Influenza Virus Human
Influenza virus is an important human pathogen. The virus causes epidemics annually resulting in an estimated 250,000 to 500,000 deaths worldwide (WHOa). In the USA alone, 5-20% of the population become infected every year resulting in an estimated 200,000 influenza related hospitalisations and 36,000 deaths (CDC). Influenza virus can also cause pandemics, with three occurring during the 20th century. The first, “Spanish flu”, began in 1918. It is estimated that “Spanish flu” resulted in 40-50 million deaths worldwide (WHOb) and is thought to have had a mortality rate higher than 2.5%, compared to a mortality rate of less than 0.1% in typical influenza epidemics (Wright et al., 2007).
The second and third pandemics of the 20th century were the “Asian flu” of 1957 and the “Hong Kong flu” of 1968. These were much milder pandemics, resulting in around 2 million and 1 million deaths worldwide, respectively (WHOb). Since 1997, it has emerged that highly pathogenic avian influenza (H5N1) can infect humans, with 335 cases worldwide, resulting in 206 deaths (WHOc, 12th November 2007). Another influenza pandemic is inevitable and the number of deaths likely to occur as a result of this can only be speculated upon.
To enable the design of new antiviral treatments and vaccines as well as allowing improved surveillance for viruses more capable of replicating in humans, an understanding of the molecular aspects of virus replication and interaction with the host cell is vital.
Influenza virus classification
Influenza virus belongs to the family Orthomyxoviridae. The Orthomyxoviridae family of viruses are single stranded, negative sense RNA viruses with segmented genomes. There are five genera: Influenza A, B, C, Thogotovirus and Isavirus (Palese and Shaw, 2007). The influenza A, B and C virus types are distinguished according to antigenic variation in the matrix protein and nucleoprotein. Distinctions are also made according to the number of gene segments as influenza A and B viruses have 8 segments, whilst influenza C viruses have only 7. Influenza B and C viruses have limited host reservoirs whereas influenza A viruses can infect a wide range of species.
The major reservoir for influenza A viruses is aquatic and shore birds. However, influenza A viruses are also capable of infecting other species including poultry, swine, horses, whales and humans (Wright et al., 2007). Influenza A viruses are further classified according to their envelope glycoproteins – haemagglutinin (HA) and neuraminidase (NA). There are sixteen known HA subtypes and nine NA subtypes. In their natural reservoir, viruses with all combinations of HA and NA subtype have been isolated. In humans, however, infections have been documented with H1N1, H1N2, H3N2, H7N7, H9N2 and H5N1 viruses.
Virus reassortment and antigenic variation
Influenza A viruses are capable of extreme antigenic variation by two processes, antigenic shift and antigenic drift. Antigenic shift allows viruses to undergo major antigenic variation of the HA and NA glycoproteins by reassortment of the genome segments (Wright et al., 2007). Antigenic drift, however, results in minor antigenic changes as a result of point mutations in the HA and NA proteins (Wright et al., 2007). In mammals, this is in response to host neutralising antibodies and it is this antigenic variation that contributes to the occurrence of annual epidemics. Antigenic shift is also important for influenza A viruses infection in humans because it is the emergence of a new HA or NA subtype into an immunologically naive human population that is the cause of pandemics. In the case of the 1957 and 1968 pandemics, this arose as a result of reassortment between a human and an avian influenza virus. In the case of the 1918 pandemic, evidence suggests that this may have been the result of direct introduction of an avian virus into humans (Wright et al., 2007).
The virus particle
At the centre of the influenza virus particle are the eight genome segments in the form of helical RNP complexes. Each RNA segment is bound by a single copy of the RNA dependent RNA polymerase complex and multiple copies of NP. The RNP complexes are surrounded by a matrix protein core. Small amounts of the nuclear export protein are thought to be associated with the matrix protein (Richardson and Akkina, 1991; Yasuda et al., 1993). Surrounding the matrix protein core is the viral envelope, derived from host cell membrane. The envelope can take many forms, ranging from spherical (80-120 nm diameter) to long and filamentous (more than 300 nm in length) in clinical isolates (Palese and Shaw, 2007). Inserted into the viral envelope are the surface glycoproteins, HA and NA as well as the viral ion channel, M2. A schematic diagram of the viral particle is shown in Figure 1.1.
Influenza virus gene segments and encoded proteins
The influenza virus genome is comprised of eight segments, numbered according to size. Five of these segments encode one protein each – polymerase basic 2 (PB2), polymerase acidic (PA), haemagglutinin (HA), neuraminidase (NA) and nucleoprotein (NP). The remaining three segments encode two proteins each. Segment 2 encodes polymerase basic 1 (PB1) and a second protein, PB1-F2, probably as a result of ribosomal scanning (Chen et al., 2001). Segments 7 and 8 encode the matrix protein and the M2 ion channel and non-structural protein 1 (NS1) and the nuclear export protein (NEP), respectively as a result of alternate splicing. A summary of the proteins encoded and a brief description of their function is shown in Table 1.1.
Segments 1-3 encode the three subunits of the RNA dependent RNA polymerase complex; polymerase basic 2 (PB2), polymerase basic 1 (PB1) and polymerase acidic (PA) proteins, respectively. The polymerase complex binds to each vRNA segment via the promoter, which is comprised of conserved sequences at the 5′ and 3′ ends of each genome segment (Fodor and Brownlee, 2002). A more detailed description of these three proteins will be given in section 1.7.
As well as encoding the PB1 protein, it has recently been described that segment 2 encodes a second protein, PB1-F2, as a result of ribosomal scanning (Chen et al., 2001). The PB1-F2 protein is targeted to the mitochondria of infected cells where it triggers a loss of mitochondrial membrane potential, resulting in apoptosis (Chen et al., 2001; Yamada et al., 2004). However, PB1-F2 does not appear to be encoded by all influenza viruses and induces apoptosis in a cell specific manner (Chen et al., 2001; Zell et al., 2007). As such, further studies are required to understand the function and importance of this protein.
Segment 4 encodes the haemagglutinin (HA) envelope glycoprotein. HA acts as the attachment protein for influenza virus by binding to sialic acid moieties on host cell surface glycoproteins and glycolipids. HA is also involved in fusion of the viral and host cell membranes, mediating entry of the viral genome into the cell (Skehel and Wiley, 2000).
The structure of HA was determined early on (Wilson et al., 1981) and revealed that HA forms a homotrimer consisting of a stalk region and a globular head. This homotrimer projects as a rod shaped spike away from the viral envelope, with the carboxy-terminal domain for each monomer inserted into the envelope (Palese and Shaw, 2007; Skehel and Wiley, 2000). Each HA monomer is synthesised as a precursor protein HA0 which is then post-translationally cleaved into HA1 and HA2. These two polypeptides remain linked by a disulphide bridge. This cleavage event is essential for the activation of the membrane fusion potential of HA.
It has become clear recently that the nature of the cleavage site is an important factor in the virulence of the virus. The cleavage site usually consists of a single arginine residue at position 329 of the HA0 precursor (Webster and Rott, 1987). This is removed extra-cellularly by trypsin-like proteases. However, HA0 molecules exist with multiple basic residues in the cleavage region which results in HA0 precursor cleavage intra-cellularly by furin (Steineke-Grober et al., 1992; Webster and Rott, 1987). Viruses containing HA molecules with the multi-basic cleavage site often have high virulence, highlighting HA as an important virulence factor (Gabriel et al., 2005; Hatta et al., 2001; Webster and Rott, 1987).
Within the globular head of the HA homotrimer is the receptor binding site. The exact receptor used by influenza viruses varies depending on the host species. In the case of avian influenza viruses, sialic acid with an α2,3 linkage acts as the receptor. However, in the case of human influenza viruses, sialic acid with an α2,6 linkage is used (Ito et al., 1998; Nobusawa et al., 1991; Rogers and D'Souza, 1989). This correlates with the most abundant form of sialic acid found in the avian gastrointestinal tract and the human airway. For an avian virus to cross the species barrier and replicate efficiently in the human population requires the mutation of the receptor binding site resulting in adaptation of HA to recognise sialic acids with an α2,6 linkage.
