive,³⁴ indicating that a considerable fraction of the interactions of the enzyme to the rest of the RNP is RNA-dependent. The atomic structure of the influenza NP has been solved in the form of a trimeric complex and a RNA-binding region has been proposed at a cleft between the body and the head in the structure.^37,38 However, the length of template RNA associated to each NP monomer is much longer than the capacity of the proposed RNA-binding site, suggesting that the viral RNA also interacts elsewhere on the NP. The atomic structure of the NP could be reasonably fitted into the volume of a NP monomer in the RNP,³⁴ suggesting that it is well preserved in a biologically active RNP (Fig. 1B). Furthermore, the NP-NP connecting loop, proposed as the main interaction between NP monomers in the crystal,^37,38 was shown to be relevant for the replication capacity of the recombinant RNP solved by electron microscopy.³⁴ However, other NP-NP interactions, not apparent in the atomic structure of the trimeric complex solved,^37,38 may be relevant for the stability and/or the dynamics of the circular mini-RNP structure or the full-length helical RNPs.

Figure 1.

Structure of an influenza virus ribonucleoprotein complex. (A) A perspective view and two side views of the three-dimensional structure of a recombinant RNP; (B) A top view of the same structure with the atomic structure of the NP docked (pdb 2IQH) into each monomer; (C) A front view of the polymerase complex present in the recombinant RNP with a proposed docking of the atomic structure of the PA(C)-PB1(N) domain (pdb 2ZNL).

Interactions between the NP and polymerase subunits PB1 and PB2 have been identified biochemically,^39,40 but so far it is not possible to correlate such information with the polymerase-NP contacts detected in the cryo-EM structure of the mini-RNP.³⁴ In fact, it is not known whether the interactions documented biochemically are relevant to the structure of the RNP or rather represent transient interactions that may be important for the function of the RNP during transcription and/or replication. In this context it is important to stress that the cryo-EM structure published represent a “fixed picture” of an RNP that, although potentially active, was not synthesizing RNA when imaged in the frozen state.

In spite of the limitations mentioned above, the cryo-EM structure of an influenza mini-RNP is the best structural information available for any functional transcription/replication unit of a negative-stranded RNA virus and should serve as a framework for the future generation of a quasi-atomic RNP model and to propose testable models for RNA transcription and replication. Such a quasi-atomic model should shed light on the interactions of the NP with the RNA template and the polymerase complex, as well as on the relevance for mRNA synthesis of the precise position of the polyadenylation signal within the structure. However, a more refined structure of the mini-RNP would not provide information on the conformation of the helical section of full-length RNPs. This information is critical to define the interaction among RNPs during the encapsidation mediated by the packaging signals. The convergence of cryo-EM of viral RNPs,³⁴ cryo-tomography of virions^41,42 and genetic analysis of RNP packaging (reviewed in ref. 43) should provide a clearer view of the influenza genome encapsidation into infectious virus.

The Polymerase Complex

The influenza RNA polymerase complex is made up of three subunits, PB1, PB2 and PA and has an aggregate molecular mass of around 250 kDa. The phenotype of classical temperature-sensitive mutants suggested that PB1 is responsible for all RNA synthesis, while PA and PB2 would be involved in RNA replication and transcription, respectively (reviewed in ref. 44). Site-directed mutations verified these hypotheses^45–47 but also showed that PB2 has a role in replication⁴⁸ and PA is involved in cap-snatching.⁴⁹ These genetic data and the results on the in vivo and in vitro activity of recombinant virus polymerase indicate that the complex is the functional enzyme during both transcription and replication and its activity is probably modulated by interaction with the RNA templates, primers and other virus and host factors.

The interaction between subunits has been studied by a combination of biochemical, genetic and structural approaches. These experiments indicated that PB1 is the core of the complex⁵⁰ and it interacts with PA by means of its N-terminal end and with PB2 through its C-terminal end^51–54 (Fig. 2). Some of these interactions have been verified by co-crystallisation^55–57 and functional studies.⁵⁸ Also, some additional PB2-PA and PB1-PB2 interactions have been described in reference ⁴⁰ and ⁵⁹, that may play structural and/or regulatory roles in the activity of the viral polymerase.

Figure 2.

Diagram of intersubunit interactions in the influenza polymerase complex. The diagram represents the polymerase subunits as bars in which the N- and C-terminal regions are indicated. The domains of each subunit whose atomic structure has been determined are indicated by underlined regions and the corresponding structures are presented (PA: Yellow; PB1: Blue; PB2: Green). The interaction domains identified biochemically that have been verified by co-crystallisation are indicated by boxes connected by dotted lines. Relevant functional domains (Endonuclease, RdRp and cap-binding domains) are also highlighted.

