Nuclear Export of Influenza A Virus mRNAs Requires Ongoing RNA Polymerase II Activity

Authors


Paul Digard, pd1@mole.bio.cam.ac.uk

Abstract

Influenza A virus transcribes its segmented negative sense RNA genome in the nuclei of infected cells in a process long known to require host RNA polymerase II (RNAP-II). RNA polymerase II synthesizes pre-mRNAs whose 5′-cap structures are scavenged by the viral RNA-dependent RNA polymerase during synthesis of viral mRNAs. Drugs that inhibit RNAP-II therefore block viral replication, but not necessarily solely by denying the viral polymerase a source of cap-donor molecules. We show here that 5,6-dichloro-1-β-D-ribofuranosyl-benzimidazole (DRB), a compound that prevents processive transcription by RNAP-II, inhibits expression of the viral HA, M1 and NS1 genes at the post-transcriptional level. Abundant quantities of functionally and structurally intact viral mRNAs are made in the presence of DRB but with the exception of NP and NS2 mRNAs, are not efficiently translated. Taking M1 and NP mRNAs as representatives of DRB-sensitive and insensitive mRNAs, respectively, we found that the block to translation operates at the level of nuclear export. Furthermore, removal of DRB reversed this block unless a variety of chemically and mechanistically distinct RNAP-II inhibitors were added instead. We conclude that influenza A virus replication requires RNAP-II activity not just to provide capped mRNA substrates but also to facilitate nuclear export of selected viral mRNAs.

Influenza A virus is a segmented RNA virus whose genome is expressed and replicated in the host cell nucleus (1). There, three species of RNA are synthesized for each of the eight segments. Initially, the viral RNA-dependent RNA polymerase transcribes the input negative sense vRNA segments into capped and polyadenylated mRNAs that undergo nuclear export and translation. After viral gene expression has become established, the viral polymerase also carries out genome replication by producing unit length positive sense cRNA copies of the genome that serve as a replicative intermediate to amplify vRNA [reviewed by Elton et al. (2)]. Both cRNAs and vRNAs are encapsidated into ribonucleoprotein (RNP) structures by the viral nucleoprotein (NP) and polymerase, but the prevailing view is that mRNAs are free of NP. The expression of individual virus genes is regulated, at least in part at the transcriptional level (3–7). At early stages of infection, expression of NP and NS1 polypeptides predominates, while later in infection, NS1 synthesis declines and M1 and HA expression increases (4). In addition, the primary mRNA transcripts from segments 7 and 8 that encode M1 and NS1 are partially spliced to produce the alternative M2 and NS2/NEP gene products, respectively (8,9).

It is well established that influenza virus replication depends on host cell RNAP-II function. A variety of drug and radiation treatments that interfere with DNA-dependent RNA synthesis block virus replication, unless (in the case of α-amanitin) the cells carry a mutant polymerase resistant to inhibition (10–15). Although initially enigmatic, an explanation for this observation was found in the mechanism of viral mRNA synthesis. Polyadenylation of the viral transcripts is achieved independently of cellular function by stuttering of the viral polymerase on a poly(U) stretch contained in each vRNA segment (2,16,17). However, capping occurs via a process known as cap snatching in which the viral polymerase generates capped oligonucleotides around 10–14 nucleotides long from cellular pre-mRNAs by endonucleolytic cleavage and then uses the capped fragment to prime transcription initiation (2,18–20). Inhibition of host RNA polymerase II (RNAP-II) at the start of infection thus deprives the viral polymerase of a source of cap-donor molecules, thereby blocking viral gene expression. However, it is not clear if this is the only reason the virus requires RNAP-II. Inhibitors such as actinomycin D (ActD), camptothecin (CP) and the isoquinoline kinase inhibitor H7 exhibit pleiotropic effects on virus replication depending on the time of addition and concentration of the drug used that cannot be satisfactorily explained solely by reference to inhibition of cap snatching (7,12,21–26). Moreover, in recent years, it has become clear that the multiple transcriptional and post-transcriptional events involved in the synthesis, nuclear export and translation of a mature cellular mRNA are functionally and physically linked through RNAP-II itself (27,28). With the exception of their first dozen or so nucleotides, fully polyadenylated influenza virus mRNAs are synthesized by the viral polymerase, yet they depend on cellular machinery for multiple downstream events, including internal methylation (29), splicing in certain cases and nuclear export. The question then arises of how and where they enter the cellular processing pathways.