It was originally proposed that this may occur in a swine host. Pigs were shown to have sialic acids of both linkages in their airways. Therefore, they were proposed to act as a vessel for viral reassortment or mutation of avian-like HAs to recognise human receptors (Ito et al., 1998). More recently, it has been demonstrated that human airways also possess sialic acid with an α2,3 linkage. Non-ciliated cells were shown to have sialic acid with an α2,6 linkage whilst some ciliated cells were shown to have sialic acid with an α2,3 linkage (Matrosovich et al., 2004b). Furthermore, avian-like viruses were capable of infecting the ciliated cells, although they grew to significantly reduced titres (Matrosovich et al., 2007). As often only minor mutations are required in the receptor binding pocket to cause a change in receptor specificity (Gamblin et al., 2004; Matrosovich et al., 2007; Rogers et al., 1983; Skehel and Wiley, 2000), this may suggest the ability of avian viruses to adapt to replicate in humans directly, without an intermediate host.
HA is the most abundant surface glycoprotein and, as such, is the target for host neutralising antibodies (Staudt and Gerhard, 1983). Antibodies targeting HA stericially block attachment of the virus to the cell and are, therefore, the most important component of the immune response. However, these antibodies are also the driving force for antigenic drift in humans.
The protein encoded by segment 5 of the influenza A virus genome is the nucleoprotein (NP) (Winter and Fields, 1981). NP is a phosphorylated, 56 kDa protein rich in arginine residues (Kistner et al., 1989; Winter and Fields, 1981) and is used as one of the determinants for distinguishing influenza A, B and C viruses. NP can form oligomers, requiring regions in the middle and C-terminus of the protein (Elton et al., 1999a) and is reported to be incorporated into ribonucleoprotein complexes in the form of dimers (Ortega et al., 2000). NP is required for encapsidation of RNA in the viral ribonucleoprotein complex.
Each NP monomer is reported to interact with 24 nucleotides of RNA (Jennings et al., 1983; Ortega et al., 2000) with interactions between NP and RNA occurring at multiple regions (Elton et al., 1999b). It has been proposed that NP does not coat the viral RNA but instead forms a core, around which the RNA winds (Baudin et al., 1994; Murti et al., 1992). Furthermore, RNA within RNP complexes remains sensitive to RNase digestion (Duesberg, 1969). NP has also been shown to interact with the polymerase complex in RNPs (Area et al., 2004; Martin-Benito et al., 2001) and to interact directly with RNP-free PB1 and PB2, although not PA (Biswas et al., 1998; Medcalf et al., 1999).
NP has been suggested to play an important role in the switch between viral transcription and replication (Fodor and Brownlee, 2002; Palese and Shaw, 2007) with recent studies suggesting that RNA binding activity of NP is important for this role (Vreede and Brownlee, 2007; Vreede et al., 2004) rather than a direct affect on polymerase function (see 18.104.22.168, below). As well as this, three nuclear localisation signals have been identified within NP (Bullido et al., 2000; Cros et al., 2005; Neumann et al., 1997; Wang et al., 1997; Weber et al., 1998) and these nuclear localisation signals are reported to control nuclear import of RNP complexes (Cros et al., 2005; Wu et al., 2007).
Furthermore, NP shuttles between the nucleus and cytoplasm (Neumann et al., 1997) and has been shown to interact with nuclear import factors karyopherin α1 and α2 (O'Neill et al., 1995; O'Neill and Palese, 1995) and nuclear export factor CRM1 (Elton et al., 2001) suggesting a role in both nuclear import and export of viral RNPs. Finally, interactions with other viral and cellular proteins including M1 (Martin and Heleniust, 1991), F-actin (Digard et al., 1999) and RAF-2p48/UAP56 (Momose et al., 2001) suggest a role for NP at multiple stages of the viral replication cycle (Portela and Digard, 2002).
Segment 6 encodes the second viral glycoprotein, neuraminidase (NA), comprised of a slender stalk and a globular head (Laver and Valentine, 1969). The crystal structure of NA has been solved identifying a homotetrameric box-shaped head (Varghese et al., 1983). Each monomer contains a neuraminidase active site in the head domain and this enzymatic activity is responsible for the release of progeny virions from the host cell surface. Cells infected with temperature sensitive mutants of the neuraminidase gene showed large aggregates of virions on the cell surface (Palese et al., 1974). Furthermore, NA functions to remove sialic acid moieties from HA on the virus surface, allowing proper function (Ohuchi et al., 1995) as well as sialic acid moieties from mucins, allowing efficient virus spread through the respiratory tract (Palese and Shaw, 2007).
The stalk of the NA homotetramer may also play an important function in determining pathogenicity and host range (Castrucci and Kawaoka, 1993). Shortening the length of the stalk resulted in wild type levels of replication in MDBK cells but a reduction in replication in embryonated chicken eggs and a reduction in pathogenicity and an inhibition of systemic virus spread in mice. Furthermore, increasing the length of the stalk resulted in an increase in replication in eggs. However, the significance of these observations is not fully understood. Finally, as well as a function in release of progeny virions, NA may also play a role at an early stage of the viral life cycle. When cells were treated with neuraminidase inhibitors prior to infection, viral replication levels were reduced (Matrosovich et al., 2004a).
As a surface glycoprotein, antibodies are made against NA which prevent viral spread and reduce severity of illness (Kilbourne et al., 2004). Inhibitors of the neuraminidase protein are available that can be useful for reducing the severity of symptoms, if administered within 48 hours of infection, and can also be used prophlactically (Gubareva et al., 2000). Due to availability, low occurrence of resistance (Reece, 2007) and proven in vitro activity against H5N1 strains of influenza A virus (Hurt et al., 2007), these drugs currently provide the best protection against a potential influenza pandemic. Thus, enough courses to treat 25% of the UK population have been stockpiled by the UK government (Department of Health website).
Segment 7 encodes two proteins, the matrix protein, M1, and the viral envelope ion-channel, M2. M1 is encoded by a collinear mRNA transcript and is the most abundant protein in the virion. M1 monomers oligomerise with positively and negatively charged residues positioned on opposite faces of the oligomer (Arzt et al., 2001; Noton et al., 2007). M1 lies beneath the lipid envelope and provides rigidity to the virus particle, interacting with the cytoplasmic tails of the viral glycoproteins, the lipid membrane (Enami and Enami, 1996; Ruigrok et al., 2000; Zhang and Lamb, 1996) and the vRNP core (Murti et al., 1992). Furthermore, M1 is proposed to be important for recruiting viral proteins to the site of budding at the plasma membrane and may be required for the budding process (Gomez-Puertas et al., 2000; Murti et al., 1992).
In addition to these functions, M1 can interact with viral RNA (Wakefield and Brownlee, 1989) and NP (Noton et al., 2007; Ye et al., 1999) and it is suggested that this may signal the end of vRNA replication and trigger assembly of vRNP complexes (Baudin et al., 2001; Huang et al., 2001). Furthermore, M1 can interact with NEP and has been shown to be important for export of vRNPs from the nucleus. An interaction between M1 and Hsc70 may be important for this (Watanabe et al., 2006), although previous work has suggested an inhibition of nuclear export by Hsp70 at elevated temperatures (Hirayama et al., 2004; Sakaguchi et al., 2003).
Segment 7 also encodes M2 via a spliced mRNA. M2 is a homotetrameric minor envelope protein consisting of two disulphide linked dimers or four disulphide linked monomers (Holsinger and Alams, 1991; Sugrue and Hay, 1991). M2 is an ion channel required for virion uncoating by allowing ions to enter the virion from the acidified endosome resulting in dissociation of M1 from the vRNP complexes. M2 also functions within the Golgi apparatus to prevent a premature conformational change of HA2 where HA0 has been cleaved intracellularly (Ciampor et al., 1992; Takeuchi and Lamb, 1994; Takeuchi et al., 1994). Furthermore, M2 may play a role in assembly and budding of the virus particle (Hughey et al., 1995; Iwatsuki-Horimoto et al., 2006; McCown and Pekosz, 2005; McCown and Pekosz, 2006; Schroeder et al., 2005). Activity of M2 can be inhibited with antivirals amantidine and rimantidine. However, due to the rapid development of resistance, these drugs are not widely recommended.