The three-dimensional structure of the influenza RNA polymerase has been determined by electron-microscopy of negative-stained samples, including polymerase associated to recombinant RNPs, RNA-devoid soluble polymerase and polymerase-vRNA template complexes generated by in vivo replication (Fig. 3).^34,60–62 These studies documented that the polymerase has a quite compact structure and is probably a flexible complex. Thus, comparing the isolated heterotrimeric complex with the RNP-bound or template-bound complexes indicates that the polymerase shows considerable conformational changes upon interaction with other viral (and presumably also cellular) partners. For the RNP-associated polymerase, the location of specific subunit domains within the structure could be determined by 3D-reconstruction of monoclonal antibody-bound or tagged RNPs.⁶⁰ The best available structure to date was obtained by cryoEM and 3D-reconstruction of polymerase associated to RNPs.³⁴ The structure has a resolution of 18 Å and shows welldefined domains, some of which could be specifically assigned to particular subunits by docking known atomic structures (Fig. 1C).³⁴

Figure 3.

Structural comparison of RNP-associated and soluble polymerase complexes. Several views are presented to compare the structures of the soluble heterotrimer devoid of template RNA (top, pink),⁶² the polymerase complex present in the recombinant RNP (middle, yellow),³⁴ and a polymerase-vRNA complex (bottom, blue).⁶¹

Although no atomic structure of the influenza RNA polymerase complex is yet available, several soluble domains of its subunits have been solved so far (Fig. 2). In addition to the subunit interfaces mentioned above, the atomic structures of domains of the PA and PB2 subunits have been determined^{47,55,56,63–67} but no information on the structure of the polymerase active site in the PB1 subunit is yet available. The data reported allowed the identification of the polymerase cap-binding site in PB2 and established that the endonuclease responsible for cap-snatching resides at the N-terminal region of PA, in agreement with previous genetic data.^46,49 The cap-binding domain shows a distinct fold, different from other cap-binding proteins, but the mechanism by which the cap structure is recognised is analogous to those previously described: the methylated base is held within an aromatic sandwich and specific aminoacid residues are responsible for the interaction with the triphosphate moiety and the methyl group.^47,68 The fold of the endonuclease domain and its biochemical properties are similar to those of type II restriction endonucleases.^63,66

In addition of the endonuclease active site, protease activities have been associated to the PA subunit of the polymerase^69–71 and a correlation has been reported between such an activity and the capacity of the polymerase to replicate.^72,73

The mechanism by which the polymerase subunits interact, the site where the trimeric complex is formed and its transport to the nucleus have been analysed in detail. Each of the individual subunits contains a functional nuclear localisation signal (NLS),^74–76 and could be imported into the nucleus independently, but several approaches have suggested that the subunits associate in the cytoplasm before nuclear import. Thus, it has been suggested that PB1 and PA form a heterodimeric complex that is transported to the nucleus by association to RanBP5,^77,78 but other studies suggested a role for Hsp90 in the transport of PB1-PB2 and PB1-PA heterodimers.⁷⁹ Recently, studies of crosscorrelation fluorescence spectroscopy have supported the formation of PB1-PA heterodimers in the cytoplasm and their transport to the nucleus independently of the PB2 subunit.⁸⁰ In addition, mutation analyses of the PB2 NLS indicated that the transport of PB2 to the nucleus and the formation of a functional polymerase complex are closely correlated events¹⁴ and represent host-regulated steps in the virus multiplication cycle.¹³

On the other hand, the polymerase complex may play a role during virus mRNA translation. Although early in vitro studies indicated that the polymerase complex dissociates from the cap-structure soon after cap-snatching,⁸¹ no in vivo studies have analysed whether normal viral mRNAs are recognised by nuclear cap-binding complex before their export to the cytoplasm or whether they associate to the eIF4F initiation factor for their translation. In fact, recent studies indicate that inactivation or depletion of eIF4E does not impair the translation of virus-specific mRNAs. This observation, together with the association of the polymerase to translation pre-initiation complexes suggest that the viral polymerase complex may play a cytoplasmic role during viral mRNA translation.⁸²

