Viral–host interactions during splicing and nuclear export of influenza virus mRNAs

Abstract

As influenza-A viruses (IAV) replicate in the host cell nucleus, intranuclear pathways are usurped for viral gene expression. The eight genomic viral ribonucleoproteins (vRNPs) segments of IAV are transcribed and two generate viral mRNAs (M and NS) that undergo alternative splicing followed by export from the nucleus. The focus of this review is on viral RNA splicing and nuclear export. Recent mechanistic advances on M and NS splicing show differential regulation by RNA-binding proteins as well as distinct intranuclear localization. After a review of IAV splicing, we will discuss the nuclear export of viral mRNAs, which occur by interacting with specific constituents of the host mRNA export machinery that translocate viral mRNAs through the nuclear pore complex for translation in the cytoplasm.

Introduction

Influenza-A viruses (IAV) remain a public health problem, causing high mortality rates annually and during pandemic years. The virus enters the host cell by interacting with sialic acids on the plasma membrane through its HA (hemagglutinin) protein. The virus is then internalized by endocytosis and the low endosomal pH induces a conformational change in the HA protein, allowing an N-terminal peptide to insert into the endosomal membrane, resulting in fusion with the viral membrane. Additionally, the viral M2 (matrix protein-2) ion channel acidifies the interior of the virus particle by pumping protons from the endosome to the interior of the virus particle. This enables dissociation of the M1 (matrix protein-1) matrix-coat protein and release of the viral genomic RNA segments (vRNAs) into the host cytoplasm upon fusion of the viral and endosomal membranes [1] (Figure 1a).

Figure 1.

Influenza virus life cycle. (a) The viral HA protein interacts with sialic acids present on the host cell membrane and the virus enters the cell via endocytosis. The viral M2 ion channel transports hydrogen ions into the viral particle, facilitating release of the vRNPs once the viral membrane fuses with the endosomal membrane. vRNPs are transported into the nucleus for transcription and replication. A subset of these viral mRNAs are spliced before export into the cytoplasm for translation. Some viral proteins are imported into the nucleus for aiding in transcription and replication. Later in the life cycle, vRNPs are copied into cRNPs, which serve as an intermediate to make additional copies of vRNPs to generate new virions. The vRNPs are exported from the nucleus to the cytoplasm and together with membrane-bound and nonmembrane-bound proteins are assembled into new viral particles that then bud from the cellular membrane. (b) Influenza virus particle. The viral membrane contains three different membrane-bound proteins, including HA, NA, and the M2 ion channel. The viral genome consists of eight different vRNPs. After transcription, the M and NS mRNAs are alternatively spliced into the depicted isoforms.

After release into the host cell cytoplasm, the eight unique vRNAs, which are bound by nucleoproteins (NP) and the subunits for the viral RNA-dependent RNA polymerase [polymerase basic 1 (PB1) subunit, polymerase basic 2 (PB2) subunit, and polymerase acidic (PA) subunit], are imported into the host cell nucleus via the nuclear pore complex (NPC). Once in the nucleus, the eight vRNAs are transcribed into mRNAs (Figure 1a). Two of the eight mRNAs undergo alternative splicing. Unspliced M1 mRNA is translated into M1 protein and is also spliced into the M2 mRNA, which encodes the M2 ion channel that functions in viral entry [1] (Figure 1b). The M2 protein is also important for viral budding and inhibition of autophagy, while M1 protein has a role in viral intracellular trafficking and as a structural component of the infectious virions. Thus, both M1 and M2 proteins are critical for viral replication. Furthermore, M1 mRNA is also spliced into mRNA3, which encodes a peptide with unknown function, and M4 mRNA that encodes an isoform of M2 protein present in specific viral strains [2] (Figure 1b). The other mRNA that is spliced is NS1 mRNA, which is generated upon transcription of the NS vRNA segment. Unspliced NS1 mRNA encodes the Non-Structural Protein-1 (NS1) and is also spliced to generate the Nuclear Export Protein or Non-Structural Protein-2 (NEP/NS2) (Figure 1b). The NS1 protein is a major virulence factor that downregulates interferon signaling, modulates the PI3K pathway to regulate apoptosis and viral replication, and inhibits host gene expression by multiple mechanisms, including effects on cellular splicing, 3’-end mRNA processing, and nuclear export [3-5]. However, NS1 also promotes viral M1 mRNA to M2 mRNA splicing [6-8] and nuclear export [6,8,9], and regulates viral RNA synthesis and translation [3]. NS2 plays a role in nuclear export of the eight genomic RNP segments (vRNPs) and therefore is also critical for viral replication [1].