The large subunit of RNAP-II contains 52 copies of a C-terminal heptamer repeat domain (CTD) whose phosphorylation state alters throughout the transcription cycle (27,30). Phosphorylation of serine 5 occurs at transcription initiation, while the transition to processive elongation is marked by phosphorylation of serine 2 and concomitantly, dephosphorylation of serine 5 (27,30,31). This regulated procession of post-translational modification is thought to play a key role in recruiting factors involved in post-transcriptional processing and transport of the pre-mRNA, many of which are loaded onto the nascent transcript (27,32). Recent work has shown that the viral polymerase preferentially associates with promoter-bound RNAP-II that is phosphorylated on serine 5 of the CTD (33,34), perhaps suggesting a means of introducing viral mRNAs into the beginning of the cellular mRNA maturation pathway. To further probe the interplay between viral and cellular transcription, we examined the effect of the adenosine analogue 5,6-dichloro-1-β-D-ribofuranosyl-benzimidazole (DRB) on virus replication. Originally characterized as an inhibitor of influenza A and B virus replication (11), subsequent work identified DRB as an inhibitor of RNAP-II that permits the synthesis of relatively short-capped RNAs but blocks the formation of long pre-mRNAs (35–37). The specific mechanism of action is thought to be inhibition of the cyclin-dependent kinase 9 component of the positive transcription elongation factor, a multienzymatic complex responsible for phosphorylation of serine 2 of the RNAP-II CTD that promotes the transition to the elongation phase of transcription (38–40). DRB thus provides a useful tool for examining the functional interplay between cellular and viral transcription, as the drug would not be expected to dramatically reduce the supply of capped pre-mRNA primers needed to initiate viral mRNA synthesis, or to directly affect the form of RNAP-II that the viral polymerase associates with. Indeed, a recent publication noted that pretreatment of cells with DRB reduced but did not abolish viral mRNA synthesis (33). We show here that when added at 90 min post-infection (p.i.), DRB inhibited virus gene expression without decreasing synthesis of capped viral mRNAs but instead blocked the nuclear export of specific transcripts. Furthermore, this blockade was not specific to DRB-mediated inhibition of RNAP-II but was mirrored by a variety of chemically and mechanistically distinct inhibitors of cellular transcription. We conclude that the influenza A virus life cycle has a second requirement for RNAP-II activity; to support nuclear export of certain viral mRNAs.

Results

To examine the interplay between RNAP-II transcription and influenza virus replication, we first tested the effect of DRB on viral gene expression. 293-T cells were infected with influenza virus A/PR/8/34 (PR8) and at 90 min post-infection, treated or mock treated with DRB, then pulse-labelled with 35S-methionine prior to harvesting at 4, 6 or 8 h p.i. In the absence of drug, the expected pattern of viral polypeptide synthesis was seen, with early synthesis of NP and NS1 and later synthesis of HA, M1 and NS2 (Figure 1A, lanes 1–3). In the presence of DRB, the overall rate of NP synthesis remained near normal, although with slightly delayed kinetics (Figure 1A, lanes 4–6). NS2 synthesis was somewhat reduced but was relatively resistant to DRB (Figure 1A, panel ii). However, synthesis of NS1, HA and M1 was significantly reduced, the latter two to below detectable levels. In confirmation of this, Western blot analysis showed that there was no substantial accumulation of HA, M1 or M2 in DRB-treated cells (Figure 1B, compare lanes 1–3 with 4–6). Similar overall results were obtained in Madin–Darby canine kidney (MDCK) cells (data not shown). Thus, DRB selectively blocks influenza virus gene expression, strongly inhibiting synthesis of three late gene products (HA, M1 and M2) and diminishing the expression of one early protein (NS1).

Figure 1.

Figure 1.

Effect of DRB on viral protein synthesis. 293-T cells were infected or mock (M) infected with PR8 virus, treated with DRB as labelled from 90 min p.i. and harvested at the indicated times p.i. Mock-infected samples were collected at 8 h p.i. A) Lysates from cells metabolically labelled with [35S]-methionine for 2-h periods ending at the indicated times p.i. were analysed by SDS–PAGE and autoradiography. Major viral polypeptides are indicated. The samples in panel ii) were analysed on a separate gel to resolve NS2. B, C) Western blot analysis of replicate samples using B) antibodies to detect i) M1, ii) HA or iii) M2 and C) RNAP-II using antisera reactive against i) unmodified serine 2 of the CTD (8wg16), ii) serine 5 phosphorylated CTD (H14) and iii) serine 2 (H5) phosphorylated form.

To test whether DRB was inhibiting phosphorylation of RNAP-II in our system, we analysed the phosphorylation status of the CTD heptamer repeat sequence by Western blot. DRB treatment or viral infection did not substantially alter the overall amounts of RNAP-II as judged by an antibody reactive against the serine 2 unphosphorylated form of the CTD repeat. Similarly, the amount of RNAP-II phosphorylated on serine 5 of the CTD repeat was not substantially altered by virus infection or drug treatment (Figure 1C, panels i and ii). However, DRB dramatically reduced the amount of phosphorylation at serine 2 in both infected and uninfected cells (Figure 1C, panel iii), consistent with previous literature (40).