To Switch or Not to Switch

The structures of the positive-polarity RNAs synthesized in influenza virus-infected cells, mRNAs and cRNAs, are drastically different (Fig. 4). Thus, the former are capped, contain extra sequences of cellular origin at their 5′-termini, lack viral sequences at their 3′-termini and are polyadenylated. On the contrary, cRNAs are complete copies of their vRNA counterparts, are neither capped nor polyadenylated and become encapsidated by association to polymerase and NP (Fig. 4). These facts suggest that the initiation and termination mechanisms for viral transcription and for (the first step of) replication are distinct and the latter is linked to the encapsidation to cRNAs into RNPs, which will serve as templates for later vRNP amplification. These mechanistic differences have been the basis to propose that the parental RNPs switch from a transcription mode early in the infection to a replication mode later on, when newly synthesized proteins would become available. The very existence of such a switch has been challenged in the late years by complementation experiments in which pre-expression of catalytically inactive polymerase was shown to allow the accumulation of cRNA from infecting RNPs in the absence of protein synthesis.⁸³ These results were interpreted to mean that the parental RNPs can synthesize cRNA by default and were consistent with the possibility to generate cRNA in vitro from purified virion RNPs.⁸⁴ Although the mechanism by which the parental RNP chooses to either initiate de novo or cap-snatch a cellular RNA is not clear in this proposal, it was reported that the concentrations of NTPs required for initiation without a primer are higher than those needed for elongation of a capped-oligonucleotide⁸⁴ and this difference might be at the basis of the discrimination of transcription versus replication by the infecting RNPs. Indeed, high concentrations of all three initiating NTPs (ATP, GTP and CTP) are required for de novo initiation of cRNA on a vRNP template and it was proposed that the initiating NTP during replication is the GTP at position +2.⁸⁵

Figure 4.

Structural and mechanistic differences between influenza virus transcription and replication processes. The diagram shows the structure of a helical vRNP (middle), including the negative-polarity template RNA (red) associated to NP monomers (orange circles). The polymerase subunits are coloured in blue (PB1), green (PB2) and yellow (PA). The promoter is depicted in its corkscrew structure¹²⁶ and the location of the polyadenylation signal (oligo U) is indicated. The structure of the mRNA product of transcription is presented at the bottom, including the capped oligonucleotide primer (boxed in green) and the poly A tail. The structure of the cRNP intermediate of replication in presented at the top, including the positive-polarity template RNA (blue) and the promoter elements.

It is possible that the association of the parental RNP to the cellular RNA polymerase II complex⁸⁶ determines its fate to transcribe and only when new viral polymerase and NP are available would the accumulation of cRNA be detectable.⁸³ However, a mechanism has to be predicted to justify the coupling between de novo initiation, encapsidation into a progeny RNP and overriding premature termination. Likewise, initiation by cap-snatching should determine that the mRNA is not encapsidated into a viral RNP and is polyadenylated. Several models have been proposed to accomplish such differential mechanistic links during mRNA versus cRNA synthesis. The presence of newly synthesized NP is needed to encapsidate the cRNA product of replication, as both RNA-binding and NP oligomerisation are needed to support RNP replication,^34,87 but, in addition, it has been proposed that direct interaction of free NP with the polymerase could be involved in favouring de novo initiation over transcription.^39,40,88 Alternatively, it has been proposed that an interaction of newly synthesized NP with the RNP template could alter the relative ability for de novo versus primed initiation.^30,89,90 However, excess expression of NP did not result in an increase of RNA replication versus mRNA synthesis in a recombinant replicon system.⁹¹ On the other hand, the NEP(NS2) protein, whose major function during infection has been assigned to nuclear export of progeny RNPs,³ was shown to alter the relative efficiency of transcription and cRNA synthesis^92,93 and to be required for the generation of short virion RNAs (svRNAs),⁹⁴ suggesting that it may be a replicase-associated virus factor.

However, none of these models sheds light on the coupling between the initiation and termination (polyadenylation) modes. Recently, a trans-model has been proposed for (the second step of) influenza virus RNA replication, by which a polymerase complex distinct from that resident in the RNP would be responsible for replicative synthesis⁹⁵ (see below). Although no direct evidence has been provided for such model to operate in the first step of RNA replication (cRNA synthesis), this would be an attractive hypothesis that could help understanding the coupling of initiation and termination modes in transcription and replication. Thus, mRNA synthesis would occur in cis, i.e., it would be performed by the polymerase present in the RNP.⁹⁵ Since the polymerase binds specifically the 5′-end of the template,^89,96,97 it would determine a steric block for termination of elongation and hence favour polyadenylation.⁹⁸ On the contrary, if cRNA synthesis would be performed in trans by a polymerase distinct from the RNP-associated polymerase, such a steric hindrance could be released and polyadenylation would be avoided. Further experimentation is needed to test these proposals.