Segment 8 also encodes 2 proteins, NS1 and NEP (NS2) (Lamb and Choppin, 1979). Non-structural protein 1 (NS1) is encoded by a collinear transcript and is a multifunctional, non-structural protein. NS1 is an important protein for controlling viral and host gene expression and inhibiting cellular anti-viral responses. Firstly, NS1 acts as an antagonist against the cellular interferon system (Geiss et al., 2002; Hayman et al., 2007; Kochs et al., 2007; Min and Krug, 2006) and virus lacking the NS1 gene is only capable of replicating in cells lacking the interferon response (Garcia-Sastre et al., 1998). Furthermore, NS1 plays a role in the viral shut-off of host cell gene expression by preventing cellular mRNA processing and nucleocytoplasmic transport of cellular mRNAs (Kochs et al., 2007; Lu et al., 1994; Nemeroff et al., 1998; Qiu and Krug, 1994; Satterly et al., 2007). Finally, NS1 functions to increase the level of viral mRNA translation initiation (de la Luna et al., 1995; Enami et al., 1994; Marion et al., 1997).
The nuclear export protein (NEP) was renamed from NS2 because it was shown to be incorporated into the virus particle (Richardson and Akkina, 1991; Yasuda et al., 1993). NEP has been shown to contain a classical nuclear export signal and interact with cellular nucleoporins (O'Neill et al., 1998). Furthermore, NEP has been shown to interact with M1 (Ortin, 1998), suggesting that NEP acts as an adapter between the vRNP complexes and the nuclear pore complex (O'Neill et al., 1998).
The virus replication cycle
Unusually for an RNA virus, the influenza virus life cycle includes steps located in the nucleus. This is due to a requirement for cellular pre-mRNAs (see 22.214.171.124, below) and cellular splicing factors for the processing of mRNA transcribed from segments 7 and 8 (see 1.5.5 and 1.5.6, above). As a result of this, influenza virus has a more complex life cycle than many RNA viruses involving nuclear import and export of viral components. The virus life cycle is summarised in figure 1.2.
Attachment and entry
The first stage of the virus life cycle is attachment to the host cell membrane. Influenza A virus achieves this via binding of the haemagglutinin envelope protein to sialic acid moieties on cell surface glycoproteins and glycolipids. Specificity of this interaction varies between different strains of influenza A virus. Avian influenza A viruses attach to sialic acid with an α-2,3 linkage, whereas human influenza A viruses attach to sialic acid with an α-2,6 linkage (Ito et al., 1998; Nobusawa et al., 1991; Rogers and D'Souza, 1989). This corresponds to the most abundant form of sialic acid found in the avian gastrointestinal tract and the human respiratory tract (Couceiro et al., 1993). This specificity for binding is dependent upon specific residues within the sialic acid binding site on HA and plays an important role in determining host range of the virus (see 1.5.2).
Following attachment to the cell surface, the virus must enter the cell. Fusion of the influenza A virus envelope does not occur at the plasma membrane. Instead, the virus is endocytosed, mainly in clathrin-coated vesicles (Matlin et al., 1981). Upon acidification of the endosome, a conformational change is triggered in HA resulting in insertion of the HA2 fusion peptide into the endosomal membrane (Bullough et al., 1994; Chen et al., 1999a). As a result, the viral and endosomal membranes fuse (Stegmann, 2000). A second affect of acidification of the endosome is that the M2 ion channel pumps protons into the virus particle. This disrupts interactions between M1 and the RNP complex (Helenius, 1992), allowing the release of free vRNPs into the cytoplasm.
Nuclear import of vRNPs
After release of the vRNPs into the cytoplasm they are rapidly transported to the nucleus (Kemler et al., 1994; Martin and Helenius, 1991), where viral transcription and replication occur. This nuclear import is via the nuclear pore and is an active process (Kemler et al., 1994; Martin and Helenius, 1991). Furthermore, NP, the major component of vRNPs, has been shown to be the main factor required for the import of vRNPs (Cros et al., 2005; Cros and Palese, 2003; O'Neill et al., 1995; Wu et al., 2007), probably via an interaction between NP and cellular import factors karyopherin α1 or α2 (O'Neill et al., 1995; O'Neill and Palese, 1995). Three nuclear localisation signals have been proposed in NP (Bullido et al., 2000; Neumann et al., 1997; Wang et al., 1997; Weber et al., 1998) with an unconventional NLS at the N-terminus being likely to play the major role in nuclear targeting of vRNP, although a second NLS in the central region of NP has also been shown to have a function (Cros et al., 2005; Wu et al., 2007).
Viral transcription and replication
Once the vRNPs have been imported into the nucleus, the viral RNA dependent RNA polymerase complex is responsible for transcribing viral mRNA and also replicating vRNA via a cRNA template (Fig. 1.3). The mechanisms for these two functions of the polymerase are very different. Viral transcription of mRNA in primed by a host derived capped oligonucleotide and transcription is terminated prematurely followed by stuttering of the polymerase at a sequence of 5-7 uridines, resulting in the synthesis of a poly-A tail. However, replication of the vRNA genome requires de novo initiation and read through of the poly-A site, resulting in a full length positive sense copy of the vRNA genome (cRNA). This cRNA template is then copied back into vRNA in a process also requiring de novo initiation (Fodor and Brownlee, 2002; Palese and Shaw, 2007).
Viral mRNA transcription is dependent upon the cellular RNA polymerase II (Pol II). This is due to a requirement for 5′ capped oligonucleotides to act as primers for initiation of mRNA synthesis (Plotch et al., 1979). Viral mRNA synthesis has been shown previously to be inhibited by treatment of cells with α-amanitin, a specific inhibitor of cellular Pol II (Mark et al., 1979). Furthermore, the influenza polymerase complex has been shown to specifically interact with the initiating form of the Pol II C-terminal domain and reduce the density of Pol II over the coding region of Pol II genes during virus infection (Chan et al., 2006; Engelhardt et al., 2005).
The viral polymerase steals capped oligonucleotides from cellular pre-mRNAs in a process known as cap-snatching. The viral RNA polymerase binds to the 5′ end of the vRNA template via a sequence conserved in all eight viral genome segments. This results in a conformational change in the PB2 subunit of the viral polymerase complex allowing recognition and binding to the cap structure of cellular pre-mRNAs (Cianci et al., 1995). Subsequently, binding of the 3′ end of the vRNA segment activates endonuclease activity of the PB1 subunit (Cianci et al., 1995; Hagen et al., 1994; Leahy et al., 2001a). This allows cleavage of the host pre-mRNA 9-17 nucleotides from the cap structure, usually after a purine residue (Beaton and Krug, 1981; Caton and Robertson, 1980; Neumann et al., 2004; Plotch et al., 1981). Transcription is then primed with the addition of a guanine opposite the cytosine at position 2 of the vRNA template (Beaton and Krug, 1981) in a process that does not require hydrogen bonding between primer and template (Krug et al., 1980). Occasionally, transcription can be initiated with the incorporation of a cytosine opposite the guanine at position 3 of the vRNA template (Fodor et al., 1995).
Transcription then proceeds until the polymerase reaches a stretch of 5-7 uridine residues located approximately 17 residues from the 5′ end of the template where the polymerase stutters and copies this poly-uridine tract many times, resulting in the synthesis of a poly-A tail (Li and Palese, 1994; Poon et al., 1999; Zheng et al., 1999). If this stretch of bases is replaced with 5-7 adenine residues, a poly-U tail is inserted instead. However, messages containing a poly-U tail are not exported from the nucleus efficiently (Poon et al., 2000). The reason for the stuttering of the polymerase at this stretch of uridine residues is proposed to be because the polymerase remains bound to the 5′ end of the vRNA template throughout mRNA synthesis, while the RNA is fed through the polymerase in a 3′ to 5′ direction. The result of this is that the polymerase becomes sterically blocked and therefore stutters at this point (Fodor et al., 1994; Hagen et al., 1994; Poon et al., 1999; Zheng et al., 1999).
Replication and the “switch” between transcription and replication
The second function of the influenza viral polymerase complex is to replicate the vRNA genome, via a cRNA intermediate. Synthesis of cRNA, like synthesis of mRNA, results in a positive sense RNA. However, the mechanism of synthesis is very different. Firstly, cRNA synthesis is a primer independent process requiring de novo initiation. This results in the 5′-terminus comprising an adenosine-triphosphate group (Hay et al., 1982). Secondly, premature termination with stuttering at the poly-A signal does not occur during cRNA synthesis, resulting in a full length copy of the vRNA segment. The mechanism by which coupling of cap-snatching and premature termination at the poly-A signal, and de novo initiation and read through of the poly-A signal occurs is unknown. In the second step of vRNA replication, the cRNA template is transcribed into new vRNA molecules in a process analogous to cRNA synthesis (Fodor and Brownlee, 2002; Palese and Shaw, 2007) (see 1.9 for further details).