To investigate the mechanism by which DRB inhibits the expression of specific influenza virus genes, we examined the accumulation of viral mRNAs. 293-T cells were infected or mock infected and treated with DRB as before. Total RNA from these cells was isolated and mRNA from segments 4, 5, 7 and 8 (encoding HA, NP, M1 and NS1, respectively) detected by primer extension analysis (41). No significant primer extension products were detected from RNA taken from mock-infected cells (Figure 2, lanes 9 and 10). In infected but untreated cells, maximal accumulation of all viral mRNA species tested was found between 4 and 6 h p.i. and amounts then declined (Figure 2, lanes 1–4). In contrast, in infected cells treated with DRB, mRNA continued to increase up to 8 h p.i. (the latest time-point taken). Furthermore, the amounts of mRNA were similar between untreated and treated cells at 4 h p.i. but thereafter increased substantially in treated cells (Figure 2, lanes 6–8). These experiments were repeated in MDCK cells with a similar outcome, except that the increase in mRNA synthesis over that of untreated cells was less pronounced for segments 4, 7 and 8 (data not shown). Therefore, DRB abolishes M1 and HA expression and severely reduces NS1 protein synthesis without significantly decreasing synthesis of their mRNAs.

Figure 2.

Figure 2.

DRB does not inhibit viral transcription. 293-T cells were infected or mock (M) infected with PR8 virus, treated with DRB as labelled from 90 min p.i. and total cellular RNA harvested at the indicated times p.i. Viral mRNAs from the indicated segments were detected by primer extension analysis followed by urea–PAGE and autoradiography. The oligonucleotides used for segments 7 and 8 are specific to the unspliced RNA species.

As DRB treatment results in the accumulation of increased amounts of viral mRNA but an apparent failure in translation of some of them, we tested whether the RNAs were functional in an in vitro system. Aliquots of micrococcal nuclease-treated rabbit reticulocyte lysate were programmed with samples of total cellular RNA isolated from cells as above and translation monitored by the incorporation of 35S-methionine. RNA from infected cells incubated in the absence of DRB resulted in the translation of NP, M1 and NS1 (Figure 3A, lanes 1–3). RNA isolated from infected, drug-treated cells also directed abundant translation of NP, M1 and NS1 (lanes 4–6), showing that although the latter two gene products are not produced in vivo, their mRNAs are not inherently untranslatable. RNA from uninfected cells produced minor quantities of a cellular protein, probably actin, that was not affected by DRB treatment (lanes 7 and 8).

Figure 3.

Figure 3.

Effect of DRB treatment on the integrity of viral mRNA. A) Total cellular RNA isolated at the indicated times p.i. from infected or mock (M) infected 293-T incubated in the presence or absence of DRB from 90 min p.i. was in vitro translated in the presence of [35S]-methionine. Radiolabelled products were separated by SDS–PAGE and detected by autoradiography. Major viral polypeptides are indicated on the left, and the position of molecular mass standards on the right. B) RNA isolated from infected DRB-treated (+DRB) or untreated (−DRB) 293-T cells at 8 h p.i was analysed by primer extension, urea–PAGE and autoradiography for the indicated viral RNAs before (T) or after fractionation into polyadenylated (+) and non-polyadenylated (−) pools.

Translation of mRNA in the reticulocyte lysate system is less dependent on a 5′-cap structure or 3′-poly(A) tail than translation in vivo (42). Therefore to further evaluate the integrity of viral mRNAs produced in the presence of DRB, we analysed them by poly(A) selection. Total RNA extracted from DRB-treated or untreated infected 293-T cells was fractionated by affinity chromatography with oligo-dT cellulose and primer extension analysis performed as before. The majority of viral mRNA from DRB-treated or untreated cells alike bound to oligo-dT cellulose and no significant reduction in the proportions of segment 4, 7 or 8 mRNAs (whose expression fails in vivo) that was retained in the poly(A) (+) fraction was seen from treated compared with untreated cells (Figure 3B, compare lanes 4–6 with 1–3). The proportion of NP mRNA (whose expression is resistant to DRB) that fractionated into poly(A) (+) and (−) species also did not change with DRB treatment, although lower amounts overall did not bind oligo-dT cellulose than with the other viral mRNAs (Figure 3B). When these experiments were repeated in MDCK cells, the proportions of poly(A)-selectable viral mRNAs were similarly unchanged by DRB treatment (data not shown). Overall, therefore, no evidence for a DRB-induced defect in viral polyadenylation was seen. Furthermore, cRNA, the minor replicative intermediate species of viral (+) sense RNA that is not capped or polyadenylated (2) was only detectable in the total and A (−) RNA pools and produced a primer extension product that as predicted was around 12 nucleotides shorter than that from mRNA (Figure 3B). This further confirms that viral mRNAs produced in vivo have a 5′ extension as expected from the process of cap snatching (19). In addition, because the mRNA endonuclease activity of the influenza virus polymerase requires doubly methylated cap-1 structures (43), it is likely that these 5′ extensions have normal cap structures. Therefore, in the presence of DRB, all the viral mRNAs tested are synthesized, polyadenylated and capped as normal but only a subset are translated in vivo.