Once transcribed, the viral mRNAs are exported from the nucleus and then translated into proteins in the cytoplasm. In the nucleus, vRNPs are transcribed into positive-strand complementary RNPs (cRNPs), an intermediate that will then be transcribed into vRNPs. The newly synthesized vRNPs are exported from the nucleus to the cytoplasm via M1 protein interaction with the NP and NEP/NS2 proteins. NEP interaction with the cellular nuclear export receptor CRM1/XPO1 (Chromosome Region Maintenance 1; Exportin 1) mediates nuclear export of vRNPs, which are then localized to the plasma membrane for new virion assembly [1] (Figure 1a).

Influenza virus RNA splicing

Splicing of the M1 mRNA occurs at nuclear speckles [6], also known as interchromatin granule clusters. Mammalian cells have ~20–50 nuclear speckles per nucleus, each containing proteins involved in splicing, RNA processing, and mRNA export [10]. Earlier models proposed that splicing factors would leave nuclear speckles to function at active transcription sites in the nucleoplasm [11]. However, active spliceosomes [12] and polyadenylated RNAs are found at nuclear speckles and a subset of mRNAs requires localization to nuclear speckles for splicing and/or nuclear export [13-18]. This is in contrast to the majority of host cell mRNAs that are cotranscriptionally spliced.

A combination of cell biological and biochemical results supports a model in which the viral NS1 protein interacts with the cellular protein NS1-BP, which together with heterogeneous nuclear ribonucleoprotein K (hnRNP K), binds M1 mRNA in the nucleoplasm and transports it to nuclear speckles [6-8,19]. This complex stabilizes the NS1-BP binding to a pY tract in M1 mRNA, which is downstream of the M2 mRNA 5’ splice site (ss), and inhibits premature splicing. Upon localization to nuclear speckles, NS1 dissociates from the complex and hnRNP K recruits U1 snRNP to the M2 5’ ss to initiate splicing. This recruitment would be facilitated by the high concentration of U1 snRNP and/or by NS1 dissociation. Recruitment of U1 snRNP is then sufficient to drive the subsequent steps of the splicing pathway (Figure 2). Perturbation of several of these components has shown the role that each play in this pathway. Mutation on the pY tract decreases the interaction between the M1 mRNA and NS1-BP, although this mutation is not able to disrupt the NS1-BP interaction with NS1 and hnRNP K. The destabilization between NS1-BP and the pY-tract mutant M1 mRNA promotes splicing and highlights the potential role of the complex in regulating splice timing. The major binding site of hnRNP K on M1 mRNA is a pC tract proximal to the M2 mRNA 5’ss. In contrast to mutating the pY tract, mutation of the pC tract or depletion of hnRNP K results in decreased splicing due to failure to properly recruit U1 snRNA [7]. Depletion of NS1-BP or infection with an influenza virus strain lacking NS1 leads to M1 transcripts no longer being localized to nuclear speckles [6], resulting in decreased splicing [6-8,19]. However, hnRNP K is still able to associate with M1 mRNA in the presence or absence of NS1; though without localizing to nuclear speckles to interact with splicing factors, no splicing can occur, even with hnRNP K binding to facilitate recruitment of these splicing factors. These observations underscore the role of NS1-BP and NS1 in transporting the M1 transcript to nuclear speckles for splicing (Figure 2).

Figure 2.

Splicing and nuclear export of influenza virus M1 mRNA. Studies in the past few years led to the identification of an intranuclear pathway for the influenza virus M1 mRNA via nuclear speckles. The results suggested a model in which the viral NS1 protein interacts with the cellular NS1-BP protein to target the M1 mRNA to nuclear speckles. At these intranuclear bodies, NS1 is dissociated from M1 mRNA, allowing the recruitment of U1 snRNA by hnRNP K to initiate the splicing reaction of M1–M2 mRNA. Nuclear speckles have active spliceosomes. The nuclear speckle assembly and splicing factor SON (son3 fragment of the human genome) binds the NS1-BP/hnRNP K complex and promotes M1 to M2 splicing. After splicing, unspliced M1 and M2 mRNAs are exported from the nucleus by associating with the depicted cellular mRNA export factors. NS1 and NS1-BP are also important at this step to promote viral mRNA export. This mechanism is currently being studied. See text for details.