During the virus life cycle, synthesis of mRNA and cRNA appears to be temporally controlled. At early time points post infection, mRNA synthesis can be detected. However, cRNA transcription can only be seen after new viral protein has been synthesised. These observations resulted in the hypothesis that incoming vRNPs were only capable of transcribing mRNA and transcription of cRNA required new protein synthesis (Hay et al., 1977). Furthermore, new synthesis of soluble NP was proposed to be responsible for this “switch” in the mode of transcription (Honda et al., 1988). The use of temperature sensitive mutants of NP seemed to support this model due to an inhibition of cRNA but not mRNA synthesis at the non-permissive temperature (Medcalf et al., 1999; Shapiro and Krug, 1988). However, recently an alternative model has been proposed suggesting that incoming vRNPs are capable of synthesising both mRNA and cRNA.
This model proposes that at early time points after infection newly synthesised cRNA is not encapsidated by NP and is therefore degraded by host cell nucleases. At later times, cRNA is stabilised by newly synthesised NP and polymerase (Vreede et al., 2004). Further evidence for this model was provided by the ability of vRNPs isolated from virus to synthesise cRNA in vitro, with addition of non-vRNP associated NP having no effect cRNA levels (Vreede and Brownlee, 2007).
Furthermore, the stabilisation model is consistent with previous work suggesting that short de novo initiated RNAs can be synthesised in vitro in the absence of NP, and in vivo observations that low levels of cRNA could be detected at early time points in the presence of cyclohexamide and that increased levels of NP do not result in increased levels of genome replication (Barrett et al., 1979; Lee et al., 2002a; Mullin et al., 2004).
Control of viral gene expression
Expression of the viral gene products has been observed to be temporally controlled (Hay et al., 1977; Smith and Hay, 1982). At early stages of infection expression of NP, PB1, PB2, PA and NS1 is favoured. However, expression of HA, NA, NEP, M1 and M2 is delayed (Hatada et al., 1989; Hay et al., 1977; Shapiro et al., 1987; Smith and Hay, 1982). The mechanism for this temporal control of gene expression is unknown. However, natural variation within the vRNA promoter may play a role. The sequence of the vRNA promoter is highly conserved. However, at position 4 from the 3′ end, variation occurs with a “C” occurring in segments 1-3 and 7 and a “U” occurring in segments 4-6 and 8.
Furthermore, the identity of position number 4 correlates with the ratio of transcription to replication with promoters containing C4 having a lower level of transcription and a higher level of replication, whereas promoters containing U4 have a higher level of transcription and a lower level of replication (Lee and Seong, 1998). Consistent with this, it has been shown that, in vivo, expression of the polymerase genes is lower than the other genes (Hay et al., 1977; Smith and Hay, 1982). Furthermore, it has been suggested by structural studies that the identity of position 4 may alter the conformation of the promoter and may therefore alter binding by the viral polymerase complex, thus altering gene expression (Lee et al., 2003a). Therefore, the identity of position 4 in the vRNA promoter may play an important role in regulating the level of gene expression.
Control of mRNA splicing is also an important factor in the control of viral gene expression. Segments 7 and 8 each encode two proteins by alternate slicing events. The virus utilises the host splicing system as shown by splicing of viral mRNA in the absence of any viral proteins (Lamb and Lai, 1982; Lamb and Lai, 1984). However, viral control of the frequency of splicing must exist because only about 10% of transcripts from segments 7 and 8 are spliced, whereas host controlled mRNA splicing in very efficient (Lamb et al., 1980; Lamb et al., 1981).
The mechanisms for this viral control of splicing are not fully understood although control of the rate of mRNA export as well as internal sequences in the NS1 transcript are involved in the control of splicing of the NS1 mRNA (Alonso-Caplen and Krug, 1991; Alonso-Caplen et al., 1992; Garaigorta and Ortin, 2007; Nemeroff et al., 1992). Control of splicing of the M1 transcript is thought to involve the viral polymerase complex and the host splice factor SF2/ASF (Shih et al., 1995; Shih and Krug, 1995).
Finally, control of viral gene expression involves the rate of translation initiation from the viral mRNAs. Preferential translation of viral mRNAs is controlled, in part, by a down-regulation of host cell gene expression (see 1.6.4). However, the mechanisms for translation initiation from viral mRNAs differ from cellular mRNAs, with cellular mRNAs requiring functional translation factor eIF4E (Burgui et al., 2007). In contrast, viral mRNAs are translated efficiently when the function of eIF4E is impaired (Burgui et al., 2007). The viral protein, NS1, plays an important role in the translational initiation of viral mRNAs. NS1 can associate with the viral mRNA and cellular factors eIF4GI and poly(A)-binding protein, recruiting the ribosome specifically to viral transcripts (Aragon et al., 2000; Burgui et al., 2003; Park and Katze, 1995).
Control of host cell gene expression
During the influenza virus life cycle, an efficient shut-off of host cell gene expression is observed and a preferential expression of viral genes. Several mechanisms are thought to control the down-regulation of host cell gene expression. The cap-snatching mechanism, as well as an inhibition of 3′ end processing and splicing of cellular pre-mRNAs by NS1 results in an inhibition of the maturation of cellular mRNAs (Chen et al., 1999b; Kochs et al., 2007; Nemeroff et al., 1998; Shimizu et al., 1999). Furthermore, export of cellular mRNAs from the nucleus is blocked by the action of NS1 in inhibiting processing of cellular mRNAs (Fortes et al., 1994; Lu et al., 1994). Finally, translation of cellular mRNAs is inhibited by inactivation of translation initiation factor eIF4E (Feigenblum and Schneider, 1993).
The virus is capable of escaping the affects of shutting off these host functions because the virus uses alternate mechanisms for viral mRNA processing, export and translation initiation.
Nuclear export of vRNPs
The first step in assembly of new virions is the assembly of vRNPs. Newly synthesised PB1, PB2, PA and NP are transported into the nucleus where they bind to newly synthesised vRNA to form vRNPs. To allow assembly of complete virus particles these vRNPs must be exported from the nucleus. M1, NEP and NP have all been implicated in this step of the virus life cycle. Interaction between M1 and the vRNPs via interactions with NP and vRNA have been observed, and this is essential for nuclear export (Bui et al., 2000; Martin and Heleniust, 1991). Despite this, M1 has not been shown to interact directly with the nuclear export machinery.
NEP, however, has been demonstrated to interact with M1 (Yasuda et al., 1993), as well as nuclear export protein CRM1 (Neumann et al., 2000) and several nucleoporins (O'Neill et al., 1998). As such, it is proposed that NEP acts as an adapter protein for export of vRNP-M1-NEP complexes. Furthermore, interactions between NP and CRM1 have also been observed (Elton et al., 2001), suggesting a direct role for NP in export of vRNP complexes from the nucleus. A role for M1 and NP has also been suggested for preventing re-entry of vRNPs into the nucleus once exported to the cytoplasm (Digard et al., 1999; Whittaker et al., 1996; Whittaker et al., 1995).
Virion assembly and budding
The influenza virus envelope glycoproteins are synthesised on ER-bound ribosomes and transported via the cellular secretory pathway to the apical surface of the plasma membrane. Following export from the nucleus, vRNPs are targeted to the viral assembly site at the plasma membrane where particles begin to assemble. M1 is thought to mediate particle formation by interacting with the cytoplasmic tails of HA, NA and M2, as well as interacting with the plasma membrane and vRNPs. Following assembly of all of the viral components, budding is triggered with an outward curvature of the membrane until the viral core is surrounded.
The membrane then fuses to release the virus particle. M1 is thought to be the driving force for the initial membrane curvature as well as completion of the budding process because M1 expressed alone was found to form virus-like particles (Gomez-Puertas et al., 2000). Despite this, a recent study reports the production of virus-like particles in the presence of HA and NA alone, without M1 (Chen et al., 2007a). Therefore, the process of budding remains controversial and further work is required to determine the factors involved in this step of the virus life cycle.