We next tested whether DRB-mediated differential inhibition of viral gene expression operated at the level of mRNA nuclear export, using fluorescence in situ hybridization (FISH) to detect positive sense RNA from segments 5 and 7. In untreated cells at 4.5 h p.i., both NP and M1 mRNAs were largely found in the cytoplasm (Figure 4A–D), while no signal above background was seen from mock-infected cells (panels i–l). In DRB-treated cells, NP mRNA remained largely cytoplasmic, although in some cells, nuclear retention of the transcript was seen (Figure 4, panels e and f). However, complete nuclear retention was observed for segment 7 mRNA after DRB treatment (panels g and h), which was maintained even at later times p.i. (see later).

Figure 4.

Figure 4.

5,6-Dichloro-1-β-D-ribofuranosyl-benzimidazole treatment results in nuclear retention of segment 7 mRNAs. 293-T cells were infected or mock infected and treated with DRB as before, harvested at 4.5 h p.i. and the indicated viral positive sense RNA species detected by FISH (false coloured in red). Cell DNA was also counterstained with 4,6-diamidino-2-phenylindole (DAPI) (false coloured green) to locate nuclei. Images were captured using a wide-field fluorescence microscope fitted with the appropriate filter sets and a 12-bit monochrome camera. Arrows on panel (e) indicate a minority population of cells where DRB treatment induced nuclear retention of NP mRNA. Scale bars: 10 μm.

Our data indicate that in the presence of DRB, M1 mRNAs are synthesized, capped and polyadenylated, but are not exported from the nucleus. To explain this, we hypothesized that ongoing RNAP-II transcription is required to load the necessary cellular factors onto the viral mRNAs in the nucleus to direct their efficient nuclear export. To test this possibility, we made use of the observation that DRB-mediated inhibition of influenza virus replication is reversible (11) in concert with other compounds known to inhibit RNAP-II. Infected cells were incubated with or without DRB as before. After 8 h, the medium was removed, the cells washed twice and reincubated either in normal medium or in medium containing DRB, ActD, H7, CP or ribavirin for a further 3 h. During this second period, cells were labelled with 35S-methionine, and HA, NP and M1 synthesis was then analysed by immunoprecipitation (IP). Camptothecin covalently links topoisomerase 1 to DNA and inhibits transcription, probably by blocking passage of RNAP-II (44,45). Actinomycin D is a DNA intercalator that blocks all DNA-dependent RNA polymerase activity (46), while H7 has an analogous mechanism to DRB in that it inhibits phosphorylation of the RNAP-II CTD, but on serine 5 rather than on serine 2 (40,47,48). The mode of action of ribavirin is complex, involving (in its triphosphate form) direct inhibition of viral transcription as well as inhibition of RNAP-II by perturbing cellular GTP metabolism (49–51). Normal synthesis of HA, NP and M1 was seen when cells were grown in the absence of DRB for the full 11 h, but as before, only NP was synthesized in appreciable quantities if DRB was maintained throughout (Figure 5A, B, compare lanes 1 and 8). If cells were incubated without drug for the first 8 h before any of the compounds were added, no significant inhibition of viral protein synthesis was seen (Figure 5A, B, compare lane 1 with lanes 2–6). This shows that when DRB or other RNAP-II inhibitors are added late in infection, they do not have a marked inhibitory effect on protein synthesis. Furthermore, 3 h after removal of DRB and incubation in the absence of drug, HA and M1 expression could be detected, demonstrating that the effect of DRB on viral gene expression was reversible (Figure 5A, B, lane 7). However, HA and M1 synthesis continued to be inhibited if DRB was substituted by the other RNAP-II inhibitors, although NP expression was maintained (Figure 5A, B lanes 9–12). This same trend was seen when replicate independent experiments were quantified by densitometry. On average, HA and M1 expression recovered to around 50% of that seen in untreated cells after removal of DRB, but synthesis of these two proteins was repressed to around 20% or less of normal levels in the presence of any of the RNAP-II inhibitors (Figure 5C). Thus, the block to late viral gene expression is maintained after replacement of DRB by a variety of transcription inhibitors, supporting the hypothesis that RNAP-II activity is required for efficient translation of M1 and HA mRNA. Furthermore, the diverse chemical structures and inhibitory mechanisms of the drugs used strongly argues for an effect mediated through RNAP-II rather than unforeseen side-effects on other cellular or viral functions.

Figure 5.

Figure 5.

5,6-Dichloro-1-β-D-ribofuranosyl-benzimidazole- mediated inhibition of viral late gene expression is reversible and can be substituted by other RNAP-II inhibitors. 293-T cells were infected (or mock infected; lane 13) with PR8 and incubated with DRB (D) or without drug (−) from 1.5 to 8 h p.i. as indicated. The cells were then washed and overlaid with replacement medium containing [35S]-methionine and either no drug, DRB, ActD (A) H7 (H), camptothecin (C) or ribavirin (R) for a further 3 h. Cell lysates were prepared and subjected to IP using A) anti-PR8 virus (recognizing HA and NP, as labelled) or B) anti-M1 followed by SDS–PAGE and autoradiography; C) M1 and HA expression was quantified by densitometry from replicate experiments and plotted as the mean percentage (± SE, n ≥ 3) of the synthesis seen in the absence of drug treatment throughout infection.