The cellular NS1-BP protein, a member of the kelch-repeat superfamily, was originally identified as a binding partner of the viral NS1 protein [20] with a pool localized at nuclear speckles [6,20]. NS1-BP has an N-terminal BTB [broad-complex, tramtrack, and bric-a-brac; also known as POZ (Pox virus and zinc finger)] domain, which is important for dimerization and splicing function [8]. The BTB domain is followed by the BACK (BTB and C-terminal Kelch) domain, which directly interacts with the viral NS1 protein, the cellular splicing factors hnRNP K and polypyrimidine tract-binding protein 1 (PTBP1), and the mRNA export factor Aly/REF (ally of AML1 and LEF1; RNA Export Factor) [8]. The latter is required for M1 and M2 mRNA nuclear export (mRNA export will be further discussed in the section below). These interactions of NS1-BP with splicing and export factors likely occur at different steps of the mRNA-processing pathway. Similar to NS1-BP localization, a pool of hnRNP K is also found at nuclear speckles [6]. While hnRNP K binds NS1-BP independently of RNA, the presence of RNA increases the interaction [19]. hnRNP K has 3 KH-type RNA-binding domains and regulates alternative splicing of viral and host mRNAs [21-24]. Based on the model described above, NS1-BP has two functions that contribute to the intranuclear trafficking and splicing of M1–M2 mRNAs. With mutation of the pY tract, NS1-BP is no longer able to directly bind M1, resulting in increased splicing without disrupting nuclear transport. These data together with depletion of NS1-BP, which inhibits M1 targeting to nuclear speckles, likely explain the differences in requirements for transport to nuclear speckles versus splicing. NS1-BP also interacts with the viral polymerase [19], which was previously shown to regulate the choice of 5’ss that favors M1–M2 splicing [25]. Therefore, it is possible that this interaction regulates M2 splicing at nuclear speckles. Overall, NS1-BP appears to be critical for spatial and temporal regulation of M1 splicing by controlling intranuclear localization of the M1 mRNA and by regulating splicing. While NS1-BP may limit access to the splicing machinery and dissociation from the pY tract may be sufficient to allow U1 snRNP access to the M2 5’ splicing site, hnRNP K is likely the RNA-binding protein that recruits U1 snRNP. This is based on mutating the pY tract, which leads to increased M2 production compared with the pCpY double mutation that did not further enhance the M2 levels [7]. In addition, hnRNP K binds constituents of U1 snRNP [26,27], supporting the model for hnRNP K-dependent U1 snRNP recruitment. However, the binding of NS1-BP to M1 mRNA is weak and we do not exclude the possibility that NS1-BP contributes to U1 snRNP recruitment. In fact, we showed that NS1-BP interacts with the U1A component of U1 snRNP in an RNA-dependent manner, which could potentially have a role in the overall stability of the NS1-BP/M1 mRNA complex [8]. There are cellular splicing events that are coregulated by NS1-BP and hnRNP K. Among them are mRNAs encoding interleukin-15 and cortactin [7]. Similar to the viral M transcript, these cellular genes have adjacent pY- and pC tracts downstream of the promoted 5’ splicing site. In sum, these data suggest that the influenza virus utilizes a noncanonical pathway for splicing and nuclear export through nuclear speckles. As most of the vRNA is not associated with nuclear speckles [6], this is likely a post-transcriptional splicing event as opposed to cotranscriptional splicing.