For production of fully functional, infectious virus, all eight vRNP complexes must be incorporated into the particle. Two models have been proposed for genome packaging. The first states that RNPs are randomly incorporated into budding particles (Enami et al., 1991), resulting in the production of a large number of non-infectious particles. The second model proposes that all eight genome segments are selectively incorporated into budding particles (Smith and Hay, 1982). This is hypothesised to occur via protein: protein and RNA: protein interactions. The selective packaging model is supported by the identification of specific packaging signals within each of the genome segments (Fujii et al., 2005; Fujii et al., 2003; Gog et al., 2007; Liang et al., 2005; Liang et al., 2007; Muramoto et al., 2006; Watanabe et al., 2003).
The viral polymerase
As mentioned earlier, the influenza polymerase is a heterotrimeric complex comprised of the PB1, PB2 and PA subunits. Although some reports describe replication activity of PB1 alone or the [PB1, PA] dimer (Honda et al., 2002; Kobayashi et al., 1996; Nakagawa et al., 1996; Toyoda et al., 1996b) and transcription activity of the [PB1, PB2] dimer (Honda et al., 2002), most studies have shown that all three subunits are required for in vivo and in vitro polymerase function (Brownlee and Sharps, 2002; Fodor et al., 2002; Hara et al., 2006; Huang et al., 1990; Lee et al., 2002b; Perales and Ortin, 1997; Poole et al., 2007). Discrepancies in these reports may be due to the method of polymerase expression. Studies suggesting activity of PB1 alone or the dimers were performed using polymerase purified from bacuolvirus infected insect cells. Studies using polymerase expressed in mammalian cells demonstrated a requirement for all three subunits.
The crystal structure of the polymerase complex, or any complete polymerase subunit, is not available. However, electron microscopy images suggest that in the polymerase complex, either alone or as part of the RNP, the polymerase subunits are tightly associated with no obvious boundaries (Area et al., 2004; Torreira et al., 2007). Furthermore, recent work has suggested that there are conformational changes in the polymerase complex upon interaction with NP or the RNP, particularly in the PB2 subunit (Torreira et al., 2007). Interaction domains between the three polymerase subunits have been mapped in a number of studies.
It has been shown that the PB1 subunit forms the core of the complex, interacting with both the PA and PB2 subunits, whilst there is no direct interaction between PA and PB2 in the absence of PB1 (Digard et al., 1989; Gonzalez et al., 1996; Ohtsu et al., 2002; Toyoda et al., 1996a). It is accepted that the C-terminus of PA interacts with the N-terminus of PB1 (Gonzalez et al., 1996; Ohtsu et al., 2002; Perez and Donis, 1995; Toyoda et al., 1996a; Zurcher et al., 1996) and that the C-terminus of PB1 interacts with the N-terminus of PB2 (Biswas and Nayak, 1996; Gonzalez et al., 1996; Ohtsu et al., 2002; Perales et al., 1996; Poole et al., 2004; Poole et al., 2007; Toyoda et al., 1996a), although a second PB1 binding site at the C-terminus of PB2 has also been reported (Poole et al., 2004). Therefore, some controversy remains as to the precise regions involved in these interactions.
Recently the polymerase complex has been shown to co-localise with the large subunit of cellular RNA polymerase II (Pol II), with trimeric complexes showing this association, but not [PB1, PA] or [PB1, PB2] dimers (Engelhardt et al., 2005). Furthermore, the influenza polymerase interacts specifically with transcriptionally active cellular Pol II resulting in a reduction in the association of Pol II with coding regions of Pol II transcribed genes. Influenza virus specifically inhibits elongation by cellular Pol II but does not affect initiation (Chan et al., 2006). This probably reflects the requirement of the viral polymerase for capped oligos to initiate viral mRNA synthesis.
The PB2 subunit of the viral polymerase complex is encoded by segment 1 of the viral genome. The main function of PB2 within the polymerase is binding of the cap structure of host cell pre-mRNAs (Blaas et al., 1982; Blass et al., 1982; Ulmanen et al., 1981). Three studies have been performed to identify the site in PB2 involved in cap binding, although the precise residues involved remain controversial. An early study identified two regions (242-282 and 538-577) involved in cap binding by UV cross-linking of 32P-labelled capped RNA to viral RNP cores followed by partial V8 protease digestion (Honda et al., 1999).
A second study identified region 544-556 as important for cap binding by cross-linking of 32P-labelled capped RNA containing a 4thio-U residue to recombinant polymerase, followed by partial V8 digestion (Li et al., 2001). In the final study, mutagenesis of aromatic amino acid residues conserved between influenza A, B, C and Thogoto viruses was performed (Fechter et al., 2003). The cap binding sites of other cap binding proteins, e.g. eIF4E, CBP20 and vaccinia virus P39, have been demonstrated to be aromatic sandwiches where the cap structure is “sandwiched” between two aromatic residues (Fechter and Brownlee, 2005). In this study, Fechter et al. (2003) suggested residues F363 and F404 are directly involved in cap binding. Discrepancies between these studies are probably due to the methods used and final evidence for the exact residues required for cap binding is unlikely to be gained until the complete crystal structure of PB2 is solved.
As well as a role in cap binding, PB2 has been shown to be essential for replication activity by the polymerase complex (Perales and Ortin, 1997) with specific residues in the N-terminal half of PB2 being required for this function (Gastaminza et al., 2003). However, the molecular mechanisms for PB2 involvement in replication have not been characterised. Other functions of PB2 have also been suggested. During the viral replication cycle, as well as being targeted to the nucleus of the host cell, PB2 is transported to the mitochondria (Carr et al., 2006). As a result of this, the membrane potential of the mitochondria is maintained, possibly resulting in maintenance of mitochondrial function (Carr et al., 2006), although the significance of this requires further investigation.
An important characteristic of PB2 identified recently is a function in host range restriction. Initially a single amino acid substitution (E627K) was identified that correlated with ability of virus to replicate in avian and mammalian cell lines (Subbarao et al., 1993). The presence of an E or K at this position is also characteristic of avian and human influenza viruses, respectively. Furthermore, the E627K mutation converted a non-lethal virus in mice into a lethal virus in mice, again, highlighting the importance of this position (Hatta et al., 2001).
Other studies have suggested that the E627K mutation does not alter cell tropism but instead increases efficiency of replication (Shinya et al., 2004), is involved in cold sensitivity of avian polymerases (Massin et al., 2001) and that a K at position 627 is required for efficient binding of NP to avian PB2 or trimeric polymerase in 293T cells (Labadie et al., 2007). However, other studies have suggested that position 627 does not always correlate with the level of virulence (Govorkova et al., 2005; Maines et al., 2005), suggesting the importance of other residues. In accordance with this, residues 701 and 714 in PB2 have also been shown to be involved in host restriction and virulence (Gabriel et al., 2005; Li et al., 2005). However, the molecular mechanisms for the increase in virulence and alteration of host restriction caused by amino acid substitutions at positions 627, 701 and 714 have not been elucidated.
The polymerase basic 1 (PB1) subunit is encoded by segment 2 of the viral genome. PB1 contains four motifs conserved within all viral RNA dependent RNA polymerases and RNA dependent DNA polymerases with four invariant residues, one in each motif (Poch et al., 1989). Motif three has been shown to be the active site for polymerisation of the growing nucleotide chain and consists of an SDD motif (Biswas and Nayak, 1994). This motif is located between positions 444-446 in PB1 and mutation of the SDD sequence inactivates the RNA synthesis function of the polymerase (Biswas and Nayak, 1994; Vreede et al., 2004).
PB1 has been shown to cross-link to the growing mRNA chain (Braam et al., 1983), consistent with a role in chain elongation. The PB1 subunit has also been shown to bind to the nucleotide substrates. The nucleotide binding site was identified by UV cross-linking PB1 or trimeric polymerase to radioactive GTP followed by partial digestion with V8 protease (Asano and Ishihama, 1997; Asano et al., 1995) or unlabelled ATP followed by extension by the addition of [α32P] GTP, followed by partial digestion with V8 protease (Kolpashchikov et al., 2004).