Next, to test if the reversible effect of DRB on viral gene expression operated at the level of mRNA export, we performed a similar drug washout and replacement experiment except that FISH was used to monitor the intracellular localization of segment 7 mRNA. Infected and mock-infected cells were incubated with or without DRB as before. After 6.5 h incubation, medium was removed, the cells washed twice and reincubated either in normal medium or in medium containing DRB, ActD or α-amanitin for 3 h before fixation and processing for FISH. In untreated cells, M1 mRNA was detected mainly in the cytoplasm at 9.5 h p.i. (Figure 6, panel a), a localization pattern that was not altered by the addition of any of the drugs at 6.5 h p.i. (panels b–d). As before, incubation in the presence of DRB from 90 min p.i. resulted in almost total nuclear retention of plus-sense segment 7 RNA (panel g). However, when the drug was removed at 6.5 h p.i. and the cells allowed to recover for 3 h, the majority of M1 mRNA was now found in the cytoplasm (panel f). Furthermore, when DRB treatment was replaced after 6.5 h p.i. by ActD or α-amanitin, segment 7 mRNA was still retained in the nucleus (panels h and i). The degree of nuclear retention in the presence of α-amanitin was less than that seen with ActD, possibly resulting from the slower kinetics with which the former drug inactivates RNAP-II (52). Overall, these data support the hypotheses that DRB inhibits viral gene expression by blocking nuclear export of specific viral mRNAs and that more generally, RNAP-II activity is required for this step in the viral life cycle.

Figure 6.

Figure 6.

5,6-Dichloro-1-β-D-ribofuranosyl-benzimidazole- mediated nuclear retention of segment 7 mRNA is reversible and can be maintained by other inhibitors of RNAP-II. 293-T cells were infected (panels a–d and f–i) or mock infected (panels e and j) with PR8 and incubated for 6.5 h either without drug (top row) or with DRB from 90 min p.i. (bottom row). After this, cells were washed twice and overlaid with medium containing no drug (including mock infected cells), DRB, ActD or α-amanitin (α-Am) as labelled for 3 h before fixation and processing for FISH to detect segment 7 positive sense RNA. Images were captured using a Leica TCS SP confocal microscope. Scale bars: 10 μm.

To further test the necessity of RNAP-II activity for segment 7 mRNA export, we tested whether ActD would inhibit export as DRB does when applied at an earlier time in infection. Cells were infected and either incubated in the absence of drug, with ActD from 90 min p.i. or with leptomycin B (LMB) from 30 min p.i. Leptomycin B is a specific inhibitor of the cellular protein CRM1/exportin 1 (53) that inhibits nuclear export of viral RNPs without significantly affecting viral gene expression (54). At 4.5 h p.i., cells were fixed and the intracellular localization of M1 mRNA monitored by FISH and (as a counterstain), NP protein by indirect immunofluorescence. No staining for either molecule was observed in uninfected cells (Figure 7, panels g and h). As expected at this time-point, NP was concentrated at the nuclear periphery in untreated and LMB-treated cells (54,55), while in ActD-treated cells, it showed a more diffuse intranuclear staining pattern (Figure 7, panels b, d and f). As before (Figure 4), plus-sense segment 7 RNA was mainly cytoplasmic in untreated cells (Figure 7, panels a, b) and this was not significantly altered by LMB treatment (panels e, f). However, after ActD treatment, segment 7 mRNA was still detectable, but was solely located in the nuclei of cells (panels c, d), confirming that the inhibition of viral mRNA nuclear export is not only a consequence of DRB treatment.

Figure 7.

Figure 7.

Actinomycin D but not LMB treatment inhibits segment 7 mRNA nuclear export. 293-T cells were infected or mock infected with PR8 and incubated without drug (including the mock infected sample) or with ActD or LMB from 90 and 30 min p.i., respectively, until 4.5 h p.i. when they were fixed and stained for segment 7 mRNA (false coloured red) and NP protein (false coloured green). Images were captured using a wide-field fluorescence microscope fitted with the appropriate filter sets and a 12-bit monochrome camera. Scale bars: 10 μm.

Discussion

We demonstrate here that DRB treatment inhibits the expression of a subset of influenza virus genes at the post-transcriptional level by blocking mRNA nuclear export and that this inhibition is reversible yet maintained in the presence of a variety of other RNAP-II inhibitors. The dependence of influenza virus replication on RNAP-II activity to supply 5′-mRNA cap structures is well established through biochemical and genetic analysis of the mechanism of viral mRNA synthesis (2,15,18–20). However, much evidence has accrued over the years suggesting this is not the only reason RNAP-II activity is required. Although insults to cellular DNA-dependent RNA synthesis abrogate viral mRNA synthesis when the damage is initiated prior to or concomitant with the onset of infection, addition of many transcription toxins later in infection is still inhibitory to the overall viral replication process but not by blocking viral mRNA synthesis. Many experiments observed specific inhibition of late viral gene expression after ultraviolet (UV) irradiation, CP or ActD treatment (7,12,21,23). In the case of ActD, several studies noted that even when added at the start of infection, the drug did not entirely inhibit viral mRNA synthesis but that (as assayed by cell fractionation) the reduced quantities of transcripts made were retained in the nucleus (3,22,23,56), consistent with the results of our FISH analysis (Figure 7).