The other viral mRNA that is alternatively spliced is NS1 mRNA, which is spliced into NS2 mRNA. This splicing event does not appear to follow the M mRNA pathway through nuclear speckles [6], therefore, it likely takes place in the nucleoplasm. While NS1 protein was shown not to alter NS1–NS2 mRNA splicing in the A/WSN/33 strain [28], a more recent study reported that NS1–NS2 splicing is inhibited by the viral NS1 protein and that NS1 directly interacts with the host splicing factor SF2 [29]. SF2 binds to multiple sites on NS1 mRNA, including an exonic splicing enhancer (ESE), to promote splicing. A G540A substitution in the ESE, which is present in a H7N9 strain, inhibited SF2 binding and splicing [29]. Additionally, the spliceosome factors RED [arginine (R)/glutamic acid (E) or arginine/aspartic acid (D) repeats] and SMU1 (suppressor of mec-8 and unc-52 1) were shown to regulate the expression levels of NS2. This regulation occurs via RED and SMU1 interactions with the RNA-dependent RNA polymerase subunits PB2 and PB1. Depletion of RED or SMU1 decreases NS2 splicing, NS2 protein levels, and viral replication [30]. This model suggests that these proteins are recruited to the weak 5’ splice site in the NS1 mRNA to promote splicing. Moreover, splicing of viral mRNAs from diverse strains can be differentially regulated by the same protein. One example is the cellular protein TRA2A that regulates splicing of NS mRNA from the PR8 strain and M mRNA from the YS/H5N1 strain [31]. TRA2A contains several RNA-recognition motifs and binds the Intronic Splicing Silencer (ISS) motif of both NS (PR8) and M (YS/H5N1) mRNAs, preventing splicing. Despite either M or NS mRNA splicing being inhibited in the respective strains, TRA2A knockdown in human cells results in increased virus replication of the avian YS/H5N1 strain and decreased replication of the human PR8 influenza strain [31]. Mutation on the ISS of M mRNA (YS/H5N1) increased M1–M2 splicing in human cells, suggesting a possible mechanism for host adaptation [31]. Other strains have developed mutations at the NS ESE to decrease splicing efficiency and slow expression of the NS2 protein. This is to properly time the viral life cycle by slowly accumulating NS2, as having a splice site optimized results in high expression levels early in the viral life cycle and decreased viral replication [32]. In sum, splicing of M and NS mRNAs is differentially regulated by specific RNA-binding protein complexes that also dictate intranuclear localization and processing of these mRNAs.

Nuclear export of influenza virus RNAs

To become export-competent, mRNAs bind the TRanscription and EXport (TREX) complex, which has multiple subunits, including the THO complex (transcription-dependent hyperrecombination complex), the RNA helicase U2AF65-associated protein 56 (UAP56), and the mRNA export adapter Aly/REF among others. Aly/REF then recruits the mRNA export receptor Nuclear RNA Export Factor 1–Nuclear Transport Factor 2-Like Export Factor 1 (NXF1–NXT1) heterodimer to the mRNA. NXF1•NXT1 docks the mRNA to the NPC and translocates it to the cytoplasm (Figure 3).

Figure 3.

Host mRNA nuclear export pathway. After transcription, splicing is followed by recruitment of mRNA nuclear export factors. The Cap-Binding Complex (CBC) bound to the 5’ cap recruits the TREX complex (THO, UAP56, and Aly/REF). Aly/REF then recruits NXF1 and NXT1 to the mRNA, followed by translocation through the NPC. Proteins in the NPC, including Nup98 (Nucleoporin 98), Rae1 (Ribonucleic Acid Export 1), Dbp5 (DEAD-Box Protein 5), and Gle1, facilitate export of the mRNA to the cytoplasm for translation.

Many cellular mRNAs are exported through the nuclear pore complex by interacting with the mRNA export receptor NXF1•NXT1, while some cellular mRNAs are exported via the CRM1 export receptor [33]. NXF1 was a hit in various siRNA screens that identified host factors involved in the IAV virus life cycle [34-36]. Knockdown of NXF1, without changes in cell viability, is reported to significantly inhibit virus replication in cells [37]. Additionally, IAV mRNAs M1, HA, NA, PB1, and NP were coimmunoprecipitated with NXF1 protein [9,38,39]. Knockdown of NXF1 and overexpression of a dominant-negative mutant of NXF1 resulted in inhibition of IAV mRNA export to the cytoplasm [37,38]. Previous results suggested that IAV mRNAs do not usurp the CRM1 pathway for their export [37-39]. Together, these results indicate that IAV mRNAs are exported via the NXF1–NXT1 pathway.