A second function of the PB1 subunit is endonucleolytic cleavage of cellular pre-mRNAs to generate capped primers for viral mRNA synthesis. Endonuclease activity is activated by binding of the polymerase to the 5′ and 3′ ends of the viral promoter, triggering conformational changes in the polymerase complex (Cianci et al., 1995; Hagen et al., 1994; Li et al., 1998). The active site for endonuclease activity was originally mapped to the PB2 subunit (Shi et al., 1995), although was later shown to be located in the 508-522 region of PB1, with amino acids E508, E519 and D522 shown to be essential for activity, similar to other endonucleases that produce 3′-OH ends (Li et al., 2001).
The PB1 subunit has also been shown to interact with cellular proteins Hsp90 (Deng et al., 2005) and Ebp1 (Honda et al., 2007). The function of interaction with Hsp90 has not been characterised, although it is likely that Hsp90, an abundant cellular chaperone, binds to newly synthesised PB1 to stabilise it, preventing aggregation and facilitating assembly of PB1 into the trimeric polymerase complex. Binding of PB1 to Ebp1 is reported to have a negative effect on polymerase activity and result in re-localisation of Ebp1 to nuclear aggregates with the polymerase complex upon viral infection (Honda, 2007; Honda et al., 2007). Honda et al. propose that Ebp1 may function as a host cell defence against viral replication (Honda et al., 2007). However, the importance of PB1 interaction with either Hsp90 or Epb1 requires further characterisation.
The polymerase acidic (PA) subunit is encoded by segment 3 of the viral genome. The function of PA within the polymerase complex is much less well defined than the functions of the PB2 and PB1 subunits, although through a number of studies, several functions have been suggested. PA is a phosphoprotein with phosphorylation at serine and threonine residues (Sanz-Ezquerro et al., 1998). Early work using temperature sensitive viruses demonstrated a role for PA in replication by the polymerase complex (Herget and Scholtissek, 1993; Krug et al., 1975; Mahy et al., 1981; Markushin and Ghendon, 1984; Scholtissek and Muller, 1983; Thierry and Danos, 1982). However, later work has also shown that PA is required for transcription. Alanine mutations at positions 510, 108 and 134 inhibited endonuclease activity whilst alanine mutation at position 102 inhibited cap binding (Fodor et al., 2002; Hara et al., 2006).
PA expression has been reported to induce proteolytic cleavage of co-expressed viral and cellular proteins (Sanz-Ezquerro et al., 1995), with the N-terminal third of the protein being important for this activity (Sanz-Ezquerro et al., 1996). Specifically, mutation of residue T157 to alanine resulted in an inhibition of proteolysis. Furthermore, mutation T157A also resulted in an inhibition of cRNA synthesis while having no affect on mRNA synthesis (Perales et al., 2000). It was therefore hypothesised that proteolysis induction by PA was required for replication activity (Perales et al., 2000). However, later reports showed that inhibition of proteolytic activity did not decrease replication (Naffakh et al., 2001).
Furthermore, a serine protease active site has been identified at position S624 (Hara et al., 2001) and although this was shown not to be essential for viral growth, it is required for maximal viral growth (Toyoda et al., 2003). However, the significance of a protease or protease induction function by the PA subunit remains to be established, although it has been suggested that this function may be important for the degradation of cellular RNA polymerase II (Pol II) late during the viral life cycle (Rodriguez et al., 2007).
Further functions have also been described for PA. Mutation at position 638 results in the production of defective interfering particles. As such, it was hypothesised that PA may have a role as an elongation factor for the polymerase complex (Fodor et al., 2003). Temperature sensitive mutation L226P suggested a role for PA in assembly and maturation of the polymerase complex (Kawaguchi et al., 2005) and double mutation G507A, R508A suggested a role for PA in assembly or release of viral particles from the infected cell (Regan et al., 2006). Mutations at positions 102, 108 and 134 were shown to inhibit vRNA and cRNA promoter binding by the polymerase complex and the 109-117 region was shown to be important for stability of the PA subunit (Hara et al., 2006).
Moreover, PA has been shown to interact with cellular proteins. PA interacts with hCLE, recently demonstrated to be a transcriptional modulator of Pol II (Huarte et al., 2001; Perez-Gonzalez et al., 2006), as well as a DNA replicative helicase MCM, required for stabilisation of virus replicative complexes, allowing elongation of the replication product (Kawaguchi and Nagata, 2007). Finally, position 515 has recently been proposed to be involved in virus pathogenicity in ducks with mutation at this position resulting in wild type levels of viral replication but reduced pathogenicity in orally infected ducks (Hulse-Post et al., 2007).
Despite the number of functional studies performed the role of PA in the polymerase complex remains unclear. As the only acidic protein in the polymerase complex, it is possible that PA may perform a role in modulating polymerase activity. However, further work is required to ascertain if this is the case, and data supporting this, is presented in Chapter 4 of this thesis.
Each segment of the viral genome has 13 and 12 conserved nucleotides its 5′ and 3′ ends. These nucleotides, together with 2-3 segment specific nucleotides at each end of the segment comprise the vRNA promoter (Fig. 1.4A). Complementary sequences in each cRNA segment comprise the cRNA promoter (Fig. 1.4B). Further characterisation of the vRNA and cRNA promoter structures is discussed in 1.8.1 and 1.8.2, below.
The vRNA promoter
Early in vivo studies of the vRNA promoter suggested that the 22 5′ terminal and 26 3′ terminal nucleotides were sufficient to signal RNA transcription, replication and packaging into particles (Luytjes et al., 1989). However, in vitro studies suggested a requirement for only the 3′ terminal nucleotides, specifically the terminal 12-14 nucleotides (Piccone et al., 1993; Seong and Brownlee, 1992a; Seong and Brownlee, 1992b; Yamanaka et al., 1991). Later, a requirement for the 5′ terminus was highlighted by the specific binding of this sequence to the polymerase (Fodor et al., 1994; Fodor et al., 1993; Klumpp et al., 1997; Tiley et al., 1994). Earlier work may not have recognised the requirement for the 5′ end of the vRNA strand due to the method of study. These studies used polymerase purified from virions and depleted of viral RNA. However, this method is unlikely to have completely removed all endogenous RNA.
Partial complementarity between the 5′ and 3′ ends of each vRNA segment result in the formation of a “panhandle” structure (Desselberger et al., 1980; Robertson, 1979; Skehel and Hay, 1978). Evidence that the panhandle structure exists in vRNA, both in virus particles and in infected cells, has been provided by cross-linking (Hsu et al., 1987) as well as NMR spectroscopy (Cheong et al., 1996), electron microscopy (Hsu et al., 1987; Klumpp et al., 1997) and structural studies using enzymatic and chemical probes (Baudin et al., 1994). Further structural analysis of the vRNA promoter proposed that base pairing between residues 11-16 and 10-15 of the 5′ and 3′ ends, respectively, was required for promoter activity whilst residues 1-10 and 1-9 of the 5′ and 3′ strands remained open (Fodor et al., 1994; Fodor et al., 1995; Kim et al., 1997; Neumann and Hobom, 1995). This is known as the RNA-fork model. Later, the RNA-fork model was refined and the corkscrew model was proposed (Flick and Hobom, 1999; Flick et al., 1996). In this model, in addition to the base-pairing described in the RNA-fork model, short hairpin loops were proposed to be formed by the nucleotides in the 5′ and 3′ termini (Fig 1.4A). This is the currently accepted model for vRNA promoter structure.
Functional studies for the sequence requirements of the vRNA promoter for different polymerase activities have been performed. It has been demonstrated that there is a requirement for binding of the 5′ end of the vRNA promoter by the polymerase to activate cap-binding activity (Cianci et al., 1995). However, recent work has suggested that binding of the 3′ end is also important (Lee et al., 2003b). Furthermore, endonuclease activity is dependent upon the short hairpin-loops structures in both the 5′ and 3′ termini (Leahy et al., 2001a; Leahy et al., 2001b), as well as the hinge at position 10 from the 5′ end (Leahy et al., 2002). However, it has been described that where endonuclease cleavage occurs following a CA sequence, only the 5′ end of the vRNA promoter is necessary (Rao et al., 2003).
1.8.2The cRNA promoter
The cRNA promoter, like the vRNA promoter, is comprised of conserved sequences at the 5′ and 3′ ends of each segment. The vRNA promoter allows cap-snatching followed by polyadenylation whilst initiation on the cRNA promoter is de novo followed by anti-termination, resulting in a full length vRNA transcript. Furthermore, there is differential regulation of vRNA and cRNA synthesis late in the virus life cycle, favouring vRNA synthesis for packaging into new virions (Fodor and Brownlee, 2002; Palese and Shaw, 2007). This suggests that there must be important differences between the vRNA and cRNA promoters to regulate these different activities.