In addition to the trans activities mediated by soluble viral factors to allow cRNA synthesis, evidence has been accumulated to show that interaction with cellular factors is required for efficient virus transcription or replication. Association of the viral polymerase to cellular RNA polymerase II⁸⁶ may improve the availability of capped RNA primers for transcription. On the other hand, the MCM replicative helicase complex was shown to interact with the PA subunit of the polymerase, be relevant for cRNA synthesis in vivo and improve the elongation of nascent cRNA in vitro.⁹⁹

Virus Genome Amplification

In the course of a virus infection, the synthesis of cRNA occurs early and during a short period of time, while vRNA synthesis takes place later and to a much higher extent.^3,9 This observation, together with the phenotype of virus mutants that are temperature-sensitive for vRNA synthesis but show normal accumulation of cRNA at restrictive temperature,^100–102 suggest that most of the cRNA synthesis that takes place in the infection uses the parental RNPs as templates. The large differences in cRNA and vRNA accumulations in infected cells may be a reflection of the strengths of vRNA- and cRNA-driven polymerase complexes. The structures of vRNA and cRNA promoter regions have been determined by NMR.^103–105 Although they show structural similarities, some marked differences are also apparent, that could account for their differential recognition by the virus polymerase. In fact, biochemical and genetic evidences suggest that distinct regions of the polymerase interact with vRNA or cRNA promoter regions.^96,106 This is in line with the proposal that initiation on cRNA- or vRNA-containing polymerase complexes is carried out by distinct mechanisms:¹⁰⁷ Whether synthesis of cRNA is initiated using as template the 3′-terminus of the vRNA template, initiation of vRNA synthesis occurs three nucleotides downstream and the pppApG initial dinucleotide should re-align to the template terminus to lead to complete replication. These observations, as well as the phenotype of temperature-sensitive and engineered NP mutants that are defective in either cRNA or vRNA synthesis^88,102 suggest that the vRNP and cRNP replication complexes are structurally and functional distinct. These differences may be at the basis of the diverse kinetics and extent of cRNA and vRNA synthesis during infection. Other virus factors, as for instance NS1 protein, may also be involved in virus genome amplification, as temperature-sensitive mutations in this protein results in diminished vRNA but not cRNA accumulations at restrictive temperature.^100,108

Another important difference between vRNPs and cRNPs refers to their capacity to perform cap-snatching. Both parental and progeny vRNPs should carry out cap-snatching during primary and secondary transcription, respectively. However, cRNPs appear not to be able to generate mRNA-like transcripts, although it is not clear at which point the restriction occurs. It has been reported that cRNPs cannot perform the cap-snatching reaction,^109,110 but other authors propose that such a step is still functional for cRNPs but the capped-oligonucleotides cannot be used as primers.¹¹¹ Nevertheless, the existence of a conserved open reading frame in the NS segment negative polarity RNA and its capacity to encode a stable protein from a recombinant vector, has been reported in reference ^112–114. Whether such a potential new virus gene, which should be transcribed from a cRNP, is expressed in influenza virus-infected cells is still an open question.

Although amplification of a vRNP from a cRNP template would appear as simpler than the opposite, several mechanistic aspects have been unclear up to recently. The presence of NP is required for vRNA accumulation in vitro^115,116 and in vivo,^88,102 although replication of short RNA templates can proceed in the absence of NP.^61,117 The incorporation of NP into new vRNPs is facilitated by the activity of cellular factors UAP56 and Tat-SF1, that may act as chaperones.^118,119 Viral polymerase should also be required stoichiometrically, as the end product is not vRNA but a vRNP. The question remains whether the polymerase complex associated to the progeny vRNP is identical or distinct to the polymerase that actually performs the replication reaction. Recent complementation experiments in vivo, using polymerase complexes that are defective in either transcription or replication, have provided strong evidence to establish that a polymerase complex different from that acting during elongation of vRNA is incorporated into the progeny vRNP.⁹⁵ These results allowed the rescue of recombinant vRNPs that are defective in replication and opened the way to test whether the polymerase complex associated to the cRNP is responsible for the synthesis in cis of vRNA or is a different, soluble polymerase complex which carries out replication in trans. Again, the results of complementation experiments in vivo documented that replication occurs in trans.⁹⁵ Although it has been reported that alternative binary complexes of the polymerase subunits might be responsible for transcription or replication,^120,121 a number of other reports indicate that the polymerase heterotrimer is the functional unit in both processes.^{8,48,49,63,66,122} Altogether, the genetic approaches indicated above enabled the proposal of a new model for influenza RNA replication, which is depicted in the diagram presented in Figure 5. Activation of vRNA synthesis would result from the interaction of a soluble polymerase complex (Non-resident polymerase-NRP1-) with the cRNP (Fig. 5A and B). Interaction among polymerase complexes has been documented previously in reference ^{50, 61} and ¹²³. Whether the NP or cellular factors are involved in this interaction is presently unknown. As a result, the incoming polymerase would gain access to the 3′-terminal sequences of cRNA and start vRNA synthesis (Fig. 5B). The recent description of short virion RNAs (svRNAs) in infected cells, that are associated to the polymerase,⁹⁴ provide a possible mechanism by which the NRP1 could recognise the 3′-terminal sequence of the template. Thus, the polymerase-svRNA could be the trans-acting replicase whereas the RNP resident polymerase (-RP-, Fig. 5B) would serve as template-recognition factor but would not participate in the synthesis step. A number of classic publications^89,124 and a recent one¹²⁵ indicated that both 5′- and 3′-terminus of the virus RNA are required for replication. The trans model depicted in Figure 5 does not contradict these data, as both termini would required for replication, but not necessarily interacting with the same polymerase complex.