Moreover, nuclear export of IAV mRNAs to the cytoplasm showed differential sensitivity to NXF1 knockdown [37,38]. In one study in HEK 293T cells, M1, NS1, and HA mRNAs showed robust inhibition of mRNA export upon NXF1 knockdown, while NP and PB2 were less inhibited by depletion of NXF1 [37]. In another study in A549 cells, HA, NP, and NA mRNA export was shown to be dependent on NXF1, while PB2, PB1, and PA mRNAs did not show strong dependence on NXF1 [38]. However, in the same study, but in HEK 293T cells, both NP and PB2 mRNAs were not significantly dependent on NXF1, while HA and NA mRNAs were dependent on NXF1 for nuclear export [38]. Additionally, reduction in UAP56 levels inhibits nuclear export of M1, M2, and NS1 mRNAs [6,37]. In one study, no significant effect on PB2, HA, and NP mRNA export was observed upon UAP56 knockdown [37]. Furthermore, the mRNA export factor Aly/REF is required for efficient export of M1 and M2 mRNAs [6]. In another report, depletion of Aly/REF had little effect on nuclear export of PB2, HA, NP, M1, and NS1 [37]. Thus, the differences in mRNA export effects between these studies may be related to knockdown efficiency or cell type. Diverse cell types may use different mRNA export adapters/factors that lead to the recruitment of NXF1. Additionally, the different influenza virus mRNAs may have specific sequences/structure that bind diverse mRNA export adapters/factors that could result in differential mRNA export efficiency.

Splicing is known to promote mRNA export and to recruit the mRNA export complex TREX to the mRNA [40-43]. During splicing, the exon junction complex interacts with the mRNA and then recruits components of the TREX complex [44,45]. Among the IAV mRNAs, 6 are intronless and 2 are exported as unspliced intron-containing mRNAs. It is not fully understood how each unspliced IAV mRNA recruits NXF1. The influenza virulence factor NS1 protein is known to interact with NXF1 [9,46,47]. Additionally, NS1 promotes M1 and HA mRNA nuclear export [6,9] (Figure 2). These results suggest that NS1 may act as an adapter to recruit NXF1 on HA and M1 mRNAs. However, NS1 interaction with NXF1 inhibits nuclear export of cellular mRNAs [46-48] while promoting viral M and HA mRNA export [6,9]. These opposing functions are likely related to different constituents of the mRNA export complex that interact with NS1 and viral mRNAs versus NS1 and cellular mRNAs. To this end, NS1-BP, which interacts with NS1 [8,20], promotes viral M mRNA export, but is not required for nuclear export of bulk cellular mRNAs [6,49]. These results indicate that NS1 differentially interacts with viral versus cellular mRNAs, resulting in different outcomes. In this regard, the 2-((1H-benzo[d]imidazole-2-yl)thio)-N-(5-bromopyridin-2-yl) acetamide was identified as a robust chemical inhibitor of M and HA mRNA export, and with a slight inhibitory effect on NS1 mRNA export, but did not inhibit bulk cellular mRNA [49].

On the other hand, NP mRNA export is independent of the NS1 protein [9], suggesting an alternative mechanism of NXF1 recruitment to NP mRNA. IAV mRNAs may contain cis-acting RNA elements such as the constitutive transport element [50] or a cytoplasmic accumulation region to promote export of unspliced and intronless mRNAs [51]. The SR proteins 9G8 and SRp20 have been shown to promote the export of intronless cellular mRNAs by directly interacting with NXF1 [52]. It is possible that intronless IAV mRNAs use SR proteins and/or other factors to recruit NXF1.

In addition to the TREX complex, various host proteins are reported to play a role in IAV mRNA export. hnRNPAB together with the viral NP inhibits viral mRNA nuclear export via interfering with viral mRNA binding to Aly/REF and NXF1 [53]. hnRNP A2/B1 binds to NS1 mRNA and inhibits NS1 mRNA nuclear export [54]. Additionally, RNase L activation has been reported to also inhibit IAV mRNA export to the cytoplasm [55]. In the future, it will be important to fully define the mRNA export complexes that mediate the export of the different influenza virus mRNAs from the nucleus.

While the studies on the molecular mechanisms involved in splicing and nuclear export of IAV mRNAs have progressed in the recent years, many key fundamental questions remain to be elucidated. The molecular mechanisms and regulation of M1 and NS1 splicing need to be further investigated to provide a comprehensive molecular understanding of nuclear speckle targeting, splicing, and exit of viral mRNAs from nuclear speckles to the nucleoplasm. Regarding nuclear export, it would be important to clarify the differential NXF1 dependence of IAV mRNAs. Since splicing facilitates export [43], it would be critical to define the mRNA export factors and RNA sequence(s) or structure(s) involved in nuclear export of intronless IAV mRNAs. Additionally, it would be key to uncover the mechanisms by which the IAV M1 and NS1 mRNAs are exported without intron removal as intron-containing mRNAs are not prone for export [56]. In-depth understanding of these steps may reveal novel vulnerabilities that could be potentially used in the development of antiviral therapy.