However, the cRNA promoter has been less intensively studied than the vRNA promoter. Initial work suggested that the cRNA promoter formed a panhandle structure (Pritlove et al., 1995; Seong and Brownlee, 1992a) and that the 3′ end of the cRNA segment formed the promoter (Seong and Brownlee, 1992a). In support of this the 3′ end of the cRNA segment was shown to be important because most mutations in this region reduced activity (Li and Palese, 1992). Another study suggested that, similar to the vRNA promoter, the cRNA promoter is comprised of sequences at both the 5′ and 3′ ends and forms a corkscrew structure (Azzeh et al., 2001). However, this study used a mutated cRNA promoter and indirect analysis of activity via a conventional CAT activity assay. Despite this, further work involving systematic mutagenesis of the cRNA 5′ and 3′ nucleotides and direct determination of in vivo activity by analysis of RNA levels supported a corkscrew structure for the cRNA promoter (Crow et al., 2004) (Fig. 1.4B).
Initiation of replication on the vRNA and cRNA promoters
As described in 1.8 above, the vRNA and cRNA promoters are both proposed to adopt corkscrew structures formed by the 5′ and 3′ terminal sequences of each vRNA and cRNA segment. Both the vRNA and the cRNA promoters signal for de novo initiation for replication of the genome. An in vivo analysis of the cRNA promoter had demonstrated that mutation at positions 4, 5 and 7 from the 3′ end of the segment inhibited replication activity (Crow et al., 2004). These mutations were unlikely to alter secondary structure of the cRNA promoter so the reason for the inhibition of activity was not understood. A recent study has analysed in detail the mechanism of initiation of replication on both promoters in vitro and has identified important differences (Deng et al., 2006b). By mutagenesis of the 3′ ends of both the vRNA and cRNA promoters it was identified that mutation of positions 1 and 2 of the vRNA promoter inhibited in vitro initiation of replication activity.
However, mutation of positions 1 and 2 of the cRNA promoter had little affect on activity and, instead, mutation of positions 4 and 5 resulted in an inhibition of initiation of replication as well as synthesis of a de novo initiated 15 nucleotide product (Deng et al., 2006b). The authors hypothesised that in the presence of the vRNA promoter, initiation occurs terminally at positions 1 and 2 (Fig. 1.5A), whereas in the presence of the cRNA promoter, initiation occurs internally at positions 4 and 5, followed by realignment of the pppApG product to positions 1 and 2 and subsequent priming of further transcription (Fig. 1.5B). Further support for this hypothesis was provided by vivo and in vitro experiments showing that the pppApG synthesised internally on the cRNA promoter could be released to prime replication at positions 1 and 2 and also to correct mutations in the cRNA promoter where position 1 had been mutated or deleted (Deng et al., 2006b). The implication of the polymerase having different positions of initiation on the two promoters is that there are likely to be different mechanisms for polymerase binding to the two promoters. Further data in support of this hypothesis is presented in Chapter 4.
Promoter binding by the polymerase
To allow transcription and replication of the vRNA genome, the polymerase complex must recognise and bind to the vRNA and cRNA promoters. It has been shown previously that binding of the polymerase complex to the vRNA promoter confers heat stability, with both the 5′ and 3′ being required for efficient heat stabilisation (Brownlee and Sharps, 2002). This implies that there are binding sites for both the 5′ and 3′ ends of the vRNA promoter within the polymerase complex.
Several studies have suggested that both the vRNA and cRNA promoters interact with all three polymerase subunits (Deng et al., 2005; Fodor et al., 1994; Fodor et al., 1993; Hara et al., 2006; Jung and Brownlee, 2006; Pritlove et al., 1995; Tiley et al., 1994). However, most studies have focused on binding of the promoters by the PB1 subunit. To identify the regions of the polymerase that interact with the vRNA promoter, Li et al. cross-linked recombinant polymerase to the 5′ end of the vRNA promoter where a 4 thio-U residue had been inserted at position 15, followed by partial protease digestion and sequencing of cross-linked peptides.
This mapped a vRNA 5′ binding site to the 560-574 region of the PB1 subunit (Li et al., 2001). Subsequent mutagenesis of basic residues in the peptide identified that positions R571 and R572 were required for binding (Li et al., 1998). Similar work was performed to identify regions of the polymerase required for binding to the 3′ end of the vRNA promoter where a 4 thio-U residue had been inserted at position 10. Region 249-260 in the PB1 subunit was mapped as the interaction domain, with F251 and F254 being essential for the interaction (Li et al., 1998).
A second study to identify the binding sites for the vRNA promoter used deletion mutants of the PB1 subunit alone, without the other polymerase subunits. In this study, regions 1-84 and 494-757 of PB1 were mapped for vRNA 5′ binding (Gonzalez and Ortin, 1999a). Furthermore, binding to the 3′ end only occurred efficiently in the presence of the 5′ end, and binding to the 5′ end, which did occur alone, was more efficient in the presence of the 3′ end of the vRNA promoter (Gonzalez and Ortin, 1999a).
In a final study, evolutionary conserved basic residues in the region proximal to R571 and R572 in PB1 identified by Li et al. (1998) were mutated to alanine to determine their involvement in vRNA promoter binding. In this study, the region 233-249 of PB1 was proposed to be the binding site for the vRNA 5′ end (Jung and Brownlee, 2006). Furthermore, in this study, mutation of R571 and R572 was not observed to inhibit vRNA binding. Therefore, much controversy still remains in the sites involved in vRNA binding.
Binding of the polymerase to the cRNA promoter has not been studied extensively. Gonzalez and Ortin demonstrated that PB1 interacts with the cRNA promoter over the regions 1-139 and 267-493 (Gonzalez and Ortin, 1999b). Competition experiments demonstrated a partial overlap in the vRNA and cRNA promoter binding sites in the N-terminus of PB1. However, further work is required to fully characterise cRNA promoter binding sites in the polymerase complex.
In addition to these studies investigating binding sites in the PB1 protein for vRNA and cRNA promoters, one study has demonstrated that alanine mutation at positions 102, 108 and 134 in the PA subunit reduced vRNA and cRNA promoter binding by the polymerase complex (Hara et al., 2006). However, further work is needed to determine the regions of PA and PB2 required for vRNA and cRNA promoter binding and to confirm the regions in PB1 required for promoter binding. Furthermore, studies to determine any differences in binding strategies by the polymerase complex for the vRNA and cRNA promoters are necessary.
Overview of cellular proteasomal protein degradation
Cellular misfolded or damaged proteins are degraded by the proteasome, a large cylindrical, multicatalytic protease complex (Peters, 1994). Proteins that have been damaged by e.g. high temperatures, oxidation or partial digestion by cytosolic proteases or nascent proteins that are unable to fold following synthesis are recognised by cellular chaperones e.g. Hsp70/40 via exposed hydrophobic domains (Goldberg, 2003; Hartl and Hayer-Hartl, 2002). This can trigger refolding of the protein or ubiquitination and targeting to the proteasome for degradation (degradation pathway summarised in Fig. 1.6).
Ubiquitin is transferred onto the E2 ubiquitin carrier proteins in a process requiring ATP and the E1 ubiquitin activating enzyme. E2, in the presence of an E3 ubiquitin ligase, transfers the ubiquitin onto the target protein. This process is repeated, generating a poly-ubiquitin chain (Ciechanover, 2006; Goldberg, 2003). Multiple E2 and E3 proteins exist within the cell and each of these possess specificity for different protein substrates, resulting in selectivity of ubiquitination (Goldberg, 2003). Following ubiquitination, target proteins are degraded by the 26S proteasome, resulting in short peptides. These are further degraded to amino acids by cytosolic proteases (Ciechanover, 2006; Goldberg, 2003).
However, it is unclear whether ubiquitination is essential for the degradation of all proteins by the proteasome. Degradation of abnormal proteins via the proteasome occurs in both bacteria and archea, but these organisms lack ubiquitin (Goldberg, 2003). Furthermore, proteins have been identified that can be degraded both in vitro and in vivo by the proteasome by a ubiquitin independent mechanism (Asher et al., 2002; Chen et al., 2007b; Tarcsa et al., 2000).