Figure 5.

A trans model for the second step in influenza RNA replication. The diagram presents the proposed process for generation of progeny vRNPs. (A) cRNP template; (B) A non-resident polymerase complex (NRP1, -semitransparent-) would access the 3′-end of the template and initiate de novo. As the vRNA product is produced, a second nonresident polymerase (NRP2, -semitransparent-) would recognise and protect the 5′-terminus, inducing the association of NP. Oligomerisation of additional NP monomers would continue protection and encapsidation of the vRNA product. Subsequent initiation events by additional NRP1 complexes could increase the replication efficiency on the same template; (C) Progression of the replication complexes formed by NRP1 would eventually displace the resident polymerase (opaque) from the 5′-terminus of the cRNA template, thus allowing complete replication; (D) Structure of the progeny vRNP.

The trans model proposed includes the possibility that multiple soluble polymerase (NRP1) complexes would initiate sequentially, allowing several vRNPs to be formed from a single cRNP complex (Fig. 5B and C). As the 5′-terminus of the progeny vRNA becomes available, it would be recognised and protected by binding to a third kind of polymerase complex, different from that performing replication and from that initially present in the cRNP (Non-resident polymerase 2-NRP2-) (Fig. 5B). Such a binding between the polymerase and the vRNA 5′-terminus would be the nucleation step in the encapsidation of vRNA into a progeny vRNP and would involve the sequential association of NP monomers. The specificity of this association would rely on the interaction of NP with the polymerase³⁹ and the capacity of the NP to oligomerise.^34,38 Finally, the replicating polymerase complex(es) would be able to displace the parental polymerase associated to the 5′-terminus of the cRNP template (Fig. 5C), thereby allowing the full-length synthesis of vRNA (Fig. 5D).⁹⁵

Concluding Remarks

A considerable improvement in our understanding of influenza virus RNA replication and gene expression has been obtained in the last years. In this review we have discussed these results and new structural models for the viral RNP and the RNA polymerase, as well as new mechanisms for virus RNA replication and transcription. These models and mechanisms would help us to design experiments and to identify information bottlenecks that have so far hindered further developments. A critical one is the lack of atomic structure of the polymerase heterotrimer, in spite of efforts in several research groups. Insofar such structure is not available, improvements in the resolution of a cryo-EM structure of the complex and docking of the available atomic structures of the polymerase domains (and those to come) should provide a quasi-atomic structure of the polymerase and the RNP, useful to test functional models. Furthermore, docking into the EM structures of various polymerase forms could help understanding the conformational changes that the heterotrimer might suffer upon interaction with other binding partners. New information is also needed on the structure of native helical RNPs to try and transpose the replication and transcription models available to a more realistic setting. In addition, the convergence of biochemical and genetic information with that collected by cryoelectron tomography, cryo EM/3D reconstruction and X-ray difraction of crystals should be very valuable to understand the network of interactions among RNPs for their encapsidation into infectious virus particles.

Abbreviations

HA

haemagglutinin

NA

neuraminidase

vRNA

virion RNA

cRNA

complementary RNA

NP

nucleoprotein

RNP

ribonucleoprotein

EM

electron microscopy

3D

three-dimensional

NLS

nuclear localisation signal

NTP

nucleoside triphosphate

NEP

nuclear export protein

NMR

nuclear magnetic resonance

NRP

non-resident polymerase