More recently, it was noted that the kinase inhibitor H7 blocks viral late gene expression (24,25). Subsequent studies indeed used this phenomenon as a tool to probe nuclear export of viral RNPs (57,58). Similarly, to DRB, H7 inhibited HA and M1 protein expression without (where examined) inhibiting mRNA synthesis. However, in contrast to our results, variable quantities of NS1 and NS2 were made following H7 treatment, with one study showing complete inhibition of NS2 accumulation (58). Kurokawa et al. (25) suggested that H7 selectively inhibited translation of viral mRNAs but consistent with our data here, Vogel et al. (26) subsequently showed that the drug retained HA but not NP mRNA in the nuclei of infected chick embryo fibroblasts. Again, similar to the effects of DRB, nuclear retention of HA mRNA was relieved on removal of the drug (26). H7 was originally characterized as an inhibitor of protein kinase C (59) and this is how most published studies on influenza virus have viewed it. However, as discussed above, it also inhibits cyclin-dependent kinase 7, a component of the transcription factor IIH complex, which is responsible for phosphorylation of serine 5 of the RNAP-II CTD (40,47,48).

Overall, therefore, we propose that RNAP-II transcription is required not just to supply pre-mRNAs to act as cap-donors, but also to support nuclear export of influenza virus mRNA. This hypothesis unifies a large number of observations made over many years with a diverse set of chemical inhibitors and suggests a new facet to the many previously characterized interactions between the virus and cellular nuclear function [reviewed by Amorim and Digard (60)]. The hypothesis makes the testable prediction that specific cellular factor(s) needed for efficient transport of the viral mRNAs are loaded onto the transcripts in the presence of active cellular transcription but not in its absence. Plausible candidates include cellular hnRNP proteins involved in mRNA export whose nucleocytoplasmic shuttling ceases after ActD treatment (61). Use of the CRM1/HuR-dependent pathway responsible for exporting a subset of AU-rich cellular mRNAs (62) seems unlikely, given the insensitivity of viral gene expression in general (54) and segment 7 mRNA in particular (Figure 7) to LMB treatment. In possible opposition to our hypothesis, however, a recent study used chromatin IP analysis to show inhibition of RNAP-II elongation in infected cells (33). However, the reduction in amount of RNAP-II associated with promoter distal regions of the two genes examined was modest, only around twofold, and this degree of inhibition may not be sufficient to interfere with mRNA nuclear export.

It seems that cellular transcription is required for a post-transcriptional step in the expression of the majority of influenza virus genes; as well as M1, M2 and HA [Figure 1; (3,12,21,23–25)], DRB and H7 also decreased expression of the viral polymerase subunits (data not shown). However, NP and (to a lesser extent) NS2 synthesis persisted after RNAP-II inhibition. For NS2, it is possible that the processing of its mRNA by the splicing machinery is sufficient to direct nuclear export even in the absence of RNAP-II transcription. Although normally cellular splicing is linked to RNAP-II transcription, at least in frog oocytes, uncoupling splicing from transcription by microinjecting a pre-mRNA rendered its processing DRB insensitive (63). This may be analogous to synthesis of the intron-containing NS1 mRNA by the influenza virus RNA polymerase. M2 synthesis would not be predicted to result from such a route, however, as expression of the influenza virus polymerase is necessary to promote use of the M2 splice site (64). Splicing, however, is not plausible to explain RNAP-II inhibitor-resistant nuclear export of NP mRNA. One possibility is that NP expression survives when DRB is added at 90 min p.i. because as an early gene, it is transcribed faster and therefore makes use of the window before the inhibitor was applied. However, this does not explain the sensitivity of the other major early viral gene, NS1 and nor do we see any notable difference in the kinetics of segment 5 mRNA accumulation compared with the other viral mRNAs (Figure 3). Furthermore, we find that NP expression is relatively resistant to DRB addition at earlier times of drug addition, including at the time of infection (data not shown). Therefore, it is tempting to speculate that segment 5 mRNA utilizes a different export pathway to the other viral mRNAs. Consistent with this hypothesis, one study examining the kinetics of viral gene expression found that the rate of HA, M1 and NS1 synthesis lagged approximately 1 h behind the accumulation of their mRNAs, but that there was no delay in NP expression (7). The authors suggested that the barrier to efficient expression of the former three mRNAs at early times p.i. might be either at the level of mRNA translation or nuclear export and that this mechanism played a part in the temporal regulation of viral gene expression. A better understanding of the intracellular trafficking of influenza virus mRNAs will allow us to test this hypothesis.