Overview of cellular classical nuclear import pathway
Trafficking of proteins from the cytoplasm to the nucleus in eukaryotic cells is a tightly controlled process. Molecules under 9 nm in diameter can diffuse passively through the nuclear pore complex (Weis, 1998). However, molecules larger than this (larger than approximately 40-60 kDa) must be actively transported into the nucleus (Stewart, 2007; Weis, 1998). In the classical nuclear import pathway (summarised in Fig 1.7), members of the karyopherin α family bind to cargo proteins via a nuclear localisation signal (NLS) (Lange et al., 2007).
Classical NLSs contain basic clusters of amino acids (Lange et al., 2007) and are classed as either monopartite or bipartite according to the number of clusters. Although, there is no strict sequence consensus, monopartite NLSs have a loose consensus of K(K/R)X(K/R) and bipartite NLSs have a loose consensus of 2 basic residues, a non-conserved 10-12 amino acid spacer and a second basic cluster containing three basic residues out of five (Dingwall and Laskey, 1991). Non-classical NLSs that are not classed as either monopartite or bipartite NLSs have also been identified (Chen et al., 2005; Wolff et al., 2002).
Following the interaction between karyopherin α and its cargo protein, karyopherin α can interact with karyopherin β. Karyopherin β then interacts with nucleoporins in the nuclear pore complex (NPC), facilitating transport of the cargo-karyopheirn α-karyopheirn β complex into the nucleus. Once in the nucleus, the higher concentration of Ran, a GTPase, in a GTP bound state, compared to the concentration in the cytoplasm triggers dissociation of the complex, release of the cargo into the nucleus and transport of karyopherins α and β back to the cytoplasm (Stewart, 2007). In human cells, there are six members of the karyopherin α family (Kohler et al., 1997; Kohler et al., 1999). As a result of this there is a degree of redundancy in binding of cargo proteins by the karyopherin α family, however, different members of the family do appear to exhibit specificity of cargo binding (Kohler et al., 1999; Quensel et al., 2004).
Alternative routes for transport of proteins into the nucleus also exist. Cargo proteins can interact directly with karyopherin β (Kurisaki et al., 2001; Moore et al., 1999). Furthermore, nuclear import of ribosomal proteins, histones and heterologous ribonucleoproteins utilise different carrier proteins as well as distinct NLSs (Stewart, 2007).
Nuclear import of the polymerase subunits
Following synthesis in the cytoplasm of the infected cell, all three subunits of the influenza RNA polymerase complex must be imported from the cytoplasm into the nucleus and this process is coupled with assembly of the trimeric complex and assembly of the RNP. All three subunits have been shown to contain nuclear localisation signals (Jones et al., 1986; Mukaigawa and Nayak, 1991; Nath and Nayak, 1990; Nieto et al., 1994; Perales et al., 1996; Tarendeau et al., 2007). Furthermore, immunofluorescence techniques have been used to show that each of the RNA polymerase subunits can be transported into the nucleus, independently of one another (Akkina et al., 1987; Fodor and Smith, 2004; Jones et al., 1986; Mukaigawa and Nayak, 1991; Nath and Nayak, 1990; Nieto et al., 1994; Nieto et al., 1992; Smith et al., 1987). However, only PB2 is efficiently transported into the nucleus when expressed alone. Efficient nuclear targeting of PA and PB1 requires co-expression of the two subunits (Fodor and Smith, 2004; Nieto et al., 1992).
Recent work has demonstrated an interaction between a bipartite NLS in the C-terminus of PB2 and karyopherin α1 (importin α5) and that this interaction is sufficient for nuclear import of PB2 (Tarendeau et al., 2007). Furthermore, another recent study has proposed an importance for Hsp90 in nuclear transport of PB2 and in assembly of the trimeric complex (Naito et al., 2007). Hsp90 may act as a chaperone to stabilise PB2 prior to assembly of the polymerase complex (Naito et al., 2007) and to facilitate interaction between PB2 and the import receptor, similar to the proposed role played by another cellular chaperone, Hsp70, in nuclear import (Mattaj and Englmeier, 1998; Melchior and Gerace, 1995).
An interaction has been demonstrated between the PB1 subunit and RanBP5, a nuclear transport receptor belonging to the karyopherin β family involved in nuclear import of ribosomal proteins and HIV-1 rev (Arnold et al., 2006; Jäkel and Görlich, 1998). This import factor was shown to interact with PB1 monomers as well as PB1 in the [PB1, PA] dimer (Deng et al., 2006a). Furthermore, RNAi knockdown of RanBP5 expression inhibited nuclear import of PB1 alone and of the [PB1, PA] dimer (Deng et al., 2006a). However, the region of PB1 required for this interaction has not been characterised.
Currently, no interactions have been described between PA and cellular nuclear import proteins. However, PA has been shown to contain two NLSs in the N-terminus of the protein, one a bipartite NLS and one a non-classical NLS (Nieto et al., 1994). Furthermore, as PA is, at least partially, imported into the nucleus when expressed alone (Fodor and Smith, 2004; Nieto et al., 1994; Nieto et al., 1992), and exceeds the size limit for passive diffusion through the nuclear pore (Stewart, 2007; Weis, 1998), PA must interact with an import protein, and these remain to be identified.
The mechanism of assembly of the trimeric complex remains controversial. Initial observations that the three subunits could enter the nucleus independently suggested that assembly occurred in the nucleus, following import of the monomers. However, it could not be ruled out that the complex assembles in the cytoplasm and is then transported into the nucleus. In fact, it has been reported that [PB1, PA] and [PB1, PB2] dimers exist. Although it has been reported that [PB1, PB2] dimers are imported into the nucleus (Naito et al., 2007), an in vitro assembly model has suggested that for functional polymerase, PB1 must bind to PA before binding to PB2 (Deng et al., 2005).
When taken in conjunction with the observation that PB1 and PA are only efficiently imported into the nucleus when co-expressed, as well as the binding of RanBP5 to the PB1 monomer and [PB1, PA] dimer, this would suggest that assembly of the [PB1, PA] dimer occurs in the cytoplasm, followed by transport into the nucleus where assembly with the PB2 monomer can occur (Deng et al., 2006a; Deng et al., 2005; Fodor and Smith, 2004). Despite the fact that nuclear import of the [PB1, PA] dimer has been demonstrated to be mediated by RanBP5 binding to PB1, due to redundancy in the import pathways in the host cell and the presence of NLSs in PA, a role for PA in mediating nuclear import cannot be ruled out. Therefore, it would be interesting to identify import factors that bind to the PA monomer as well as the [PB1, PA] dimer. Data is presented in Chapter 6 that shows an interaction between the PA monomer and members of the karyopherin α family.
Outline of thesis
This thesis focuses on the characterisation of the PA subunit of the influenza A virus polymerase complex. To allow accurate comparison of the activity of polymerases with different PA subunits, a technique was developed for quantitating relative levels of PA by SDS-PAGE followed by silver staining (Chapter 3). Additionally, a linear dependence of activity upon polymerase concentration in four in vitro polymerase activity assays was confirmed (Chapter 3).
Since differences in the mechanism of polymerase binding to the vRNA and cRNA promoters has not been extensively studied, analysis of the involvement of the three polymerase subunits in binding to the vRNA and cRNA promoters was performed (Chapter 4). Further to these experiments, a region towards the N-terminus of PA was studied by alanine scanning mutagenesis followed by systematic in vitro and in vivo analysis of polymerase activity. Through this analysis, a region of PA was found to be directly or indirectly involved in cRNA promoter binding (Chapter 4). Furthermore, residues were identified that were important for efficient polymerase complex formation. Further analysis identified that mutations of these residues resulted in proteasomal degradation of PA, possibly as a result of protein misfolding (Chapter 5).
A final aim of this thesis was to study nuclear import of PA. PA was found to interact with cellular karyopherin αs but this was not mediated by the proposed bipartite NLS in the region 124-139 of PA (Chapter 6). However, nuclear localisation of PA was partially inhibited by mutation of the bipartite NLS, supporting a functional role for this NLS. In contrast, the bipartite NLS was found to have only a minor role in nuclear targeting of the [PB1, PA] dimer (Chapter 6).