Materials and methods

Virus, cells, plasmids, antibodies, primers and other compounds

Influenza virus strain PR8 was propagated in 10-day-old embryonated eggs as described previously (54). Infections were carried out at an m.o.i of 10. Human embryonic kidney 293-T cells and MDCK cells were cultured in DMEM supplemented with L-glutamine, penicillin, streptomycin and 10% fetal calf serum. Plasmids pCDNA-NP (65) and pCDNA-M1 (generous gift of Dr Debra Elton) contain cDNA copies of segments 5 and 7 of PR8 virus, respectively, flanked by the bacteriophage T7 (positive sense) and SP6 promoters (negative sense). Antibodies directed against the CTD repeat of RNAP-II were obtained from Covance [Berkeley, CA, USA (H5, H14 and 8wg16, to detect phosphorylated serine 2, 5 and unphosphorylated serine 2, respectively) (31)]. Antisera against LAP2 were purchased from Transduction Laboratories (Lexington, KY, USA). Anti-digoxigenin and anti-M2 (14C2) antisera were purchased from AbCam (Cambridge, UK). Polyclonal anti-PR8 serum was made by immunizing rabbits with UV-inactivated whole virus. Antisera against PR8 M1 was prepared by immunizing rabbits with a glutathione-S-transferase-M1 fusion protein purified from Escherichia coli. Species-specific fluorescent or horseradish peroxidase-conjugated secondary anti-IgG antibodies were purchased from Molecular Probes (Paisley, UK) and DAKO (Ely, UK). 5,6-Dichloro-1-β-D-ribofuranosyl-benzimidazole was purchased from Calbiochem (Darmstadt, Germany) and reconstituted to a stock solution of 150 mM in dimethyl sulphoxide (DMSO). It was generally used at a final concentration of 150 μM for experiments in 293-T cells and 100 μM for experiments in MDCK cells. These concentrations were chosen on the basis of titration experiments measuring inhibition of HA and M1 expression but essentially the same effects were seen with DRB concentrations ranging from 75 to 200 μM (data not shown). Actinomycin D (Sigma, Surrey, UK) was solubilized in methanol to a stock of 2 mg/mL and used at 0.1 μg/mL (13). H7 was purchased from LC Laboratories (Woburn, MA, USA), dissolved in water to 25 mg/mL and used at 25 μg/mL. Leptomycin B was purchased from LC Laboratories, stored in ethanol and used at a concentration of 11 nM (54). Camptothecin and ribavirin (Sigma) were dissolved in DMSO and water to 20 and 200 mg/mL and used at 20 μM and 200 μM, respectively. L-[35S] Methionine (800 Ci/mmol) and γ[32P] ATP (3000 Ci/mmol) were bought from Amersham/GE Healthcare International (Little Chalfont, Buckinghamshire, UK).

Virus infection and protein analyses

293-T cells (2 × 105) or MDCK cells (4 × 105) were infected in serum-free medium and at 90 min p.i., overlaid with complete medium containing drugs as necessary. At the appropriate time, cells were collected and solubilized in 200–250 μL of IP buffer (100 mM KCl, 50 mM Tris pH 7.6, 5 mM MgCl2), supplemented with 1% Triton-X-100, 0.1% sodium dodecyl sulfate (SDS), 1% Na deoxycholate. For IP analysis, 10 μL of the cell lysates were diluted in 90 μL of IP buffer supplemented with 1 mM DTT and 0.1% Nonidet P-40 and incubated on ice for 60 min with the desired antisera before 40 μL of a 50% (v/v) slurry of protein A Sephadex beads were added and incubated with rotation for a further 30 min at 4°C. The beads were then collected and washed three times by centrifugation before bound material was eluted by boiling in Laemmli's sample buffer and analysed by SDS–PAGE and autoradiography. Densitometry of exposed X-ray film was carried out using the programme NIH image (http://rsb.info.nih.gov/nih-image/). For Western blot analysis, samples were separated by SDS–PAGE (or tris-tricine-SDS–PAGE to detect M2), transferred to nitrocellulose using a semi-dry transfer cell and bound proteins detected by primary and secondary antibodies followed by chemiluminescent development using standard protocols (ECL, Amersham, Princeton, NJ, USA/GE Healthcare).

RNA analysis

Total cellular RNA was isolated using a commercial kit (SV total RNA Isolation System, Promega, Southampton, UK). For in vitro translation reactions, aliquots of the isolated RNA were added to micrococcal nuclease-treated rabbit reticulocyte lysate (Promega) supplemented with 0.6 μCi/μL [35S]-methionine and incubated at 37°C for 1.5 h. Primer extension reactions to detect influenza virus RNAs were performed as previously described (41). Oligonucleotides to detect positive sense segment 5 and segments 7 and 8 (unspliced forms only) RNAs have been previously described (41,66). To detect positive sense RNA from segment 4, the oligonucleotide 5′ TCA CTG TCA CAT TCT TCT GCA GC 3′ was used. The predicted size of primer extension products were confirmed with respect to matching DNA sequencing ladders (data not shown).

To generate FISH probes, plasmids pCDNA-NP and pCDNA-M1 were linearized by digestion with KpnI or HindIII, respectively, and used as templates for in vitro transcription after purification using a Qiagen QiaQuick PCR Purification kit. Transcription with 40 U of SP6 RNA polymerase (Roche, Welwyn Garden City, UK) was carried out in manufacturer's buffer supplemented with 5 mM DTT, 0.5 mM each of ATP, CTP, GTP, 12 μM UTP, 0.25 mM digoxigenin-UTP (Roche), 0.5 U/μL RNAguard ribonuclease inhibitor (Pharmacia, NY, USA), and 1 μg template DNA in a final volume of 20 μL. The reaction mixture was incubated at 37°C for 1 h 15 min. before adding 1 U of RNAse-free DNAse I (Promega). After 15 min at 37°C, 80 μL of a solution containing 6.25 mM ethylenediaminetetraacetic acid (EDTA) pH 8, 1 M LiCl, 2.5 mg/mL single-stranded salmon sperm DNA was added and nucleic acid recovered by ethanol precipitation. RNA probes were dissolved in 100-μL 10 mM Tris pH 8, 1 mM DTT supplemented with 0.2 U/μL RNAguard ribonuclease inhibitor. Their integrity and concentration was assayed by spotting serial 10-fold dilutions onto nitrocellulose. After UV cross-linking, bound labelled RNA was detected colorimetrically by incubation with an alkaline-phosphatase-conjugated anti-digoxigenin antibody (Roche) followed by development with 4-nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl-phosphate. Useful probe preparations gave signals down to 10−3.

Positive sense RNA probes produced by the same method and used in the FISH procedure (see below) are capable of detecting vRNA in infected cells (M.-J. Amorim and P. Digard, unpublished data). It is likely therefore that the negative sense probes described here also detect cRNA. However, cRNA represents a minor proportion (5–10%) of the total positive sense RNA in infected cells [Figure 3B; (3,41)] and so is unlikely to significantly interfere with viral mRNA detection. In the case of segment 7, the negative sense probe was prepared from a cDNA clone of the entire segment and would therefore be predicted to hybridize with both spliced and unspliced mRNA products. However, most segment 7 mRNA remains unspliced in infected cells (8), so the majority of the FISH signal is predicted to result from M1 mRNA.

FISH and immunofluorescence analysis

For FISH analysis, 293-T cells were seeded at a density of 2 × 105 cells on glass coverslips previously coated with 0.1 mg/mL poly-D-lysine. After infection, cells were fixed for 20 min using 4% formaldehyde in PBS, washed three times for 5 min in PBS, then dehydrated by rapid washes with increasing concentrations of ethanol [25, 50, 75, 100% (v/v)]. If necessary, cells were stored overnight at 4°C in 100% ethanol. Next, samples were rehydrated using the same ethanol series in reverse, washed again with PBS and permeabilized using 0.2% (v/v) Triton-X-100 in PBS for 7 min followed by more PBS washes. Cells were then digested with 2.5 μg/mL of Pronase (Roche) in PBS for 15 min at 37°C before quenching with 2 mg/mL glycine in PBS for 10 min. Cells were washed again, postfixed using a 1% (v/v) formaldehyde solution in PBS for 10 min and washed again with PBS. The prehybridization mix (60% formamide, 0.3 M sodium chloride, 30 mM sodium citrate pH 7.0, 10 mM EDTA pH 8, 35 mM NaH2PO4, 5% dextran sulphate (w/v), 250 ng/mL tRNA) was added to the cells and incubated for 1 h at 37°C. During this period, digoxigenin-labelled ribonucleotide probes were heated to 90°C for 5 min and placed on ice for further 5 min prior to dilution to a final concentration of 0.3% (v/v) in 0.5 mL of prehybridization solution containing 20 U/mL RNAguard RNAse inhibitor. Cells were then incubated with the probe mix for at least 16 h at 37°C then washed three times for 15 min at room temperature with 60% formamide, 0.3 M sodium chloride, 30 mM sodium citrate pH 7.0. Further washes in PBS for 5 min and 100 mM Tris pH 7.5, 150 mM NaCl (Buffer 1) for 2 min followed before the addition of blocking solution [1% Boehringer Mannheim (Mannheim, Germany) blocking reagent in buffer 1] for 30 min at room temperature. Bound probe was then detected by indirect immunofluorescence with anti-digoxigenin antibodies, along with parallel immunostaining of proteins as appropriate (65). Images were captured with a Leica TCS NT confocal microscope or an Olympus IX70 fluorescence microscope fitted with a Retiga 2000R cooled 12-bit monochrome camera. Image capture parameters were kept constant within individual experiments. Postcapture processing was carried out using Adobe Photoshop and brightness/contrast adjustments were applied evenly across figure panels to allow daylight visualization of images.

Acknowledgments

We thank Drs Ian Goodfellow and Debra Elton for advice, discussion and gifts of reagents. This work was supported by grants from the Wellcome Trust (no. 073126), BBSRC (no. S18874) and Medical Research Council (no. G9901213), to P. D. M. J. A. is supported by the Gulbenkian PhD program in Biomedicine and the Fundação para a Ciência e Tecnologia.

Ancillary